|USFDC Home | USF Electronic Theses and Dissertations||| RSS|
This item is only available as the following downloads:
xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim
leader nam 22 Ka 4500
controlfield tag 007 cr-bnu---uuuuu
008 s2010 flu s 000 0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0004711
Examination of the effect of child abuse case characteristics on the time a caseworker devotes to a case
h [electronic resource] /
by Christopher Card.
[Tampa, Fla] :
b University of South Florida,
Title from PDF of title page.
Document formatted into pages; contains X pages.
Dissertation (PHD)--University of South Florida, 2010.
Includes bibliographical references.
Text (Electronic thesis) in PDF format.
Mode of access: World Wide Web.
System requirements: World Wide Web browser and PDF reader.
ABSTRACT: This study used an explanatory research model that determined the effect on caseworker time and therefore workload caused by specific characteristics of cases assigned after the child abuse investigation is complete. The purpose of this study was to explain the relationship between child protection case characteristics and the time an assigned caseworker devotes to a case. With this knowledge an informed methodology to assess the current workload of a caseworker could be used to assure that the caseworker is able to successfully complete the tasks required for each child assigned. Further, the knowledge of the amount of time spent on a case with specific characteristics allows supervisors to assess and properly assign cases. Utilizing focus groups and a secondary data analysis of the Florida State Automated Child Welfare Service Information System (SACWSIS) the case characteristics of race/ethnicity, living arrangement, placement, removal and prior removal were found to significantly affect caseworker time spent on a case. Additionally, the case characteristics of gender, age, type of maltreatment, and disability were not found to affect caseworker time spent on a case.
Advisor: William Rowe, D.S.W.
x Social Work
t USF Electronic Theses and Dissertations.
Examination of the Effect of Child Abuse Case Characteristics on the Time a Caseworker Devotes to a Case by Christopher J. Card A dissertation submitted in fulfillment of the requirements for the degree of Doctor of Philosophy School of Social W ork College of Behavioral and Community Sciences University of South Florida Co Major Professor: William Rowe, DSW. Co Major Professor: Mary Armstrong, Ph.D. Roger Boothroyd, Ph.D. Steve Freedman, Ph.D. Date of Approval: October 27, 2010 Keywords : caseload, workload, child abuse, case assignment, workload analysis Copyright 2010 Christopher J. Card
Acknowledgements I want acknowledge Dr. Svetlana Yampolskaya and Mr. Keith Perlman for their assistance in securing the SACWSIS data and in inter preting and organizing this large data base
i Table of Contents List of Tables iii List of Figures v Abstract vi Chapter I. Introduction 1 Purpose of Study 10 Social Work Practice Theoretical Hypothesis 12 Research Questions 13 Chapter II. Literature Rev iew 15 Background 15 Caseworker Survey Results 17 Caseload Studies 21 Summit County Ohio 21 Lutheran Social Services of Washington and Idaho 22 New Mexico Department of Human Services 24 Mid West Non Profit Agency 28 New York State Child Welfare Workload Study 30 Californ ia Child Welfare Workload Study 33 Summary of Caseload Studies 36 Child Welfare League of America Standards of Excellence 38 Council on Accreditation Standards 40 Summary of Literature 40 Chapter III. Methods 42 Caseworker and Supervis or Focus Groups 47 Caseworker Job Descriptions 51 Florida State Automated Child Welfare Services Information System (SACWSIS) data set 53 SACWSIS Data Collection Procedures 57 Case Characteristics, the Independent Variables 60 Analysis and Description of the Variables 63 Caseworker Time, the Dependent Variable 78 Chapter IV. Results 88 Caseworker Focus Group Results 88 Supervisor Focus Group Results 92 Job Description Results 96
ii Analytical Phase I: 100 Focus Groups, job descriptions, and SACWSIS data set independent variables and general activity categories comparison. Analytical Phase II: 111 Comparison of variables between lead agency study cases and nonlead agency cases using the independent samples t test and chi square. Analytical Phase III : 144 Bivariate Relationship between each Independent Variable and the Dependent Variable Analytical Phase IV: 147 Multivariate Regression and Stepwise Analysis Chapter V. Discussion and Findings 161 Finding I 161 Finding II 169 Finding III 175 Finding IV 181 Limitation of Study 194 Practice, Policy, and Research Recommendations 196 Practice Recommendations 197 Policy Recommendations 198 Research Opportunities 199 References 201 Appendix A. CWLA Standards of Excellence 213 Appendix B. Guidelines f or Caseload Standards 215 Appendix C. Caseworker and Supervisor Focus Group Questions 220 Appendix D. Voluntary Consent for Focus Group Participation 221
iii List of Tables Table 1. Cases Open and Closed within the study time frame 59 Tab le 2 Caseworker I dentified Characteristics 89 Table 3 Supervisor Identified Characteristics 92 Table 4 Job Description Tasks by Agency 97 Table 5 Mental Health Condition: Lead Agency Study Cases 101, 116 Table 6 Developmental Disability: Lead Agency Study Cases 102, 120 Table 7 Case Characteristics Comparison 106 Table 8 Caseworker Task Matrix 110 Table 9 Age: Lead Agency Study Cases 113 Table 10. Gender: Lead Agency Study Cases 115 Table 11. Removal: Lead Agency Study Cases 117 Table 12. Race/Ethnicity of the Child: Lead Agency Study Cases 118 Table 1 3. Living Arrangement: Lead Agency Study Cases 121 Table 14. Living Arrangement nonLead Agency Cases 122 Table 15. Living Arrangement Chi Square Values 123 Table 16. Maltreatment Type: Lead Agency Study Ca ses 124 Table 17. Maltreatment Type: nonLead Agency Cases 126 Table 18. Maltreatment Type Chi Square Values 127 Table 19. Substance Abuse: Lead Agency Study Cases 128 Table 20. Domestic Violence: Lead Agency Study Cases 130
iv Tab le 21. Placement T ype: Lead Agency Study Cases 133 Table 22. Placement Type: nonLead Agency Cases 134 Table 23. Placement Type Chi Square Values 135 Table 24. Prior R emoval: Lead Agency Study Cases 137 Table 25. Caseworker Time: Lead Agency Study Cases 139 Table 26. Comple te Activities by Type: Lead Agency Cases 141 Table 27. Time by Activity Type: Lead Agency Study Cases 142 Table 28. Bivariate Relationship between Independent and Dependent Variables 145 Table 29. Removal Model Regression 150 Ta ble 30. Removal Model S tepwise Regression 153 Table 31. Placement Model Regression 156 Table 32. Placement Model Stepwise Regression 159 Table 3 3. Comparison of Activities Used in Caseworker Time Studies 164
v List of Figures Figure 1. Knowledge and Workload Impact in Case Assignments for Child Welfare Caseworkers 12 Figure 2. Multiple Regression Equation with removal (all variables Model) 85 Figure 3. Multiple Regression Equation without removal, (Placement Model) 85 Figure 4. Multivariate Step wise Regression Equati on 86 Figure 5. Time for Complete A ctivities of 8 Hours or Less 139 F igure 6. Dependent Variable2 140
vi Abstract Examination of the Effect of Child Abuse Case Characteristics on the Time a Caseworker Devotes to a Case Christopher J. Card This study u sed an explanatory research model that determine d the effect on caseworker time and therefore workload caused by specific characteristics of cases assigned after the child abuse investigation is complete. The purpose of this study was to explain the relati onship between child protection case characteristics and the time an assigned caseworker devotes to a case. With this knowledge an informed methodology to assess the current workload of a caseworker could be used to assure that the caseworker is able to su ccessfully complete the tasks required for each child assigned Further, the knowledge of the amount of time spent on a case wit h specific characteristics allow s supervisors to assess and properly assign cases. Utilizing focus groups and a secondary data analysis of the Florida State Automated Child Welfare Service Information System (SACWSIS) the case characteristics of race/ethnicity, living arrangement, placement, removal and prior removal were found to significantly affect caseworker time spent on a case. Additionally, the case characteristics of gender, age, type of maltreatment, and disability were not found to affect caseworker time spent on a case.
1 Chapter I. Introduction The position of a child protection caseworker has long been one of the most demanding and difficult in the human service s field. Th e caseworker position is the core intervention strategy every state uses to protect children from being abused and neglected and to determin e long term living arrangements of these children. Because of this key role, caseworkers perform ance determines the success or failure of our child protection system for each child and family that enters. The caseworker s performance is also the primary factor in determining the overall cost of service and the ultimate safety and permanency of each child. This study looked at the impact of specific case characteristics on the amount of time a caseworker spends on a particular case. The study utilize d data specific to the S tate of Florida and discusses the generalizability of the findings The study consist ed of two general phases of analysis. First data w ere collected from focus groups of caseworkers and supervisors. This data indicate d which case characteristics these active professionals believe impact caseworker time the most. Additionally, the focus groups identif ied which tasks caseworkers are required to perform that are critical to understanding and measuring the actual time a caseworker spends on a specific case. Included in th is first phase of analysis was the collection of job descriptions from various agencies employing child welfare caseworkers in Florida. The job descriptions w ere analyzed to determine the professional tasks commonly required of the caseworkers. The data from these sources w ere then compared with the actual case characteristics and tasks that are recorded in the Florida State Automated Child Welfare Services Information System
2 (SACWSIS). This first phase analysis indicate d the strength of the SACWSIS data base by identifying the case char acteristics and job tasks common to the data base, focus group findings of the active professionals, and the job descriptions. The gaps between the database, the focus group results and the job descriptions will be discussed as validation or gaps in the database. The second phase of analysis w as a secondary data analysis of the SACWSIS data The dependent variable was the recorded caseworker time and the independent variables were the case characteristics in the database. This analysis reveal ed those cha racteristics that have a significant impact on the amount of caseworker time spent on a case. This study was intended to provide the assigning supervisors a basis for caseload distribution and performance expectation based upon the presenting case charac teristics. Further this study will assist in the assessment of a specific caseload and the potential time commitment to completing it. With this knowledge workloads for caseworkers can be managed to allow for successful completion of tasks within a reasonable amount of time. Uniformly the tasks assigned to caseworkers are numerous, intense, and carry a high degree of responsibility that all others involved in the case are dependent upon (Child Welfare League of America, 2002; Zlotnik, DePanfilis, Daining, & Lane 2005) These tasks center on coordination and communication. The caseworker is tasked to negotiate a service plan that is focused on keeping children safe and resolving the issues of abuse to the extent that a permanent living situation can be achi eved for the child as soon as possible. Once the plan is negotiated and other vested parties are in agreement
3 then the caseworker must coordinate the services and assure that the child and family are receiving and benefiting from the services. This coordination can be quite involved and can include arranging for the referral to the identified service and making certain that the child or family member is enrolled into the service. Often there are eligibility issues, fee issues, transportation hurdles, and wa iting lists that must be navigated and resolved. T ransportation to services is often one of the biggest hurdles for families Subsequently, t he caseworker is often burdened with the responsibility for making certain the child and/or family member is able to get to the service. Transportation is often c ited by caseworkers as a time consuming and stressful issue they must contend with (American Federation of State, County, and Municipal Employees (AFSCME) Survey 1998; Cyphers, American Public Human Services Association (APHSA) 2001; Child Welfare League of America, 2002). This is often complicated by the family not having a working phone and moving frequently among other issues. Further, the caseworker must assure that the services are thoroughly complete d and have been effective in resolving the issues that indicated the particular service was necessary. This requires caseworkers to stay in constant communication with collateral professionals and agencies involved with the services and/or case as well as collecting copies of any certificates of completion or letters of completion from the various service providers These professionals and agencies include the school, the Court Appointed volunteer (Court Appointed Special Advocates or Guardian Ad Litems), t he therapists, homemakers, child care center, attorneys for the state/parents/agencies, foster parents, domestic violence professionals, substance abuse professionals and the housing authority, to name a few Likewise, these professionals must be kept inf ormed in a
4 timely manner of the events and progress of the child and family on the service plan and the goals. Added to these tasks is the responsibilit y of determining whether a child stays in his or her family home, or is placed with a relative or pl aced in foster care. After such a determination, the caseworker must continuously assure the childs safety in their current environment by monitoring the placement Further the caseworker must prepare and present comprehensive reports to the court, docume nt all activities in an elaborate data system assure the case file is up to date and complete, gather all necessary documents related to the child and family, and complete the significant amount of paper and electronic documentation required by federal an d state laws and reg ulations. Child welfare caseworkers are given the responsibility of protecting childrens lives. The stories that usually make the headlines are regarding children that are killed by their parents, moms boyfriend, or the caregivers with whom the child is living. Each time there is immense scrutiny over every task and phone call the caseworker has performed. Every note and each piece of documentation in the case record is carefully reviewed and matched to see if the extensive policies and procedures were followed exactly. All of this is done to determine whether the caseworker made an error. If an error is found the caseworker can be front page news, terminated, and even prosecuted. As a key component of the responsibility of protec tion, the caseworker is the primary source /witness of information regarding all civil levels of protection from abuse and neglect for children in our country. Without the caseworkers testimony and recommendations the court would be seriously crippled in m aking a decision regarding the safety and placement of the child. Further, the caseworker often is the primary
5 witness to the parents progress and completion of case plan tasks, sobriety, employment, and housing. Often, if the court returns a child to the parents and the child is hurt again the court claims it only acted on the information that the caseworker provided, thus putting the full burden of responsibility for the error in judgment on the caseworker. Equally important and stressful is the fact th at the caseworker often plays an important role in any criminal investigation and prosecution of the child abuser This role can further alienate the worker from the family if the abuser is a family member. The time, effort and concentration it takes for t he caseworker to properly prepare for such a critical role in the court system is often neglected as a key task. Caseworkers have reported that they are not prepared for these high level civil and criminal responsibilities (Lieberman, Hornby & Russell, 1988). Cases presented for child protection services vary a great deal. The case requirements for successful completion are likely driven by the demographics of the case such as the age of the victim child, the number of victim children, other siblings, menta l health and behavior patterns of the victim child(ren), location of the family (rural versus urban), the mothers condition and availability, family income, the fathers condition and availability, relatives availability, substance abuse and/or mental hea lth issues of the care takers/parents, among other factors. Other components that may affect the time commitment of the caseworker include the type of abuse (sexual, physical, neglect, abandonment), any history of abuse, placement options and the number of placements for the victim child(ren), the stability of the family (place to live, livable wage), access to services needed, and the willingness of the family to resolve the issues at hand. Some of these factors are difficult to measure and may not become clear until the case is well
6 underway. This study examined at these characteristics and their affect on the amount of time and work needed to complete and close the case. The burden on child protection caseworkers is tremendous. Studies have indicated tha t c hild protection c aseworkers show more depersonalization, less job satisfaction, more values conflict, role conflict and confusion than workers in family service settings and community mental health agencies (Jayaratne & Chess, 1984; Laird, 1985) The child protection caseworker must cope with the constant threat of civil and criminal liability without protections of immunity (Alexander & Alexander, 1995) and the ridicule of the media and other professionals that focus on the tragedies of child abuse ( Ell et 2006; Morris 2005). Yet, t hese same caseworkers provide legally mandated services and are often the sole person standing between the safety and the harm of a child. The Child Welfare League of America (2002) and the United States Government Accountabi lity Office (GAO) (2003, 2006) report the number of children needing child protection services continues to increase yet the retention of qualified caseworkers is decreasing. State and child welfare agencies from across the country have tried to address t his crisis of caseworker turnover with no notable success (GAO 2003, 2006). Caseworker caseload has been found to be a critical factor in the turnover crisis through both qualitative and quantitative research ( Zlotnik, DePanfilis, Daining, & Lane 2005). According to a systematic review of research and outcomes related to recruitment and retention of child welfare professionals published by the Institute for the Advancement of Social Work Research, professional commitment and level of education are the m ost consistent personal characteristics and supervisory support and workload/caseload are the
7 most consistent organizational factors identified as influencing recruitment and retention ( Zlotnik, et. al, 2005) Caseloads for child protection caseworkers ar e typically not weighted or assigned with any thought of matching the skills of a caseworker with the needs of the case or with the time a particular case may consume and the time a caseworker may have available. As indicated in the second chapter of this study there is no valid methodology utilized widely that objectively assesses a child protection case for workload impact and assigns the case in an informed manner to an appropriate caseworker. Over the last 10 years the American Humane Association has co nducted several time studies with child protection caseworkers from various states. These studies documented all qualifying staff and all of their work time over a limited number of weeks. These studies document well how caseworkers are currently spending their time and each study made some recommendation regarding caseloads, (Tooman & Fluke, 2002). A select number of these studies are discussed in the following chapter. The most quoted and referenced standards for caseloads in child protection are the Chi ld Welfare League of America (CWLA) Standards of Excellence for Services for Abused or Neglected Children and Their Families and the Standards of Excellence for Family Foster Care Services (CWLA Standards of Excellence, 1999) These excellent manuals make specific recommendations on caseload size and outline items to consider in determining appropriate caseloads. However, there is a dearth of empirical evidence that evaluates these components or any others in determining appropriate caseload size. Further the CWLA recommendations for caseload size do not differentiate types of cases and how much time each case may require, there is no discussion of any recommendation considering the variation of casework necessary for a child at age 3
8 versus a child at ag e 15, nor for a physical abuse case versus a sexual abuse or neglect case. The discussion for caseload size appears to be limited to the tasks assigned to the caseworker, the support systems available, and the expectations of the agency or community on the intensity of work expected. While these are valid considerations, it seems logical to assume the amount of tasks and the type of tasks a caseworker must accomplish will vary based on the characteristics of each case, each child, and each family. Additionally, the CWLA introduced a national accreditation specifically directed at child protection services in 1977. The Council on Accreditation (COA) developed comprehensive standards covering 38 different services areas and over 60 types of programs (Council On Accreditation Standards, 2006) These standards were developed through a series of consensus meetings and reviews of current practices by selected professionals from around the country. Over the years these standards have been refined significantly and currently a number of states require COA accreditation for community agencies providing child welfare services and several state agencies have become accredited themselves. COA articulates within their standards that the known literature about the child w elfare workforce suggests that high caseloads and timeconsuming paperwork are primary factors in turnover. According to the Government Accountability Office ( 2003) their analysis of federal Child and Family Service Reviews ( CFSRs ) corroborated caseworker accounts, showing that large caseloads and worker turnover delay the timeliness of investigations and limit the frequency of worker visits with children, thereby hampering agencies attainment of some key federal safety and permanency goals (GAO, 2003) R esearch and literature also suggest that high turnover
9 rates impact timeliness of reunification and foster parent retention ( Ryan, Garnier, Zyphur, & Zhai 2005) COA recommends caseloads not to exceed 18 children or 8 children with special therapeutic nee ds for foster care and kinship care caseworkers, 12 to 25 families for adoption caseworkers, 12 to 18 families for family preservation services and 2 to 6 families providing intensive family preservation and stabilization services. COA further recommends no more than 15 active investigations for Child Protection Investigation caseworkers with no more than 15 to 30 open cases. Dalton and Morelli (1988) discuss the concept of casemix using a formula that measures the completion of services assigned to a case based upon a diagnosis of the presenting issues. In their study they emphasize the characteristics of a case as drivers to the services needed. Thus different characteristics lead to varying degrees of effort needed to resolve the issues. The casemi x is the variety of cases needing different services that make up a caseload. Caseload studies and consideration cannot overlook special cha racteristics of cases (Toolman & Fluke, 2002) Given these references it is curio us that COA does not consider other case characteristics in setting the standards for caseloads other than special therapeutic needs. COA further qualifies the recommended caseloads by stating that a manageable workload w ould allow caseworkers to meet practice requirements, would not impede the achievement of outcomes, and would take into consideration the qualifications and competencies of the worker as well as the case status. Some o ther characteristics that ought to be considered are the complexity of the case, ages and number of children involved, the competencies, strengths, and weaknesses of the parents, and the availability of services ( Walter R. McDonald & Associates and American H umane 2006). COA also appears to assume that a caseworker will have all of
10 the same type of cases, either all with special therapeutic needs or not. COA emphasizes the intensity of services that are provided to in home family cases by differentiating the caseload recommendations based upon whether intensive family preservation services are being provided or not. The caseload and workload research cited in this study consistently reference CWLA and COA caseload recommendations as the standard to measure against These referenced caseload recommendations do not consider the unique characteristics of children an d families that may demand a larger portion of a caseworkers time. These recommendations rely on consensus and at best, the type of service that is being offered rather than empirical evidence Th ese caseload recommendations do not seem to enable casework ers and their supervisors to accurately assess the caseload time commitment at the point of case assignment Further, the supervisor has little frame of reference in determining which caseworker within the unit is most capable of taking on a particular case or how much time a caseworker ought to devote to a particular case. This lack of knowledge and assessment leads to further chaos and unregulated workloads for the critical requirements of the caseworker within our child protection system. Purpose of Study This was a mixed methods research project. T he significance of the influence that each case characteristic has upon the time a caseworker spends on a specific case w as determined through the use of a multiple and a stepwise regression analysis. The job descriptions from several participating agencies for caseworkers were analyzed to identify the stated and designed job tasks. Additionally, four focus groups of caseworkers
11 and supervisors w ere conducted using qualitative methods of comparison and prio ritization of data analysis. This study use d an explanatory research model that seeks to determine the effect on caseworker time and therefore workload caused by specific characteristics of cases assigned after the child abuse investigation is complete. Th e p urpose of this study wa s to explain the relationship between child protection case characteristics and the time an assigned caseworker devotes to a case. It was hoped that with this knowledge an informed methodology to asses s the current workload of a c aseworker could be used to assure that the caseworker is able to successfully complete the tasks required for each child assigned Further, the knowledge of the amount of time spent on a case with specific characteristics will allow supervisors to assess and properly assign cases. The child welfare case assignment process is currently viewed as haphazard without such a methodology and is an impediment to caseworker retention and successful performance (GAO, 2003 & 2006; Zlotnik, 2005) Below (see Figure 1) is the model that this study is proposing for how an evidence based methodology can be used by supervisors in making case assignments. The figure shows the beginning of a suspected child abuse case through the transfer from investigation units to the fi el d units of caseworkers for ongoing services1 It is at this point of transfer and the decision by the supervisor on case assignment that this study is focused. 1 The Florida Child Welfare system separates investigations from ongoing child welfare casework. This is a transfer of case responsibility between distinct agencies as well as distinct caseworkers.
12 Report of Child Abuse received by the hotline Report accepted and Investigated Families are determined Dependent requiring ongoing services Investigation units refer case to field units Casework Supervisors review case characteristics to determine workload impact Casework Supervisor assigns the case considering caseworker s workload Caseworkers have a managed and reasonable workload Caseworker success and stability increases Children and families are safer and better served Figure 1. Knowledge and Workload Impact in Case Assignments for Child Welfare Caseworkers. Social Work Practice Theoretical Hypothesis The social work practice theoretical hypothesis considered in this study is; Specific case characteristics have significant impact on the amount of time a caseworker will spend on the particular case. With this k nowledge supervisors can make informed decisions about case assignment taking into account the amount of time a particular case will require and the current time commitment (based on the caseload already assigned) of the caseworker receiving the case. This informed decision of case assignment, and the resulting management of the workload for all caseworkers, would assist caseworkers with their attainment of child welfare outcomes and improve the retention of the caseworkers. This improved retention and per formance would lead to better outcomes for children and families served (Annie E Casey Foundation, 2003; GAO, 2003 ; GAO, 2006 )
13 To date the studies that have been completed have identified the personal and organizational factors leading to recruitment an d retention failures and the current distribution of time caseworkers spend based upon their specific position category. The current range of knowledge is missing the relationship between child and family characteristics and workload. This study will contr ibute to closing this gap in our knowledge base by examining the relationship of age, gender, race, living arrangement initial placement, prior removal mental health condition of the child, developmental disability of the child, existing domestic violenc e, existing substance abuse, decision to remove the child, type of abuse, to the time a caseworker spends on the case. Th e findings will allow a determination of the effect of these characteristics on the workload of the caseworker. The research questions to be addressed in this study are ; Research Questions 1. What specific case characteristics and job tasks do current caseworkers and supervisors identify that causes more time to be devoted to a specific case? 2. Which critical case characteristics and tasks ide ntified by the job descriptions, caseworkers, and supervisors that are not currently within the SACWSIS data set ? 3. Is there a significant difference in the independent variables and time spent per case between the lead agency study cases and the non lead ag ency cases? 4. Are there specific child protection cases characteristics that significantly influence the time caseworkers spend on a particular case?
14 The first phase of this study utilize d the focus group and job description data to measure the completeness and strength of the SACWSIS data set. The second phase then perform ed a secondary anal ysi s of the SACWSIS data set for significant case characteristics. The combination of these two phases allow ed for a robust discussion of the strength of the regression m odel and potential recommendations for future studies.
15 Chapter II. Literature Review Background The Government Accountability Office (GAO) reported in 2006 that states continue to receive over 2 million reports of child maltreatment each year with over half a million children in foster care. Upon completion of the first round of the Child and Family Service Reviews (CFSR) for each states child protection program in 2006, the U. S. Department of Health and Human Services found that no state had achie ved all of the child welfare outcome measures regarding safety, well being, and permanency of children (GAO, 2006) The literature is rich with studies that identify caseworker turnover as the most critical factor in failure to achieve identified outcomes (Bednar 2003; Drake & Yadama, 1996; Ellett, Ellis, & Westbrook 2007; GAO, 2003; Mor Barak, Nissly, & Levin 2001; Zlotnik, et. a l ., 2005). Further, the turnover of caseworkers in child protection services has been a historic problem that appears to date back to the earliest published studies on the position. The GAO report of 2003 traces the history of the issue of caseworker turnover affecting policy decisions to well over 45 years ago. In a 1960 study child protection agency directors reported that staf f turnover handicapped their efforts to provide effect ive social services (Tollen 1960). The Institute for the Advancement of Social Work Research conducted a comprehensive review of the research and literature including non published dissertations and bodies of work cover ing the span from 1974 through 2004 (Zlotnik, et. a l 2005) This study identif ied caseworker
16 turnover as a consistent critical problem to the success of child protection services throughout the three decades of work reviewed. Casew orker turnover has been a source of concern in child protective services because of its impact on the achievement of safety, permanency, and well being of children in care CWLA (2002) reported that there was a direct impact on outcome performance in child protective services due to staff turnover. Depersonalization by caseworkers has been found to be a key component of worker burnout and turnover (Maslach & Jackson 1986) and has been directly associated with failure to achieve performance expectations in child protective services (Drake & Yadama, 1996). Caseworker turnover has also been found to be a significant factor in extending the length of stay for children in foster care and for reducing the likelihood of reunification (Ryan, Garnier, Zyphur, & Zha i, 2006). Another condition resulting from caseworker turnover is the constant fluctuation in the size of caseloads for those caseworkers that remain on the job and take on the work of the vacant positions (Annie E. Casey Foundation, 2003; GAO, 2003 & 2006; Mor Barak, et. al 2001) The hiring of non social work degreed staff to fill positions, lack of perceived respect for the position of caseworker and the need to expend limited public resources on recruitment and training of new caseworkers were also found to be significant contributors to turnover and performance failure ( Ellet, 2006 ; Freund, 2005; GAO, 2003 & 2006; Lee 1979). The literature repeatedly reports p ersonal family conflicts generated by the job demands, a lack of a personal support network, and emotional exhaustion to be significant contributing factors in a caseworkers decision to leave his or her job
17 ( DePanfilis 2006; Freund 2005; Zlotnik, et. al ., 2005) Factors directly related to the demands of the job and required to complete cas ework tasks such as being on call for emergencies, having to work nights and weekends to visit families and children when they are available, and the extended hours ( well past fifty hours each week ) were found to be the underlying causes leading to caseworker turnover (Ellet, 2006; Gillespie & Cohen 1984). Other concerns that affect the tenure of caseworkers include their level of commitment to working with children and families, previous work experience, education, age, eff ectiveness, and job satisfact ion. Additionally, the research over the last 30 years repeatedly points to organizational factors related to turnover including salary, supervisory support, opportunities for advancement, coworker support, flexibility of work schedule, organizational com mitment to the mission of helping children and families, caseload size, and the organization s valuing of the employees ( CWLA 2002; DePanfilis 2006; Ellet, 2006; Freund 2005; Gillespie & Cohen, 1984; Morris 2005 ; Mor Barak, et. al ., 2001 ; Nunno, 2006, 2008; Zell 2006; Zlotnik, et. al ., 2005). Caseworker Survey Results Two comprehensive surveys of child protection caseworker perspectives on their jobs can be found in the literature. These surveys represent the opinion of the caseworker staff on issues including salaries, qualifications, training, and violence in the workplace. Both surveys speak specifically to caseload and the impact this has on retention of caseworkers and successful performance in child protection. The first survey was conducted by the American Federation of State, County, and Municipal Employees (AFSCME) in 1998. Survey responses were received from 29
18 AFSCME affiliates representing 13,380 child protection caseworkers in 10 different states. Surveys were completed by union repres entatives in consultation with caseworkers employed by the state or local government. These caseworkers identified a variety of factors influencing job performance and satisfaction. Over 70% of the respondents reported that front line workers have been vic tims of violence or threats in the line of duty. Caseworkers reported they are often required to go into neighborhoods that are known to be dangerous, and the nature of their job includes asking questions and taking steps which may be perceived as threaten ing by those they seek to help. Time spent on paper work, court, staff meetings and case meetings occupied a disproportionate amount of a caseworkers workload, further interfering with the caseworkers attempts to successfully perform their job. The wag es paid to caseworkers are not commensurate with the educational requirements (four year degree minimum) and the job duties. Most entry level salaries were found to be in the mid $20,000 range (approximately $25,000 in 2006 dollars) Workers reported the t raining they received was inadequate and lacked practical input. Over 50% of the respondents state d that they carry caseloads well over the Child Welfare League of America (CWLA) recommendation of 12 to 15 children per caseworker. Just 25% of those respon ding claimed to be at or below the CWLA recommendations for investigations and only 11% reported to be within these guidelines for out of home care caseloads. Several agencies reported caseloads exceeding 50 per caseworker which is three times larger than the recommended caseload
19 The authors indicate that the caseloads may be even worse than reported in relation to the CWLA standards, since most of the reporting affiliates measured caseloads by families, many of which had more than one child. Further, the survey found that few child welfare agencies have taken steps towards controlling the caseloads of child protection caseworkers. For example, taking the most basic step of determining the maximum number of clients that may be assigned to a caseworker had not been implemented. With the unpredictable and volatile nature of children needing protection from abuse, caseloads can and often have spiked to very dangerous levels that no one could manage. The affiliates reported the overwhelming caseloads as a prim ary cause of worker turnover. A second comprehensive study was done with state and local government employees responsible for child protection services. A collaborative survey was done between the American Public Human Services Association (APHSA) and t he CWLA in the fall of 2000. A total of 43 states and the District of Columbia (84%) completed the survey with 8 of these states using a locally administered child protection system and 35 using a state administered system. The survey results indicated tha t half of the turnover of child protection caseworkers was preventable. Preventable turnover was defined by the authors as a competent worker leaving a position for a similar paying job in the same community. In this study, the causes of turnover were ra ted by the states, and the number one issue was caseloads that were too high and too demanding. Other issues identified included too much time spent on paperwork, court, travel, and meetings. Caseworkers reported the salaries were too low for the requireme nts and work expectations as well as a
20 general feeling of not being valued by the agency. Supervision was reported to be insufficient, and there were inadequate resources for the children and the families, causing professional caseworkers to leave. The sur vey found high turnover rates but low vacancy rates leading to the conclusion that the high turnover rates had produced efficiency in recruitment and hiring but not in retention. The agency directors emphasized the success of the relationships they establi shed with local universities to assist in recruitment of staff. These collaborative relationships appeared to be yielding results and filling jobs. Child protection systems partnering with universities and leveraging the Title IV E funding has also proven to be an effective tool in recruitment (GOA, 2006; Hopkins, Mudrick, & Rudolph, 1999). The strategies implemented in an attempt to prevent or diminish the caseworker turnover can be divided into two categories. The first includes those strategies that re quire substantial increased funding such as salary increases, reduced caseloads, and educational financial support. The second category of strategies was labeled as softer strategies not requiring a significant increase in funding such as the valuing of workers, training, flex time, mentoring, and caseworker safety. Only 8% of the combined strategies were judged to be highly effective by the states implementing them. Only one state of 14 that used increased salaries as a strategy found it to be highly effective. Over 64% of the agencies reported not implementing any specific strategy to attempt to retain staff. Agencies that were successful in improving caseworker retention reported the reduction of caseload as one of the top solutions contributing to t his success. Further, agencies that reported outcomes of caseworker satisfaction surveys found that reduced caseloads and workload management were second only to having good supervision for
21 improved caseworker satisfaction. These findings support Herzberg s two factor theory of job satisfaction, motivation and enrichment (Herzberg, 1966). Th e survey results confirm that manageable caseloads and competent supervision are necessary components to staff recruitment and retention (ASPHA and CWLA, 2000) Neithe r survey report discussed case assignment nor any method of workload distribution other than when caseworkers left the caseloads had to be distributed to those remaining until a replacement was hired and trained. Also, the surveys mentioned that in the chi ld protection business there is no option for a waiting list or a denial of services due to insufficient number of caseworkers. All cases have to be accepted and worked regardless of the capacity of the caseworkers and supervisors. Caseload studies There have been some caseload studies completed. From the published literature and published reports a few studies have been found. The attempts to find a formula for determining a proper workload for caseworkers ha ve come about primarily in response to poor per formance, lawsuits addressing egregious harm or failure to perform required tasks, and attempts to get turnover under control. Following is a discussion of a number of the studies found. Summit County Ohio In 2003 the Summit County (Ohio) Children Services Board child protection caseworkers went on strike demanding a caseload cap. Two journalists (David Knox and John Higgins) from t he Akron Beacon Journal wrote a news article regarding the dispute over the caseload cap (Knox & Higgins 2003) In this article they sta ted that they received time study information regarding child protection investigation workers from the
22 Public Childrens Service Organization of Ohio. The time study components that the journalists delineate were as follows: Initial research into the familys history with the agency: 30 minutes Initial Home visit ; 3 hours Calls to police, teachers, counselors, doctors and other sources; 1 hour and 20 minutes. Completing a child risk assessment form ; 90 minutes Paperwork; 2 hours and 30 minutes Ot her duties (not defined) ; 5 hours and 48 minutes. This totaled 14 hours and 38 minutes. From this calculation the journalists concluded that a caseworker could perform 11 investigations in an average month, working 40 hours per week (Knox & Higgins, 2003). This news paper article is a noted reference for the CWLA caseload standards (CWLA Standards of Excellence, 1999) Lutheran Social Services of Washington and Idaho A permanency study was conducted between July of 1986 and February of 1988 with Lutheran S ocial Services of Washington and Idaho (Katz 1990) The study measured the impact of five permanency planning strategies from existing child welfare knowledge. These strategies were; limiting the caseload to 10 children for each caseworker, early case planning with significant participation by the family, intensive services to the parents, contracting with the parents to assure clarity of tasks and expectations, and an emphasis on parental visitation assuring weekly visits and more frequently with infants In addition to these five strategies the study incorporated three techniques in child protection work. The first technique was concurrent case planning,
23 wherein the primary effort is to reunite the family but the work necessary to move into an adoption go al is completed as well. The second technique was to place the children in foster to adopt families. T he third technique deployed was open adoptions wherein the biological family will have continued access and contact with the adopting family. This techn ique has been credited with helping birth mothers view the adoption process positively (Berry 1993) Additionally, Lutheran Social Services, as part of this study, eliminated the practice of transferring the case from a foster care worker to an adoption worker when the goal changed from reunification to adoption. A second structural change was the addition of specialized legal assistance. These cases were all active in the dependency court and to assure expedience in the court room as the cases progressed, Lutheran Social Services hired specialized legal counsel. There were a number of other factors regarding the selection of the children and families that were involved in this study, specifically all of the children were under the age of 10 and 77% we re under 3.5 years old. Only 6% of the children were reunified, with 94% having adoption as their permanency outcome. The average length of time to achieve permanency for the 30 children completing the study was 13.1 months. The study does not reference wh at the average time to permanency is for any other group of children outside of the study so it is difficult to determine if this is any improvement over other commonly used practices The author does include information regarding the length of time of prior placements, which are placements for the children before entering the study project. The cohort averaged 2 prior placements with 46% of the children having 1 or none.
24 This study is often used as support for a reduced caseload resulting in expedited per manency. The study does no analysis regarding the impact of the multitude of variables described above and does not discuss the reduction in caseload in any detail. Without any description of how the principals arrived at a caseload of 10, how this caselo ad differed from the other caseworkers, what additional tasks were expected of this group of caseworkers, or what the impact on the caseworker was of this reduced caseload it seems to be quite a reach to claim this study supports a reduced caseload as impacting permanency. New Mexico Department of Human Services In 1983 in response to a lawsuit a court ordered settlement was accepted by the New Mexico Department of Human Services requiring the development of a case weighting formula. Stein, Callaghan, McGee and Douglas ( 1990) describe the process and results of the departments efforts at meeting this requirement. Three methods of defining manageable caseloads for child protection workers were identified and considered. The first method utiliz ed the judgm ent of professional staff ( based on their experience, training and education) to determine the amount of time needed to provide the services and tasks assigned to a child protection caseworker for various case types. A second method w as to conduct detailed workload studies using coded time logs. This method detailed the full array of caseworker activities to uniform units of time so that a caseload c ould be determined based on the number of units a full time caseworker could carry. These first two methods w ere described as potentially limited in their effectiveness based on the degree to which the staff contributing their judgment or documenting their
25 time were actually successful caseworkers. Measuring unsuccessful or marginal caseworkers w ould not yield a useful weighting formula. The third method involved examining the caseloads of workers that achieved stated outcomes consistently, using their caseload size as a standard. Studies using successful caseworkers as the standard ha ve been completed in Al a me da County, California ( recommending 20 families per caseworker ) and in Oregon ( recommending 15 children per caseworker ) (Emlen, Lahti, Downs, McKay, & Downs, 1977; Stein, Gambrill, & Wiltse 1978). The department intended to establish a case weighting form ula that would be used to project the number of caseworkers needed and provide empirical support for a request for funding from the state legislature Additionally, the department hoped to establish an upper limit on the caseload an experienced and skilled caseworker could manage and successfully perform the goal oriented services. The authors argued against using a time study methodology due to the limited information known about a case when an assignment needed to be made. As the case progressed and more information was gathered the case would need to be re evaluated and potentially re assigned. For the authors, t h e time study methodology present ed too high a risk of constant re assignment of cases for the authors to support this technique A model case w ith seven activity categories, dividing cases by goal (in home, reunification, adoption, long term care) was then developed. A uniform percentage of time was assigned to each activity by goal using the judgment of staff from around the state. T he judgment of professional staff was used to identify a common set of administrative activities (meetings, training, supervision, etc.) and a time allowed for each. The
26 assumption was that cases with the same goal may need different amounts of time for each activity but no case would need more time than assigned to the model case. The department primarily utilized the first method of professional judgment in setting a weighting standard for caseload. The authors ruled out differentiating caseworker time based on wh ether the case was an in home case or an out of home placement case. In both settings the authors assumed an equal amount of time. This was justified by an expectation that the increased attention the parents would need in an in home case would be offset b y the reduced amount of time the parents would require in an out of home placement case. The authors also decided there was no time or workload difference between rural cases and urban cases. Rural cases require longer traveling distances, but urban cases have severe traffic issues to deal with and these would be offsetting factors. The state did determine that cases involved in the court system would require more time for paperwork and court appearances. The formula developed sums the percentage of time assigned to the activity categories for each case type and determines the number of points a particular case generates. The model case established a maximum of 100 points for any one case and the department determined no caseworker could have more than 2,000 points resulting in a maximum limit of 20 families or 35 children. The department administrators made adjustments to accommodate child abuse investigations that are often presented in an unpredictable pattern affected by events and media coverage. The department set the number of investigations per caseworker at 21. The department reduced the upper limit of noninvestigation case points to allow for some caseworkers to carry a mixed caseload and have some capacity to accept the next investigation without going over the total
27 points of 2,000. The department reported developing a data base t o collect and report the number of points for each case assigned and the total points for each caseworker allowing for the caseload to be monitored closely and potentia lly controlled. However, the study documents available never discuss the use or report any analysis of this information. Further, t he authors do not discuss any results or effects of this methodology. In reviewing this study describing a caseload weightin g formula there are several points worth noting. T he department and the authors did not include any discussion of highly disturbed children including those who frequently disrupt placements or run away, medically complex cases, or highly contested cases, any of which can significantly a ffect the amount of time a caseworker must spend on a given case (Ellet 2006; Mor Barak, Nissly, & Levin 2001; Nunno, 2006; Zell 2006). Further the authors do not discuss the impact of other case characteristics such as are serving as the independent variables in this study. Additionally, the authors emphasize the logic of utilizing caseworkers that were successful in determining the proper amount of time needed for specific activities, but they give no indication of the e xperience or expertise of staff serving on committees that determin e d the assigned time for activities. The authors state that the caseload standard is based on empirical data, but there is no reference to this data. Further, the model case sets the maxi mum points for any particular case at 100 but t here was no discussion of how much time those 100 points actually represent. Did it represent time per week, pay period, or month? This might have explained how the department arrived at a cap of 20 families and 35 children. There is further confusion in this study as the formula application shown indicates that caseworkers could carry more than 20 families. With this system, a caseload could have as few as 20 adoption cases or
28 as many as 50 longterm care fami lies and still represent 2,000 points. Finally, t he article does not describe the impact of the weighting formula except that it has aided in the decision to set the number of families and children caps. There is no discussion on implementation or casework er and supervisor reaction to the use of the formula. Mid West Not For Profit Agency Mills and Ivory ( 1991) describe a not for profit agencys (the agency) process of developing a workload management system. This system had the goals of distributing workload equitably, improving morale, reducing turnover, and improving outcomes through increased contact time with the child and family. The agency was motivated to undertake this task due to a rapidly increasing number of new cases and the high turnover of caseworker s causing the failure of the agency to meet contract performance expectations. Interviews with skilled and committed staff that were leaving their jobs revealed that workload was a major consideration in their decisions to leave. It was assumed th at certain case characteristics translated into work and time demands and a committee of foster care staff identif ied the specific time allotments for the various cases. The committee determined two factors that measured work and time demands : location of the child, and severity of the case. The agency categorized cases by neglect, physical abuse, and sexual abuse and then described, through case assessment, specific types of actions within these categories to determine the severity of each case based upon the level of abuse. The severity levels were identified as: slight, moderate, high, and critical (life threatening). The agency developed a second categorization by placement location, which included foster care, first 30 days of aftercare, relative care, inhome care (all of which
29 required weekly contact) aftercare beyond 30 days and residential care ( requiring bi weekly or monthly contact ) The agency then looked at the amount of caseworker time that was required for several common activities such as travel, visitation time, paperwork, and administrative time They reach ed a conclusion that a caseworker would spend 3 or more hours each week on cases rated with the critical severity level and that a reasonable caseload would need no more than 75 hours over a two week period. With this time determination the agency then decided that an average of 25 moderate level cases could be handled by a caseworker and used this as the base line for the weighting formula. The method described by Mills and Ivory has s everal strong components. C apping the number of hours that a caseworkers caseload could generate bi weekly at 75 establishes a clear limit for case load. T he authors do not discuss if this limit was a hard and fast rule or a guide. T he method to determine the severity of a case is a strength of the weighting system developed.T his formula is represented by a mathematical equation (placement location multiplied by case worker time as determined by the severity of the case) making it relatively simple to weig ht any case This weighting model also raises several questions and concerns. It is not known if the agency was responsible for all child protection casework duties and t here was no discussion of adoption or court related activities. The formula did not define requirements for seeing parents and the child (ren) separately or together. Further it did not account for evolving circumstances of the cases wh ich could result in a new placement or a change on the scale of severity. Using staff estimates to differentiate the amount of caseworker time each level of severity would require, a critical severity level was determined to
30 require four times the amount of caseworker time as that of slight severity. This was a professional estimate based on experience and consensus but t here was no time study to support these assignments. The authors state that the implementation of this formula created a perception that incoming cases were being assigned more equitably Additionally the authors reported, staff morale w as improved, the caseworkers were having more contact time with their clients, and the caseworkers were able to concentrate on service delivery and plan development. The authors did not offer any quantitative or qualitative data to support these claim s. Further, the authors did not address any change in performance or outcome measures by the age ncy. New York State Child Welfare Workload Study The New York State Child Welfare Workload Study, (Walter R. McDonald and Associates and American Humane, 2006) looked at actual time studies of caseworkers over a two week period and reported time spent on a variety of tasks by caseworkers with recommendations for 11 to 12 children for foster care caseworkers and 12 to 16 families for preventive case planning service s (in home services). This study include d time study data from 2,208 State of New York caseworkers representing 15% of the total state child welfare workforce. Data on specific tasks in each program or service area w as collected by worker and by case. Ac cording to the authors, the approach used in this study was perhaps the most rigorous in that it used actual time spent by caseworkers to estimate how much time is needed per case. Time that was not used on direct casework was also collected to estimate ho w much time is available during a year for casework. The total amount of time spent was computed over the period of 2 weeks, as well as by time spent
31 on specific tasks, such as faceto face contact with the child. Estimates were then made as to how much ti me per case is needed per month. The number of caseworkers that are needed in any given month would then be computed based on multiplying the number of cases open in a month by the average number of hours needed to provide services to that case type. This number would then be divided by the average amount of time available per caseworker for case related services. Such estimates are further informed by policy and best practice guidelines as to tasks that should be conducted to establish estimates that repre sent both what is and what should be. The authors found that case related time (time spent performing task s specifically related to child welfare services) consumed 77.8% of the caseworkers time. N oncaserelated time (leave time and non case specifi c administrative tasks) consumed 22.2% of the caseworkers time with leave time accounting for nearly half of this noncase related time Foster care services took 39.8% of all time recorded, with 29.1% of the time spent on child protective investigations, 22.2% on preventive (inhome) services, and 8.9% spent on case support services. Further, this study found of the time that was spent on caserelated work: 16.9% is in direct face to face contact with children or their parents; 7.0% involved other forms of communication with children and parents; 20.5% is spent in other case related activities, including contacts with service providers, collaterals, supervisors or managers, and peer consultation regarding cases, as well as preparing for, participating in, or waiting for various meetings; 6.5% is spent preparing for, appearing in, or waiting to appear in court;
32 30.8% is spent in various forms of documentation; 11.2% is spent traveling to and from client homes, service providers, or other meetings; 7.2% is s pent in casesupportive activities such as training or community outreach. Only 17% of the caseworkers time was in direct face to face contact with the children or family and 31% of their time was spent in various forms of documentation. The family preve ntive (in home) services had the highest percentage of case related time spent on direct face to face contact with 23.1%. D uring the two week data collection period, caseworkers reported 75.9 hours of work. The findings indicated an average of c ase related time for child protection investigations was 52.6 hours, for foster care services it was 48.8 hours, and for inhome services it was 36.5 hours. By dividing the average number of hours per case type by the average number of available hours for direct c ase related work per individual case the authors determined that a child protection caseworker could carry a caseload of 2 3 (126/5.5) investigations, foster care services could carry 16 (126/7.7) children, family prevention services (inhome) could carry 22 (126/5.7) families, and adoption caseworkers could carry 27 (126/4.6) cases. These caseloads are based on the current activities and behavior of the New York state caseworkers. The state is able to use these standards to determine if a particular casewor ker is carrying more or less cases than the current statewide practice would dictate and adjust accordingly if they chose to. The authors went a step further and
33 analyzed the performance of the state child protective services and found that the state has f ailed to meet a significant number of outcome measures and performance thresholds, implying that the current caseload and time allocation was not effective with meeting the stated outcomes. Utilizing the COA and CWLA standards as well as a determination th at increased face to face contact would yield improved performance, the authors de veloped the above stated caseload recommendations. This study defines how the New York caseworkers are currently spending their time; it does not seem to address what factors are driving this particular time allocation. Again, this study only differentiates cases by in home, foster care, investigation or adoption. There is no consideration of the circumstances or characteristics of the children and families other than whether the child has been removed from the home or remains in the home. However, t his study took a significant step in revealing the current condition of caseworkers time and the results California State Child Welfare Workload Study This same team of Walter R. McDonald and American Humane conducted a similar study in California in 2000. This study analyzed time study data from almost 16,000 child protection caseworkers documenting 1,140,667 hours of time. This accounts for an average of 72 hours of work time per caseworker for the two week sample period. The methodology was very similar to the New York study, calculating the time worked by specific position (Emergency Response Worker, Hotline/Intake/Screening Worker, Family Maintenance Worker, Family Reunificat ion Worker, and Permanent Placement Worker) performing specific tasks T ime was collected from the br oad categories of primary casework er staff time (44%) nonprimary casework staff (non primary casework staff that perform some case related activities)
34 time (45%) and administrative staff (staff that only recorded administrative units of time during the study) time (11%) The California study showed 67% of the caseworkers time was spent in case related activities and 33% in non case related activities. The caseload recommendations resulting from this study are: Standards Optimum M aximum Screening/Hotline/Intake caseworker 140 to 232 calls Emergency Response Caseworker 10 to 13 investigations Family Maintenance Caseworker 10 to 14 families F amily Reunification Caseworker 12 to 16 families Permanent Placement Caseworker 16 to 24 children T his study d id not take into consideration the special and unique characteristics of the children and the families in determining a manageable workload. T his study did find that the current caseloads were so overwhelming that the reported time did not include all assigned cases. In other words, a significant number of cases were simply not receiving any services each month. Therefore one of the methods use d in calculating the above recommended caseloads was using a factor to include appropriate level of services to all of the assigned cases each month. The problem is that there is no determination of how much time or how much service is appropriate to each case based upon the needs of the case. The team also utilized the CWLA and COA standards to identify the recommended caseloads. A summary of the findings from this workload California workload study and one conducted in Arizona by the AHA and Walter R. McDonald and Associates found that the CWLA workload standards were out of date and not relevant for current practices and service configurations (Edwards 2002). This comparison further found that workload
35 estimates are similar for each state despite organizational variances, suggesting that it is possible to compare estimates between jurisdictions. Additional studies completed on caseworker caseload and times include a Florida study of caseworker tasks and time associated with these tasks (Perry & Murphy 2008). In this study Perry and Murphy found that 98% of the tasks caseworkers were reporting were case management tasks. This study used two methods of data collection. First they used a time log completed by each caseworker and second they had trained as sociates shadow caseworkers and documenting the activities and time spent. This study did not capture the different types of tasks comprising case management but did find 38.4% of the caseworkers time was dedicated to out of home care services, 32.5% was dedicated to multitasking, 6.6% of their time was dedicated to inhome services, and 10.5% was dedicated to non case related activities. They also found that 28.7% of the caseworker time was dedicated to contact with the client. This study again measures how the caseworker is spending their time currently and documenting the activities that consume caseworker time. A recent caseload study deploying three data sources of focus groups, job shadowing, and the state (Pennsylvania) data set for the Administr ation for Children and Youth agency (Yamatani, Engel, & Spjeldnes 2009). Using the job shadowing methodology the researchers determined that caseworkers spent 3.29 hours per month per case on family visits and 3.55 hours per case per month on casework processing. This came to a total of 7.2 hours per active case per month. The researchers then determined that there were 118 work hours available per month for the average caseworker after adjusting for sick, vacation, and holiday time. Then through a simple division of average
36 time per case and available time per month the researchers found that the caseworkers could handle a caseload of 16 cases. The questions raised here is that each case will fluctuate during the life of the case and will not always requir e 7.2 hours of time each month. The assumption is that some cases will need more time in a given month and this will balance out. This is another example of the approach to caseload determination based on the activities of the caseworker and how the casewo rker spends their time. This is a valuable approach and has many studies that have utilized this methodology. This study takes a different approach in looking at the amount of time documented by a caseworker over the entire life of the case and what case characteristics impact the amount of time a caseworker spent on a specific case. Summary of Caseload Studies Both of the studies by Walter R. McDonald and Associates and American Humane are significant pieces of work in determining appropriate caseloads fo r caseworkers. However, both use the same methodology based upon current practice and current time spent by the caseworkers The current practices and time utilization have apparently resulted in failed performance and even neglect ( in California ) of a nu mber of cases as reported in the studies. This failure of performance, as well as high turnover of staff, was cited as being a critical catalyst for conducting the time studies. In addition, there is no discussion of the unique needs of the children and fa milies nor any consideration of the time required by the caseworker as determined by the characteristics of these children and families. In both studies t he authors rely upon the CWLA a nd COA standards to extrapolate their recommendations for caseload stan dards.
37 The studies completed by Walter R. McDonald and American Humane are significant efforts to capture current time spent by caseworkers in New York and California on their caseloads. These studies while very significant and informative only serve to develop an average amount of time for each case a caseworker would provide So regardless of the age, prior history, medical condition, strengths and condition of the parents, academic status, type of abuse, availability of services, etc. the in home caseworker would be expected to perform all tasks within the same amount of hours for each case or at least their assigned caseload ought to average out. It seems that the next iteration of study needs to be a closer look at the case characteristics and the im pact of specific characteristics on caseworker time. All of the studies discussed above provide significant insight into the difficulty and complexity of determining the caseload/workload for child protection caseworkers. These studies considered the job classification of the caseworkers and some very broad case classifications. They look to average amounts of time spent by caseworkers on separate job functions and draw conclusions that estimate improved performance and retention. Ho wever, these studies do not consider the effect on a caseworkers workload by specific known case (child and family) characteristics, nor do they follow through with any evidence that the utilization of the formulas or knowledge developed actually improved retention of caseworke rs or outcomes for children and their families. In a review of 50 selected case records of children and youth receiving Intensive Case Management (ICM) for severe behavioral health issues, Gail B. Werrbach found that ICM caseworkers spent more time with ch ildren and youth that were hospitalized, with higher rates of child functioning difficulties and who were male (Werrbach, 2002).
38 These findings give evidence that there are correlations between case characteristics and time spent by caseworkers. Th e limitations of the current study of child protection casework clearly support the need for further study, specifically to determine the effect of case characteristics upon the workload of the child protection caseworker. Child Welfare League of America Standar ds of Excellence The literature review for this study revealed that most caseload or workload studies referenced, compared, or utilized the Child Welfare League of America Standards of Excellence and/or the Council on Accreditation caseload standards. Ther efore, it is critical to review the manuals and standards of these national leaders in setting caseload expectations for child protection work in America. Including the standards of CWLA and COA provide for a comprehensive review of the current knowledge r egarding caseload/workload in child protection. The Child Welfare League of America (CWLA) publishes Standards of Excellence for Services for Abused or Neglected Children and their Families (originally published in 1999, revised and reprinted) and Standards of Excellence for Family Foster Care Services (originally published in 1995, revised and reprinted). These manuals are a comprehensive set of standards for the practice of child protection work. Included in these standards are specific recommendations for caseloads of child protection case managers. These standards are generally accepted and quoted throughout child protection and are currently the only measure this industry has for best practice. The standards for Family Foster Care recommend a caselo ad between 12 and 15 children, depending upon the level of service required to meet the assessed needs of the child. The CWLA provides a list of several factors that should be considered in
39 determining the appropriate caseload size, including the comple xity and needs of the child and family and the specific functions assigned to the caseworker (a full list can be found in Appendix A) The standards for Abused or Neglected Children and their Families recommend that workload should be based upon the tasks and activities expected and allow caseworkers to complete the required tasks and activities (a full list can be found in Appendix A) In this manual CWLA recommends a workload analysis to determine the appropriate workload standards, but without this ana lysis it is recommended that caseworkers doing initial assessments should involve no more than 12 active reports each month. This is one of the motivations for completing this study using the federally mandated SACWSIS data system. Assuming approximately 21 work days in the average month this would give an investigator 14 hours of work time to investigate an active case each month. CWLA does not offer any formula for making this recommendation in fact the Guidelines for Computing Caseload Standards (see App endix B) states that there is no universally accepted formula for computing caseloads (www.cwla.com) This manual goes on to recommend for caseworkers serving intact families under the supervision of the child protection agency have no more than 17 fam ilies on their caseload and receive no more than 1 new case for every 6 open cases. It is not clear if this means the assignment of cases at this ratio each month or some other time period. However, this ratio of 1 new case for every 6 opened would seem t o assign more new cases to the caseworker with the largest number of cases. A combined caseload of initial assessments and intact f amilies is also addressed with a caseload recommendation of no more than 10 intact families and 4 initial assessments being t he standard. CWLA lists
40 several factors in to consider in developing workload standards including, t he specific assigned functions and time required for each task (e.g. intake, assessment/investigation, placement services, court activities, community development, provision of services) and t he competencies needed for each social work function (knowledge, skills, experience) These standards and the Guidelines for Computing Caseload Standards (CWLA) represent the industry best practice to provide the child protection industry a measure of manageable caseloads. Council on Accreditation Standards The Council on Accreditation (COA) was founded by CWLA in 1977. This national accreditation body has a focus on child protection services and has established a compre hensive set of standards for these services. COA recommends caseloads not to exceed 18 children or 8 children with special therapeutic needs for foster care, kinship care caseworkers, 12 to 25 families for adoption caseworkers, 12 to 18 families for family preservation services and 2 to 6 families providing intensive family preservation and stabilization services. COA further recommends no more than 15 active (new each month) investigations for Child Protection Investigation caseworkers with no more than 15 to 30 open cases at any time COA supports these standards with a list of literature references but does not offer any specific methodology for how these specific numbers were derived. Summary of Literature Review Throughout the literature it is clea r that the caseload size and the amount of work required by caseworkers has direct impact on the achievement of safety, permanency, and well being for abused and neglected children. Further affected are the
41 caseworkers with high rates of burnout and turnover. There have been several studies focused on determining what the current workload and time utilization is for caseworkers, but very little focused on what drives the time a caseworker spends on a case. Excellent work is being done to measure how casewor kers spend their time and efforts have been made to determine how they should spend their time W hat this study intends to discover is what factors are driving caseworkers time. Caseload management formulas have only been attempted in a few small pockets o f our country with most recommendations and states relying on the CWLA and COA standards. In reviewing these standards and the references cite d there remains a void in determining what the caseworker time drivers are in the child welfare industry.
42 Chapte r III Methods This study endeavor ed to answer the question of the impact of identifiable case characteristics on the amount of time a caseworker spends on a case. The dependent variable was the time in hours recorded by a caseworker and the independent variables are a list of case characteristics including; the age, race, gender, the type of maltreatment, was the child removed, does the child have a diagnosed mental illness or behavior problem, does the child have a disability, placement type, living arra ngement at the time of investigation, family domestic violence, family substance abuse and did this child have a prior removal in a previous case of child abuse This study was limited to examining data within the State of Florida utilizing cases that wer e open for services in calendar year 2003 and closed prior to January 1, 2007. Florida is organized very differently than most states in the structure of the child protection system as explained below. To grasp the methods structure of this study a brief explanation of Floridas child welfare system is required. The Florida Department of Children and Families under the requirements of Florida statute (f. s. 409.1671) outsources, utilizing a capitated funding model, all child protection services other than legal services and protective investigations. The capitated model is a global budget model, simply meaning that there is a fixed amount of funding for a community to spend on all child protection services for any and all children that qualify for such se rvices. These capitated service contracts are with locally based nonprofit agencies called Lead Agencies. The Lead Agency is responsible
43 for the coordination, facilitation and management of the child welfare system of care for a defined community (a count y or group of counties). These Lead Agencies then subcontract most of the direct service work to other community agencies or specialty agencies. The largest of these subcontracts are to agencies that perform all of the casework services within the child w elfare system of care. For the purposes of this study these agencies are referred to as sub contracting casework service organizations. This system structure is referred to as Community Based Care (Vargo, Armstrong, Jordon, Kershaw, Pedraza, Romney, Yampol skaya 2006). The Florida Department of Children and Families further organizes its services into six regions covering all counties and judicial circuits within the state. This study references Lead Agencies, Regions, and sub contracting caseworker service organizations as parts of the operational components involved with the structure of the study and data collection. T his study utilize d three sources of data; focus groups with child welfare caseworkers and supervisors, caseworker job descriptions, and the State Automated Child Welfare Services Information System (SACWSIS) data set, to answer the research questions stated above in Chapter 1. These data sources were chosen to provide the most complete information regarding the case characteristics and casewo rker job tasks that determine the amount of time required to complete the case The first source of data was a set of focus groups with current caseworkers and supervisors that have a minimum of two years experience. The focus groups identified specific case characteristics and tasks that they found caused more time to be spent on any particular case. Additionally these focus groups provided experiential information as to what case characteristics are available at the time the case is assigned from the investigator. The second data source
44 was a collection of job descriptions and an analysis of the various tasks expected of the caseworkers by their employers (the sub contracting casework service organizations). The third source was the SACWSIS data set. Thi s data set was first identified as HomeSafenet (HSn) at its inception in 2001. While known as HSn, the Community Based Care caseworkers, working for the sub contracting caseworker service organizations, were required to enter all of their activities with b eginning and end time, therefore offering the ability to determine the amount of time a caseworker spent on a specific case while performing specific tasks. In late 2006 the HSn data base was updated and converted into a new system called Florida Safe Fam ily Network (FSFN). Throughout this study the author uses the term SACWSIS data set to refer to both of these data bases, HSn and FSFN. Additionally, i t is noted that throughout the literature the terms workload and caseload are often used to describe similar activities. Throughout this study the author uses the term caseload as a comprehensive term to include any activities commonly associated with the term workload or caseload These three data sources provided the expectation and design of the casewor kers tasks (job descriptions), the practical and real life tasks performed and the identity of key case characteristics effecting time spent by the caseworkers (focus groups), and the tasks that are documented and have actual caseworker time documentation (SACWSIS data set). These three sources allowed for a comparative analysis defining the completeness of the SACWSIS data set regarding the tasks and case characteristics being studied determining the impact upon the time caseworkers spent on a particular c ase as well as any key tasks or case characteristics not found in the SACWSIS data set. The illumination of key tasks or case characteristics missing reveals the limitations of the
45 regression model and leads to the recommendation for further study. Convers ely where these data sources match tasks and case characteristics the model is strengthened. Utilizing multiple and step wise regression models the impact and significance of specific case characteristics (independent variables) on caseworker time (depend ent variable) was defined. This study utilized data specific to the State of Florida with a discussion of the generalizability of the findings. The study consist ed of three analytic phases. First, data from the focus groups and job descriptions were compa red with the actual case characteristics and tasks that were recorded in the SACWSIS data set through a matrix identifying and matching the common data and identifying any unique data. The second analytical phase compared the independent variables, the dependent variable, and general data from the lead agency counties (12 of 67) with the nonlead agency counties (55 of 67) utilizing an independent samples t test and a Chi square test. This analysis allowed a determination regarding any significant diffe rences in the two groups of counties making an argument for generalizability. The t test was selected as a test of means between two independent groups and was used for the age characteristic as the only independent variable that is captured in an interval level of measurement. Each independent variable represents a case characteristic. Each independent variable (except for age) was a categorical variable and indicates the presence of the characteristic (substance abuse, removal, mental health condition, et c.) or subcharacteristic (race white, black, Hispanic; living arrangement with one parent, with relative, with nonrelative, with two parents, etc.). The independent variables with subcharacteristics, race, type of maltreatment, living arrangement, and placement type were re coded to create an
46 independent variable for each sub characteristic. Therefore, each independent variable (except for age) has two values, 0 indicating the characteristic is not present and 1 indicating the characteristic is present. The Chi square test was chosen to test the significance between two categorical variables between the two groups of cases. Chi s quare is used to determine whether the proportion of cases with the same characteristic is the same for one group as for another (Montcalm and Royse 2002). If the proportions are quite similar chi square will be small and insignificant and when they are not similar chi square will be larger and significant. The third analytical phase utilized three analytical tests. First, a bivar iate correlation analysis was used to assess the strength of the bivariate relationships between each independent variable and the dependent variable determining a Pearsons R for measurement of the relationship Second, a multiple regression analysis was completed and determined the amount of variation in the dependent variable explained by the group of independent variables. The regression analysis was broken down into two separate equations. This was done due to the possible interaction between the inde pendent variables of placement and removal. The independent variable removal is technically a subcategory of the independent variable placement. The independent variable removal, as explained in the next chapter, captures those cases in which the child was removed from their family within the first 30 days following the case begin date. Each of these cases will also have a placement as a result of that removal. However not all cases that have a recorded placement will have had a removal within the first 30 days of the case. Therefore, a n all variables model with the independent variable removal included and
47 then a placement model with the independent variable removal excluded make up the two models utilized. Finally, a step wise regression analysis was conducted, with both regression equations and determined those independent variables that account for the variance in the dependent variable and eliminating any independent variable that duplicates the same explanation of variance as another independent v ariable. In Florida, as well as many other states, all cases of suspected child abuse are first investigated by specialty units that only perform the investigative duties and make a determination of whether child abuse has occurred as defined by Florid a statute. All cases in Florida wherein a determination of abuse has occurred are petitioned for court supervision and transferred to an ongoing field caseworker working for the lead agency or for a subcontracting casework service organization. It is at this point of transfer that this study was focused. Cases that were opened and investigated but abuse was not determined to have occurred and the case was not opened for services were not included in the study. This study identified those case characterist ics most likely to be known at this point of transfer from investigations to the field caseworker that significantly impact the amount of time the assigned caseworker will likely need to devote. Caseworker and Supervisor Focus Groups The Nominal Group Technique (NGT) was originally developed by Delbecq and VandeVen (1963). The focus groups were conducted utilizing this technique as described by Buddenbaum and Novak (2001). This technique utilizes a method of idea generation wherein each participant dev elops their personal list of responses to the questions (i.e. which case characteristics impact caseworker time). The second step is idea sharing
48 wherein all ideas generated are placed on a group list and then there is group comment and evaluation wherein each member has an opportunity to comment about each idea. Finally, there is the conclusion where the lists of ideas are prioritized by the participants. This method generates a complete list and a prioritization by the focus group participants of the case characteristics and the caseworker tasks that determine the amount of time a caseworker spends on a particular case. Drennan, Walters Lenihan Cohen Myerson & Iliffe, (2007) utilized the nominal group technique to determine the unmet needs of older people utilizing primary medical care professionals. This study of caseworker time used homogenous groups of caseworkers and supervisors in the focus groups to determine the perceptions of active professionals regarding the case characteristics and caseworker tasks being studied. Focus group participants were selected using criteria to maximize the knowledge and experience base of the participants. To assure that the participan ts had sufficient experience to respond thoroughly to the questions, a threshold of two continuous years of casework experience was required. The supervisors also were required to have two years of continuous experience as a supervisor. Further, participants were selected from a pool of caseworkers/supervisors meeting this criteria based upon their willingness to participate. Each supervisor participating had experience being a caseworker previously. Caseworkers and Supervisors represented various agencies within each focus group to avoid the potential of peer influences and agency responses. Participating caseworkers identified themselves as being successful in their positions with no other criteria used to determine if they were successful. Superviso rs participating had case assignment responsibility.
49 This methodology was included in the study to; 1.) validate the caseworker tasks identified by the job descriptions and those included in the SACWSIS data set, 2.) identify what case characteristics ca seworkers and supervisors find are available at the point of transfer from the investigating units, and 3.) to compare and contrast the impressions and opinions of current child welfare caseworkers and supervisors regarding what they experience and interpr et as the case characteristics driving their time on a particular case with the SACWSIS data set findings. The focus groups identified characteristics available at the time of case assignment that increase or decrease the amount of time the caseworker ne ed s to devote to the case. It is important to differentiate what characteristics are likely to be known at the time of case assignment from characteristics that may develop or become known after the case is assigned as this study seeks to develop informat ion pertinent to the initial assignment of the case by the supervisor to a field caseworker. Using the NGT participants in each focus group were asked to sort the case characteristics into three groups of case characteristics that impact caseworker time. First was a complete list of each case characteristic any member of the focus groups named. Second was a list of each case characteristic any member of the focus group named that would be known at the time of case assignment. The third group consisted of each characteristic that any member of the focus group identified that is not likely to be known at the time of case assignment. Each group then prioritized each case characteristic that any member of the focus group identified as increasing the time they would expect to spend on the case.
50 This same three step process was then used to have each focus group identify and prioritize specific tasks caseworkers perform that consume their time. Assumptions and limitations of using focus groups have to do w ith the selection of participants as well as the interaction between participants during the group process. Asking for volunteer participants with two continuous years of experience and who describe themselves as successful may render the input ungenerali zable to the entire caseworker population. However, these criteria created a group of participants that were well informed and experienced in the complexity and wide range of tasks and case characteristics being discussed. Specific questions for the group can be found in Appendix C. Four focus groups were conducted, consisting of two groups of caseworkers and two groups of supervisors. A total of 15 caseworkers and nine supervisors participated in the groups held on June 5, 2009 at the Hillsborough Kids, Inc. (lead agency) West Care Center in Tampa Florida and on July 15, 2009 at the Childnet, Inc. (lead agency) central offices in Ft. Lauderdale Florida. Caseworkers represented four separate sub contracting casework service agencies while the supervisors r epresented three separate agencies. There was representation from sub contracting casework service agencies that were specific to one community, agencies that had multiple contracts in various communities, and an agency wherein the Lead Agency employed the casework staff directly. This array of caseworkers and supervisors represented each of the organizational models of delivering child welfare casework in Florida allowing for any variation in tasks and focus to be addressed. These variations were expected to be slight due to the administrative rules regulating casework practice by the State of Florida.
51 Each c aseworker s ervices organization and Lead Agency serving in Suncoast Region and the Southeast Region were contacted by e mail and asked to identify caseworker and supervisor volunteers to participate in the focus groups The individual caseworkers and supervisors that volunteered to participate were then directly contacted with an explanation of the study, an explanation of the focus group process, a copy of the informed consent, and with the date, time and location of the focus group. Th e Suncoast and Southeast regions w ere selected to incorporate the variation of the agencies that perform casework services. Each participant completed and signed an i nformed consent. See Appendix D for the informed consent form. Caseworker Job Descriptions Caseworker job descriptions were requested from each of the caseworker service organizations within the selected regions. The request was through e mail correspondence with the CEO of each sub contracting casework services organization and then through whomever the CEO referred to for handling the request. The request was specifically for the job descriptions of the child welfare caseworker position within their organ ization. Eight agencies of the 17 contacted responded with their job descriptions. Those agencies not responding were contacted on at least three separate occasions. The eight responding agencies do represent a crosssection of the caseworker service organizations state wide. The tasks on the job descriptions were then placed into a table identifying all of the job tasks within the collection of job descriptions including the most common job tasks to those that are unique to one or two agencies. The questions addressed through this methodology include;
52 (a) What are the most common job tasks found among the caseworker service organizations caseworker job descriptions? (b) What common job tasks found on the job descriptions are not collected in the SACWSIS dat a set? (c) What job tasks do caseworkers and supervisors identify that are not captured by the job descriptions? This data source was selected to determine what tasks and requirements are expected of the caseworker. In measuring the caseload of a caseworker it is critically important to understand what tasks are expected. In the studies described in the literature review above, the tasks of caseworkers were varied in attempts to improve outcomes for children as well as trying to determine a caseload that w as manageable. This data source was also selected as a validation method for the data variables in the SACWSIS data set. The recorded data in SACWSIS may only represent a portion of the caseworker tasks and this must be considered as critical factor in the evaluation of the caseworker time recorded in SACWSIS for any particular case in this study. Each organization providing child protection services within the State of Florida are required to have specific job descriptions defining and delineating the tasks and responsibilities with each position. These job descriptions are developed and owned by each individual organization. However, the State of Florida requires that the job descriptions for caseworkers, from all agencies providing caseworker services, are reviewed by the Florida Department of Children and Families to assure that the statutory and regulatory (both state and federal) requirements are included and addressed properly (Section B paragraph 2 of the standard Attachment I to the Community Based Care contract between the Department of Children
53 and Families and a lead agency). This process of state review provides a level of assurance that each job description minimally contains the core requirements of the child protection casework duties within F lorida. The caseworker job descriptions for this study were defined as noninvestigatory positions that receive a case after the investigation is complete and a determination for ongoing work has been made. In Florida, the child protection system operates under a Community Based Care model. This model has all services after investigations contracted out to community agencies. According to the Florida Office of Program Policy Analysis and Government Accountability (Dufoe, Carroll, Wipple, Vickers, Vanlandignham, 2006), the 21 Lead Agencies in the state held 69 contracts with caseworker services agencies in 2006. These are not all unique agencies as there were a number of agencies that provided caseworker services in multiple districts. There are four Lead Age nc ie s directly providing caseworker services with the other 17 Lead Agencies issuing the 69 subcontracts for caseworker services to other agencies. The sample of job descriptions are from unique agencies within the two identified regions and include six ca sework agencies serving multiple areas, seven nonprofit agencies, one for profit agency, two agencies that serve only one identified community, three agencies operating in rural areas, and six agencies that operate in urban areas. Some the caseworker serv ice agencies responding operate in both rural and urban areas. SACWSIS Data Set The SACWSIS data set is known currently as the Florida Safe Families Network (FSFN). This data source was chosen for the robust information it contains and because it is Flor idas version of the federally required State Automated Child Welfare Services
54 Information System (SACWSIS). The federal government pays a significant portion of the cost for these systems, outlines numerous requirements for the system, and is in the proce ss of certifying those systems that have met all of the criteria. Florida is undergoing the development and incorporation of additional modules to meet the national standards for completion. However, the current system as well as its predecessor known as H omeSafenet is accepted by the DHHS as the sole source of information from Florida regarding child protection services and federal and state law compliance. Therefore the expectation is that all child protection cases, within the State of Florida, are thoro ughly documented within this system. However, the SACWSIS data set for this study does include various pockets of missing data. There are two variables with missing data that are of concern, the living arrangement (32.5%) and the type of maltreatment (28.9%). This is attributed to the failure of the caseworker to document this information in the SACWSIS data set and is discussed in more detail below. There are thousands of cases recorded in this system from the year 2002 to the present time. The state also transitioned the legacy data they had in the previous state data base (Client Information System, CIS) they had been using. This comprehensive data set includes the most expansive information available regarding how caseworkers spend their time. The SACWS IS data set, as defined in Florida Administrative Code 65C 30.001, is the official record for the child protection system in Florida. State administrative code 65C 30.001 requires thorough documentation of each case entered into the SACWSIS system. The Dep artments contract with each of the CBC lead agencies as well as the lead agency contracts with each sub contracting casework services agency includes specific language requiring documentation into the SACWSIS system of all activities within 24 to 72 hours
55 of the activity. From early 2002 when the first version of the SACWSIS was implemented statewide through the end of 2006 every caseworker had to directly enter specific information regarding all activities into the system on a routine basis including case notes describing every task completed with both a code of the type of activity and the time the activity took to complete by requiring a start and end time. The requirement to record time was more significant for sub contracting caseworker services agenci es due to the requirement to report caseworker time in order to earn Federal Title VIE funds. The state had an existing agreement with the federal agency to allow for random moment sampling to capture this information, however the lead agency outsourcing model was very new and unique so the federal agency would not agree to the same terms and required complete documentation by each caseworker. One noteworthy security feature is the inability to change any documentation that has been entered and saved so a history of all activities, placements, caseworker notes as well as several other types of information are kept as well as a progression of when specific tasks were recorded, no one can change any data component once it is entered and saved. In late 2006 F lorida received a federal Title IV E waiver that removed the eligibility requirements from this federal funding source and discontinued the need for caseworkers time to be meticulously documented. Therefore the cases chosen for this particular study are a ll completed and closed prior to January 1, 2007. A requirement of the quality assurance reviews the State of Florida has conducted each year throughout the entire child protection system includes a validation of the data within SACWSIS data set. These rev iews are called the Child Welfare Integrated Quality Assurance reviews (CWIQA). The domain for the Data Validation measure in the
56 CWIQA is defined as; This domain addresses the timeliness, accuracy, and reliability of case management information entered i n the statewide automated child welfares services information system. Accurate, detailed documentation is one of the most critical components of strong case management as detailed in Chapter 39, F. S. and is consistent with the Adoption Safe Family Act per formance standards. In these quality reviews completed by the state the CBC lead agencies responsible for the cases selected for this study scored over 90% validation on this measure. Permission to access the SACWSIS data set was provided by the Florida Department of Children and Families (DCF). The DCF is the state agency that has responsibility and authority to manage the Florida child welfare system. The DCF currently provides an extract of the SACWSIS data to the Florida Mental Health Institute (FMHI ) within the Department of Child and Family Studies at the University of South Florida (USF) for the purposes of statewide evaluation and analysis of the performance of the Florida child welfare system. FMHI agreed to add the necessary data elements for th is study to the extract they currently receive from DCF. Further, the USF Department of Child and Family Studies agreed to incorporate the requirements of this study and necessary SACWSIS data set access for the principal investigator of this study into th e existing Privacy and Security Agreement between DCF and FMHI. A request for permission for the principal investigator to work with the data set at FMHI was approved by DCF along with the amended Privacy and Security Agreement. The data elements necessary to add to the current data set FMHI receives included, caseworker time data that was recorded and living arrangement.
57 The secondary analysis of this data set permitted the identification of the specific child protection case characteristics that significantly influence the time caseworkers spend on a specific case. As described above the SACWSIS data set will also provide a comparable list of tasks and case characteristics as a component of the analysis of the job description information and the focus group data. SACWSIS Data Collection Procedures The Department of Children and Families sent a complete file of all cases opened from January 1, 2003 through December 31, 2006. From this file the Florida Mental Health Institute (FMHI ) staff removed all iden tification information so that no child or family could be identified. Each case was assigned a unique identification number. Then the FMHI staff selected the cases with an initial open date in the calendar year 2003 and with a closing date prior to Januar y 1, 2007. The open date is listed within the SACWSIS data set as the begin date. This SACWSIS data subset was then uploaded into SPSS Statistics software version 18 by the FMHI staff, which is the common practice for FMHI staff to analyze data from the state SACWSIS system. This SPSS file was then made available to the principal investigator. The methods the PI utilized to create the dependent and independent variables and isolate the cases for this study are described below. This study utilize d all co mpleted cases that had a begin date within the calendar year 2003 (January 1, 2003 through December 31, 2003) and an end date prior to January 1, 2007. These are completed cases as services have finished and all of the time spent on the case has been compl eted. Each case reflects an individual child as this is how Florida trac ks t heir caseload and case assignment. Therefore this study is measuring time caseworkers spend working on an individual childs case.
58 The SACWSIS data set was then desegregated betw een those cases opened for services in 2003 to C ommunity B ased C are (CBC) Lead Agencies (representing 12 counties and about 23% of the total cases opened in 2003 and closed by January 1, 2007) and those opened for services in 2003 to the DCF (representing 55 counties and about 77% of the total cases). This distinction allowed for the isolation of the CBC lead agency cases which is the population selected for this study. The Community Based Care Lead Agencies that were fully operational prior to January 1, 2 003 providing comprehensive child welfare services and receiving all new cases from the investigation units were in the following 12 counti es representing five lead agencies (Evaluation of the Florida Department of Children and Families Community Based Car e Initiative 2003). County Fully Operational Date Sarasota June 1997 Manatee February 2000 Pinellas February 2001 Pasco April 2001 DeSoto November 2001 Flagler April 2002 Volusia April 2002 Santa Rosa November 2002 Okal oosa November 2002 Walton November 2002 Escambia November 2002 Hillsborough November 2002 For the purposes of this study, the start date of a case was operationalized as the start date indicated within the SACWSIS data set and the end date was the end date indicated within the SACWSIS data set. This limits the universe of cases considered completed cases only, allowing for the entirety of time spent on a case to be considered. With the data set being contained to 2003 through 2006 all of the cases completed within that period will be four years or less in duration. This limitation of time is due to the
59 SACWSIS data system being fully deployed in 2002 and the requirement of Community Based Care Lead Agencies to have all caseworker time document ed into the system being lifted in the third quarter of 2006 as a result of the approval of Social Security Act Title IV E waiver being granted to Florida. It is important to note that the DCF continued to provide child welfare services in other counties a round the state until the last CBC Lead Agency model was fully operational in 2004. Each of the remaining 17 lead agencies and the counties within their project scope became fully operational at various times through the end of 2004 when the final lead age ncy became fully operational. Table 1 below outlines the number of cases and is grouped by those CBC lead agency cases selected for this study and the non CBC cases. Table 1 Cases Opened and Closed Within the Study Time Frame CBC Lead Agency Cases Non CBC Lead Agency Cases Total Cases Cases Opened for Services in 2003 9,492 27,328 36,820 Cases Closed by 12/31/2006 8,895 26,852 35,747 Percentage of Cases Closed 94% 98% 97% It is important to note that 104,037 cases were opened in 2003 in the state of Florida for investigation of child abuse of these 36,820 (35%) were determined that abuse did occur and were opened for services. According to the U. S. Department of Health and Human Services Administration for Children and Youth Bureau Maltreatment Report 2007 (2009) the national average of substantiated abuse reports to investigations completed is 25%. Florida having a 35% substantiation rate is above average but not out of the general range that states fall within. The data provided through the SACWSI S data set meeting the above criteria includes 36,820 cases that were opened for services
60 statewide in calendar year 2003. Of these 35,747 had closing dates in the SACWSIS data set prior to January 1, 2007 (97%). Within the selected counties with active Co mmunity Based Care Lead Agencies there were 9,492 cases opened in 2003 (26% of the state total cases opened in 2003). Of these 8,895 cases (25% of the state total of cases closed by January 1, 2007) had closing dates within the SACWSIS data set prior to Ja nuary 1, 2007, representing 94% of the total cases opened in 2003 in the selected counties. The cases opened to the DCF services in 2003 were 27,328 representing 74% of the total. The DCF cases closed by 2007 were 26,852 (75% of the total number of cases closed by January 1, 2007) representing 98% of the cases opened in the DCF counties. By definition within the SACWSIS data set the begin date is the date that the investigation started and therefore some cases have been closed without referral to services as an invalid case. To assure that these cases are eliminated to the greatest extent possible the CBC lead agency data set of cases was filtered removing cases with a length of stay of less than 60 days. This date range was chosen to mirror the required ti meline for investigations to be completed and closed within 60 days in accordance with Florida Administrative Code 65C 10.003. This reduced the number cases within the CBC lead agency data set by 8% from 8,895 to 8,183. Case Characteristics, the Independe nt Variables Specific case characteristics are identified as the independent variables in this study with the dependent variable being caseworker time. These variables are limited to the information that is routinely collected and documented within the SAC WSIS data set. Further, the variables studied were determined by the focus groups to be available or potentially available at the time of the case transfer from the investigator to the ongoing
61 caseworker allowing this study to be valuable in aiding in the rational assignment of workload. The case characteristic variables have been collected as categorical or numeric values. The independent variables studied to determine if there is a significant effect on the amount of time a caseworker spen ds on a case ar e found in listed below. The dependent variable was constructed by adding the documented time of each casework task associated with a specific case. These tasks occurred throughout the life time of the case (from the defined start date to the defined end d ate) and represent the documented time by every caseworker assigned to the specific case over the life of the case. The final set of independent variables was selected based on three factors; the focus group results, the literature support, and the existe nce of the data within the SACWSIS data set. The focus groups did identify two additional variables that were present in the SACWSIS data set, the mental health status of the child and if the child was identified with a developmental disability. These vari ables are supported in the literature as significant issues within child welfare services. Bruhn 2003 found that children with disabilities were overrepresented in the child welfare systems as being more likely to enter the child welfare system than children without disabilities and staying longer in the system. Authors dosReis, Zito, Safer, and Soeken (2001) found that the prevalence of mental health issues among children in the child welfare system was nearly 60%, far above the mental health issues presen t among the general population. As discussed below the variable representing a mental health condition characteristic appeared to be so significantly under reported that it was determined to be invalid and was not included in the final analysis. The develo pmental disability variable while found to most likely be under reported was not as far off from the national averages and
62 remained in the final analysis. Additionally, the focus groups did not identify the independent variables of race or gender. Race is strongly supported by the literature (Casey Family Programs, 2007, GAO 2008) and was retained in the model. However there were no strong indications within the literature that gender was a significant factor regarding length of stay or time in the child we lfare system. The independent variable gender remained in the model to account for general demographics and to determine if there was a potential for further research on this issue. Below there is a more robust discussion of these variables and the focus g roup comparisons with the initial list of independent variables The independent variables included in this study are: a. Age of child (017) b. Gender c. Mental Health Condition, identified as having a clinical diagnosis or emotional disturbance. d. Removal, child was removed from their home within 30 days of case start date. e. Race/Ethnicity, identified as White, African American, Hispanic, or other. f. Developmental Disability Condition, identified as mentally retarded (SACWSIS term) or physically disabled. g. Living arr angement (family configuration); one parent, non relative, mother and father, relative, other. h. Maltreatment, Type of abuse was identified as physical abuse, sexual abuse, medical neglect, supervision neglect, or general neglect i. Substance abuse, Reason for services Substance Abuse was selected. j. Domestic Violence, Reason for services Domestic Violence was selected. k. Placement type, first placement within 30 days of case start, no placement type identified meant the child remained in their home, placement types included; relative, non relative, shelter, foster home, therapeutic foster home, group home, institution, Department of Juvenile Justice, Developmental Disabilities, runaway, other. l. Prior Removal, a documented removal prior to the case begin date.
63 Ana lysis and description of the Variables The following section describes each independent variable, how it was calculated, some degree of literature support for the variable, and any concerns with the variable. Utilizing the Administration for Children and Families Child Maltreatment Report for 2007, published in 2009 the case characteristics included as independent variables can be compared to the national averages of those characteristics for approximately the same time period. The independent variable of age was a simple calculation of how old the child was on the begin date of the case, using the childs date of birth and the begin date of the case. According to the Administration for Children and Families (2009) Child Maltreatment 2007 report, the nation al data for 2007 shows that 12% of victims were under one year of age, 31.9% of victims were ages 3 and younger, and 55.7% of child victims were 7 years and younger. The age distribution within this study is relatively consistent with the national age dist ribution of child abuse victims although slightly younger as noted in the higher percentages of children within the younger age brackets, 16.1% under the age of one, 36.7% under the age of 3, and 60.4% under the age of 7 years. This is discussed further in the next chapter. The independent variable of gender is constructed from specific gender fields within the SACWSIS data set. This variable has an equal split between males and females in the selected cases. This is consistent with the state wide data for cases with a start date in calendar year 2003 and with an end date prior to January 1, 2007. According to the Administration for Children and Families 2007 publication of Child Maltreatment,
64 48.2% of all child victims are male and 51.5% are female. This s eems to be very consistent with the selected cases gender distribution. The independent variable of mental health condition was constructed by identifying cases that were documented as having a specific clinical diagnosis within the SACWSIS data field cli nical disability description or an identified emotional disturbance or both. Therefore any case with either condition documented or with both was included in the variable. The independent variable of mental health condition was only documented in 4.2% of the cases selected for this study. This prevalence in a foster care population is well below the 57% prevalence level found by dosReis, et. al (2001). Also, Clausen, Landsverk, Ganger, Chadwick, & Litrownik (1998) found that the prevalence was consistent ly high in the foster care population and that behavioral problems were found to be two and one half times higher than the general population. During the time frame of this study mental health information was not reported by the Florida Department of Children and Families and was not accounted for in any of the quality assurance reviews or contract requirements reviewed for this study. Additionally, according to the Administration for Children and Families (ACF) 2007 Child Maltreatment Report (2009), Florida reported 1.3% of child victims with emotional disturbance and/or behavior problems for the year 2007. The ACF report did not have a category for clinical diagnosis or for mental health condition and the Florida SACWSIS system in 2003 did not have a cate gory for behavior problems. There may be differences in how this information is being reported that would account for the prevalence differences. Nationally, the 2007 Child Maltreatment Report indicates 5% of the child victims have emotional disturbance o r behavior problems. Florida has recently began
65 collecting data on children in the foster care system whom are prescribed psychotropic drugs and a recent report by the Florida Department of Children and Families (2010) indicates that 14% of the children ar e on psychotropic medications. There are many children with a mental health diagnosis and receiving mental health services who are not on psychotropic medications. With these many factors indicating that the documentation of the mental health condition for the study cases is significantly under reported this variable will not be included in the regression equations. The independent variable removal was measured by comparing the removal date (if one exists) and the start date found in the SACWSIS data set. I f the removal date is within 30 days of the case start date then this variable is recorded as a yes or the variable removal is present This 30 day limit is an attempt to capture only those cases wherein the child was removed as part of the initial inves tigation. The DCF child protection investigators have a standard of 60 days to complete their investigation and turn the case over for services or close the case (Florida Administrative Code 65C 10.003). Again it is essential to focus on variables that would be known at the time of case transfer from the investigating unit to the caseworker service organization. A removal within 30 days of the start date indicates that the removal was a direct result of the initial investigation and would be known at the ti me of transfer. Again, according to the U. S. Department of Health and Human Services Administration of Children and Families 2007 Child Maltreatment report, n ationally, it is estimated that 269,000 children were removed from their homes as a result of a child maltreatment investigation. Approximately one fifth of victims (20.7%) were placed in foster care as a result of an invest igation compared to 21.5% for F ederal F iscal Y ear
66 2006. Although the national percentage of victims who were removed from home or received foster care services at the time of the in vestigation is 20.7%, several s tat es reported more than 40% of victims received foster care services. In addition, 3.8% of nonvictims experienced removal. This is the first year that the Administration of Children and Families reported on children removed as the result of an investigation. The Administration for Children and Families did not indicate any time frame for when the removal took place only that it was the result of an investigation. The data set for this study shows a higher removal rate of 36.8% than the national average of 20.7%. The independent variable of race was constructed through a combination of utilizing the SACWSIS data set categories of race and ethnicity. This is done to allow f or the inclusion of Hispanic children as a specific category within this variable. In the SACWSIS data set Hispanic is only listed as an ethnicity along with Haitian and other ethnicities. The race options include Asian, Black (SACWSIS data set term), Whit e, Hawaiian, and Indian. The data were organized into four categories of Black, White, Hispanic or other. The other field includes Asian, Hawaiian, and Indian. The Hispanic category within the variable was constructed by recoding the race selection to the ethnicity selection of Hispanic, if the child was marked as Hispanic under ethnicity. This process avoided any duplicate counting of children identifying those children with a Hispanic or Latino origin and is the common practice used by other researchers s uch as the Armstrong, et. al (2005) accessing the SACWSIS data set. According to the Administration for Children and Families (ACF) 2007 Child Maltreatment Report (2009), Black children, American Indian or Alaska Native children, and children of multiple races had the highest rates of victimization at 16.7, 14.2, and
67 14.0 per 1,000 children of the same race or ethnicity, respectively. Hispanic children and White children had rates of 10.3 and 9.1 per 1,000 children of the same race or ethnicity, respective ly. Asian children had the lowest rate of 2.4 per 1,000 children of the same race or ethnicity. Nearly one half of all victims were White (46.1%), one fifth (21.7%) were Black and one fifth (20.8%) were Hispanic. The Hispanic population (5.5%) in our case s selected for this study represents a much smaller percentage of the total study group while the Black (32.8%) and White (60.8%) populations represent a larger percentage of the cases than what the ACF reports as national race/ethnicity percentages. T he independent variable developmental disability was constructed by utilizing the SACWSIS data set fields of mental retardation (SACWSIS data set term), visually impaired, physically disabled, and specific diagnosis types under the SACWSIS data field labeled clinical disability description. These data fields were combined, meaning that if one or any combination of them were documented on the case it was considered to have the developmental disability variable present. This combination of SACWSIS data set fields accounts for both physical and mental disabilities. According to Bruhn (2003) approximately 6.6% of the general population under the age of 18 has identified disabilities. Bruhn further found that children with disabilities were over represented in the child welfare system. However, according to the Administration for Children and Families (ACF) 2007 Child Maltreatment Report (2009), Florida reported 2.7% of child victims were mentally retarded and/or physically disabled. Nationally the states repor ted 3.9% of child victims were mentally retarded and/or physically disabled. The prevalence of development disability condition within the study group of cases is over twice that of the prevalence reported by Florida for 2007 and 160%
68 of the prevalence rep orted nationally. The difference in the Florida reporting from 2003 (year this study is analyzing) of 6.7% for all cases opened for services and 2007 (year reported to ACF) of 2.7% could be that this study is not looking at the children active in 2003 just the children with cases opened for services that year, while the report sent to ACF included all children active in 2007. The lead agency study cases (6.3% indicating developmental disabilities) and those cases open for services in Florida in 2003 (6.7% i ndicating developmental disabilities) appear to be consistent with what Bruhn found in the general population under the age of 18 but not an over representation of the prevalence of this characteristic as Bruhn found. The independent variable living arrang ement is represented in the SACWSIS data set by 28 possible living arrangement options for the investigation units to choose when the case is initiated in the system as well as caseworkers to choose from during the course of the case. These living arrangem ents may or may not be documented at the time of case initiation. This is the last living arrangement documented in the case record. Some of these children may have experienced a placement during the course of their case. The SACWSIS data set does not hav e a mechanism to determine if this was the only living arrangement during the course of the case. These living arrangement options include several categories of non relative and relative placements as well as the full array of placement types offered to children in the system. For the purposes of this study these 28 options were consolidated into five primary categories of; one parent, mother and father, relative, nonrelative, and other. This study is looking at this variable at a higher level of abstract ion to determine if these general areas of living arrangement influence the amount of time a caseworker spends on
69 a particular case and not the more narrow subcategories. Additionally, these more general consolidated categories are consistent with the Admi nistration for Children and Families (ACF) 2007 Child Maltreatment Report (2009) categories for living arrangement as reported by each state. Each of the subcategories included in the living arrangement options in the SACWSIS data set that were consoli dated into each primary cat egory are listed below
70 Living Arrangement Recoded: 1. Living with one parent a. Domestic Violence Shelter with parent/caregiver b. Living with one parent 2. Living with a nonrelative a. Non relative b. Living with a married couple nonrelative c. Living with a single female nonrelative d. Living with a single man nonrelative 3. Living with mother and father a. Living with two parents 4. Living with a relative a. Relative b. Shelter home relative c. Living with a married couple relative d. Living with a single female re lative e. Living with a single male relative 5. Other living arrangements a. Adoptive home b. Agency group home c. DJJ residential program d. Department foster family group home nonrelative e. Department foster family home nonrelative f. Other placement g. Private agency foster fa mily home nonrelative h. Child caring facility i. Runaway status for more than 30 days j. Shelter home nonrelative k. Therapeutic foster home nonrelative l. Foster family (age 18+) m. Group care (age 18+) n. Independent living (age 18+) o. Developmental services program p. Hospi tal According to the Administration for Children and Families (ACF) 2007 Child Maltreatment Report (2009) of the 22 States that reported living arrangement data, 25.5% of victims were living with a single mother. 19.1% of victims were living with married parents, while approximately 21% of victims (20.9%) were living with both parents, but
71 the marital status was unknown. These 22 states reported 84.3% of the child victims living with one or both parents. The independent variable maltreatment type is represented in the SACWSIS data set by 48 possible maltr eatment types These include some very specific types of injuries or harm. For the purposes of this study these maltreatment types were condensed into the five primary categories of; physical abuse, sexual abuse, medical neglect, supervision neglect, and general neglect. This study is examining a higher level of abstraction regarding maltreatment types similar to living arrangement and, as will be explained later, placement type. The condensing of this vari able into five primary categories also allows for more universal discussion of maltreatment. Florida has many subcategories for physical abuse, for neglect and sexual abuse that may be unique to Florida, therefore grouping these types of maltreatment at a higher abstraction level allows for a more universal understanding and discussion. This recoding of these maltreatment types also aligns more closely with the Children and Families (ACF) 2007 Child Maltreatment Report (2009) maltreatment categories collect ed from all of the states. The ACF report includes psychological abuse and a multiple abuse type that are not collected in Florida. Each subcategory maltreatment type from the SACWSIS data set that was recoded into a primary ca tegory is listed below
72 Mal treatment Type Recod ed : 1. Physical Abuse a. Bruises/welts b. Bites/punctures/cuts c. Burns/scalds d. Dislocation e. Bone Fracture f. Internal injuries g. Skull fracture/brain or spinal cord damage h. Asphyxiation/suffocating/drowning i. Deadly Weapon injury j. Beatings k. Excessive corpora l punishment l. Other physical injury m. Inappropriate/excessive restraint n. Inappropriate/excessive isolation o. Confinement/bizarre punishment p. Other mental or psychological injury q. Family violence threatens child r. Failure to protect from inflicted injury s. Other child dead abuse/neglect t. Death due to abuse 2. Sexual Abuse a. Sexual battery (incest) b. Sexual battery (not incest) c. Sexual molestation d. Sexual exploitation e. Sexual abuse other child f. Child on child sexual abuse 3. Medical Neglect a. Physically drug dependent newborn b. Substance m isuse c. Substance exposed child d. Alcohol exposed child e. Poisoning f. Failure to thrive g. Failure to provide medical care h. Malnutrition/dehydration i. Medical neglect 4. Supervision Neglect a. Inadequate supervision caretaker present b. Inadequate supervision caretaker not prese nt c. Abandonment d. Inadequate supervision (Department of Juvenile Justice, DJJ) e. Caregiver incarcerated f. Caregiver hospitalized
73 g. Caregiver deceased 5. General Neglect a. Conditions hazardous to health b. Inadequate shelter c. Inadequate clothing d. Inadequate food e. Foster care r eferral f. Parents need assistance The general categories and assignment of the specific SACWSIS data set maltreatment types were reviewed independently by two individuals with extensive experience in child welfare services in the state of Florida. Both revi ewers agreed this use of general categories and the specific assignments of maltreatment types were appropriate and consistent with the standards and professional protocols within Florida. Again, according to the Administration for Children and Families (A CF) 2007 Child Maltreatment Report (2009) 59.9% of children in the child welfare system were victims of neglect (Medical Neglect and Neglect categories). The ACF does not delineate what the subcategories are under the neglect title so it may include types of abuse that are included in the physical abuse or sexual abuse categories listed above. The ACF reported 10.8% of children were victims of physical abuse 7.6% of children are sexually abused, a 13.1% of children with multiple maltreatments (this study d id not look at multiple maltreatments and only captured the primary maltreatment identified), and a 4.2% psychological maltreatment (this is not a maltreatment type available in Florida). The independent variable substance abuse is captured with in the SAC WSIS data set through the child protection investigator documenting this as a reason for referral. The SACWSIS data system allows for up to five reasons for referral to be documented. For the purposes of this study the substance abuse characteristic was marked affirmative
74 if substance abuse was given as reason for referral within any of these five opportunities, whether it was the first reason or the fifth reason or the first and third reason, etc. The statistics vary, but studies have shown that betwee n one third and twothirds of child maltreatment cases involve substance abuse (U.S. Department of Health and Human Services, 1999). In a recent survey by the National Center on Child Abuse Prevention Research, 85 percent of States reported substance abuse was one of the two major problems exhibited by families in which maltreatment was suspected (National Center on Child Abuse Prevention Research, 2001). The prevalence of substance abuse as a reason for referral in the lead agency study cases is at the low end of the range of the national prevalence of substance abuse in child maltreatment cases. The independent variable domestic violence was calculated in the same fashion as the independent variable substance abuse above. The child protection investigator or services caseworker identify reasons for referral to services for each case. The SACWSIS data system allows for up to five reasons for referral and the domestic violence variable is present if the case has domestic violence documented in anyone or sever al of the five opportunities. In the 2009 report by Administration for Children and Families 34 states reported data regarding domestic violence as a risk factor for the caregiver of the child victim. These states reported 14.9% of victims and 2.6% of nonvictim children in the home had a caregiver risk factor of Domestic Violence. The independent variable placement type represents the first placement for each child. If a child was not placed during the course of their case then it was determined that they were placed at home. The SACWSIS data set contains 49 possible placement type options for the investigator or caseworker to choose when the child is placed. These
75 placement type options include several categories of similar placements as well as the full a rray of placement types offered to children in the system. For the purposes of this study these 49 options were consolidated into 11 primary categories of; relative, nonrelative, shelter, foster care, therapeutic foster care, group home, institution, Depa rtment of Juvenile Justice, development disabilities placement, runaway, and other. This consolidation simply combined the like placement types under the common general category such as under Relative placement; Approved Relative, Family Shelter Relative, Foster Home Relative, Non Licensed Shelter Relative, and Medical Foster Home Relative were included. This study is examining a higher level of abstraction regarding placement types similar to living arrangement and, maltreatment types discussed above. Fl orida has many subcategories for foster care, institution, Department of Juvenile Justice, relative and nonrelative that are unique to Florida, therefore grouping these types of maltreatment at a higher abstraction level allows for a more universal unders tanding and discussion. As an example, several of the subcategories below have a reference to under 12 or over 12 which refers to the number of beds within the facility. This distinction of the number of licensed beds within a facility is documented due to the variance of the Medicaid funding requirements and state licensure rules depending on the size of the facility. This recoding of these maltreatment types also aligns more closely with the Children and Families (ACF) 2007 Child Maltreatment Report (2009) placement categories collected from all of the states. Each of these placement subcategories in the SACWSIS data set that were consolidated into a primary category are listed below.
76 Placement Type Recoded : 1. Relative a. Approved relative b. Family Shelte r Home, relative c. Foster Home, relative d. Non licensed shelter bed relative e. Medical foster home, relative 2. Non Relative a. Approved nonrelative b. Non custodial prospective parent c. Non licensed shelter bed, nonrelative d. Temporary unlicensed facility e. Temporary unlice nsed placement f. Pre adoptive home 3. Shelter a. Runaway Shelter under 12 b. Runaway shelter over 12 c. Shelter Facility under 12 d. Shelter Facility over 12 4. Foster Care a. Adoptive Home b. Family shelter home non relative c. Foster home nonrelative d. Medical foster home non relativ e 5. Specialized Therapeutic Foster Care a. Therapeutic foster home nonrelative b. Therapeutic foster home relative 6. Group Home a. Group Home under 12 b. Group home over 12 7. Institution a. Hospital b. Mental health facility under 12 c. Mental health facility over 12 d. Residential Tr eatment under 12 e. Residential treatment over 12 f. State Hospital g. Substance Abuse detox under 12 h. Substance Abuse detox over 12 i. Substance Abuse residential treatment under 12 j. Substance Abuse residential treatment over 12 8. Department of Juvenile Justice (DJJ) a. DJJ detention center under 12 b. DJJ detention center over 12 c. DJJ facility under 12 d. DJJ facility over 12
77 e. Jail/prison 9. Developmental Disabilities Placement (DD) a. DD foster home relative b. DD foster home nonrelative c. DD group home under 12 d. DD group home over 12 10. Runawa y 11. Other a. Abducted b. Converted (this term is not defined within the data set) c. Parent/adult absconded d. Placement ICPC other state e. Supervised practice independent living f. Other This variable was constructed by selecting the first placement type during the life of the case. This first placement could have occurred at any time during the life of the case. Therefore, this variable may or may not be known at the time of case transfer from the investigation staff to the subcontracting caseworker services agency. This is different than the definition of the independent variable removal discussed above as that variable was only gathered within the first 30 days of the case. The data indicated that nearly 63% of the children were not removed from their primary caregiver within the first 30 days of the case. Th e independent variable prior removal was constructed by comparing removal dates associated with the case and the case begin date. As discussed earlier the SACWSIS data set included some data prior to 2003 and actuall y included the Florida Department of Children and Families legacy data from the client information system that was utilized prior to the SACWSIS system. Therefore, some of the lead agency study cases have removal dates prior to the case begin date associat ed with prior cases that had been closed. These are the cases identified to have the prior removal characteristic. In looking
78 at the SACWSIS data set made available for this study this is the best indicator that the case has been active in the past and sp ecific to the same child that is included in this study. Prior abuse reports were identified by family and the data was not accessible for this study. Additionally, prior placements were considered as a possible source to determine previous involvement or a history of abuse however, placements were not documented in the client information system. The focus groups identified prior removal as a key characteristic that would likely lead to more caseworker time spent on the case. According to the US Departmen t of Health and Human Services Administration for Children and Families (2007), states reported between 1.6% and 13.7% of children were again victims of abuse within 6 months following the initial abuse report. While the national data from the Administrati on for Children and Families documents a specific time frame and includes any occurrence of substantiated abuse this variable only reflects actual removal of the specific child resulting from a previous case. Caseworker Time, Dependent Variable Once Flori da began to shift from a system wherein all the caseworkers were state employees under the DCF to a system where all the caseworkers are employed by Community Based Care caseworker service organizations, Florida placed a requirement in each contract with t he Community Based Care lead agency requiring all caseworkers employed under the contract to enter each activity and the time related to that activity into the SACWSIS data set. The state further require d completion of start and end time fields be completed for each case note or activity entered regarding a specific case/child. The SACWSIS data set, as defined in Florida Administrative Code 65C 30.001, is the
79 official record for the child protection system in Florida. State administrative code 65C 30.001 re quires thorough documentation of each case entered into the SACWSIS system. The dependent variable was constructed by adding the time indicated for each activity recorded on a specific case with a beginning time and an ending time (complete activity). A recorded activity needed to have a beginning time and an ending time to define the amount of time a caseworker spent on the activity. Some activities were only recorded with a beginning time and thus could not be included in this study. Additionally, any a ctivities that indicated more than eight hours of continuous time were disregarded as a data entry error that would invalidate the variable. In examining the activities that were used to record a caseworkers time (telephone contacts, face to face visits, field visits, family visitations, staffing events, note to file, and case reviews) it becomes clear that any activity that has more than eight hours associated with it is unreasonable. It may be that the argument could be for a cut off at some other amount time (6 or 4 hours), this was just a self imposed limit to meet a reasonableness standard. The SACWSIS data set contained 53 activity options for caseworkers and investigators to choose from when recording their work. For the purposes of this study the se 53 options were consolidated into 7 general categories of activities (Telephone Contacts, Face to Face Visits, Filed Visits with Collaterals, Family Visitation, Staffing Events, Note to File, and Case Reviews). This consolidation grouped several variati ons of similar activities under a common general theme as can be seen below. Again, this study is examining the data at a higher level of abstraction. As shown in the caseworker activities recorded list there are several types of specific activities under each general category that may be unique to Florida but they logically role up to
80 the general category. This consolidation did not have any impact on the number of activities associated with each case, the overall number of activities, or the individual or collective time recorded with the activities as these values were all aggregated in the general category. This consolidation was done to help understand the general types of activities the caseworkers were spending their time. The general categories were used to compare the SACWSIS data base activities/tasks with what was gathered through the focus groups and the job descriptions as discussed above. Caseworker Activities Recoded: 1. Telephone Contacts a. Telephone Contact 2. Face to Face Visits with the Child a. Clien t Contact School Visit b. Home Visit Childs Current Residence c. Home Visit Home of Other Parent d. Home Visit Home of Parent/Caregiver Removed From 3. Field Visits with Collaterals a. Contact CPT/SATP b. Contact Non Caregiver Relative c. Contact School Officials d. Contact Serv ice Provider e. Filed/Office VisitCourthouse f. Field/Office VisitHome of Contacted Person g. Field/Office VisitLaw Enforcement Facility h. Field/Office VisitMedical/Psych. Office/Facility i. Field/Office Visit Other j. Field/Office VisitSchool k. Field/Office VisitWorkplace/Office l. Medical Provider m. Relative/Non Relative Collateral 4. Family Visitation a. Visitation b. Visitation Other c. Visitation Other Parent d. Visitation Parent/Caregiver Removed From e. Visitation Relative f. Visitation Sibling 5. Staffing Events a. Case Transfer Summary b. ESI St affing
81 c. Staffing Case Transfer/ESI d. Staffing Legal e. Staffing Multidisciplinary f. Staffing Other g. Staffing Permanency h. Staffing Reunification i. Staffing Separated Sibling 6. Note to File a. Birth Verification b. Converted (this is not defined) c. Courtesy Supervision Quarterly Summary d. Fingerprints Obtained e. Modification of Placement Summary f. Note to File General g. Note to File Interstate Compact h. Note to File Legal i. Note to File Summary j. Out of Home Placement FC/RGC k. Out of Home Placement Shelter l. Photographs Obtained m. Six Month Family Pr ogress Assessment n. Termination Summary 7. Case Reviews a. Reviews Administrative (no removal) b. Reviews CWLS c. Reviews Judicial (no removal) d. Reviews Other e. Reviews QA f. Reviews Supervisory The SACWSIS data set that was first received contained 4,508,391 activities reco rded for all 104,037 cases opened for investigation in 2003. This list of activities was modified by only considering those activities that were complete (begin time and end time) and less than 8 hours in duration. These two modifications were done to elim inate those activities that had unlimited time associated with them (no end time) and those that seemed unreasonably long over 8 hours. The assumption here is that no activity by a caseworker would take longer than 8 hours and there must have been a data e ntry error. With these two modifications the number of activities was reduced by 26% to 3,343,151.
82 Then the number of cases was modified, as has been described above, by eliminating cases that were not opened for services, where the child was identified as being 18 years old on the day the case was opened (begin date) and where cases were opened for less than 60 days. The reasons for these modifications have been discussed. Cases for young adults 18 years or older have completely different criteria and du ring the time frames of this study most activities for these young adults were not required to be recorded in the SACWSIS data set. However these cases could have had the dates of birth entered in error and still had significant activities associated with them but this distinction is not within the scope of this study. Those cases opened for less than 60 days were likely closed by the investigator and never transferred to the service units, thus these cases are not appropriate for this study of the time a c aseworker spends on a case referred for services. Along with the consideration of cases with duration of less than 60 days is the elimination of any cases not opened for services. As stated above, there were 104,037 cases opened for investigation in 2003. All of the cases opened for investigation had recorded activities associated with them. According to the ACF 2007 Child Welfare Report (2009) states averaged around 25% rate of substantiated findings of child abuse of those cases investigated. In 2003 Flor ida opened 36,820 cases for services indicating the abuse allegations were substantiated or services were required to mitigate imminent threat of abuse reflecting a 35% rate of substantiation. These modifications reduced the number of cases in the study to 31,564 cases that were determined to meet all the criteria for the study with 818,998 complete activities. Of these cases meeting the criteria 8,133 cases comprised the lead agency study cases with 330,311 documented activities and 23,431 cases comprised the nonlead agency data set with 488,685 documented activities.
83 Again, it needs to be noted that these nonlead agency cases were all transitioned over to new lead agency projects by December 31, 2004. In the fol lowing chapter this data is analyzed in four steps. First, through a series of matrices, the focus group identified characteristics and tasks, the job description tasks, and the SACWSIS data set general activity categories were compared and analyzed for consistency and outliers. Second, each indepe ndent variable and the dependent variable were compared between the lead agency study cases and the nonlead agency cases to determine if the cases are similar and discuss any significant differences found. An Independent Samples t test was utilized to det ermine any difference and the significance of the difference between the two independent groups for the independent variable age and the dependent variable, as the only interval level of measurement variable s A Cohens d is calculated to measure the effect size of any significant finding with these variables and to describe the strength of the association as well. All of the remaining independent variables were compared between the lead agency study cases and the non lead agency cases by calculating a chi square value. Chi square is used to describe the relationship between independent groups with categorical data and each of the independent variables except for age is a categorical variable. Additionally, an odds ratio or a Cramers V value is calculated f or any significant finding to measure the effect size and describe the strength of the association. Odds ratios were used to evaluate the importance of predictors. The statistically significant predictors that change the odds of the outcome the most were i nterpreted as most important, that is, the greater deviation of odds ratio from 1, the more influential the predictor. This analysis indicates the comparability and generalizability of the study group and the nonlead agency cases.
84 These two groups togeth er account for all of the cases opened for services in Florida during the calendar year 2003. Third, the independent variables were compared individually using a bivariate analysis to describe the relationship of each independent variable to the dependent variable. This analysis provided insight into the variance in the dependent variable explained by the individual independent variables. Finally, two types of regression analysis were used to analyze and determine the effect on the variance within the depen dent variable. First two multiple regression analysis w ere conducted with two groups of independent variables. The first analysis equation included the independent variable removal. The second equation excluded the independent variable removal. The variabl es placement and removal and their interaction was described above. This analysis determined the total amount of variance in the dependent variable explained by the groups of independent variables. The coefficient of each independent variable explains the variables contribution to the dependent variable variance when all of the other independent variables are held constant. The two analysis equations are found in Figure 2 and Figure 3 below. The equations are referred to as the all variables Model (with th e removal variable) Figure 2, and the Placement Model (without the removal variable), Figure 3.
85 Caseworker time = a+b(age)+c(gender)+d (race) +e(race2)+f(mental health cond.) +g(developmentaldisability)+h(maltreatment1)+i(maltreatment2)+j(maltreatment3) +k(maltreatment4)+l(substance abuse)+m(domestic violence) +n(living arrangement1)+o(living arrangement2)+p(living arrangement3) +q(livingarrangement4)+r(placement1)+s(placement2)+t(placement3)+u(placement4) +v(placement5)+w(placement6)+x(placement7)+y(placement8)+z(placement9) +aa(placement10)+bb(prior removal) + cc(removal)+ residual Figure 2. Multiple Regression Equation with Removal, (all variables Model) Caseworker time = a+b(age)+c(gender)+d (race) +e(race2)+f(mental health cond.) +g(developmental disabi lity)+h(maltreatment1)+i(maltreatment2)+j(maltreatment3) +k(maltreatment4)+l(substance abuse)+m(domestic violence)+n(living arrangement1) +o(living arrangement2)+p(living arrangement3)+q(living arrangement4) +r(placement1)+s(placement2)+t(placement3)+u(placement4)+v(placement5) +w(placement6)+x(placement7)+y(placement8)+z(placement9)+aa(placement10) +bb(prior removal) +residual Figure 3. Multiple Regression Equation without Removal, (Placement Model) Within these equations a is the constant and b through cc represent the coefficients and size of the effect created by the associated independent variable. A Beta co efficient was determined for each independent variable allowing for standardized comparison of effect on the dependent variable. The residual is the portion of the variance in the dependent variable that is not explained.
86 The analysis computed an adjusted R2 adjusting for the number of variables and sample size while measuring the amount of variance these variables explain in the dependent variable The data was then analyzed through a multivariate stepwise regression analysis for each equation above. This analysis identified the independent variable that accounts for the largest portion of the variance in the dependent variable and then added the next independent variable that accounts for the largest portion of the remaining variance, etc. This form of analysis continued to step through the independent variables until there was no independent variable remaining that accounted for any of the rema ining variance in the dependent variable. Through this analysis specific independent variables were identified that do not explain any unique portion of the dependent variable variance and therefore are not necessary to be included in the final model. The equation for the stepwise regression analysis is found in Figure 4 below. Caseworker time=a+b(independent variable 1)+c(independent variable 2) if significant .+residual Figure 4. Multivariate Step wise Regression equation. Within this equation a is th e constant and b, c, and additional coefficients through cc, if needed to account for additional variance in the dependent variable, represent the coefficients and size of the effect created by the associated independent variable. A Beta co efficient was determined for each independent variable allowing for standardized comparison of effect on the dependent variable. The residual is the portion of the variance in the dependent variable that is not explained.
87 The analysis computed an adjusted R2 adjusting f or the number of variables and sample size while measuring the amount of variance these variables explain in the dependent variable.
88 C hapter I V Results This chapter reviews the results of the data collection and analysis as described in the previous chapter. First, the results of the caseworker and supervisor focus groups describing the tasks and the characteristics identified and a matrix of case characteristics are developed including the focus group results and the SACWSIS data set independent var iables. Then the job description findings are described and the matrix of caseworker tasks including tasks from the focus groups, the job descriptions and the SACWSIS data set (activity codes for caseworker time) is developed. Following these results are d ata descriptions of each independent variable and the dependent variable along with bivariate and regression analysis using the equations described in the Methods Chapter. Caseworker Focus Groups Two focus groups were held with caseworkers meeting the ide ntified criteria. Below, in Table 2, is the list of the caseworker focus group results regarding case characteristics. This list is all inclusive from both focus groups using the nominal group technique. This table also indicates the case characteristics i dentified by the caseworkers as known at the time of transfer, and those that may or may be known at the point of transfer.
89 Table 2 Caseworker Identified Characteristics Case Characteristics Characteristics Known at Transfer May/May not be Known at Transfer Teenagers/Age of the Child Teenagers/Age of the Child Children on Psychotropic Medication Children on Psychotropic Medication Substance Abuse Substance Abuse Removal Removal Missing Parents or Unknown Missing Parents or Unknown Teen Parent s Teen Parents Large Sibling Groups/Number of Placements within the Sibling Group Large Sibling Groups/Number of Placements within the Sibling Group Re entry into the system (Re open case) Re entry into the system (Re open case) Sexually Abused Child ren Sexually Abused Children Caregiver Mental Health Disorder Caregiver Mental Health Disorder Expedited TPR Expedited TPR Child/Teens With Mental Health Disorders Child/Teens With Mental Health Disorders High Profile Media Coverage (Gov., LE, DCF ) High Profile Media Coverage (Gov., LE, DCF) Age of the Parent Age of the Parent History of Abuse/Return to Care History of Abuse/Return to Care Domestic Violence Domestic Violence Child with Criminal Record Child with Criminal Record Prior Rem oval Prior Removal Children with Development Disorders Children with Development Disorders Parent Non English Speaking Parent Non English Speaking Medically Needy Children (Medical Neglect) Medically Needy Children (Medical Neglect) Placement Type Placement Type Location of Family (Rural vs Urban) Location of Family (Rural vs Urban) Number of Placements Number of Placements 2 2 Number of Placements will not be known until the case is completed and closed. This characteristic was identified by the focus group but is not a consideration for inclusion in this study as an independent variable.
90 Noticeably absent from the list of case characteristics that were identified by the caseworker focus groups as indicators that a specific case would take more time were race and gender. The caseworkers were asked about this following the focus group and they did not see these characteristics as significant factors. They acknowledged that some of the other characteristics identified may be disproportionately distributed to specific racial groups or gender group but they did not see this as a race or gender issue. The caseworkers then identified those characteristics that could/would be known at the time of case transfer. T he caseworkers agreed that some of these characteristics may be known at the point of transfer or may not be known. The one characteristic that was identified that could not be known at the time of case transfer is the number of placements a child has h ad. The caseworkers related that as the number of placements for a child increased the amount of time spent on the case also increased due to finding and stabilizing placements, working through the disruptions with the child and the caregiver, going to court and documenting the case record, as well as increased staffing and supervisory oversight on the case. The caseworkers further discussed that they discover many facts about a family and the history of the family throughout the life of the case. Not everything needed to properly assess a case is known at the point of transfer. The caseworkers stated that the known characteristics at the time of transfer significantly depend on the quality of the investigation and the amount of information the investigator has collected. The caseworkers each described a current or recent case that took up a disproportionate amount of their time. These examples were focused on the ability or the availability of the parents to be engaged and worked with, the behavior and diagnosis of
91 the child, and the requirements of the state in tracking and intervening regarding specific circumstances (placement type, medications, etc.). The group discussed the burden of high profile cases. These are cases where there is media attention, G overnor or Secretary of the Department of Children and Families attention, legislators involvement or any combination of these. It was a consensus that high profile cases did result in more time spent on a case and that often this is a known characteristi c at the time of referral. But the group also acknowledged that these were rare cases and that there was nothing in the SACWSIS data set that would indicate a case as high profile. The caseworkers discussed child death cases as being very intense but for a relatively short period of time. It was also discussed that when there were siblings in a child death case it did take longer to resolve. The second part of the caseworker focus group was to identify specific tasks that they spent their time on when w orking on a case. These tasks did not include administrative tasks such as leave, community meetings, or training as the SACWSIS data set did not capture these items and they are not assignable to a specific case. Below is the list of tasks identified by t he caseworkers. Home Visits (required at a minimum every 30 days) Transportation Placement (getting children accepted and stable) Service Engagement (making referrals, getting clients started, follow up) Psychotropic Medication Protocols SACWSIS data entry School Visits Telephone Contacts Supervision Staffings/Meetings (on specific cases) The caseworkers emphasized how much traveling and transportation they do and then still have to document everything in the SACWSIS data system. The above list is
92 exhausti ve; the group did not mention any other items. Home visits varied with the minimal of one every 30 days. Some cases required more visits to be certain the situation was safe and stable for the children. The caseworkers spoke of visiting the school being time consuming due to the protocols at each school. Caseworkers often had to speak with 2 or 3 school officials before being able to see the child. They were also often required to bring documentation of the childs school attendance and performance history. These identified tasks are not listed in any particular order. The request of the focus group was just to identify the tasks that caseworkers perform. Supervisor Focus Groups Results Two focus groups were held with supervisors meeting the above stated c riteria. Below, in Table 3, is the list of the case characteristics identified by the supervisors using the nominal group technique. Again, the Gabriel Myers case and resulting changes in psychotropic medication protocols were an influencing factor with th e supervisors and this may not have played such a prominent role only a few months earlier. Table 3 also includes those case characteristics identified by the supervisors as known at the time of case transfer and those that may or may not be known at the time of case transfer. Table 3 Supervisor Identified Characteristics Case Characteristics Characteristics Known at Transfer May/May not be Known at Transfer Substance Abuse Caregiver Substance Abuse Caregiver 0 to 5 Year Olds (age of the child) 0 to 5 Year Olds (age of the child) Removal Removal Uncooperative Parents/Parent Non Communicative (Missing, Unknown) Uncooperative Parents/Parent Non Communicative (Missing, Unknown) Children on Psychotropic Medications Children on Psychotropic Med s P rior History of Abuse/Prior Removal Prior History of
93 Case Characteristics Characteristics Known at Transfer May/May not be Known at Transfer Abuse/Prior Removal Child/Teens with Mental Health Disorders Child/Teens with Mental Health Disorders Caregiver with Mental Health Disorder Caregiver with Mental Health Disorder Domestic Violence i n the Home Domestic Violence in the Home Physically Abused Infants Physically Abused Infants High Profile Media Coverage High Profile Media Coverage Placement Type Placement Type Teen Parents Teen Parents Sexually Abused Children Sexually Abused Children Substance Exposed Children Substance Exposed Children Location of Family (distance to services, supports) Location of Family (distance to services, supports) Children with Delinquency Involvement Children with Delinquency Involvement Teens with Multiple Placements Teens with Multiple Placements Teens with Multiple Runaway Episodes Teens with Multiple Runaway Episodes Parents who were former foster children Parents who were former foster children Children with Developmental Disorders Children with Developmental Disorders Large Sibling Groups Large Sibling Groups Multiple Fathers Multiple Fathers Parent Non English Speaking Parent Non English Speaking Dual Diagnosis Children 3 Dual Diagnosis Children Cases Presenting Immigration Issues Cases Presenting Immigration Issues Medically Needy Children Medically Needy Children Native Indian Cases Native Indian Cases Failure to Thrive cases Failure to Thrive cases Sibling Death Sibling Death Non Court Cases Non Court Cases Ex pedited TPR Expedited TPR Criminal Charges Against Caregiver Criminal Charges Against Caregiver 3 Dual Diagnosis is defined as children with mental health and substance abuse diagnosed problems.
94 Again, noticeably absent from the list of case characteristics that were identified by the supervisor focus groups as indicators that a specific case woul d take more time were race and gender with the exception that the supervisor focus group identified Native American Indian cases. This racial/ethnicity characteristic was identified by supervis ors due to the additional work and coordination required under the Indian Child Welfare Act of 1978. This involves close coordination with tribal social workers and leaders as well as additional documentation. The identification of American Indians potentially taking more caseworker time was not because of anything specific about the race/ethnicity of American Indians, but it was due to the burden of working with another bureaucracy and additional required documentation. The supervisors were asked about this following the focus group and they generally did not see rac e/ethnicity as a factor, but more as an indicator that there would be a collection of the characteristics that they named above for minority children, particularly African American males. The supervisors discussed cases that did not have the identified cha racteristics and went very smoothly and quickly regardless of gender or race. The supervisors then identified those characteristics that would be known at the time of case transfer. The supervisors also discussed and identified the characteristics that may or may not be known at the point of transfer. The characteristics that were identified by the supervisors that could not be known at the time of case transfer are Teens with Multiple Placements and Teens with Multiple Runaway Episodes. The supervisors like the caseworkers, related that as the number of placements for a child increased the amount of time spent on the case also increased. The supervisors discussed the issues with teens that are runners and have multiple placements. Both focus groups of supervisors acknowledged that these factors
95 significantly increase the time spent on a case by the caseworkers, but they also stated that this would not be a known factor at the time of assignment. Case assignment by supervisors to ongoing service caseworkers occurs within the first 60 or so days of the case, Florida Administrative Code 65C 10.003 requires child protection investigators to finish the investigation within 60 days of referral and transfer the case to the ongoing service caseworker Placemen ts and runaway episodes occur during the lifetime of the case and therefore are not going to be known at the time of case assignment. The supervisors stated that these characteristics c ould be related to the age of the child, insinuating that the older the child the more likely they are to have multiple placements and runaway episodes during the term of their case. The supervisors also acknowledged that several factors about the children and families associated with a case are discovered throughout the life of the case. Not everything needed to properly assess a case is known at the point of transfer. Both supervisors and caseworkers stated that the known characteristics also depend upon the quality of the investigation. The supervisors did emphasize that t hey believed the entire list of characteristics were important. The supervisors also spoke of current or recent cases that they believed took more time by the caseworker. The supervisors spoke of the younger children being at higher risk and requiring mor e attention including day care, arranged/supervised visits with parents, and additional medical appointments. The supervisors emphasized the ability and level of cooperation of parents as being a factor in the amount of time a case would consume. Follo wing the identification of case characteristics by the supervisor groups there was an impromptu discussion of how the supervisors made their case assignment decisions. The discussion of case characteristics effecting the caseworker time stimulated
96 this dis cussion in both focus groups. Most supervisors reported making case assignment decisions based on what they believed the caseworkers capacity to be and the circumstances of the case. The supervisors reported trying to match the abilities of the worker and the capacity of the worker with the type of case. In looking at the case record the supervisors reported reviewing several factors including the type of abuse, whether the child was removed and where the child was placed, the age of the child, the circums tances of the legal parents, any known prior history, and the number of siblings. Next the supervisors identified the tasks they found caseworkers spent time on in performing their duties on each case. The supervisors were asked to limit the tasks to nonadministrative tasks that were case specific activities. The task list the supervisors arrived at is as follows; Home Visits (required at a minimum every 30 days) Placement (getting children accepted and stable) Psychotropic Medication Protocols SACWSIS Da ta Entry Supervision Staffings/Meetings (on specific cases) Court Activities The supervisors emphasized the amount of time caseworkers spent on court activities, preparing documents, staffing the case in preparation for court, and the time spent at court, waiting for the various hearings. Job Description Results In Table 4 below the job description tasks are listed on the left column with the responding agencies listed across the top row. This table is a com pilation of the job descriptions collected from 8 subcontracting casework service organizations. Each subcontracting caseworker service organization within the 12 counties covered by the lead
97 agencies included in this study was contacted and asked to submit a job description for analysis and inclusion in this study. Table 4 Job Description Tasks by Agency Job Tasks From Agency Job Descriptions Kids Hope United Lutheran Services Florida Childrens Home Society Childrens Home, Inc. Youth and Family Alternatives Family Preservation Services Camelot C ommunity Care Devereux Assessment X X X X X X X X Planning X X X X X X X X Linking Families to Resources X X X X X X X X Monitoring X X X X X X X X Advocacy and Community Outreach X X X X X X X Medicaid Comp. Assessment X X Case Plan Develop ment X X X X X X X X Case Plan Review (Monitoring) X X X X X X X X Evaluate Reports from Providers X X X X Develop and Maintain Case Record, Data Entry X X X X X X X X Court Testimony and Reports X X X X X X X Diligent Search X X X X Requir ed Reports X X X X X X X X Education to Family X X X X X X X Direction to Resources and training for Family X X X X X X Indentify and Monitor Risk (Safety Assessment) X X X X X X X X Home Studies X X X X X Interstate Compact X X Placement X X X X X
98 Job Tasks From Agency Job Descriptions Kids Hope United Lutheran Services Florida Childrens Home Society Childrens Home, Inc. Youth and Family Alternatives Family Preservation Services Camelot C ommunity Care Devereux Independent Living Services X X Arrange Trans. X X X X Arrange for Drug Screening X X Participate in CQI X X Assist with Medical coverage and Subsidies X X X X Coordinate Services X X X X X X X X Team/Staff Meetings X X X X X X X X Client Care Reviews X X X X X X X X Staff Development X X X X X Effective Working Relationships X X X X X X X X On Call Support X X X Removal of Children X X X X Child Visitations X X X X X X X Sibling Visits X X Incident Reports X X X Other Duties X Table 4 above indicates a total of 35 independent job tasks identified by the eight subcontracting casework service organizations reporting. Of these 35 tasks 13 were included in all 8 job desc riptions. An additional eight tasks were common to five or more agency job descriptions. Below is a list of these 21 common tasks. Assessment Planning Linking Families to Resources Monitoring Advocacy and Community Outreach Case Plan Development Case Pl an Review Develop and Maintain Case Record, Data Entry Court Testimony and Reports Required Reports
99 Education to Family Direction to Resources and Training to Family Identify and Monitor Risk (safety assessment) Home Studies Placement Coordinate Services to Children and Families Team/Staff Meetings Client Care Reviews Staff Development Effective Working Relationships Child Visitations Four agencies included removal of a child as a task. This task should not be confused with the activity around placement once a child is removed from the home. The appearance of this task raises some questions as the Community Based Care system in Florida was designed to separate the investigative activities from the ongoing service activities and it is necessary to determin e a child is at immediate risk of abuse through a child protection investigation in order to remove a child from her/his home (Florida Statutes 39.301). Similarly, four agencies listed the arrangement of transportation as a job task while all of the agenci es listed linking families to resources as a task. Additionally, only two agencies included the tasks of sibling visits. This is the arrangement of visits between siblings who are currently under the protective custody of the state but placed in separate locations or one sibling is under the protective custody of the state and is visiting siblings who are not under protective custody. Incident reporting is a critical requirement under the States regulations, Child and Family Operating Procedure, CFOP 2156, yet this was only found to be a listed task in three of the submitted job descriptions. Also Continuous Quality Improvement was reported in two job descriptions yet is a required task under Florida statute 409.1671 (4) (a). Independent Living is listed in just two job descriptions. Independent living refers to services provided to foster children ages 13 through age 22 and are mandated by Florida Statute 409.1451. Arranging transportation was included in four job descriptions of the eight. Finally only one job description included the proverbial, and other tasks as assigned.
100 Analytic al Phase I Focus Groups, job descriptions and SACWSIS data set independent variables and general activity categories comparison. Table 7 below is a matrix of the critic al case characteristics the focus groups identified and what was found in the SACWSIS data set. What is quickly identified is where all three, the caseworkers, the supervisors, and the selected independent variables from the SACWSIS data set match. This co mmon set of identified case characteristics includes; Substance Abuse, Age of the Child (caseworkers and supervisors were mostly concerned with teens and to a lesser extent preschool children), Domestic Violence, Placement, and some specific types of malt reatment and living arrangements. As has been mentioned earlier neither focus group mentioned race or gender as being a critical case characteristic, except for the characteristic mentioned by the supervisors regarding American Native Indians and the tribal issues that arise. Tribes in Flo r ida have their own social workers and cases must be coordinated and conducted together as required under the federal Indian Child Welfare Act Critical characteristics identified by the caseworkers and supervisors that were not included in the original model but could be known at the point of case assignment include; The mental health status of the primary caregiver (parent or other person child is living with when the abuse occurred) and the child (teen or child) The mental health status of the parent was not available in the SACWSIS data base. However, the SACWSIS data base does include information on whether a child has a clinical disability description and if the child is emotionally disturbed. These two character istics provide an indication of a mental health condition. These
101 characteristics within the SACWSIS data set were combined (using specific mental health disability descriptions); mean ing one or both were documented on the case, to create a mental health co ndition characteristic .Th is characteristic was then evaluated for being added to the model. The distribution of t his characteristic within the study group is shown be low in Table 5 As indicated in the table 4.2% of the study group population is identified with mental health condition using the two SACWSIS data set indicators. As discussed further in Chapter V this level of mental health condition prevalence in a foster care population falls dramatically below the research and literature findings. During th e time frame of this study mental health information was not reported by the Florida Department of Children and Families and was not accounted for in any of the quality assurance reviews or contract requirements reviewed for this study. This information a ppears to be so dramatically under reported it is not going to be included in the model Table 5 Mental Health Condition: Lead Agency Study Cases Frequency Percent Valid Percent Valid No Mental Health Condition 7793 95.8 95.8 Child has Mental Health Condition 340 4.2 4.2 Total 8133 100.0 100.0 The child has a developmental disability was identified by both caseworkers and supervisors in the focus groups as a characteristic effecting the time of a caseworker The SACWSIS data set does include information on whether a child is mentally retarded visually impaired, specific clinical disability descriptions, and if the child i s physically disabled. These data fields were combined meaning
102 one or any combination of them that were documented on the case, to create the Developmental Disability Condition characteristic. This combination of SACWSIS data set fields accounts for both physical and mental disabilities. The frequency distribution of this characteristic is shown in Table 6 below. This Development Disability Condition is present in 6.3 % of the study group population. Table 6 Developmental Disabilities: Lead Agency Study Ca ses Frequency Percent Valid Percent Valid No Developmental Disabilities 7617 93.7 93.7 Child has Developmental Disabilities 516 6.3 6.3 Total 8133 100.0 100.0 According to Bruhn (2003), 6.62% of the US population under 18 were reported with disabil ities in the 2001 Census. Further, Bruhn found that children with disabilities are overrepresented in the child welfare system because they are more likely to enter foster care than children without disabilities and less likely to leave. Although Bruhn di d not provide specific statistical data her findings clearly indicated that the number of children with disabilities in the child welfare system would be greater than the number of children in the general population with disabilities. Therefore, the SACWSI S data set identifying 6.3% of the children in the study group as having a disability seems very consistent with the general population prevalence found by Bruhn. This variable was added to the model as an independent variable for analysis in this study.
103 The childs delinquency involvement (this information could become available in the future as the Florida Department of Children and Families and the Florida Department of Juvenile Justice work to integrate their data systems), however this is not information available in the current SACWSIS data set. H istory of parents as foster children, and prior history of abuse are not all cons istently available to the supervisors at the point of transfer and the SACWSIS data set did not have this information in a reliable format for this study. The history of abuse as been historically linked to the parent and with the changes in the SACWSIS da ta set this information is not consistently captured or documented for a specific child. This study is looking at specific children and the case characteristics associated with them. Children on psychotropic medications have just recently been added as a data component in the SACWSIS data set. T his information will be available for future studies. High Profile Media coverage cases would often be known to the supervisor but is not collected in the SACWSIS data set. Large Sibling groups and the initial plac ement of these siblings are in the SACWSIS data set However, this information was not consistent enough to include in this study during the time frame being evaluated. Children are not always clearly identified as siblings in the SACWSIS data set. Due to the number of different family names and relations documented within the SACWSIS data set it requires a specific study of this variable to properly determine and define siblings.
104 Expedited T ermination of P arental R ights (TPR) is not uniquely identified wit hin the SACWSIS data base. This is a decision that is made early in the case and can be made before the transfer to the ongoing services unit from the investigations unit but the data is not available. Location of the family relative to services was not a characteristic considered for this study. The SACWSIS data set does have the addresses of the families and through various mapping techniques a determination of the families distance from services and caseworker offices could be determined. This could be the basis of future studies There are a number of case characteristics that were identified by the focus groups that are not in the SACWSIS data set and may or may not be known at the time of transfer. These include; uncooperative parents, nonEnglish spe aking parents, dual diagnosed children, immigration status, and criminal charges against the parents. Due to the absence of data on these characteristics they could not be included in the study. Non court cases or voluntary cases represent a very small nu mber of active cases, they are routinely closed within 6 months, and in some communities the documentation is very limited or not included in the SACWSIS data set so this characteristic was not included. Tee n parent was not a characteristic considered for this study due to the absence of data within the SACWSIS data set Finally some focus group identified characteristics were eliminated from consideration because they would never be known at the transfer of the case, such
105 as; independent living youth, as this only occurs after a case has been opened for a period of time and the child is over the age of 12, adoptions, again thi s only occurs after a case has been opened for a period of time and the parental rights have been terminated, and number of placements as discussed above.
106 Table 7 Case Characteristics Comparison Focus Groups Identified Case Characteristics SACWSIS Caseworker Identified Characteristics Supervisor Identified Characteristics Independent Variables Substance Abuse Substance Abuse Caregiv er Substance Abuse Teenagers Teens with Multiple Placements Age of the Child Teens with Multiple Runaway Episodes Age of the Child Teen Parents Teen Parents Teens With Mental Health Disorders Teens with Mental Health Disorders Mental Health Condition Child with Mental Health Disorders Child with Mental Health Disorder Mental Health Condition Caregiver Mental health Disorder Careg iver with Mental Health Dx. Not Present in Data Set Child with Criminal Record Child ren with Delinquency Hx. Not Present in Data Set Re entr y into the system Prior History of Abuse Prior Removal Prior Removal Prior Removal Sexually Abused Children Sexually Abused Children Maltreatment Type Age of the child 0 to 5 year old victims Age of the Child Physically Abused Ba bies Maltreatment Type Domestic Violence Domestic Violence in the Home Domestic Violence Children on Psychotropic Meds Children on Psychotropic Medication Not Present in Data Set Substance Exposed Children Maltreatment Type Parents former foster c hildren High Profile Media Coverage (Gov.,LE,DCF) High Profile Media Coverage Not Present in Data Set Missing Parents or Unknown Parent Non Communicative (Missing, Unknown) Living Arrangement Children with Development Disorders Children with Developme ntal Disorders Developmental Disabilities Cond. Large Sibling Groups Large Sibling Groups Multiple Fathers Living Arrangement Uncooperative Parents Not Present in Data Set Parent Non English Speaking Parent Non English Speaking Dual Diagnosis Ch ildren Not Present in Data Set Cases Presenting Immigration Issues Not Present in Data Set Medically Needy Children (Medical Neglect) Medically Needy Children Maltreatment Type Native Indian Cases Tribal and Reservation issues Race Removal Judicial Assignment Removal Failure to Thrive cases Maltreatment Type Sibling Death Maltreatment Type Non Court Cases Placement Type Placement Type Placement Type Expedited TPR Expedited TPR Not Present in Data Set Criminal Charges Against Caregiver Not Present in Data Set Location of Family (Rural vs Urban) Location of Family (distance to services, supports) Number of Placements within a Sibling Group Gender Race
107 There are 39 independent characteristics identified in Table 7 after two characteristics of independent living and adoptions were eliminated as stated above Of these 17 characteristics identified are not included as independent variables, ten of these were not present in the data base, four of these 17 characteristics (sibling groups, sibling placements, history of parents as foster children, history of abuse) did not have consistent and reliable data in a usable format t o be included in the analysis (as described above) three were not considered (noncourt cases, location of parents/caregiver, teen parents) for various reasons explained above Three of the characteristics identified by the focus groups were evaluated as independent variables (mental health of the child, mental health of teen s, and developmental disabilities of the child) with developmental disabilities being included in the final analysis of the study There were two characteristics that were selected as independent variables that the focus groups did not mention, race and gender. Race has been briefly addressed above as being of concern. There are numerous studies and reports of the overrepresentation of minority children in the child welfare system. Most of these studies indicate the real factor as poverty and family dynamics. Gender as an issue in the child wel fare system is not a widely discussed issue. The information available indicates some gender differences by type of abuse. Girls are more likely to be sexually abused (Finkelhor, 1994) and boys are more likely to be shaken (Carbaugh, 2004). Gender will be included in the analysis to control for general demographic information It is noteworthy that 1 9 characteristics identified were already identified in the independent variable group. If we eliminate the characteristics that are not in the SACWSIS data set ( 10 ) and those that did not have good data in the SACWSIS data set
108 ( 4) then there are 2 5 possible characteristics identified by the focus groups. The independent variables already capture 1 9 of these and the result of this analysis has lead to adding 1 ad ditional characteristic to the study. Therefore, the study includes 2 0 of the 25 possible characteristics or 80% of the focus group identified characteristics This indicates the model has validi ty from the field. Tabl e 8 below is a matrix of the tasks identified by the focus group participants, the subcontract caseworker services organizations job descriptions, and the general category of activities within the SACWSIS data set. There are 23 tasks identified by the focus groups and the job description ana lysis (common tasks identified above). In matching up the tasks in the SACWSIS data set and the tasks identified by the focus groups and job descriptions the consolidated SACWSIS data set tasks were considered with reference to the sub category tasks they contained as identified above. The Table 8 task matrix shows that 5 of the focus group and job description identified tasks are not captured by the SACWSIS data set tasks. These tasks include transportation, psychotropic medication compliance, team and staff meetings, staff development, and effective working relationships. The se tasks are simply not in the SACWSIS data set. The psychotropic medication issue has already been discussed above. This is a current hot button issue with the Florida child welfare system due to a couple of horrific cases. The state recently changed the reporting and requirements protocols for children on psychotropic medications and this happened outside the time frame of this study. Transportation was identified as a task by the c ase workers. It is interesting that this is not a specific task identified by the job descriptions or by the supervisors. The SACWSIS data set does not include this as a task. Some caseworkers could include the
109 time spent on transportation in the other tas ks they do that are in the SACWSIS data set (i.e. face to face visits with the child, family visitation, field visits with collaterals). There is no way to know at the time of this study and likely it is a different practice around the state. The remaining tasks not included in the SACWSIS data set (team and staff meetings, staff development, and effective working relationships) are internal staff activities conducted by the sub contracting caseworker services organizations and arguably would never belong i n the SACWSIS data set. The SACWSIS data set was never intended to be a time sheet. These tasks were not identified by the focus groups but only identified through the job description analysis. If the 5 tasks that are not included in the SACWSIS data set a re eliminated due to their absence then the SACWSIS tasks included in this study capture 100% of the available tasks identified by the focus groups and job description analysis.
110 Table 8 Caseworker Task Matrix Focus Group Caseworkers Focus Group Superviso rs Job Descriptions SACWSIS Data Set Home Visits Home Visits Child Visitations Face To Face Child Visits Transportation Placement Placement Placement Removal Service Engagement Linking Families To Resources. Direction to Resources and Training to Family. Filed Visits with Collaterals Psychotropic Medications Psychotropic Medications SACWSIS Data Entry SACWSIS Data Entry Develop and Maintain Case Record, Data Entry Note To File School Visits Coordinate Services to Children and Families Field Visits with Collaterals Telephone Contacts Telephone Contacts Supervision Supervision Client Care Reviews Case Reviews Staffing Meetings Staffing Meetings Case Plan Review Staffing Events Court Activities Court Testimony and Reports Field Visits wit h Collaterals Assessment Face To Face Child Visit Family Visitation Planning Staffing Events Monitoring Case Reviews Advocacy and Community Outreach Field Visits with Collaterals Case Plan Development Staffing Events Team/Staff Meetings Required Reports Note To File Education to Family Field Visits with Collaterals, Family Visitation Identify and Monitor Risk Face to Face Child Visits Family Visitation Home Studies Face to Face Child Visits Staff Development Effective Working Relationships
111 Again, this indicates a high correlation between the model and the field and supports the strength and accuracy of the model. Analytic al Phase II Comparison of variables between l ead agency study cases and nonlead agency cases us ing the independent samples t test and chi square. As has been discussed above this study is focused on the cases opened for services in 2003 and under the management of the Lead Agencies. This study group includes five Lead Agencies that were fully operat ional prior to January 1, 2003 providing services within 12 counties. At the same time of this study the Florida Department of Children and Families (DCF) were providing and transitioning services in the remaining 55 Florida counties to an additional 17 Le ad Agencies. The cases that were opened for services in 2003 in these nonlead agency counties are referred to as the non lead agency cases for the purposes of this study. By mid 2005 all child welfare services except investigations and legal services were being managed by a Lead Agency in all 67 Florida counties (Vargo, Armstrong, Jordan, Kershaw, Pedraza, Yampolskaya 2006). As discussed earlier, the number of cases included in the study after the adjustments for age and cases opened beyond 60 days for the CBC counties came to 8,133. The number of remaining cases referred to here as the nonlead agency cases came to 23,431. This Phase II analysis describes the frequency of each independent and dependent variable and compares the variables of the lead agency cases (study cases) with the variables of the nonlead agency cases. Throughout this analysis and further with the regression analysis to follow this study will use the standard of p < .01 to determine the significance of a relationship. This
112 means that the findings reported would have occurred by chance less than 1% of the time. Additionally, in working with the SACWSIS data set in this secondary analysis there are some issues with missing data. As described above, there are 8,133 lead agency study cases. In the following pages, as the independent and dependent variables are described it will be noted that numerous cases have missing data. This missing data represents the caseworkers failure to document the status of a particular variable in the SACWSI S data set. While this is no surprise in the child welfare system it is a concern that must be addressed. Cases with documentation of each variable (present or not present but documented) are included in the regression analysis and those cases without documentation of each variable are not included in the regression analysis. This results in 3,855 cases being included in the final regression analysis. This is standard procedure in the type of regression analysis being conducted. All variables with missing d ata are identified and the missing data is discussed. Further, following the regression analysis at the end of this chapter a comparison of the cases included in the analysis with those excluded is completed to verify that the cases are not significantly d ifferent. Independent Variable 1, a ge was a simple calculation of how old the child was on the begin date of the case, using the childs date of birth and the begin date of the case. Table 9 below shows the frequency of lead agency study cases by age.
113 Ta ble 9 Age: Lead Agency Study Cases Frequency Percent Valid Percent Valid 0 1307 16.1 16.1 1 635 7.8 7.8 2 549 6.8 6.8 3 563 6.9 6.9 4 495 6.1 6.1 5 507 6.2 6.2 6 450 5.5 5.5 7 403 5.0 5.0 8 423 5.2 5.2 9 358 4.4 4.4 10 385 4.7 4.7 1 1 363 4.5 4.5 12 362 4.5 4.5 13 335 4.1 4.1 14 333 4.1 4.1 15 317 3.9 3.9 16 243 3.0 3.0 17 105 1.3 1.3 Total 8133 100.0 100.0 There were 50 cases of children with ages of 18 years or more on the recorded begin date of the case that were removed from the sample because cases with young adults of 18 years old are treated very differently by the caseworkers in accordance to Florida Statute 409.1451. This brings the sample to 8,133. These young adults are not required to be seen as often and th e case record documentation requirements are very relaxed as these young adults are no longer formal cases due to their age. It is also recognized that these situations could be data entry errors, but impossible to determine so the cases were eliminated fr om the study group. As seen in T able 9 1,307 infants or 16.1% of the children in the lead agency study cases are less than one year old with 37.6% of the children being ages three and younger and 60.4% are 7 years old or
114 younger. The age distribution with in this study is relatively consistent with the national age distribution of child abuse victims, possibly our group is slightly younger. This distribution of age also has some consistency with the total group of Florida children associated with cases open ed in 2003 and closed prior to January 1, 2007. For the nonlead agency cases (23,431 cases meeting the same criteria as the study cases) there were fewer children under one year of age (14.3%), 35.4% were under the age of 3 years and 58.5% were 7 years ol d or younger. The independent samples t test comparing the age variable between the lead agency cases and the non lead agency cases reveals the average age for the lead agency children was 6.4 years and the average age for the nonlead agency cases was 6.6 years. Levenes test for equality of variances indicated there is not a lot of distortion in the data, the variance in the two groups was not significant p > .01 ( .355) Therefore, assuming equal variances the t (31,562) = 3.35, p < .01. This indicates a significant difference in the average ages between the two groups although the means are only .2 years apart, with the lead agency study cases representing a younger group. With this small of a difference in average age a Cohens d was calculated resultin g in an effect size of .02. According to Valentine & Cooper (2003) an effect size of .2 is small, .5 is medium, and .8 is large. So the effect size of the difference in age is extremely small and therefore it can be concluded that there is no real differen ce in age. Due to the large number of cases within this study very slight differences can be found significant and therefor e an effect size shall be calculated in these circumstances.
115 Independent Variable 2, gender is constructed from specific gender f ields within th e SACWSIS data set. Table 10 below shows the frequency for each gender within the selected lead agency cases for this study. Table 10 Gender : Lead Agency Study Cases Frequency Percent Valid Percent Cumulative Percent Valid Female 4068 5 0.0 50.0 50.0 Male 4065 50.0 50.0 100.0 Total 8133 100.0 100.0 This variable has an equal split between males and females in the selected cases. This is consistent with the state wide data for cases with a start date in calendar year 2003 and wit h an end date prior to January 1, 2007. For the nonlead agency cases (23,431 cases meeting the same criteria as the study cases) there were 50.1% of the cases were female and 49.9% were male. The chi square test comparing the gender variable between th e lead agency study cases and the non lead agency cases indicates that there is not a significant difference in the gender prevalence between the groups. X2(1, N = 31,557) = .004. Independent Vari able 3, mental health c ondition, was measured by identif ying cases that were documented as having a specific clinical diagnosis within the SACWSIS data field clinical disability description or an emotional disturbance or both. Table 5 below shows the frequency of mental health c ondition in the lead agency study cases.
116 Table 5 Mental Health Condition: Lead Agency Study Cases Frequency Percent Valid Percent Valid No Mental Health Condition 7793 95.8 95.8 Child has Mental Health Condition 340 4.2 4.2 Total 8133 100.0 100.0 As indicated in Table 5 there are 340 children or 4.2% of the lead agency study group population with an identified mental health condition using the two SACWSIS data set indicators. This level of mental health condition prevalence in a foster care population is well below the 57% pre valence level found by dosReis, Zito, Safer, and Soeken (2001). During the time frame of this study mental health information was not reported by the Florida Department of Children and Families and was not accounted for in any of the quality assurance revi ews or contract requirements reviewed for this study. For the non lead agency cases (23,431 cases meeting the same criteria as the study cases) there were 3.8% of the cases with an identified mental health condition. The chi square test comparing the ment al health condition variable between the lead agency study cases and the non lead agency cases indicates that there is not a significant difference in the prevalence of mental health issues between the groups. X2(1, N = 31, 564) = 1.940. Within the lead agency study cases that captured this information there were 4.2 % with this characteristic and within the no nlead agency cases there were 3.8% with this characteristic. With this prevalence level so significantly below the levels found in the literature and with the current information produced by the Florida Department of Children and Families regarding the children in foster care prescribed psychotropic medications this variable will not be included in the final regression analysis as it is unreliable.
117 In dependent Variable 4 r emoval was measured by comparing the removal date and the start date found in the SACWSIS data set. Table 11 below indicates there are 2,989 cases with a removal date within 30 days of the begin date. An additional 654 cases with th e lead agency study cases had removal dates after the first 30 days of case. Also as discussed below, the placement data indicates an additional 2,156 children were placed after the first thirty days outside of their home sometime during the life time of their case. Not all children who were placed outside of their home would necessarily have a removal date as the placement can be decided by the court during an active judicial review and a placement change order issued. Removals are primarily referring to an active abuse investigation resulting in the removal of the child for safety purposes. Table 11 below shows the frequency of removal and nonremoval within the cases selected for this study. Table 1 1 Removal within 30 days of Begin Date: Lead Agency Stud y Cases Frequency Percent Valid Percent Valid No Removal within 30 days 5144 63.2 63.2 Removal within 30 days 2989 36.8 36.8 Total 8133 100.0 100.0 As seen in Table 11 above, 2,989 children or 36.8% of the children in the lead agency study cases w ere removed from their family within the first 30 days of the case begin date. The data set for this study shows a higher removal rate of 36.8% than the national average of 20.7%. For the nonlead agency cases (23,431 cases meeting the same criteria as the study cases) there was a smaller ratio of children removed within 30 days of the case begin date (32.2%).
118 The chi square test comparing the removal variable between the lead agency study cases and the nonlead agency cases indicates that there is a sig nificant difference in the rate of removal between the groups. X2(1, N = 31,5 64) = 62.97, p < .01, odds ratio = 1.24. Within the lead agency study cases that captured this information the re were 36.8% of the children removed from their home within the fi rst 30 days and within the nonlead agency cases there were 32.2 % of the children removed with in the first 30 days The odds ratio indicates strength of association supporting a significant difference between the groups regarding the variable removal. Ther efore the lead agency study cases had a significantly higher prevalence of removal within the first 30 days of the case. Independent Variable 5, race/ethnicity was constructed through a combination of utilizing the SACWSIS data set categories of race and ethnicity. Any child identified with a Hispanic or Latino origin was included in the Hispanic category. Table 12 below shows the frequency of White, Black (SACWSIS data set term), Hispanic and other in the lead agency cases selected for this study. Table 1 2 Race/Ethnicity of the Child: Lead Agency Study Cases Frequency Percent Valid Percent Valid Other 68 .8 .8 White 4944 60.8 60.8 Black 2671 32.8 32.8 Hispanic 450 5.5 5.5 Total 8133 100.0 100.0 As indicated in Table 12 60.8% of the children in the lead agency study cases are white, 32.8% are Black, and 5.5% are Hispanic. As discussed above this distribution does represent a smaller portion of Hispanic children and larger portions of white and black children than the national averages. For the nonlead agency cases (23,431 cases meeting
119 the same criteria as the study cases) there was a smaller percentage of white children (49.2%) and larger percentages of black children (38.8%) and Hispanic children (11%). The chi square test comparing the race/e thnicity variable between the lead agency study cases and the nonlead agency cases reveals a significant difference in the race/ethnicity characteristics between the groups, X2(3 N = 31,564 ) = 401 .842, p < .01 ( .000) Additionally, the chi square test w as used to compare each race/ethnicity sub characteristics individually. When isolating and comparing the white children in the two groups approximately 61% of the lead agency study cases are white while 49% of the nonlead agency cases are white. The chi square test reveals this is significant and the lead agency study cases have a greater prevalence of white children X2(1, N = 31,564) = 323.81, p < .01, odds ratio = 2.33. Further testing indicated that there were significantly fewer black children X2( 1, N = 31,654) = 90.76, p < .01, odds ratio = 1.43 and significantly fewer Hispanic children X2( 1, N = 31,654) = 218.98, p < .01, odds ratio = 2.38 in the lead agency study group cases than in the nonlead agency cases. This indicates a significant difference in the race/ethnicity characteristics between the two groups. The odds ratio for each type of race ethnicity indicates a strong association supporting the difference in race/ethnicity between the groups. Independent Variable 6, developmental d isabiliti es, was constructed by utilizing the SACWSIS data set fields of mental retardation (SACWSIS data set term), visually impaired, physically disabled, and specific diagnosis types under the SACWSIS data field labeled clinical disability description This c ombination of SACWSIS data set fields accounts for both physical and mental disabilities. The frequency distribution of
120 this characteristic is shown in Table 6 below. This Development Disabili ty Condition is present in 6.3% of the study group population eq uating to 516 children. Table 6 Developmental Disabilities: Lead Agency Study Cases Frequency Percent Valid Percent Valid No Developmental Disabilities 7617 93.7 93.7 Child has Developmental Disabilities 516 6.3 6.3 Total 8133 100.0 100.0 The lead agency study cases with a 6.3% prevalence of developmental disabilities and those cases open for services in Florida in 2003 with a 6.7% prevalence of developmental disabilities appear to be consistent with the prevalence within the general population under the age of 18 of 6.6% ( Bruhn, 2003) but not an over representation of the prevalence of this characteristic as Bruhn found. For the nonlead agency cases (23,431 cases meeting the same criteria as the study cases) there were 1,613 children or 6.9% of t he cases with an identified developmental disabilities condition. The chi square test comparing the developmental disabilities characteristic between the lead agency study cases and the non lead agency cases indicates that there is not a significant diffe rence in the prevalence of developmental disability issues between the groups. X2(1, N = 31,564) = 2.794. Within the lead agency study cases that captured this information the re were 6.3 % of the children had a developmental disabilities characteristic docu mented and within the nonlead agency cases there were 6.9 % of the children had this characteristic.
121 Independent Variable 7, living a rrangement was constructed from the 28 options of living arrangement found in the SACWSIS data set into the 5 subcategor ies for this variable of; one parent, nonrelative, mother and father, relative, and other. Table 13 below indicates that the other living category while having numerous living arrangement suboptions rolled up into it, as described in the previous chapter actually captures only 1.5% of the total. These living arrangement options included in the other category are also mostly licensed placements. Table 13 Living Arrangements: Lead Agency Study Cases Frequency Percent Valid Percent Cumulative Percent V alid Living with one parent 3920 48.2 71.4 71.4 Living with non relative 24 .3 .4 71.8 Living with mother and father 1092 13.4 19.9 91.7 Living with relative 335 4.1 6.1 97.8 Other living arrangements 121 1.5 2.2 100.0 Total 5492 67.5 100.0 Mis sing 2641 32.5 Total 8133 100.0 This variable revealed 3,920 children, or 48.2 % of the children in the cases selected for this study were living with one parent and 1,092 children, or 13.4% of the children were living with their mother and father. This indicates that 61.6% of these children were living with one or both parents This percentage could be even larger but for the fact that there is 32.5% of the selected cases w ith missing data for this variable. It seems fairly reasonable to assume that every child has some sort of living arrangement on the begin date of the case. Therefore, the missing data is assumed to be due to the failure of the caseworker to document this in the SACWSIS data set. The 2,641 cases with missing data will be dropped out of the final regression analysis as a result. A comparison
122 of the cases retained in the model due to having all variables accounted for and those cases that are dropped due to missing cases is completed later in the next chapter. If only those cases rep orting a living arrangement are considered then the percent of children in the lead agency study cases living with one or both parents becomes 91.3%. Table 14 reflects the living arrangements for the nonlead agency cases. The non lead agency cases reflect 49.1% of the children were living with one parent and 17.6% were living with both parents This indicates that 66.7% of the children were living with one or both parents. This variable has 27.2% of the cases missing documentation of a living arrangement. The assumption is again that this is due to the failure of the caseworker to document this information in the SACWSIS data set. Again, only considering those cases with a living arrangement documented the percentage of children living with one or both pare nts for the nonlead agency cases becomes 91.6%. Table 1 4 Living Arrangements: Non Lead Agency Cases Frequency Percent Valid Percent Cumulative Percent Valid Living with one parent 11509 49.1 67.4 67.4 Living with non relative 67 .3 .4 67.8 Living with mother and father 4133 17.6 24.2 92.0 Living with relative 998 4.3 5.8 97.9 Other living arrangements 362 1.5 2.1 100.0 Total 17069 72.8 100.0 Missing 6362 27.2 Total 23431 100.0 The chi square test comparing the living arrangement var iable between the lead agency cases and the non lead agency cases was conducted for each type of living arrangement. Table 15 below shows the results of the chi square by each general category of living arrangement.
123 Table 15 Living Arrangement chi squar e values N X 2 D f Sig. (2 tailed) p < .0 1 Living with One Parent 22,561 2 9 99 1 000 Living with Non Relative 22,561 2 0 1 6 3 3 Living with Mother and Father 22,561 43. 7 8 1 .000 Living with Relative 22,561 4 8 1 491 Living Other 22,561 .1 4 1 .71 1 The living arrangement s of Living with Mother and Father and Living with One Parent are the arrangement s that are significantly different between the two groups. The lead agency study cases had the mother and father living arrangement for 13.4% of the cases (19.9% of the cases when only considering those cases where living arrangement was documented) and 17.6% of the nonlead agency cases were living with the mother and father (24.2% of the cases when only considering documented cases). For thi s specific living arrangement the mean for the lead agency study cases is .13 (range from 0 to 1) and the mean for the nonlead agency cases to be .18 (range from 0 to 1). The chi square X2(1, N = 22,561 = 43.78, p < .01, odds ratio = 1.27. The odds rati o indicates a strength of association and supports the finding of a significant difference regarding this living arrangement type. This indicates there are significantly fewer children with a living arrangement of Living with Mother and Father for the le ad agency study cases than those in the non lead agency cases. For the variable living with one parent the lead agency study cases had 48.2% of the cases with living arrangement documented and the non lead agency cases had 49.1% with this living arrangemen t. The chi square X2(1, N = 22,561) = 29.99, p < .01, odds ratio = 1.01. This indicates that the lead agency study cases had significantly fewer children with a living arrangement of living with one parent in comparison with the non lead agency cases. Howe ver, the odds
124 ratio indicates the association is extremely weak and thus a significant difference cannot be assumed regarding the living arrangement type living with one parent All of the other types of living arrangements were not significantly differe nt. Independent Variable 8, m altreatment type of abuse was constructed by consolidating the 48 maltreatment type options within the SACWSIS data set into five primary categories as explained earlier. Table 16 below shows the frequency of each general category of maltreatment type for the lead agency study cases. Table 16 Maltreatment Type : Lead Agency Study Cases Frequency Percent Valid Percent Valid Physical abuse 3120 38.4 54.0 Sexual abuse 370 4.5 6.4 Medical neglect 1448 17.8 25.1 Supervisio n neglect 550 6.8 9.5 General neglect 291 3.6 5.0 Total 5780 71.1 100.0 Missing 2353 28.9 Total 8133 100.0 This data field is completed by the Child Protection Investigator prior to the case being transferred to the subcontracting caseworker services agency. The missing data represents approximately 29% of the cases without a maltreatment type. It seems reasonable to assume that every child referred to services, thus transferred to a sub contracting caseworker services agency would have an ide ntified maltreatment type. Therefore the missing data is assumed to be due to the failure of the child protection investigator and/or caseworker to document the maltreatment type in the SACWSIS data system. The 2,353 cases with missing data will be dropped out of the final regression analysis as a result. A comparison of the cases retained in the model due to having all variables accounted for and those cases that are dropped due to missing cases is
125 completed later in the next chapter The Medical Neglect category represents a large segment of this variable as a result of the specific SACWSIS categories rolled into this primary category. As stated above th e Administration for Children and Families (ACF) 2007 Child Maltreatment Report (2009) 59.9% of children in the child welfare system were victims of neglect (Medical Neglect and Neglect categories). In the lead agency study cases only 28.2% of the cases reported neglect as the maltreatment type. Removing the missing data cases changes this percentage of n eglect cases to 39.6% for the lead agency study cases. The ACF does not delineate what the subcategories are under the neglect title so it may include types of abuse that are included in the physical abuse or sexual abuse categories listed above. The ACF reported 10.8% of children were victims of physical abuse compare to 38.4% (54% if the cases missing this variable are removed) of the cases reporting this maltreatment type in our study. The lead agency study cases appear to be dealing with much more physical abuse and much less neglect than the national norm. This difference may indicate an influence of the CBC model on which type of cases are actually accepted into the system for services and which are referred for services within the community. The AC F reports that 7.6% of children are sexually abused compared with 4.5% of the victims with this maltreatment type identified in the lead agency study cases (6.4% if the cases missing this variable are removed). Table 17 reflects the distribution of the maltreatment variable within the non lead agency cases. The missing data represents approximately 28% of the cases without a maltreatment type. Compared to the nearly 29% missing data from the lead agency study cases. For the same reasons as stated above t he missing data is assumed to be due to the
126 failure of the child protection investigator to document the maltreatment type in the SACWSIS data system. Table 17 Maltreatment Type : Non Lead Agency Cases Frequency Percent Valid Percent Physical abuse 9403 40.1 55.6 Sexual abuse 989 4.2 5.8 Medical neglect 3817 16.3 22.6 Supervision neglect 1691 7.2 10.0 General neglect 1018 4.3 6.0 Total 16923 72.2 100.0 Missing 6508 27.8 Total 23431 100.0 The consolidated neglect maltreatment type represents 27.8% of the nonlead agency cases compared with 28.2% of the lead agency study cases. Physical abuse as a maltreatment type represents 40.1% of the nonlead agency cases compared with the 38.4% in the lead agency study cases. 4.2% of these cases have a maltreatment type of sexual abuse while 4.5% of the study cases reported a maltreatment type of sexual abuse. The chi square test comparing the maltreatment type variable between the lead agency cases and the non lead agency cases was conducted for each general catego ry of maltreatment. Table 18 below shows the results of the chi square test by each type of maltreatment.
127 Table 18 Maltreatment Type chi square values Using a significance level of .01, t here is no significant difference found between the groups regarding the maltreatment types of sexual abuse physical abuse, and supervision neglect. However there is significant difference between the groups regarding two of the five maltreatment types. For the variable medical neglect, after removing the cases with missing data, the lead agency study cases had 25.1% documented with this maltreatment type and the nonlead agency cases had 22.6%. The chi square X2(1, N = 22,703) = 15.08, p < .01, odds ratio = 1.14. With such a large number of cases included in this analysis even small differences can be determined to be significant. However, the odds ratio indicates a very weak association and does not support a significant difference in the prevalence of medical neglect between the groups. Therefore, it is determined that the prevalence of the maltreatment type, medical neglect, is the same between the groups. For the variable general neglect, after removing the cases with missing data, the lead agency study cases had 5% documented with this maltreatment type and the nonlead agency cases had 6%. The chi square X2(1, N = 22,703) = 7.6, p < .01, odds ratio = 1.16. As stated above, with such a large number of cases included in this analysis even small differences can be determined to be significant when using the chi square analysis. However, the odds ratio indicates a very weak association and does not support a significant difference in the prevalence of general neglect between the groups. Therefore, N X 2 Df Sig. (2 tailed) p < .0 1 Physical Abuse 22,703 4.73 1 .037 Sexual Ab use 22,703 2.3 8 1 .123 Medical Neglect 22,703 15.08 1 .000 Supervision 22,703 1.10 1 .294 General Neglect 22,703 7.6 3 1 .006
128 it is determined that the prevalence of the maltreatment type, general neglect, is the same between the groups. In examining the chi square values independently for these variables there appears to be less general neglect and significantly more medical neglect in the lead agency study cases than in the non lead agency cases. However, when consider ing the very small differences in the prevalence percentages of these variables between the group it is unreasonable to assume the differences are significant. Independent Variable 9, substance abuse as a characteristic in a case wa s captured with the SAC WSIS data set through the child protection investigator documenting this as a reason for referral. The SACWSIS data system allows for up to five reasons for referral to be documented. For the purposes of this study the substance abuse characteristic was marked affirmative if substance abuse was given as reason for referral within any of these five opportunities, whether it was the first reason or the fifth reason or the first and third reason, etc. Table 19 below reflects the number of cases with one or more instances where substance abuse was indicated as a reason for referral within the five opportunities. Table 19 Substance Abuse as a Reason for Referral: Lead Agency Cases Frequency Percent Valid Percent Valid No Substance Abuse Indicated 4854 59. 7 60.4 Substance Abuse Indicated 3176 39.1 39.6 Total 8030 98.7 100.0 Missing 103 1.3 Total 8133 100.0 This data indicates that 8,030 cases completed the reason for referral information and of those cases 59.7% did not indicate a reason for ref erral as substance abuse. Table 19 above indicates 3,176 or 39% of the cases indicated substance abuse as a reason for referral. The 103 cases with missing data will be dropped from the final
129 analysis. A comparison of the cases retained in the model due to having all variables accounted for and those cases that are dropped due to missing cases is completed later in the next chapter. For the nonlead agency cases there were 31% of the cases with substance abuse identified as a reason for referral compared with 39% of the lead agency study cases. The chi square test comparing substance abuse as a reason for referral variable between the lead agency cases and the non lead agency cases indicates a significant difference of the prevalence of substance abuse between the two groups X2(1, N = 30,906) = 161.38, p < .01, odds ratio = 1.40. The odds ratio indicates strength of association and supports a significant difference in the prevalence of substance abuse between the groups. This indicates that there is a significant difference in the prevalence of substance abuse as a reason for referral between the two groups with the lead agency study cases having a significantly greater prevalence than the non lead agency cases. Independent Variable 10 domestic v iole nce violence was calculated in the same fashion as the independent variable substance abuse above. The child protection investigator or services caseworker identify reasons for referral to services for each case. The SACWSIS data system allows for up to f ive reasons for referral and the domestic violence variable is present if the case has domestic violence documented in anyone or several of the fiv e opportunities. Table 20 below indicates that of the 8,030 lead agency study cases reporting a reason for re ferral and within that group 1,746 or 21.7% reported domestic violence as a reason for referral to services. The 103 cases with missing data will be dropped from the final analysis. A comparison of the cases retained in the model
130 due to having all variable s accounted for and those cases that are dropped due to missing cases is completed later in the next chapter. Table 2 0 Domestic Violence: Lead Agency Study Cases Frequency Percent Valid Percent Valid No Domestic Violence Indicated 6284 77.3 78.3 Dom estic Violence Indicated 1746 21.5 21.7 Total 8030 98.7 100.0 Missing 103 1.3 Total 8133 100.0 As discussed in the previous chapter the 2009 report by the ACF indicates that nationally the average prevalence for domestic violence is just fewer t han 15%. The lead agency study cases have a greater prevalence of domestic violence documented than the national average as reported by ACF. For the nonlead agency cases there were 18.4% of the cases with domestic violence identified as a reason for refer ral compared with 21.5% of the lead agency study cases. The chi square test comparing domestic violence as a reason for referral variable between the lead agency cases and the non lead agency cases indicates a significant difference of the prevalence of d omestic violence between the two groups X2(1, N = 30,906) = 32.29, p < .01, odds ratio = 1.19. This indicates there is a significant difference in the prevalence of domestic violence as a reason for referral between the two groups with the lead agency study cases having a significantly greater prevalence than the non lead agency cases. However, the odds ratio indicates a very weak association and does not support a real difference in prevalence of domestic violence between the groups.
131 Therefore, it is determined that there is not a significant difference in the prevalence of domestic violence between the groups. Independent Variable 11 placement t ype represents the first placement for each child. If a child was not placed during the course of their case then it was determined that they were placed at home. The SACWSIS data set contains 49 possible placement type options for the investigator or caseworker to choose when the child is placed. These placement type options include several categories of similar placements as well as the full array of placement types offered to children in the system. For the purposes of this study and as described in the previous chapter these 49 options were consolidated into 11 primary categories of; relative, non relative, s helter, foster care, therapeutic foster care, group home, institution, Department of Juvenile Justice, development disabilities placement, runaway, and other. Table 21 below indicates the number and percent of cases in each of these primary categories. Additionally, 36.7% of the children in the lead agency study cases are placed at their own home. Table 21 below reveals that 36.7% of the children in the lead agency study cases were not given a placement type during the life of their case. This means that during the life of the case the child was never removed from their own home or there is a documentation error (i.e. placement type was not documented, or documented in error). This study cannot distinguish which is correct. According to the University of South Florida (USF) Florida Mental Health Institutes (FMHI) Evaluation of the Department of Children and Families CommunityBased Care Initiative Fiscal Year 2006 2007 Report to the Legislature (2008), in December of 2006 the five lead agencies having the cases
132 chosen for this study were serving a total of 14,080 children and of these 9,278 were in out of home care, meaning they were in a placement. This indicates that 66% of the children served at that moment in time had a placement type and that 34% of the children were at home. The USF/FMHI report captures the placement status for a moment in time (December 2006) in comparison to the lead agency study cases reporting the first placement during the life of the case. The USF/FMHI study also contains very few of the lead agency study cases as reported earlier, 94% of the study cases were closed by January 1, 2007 and the USF/FMHI study was a point in time snapshot in December of 2006. The USF/FMHI report provides a comparison point to support the assumption that 36.7% of the lead agency study cases were placed at home. Comparing the children not being removed within thirty days to those children that were potentially never removed during the life of the case indicates that approximately 26% of the children i n the lead agency study cases were removed from their primary caregiver during the life of their case after the first 30 days of the case. Within the first 30 days of the case 5,144 children of this data set were not removed. However, after the first 30 da ys and prior to the end of the case 2,156 children of those not removed were placed outside of their primary caregivers home. This indicates that approximately 42% of the children not removed within 30 days were eventually removed. This is an important ph enomenon in weighing the workload and considering the impact of removal at the time of case assignment. This is discussed further in the final chapter of this study. Examining the data presented in Table 21 yields the number of children actually placed t o be 5,145 (8,133 2,988). From this it is noteworthy to recognize that 52%
133 (2,700/5,145) of the children placed were placed with relatives and 28.4% (1,458/5,145) are placed in licensed foster care. Of the children placed nearly 9% (455/5,145) are placed with nonrelatives with only 5.1% placed in group homes or institutions. Table 2 1 Placement Type: Lead Agency Study Cases Frequency Percent Valid Percent Cumulative Percent Valid Home 2988 36.7 36.7 36.7 Relative 2700 33.2 33.2 69.9 Non relative 455 5.6 5.6 75.5 Shelter 208 2.6 2.6 78.1 Foster care 1458 17.9 17.9 96.0 Therapeutic foster care 7 .1 .1 96.1 Group home 125 1.5 1.5 97.6 Institution 149 1.8 1.8 99.5 DJJ 8 .1 .1 99.6 DD 8 .1 .1 99.7 Runaway 12 .1 .1 99.8 Other 15 .2 .2 100.0 Total 8133 100.0 100.0 Table 22 reflects the distribution of the placement type variable within the non lead agency cases. The missing data represents approximately 60.6% of the cases without a placement type. This indicates that 60.6% of the nonlead agency cases do not have a placement type for the life of the case. As discussed above the 36.7% of the lead agency study cases without a placement type seemed reasonable to assume that these children were at home as supported by the USF/ FMHI stu dy findings (2008). However, the 60.6% of cases without a placement type for the nonlead agency cases presents a much more inconsistent situation and increases the likelihood that there are data errors present in the
134 nonlead agency cases. According to the USF/FMHI report (2008) the 17 non lead agencies were serving 33,537 children in December 2006 and of these 19,536 were in out of home care, they had a placement type. This indicates that 42% of the nonlead agency children were at home when this study wa s completed With the table below indicating 60.6% it may be reasonable to assume the data is not correct. Table 2 2 Placement Type: Non Lead Agency Cases Frequency Percent Valid Percent Valid Home (Missing) 14199 60.6 60.6 Relative 4865 20.8 20.8 Non relative 711 3.0 3.0 Shelter 635 2.7 2.7 Foster care 2359 10.1 10.1 Therapeutic foster care 17 .1 .1 Group home 302 1.3 1.3 Institution 232 1.0 1.0 DJJ 21 .1 .1 DD 23 .1 .1 Runaway 22 .1 .1 Other 45 .2 .2 Total 23431 100.0 100.0 Examining the data presented in Table 22 yields the number of children actually with a placement type documented to be 9,232 (23,431 14,199). From this it is noteworthy to recognize that 52.7% of the children placed were placed with relatives (lead agency study cases had 52% in relative placement) and 25.8% are placed in licensed foster care (lead agency study cases had 28.4% in licensed foster care). Of the nonlead agency children with a placement type documented 7.7% are placed with non-
135 relatives (this was 9% for the lead agency study cases) with only 5.8% placed in group homes or institutions (5.1% for lead agency study cases). The chi square test comparing the placement type variable between the lead agency cases and the non lead agency cases was co nducted for each general category type of placement. Table 23 below shows the results of the chi square test by each type of placement. Table 2 3 Maltreatment Type chi square values N X2 D f Exact Sig. (2 tailed) p < .0 1 Placement Home 31,564 1385 .8 2 1 .000 Placement Relative 31,564 512. 3 1 .000 Placement Non Relative 31,564 111.22 1 .000 Placement Shelter 31,564 .54 1 .4 7 2 Placement Foster Care 31,564 350.7 9 1 .000 Placement Therapeutic FC 31,564 .14 1 647 Placement Group Home 31,564 2.78 1 1 05 Placement Institution 31,564 35.88 1 .000 Placement DJJ 31,564 .050 1 .8 3 2 Placement Developmental Disabilities 31,564 .000 1 1 00 Placement Runaway 31,564 1.61 1 .2 38 Placement Other 31,564 .018 1 9 9 Table 23 indicates there are 7 placement types that are not significantly different between the groups and five that are significantly different. As can be seen in Table 23 above, t he placement types that were significantly different between the groups consist of Home, X2(1, N = 31,564) = 1,385.82, p < .01, odds ratio = 1.58, Relative, X2(1, N = 31,564) = 512.3, p < .01, odds ratio = 1.66, NonRelative, X2(1, N = 31,564) = 111.22, p < .01, odds ratio = 1.92, Foster Care, X2(1, N = 31,564) = 350.79, p < .01, odds ratio = 1.85, and Institutions, X2(1, N = 31,564) = 35.88, p < .01, odds ratio = 1.92 include the majority of cases, 94% of the lead agency study cases and 97% of the nonlead agency
136 cases. The odds ratio for each significant placement type indicates a strength of association and support s the significant difference in prevalence of these placement types between the groups. It is important to note the discussion above regarding the potential errors in the nonlead agency data for this variable. This analysis indicates that the non lead agency cases have a significantly larger number of children placed in their own home than the lead agency study cases. The lead agency study cases have significantly more children placed in relative care, nonrelative care, foster care, and institutions. The issue of a large portion of missing data that cannot be collaborated to indicate a placement at home with the nonlead agency cases regarding this characteristic may be an indication of the difference in the intensity of the documentation requirements betw een the lead agency projects and the nonlead agency projects in 2003 as discussed above. If the data is flawed as suspected the real difference between these groups regarding placement types is unable to be determined. Independent Variable 12, prior r em oval was constructed by comparing removal dates associated with the case and the case begin date. In looking at the SACWSIS data set made available for this study this is the best indicator that the case has been active in the past and specific to the sam e child that is included in this study. Prior abuse reports were identified by family and the data was not accessible for this study. Additionally, prior placements were considered as a possible source to determine previous involvement or a history of abus e however, placements were not documented in the client information system. As shown in Table 24 below .8% of the lead agency study cases have a documented prior removal.
137 Table 2 4 Prior Removal: Lead Agency Study Cases Frequency Percent Valid Perce nt Valid No Prior Removal 8065 99.2 99.2 Prior Removal 68 .8 .8 Total 8133 100.0 100.0 The lead agency study cases include 68 children or .8% of the cases with the characteristic of prior removal. The nonlead agency cases include 167 children or .7 % of the cases with the characteristic of prior removal. The chi square test comparing the prior removal variable between the lead agency cases and the non lead agency cases indicates there is not a significant difference of the prevalence of prior removal between the two groups X2(1, N = 31,564) = 1.24. This indicates there is a no significant difference in the prevalence of prior removal between the two groups. The Dependent Variable, caseworker t ime in this study is the recorded time by caseworkers in hours aggregated for each case. As has been stated above, the caseworkers with either the sub contracted caseworker services organizations or with the DCF were required to enter their activities into the SACWSIS data system. Each activity was to have a beg inning time and an ending time. For the purposes of this study a complete activity is one that has both a beginning time and an ending time and the activity was less than 8 hours in length. Any activity longer than 8 hours was assumed to be an error in data entry as the list of activities conducted did not include any activity that would reasonably take longer than 8 hours. The SACWSIS data set contained 53 activity options for caseworkers and investigators to choose from when recording their work. For the purposes of this study these 53 options were consolidated into 7 general categories of activities (Telephone
138 Contacts, Face to Face Visits, Filed Visits with Collaterals, Family Visitation, Staffing Events, Note to File, and Case Reviews) as described in the previous chapter. The time is captured for any caseworker or supervisor assigned to the case and recording activities on the case in the SACWSIS data set. This study does not differentiate between caseworkers nor does it account for how many differe nt caseworkers actually worked and recorded activities on a particular case. As stated earlier, the cases within this study are complete cases and the recorded caseworker time associated with the case from the begin date of the case to the closure of the case is captured regardless of who documented the activity. The time for the lead agency cases with complete activities is shown in Table 25. This table indicates there are 150 cases with no recorded time for the complete activities (2%) ; the mean number of hours is 43.6 per case with a median at one half of the mean, 21.3 hours and a mode of 4.5 hours, with a standard deviation of 69.9 hours. This clearly indicates a very positively skewed distribution (skewness = 6.925, standard error .027) with a large majority (71%) of the cases having fewer hours than the mean. The Kurtosis of this distribution is also significant (kurtosis = 101.954, standard error .055). According to Montbalm and Royse (2003) each of these measures ought to be within a multiple of 2 of their respective standard error values to render them insignificant, this is clearly not even remotely the case.
139 Table 25 Caseworker Time: Lead Agency Cases N Valid 7983 Missing 150 Mean 43.6078 Std. Error of Mean .78271 Median 21.3333 Mode 4.5 0 Std. Deviation 69.93345 Skewness 6.925 Std. Error of Skewness .027 Kurtosis 101.954 Std. Error of Kurtosis .055 Minimum .00 Maximum 1898.90 The histogram below provides a visual portrayal of the skewness of the dependent variable. Figure 5. Ti me for Complete Activities of 8 Hours or less To correct for this skewness and kurtosis the variable will be transformed using a Log10 SPSS transformation procedure. This procedure is to reduce the nonnormality of the variable by changing the gaps between the data values. The data remains as an ordinal
140 level of measurement and retains the sequential placement of values (least number of hours to most number of hours) as the original data. J. W. Osborne (2008) recommends this type of transformation for posi tively skewed data. Transformation is supported when the data is skewed and not normally distributed (Ferketich & Verran, 1994). Further, Tabachnick & Fidell (1989) go so far as to recommend transforming unless there is a substantial reason not to do so. T he histogram below shows the dependent variable after the log 10 transformation. Figure 6. Dependent Variable2 The above Histogram of the transformed data shows a very normal distribution of the data and indicates the range of the data to be 0.00 to 3.28. These extreme points represent a minimal number of cases. This will allow for the study to meet the basic assumption of normality. Of those lead agency study cases with recorded activities the average number of activities per case came to be 41.3 with the average length of stay being 474 days for all of the 8,133 lead agency study cases, yielding an average of one activity recorded about
141 every 11 days per case. Of those nonlead agency cases with recorded complete activities the average number of activi ties per case came to be 36 with the average length of stay being 392 days for all 23,431 nonlead agency cases, yielding an average of about one activity recorded every 11 days per case. Table 26 below is a summary of the number of activities by type for the lead agency study cases. Table 26 Number of Complete Activities by Type: Lead Agency Study Cases N Minimu m Maximum Sum of Activities Percent of Total Telephone Contacts 3416 1.00 408 55,975 16.9% Face to Face Visits with Child 7954 1.00 158 166,1 03 50.3% Field Visits with Collaterals 3961 1.00 123 22,108 6.7% Family Visitation 1291 1.00 79 7,258 2.2% Staffing Events 2509 1.00 27 6,187 1.9% Note to File 3893 1.00 590 63,518 19.2% Case Reviews 2553 1.00 45 9,162 2.8% Total Cases with a Minimum of One Complete Activity 7983 1.00 979 330,311 100% Table 27 below shows the time (hours) by generated by these activities for the lead agency study cases. This table indicates that nearly 53% of the recorded caseworker time was spent on face to face visits with the child. Additionally, 20% was spent documenting case activities through the note to file general activity, and 11% was spent making contacts with collaterals in the field. Caseworkers recorded only 7% of their time on telephone cal ls and about 5% of their time on family visitation. Approximately 2% is recorded for staffing events and only 2% is recorded for case reviews. In the national survey conducted by AFSMCE in 1998 the caseworkers complained about the amount of
142 time they spen t on paperwork and documentation. It could be that the Florida caseworkers did not routinely document the time they spent documenting records but clearly what is recorded in the SACWSIS data set does not support an inordinate amount of time on documentation. In fact it supports a vast majority of time spent with the child and family. Additionally the focus groups did not emphasize documentation as an extraordinary time consuming task. It is noteworthy to mention that the Florida caseworkers were not using this SACWSIS data set as a time sheet but to record specific activities conducted for the case and document the time that activity took. In other words the caseworkers may not have recorded the time they spent on paperwork although there were activity opt ions for documentation. Table 27 Time in Hours by Activity Type: Lead Agency Study Cases N umber of cases with Recorded Activity Sum of Hours Recorded for Activity Percent of Total Hours Telephone Calls 3416 24,577 7% Face to Face Visits 7954 183,628 53% Filed Visits with Collaterals 3961 38,929 11% Family Visitation 1291 17,243 5% Staffing Events 2509 5,605 2% Note To File 3893 70,213 20% Case Reviews 2553 7,273 2% Total Hours 347,468 100% Time recorded has been aggregated by case to include all complete activities and is reported in hours. As noted above the lead agency study cases had 150 cases missing the recorded time variable. This represents 2% of the lead agency study cases. The non lead agency cases had 9,796 cases missing the time recorded variable representing 42% of the total nonlead agency cases. The reason there are any cases missing caseworker
143 time documentation is due to the absence of recorded activities associated with the case. As described earlier activities needed to have a begin time and an end time as well as having a duration of less than 8 hours to be included in the calculation of the dependent variable. Therefore, some cases could very well have no activities meeting these standards and thus have no time recorded for the dependent variable. An independent samples t test was used to compare the dependent variable between the lead agency study cases and the non lead agency cases. For the lead agency study cases the mean was 43.6 hours compared to 34.6 hours for the nonl ead agency cases. The standard deviation for the lead agency study cases was 69.9 hours, while the standard deviation for the nonlead agency cases was 46.5 hours. The t test results indicate a significant difference in the caseworker time between the two groups. Levenes test for equality of variances indicated there is a lot of distortion in the data, the variance s in the two groups w ere significant ly different, p < .01. Therefore, assuming the variances are not equal, t (1 2,166 ) = 10.24, p < .01. This in dicates a significant difference in the average caseworker time per case between the two groups, with the lead agency study cases having a significantly larger amount of average time documented for the cases. Cohens d was calculated resulting in an effect size of .159. According to Valentine & Cooper (2003) an effect size of .2 is small, .5 is medium, and .8 is large. Therefore the effect size of the difference between these two independent groups regarding caseworker time is small and does not support a significant difference between the groups regarding caseworker time. Ther e fore, it is determined that there is not a significant difference in the dependent variable between the two groups.
144 Analytical Phase III Bivariate relationships between each independent variable and the dependent variable. The next step is to look at the strength of the bivariate relationship for each individual independent variable with the dependent variable. Table 28 below shows the bivariate relationship between each independent variable and associated subcategory of each variable and the dependent variable. As explained above each independent variable except for age is a categorical variable. Each independent variable and subcategory was constructed to indicate the presence or absence of the variable for each case. Each independent variable and subcategory, except for age is coded 0 for the absence of the variable and 1 for the presence of the variable. Table 28 below shows the specific bivariate relationship between each independent variable and subcategory to that variable with the dependent variable.
145 Table 28 Bivariate Relationship between Independent and the Dependent Variable s Variable N Pearsons R Sig. (2 tailed) P < .01 1. Age 7983 .071 .000 2.Male 7983 .005 .637 3 .Removal 7983 .145 .000 4 .a.Black 7983 .039 .001 4 .b.White 7983 .078 .000 4 .c.Hispanic 7983 .098 .000 5 .Developmental Dis. 7983 .065 .000 6 .a.Living 1 Parent 5370 .130 .000 6 .b.Living Non Rel 5370 .022 .103 6 .c.Living 2 Parent 5370 .107 .000 6 .d.Living with Rel 5370 .086 .000 6 .e.Living Other 5370 .039 .004 7 .a.Physical Abuse 5720 .02 .127 7 .b.Sexual Abuse 5720 .039 .003 7 .c.Medical Neglect 5720 .034 .010 7 .d.Supervision 5720 .03 .022 7 .e.General Neglect 5720 .031 .021 8 .S ubstance Abuse 7883 .04 .000 9 .Domestic Violence 7883 .077 .000 1 0 .a.Placement Home 7983 .377 .000 1 0 .b.Placement Rel 7983 .096 .000 1 0 .c.Placement Non Rel 7983 .07 .000 1 0 .d.Placement Shelter 7983 .071 .000 1 0 .d. Placement FC 7983 .232 .000 1 0 .e. Placement Therapeutic FC 7983 .001 .921 1 0 .f. Placement GH 7983 .059 .000 1 0 .g. Placement Inst. 7983 .077 .000 1 0 .h. Placement DJJ 7983 .005 .685 1 0 .i. Placement DD 7983 .000 .973 1 0 .j. Placement Runaway 7983 .007 .509 1 0 .k. Placement Other 7983 003 .779 1 1 .Prior Removal 7983 .064 .000 Table 28 reveals that gender (male), living arrangement nonrelative maltreatment type physical abuse, maltreatment supervision, maltreatment general neglect, and the placement types of therapeutic foster care, Department of Juvenile Justice, developmental disabilities, runaway, and other are not significant variables in effecting the dependent variable caseworker time. So of the 32 independent variables and subcategories 10 or 31% do not have a significant bivariate relationship with the dependent variable. Of the remaining 22 independent variables 14 have a significant and
146 positive bivariate relationship with the dependent variable, caseworker time. This equates to 44% of the independent variables and subcategories significantly effecting more caseworker time in a bivariate relationship As can seen in Table 28 above these significant characteristics with a positive bivariate relationship include ; if the child was removed within 30 days of the case begin date, if the child is black or Hispanic, if the child has a developmental disability, living with one parent, living with other, when the maltreatment type is sexual abuse, when the placement is in foster care, with a relative, with a nonrelative, shelter, group home, or institution, and when there is a prior removal. The mental health condition of the child variable was removed from further analysis as discussed earlier. Of the 2 2 independent variables and subcategories that have a significant bivariate relatio nship with caseworker time, 8 have a negative bivariate relationship as indicated by the negative Pearsons R This equates to 25 % of the independent variables and subcategories effecting less caseworker time when they are a characteristic that is present. The negative relationship indicates that when this characteristic is present the caseworker spends less time on the case. Under race/ethnicity children who are white have a negative bivariate relationship with caseworker time indicating they receive less time. T he living arrangements of living with mother and father and living with a relative also have a negative bivariate relationship with caseworker time, indicating that these living arrangements reduce the time caseworkers spend on the case. These liv ing arrangements may or may not be at the time of case initiation. This variable represents the last living arrangement documented in the case record. Some of these children may have experienced a placement during the course of their case. The SACWSIS data set
147 does not have a mechanism to determine if this was the only living arrangement during the course of the case. Looking at the maltreatment type variables Medical neglect also has a negative bivariate relationship with caseworker time while sexual abu se, supervision and general neglect all have positive bivariate relationships, (physical abuse did not have a significant bivariate relationship). Both substance abuse and domestic violence have a negative bivariate relationship with caseworker time. Ana lytical Phase IV Multivariate regression and stepwise analysis. The first regression analysis equation includes 26 independent variables and sub categories. The independent variables are gender, age, race/ethnicity, domestic violence, substance abuse, developmental disabilities, placement type, maltreatment type, living condition, prior removal and removal. The adjusted R2 for this model (referred to as the all variables Model for the purposes of this study) is .22 (N = 3,855) indicating this model explains 22% of the variance within the dependent variable. The model is significant in explaining the variance in the dependent variable, p < .01. The number of cases included in the analysis (n) is 3,855. This is a result of the cases that do not account for al l of the variables (have missing data for one or more variables as discussed earlier) are not included in the model. Such a significant drop from 8,133 cases that were qualified for the study to 3,855 cases included in the final model analysis can be a concern. The reduction in the cases included in the regression analysis is directly related to the independent variables and dependent variable with missing data. The independent variables with missing data
148 include; substance abuse (103), domestic violence (103), maltreatment type (2,353), and living arrangement (2,641). Additionally, the dependent variable caseworker time, had 150 cases with missing data. Therefore, a comparative analysis was constructed to determine if the cases in the model were significantly different on key demographic indicators (race, gender, age) than those cases excluded due to missing data. An independent samples t test was conducted to determine the relationship b etween the groups regarding age. The average age for the cases incl uded in the analysis was 6.3 years and the average age for those cases excluded was 6.4 years. The Levenes test for equality of variances was significant indicating the variances are significantly different. It was found that there is not a significant di fference between the groups regarding age, assuming the variances are not equal, t (8,129) = .842. The cases included in the analysis are made up of 31.2% black, 62.2% white, and 6.1% Hispanic children. The cases excluded from the analysis include 34.3% black, 59.6% white, and 5.7% Hispanic children. While a chi square test indicates there are significantly more black children in the excluded group, X2(1, N = 8,133) = 8.88, p < .01, odds ratio = 1.14 there is only a slight difference between 31.2% and 34.3%. The odds ratio indicates there is a very weak association and does not support a significant difference in the prevalence of black children between the groups. The chi square test for white children and Hispanic children did not indicate a significant difference between the groups regarding these race/ethnicity characteristics. Therefore, it is determined that there is not a significant difference between the groups regarding the characteristic of race/ethnicity. Also calculating a chi square for gender between the two groups yields that the groups are not significantly different regarding gender, X2(1, N = 8,133) = 3.03.
149 Therefore, it is determined that these two groups are not significantly different and the results of the regression analysis are generalizable to all of the lead agency study cases. Table 29 below contains the results of the regr ession analysis for the all variables model.
150 Table 29 Regression all variables Model: Lead Agency Study Cases Model Standardized Coefficients T Sig. p < .01 C ollinearity Statistics Beta Tolerance VIF 1 (Constant) 54.701 .000 Child's gender .000 .012 .990 .979 1.021 Black .063 4.253 .000 .909 1.100 Hispanic .096 6.584 .000 .954 1.048 Developmental Disability .007 .461 .645 .937 1.067 Age .023 1.557 .120 .900 1.112 Domestic Violence .026 1.741 .082 .878 1.138 Substance Abuse .065 4.158 .000 .839 1.192 Placement Foster Care .383 23.373 .000 .752 1.329 Placement Relative .401 24.207 .000 .737 1.357 Placement Non Relative .157 10.560 .000 .913 1.095 Placement Shelter .130 8.841 .000 .932 1.074 Placement Group Home .119 8.166 .000 .953 1.050 Placement Institution .109 7.360 .000 .927 1.079 Placement DJJ .005 .321 .748 .984 1.017 Placement Runaway .033 2.292 .022 .992 1.008 P lacement Other .041 2.876 .004 .994 1.006 Sexual Abuse .020 1.376 .169 .926 1.080 Medical Neglect .013 .786 .432 .784 1.276 Supervision Neglect .012 .821 .412 .900 1.111 General Neglect .012 .832 .405 .939 1.065 Living Non Relative .025 1.745 .081 .995 1.005 Living Mother and Father .040 2.700 .007 .927 1.079 Living Relative .064 4.391 .000 .965 1.036 Living Other .001 .062 .951 .960 1.041 Prior Removal .026 1.787 .074 .988 1.012 Removal within 30 Days .151 9.544 .000 .811 1.23 4 This all variables Model includes all of the variables as discussed in earlier chapters and sections of this study. As can be seen in Table 29 above race is significant
151 with both subcategories being significant (Black p < .01 and Hispanic p < .01) S ubstance Abuse ( p < .01) also is significant but ha s a negative relationship with caseworker time as indicated by the negative beta coefficient. The placement types of therapeutic foster care and developmental disabilities were dropped from this equation due to neither having a presence in any of the remaining cases. This means that of the 3,855 cases that accounted for all of the independent variables none of them had either of these two placement types. Of the placement types that were present in the cases included in the analysis o nly the placement types of runaway and DJJ are not significant. This is not consistent with the focus groups who thought these placement cha racteristics would significantly increase a caseworkers time spent on a case. The rem aining placement types are significant ( p < .01) and cause more time to be spent by the caseworker when they are present. Not a single maltreatment type is significant in effecting the time a caseworker spends on a case. Again the focus groups expected sexual abuse cases to take more time. Living arrangements of mother and father ( p < .01) and relative ( p < .01) are both significant with a negative relationship to caseworker time indicating that these living arrangements result in less time spent by the caseworker. This result appears logical due to placements with families and relatives would indicate a much more stable and permanent situation. Prior removal is not significant in this model. Again, this is not consistent with what the focus groups indicated. They believed that they spent more time on cases with a prior removal. Finally, removal is significant ( p < .01) but also has a negative coefficient indicating a caseworker spends less time on the case when this characteristic is present. Also, Ta ble 29 provides an analysis of col linearity. As can be
152 seen the tolerance levels range between .737 and .995, according OBrien (2007) a tolerance level of less than .2 indicates a multicollinearity problem. Similarly the VIF values range from 1.005 to 1.357, again according to OBrien (2007) a VIF value above 5 indicates a multicollinearity problem. Additionally, the condition index does not approach 30 with a high value under 10. Therefore, there is no multicollinearity and the null hypothesis can be reje cted The stepwise regression analysis for the all variables model yields a ranking of the variables that explain the most variance. Table 30 below lists the 13 variables that explain unique portions of the variance in caseworker time by order of their im pact.
153 Table 3 0 Stepwise Regression for all variables Model Stepwise all variables Regression Model Standardized Coefficients Adjusted R2 T Sig. Beta 1 (Constant) 158.270 .000 Placement Foster Care .224 .050 14.280 .000 2 ADD Placement Re lative .292 .129 18.739 .000 3 ADD Placement Non Relative .119 .143 7.866 .000 4 ADD Placement Shelter .116 .156 7.773 .000 5 ADD Removal within 30 Days .126 .169 7.922 .000 6 ADD Placement Group Home .117 .182 7.926 .000 7 A DD Placement Institution .116 .195 7.934 .000 8 ADD Hispanic .090 .203 6.251 .000 9 ADD Substance Abuse .082 .209 5.668 .000 10 ADD Black .068 .214 4.653 .000 11 ADD Living with Relative .053 .216 3.687 .000 12 ADD Living with Mother and Father .042 .218 2.897 .004 13 ADD Placement Other .041 .219 2.846 .004
154 The independent variable that explains the largest amount of variation of caseworker time is the placement in foster care. This indi cates that caseworkers spend more time with children in foster care and this characteristic explains the most unique variance in the dependent variable of all the variables. The second independent variable explaining nearly an equal amount of variation in the depende nt variable is the placement type relative. When combined with the first variable (placement type foster care) these variables explain 13% of the unique variance in the dependent variable. The other variables that explain the variance within the dependent variable include placement with a nonrelative, placement in a shelter, removal within 30 days of the begin date (negative relationship) placement in a group home, placement in an institution, Hispanic, substance abuse (negative relationship), black, living wit h a relative (negative relationship) living with mother and father (negative relationship), placement other, and placement runaway. Of the 14 variables that explain unique portions of the variance within the dependent variable four of them have a negative relationship indicating the caseworkers spends less time on the case when these characteristics are present. The several variables in the all variables model changed from being significant in the bivariate relationship to not being significant in the regression model including the following independent variable s; age, developmental disabilities, living arrangement other, maltreatment type sexual abuse, domestic violence, and prior removal. This indicates that these variables did have a significant influence on caseworkers time in a bivariate analysis but when included in the model the variance these variables explained is now covered by other variables in the model. Only the independent variable placement
155 other changed from being insignificant in a bivariate relationship to being significant in the regression model. The second regression equation includes 25 independent variables and subcategories. The independent variables are gender, race/ethnicity, age, domestic violence, substance abuse, devel opmental disabilities, maltreatment type, living arrangement, prior removal, and placement. For the purposes of this study this equation is referred to as the Placement Model. The adjusted R2 for the Placement Model is 203 (N = 3,855), indicating that th is model explains 20.3% of the variance in the dependent variable. Two independent variables were dropped from the equation by the SPSS due to them being a constant or having no beta value. These are Therapeutic Foster Care Placement (N = 7) and Developmen tal Disabilities Placement (N = 8). These placement types were active on very few cases and likely their cases were removed due to the 4,278 cases with missing data being dropped in the model. The mo del is significant in explaining the variance in the depe ndent variable p < .01. As explained above the model dropping 4,278 cases from the originally qualified 8,133 cases is due to missing data. An analysis of the two groups (included cases and dropped cases) determined there is no significant difference between the groups. Table 3 1 below shows the results of the regression analysis.
156 Table 3 1 Placement Regression Model Placement Model Standardized Coefficients T Sig. Collinearity Statistics Beta Tolerance VIF 1 (Constant) 53.992 .000 Child's Gender .000 .005 .996 .979 1.021 Black .066 4.389 .000 .909 1.100 Hispanic .096 6.496 .000 .954 1.048 Developmental Disability .005 .314 .754 .937 1.067 Age .023 1.491 .136 .900 1.112 Domestic Violence .023 1.484 .138 .879 1.138 Substance Abuse .0 61 3.870 .000 .839 1.191 Placement Non Rel ative .132 8.895 .000 .943 1.060 Placement Shelter .114 7.684 .000 .945 1.059 Placement Group Home .104 7.129 .000 .963 1.038 Placement Institution .093 6.244 .000 .939 1.065 Placement DJJ .000 .030 .976 .985 1.016 Placement Runaway .030 2.096 .036 .992 1.008 Placement Other .041 2.875 .004 .994 1.006 Prior Removal .033 2.278 .023 .991 1.009 Living Non Relative .026 1.832 .067 .995 1.005 Living Mother and Father .040 2.706 .007 .927 1.079 Living Relative .073 4.968 .000 .969 1.032 Living Other .010 .698 .485 .965 1.037 Sexual Abuse .024 1.595 .111 .927 1.079 Medical Neglect .016 1.005 .315 .784 1.275 Supervision Neglect .005 .347 .728 .902 1.108 General Neglect .015 1.038 .299 .940 1.064 Placement Foster Care .333 21.205 .000 .837 1.195 Placement Relative .349 22.048 .000 .828 1.208
157 The Placement Model explains slightly less of the variance in the dependent variable without the removal variable. As can be seen in Tabl e 31 the Placement M odel has the following significant independent variables; Black ( p < .01) Hispanic ( p < .01) Substance Abuse ( p < .01) Living with mother and father ( p < .01) living with a relative ( p < .01) and the placement types of foster care ( p < .01) relative ( p < .01) nonrelative ( p < .01) shelter ( p < .01) group home ( p < .01) institution ( p < .01) and other ( p < .01) These significant independent variables include both positive and negative relationships with the dependent variabl e. Substance Abuse, living arrangement with mother and father, and living arrangement with a relative had a negative relationship with the dependent variable and indicate less time was spent by the caseworker on a case with these characteristics. The sign ificant variables for race and placement had a positive relationship with the dependent variable and when present resulted in more time being spent on the case by the caseworker. The independent variables that are significant in the Placement Model as well as the all variables Model are Black and Hispanic, substance abuse, living with a relative living with mother and father and the placement types; foster care, relative, non relative, shelter, group home, institution, and other Of these significant independent variables the directional relationship remains the same in both models with substance abuse, living arrangement with a relative and living arrangement with mother and father having a negative direction indicating less caseworker time if this charac teristic is present. The independent variables Black, Hispanic, and placement types that are significant show a positive direction indicating that the caseworkers spend more time on the case when these characteristics are present.
158 The following variables were not significant in either the all variables or the Placement models presented; gender, developmental disabilities, age, domestic violence, sexual abuse, medical neglect, general neglect, supervision, living with a nonrelative, living other prior re moval, and the placement types DJJ and runaway The Placement Model includes 9 placement independent variables with 7 of these options being significant in explaining the variance of the dependent variable. The only placement variables that are not significant are placement in a DJJ facility and placement as a runaway All of the significant placement independent variables show a positive direction with the dependent variable indicating that if these characteristics are present a caseworker spends more t ime on the case. Of these placement independent variables placement other w as not significant in the bivariate relationship but is significant in the Placement Model. Placement types DJJ, runaway, developmental disabilities, and therapeutic foster care were not significant in the bivariate relationship or in the Placement Model. None of the maltreatment type independent variables were significant in the Placement Model indicating the type of abuse did not influence the time a caseworker spent on the case. Also, Table 31 provides an analysis of collinearity. As can be seen the tolerance levels range between .784 and .995, according OBrien (2007) a tolerance level of less than .2 indicates a multicollinearity problem. Similarly the VIF values range from 1.0 05 to 1.275, again according to OBrien (2007) a VIF value above 5 indicates a multicollinearity problem. Additionally, the condition index does not approach 30 with a high value under 10. Therefore, there is no multicollinearity and the null hypothesis ca n be rejected.
159 The step wise regression analysis for the Placemen t Model is shown in Table 32 below. This analysis ranks the 12 independent variables that explain unique amounts of variance in the dependent variable within the Placement Model. Table 32 Ste pwise Regression Placement Model Placement Model Standardized Coefficients Adjusted R 2 t Sig. Beta 1 Placement Foster Care .224 .050 14.280 .000 2 ADD Placement Relative .292 .129 18.739 .000 3 ADD Placement Non Relative .119 .14 3 7.866 .000 4 ADD Placement Shelter .116 .156 7.773 .000 5 ADD Placement Group Home .104 .166 6.992 .000 6 ADD Placement Institution .099 .176 6.723 .000 7 ADD Hispanic .090 .183 6.170 .000 8 ADD Black .082 .190 5.573 .0 00 9 ADD Substance Abuse .066 .194 4.480 .000 10 ADD Living with Relative .063 .197 4.321 .000 11 ADD Living Mother and Father .044 .199 2.942 .003 12 ADD Placement Other .041 .201 2.830 .005 As seen in Table 32 above utilizing a stepwise regression model indicates that each placement type that has a significant relationship with the dependent variable
160 contributes a unique portion of the overall explanation of variance in the dependent variable. The first six independent variables in the stepwise model are all placement types and explain 87% of the variance in the dependent variable that is explained by the Placement Model. All of the placement independent variables have a positive direction in the relationship with the d ependent variable indicating that when present the caseworker spent more time on the case. The other significant independent variables include substance abuse, Hispanic, Black, living with a relative, and living with mother and father. T he independent vari able s substance abuse, living with a relative and living with mother and father show a negative direction wi th the dependent variable and these relationships have been discussed earlier The following chapter provides a thorough discussion of these resul ts, some recommendations for supervisors in assigning cases, and recommendations for further study.
161 C hapter V D iscussion In this chapter each of the research questions are answered with specific findings and a thorough discussion of these findings. Spe cific study limitations are discussed and then a list of practice, policy, and research recommendations are made based upon the findings of this study. The first research question posed was; what specific case characteristics and job tasks do current cas eworkers and supervisors identify that causes more time to be devoted to a specific case? FINDING I Although there are some characteristics identified by the focus groups not present in the data set and some characteristics identified by the focus groups that were not collectible for this study, the independent variables included in the model represented a majority (80%) of the characteristics identified by the focus groups that were available and usable within the SACWSIS data set, and therefore present s a strong model supported by the focus group findings. The focus groups were very informative in this research project. Between the supervisors and the caseworkers the focus groups identified 39 case characteristics that from their experience would add tim e spent by a caseworker on a case. After evaluating the characteristics identified by the focus groups it was found that 10 were simply not in the SACWSIS data set ( caregiver mental health, delinquency history of children, children
162 on psychotropic medicati ons, high profile cases, uncooperative parents, nonEnglish speaking parents, children with dual diagnosis, immigration issues, expedited TPR, and criminal charges against the caregiver/parent ) and four were of a nature that the data set did not have a cl ear and consistent field where this information was collected (sibling groups, placement of sibling groups, history of abuse, and parents who were former foster children). Eliminating these 1 4 characteristics, the focus groups identified 2 5 potential characteristics with the supervisors identifying all but one of the characteristics the caseworkers identified. The independent variables selected for the study included 20 of these 2 5 characteristics or 8 0%. While this percentage of characteristics contained w ithin the study allows a strong claim that this study did capture the core characteristics that may drive time spent by caseworkers on a particular case it does leave the question of what could be revealed if the additional 19 characteristics identified by the focus groups were included. This finding emphasizes the need for further improvement in the data captured by the SACWSIS data set and the level of documentation by caseworkers within the data set. In addition to better documentation and more data ele ments the SACWSIS data set needs to more clearly define and link the data For example, connecting information on sibling groups within the data set ought to be a very simple process and defining the context and timeliness of a living arrangement ought to be imbedded in the system These improvements do not seem to be overwhelming and would move the SACWSIS data set to a greater granularity allowing for more refined research and information for the field. Comparing the case characteristics and the activiti es utilized within this study with other caseworker time studies further demonstrates the strength of this study in
163 capturing the common elements as well as the uniqueness of this study. As discussed in Chapter II above caseworker time studies that have b een completed typically had the caseworker track their activities for a short period of time such as 2 weeks T hese studies include the Summit Ohio study, (Knox & Higgins, 2003 relating to child protection investigators only) New York study (Walter R. Mc Donald and Associates and American Humane, 2006) the California Workload Study (Walter R. McDonald and Associates and American Humane, 2000) and the Florida Caseworker Time study (Perry & Murphy, 2008) These time studies looked at the amount of time the caseworker spent on specific activities The Florida study used unconventional terms not used in the other stud ies such as in home services, out of home services, and multi tasking but the study did report on client contact. Similarly the California study did specify the activities only reporting on whether it was a case related activity or a non case related activity. Table 33 below shows the activities captured by these studies as well as a couple of other studies that did not just capture caseworker tim e and compares these activities to those included in this study (identified as the Card study). As can been seen in T able 33 visiting the child/family and documentation are the common activities for all 6 studies The only activities captured in the other time studies and not included in this study are transportation/travel, research/risk assessment, and administrative activities. As has been discussed the absence of transportation/travel is a limitation of this study and specifically identifying this acti vity either as part of or included with the other activities or as a separate activity is a very real need of the SACWSIS data set. The research/risk assessment activities are likely included in the Face to Face activities included in this study. However t he Summit Ohio study identified this activity separately so it is addressed separately here.
164 Table 33 Comparison of Activities Used in Caseworker Time Studies Caseload Study Time Variables Comparison Card Study Summit County Florida Mid West New York P ennsylvania Telephone Calls (7%) Communication with Child & Family (7%) Face To Face Visits (53%) Home Visit (20.5%) Cli ent Contact (28.7%) Visitation Time Face to Face Children and Families (17%) Family Visits (46%) Field Visits (11%) Collateral Co ntacts (9%) Case Related Activities (20.5%) Family Visitation (5%) Staffing Events (2%) Case Supportive Activities (7%) Note To File (20%) Paperwork (17%) Paperwork Documentation (31%) Casework Processing (54%) Case Reviews (2%) Court Related Activities (6.5%) Travel Travel (11%) Research and Risk Assessment (13.6%) Other Duties (39.6%) Administrative The Mid West Not For Profit Agency study ( Mills & Ivory, 1991) looked at the case characteristics of maltreatmen t type and location (placement) of the child and assigned severity ratings to the cases. This study and the Florida study were the only studies that specifically addressed case characteristics with any similarity to this study. The Lutheran Services of Washington study (Katz 1990) studied the impact on the time to reach permanency when caseloads were small and control l ed as well as additional services were provided.
165 This comparison of other studies within the literature clearly demonstrates the uniquene ss of this study. The other studies always looked at how the caseworker spent their time while this study considers the case specific characteristics that drive the caseworkers time. Even those studies that looked at case characteristics it was more alig ned with defining what cases caseworkers spent their time on but not whether the case characteristics caused the caseworkers to spend more or less time on the case. In addition to demonstrating a strong level of consistency with other caseworker time st udies reg arding caseworker tasks Table 33 also demonstrates that the caseworkers with in this study spent 53% of their recorded time on face to face visits with the children and families. The literature and to some extent the other time studies indicate th at caseworkers complain about the amount of paperwork and documentation requirements. This study does not seem to support that documentation represents an overwhelming amount of caseworker time as do the New York and Pennsylvania studies. This finding provides an argument for the recommendation noted at the end of this chapter for more documentation and more thorough requirements. The four characteristics with problematic data in the SACWSIS data set may warrant attention in a future study. These charact eristics are often seen as requiring more time and attention by the caseworkers The number of siblings present s a problem within the data set due to the cases not always linked with the siblings. The SACWSIS data set has unique identifier numbers for each child and each child within a sibling group is suppose to be linked through the mothers name and her unique case number. However, if the mothers case name was used to capture cases with sibling groups the data available for many of the case characterist ics used as independent variables would only reflect that
166 of the primary child. Each mothers case has an identified primary victim child and this would be the data on all of the case characteristics for that case. To capture the number of children and e ach of their associated characteristics the childs unique identifier needed to be used and this resulted in not being able to see which children were siblings. For this same reasoning it was impossible to identify the variety of placement settings sibling s were in or if they were all placed together. The history of abuse is kept in a section of the SACWSIS data set that is accessible only by the investigating caseworker and only becomes available to the ongoing caseworker when the full record is transfer red and read, it has not been available as part of the electronic record in the SACWSIS data set for the ongoing caseworker The state is working on the SACWSIS data system to connect these sections and this would be an excellent subject of a future study. The data available for the time frame of this study did include information regarding prior removal and this is an indication of history of abuse and is discussed in more detail later in this chapter. The history of abuse for a child has been historically linked to the parent and with the changes in the SACWSIS data set this information is not consistently captured or documented for a specific child. This study is looking at specific children and the case characteristics associated with them. If the data d oes become usable in the future th ese factor s would c ertainly be worth investigating in a study of this nature. The model does include an independent variable for prior removal which does address similar information regarding a pre sence of a history of abuse, and is specific to a particular child. Using the prior removal variable is the closest and most appropriate indicator of prior abuse for this study.
167 The mental health condition characteristic was so under representative of the literature and findings from various studies that it was dropped from the final analysis. Additionally, the focus groups believed that criminal histories and active criminal charges for the children and caregivers were important characteristics in determining the amount of time a caseworker would spend on a case. This data is captured in other data bases and could possibly be accessed for additional study. M ost of the children in the SACWSIS data set receive Medicaid benefits. If this data source were merged or linked to the SACW SIS data set additional characteristics could be studied as well as refining others such as mental health condition and developmental disabilities. As stated t his study is seeking to identify the case characteristics that cause more caseworker time to be devoted to the case with a practical application of assisting supervisors in assigning cases and assessing current caseloads of their staff. Most supervisors reported making case assignment decisions based on what they believed the caseworkers capacity t o be and the circumstances of the case. The supervisors reported trying to match the abilities of the worker and the capacity of the worker with the type of case. In looking at the case record the supervisors reported reviewing several factors including th e type of abuse, whether the child was removed and where the child was placed, the age of the child, the circumstances of the legal parents, any known prior history, and the number of siblings. The supervisors indicate that they assessed their staff to ide ntify those caseworkers that understood or could cope with different types of abuse. For example they had identified staff that understood physical injuries better, knew more about the treatment protocols and safety plans, anger management, etcetera and w ould be more likely to assign this caseworker to a physical abuse case. Some caseworkers had
168 more knowledge and skill in substance abuse or domestic violence, or working with teens. Supervisors used their knowledge of the caseworkers strengths and skills to assign the case. Further, the supervisors knew the current workload of the caseworkers and would take this into consideration when assigning a case. The supervisors often knew about a caseworkers personal stressors (getting married, divorced, financia l issues, health concerns, etc.) as well as their job workload and determined if the caseworker could manage the assignment before making such assignment. This process is all based upon general knowledge and assumptions without any data, research, or facts to support it. The supervisors strongly felt that if they could weight the cases as well as the current caseloads based upon the characteristics of the case this would be a significant benefit to their job. They reported believing that the workload for c aseworkers was a major factor in caseworker burnout and turnover; this is strongly supported in the literature ( Zlotnik, et. al ., 2005) Therefore, if they could balance the workload of the caseworkers by identifying time consuming cases they could retain their caseworkers for longer durations and this would improve outcomes for children. Of additional interest were the two characteristics that were not identified by the focus groups of race/ethnicity and gender. The supervisor focus group did identify the race/ethnicity characteristic of Native American Indian due to the additional work required to coordinate with the tribal authorities. Gender was not supported by the literature as being a significant characteristic within the child welfare services other than increasing the risk for certain types of abuse. However race/ethnicity is strongly supported as a characteristic that is found to lengthen the stay of children in child welfare services. According to the U. S. Government Accountability Office report A frican American Children in Foster Care,
169 (2008) African American children are over represented in the foster care system and stay much longer with contributing factors cited including; older youth in foster tend to be African American, more African America n children are diagnosed as special needs children, there is a mistrust by child welfare decision makers of the African American families, and services for African Americans or culturally appropriate are not readily or consistently available. Additionally the Casey Family programs published An Analysis of Disproportionality and Disparity at the National State and County Levels. The second research question asked; W hich critical case characteristics and tasks identified by the job descriptions, caseworkers, and supervisors that are not currently within the SACWSIS data set? FINDING II The SACWSIS data set captures 18 of the 20 (90%) common caseworker job tasks identified from the focus groups and the job descriptions. The time documented from these tasks wi thin the SACWSIS data set formed the dependent variable. The model developed for this study included 90% of the tasks identified from these sources and thus is a very strong model representing the caseworker activities and time spent. The answer regarding the case characteristics is included in the above discussion. The job descriptions were very robust in the inclusion of job tasks identifying a total of 35 job tasks and 20 common tasks for the eight casework service organizations responding. The focus groups added 3 additional unique tasks that were not part of the job descriptions (transportation, psychotropic medication requirements, and telephone contacts) The job descriptions included 12 tasks that were not identified by the focus groups. The focus groups were limiting their observations to tasks that caseworker
170 specifically perform on cases while the job descriptions included tasks that are more generic to general employ ment such as staff development, training, supervision, etc The SACWSIS data se t contained 53 activity types that the caseworkers could select from. These were consolidated into 7 general categories of activities. The analysis of the three data sources for job tasks (focus groups, job descriptions and the SACWSIS data set) indicate s that the SACWSIS data set accounts for 18 of the 23 job tasks identified. If the job tasks that are specific to general employ ment of the subcontracting caseworker agencies are eliminated (employee development, working relationships, and staff meetings) then there are only two tasks identified that are not captured by the SACWSIS data set. These tasks were psychotropic medication management and transportation. The recent suicide death of a seven year old foster child, Gabriel Myers was a catalyst for the State of Florida to re evaluate and revamp the treatment of children within the child welfare system who were being treated with psych o tropic medications. This childs death sent shock waves through the state. The investigation found that his psychotr opic medications may have played a role in his death. This resulted in the state aggressively reviewing public policies and departmental rules regarding children being prescribed psychotropic medications. This death resulted in the formation of a state wid e task force to evaluate the issues surrounding this childs death and to make recommendations for improvements in the state system. As a result the state recently made significant changes to the procedural and documentation requirements when a child is pr escribed any type of psychotropic medication. Therefore the amount of work required of the caseworker increased because a child is on these types of medications. Thus the focus groups identified this issue as a case characteristic causing more time to
171 be s pent on a particular case. This issue may not have been identified a few months earlier, but clearly is a time consuming activity currently. The tracking of psychotropic medications by child and the required consent forms and legal documents are new items recently added to the SACWSIS data set. This represents a limitation in the study resulting from the selected study time frame of 2003 to 2007 and the focus groups were held in 2009 highlighting some changes in the focus and priorities of the system. This addition of information on psychotropic medication prescriptions to the SACWSIS data set w ill also assist in capturing a better picture of the children with mental health condition. Florida has recently began collecting data on children in the foster care system whom are prescribed psychotropic drugs within the SACWSIS data set and a recent report by the Florida Department of Children and Families (2010) indicates that 14% of the children are on psychotropic medications. This would represent 539 children within the 3,855 cases included in the regression analysis. This 14% prevalence was found after a massive effort by the state and the lead agencies to minimize the number of children on psychotropic medications. A study conducted by Tufts Clinical and T ransla tional Science Institute (Leslie, et al., 2010) found that 26 states have policies and procedures regarding the use of psychotropic medications by children in state care, 13 were currently developing policies, and 9 states did not have anything and w ere not developing policies. This further identified that 13 states were not tracking data on psychotropic medication use among children in care, eight states had developed the tracking capacity within their SACWSIS data set, 9 states were tracking use thr ough Medicaid data sets, and 17 states were using
172 multiple retrievable data sets to track the use of psychotropic medication use among children in care. Raghavan, Lama, Kohl, and Hamilton (2010) used data from the National Survey of Child and Adolescent We llBeing (NSCAW), the Kaiser Family Foundation, and the Area Resource File to examine interstate variations in psychotropic medication use among children coming into contact with child welfare agencies. They found that the range for prevalence of psychotropic medication use among children in care ranged from 7.1% in California to 20.1% in Texas. Further they determined that the difference in medication use was based on the child characteristics of age, gender, foster care placement revealing that interstate variations in psychotropic medication use are driven by child characteristics, rather than by mental health need. This finding further supports the approach of evaluating the services and performance of the child welfare system by case characteristics. Each state must develop specific polici es and procedures as well as a thorough tracking system of psychotropic medication use among children in care in the SACWSIS data set. Using other data sets eliminates the likelihood that case characteristics and ment al health conditions would be easily associated with the use of psychotropic medications. As Raghavan et al (2010) found, psychotropic medication use is determined more by case characteristics driven by stakeholders in the child welfare system (foster par ents, group home providers, schools, courts) wanting to control a child s behavior than by a proper diagnosis of a mental health condition. Add to this the ongoing concern and debate regarding children receiving psychotropic medication that are not properl y tested for children and there is high risk of misuse that can harm the childs ability to achieve
173 safety, permanency and well being, as well as creating additional work for caseworkers that is unnecessary. The report by the Gabriel Myers Workgroup (Flori da Department of Children and Families 2009) found that the child welfare system had not been utilizing t he Agency for Health Care Administrations Medicaid Drug Therapy Management Program for Behavioral Health which had developed evidence based guideline s, reviewed every two years, for the use of psychotropic medications for children and therefore had no mechanism to judge whether prescribed medications were appropriate for children let alone for their mental health condition. T ransportation is not a specifically identified task in the job descriptions or the SACWSIS data set. This task was identified by the caseworkers focus groups only and is an important consideration in the time spent by a caseworker. It is unknown if caseworkers ever include the tr avel time when documenting other activities (face to face visits, collateral visits, field visits, etc.). There are several issues related to the travel/transportation component of child welfare services. The task is to support the children and families en gaged in the child welfare system to be able to get to the needed services and supports, to visit with each other, and to attend the various staffing and court hearings being held to discuss and make decisions about their case. Further the caseworker must travel to various meetings, visits, and locations during the daily course of their work. There is a wide variety of how transportation for these stated purposes is accomplished. The family has/provides their own transportation, the foster parents/group hom e staff provides the transportation, the caseworker provides the child and/or family with tickets/passes/tokens to take the public transportation available, the caseworker provides the transportation in his/her own car, the caseworker provides the transpor tation
174 in an agency vehicle, or the caseworker has access to an agency transporter who provides the transportation. Under any circumstance this takes time to do the transportation or to arrange for it. Further the caseworker often must travel several miles to visit a child, meet with the family, make a collateral contact, obtain forms and/or documents, etc. Most of the studies regarding caseworker turnover, retention, or time do not identify travel/transportation separately it is commonly included in the workload description (Zlotnik et. a l ., 2005). The Midwest ( Mills & Ivory 1991) and New York time studies referred to throughout this text did attempt to capture this variable independently with the New York study (Walter R. McDonald and Associates and American Humane, 2006) finding that 11% of the caseworker s time was spent on tra vel. However, the study did specify that the travel did not include transporting clients but was limited to only the caseworkers job related travel. In the New Mexico Depar tment of Human Services (Stein, Callaghan, McGee, & Douglas 1990) study discussed earlier the state determined there was no difference regarding travel between rural and urban cases. The f act that this task is missing from the job descriptions and SACWSIS data set is a significant flaw in the current oversight of child welfare services. Travel and transportation as separate data points need to be included in the SACWSIS data set nationally. This would allow for further delineation and analysis of how casew orkers spend their time and what additional resources could be developed. One strategic tool that is beginning to take hold in mental health and in health care is the use of video communications. Can visits, therapy sessions, sibling visits, staffing, cour t appearances, and other meetings be held through video or telecommunications tools and reduce a significant amount of the travel and missed appointments? Additionally, the further
175 utilization of combining various state and community data sources into an e lectronic record would significantly reduce the amount of time caseworkers or their support staff from tracking down documents that need to be included in the child welfare records as well as providing critical information as background and as a communicat ion tool to caseworkers, judges, teachers, therapists, medical professionals, and othe rs. This is the age of electronic storage and communication and the child welfare systems need to take advantage of these tools and time savers. T he third research ques tion asked ; I s there a significant difference in the independent variables and time spent per case between the lead agency study cases and the nonlead agency cases? FINDING III The two groups have been found to be significantly different regarding three specific independent variables, race/ethnicity, removal within 30 days, and substance abuse. Additionally, there were significant differences found with specific subcategories of the independent variables of living arrangement and placement. The validity of the placement variable data within the non lead agency cases is questionable and a real difference between the groups for this variable is not clear. There were 7 independent variables and the dependent vari able that were found not be significantly diff erent between the groups. It is important to note that the variables that are different between the groups are also the variables that are significant in explaining the dependent variable of caseworker time. With these findings it is reasonable to assume t here was some difference between the groups and that the findings for the lead agency study group cannot be simply extrapolated to the entire group of cases opened for services in 2003 in
176 Florida. However, these comparisons did not correct for experiment w ide errors that could result from the 36 statistical tests that were completed. Less than 10% of the tests were significant and with an expectation of up to 5% of these tests to be significant by chance it is difficult to determine the definitive value in these findings. There would need to be further study regarding the two groups to have confidence in making this extrapolation. Future studies can include samples drawn from the entire population as all of the Florida child welfare system is operating under a lead agency model as of the end of 2004. This study focused on the Florida child welfare cases opened for services in 2003 and completed/closed by January 1, 2007. As stated above there were 36,820 cases referred for services throughout Florida in 2003 and 97% of these cases were closed by January 1, 2007. This study focused on a sub group of these cases that were under the supervision of the existing lead agencies in Florida in 2003. The lead agency study cases made up 26% of the total cases. The compar ison of variables between the study group and the rest of the cases was conducted to determine if these variables were the same or significantly different between the groups. This comparison using independent samples t test and chi square analysis, and uti lizing a significant standard of p < .01, found that 7 of the 12 independent variables were not significantly different between the groups. These variables were age, gender, mental health condition, developmental disabilities, maltreatment, domestic violen ce, and prior removal. The variable mental health condition was found to be significantly under representative of the prevalence commonly found in child welfare populations and was dropped from the final analysis. This variable represented similar prevalen ce levels between the lead agency study cases (4.2%) and the
177 nonlead agency cases (3.8%) studies and reports throughout the literature commonly find the prevalence level of mental health condition to be between 40% and 80% (Halfonn, Zepeda, & Inkelas, 2002) None of the 7 variables that were not significantly different between the groups were found to be significant predictors of caseworker time in the regression analysis. The independent vari ables of removal within 30 days and substance abuse were signifi cantly different between the groups with the lead agency study cases having significantly higher prevalence with each of these characteristics. R emoval within 30 days w as found to be a significant factor in influencing caseworker time in the all variables Model and substance abuse was found to be significant in the both models. Removal had a negative relationship with the dependent variable in the all variables model. As discussed earlier this is counterintuitive and thus a second analysis was conducted without this variable. Substance abuse has a negative relationship with the dependent variable as well in both models indicating that when this characteristic is identified as a reason for referral it reduces the time a caseworker spends on the case. As st ated above this was not the expectation of the focus groups but may be a result of this characteristic being specific to the adults involved in the case and the documentation within the SACWSIS data set being primarily child specific. The independent vari ables of living arrangement and placement were found to have some subcategories with significant difference between the groups. Only the living arrangement with mother and father was found to be significantly different between the groups indicating the lea d agency study cases had significantly fewer children with a living arrangement of mother and father. The independent variable placement has a concern due to the high number of cases with missing data in the nonlead agency group.
178 The nonlead agency group had 60% of the cases without a placement documented throughout the life of the case (missing data) while the lead agency study cases had 36.7% of cases without a documented placement This high percentage among the nonlead agency cases indicates a potential problem with the data. The cases without a placement documented throughout the life of the case were determined to indicate the child was placed at home for the purposes of this study. When checked against current data within the USF/FMHI report (200 8) the lead agency study cases were supported in having approximately 37% of the placements at home. However the non lead agency cases were not supported by the same study of having 60% of the cases placed at home. The significant difference between the g roups with this variable was directly influenced by the number of children placed at home. The comparison resulted in a significant difference with the placement types of home, relative, nonrelative, foster care, and institution. These five placement types account for 94% of all the children in the two groups. While the statistical analysis indicates a significant difference in the placement variable between the groups the non lead agency data is not supportable as an accurate reflection of the prevalence of the various placement types. Therefore, no conclusion can be drawn as to whether these groups actually differ regarding this variable. The independent variable race/ethnicity was found to be significantly different between the two groups. The lead agenc y study cases had significantly more white children and fewer black and Hispanic children Both regression models indicated that race/ethnicity were significant characteristics in explaining the time spent by caseworkers.
179 The placement type independent variables of foster care, relative, non relative, shelter, group home, institution, and other were significant in both regression models. The placement variables of runaway and DJJ were not significant in either regression model. T he placement types of therapeutic foster care and developmental disabilities were excluded from the model due to them being a constant or not having a coefficient value. This is likely due to the very small number of cases that had these specific placement types. There were o nly seven cases with therapeutic foster care placements and 8 cases with developmental disabilities cases 53% of the cases (4,279) were dropped from the model because of missing data from some of the variables. The few cases with placement types of therap eutic foster care and developmental disabilities were likely within th e group of cases dropped from the model and therefore did not generate a beta coefficient The dependent variable of caseworker time was found not to be significantly different between t he two groups The chi square test indicated a significant difference with the lead agency study cases having significantly more tim e documented. However, Cohens d was calculated to equal .159 indicating a very weak association that does not support a sig nificant difference. Therefore, it was determined that there was not a significant difference between the lead agency study cases and the non lead agency cases regarding the time spent by caseworkers. Again, there may be a data issue with this finding as w ell due to the fact that 42% of the nonlead agency cases did not have any time documented. The lead agency study cases had time documented on 98% of the
180 cases. Those non lead agency cases with documented time averaged 9 hours less time documented per case than the lead agency study cases. This data issue for the dependent variable and the placement variable within the nonlead agency cases, supports and explains one of the key reasons for the selection of the lead agency study cases. As discussed earlier i t was expected that the lead agency cases would have better documentation of the dependent variable and the independent variables due to the unique operating conditions and requirements governing the lead agencies in Florida. The selection of the lead a gency counties as the focus of the study was primarily based on the expectation that the data for the dependent variable and the independent variables would be more thorugh and complete. As seen throughout this study this has proven to be the case. The lea d agencies were under a significantly greater degree of scrutiny rregarding documentation and the data was being used very publically to measure their performance. This drove the lead agencies to collect much better data. With the better data we have a mo re sound study and more valid results. As indicated above the difference in the lead agency study cases and the non lead agency cases is very minimal and weak, so the distinction was not to compare the two systems but to capture the most robust data. The i ssue of missing data is critical not only for this study but for the child welfare systems. According to the Administration for Children and Families, t here are 32 states that have active SACWSIS systems with 9 of these stats having achieved full federal compliance. Florida is currently being assessed by the federal government to judge compliance. There are 9 states that have partial SACWSIS systems and 10 states that have non SACWSIS systems. The Administration for Children and Families uses
181 the SACWSIS systems to collect data and provide analysis of performance and trends by state as well as nationally. The missing data within the Florida SACWSIS data set found through this study supports the need for the Administration for Children and Families to requi re validation and completeness of case information in the SACWSIS systems operating in each state and each state should ensure data is entered thoroughly and timely. The resource these 32 state SACWSIS data sets offer to better understand our child welfar e system and to develop state and national solutions to issues that have plagued child welfare services forever is immeasurable. More studies are coming out utilizing this administrative resource and the federal government is depending on this source to assist in setting policy and funding. The importance here cannot be over stressed. As demonstrated by this study the lack of completeness of the data fields that exist within the SACWSIS data set reduced the number of cases included in the regression analysi s by 52%. The finding that 29% of the cases did not have a documented maltreatment is inexcusable. Every case must be required to have documentation of the maltreatment type that brought the case to the attention of the state The fourth research question asked ; A re there specific child protection case characteristics that significantly influence the time caseworkers spend on a particular case? FINDING IV The case characteristics of race/ethnicity, living arrangement, removal, substance abuse, and placem ent influence the amount of time a caseworker will spend on a particular case. Additionally, gender, age, disability, prior removal, domestic violence
182 and maltreatment type do not influence the time a caseworker will spend on a specific case. From the two regression models the simple answer to this question is yes. The characteristics that were significant in both models are race/ethnicity substance abuse, placement types of foster care, relative, non relative, shelter, group home, and institution, and the living arrangements of living with mother and father and living with a relative. The living arrangement variables as well as the substance abuse variable have a negative relationship with the dependent variable in both models indicating that when this type of living arrangement is present the caseworker spent less time on the case. Regarding the living arrangements t his is a logical conclusion as these living arrangements with the childs family (mother and father and relative) are understandably more stable and permanent. These living arrangements of two parents and relative also have a negative bivariate relationship with caseworker time. Further supporting that living with both parents and a relative are more stable and safe environments leading to a sh orter intervention time as permanency appears eminent. Living with one parent increases caseworker time and may be a result of others in the home, the economic stability of the single parent and the case plan for permanency. The category of other living a rrangement has a positive bivariate relationship. This may be influenced by the types of living arrangements within this subcategory. The living arrangement list above indicates that the living arrangements comprising this subcategory consist mainly of fo rmal licensed placements indicating the child was removed from their home. Again, the living arrangement variable is suppose to reflect the living conditions at the point of case initiation, however this was not clear in the SACWSIS data set and
183 without in terviewing staff and finding out what their intent and timing w ere when entering this information into the SACWSIS data set it is unknown at what point in the life of the case record this information represents. Additionally, caseworkers spent less time on cases with the characteristic of substance abuse. This result is not what the focus groups expected and may seem counterintuitive at first consideration. While the literature is very strong in reporting the prevalence of substance abuse in child welfare cases and the risk of child abuse when a caregiver/parent has a substance abuse problem, there is no specific evidence that this characteristic causes a caseworker to spend more or less time on a case prior to this study It seems, as has been stated thr oughout this study, that the services and caseworker time associated with working on a case with substance abuse as a characteristic may not have a proper place for documentation with the SACWSIS data set because this characteristic is specific to the caregiver/parent and the SACWSIS data set is focused on documenting activities specific to the child. Therefore, the caseworker is not recording all of their time on these type cases. This may be due to these characteristics reflecting issues with the parents and not with the children. The focus groups could be describing more time spent with the parents and the services needed to assist the parents than time spent with the child. This may indicate that caseworkers are not documenting their time spent with the adults/parents in the same fashion they document time spent on the child. This is a significant question as to why this characteristic has a negative relationship with the dependent variable and may warrant further study. The focus groups were very clear t hat these characteristics would lead to more time needed by the caseworker T his characteristic also has a very high recidivism/relapse rate that can often prolong a case.
184 According to Young, Boles, and Otero (2007) approximately 11% of families in the child welfare system that had the children remain in their home (no removal) had substance abuse issues identified. This is very similar to the general population. Studies conducted using case review procedures specifically looking for notations of substance use problems have found rates from 43% to 79% for children placed in out of home care This study found 39.1% of the case indicated substance abuse as a reason for referral. Throughout the literature the r e are studies and reports that validate shorter len gths of stay for children in foster care when the adults receive specific treatment. A report by Alcoholism and Drug Abuse Weekly (2010) stated that an Oregon program providing substance abuse services in an intensive treatment model was able to reduce the average length of stay by 2 months. Green, Rockhill, and Furrer (2007), found that mothers with children removed from them for child abuse receiving substance abuse treatment quicker, staying in treatment longer, and completing at least one episode of tr eatment their children spent fewer days in foster care and were more likely to be reunified. Also Potter and KleinRothschild (2002) found that families with substance abuse issues were more likely to achieve timely permanency in Colorado s Expedited Permanency Planning program. The Florida SACWSIS data set does not collect data on treatment provided to or completed by adults/care givers. Florida does have a strong relationship between the substance abuse services and the child welfare system with the Fam ily Intervention Specialists (FIS). These are substance abuse/mental health specialists paid out of the states mental health substance abuse budget that are assigned to work in each community specifically with parents with substance abuse and mental heal th services. This program may be causing the adults to receive the proper treatment in an expedient
185 manner and thus reduces the length of stay of their children. The literature strongly supports this type of collaboration and intensive substance abuse serv ices do in fact reduce the length of stay for the children. This may be a better reason for the negative relationship between the case characteristic of substance abuse and the time caseworkers spend on the case. To fully understand this and verify the rel ationship between the FIS program and less time caseworkers spend on substance abuse cases a specific study needs to be conducted. The finding that race/ethnicity is a significant characteristic in both models is contrary to what the focus groups indicate d yet consistent with the literature. The focus groups were specifically asked about race after they did not identify this characteristic and still reported that this was not a factor they experienced may warrant further study. Specifically, a study to de termine if there is a significantly higher prevalence of the independent variables that influenced caseworker time associated with specific race/ethnicity groups. Such a study could support what the focus groups stated. In the bivariate analysis the race/ et hnicity was significant for black and Hispanic with a positive relationship with the dependent variable and for white it was also significant with a negative relationship indicating that minority children received more time by the caseworkers and white c hildren received less. Again, according to the Government Accountability Office (2008) and Hill (2007) race is a factor of length of stay and entry into the system. African Americans and Hispanics are over represented in child welfare systems throughout t he country. This study confirms that caseworkers spent more time on the lead agency study cases where the characteristic of black racial identity or Hispanic ethnicity were present. The focus groups did not see this factor as being significant. This
186 may be the result of the minority children in the system have a combination of the other characteristics that the focus groups did see as effecting the caseworker time such as placement, prior removal, and removal and the focus groups did not see this as an issu e of race. Most of the literature focuses on the impact of race and ethnicity in entry into the foster care system through investigations and substantiation, (Fluke, Yaun, Hedderson, & Curtis, 2003) or the impact of race on the time to permanency, (Kemp & Bodonyi, 2002) In both of these studies the findings supported that black and Hispanic children were more likely to enter the child welfare system and more likely to have longer lengths of stay than white children. This supports the finding of this stud y that black and Hispanic children received more caseworker time than white children. The practice implications of this finding are significant. It is as simple as evaluating caseloads with the understanding that minority children are going to require mor e time by the caseworker and as complicated as adjusting the case assignment process of supervisors. More specifically there is a real training and educational need presented with the gap between what the focus groups believe and what the study and literat ure support. Supervisors and caseworkers need to be educated on the importance of race/ethnicity in the cases under their responsibility and how this plays a role in the caseworkers ability to be successful. Given more attention to this finding and a bett er distribution of workload could result in shorter lengths of stay in the system for all children. The characteristic prior removal was identified by the focus groups as a characteristic that would cause more time to be spent on a case. However, possibly due to
187 the small number of cases having this characteristic in the study ( N = 68) this characteristic proved not to be significant in either model. Maltreatment type was not found to be significant in either model. Basically this finding indicates that th e type of abuse does not influence the time a caseworker spends on a case. This result may reflect the response and time spent by the caseworker is defined more by the living condition of the child (placement, living arrangement) than by the type of abuse. Possibly some children move through the system quicker to permanency and closure due to family support or the interest of adoptive parents and not due to anything related to the type of abuse that was found to have occurred. Other studies looking at prima rily length of stay have found mixed results regarding the impact of the type of maltreatment the child was referred into the system for. Benedict (1991) found that children referred for a physical abuse maltreatment had shorter lengths of stay than the me dian length of stay in her study group and those referred for neglect or sexual abuse had longer lengths of stay than the median. Yampolskaya, Armstrong and Vargo (2007) found that children referred to services for physical abuse and neglect were less like ly to be discharged from care within 24 months if their parent s rights were terminated. However, children referred for services due to physical abuse were more likely to discharge from care within 12 months if their parent s rights were not terminated. C onnell, Katz, Saunders and Tebes (2006) found that removal due to sexual abuse was associated with delays in achieving permanency. The literature does not address a relationship between type of abuse and the time a caseworker spends on the case. Further, m ost of the studies found in the literature are only considering
188 children/cases that were removed from their home. This study looks at both inhome and out of home cases. Even considering that studies have proven that certain maltreatment types may be associated with extended lengths of stay or delays in achieving permanency there is no relationship or evidence that this resulted in additional time spent by a caseworker. This may seem obvious that cases with longer lengths of stay result in more time spent on the case by the caseworker but there is no support for this conclusion in the literature. This would be a good follow up study. Therefore, the finding that maltreatment type was not a significant factor in determining the variance within the dependent variable of caseworker time is not specifically inconsistent with the literature, but there is a need for further study in this area to determine if the length of stay in the system is a significant factor in determining the amount of time a caseworker spe nds on a case. It is also equally important to note that the independent variables of gender, age, developmental disabilities, and domestic violence were not significant in either model. The idea that age does not make a difference is somewhat surpr is ing. T he focus groups emphasized th is characteristic as causing significant time needing to be spent for older children The focus groups were very vocal about older children resulting in more time spent by caseworkers. This did not prove to be true. In fact in the bivariate analysis the relationship between age and the dependent variable was significant but it was also negative Age has a unique interpretation because it is the only variable that is not categorical. The negative bivariate relationship b etween age and caseworker time indicates that as age increases the amount of time a caseworker spends on the case
189 decreases. This is a finding that could change the case assignment considerations for supervisors. T his finding sh ould provide insight for supervisors assigning cases. Younger children require more time by a caseworker and older children may stay in the system longer but receive less time. Again this specific issue may warrant further study with a specific focus on the influence of age on time s pent on the case. There was significant consensus with the focus groups and with the literature that gender was not a significant characteristic and that was supported by the findings in this study. T he characteristics of domestic violence and developm ental disabilit y were found not to be significant in both model s As stated earlier the characteristic of developmental disabilities w as found to have a much greater prevalence when looking at national reporting for child welfare cases (Administration for Children and Families 2007) D omestic violence w as a characteristic that had a significant bivariate relationship with the dependent variable but then was not found to significant in either regression model The bivariate relationship was negative similar to substance abuse. That is to say that when domestic violence was documented as a reason for referral the caseworker spent less time on the case. As discussed earlier this may be due to the fact that th is characteristic like substance abuse is a specifi c adult characteristic and a reason for referral to services in addition to the maltreatment finding. As an adult characteristic the adult needs more attention and this could result in the caseworker allowing the adult to take the actions necessary to reso lve their problem and thereby spending less time on the case or the caseworker did not document the time they spent serving the adult because the
190 SACWSIS data set is more geared towards collecting data about the child. Bills, Shin and Edleson (2010) found that the prevalence of domestic violence in child welfare cases ranges from 33% to 47% with the general population prevalence of 16.3%. Kohl, Edleson, English and Barth (2005) found in a nationally re presentative sample of child welfare cases that families with active domestic violence were substantiated for child maltreatment at higher rates than others, but the presence of domestic violence did not contribute strongly to the workers' decision making. In addition, the categories of maltreatment for which f amilies were substantiated were not different between those with or without domestic violence. Again, the research has been focused on the prevalence and identification of domestic violence in child welfare cases but not on the length of stay or caseworker time commitment when domestic violence is present. Due to domestic violence being found broadly across maltreatment types, race and age groups may be an explanation as to why this is not a significant factor in determining caseworker time spent on a case. Further, in domestic violence cases the perpetrator of the domestic violence is asked to leave the home and the children remain with the non offending parent which may reduce the time a caseworker spends on the case as well as reducing the time a case is open in the child welfare system. Removal within 30 days was found to be a significant factor in the all variables Model However, the relationship with the dependent variable was negative indicating that when the child was removed within the first 30 days of the case, caseworkers spent less time on the case. The study was conducted using these two models due to an interaction between the independent variables of placement and removal within 30 days The independent variable removal within 30 days is technically a subcategory of the
191 placement independent variable. As explained earlier every case that has a removal within 30 days will have an associated placement for that removal. However not all placements will have a removal within 30 days of the case beg in date. As was found all of the placements are associated with a removal event, 42% of the cases that did not have a removal within 30 days of the case begin date eventually had a removal and placement falling between day 31 following the case begin dat e and the closure of the case. The placement variable was constructed using only the first placement, therefore subsequent or multiple placements were not included. Because of this relationship between these independent variables the relationship between t he dependent variable and removal within 30 days became negative when both variables were included in the model. The removal variable bet a coefficient became negative when the placement variables of foster care and relative care were added to the model. Th ese two placement types explain over 65% of the variance in the dependent variable and are the top two variables that explain unique portions of the variance as indicated in the stepwise regression analysis performed. This appears to indicate some type of interaction within these variables that is not collinearity or multicollinearity as the analysis did not reveal this type of duplicative explanation of variance. An argument could be made that removals that occurred at the outset of a case (within 30 days) are more likely to achieve permanency faster through termination of parental rights and adoption or placement with a relative guardian. The cases included in the all variables model regression analysis included 619 cases with the characteristic of removal within 30 days. This represents about 21% of the lead agency study cases that included this characteristic. This small percentage of cases with this characteristic making it into the final regression equation may also have impacted the
192 coefficient direction. Also the age of children with the characteristic of removal within 30 days is younger (average age 5.9 years) than the larger group of lead agency study cases (average age 6.4 years). Again the age difference may have an impact on the coefficient dire ction for the removal variable. However, the focus group participants representing current experienced professionals in the field were clear that a removal leading to placement results in more time spent by the caseworker. Without clearly understanding why this counterintuitive result occurred a second regression was performed without the removal variable. In the all variables model the independent variable removal did explain some unique portion of the variance in the dependent variable as show in the stepwise regression analysis. With the removal variable included the model explained 22% of the variance (adjusted R2 = .22) in the dependent variable and without removal the model explained 20% of the variance (adjusted R2 = .203). In both models the strongest influence on the time a caseworker spends on a case is the placement types. The chief difficulty here is that at the point of transfer the supervisor may not know if the child will have a placement event during the life of the case. So although placem ent is a strong predictor of caseworker time it will not always be known at the time of case transfer and assignment in fact 42% of the cases that did not have a removal within the first 30 days of the case did have a removal and placement prior to the en d of the case The timing of the case transfer can impact the amount of knowledge that is available to the assigning supervisor. In general the case is to be transferred once the investigation is complete and the court has agreed to the need for ongoing se rvices. This can happen within a few days of the case begin date or within 60+ days of the case begin date.
193 In summary, t his study was intended to inform supervisors of the potential time and workload impact the assignment of a particular case would have on a particular caseworker. This study clearly demonstrates that the characteristics of race/ethnicity placement in foster care, with a relative or non relative, in a shelter, group home, or institution, or other placement will result in more time spent on a case by the caseworker. When a supervisor is assigning a case with one or more of these characteristics they can be confident it will result in additional time sp ent by the caseworker. Additionally, the presence of the characteristics of substance abuse as a reason for referral, or a living arrangement with mother and father or with a relative indicates that a caseworker will spend less time on a specific case. Further, an existing caseload can be evaluated by identifying cases with these characteristics as being cases that will consume more or less time than others and the caseload can be weighted by the number of cases that have multiples of these characteristics. This will allow the supervisor to evaluate the capacity of a caseworker to handle a new case with these characteristics or to handle a case without these characteristics. It is understood that child welfare cases are very dynamic and circumstances do change but a strength of this study is that it considered only complete cases. The analysis wa s able to capture the caseworker time, regardless of which caseworker was assigned, dedicated to the case through the life of the case The time captured was limited by the adjustments indicated above in the description of the dependent variable and the la ck of documentation in the SACWSIS data set An additional question for another study would be to ask if the time spent on a case by the assigned caseworker is affected by changes in caseworkers. It is a common
194 understanding with strong support in the lite rature that caseworkers change frequently and turnover is very high in these positions. Changes in caseworkers did not affect the data collection for this study as the identification of the caseworker was not considered just the time documented by any case worker. Limitations of the Study There are a few limitations of the study that need to be clearly articulated. There are limitations in the SACWSIS data set regarding variables that are not in the data set, missing data, under reported data, and lack of c omplete time documentation (begin time and end time) for each activity. Further it remains unclear if the caseworker travel and or transportation is included in the time documented for an activity. This is a critical factor as travel and transportation wa s cited as a significant time consuming activity in all of the time studies found and the focus groups. It is a key recommendation to add this to the SACWSIS data set. The SACWSIS data set presented numerous challenges. One was due to the lack of end tim es on activities and therefore a large number of activities could not be included in this study. The combined lead agency study cases and nonlead agency cases of 31 ,564 carried 818,998 completed activities but at least 26% of the activities were lost due to the lack of an end time or a duration of over 8 hours which is not logical and indicated a documentation error. Additionally, a large number of activities (75%) appeared to be associated with the cases that were investigated but not opened for services. Verifying this large portion of activities to be correct ly associated with cases not opened for services was beyond the scope of this study but can be seen as a limitation on the completeness of the data that was included in the study.
195 T he missing data reduced the size of the study sample by 52% (8,133 to 3,855). This smaller sample size resulted in small numbers of cases including some of the independent variables and eliminating two of the placement subcategories (Therapeutic Foster Care and Developmental Disability placement types). The impact of the missing data on the results would need to be further evaluated in a future study. The two independent variables with the missing data were living arrangement and maltreatment type both of which are requir ed data fields with the Florida administrative regulations. An additional limitation included in the SACWSIS data set is the lack of clarity over the time specific data elements are entered during the life time of the case. It was not possible to only vie w data that was entered at the point of transfer to the subcontracting casework service organizations. Specifically at what point in the case was the living arrangement documented. Additional limitations surrounded the characteristics and their relationship with the dependent variable. Both substance abuse and domestic violence have a negative bivariate relationship with caseworker time. This again is not what the focus groups predicted. The focus groups indicated that these characteristics would cause more time expended by the caseworker. This may be due to these characteristics reflecting issues with the parents and not with the children. The focus groups could be describing more time spent with the parents and the services needed to assist the parent s than time spent with the child. This may indicate that caseworkers are not documenting their time spent with the adults/parents in the same fashion they document time spent on the child. Also it was found that the mental health condition of children wi thin the lead agency study cases was significantly under reported. The characteristic of mental health
196 condition of the child was found in only 4.2% of the cases. This level of mental health condition prevalence in a foster care population is well below the 57% prevalence level found by dosReis, Zito, Safer, and Soeken (2001). The National Court Appointed Special Advocates Association reported the prevalence of mental illness among foster children to range between 40% and 85% (The Connection, 2004). Accordi ng to Halfonn, Zepeda, and Inkelas (2002), several studies indicate that from 50 to 80 percent of the children in foster care suffer from moderate to severe mental health problems. There are many more studies to support the significant number of children i n foster care with mental illness. The state of Florida has recently begun to collect data on children on psychotropic medications within the foster care system and currently reports 14% of the children receiving these types of medications. While this does not include all children with mental illness it certainly reinforces that the 4.2% found in the SACWSIS data set to be a significant under reporting. Regardless of these limitations, the SACWSIS data set is the most robust source of information regarding child welfare services and outcomes in the state of Florida. It is comparable to every other state information as the development of a SACWSIS data system is a requirement from the federal government and each state has developed such systems. This data sou rce is a fairly untapped resource for research and should be more aggressively pursued in future studies of this social issue. Practice, Policy, and Research Recommendations The following practice, policy and research recommendations are simply a more c oncise reiteration of the discussion and findings above.
197 Practice recommendations 1. Cases should be evaluated by the case characteristics of placement, living arrangement, race/ethnicity, and substance abuse as indicators of the amount of time a casewor ker will need to devote to the case. Further each caseload can be assessed using these same characteristics to determine the caseloads needing more caseworker time. Utilizing the findings of this study in this manner will assist caseworkers and supervisor s to manage their time better, create appropriate workloads, assign cases that will not overwhelm caseworkers, improve caseworker performance and retention. Supervisors will be able to evaluate each case they are asked to assign to a caseworker and make st rategic decisions based on the amount of time a current caseworkers caseload consumes and the new case will add. Also supervisors will be able to better judge cases that will consume less time for new caseworkers or for caseworkers that have other confli cts. 2. Caseworkers must document thoroughly in the SACWSIS data set. Caseworkers need to be trained and retrained a minimum of semi annually to refresh their understanding of the complete data set, the terms and definitions, and the timeliness of the dat a entry. 3. Increase the number of required fields in the SACWSIS data set to assure completion of the information. This would include caseworker travel, transportation, removal dates on every child that is removed from their primary caregiver (parent/gu ardian) at any point in the life of the case begin time and end time for each activity, maltreatment type, history of abuse, siblings, living arrangement, identify inhome cases with no removal event in the life of the case, etc.
198 4. Services to the parent s must be documented and collected to define which services are effective in achieving specific child protection outcomes as well as to collect the full effort of work and service being provided on a case. Once this information is being collected further s tudies will be able to be conducted. A case record review might reveal documentation in the paper record of services provided to the parents/caregivers and could be the data source for a study. Putting a system like this described in these two recommendat ions would minimize the missing data and standardize the reporting of information. This would allow for the use of case characteristics of sibling groups, placement of sibling groups, history of abuse, and parents who were former foster children. Policy Re commendations 1. Aggregate data systems: The data missing from the SACWSIS data set that was identified by the focus group can in part be found in other state data systems. Specifically the data can be found in the Department of Juvenile Justice, the Medi caid data base, the Department of Education, the Agency for Persons with Disabilities, the Child Protection Investigations data within the current SACWSIS data set need to be accessible to identify a history of abuse, and the Florida Department of law enforcement. These data basis could provide information regarding; delinquency history of children, children on psychotropic medications, children with dual diagnosis, and criminal charges against the caregiver/parent. These systems can be combined with a comm on child identifier and a search engine. 2. Add new fields to the SACWSIS data set to include, caregiver mental health condition, high profile cases (cases with media coverage, involvement of political figure
199 or community leader, etc.), uncooperative par ents (this may be a difficult data field to incorporate due to general subjective nature of it), nonE nglish speaking parents, immigration issues, and expedited termination of parental rights. 3. All states need to be tracking the mental health condition a nd the use of psychotropic medications for children to assure the proper treatment is being provided and psychotropic medications are not being used as behavior management tools. It is recommended that the Administration for Children and Families under the Federal Department of Health and Human Services require every state SACWSIS data set to add these fields as mandatory data for each child in the system. Research Opportunities 1. The system continues to keep paper records so one research opportunity would be to review a sample of paper records to verify the SACWSIS data set and determine on specific case characteristics the availability of the data as well as the accuracy of the data. This could include mental health condition, prior abuse, and sibling status. 2. Additional research is recommended to determine the impact upon the time caseworkers spend on a case of the history of abuse (data collected through an improved SACWSIS data set or through case review), removal data to determine more specificall y why this characteristic had a negative beta coefficient with caseworker time as the dependent variable, and pulling a sample of cases with siblings. 3. Further research into the impact of the characteristics of substance abuse and domestic violence in an effort to determine if caseworker is spent on these cases but not recorded in the SACWSIS data set, if treatment of parents is taking place and actually
200 reducing the time a case is opened for services and the time a caseworker spends on the case, and if these characteristics are being captured and identified correctly. 4. Additional research into the impact of missing data on the accuracy and utility of the SACWSIS data set. This focused study should look at the national data and determine how much is mis sing and in what areas as well to create a national and state comparative view. 5. A study to determine if the frequent changes in caseworkers have an impact on the sum of caseworker time spent on a case. Does a new caseworker spend more time on a case th ey receive at a mid point in the life of the case? 6. This study focused on case characteristics and their impact on caseworker time a follow up study to evaluate the caseworker characteristics and the impact on their time spent on a case would provide a complete picture. It seems reasonable to expect that different skill, training, organizational, personality traits among the caseworkers could very well determine how much time it takes to accomplish the activities and to perform the necessary work to reso lve a case. This is a study that is strongly recommended and would provide the supervisors assigning the cases the second half of the equation needed to properly match a case with a caseworker for the best results.
201 References Agresti, A., (2002). Categorical Data Analysis (2nd ed.) Hoboken, New Jersey: Wiley Interscience. Alcoholism & Drug Abuse Weekly (2010). Intensive parent treatment in Oregon leads to shortened foster care stays. Alcoholism & Drug Abuse Weekly ,22(7), 1 8. Alexander Jr ., R., & Alexander, C. L. (1995). Criminal prosecution of c hild protectio n w orkers. Social Work 40 (6) 809814. Annie E. Casey Foundation. (2003). The unsolved c hallenges of s ystems r eform: The c ondition of the f rontline human s ervices w orkforce. Balti more, MD: Annie E. Casey Foundation. American Federation of State, County, and Municipal Employees ( 1998) Double j eopardy: c aseworkers at r isk helping a t r isk k ids. AFSME Publications Retrieved September 10,2007 from http://www.afscme.org/publications/ 1331.cfm Armstrong, M., et al (2008). Report to the Legislature Evaluation of the Department of Children and Families CommunityBased Care Initiative Fiscal Year 2006 2007. The Louis de la Parte Florida Mental Health Institute Armstrong, M., et al (20 06). Community Based Care White Paper. The Louis de la Parte Florida Mental Health Institute Armstrong, M., et al (2005). Statewide Evaluation of Floridas Community Based Care: 2005 Final Report. The Louis de la Parte Florida Mental Health Institute
202 Bednar, S. G. (2003). Elements of s atisfying o rganizational c limates in c hild w elfare a gencies. Families in Society, 84 (1), 7 12. Benedict, M. and White, R. (1991). Fa ctors associated with foster care length of stay. Child Welfare League 70 4558. Berry, M. (1993). Adoptive parents' perceptions of, and comfort with, open adoption. Child Welfare 77(3), 231 253. Bills, J., Shin, N., LaLiberte, T., and Edleson, J. (2010). Child welfare professionals' responses to domestic violence exposure among children Children and Youth Services Review 32(12), 16401647. Bruhn, C. (2003). C hildren with disabilities: Abuse, neglect and the child welfare system. Hawthorn Maltreatment and Trauma Press 8(15/16), 173203. Buddenbaum, J. M. and Novak K. B. (2001). Applied communication research. Ames IA. I owa State University Press, Blackwell Publishing. Carbaugh, S. F. (2004). Understanding shaken baby syndrome. Advances in Neonatal Care 4(2), 105117. Child Welfare League of America. (1999). CWLA s tandards of e xcellence for s ervices for abused or neglected c hildren and t heir f amilies. Wa shington, DC: Author. Child Welfare League of America. (1999). CWLA s tandards of e xcellence for f amily f oster c are. Washington, DC: Author. Child Welfare League of America. (2002). Research r oundup: Child w elfare w orkforce. Washington, DC: Author. Clausen J. M., Landsverk, J., Ganger W., Chadwick D., & Litrownik A. (1998).
203 Mental health problems of children in foster care. Journal of Child and Family Studies 7 (3) 283296. Connell, C., Katz, K., Saunders, L., & Tebes, J. (2006). Leaving foster care the influence of child and case characteristics on foster care exit rates. Children & Youth Services Review 28 (7), 780798. Council On Accreditation Standards, 8th Edition ( 2006) Retrieved October 20, 2007 f rom http://www.coastandards.org/standards.php. Court Appointed Special Advocate (CASA) National Association (2004). Mental Health Needs of Youth in Foster Care: Challenges and Strategies. The Connection, 20(4), 613. Creswell, John W. (1998). Qualitative inquiry and research design, choosing among five traditions. Thousand Oaks, CA : Sage Publications Cyphers, G. (2001). Report from the c hild w elfare w orkforce s urvey: State and c ounty data f indings. Washington, DC: American Public Human Services Association. Dalton, G., & Morelli, P. (1988). Casemix and c aseload: Measurement of output of a s ocial w ork a gency. Administration in Social Work 14(5), 351 357. Depanfilis, D. (2006). Compassion f atigue, burnout, and c ompassion s atisfact ion: i mplications for r etention of w orkers. Child Abuse & Neglect, 30(10), 10671069. Dos Reis, S., Zito, J M., Safer, D J., & Soeken, K L (2001). Mental health services for youths in foster care and disabled youths. American Journal of Public Health, 91(7), 10941091. Drake, B., & Yadama, G. N. (1996). A s tructural e quation m odel of burnout and j ob
204 e xit among c hild p rotective s ervices w orkers. Social Work Research, 20 (3) 179188. Drennan V Walters K Le nihan P Cohen S Myerson S & Iliffe S ; SPICE Research Group. (2007). Priorities in identifying unmet need in older people attending general practice: a nominal group technique study. Family Practice, 24(5) 45460. Dufoe, N., Carroll, L., Wipple, J., Vic kers, B., & Vanlandignham, G. (2006). Florida Office of Program Policy Analysis and Government Accountability (OPPAGA); Child w elfare s ystem performance m ixed in f irst y ear of s tate w ide c ommunity based c are. St ate of Florida OPPAGA Office Report Number 0650. Dunnette,M. D., Campbell, J. D., and Jaastad, K. (1963) The Effect of Group Participation no Brainsto r ming Effectiveness for Two Industrial Samples Journal of Applied Psychology 47 (1), 3037. Edwards, M. (2002). Child Welfare Workload Standards: Reality and Myths. Protecting Children, 17 (3), 4046. Ellett, A. J. (2006). Broad s tudy but narrow question: A f riendly c ritique of perrys a rticle. Research on Social Work 16 ( 4) 406411. E llett, A. J., Ellis, J. I., & Westbrook, T. M. (2007). A qualitative s tudy of 369 c hild w elfare professionals' perspectives about f actors c ontributing to e mployee r etention and t urnover. Children & Youth Services Review, 29 (2), 26481. Emlen, A., Lahti, J., Downs, G., McKay, A., & Downs, S. (1977). Overcoming ba rriers to planning for c hildren in f oster c are. Regional Research Institute, Portland State University
205 Ferketich, S. & Verran, J. (1994). An overview of data transformation. Research in Nursing and Health, 5, 393404. Finkelhor D. (1994). Sexual abuse o f children. The Future of Childr en 4(2), 3153. Florida Department of Children and Families (2010). Report on Psychotropic Medication use among children in care. Florida Department of Children and Families Tallahassee, FL Fluke, J., Yuan, Y., Hedders on, J., Curtis, P. (2003). Disproportionate representation of race and ethnicity in child m altreatment: investigation and victimization. Children and Youth Services Review 25( 56) 359373. Fowler, C., McDonald, J., & Sumsk i, M. (2005). Review of t urnover in M ilwaukee C ounty private a gency c hild w elfare ongoing c ase m anagement s taff. Child Welfare Associates. Retrieved September 30, 2007 from http://www.childwelfare associates.com/ Freund, A. (2005). Commitment and j ob s atisfaction as predictors of t urnover i ntentions among w elfare w orkers. Administration in Social Work, 29(2), 5 21. Garnier, P., Zyphur, M., & Zhai, F. (2006). Investigating the e ffects of c asewo rker c haracteristics in c hild w elfare. Children and Youth Services Review, 28(9), 9931006. Gillespie, D. F., & Cohen, S. E. (1984). Causes of w orker burnout. Children and Youth Services Review 6(2), 115124. Glass, G. V. and Hopkins, K. D. (1996). Stat istical methods in education and psychology (3rd ed.). Boston: Allyn and Bacon, Inc.
206 Government Accountability Office. (2003). Child Welfare: HHS Could Play a Greater Role in Helping Child Welfare Agencies Recruit and Retain Staff. GAO03357. Washingt on, DC: Author. Government Accountability Office (2006). Improving Social Service Program, Training and Technical Assistance would Help Address Long Standing Service Level and W orkforce Challenges. GAO 0775. Washington, DC: Author. Government Accountabi lity Office (2008). African American Children in Foster Care, GAO081064T Washington, D. C: Author. Green, B., Rockhill, A., & Furrer, C. (2007). Does substance abuse treatment make a difference for child welfar e case outcomes? A statewide longitudinal analysis. Children and Youth Services Review 29 (4) 460473. Halfon, N., Zepeda, A., Inkelas, M. (2002). Mental Health Services for Children in Foster Care. UCLA Center for Healthier Children, Families and Communities, 4(9):1 13. Herzberg, F. (1966). Work and the nature of man. Cleveland: World Pub. Hill, R. B. (2007). An Analysis of Disproportionality and Disparity at the National State and County Levels. Casey Family Programs Hopkins, K. M., Mudrick, N. R., & Rudolph, C. S. ( 1999). Impact of university/a gency partnerships in c hild w elfare on organizations, w orkers, and w ork a ctivities. Child Welfare, 78(6) 749773. Jayaratne, S., & Chess, W. (1984). Job s atisfaction, burnout and t urnover: A national s tudy. Social Work 29 (5) 448453.
207 Katz, L (1990). Effective permanency planning for c hildren in f oster c are. Social W ork, 35(3), 220. Kemp, S., & Bodonyi, J. (2002). Beyond termination: length of stay and predictors of permanency for legally free children. Child Welfare 81(1), 58 86. Knox, D., & Higgins, J. (2003). Caseload definition in dispute. Akron Beacon Journal Retrieved August 18, 2007, from ht tp://www.ohio.com/archives Kohl, P. L., Edleson, J. L., English, D. J., & Barth, R. P. (2005). Domestic violence and pathways into child welfare services: Findings from the National Survey of Child and Adolescent Well Being. Children and Youth Services R eview 27(11) 1167 Laird, J., & Hartman, A. (1985). A h andbook of c hild w elfare: context, knowledge and practice. New York: Free Press. Lee, D. (1979). Staying a live in c hild protective s ervices: s urvival s kills for w orker and s upervisor. Arete, 5 (4) 195253. Leslie, L., Mackie, T., Dawson, E., Bellonci, C., Schoonover, D., Rodday, A., Hayek, M., Hyde J. ( 2010) Multi State study on psychotropic medication oversight in foster care. Tufts Clinical and Translational Science Institute. Lieberman, A. A., Hornby, H., & Russell, M. (1988) Analyzing the educational backgrounds and work experiences of child welfare personnel: A national study. Social work, 33(6), 485489. Maslach, C., & Jackson, S. (1986). Maslach burnout i nventory m anual (2nd ed.). Palo Alto, CA: Consulting Psychologists Press. Mills, C. S., & Ivery, C. (1991). A s trategy for w orkload m anagement in c hild
208 protective p ractice. Child Welfare, 70 (1) 3543. Montcalm, D., & Royse, D. (2002). Data Analysis for Social Workers .Boston: MA: Allyn and Bacon. Mor Barak, M. E., Nissly, J. A., & Levin, A. (2001). Antecedents to r etention and t urnover among c hild w elfare, s ocial w ork, and other human s ervice e mployees: What can w e l earn from past r esearch? A r eview and m etanalysis. Social Service Review 75(4), 625661. Mor Barak, M. E., Levin, A., Nissly, J. A., & Lane, C. J. (2006/5). Why do t hey l eave? Modeling c hild w elfare w orkers' t urnover i ntentions. Children and Youth Services Review 28(5), 548577. Morris, J. (2005). For the c hild ren: A ccounting for c areers in c hild p rotective s ervices. Journal of Sociology and Social Welfare, 32(2) 13145. National Center on Child Abuse Prevention Research. (2001). Current trends in child abuse prevention, reporting, and fatalities: The 1999 fifty State survey Chicago: Prevent Child Abuse America. Retrieved June 19, 2010 from: www.preventchildabuse.org/learn_more/research_docs/1999_50_survey.pdf. Nunno, M. (2006). The e ffects of the ARC organizational i ntervention on c aseworker t urnover, c lim ate, and c ulture in c hildren's s ervices s ystems. Child Abuse & Neglect, 30 (8), 849854. Osborne, J. W., Ryan, J. (2008). Best Practices in Quantitative Methods Thousand Oaks, CA: Sage Publications O Brie n, R. M. (2007). A Caution Regarding Rules of T humb for Variance Inflation Factors. Quality and Quantity, 41(5), 673690.
209 Perry, R., & Murphy, S. (2008). A critical appraisal of what child welfare workers do: Findings from a task analysis study in Florida. Protecting Children 23 (3), 4473. Potter, C., & Klein Rothschild, S. (2002). Getting home on time: Predicting timely permanence for young children. Child Welfare 81 (2), 123150 Raghavan R., Lama G., Kohl P., H amilton B.H. (2010). Interstate variations in psychotropic medication use among a national sample of children in the child welfare system. Child Maltreatment, 15(2) 121131. Sewell, J., Janes, W., Rosenberg, R., & Tandon, R., Wells, A. (2009). The Gabriel Myers workgroup report. Florida Department of Children and Families Stein, T. J., Callaghan, J., McGee, L., & Douglas, S. ( 1990) A c aseload w eighting f ormula for c hild w elfare s ervices. Child Welfare 69(1) 3342. Stein, T. J., Gambrill, E. D., & Wi ltse, K. T. (1978). Children in f oster homes; Achieving c ontinuity of c are. New York : Praeger Publishers. Tabachnick, B. G., & Fidell, L. S. (1989). Using multivariate statistics (2nd ed.) New York: Harper & Row. Tabachnick, B. G., & Fidell, L. S. (2 006). Using multivariate statistics (5th ed.). Boston, MA: Allyn and Bacon. Tooman, G., & Fluke, J. (2002). Beyond c aseload: What w orkload s tudies c an t ell us a bout e nduring i ssues in the w orkplace. P rotecting Children, 17(3), 18. Tollen, W B. ( 1960) Study of s taff l osses in c hild w elfare and f amily s ervice Agencies. Childrens Bureau Publication no. 3831960. Washington, D.C.: U.S. Government Printing Office.
210 U.S. Department of Health and Human Services, Administration on Children, Youth and Fami lies (2009) Child Maltreatment 2007 Washington, DC: U.S. Government Printing Office. U. S. Department of Health and Human Services, Administration for Children and Families Childrens Bureau (2007). Child Welfare Outcomes 20032006: Report to Congress U. S. Government Printing Office. U.S. Department of Health and Human Services. (1999). Blending perspectives and building common ground: A report to Congress on substance abuse and child protection. Washington, DC: U.S. Government Printing Office. Vale ntine, J. C. & Cooper, H. (2003). Effect size substantive interpretation guidelines: Issues in the interpretation of effect sizes. Washington, DC: What Works Clearinghouse. Walter R. McDonald & Associates, Inc., American Humane Association (2000). SB 2030 Child Welfare Services Workload Study: Final Report. American Humane Association. Retrieved on September 28, 2007 from http://www.dss.cahwnet.gov/cdssweb/sb2030chil_287.htm Walter R. McDonald & Associates, Inc., American Humane Association (2006). New York State Child Welfare Workload Study: Final Report. American Humane Association. Retrieved on September 28, 2007 from http://www.ocfs.state.ny.us/main/reports/WorkloadStudy.pdf Werrbach, G. B. (2002). Providing intensive child case management services: How do case managers spend their time. Child and Adolescent Social Work Journal 19(6) 473486
211 Yamatani, H., Engle, R., & Spjeldnes, S. (2009). Child Welfare Worker Caseload Whats Just Right? Social Work, 54(4), 361368. Yampolskaya, S., Armstrong, M., & Vargo, A. (2007). Factors associated with exiting and reentry into out of home care under Community Ba sed Care in Florida. Children & Youth Services Review 29 (10), 13521367. Young, N., Boles, S., & Otero, C. (2007). Parental Substance Use Disorders and Child Maltreatment: Overlap, Gaps, and Opportunities. Child Maltreatment 12( 2) 137149. Zell, M. C. (2006). Child w elfare w orkers: Who t hey are and how t hey view the c hild w elfare s ystem. Child Welfare 85(1) 83103. Zlotnik, J., DePanfilis, D., Daining, C., & Lane M. (2005). Factors i nfluencing the r etention of c hild w elfare s t aff: A s ystematic r eview of r esearch. Institute for the Advancement of Social Work Research Report from the Institute for the Advancement of Social Work Research, University of Maryland School of Social Work, Center for Families & Institute for Human Serv ices Policy.
213 Appendix A Child Welfare League of America Standards of Excellence I. Standards of Excellence for Family Foster Care Services, Recommended Caseload Standard Criteria. The complexity of the needs of the child and family The le vel of competency of the social worker, including skills and experience The specific functions assigned, including intake responsibilities and court work, and the concomitant time requirements of each The geographic area served and the time required for tr avel for service provision The availability for services and resources required by the clients The number of other agencies involved in providing services to the cases within the caseload The time required for case documentation and court related activitie s, and The time needed for agency activities such as meetings, professional development, and administrative functions II. Standards of Excellence for Abused or Neglected Children and their Families, Recommended Caseload Standard Criteria The specific assig ned functions and time required for each task (e.g. intake, assessment/investigation, placement services, court activities, community development, provision of services); The competencies needed for each social work function (knowledge, skills, experience) ;
214 Appendix A (Continued) The time required for travel and other necessary but noncasework tasks; Standards of sound practice; The availability of paraprofessionals and professionals from other services to help with routine activities (e.g. foster families in home aides); The intensity of services that the agency and the community considers appropriate; The number of other agencies, individuals, or services involved with the family and the amount of time needed to communicate effectively with other communi ty partners; The amount of time needed for community outreach or other activities not tied to a specific family; and The amount of time allocated for activities such as staff meetings, training and development, administrative functions, and personal leave.
215 Appendix B Guidelines for Computing Caseload Standards The most requested CWLA Standards are those that provide recommended caseload ratios for workers in child welfare program areas, such as child protective services, foster care, adoption, and residential services. These ratios of clients to staff members offer guidance based on the field's consensus of what constitutes best practice. They're also supported by the findings of caseload and workload studies and by projects that show particular success in reaching agency goals. The following broad principles provide a context for agencies as they approach the task of computing caselo ads for child welfare workers: People are the key ingredient in an effective child welfare system. Child welfare work is labor intensive. Caseworkers must be able to engage families through face to face contacts, assess the safety of children at risk of harm, monitor case progress, ensure that essential services and supports are provided, and facilitate the attainment of the desired permanency plan. This cannot be done if workers are unable to spend quality time with children, families, and caregivers. Computing caseloads is an inexact science. When in doubt, err on the side of safety. When systems are short staffed, bad th ings can happen. Studies of critical incidents, including child deaths, child injuries, and children missing from foster care, almost always involve an overworked caseworker who didn't have sufficient time to
216 Appendix B (Continued) adequately assess or mo nitor the child's situation. In addition to leading to such tragedies, insufficient staffing results in inefficient services. Our goal is to ensure safety, permanency, and well being for all children who come to the attention of the child welfare system We need to focus on what it takes to achieve these service goals. In the federal Child and Family Service Reviews, those states that showed strength in such areas as family involvement and worker contact with children in foster care were more likely to achieve safety and permanency goals. Caseloads must permit such activities and opportunities. Currently no universally accepted formula for computing caseloads exists. But the following general rules of thumb can guide jurisdictions in determining the numb er of workers necessary to meet CWLA's recommended standards: The CWLA caseload standards are expressed in terms of maximum cases per worker. Any formula should result in caseloads no greater than the maximum recommended number, rather than exceed it. For example, anticipated vacation and sick leave time, agency holidays, and regularly scheduled training events should be deducted from the number of calendar days to arrive at the total actual workdays available per worker per month. This should be done bef ore computing caseloads. Some caseload ratios are expressed in terms of cases per month, whereas others are expressed in terms of the number of cases on any given day. These variations need to be accounted for in computing cases. For example, for investi gative workers in child protective services, the recommended caseload is 12 active
217 Appendix B (Continued) cases per month. This number should not be construed to mean 12 active cases at any point in time, but 12 active cases in the workdays available duri ng a designated 30day period or month. Moreover, if the worker is carrying forward cases from the previous month, the number of new cases should be reduced accordingly. Caseloads should be computed separately for each worker category. For example, when computing any category of workers, staff who may play a role in service delivery but are not performing the specific functions of this category, should not be included in the worker count. Though helpful, case aides, supervisors, and others who may assist with cases, do not perform the same functions, and including them provides a misleading caseload count. Case transfers and changes in case status should receive careful consideration. Caseload counts should accrue to the worker, not to the case. Multip le workers may address the practice needs of a family and its children in a given period. Whenever cases transfer from one worker to another within a specified period, they should be counted on each worker's caseload. The fact that this is a single case do es not negate the need to count it as part of each worker's caseload. The same principle applies to changes in case status.
218 Appendix B (Continued) Caseloads and workloads A U.S. Children's Bureau document, Workload Standards for Children and Family Soc ial Services, differentiates caseload and workload measures as follows: Caseloads are defined as the amount of time workers devote to direct contacts with clients. Workloads are defined as the amount of time required to perform a specific task. Although CWLA recommends caseload ratios for each area of child welfare practice, workloads are best determined through careful time studies conducted within the individual agency. They should be based on the responsibilities assigned to complete a specific set of tasks or units of work for which the worker is responsible. For those agencies interested in developing their own specific workload figures, time required to conduct the following tasks should be calculated: travel; collateral visits, outreach activitie s, and court schedules; emergencies that interrupt regular work schedules; supervision, consultation, and collaboration; work with community groups; attendance at staff meetings, staff development, professional conferences, and administrative functions ; case management; and
219 Appendix B (Continued) telephone contacts, reading of records, case recording or computer entry, and reports of conferences and consultations.
220 Appendix C Caseworker and Supervisor Focus Group Questions Specific questions for the group include: o What case characteristics do you generally know about a case when it is assigned to you or you are assigning the case? o Which of these characteristics do you think determine the amount of time caseworkers will have to spend on a case? Focus group participants will be asked to rate the case characteristics impact on the amount of time a caseworker devotes to a case as 1 (low), 2 (medium), or 3 (high) o What tasks do caseworkers perform that take up the most amount of time? Focus group participan ts will be asked to rate the time for tasks identified as 1 (low), 2 (medium), or 3 (high) o What case characteristics do you believe would be present in a case that would take minimal time? o How many cases do you think a caseworker ought to have? Does it ma ke a difference the type of cases assigned? What types of cases make a difference and why? o Do you count cases by child or by family? o How many hours do you work each week?
221 Appendix C (Continued) o What do you think is the usual number of hours caseworkers work each week?
222 Appendix D IRB # 107640 I Examination of the Effect of Child Abuse Case Characteristics on the Time a Caseworker Devotes to the Case Research Project University of South Florida School of Social Work Informed Consent This document is designed to inform any prospective participants in the Examination of the Effect of Child Abuse Case Characteristics on the Time a Caseworker Devotes to the Case Research Project (the Project) of all aspects of --and their participation in --the project. This is a completely voluntary effort. If you agree to participate, you are free to withdraw your consent at any time, and you may discontinue your participation in the focus group at any time without consequence. You do not have to answer any question you do not wish to answer. Purpose and Duration As part of a dissertation project, focus groups with active case managers and supervisors are being conducted, the purpose of which is to capture the opinions of active and experienced caseworkers and supervisors regarding the case characteristics and key caseworker tasks that determine the amount of time a caseworker will need to devote to a specific case. You are being asked to participate in a focus group lasting no longer than 90 minutes. The focus gr oups will be conducted privately during business hours at the offices of the case managers and supervisors, and all focus groups will be completed during the months of October and November, 2009. Procedure The researchers will identify the case manager s and supervisors through the management of the organizations in which they are employed. The management will give consent for the research, and participating case managers and supervisors will voluntarily sign
223 Appendix D (Continued) this informed consen t document. An introductory and scheduling call will be made to the case managers by one of the researchers. The interview will be scheduled at a convenient time for the participants. All participants will be asked to answer 7 questions and then provid e some background information. Risks and Benefits The risk to participating in this research is the loss of 90 minutes of precious time for the case manager and or supervisor. We readily acknowledge this is a sacrifice and are very appreciative to the participants. The researchers are all former chi ld welfare case managers and do know the value this time holds to the case manager. If you feel that you may have been harmed in any way by participation in this study, please contact the University of South Florida School of Social Work at (813) 974-2063, or the Principal Investigator or Faculty Advisor at the numbers listed under Contact Information below. The benefit to participation in this study is in the ability to build the knowledge base and provide an informed list of key case characteristics and caseworker tasks that drive the amount of time a caseworker must spend on a specific case. The researchers believe that the core to success for children and families involved in the child welfare system is based on the performance of the case manager. Defining and allowing the management of the workload of the case manager is a key to building success for each child and family. Confidentiality The answers provided will never be identified with the participant, nor will any of the participants ever be identified by name or any other descriptor that is individually applied. The results will be reported as group responses, and full confidentiality shall be maintained at all times. Participants will be from multiple Florida counties, adding to the assur ance of confidentiality. Because of the group setting, absolute confidentiality cannot be guaranteed.
224 Appendix D (Continued) However, we ask that you keep what is discussed during the group confidential and not disclosed to others outside of the group. Contact Information If you have any questions regarding this study or participants rights, please contact: Chris Card, Principal Investigator, at (813) 8431827, or at email@example.com; or the Faculty Advisor, Dr. Bill Rowe, Professor and Director, School of Social Work, University of South Florida, 4202 East Fowler Avenue, MGY 132, Tampa, FL, 336206600 at (813) 974-2706, or at firstname.lastname@example.org. Personal Statement and Consent I have read this statement and understand the purpose, procedure, risks and benefits of the Examination of the Effect of Child Abuse Case Characteristics on the Time a Caseworker Devotes to the Case research project by the University of South Florida School of Social Work doctoral students. I understand that, at any time, I may withdraw without consequence. I voluntarily agree to participate in this study. ________________________ _______________________________ Participant Signature/Date Team Member/Date + + + + + + + + + + + + + + + + + + + + I voluntarily agree to have the interview audio taped by the team member. (The tape is only to be used by the study team to assure accuracy of informati on for data analysis purposes.) ______________________________ ____________________________________ Participant Signature/Date Team Member/Date