USF Libraries
USF Digital Collections

Connections count

MISSING IMAGE

Material Information

Title:
Connections count understanding gender and race differences in school-based problem behavior during adolescence
Portion of title:
Understanding gender and race differences in school-based problem behavior during adolescence
Physical Description:
Book
Language:
English
Creator:
Santa-Lucia, Raymond C
Publisher:
University of South Florida
Place of Publication:
Tampa, Fla.
Publication Date:

Subjects

Subjects / Keywords:
discipline
referrals
suspensions
dropout
longitudinal
Dissertations, Academic -- Psychology -- Doctoral -- USF   ( lcsh )
Genre:
government publication (state, provincial, terriorial, dependent)   ( marcgt )
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )

Notes

Summary:
ABSTRACT: Utilization of a large, diverse sample provided a rare opportunity to advance our understanding of gender, race, and socioeconomic differences in school-based problem behavior. Yearly assessment of discipline referrals and suspensions received within the school context from 5th- through 11th-grade, as well as assessment of school dropout, provided an opportunity to examine these issues through an extended prospective longitudinal design. Results highlight the middle school transition as a time when discipline referrals and suspensions increase markedly, while student reports of connections to others, motivation, and optimism decline sharply. Results indicate that boys, African-American students, and students from low socioeconomic status backgrounds report lower levels of connections to others, motivation, and optimism in 5th-grade. Boys, African-American students, and students from low socioeconomic status backgrounds also receive more discipline referrals and suspensions from 5th-grade onward and are more likely to experience dropout. However, regardless of demographic group membership, students who report stronger connections to others, motivation, and optimism in 5th-grade receive fewer referrals and suspensions from 5th- through 11th grade, and are much less likely to dropout of school than are students who report lower levels of connections, motivation, and optimism in 5th-grade. These results highlight the need to address students' sharp declines in functioning across the middle school transition through both ecological and person-centered prevention and school restructuring efforts. Results also highlight the utility of movement away from a static, demographic based understanding of problem behavior toward a clearer understanding of person and environment factors that may underlie both between and within demographic group differences in outcomes. Placing emphasis upon factors that are potentially amenable through school based prevention efforts considerably increases the likelihood that all of our nation's children are provided with equal opportunity to achieve their fullest potential.
Thesis:
Thesis (Ph.D.)--University of South Florida, 2004.
Bibliography:
Includes bibliographical references.
System Details:
System requirements: World Wide Web browser and PDF reader.
System Details:
Mode of access: World Wide Web.
Statement of Responsibility:
by Raymond C. Santa Lucia.
General Note:
Includes vita.
General Note:
Title from PDF of title page.
General Note:
Document formatted into pages; contains 124 pages.

Record Information

Source Institution:
University of South Florida Library
Holding Location:
University of South Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 001469392
oclc - 55731003
notis - AJR1146
usfldc doi - E14-SFE0000293
usfldc handle - e14.293
System ID:
SFS0024988:00001


This item is only available as the following downloads:


Full Text
xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader nam Ka
controlfield tag 001 001469392
003 fts
006 m||||e|||d||||||||
007 cr mnu|||uuuuu
008 040524s2004 flua sbm s000|0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0000293
035
(OCoLC)55731003
9
AJR1146
b SE
SFE0000293
040
FHM
c FHM
090
BF121
1 100
Santa-Lucia, Raymond C.
0 245
Connections count
h [electronic resource] :
understanding gender and race differences in school-based problem behavior during adolescence /
by Raymond C. Santa Lucia.
3 246
Understanding gender and race differences in school-based problem behavior during adolescence
260
[Tampa, Fla.] :
University of South Florida,
2004.
502
Thesis (Ph.D.)--University of South Florida, 2004.
504
Includes bibliographical references.
500
Includes vita.
516
Text (Electronic thesis) in PDF format.
538
System requirements: World Wide Web browser and PDF reader.
Mode of access: World Wide Web.
Title from PDF of title page.
Document formatted into pages; contains 124 pages.
520
ABSTRACT: Utilization of a large, diverse sample provided a rare opportunity to advance our understanding of gender, race, and socioeconomic differences in school-based problem behavior. Yearly assessment of discipline referrals and suspensions received within the school context from 5th- through 11th-grade, as well as assessment of school dropout, provided an opportunity to examine these issues through an extended prospective longitudinal design. Results highlight the middle school transition as a time when discipline referrals and suspensions increase markedly, while student reports of connections to others, motivation, and optimism decline sharply. Results indicate that boys, African-American students, and students from low socioeconomic status backgrounds report lower levels of connections to others, motivation, and optimism in 5th-grade. Boys, African-American students, and students from low socioeconomic status backgrounds also receive more discipline referrals and suspensions from 5th-grade onward and are more likely to experience dropout. However, regardless of demographic group membership, students who report stronger connections to others, motivation, and optimism in 5th-grade receive fewer referrals and suspensions from 5th- through 11th grade, and are much less likely to dropout of school than are students who report lower levels of connections, motivation, and optimism in 5th-grade. These results highlight the need to address students' sharp declines in functioning across the middle school transition through both ecological and person-centered prevention and school restructuring efforts. Results also highlight the utility of movement away from a static, demographic based understanding of problem behavior toward a clearer understanding of person and environment factors that may underlie both between and within demographic group differences in outcomes. Placing emphasis upon factors that are potentially amenable through school based prevention efforts considerably increases the likelihood that all of our nation's children are provided with equal opportunity to achieve their fullest potential.
590
Adviser: Gesten, Ellis
653
discipline.
referrals.
suspensions.
dropout.
longitudinal.
690
Dissertations, Academic
z USF
x Psychology
Doctoral.
773
t USF Electronic Theses and Dissertations.
4 856
u http://digital.lib.usf.edu/?e14.293



PAGE 1

Connections Count: Understanding Gender And Race Differences in School Based Problem Behavior During Adolescence by Raymond C. Santa Lucia A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Phil osophy Department of Psychology College of Arts and Sciences University of South Florida Major Professor: Ellis Gesten, Ph.D Wally Borman, Ph.D. Bob Friedman, Ph.D. Vicky Phares, Ph.D. Charles D. Spielberger, Ph.D. Date of Approval: January 23, 2004 Keywords: (discipline, referrals, suspensions, dropout, longitudinal) Copyright 2004, Raymond C. Santa Lucia

PAGE 2

Dedication This paper is dedicated to my grandparents; Charles Santa Lucia, Josephine Santa Lucia, John Mazza, and Florence Mazza, for overcoming the challenges faced by immigrants to America to make a better life for their families. They are all in my heart. It is dedicated to my sister Tammy, for overcoming her challenges and making a big difference in the lives of many children and fam ilies through her work as a Licensed Clinical Social Worker. It is dedicated to my friends; Ryan Russon, Octavio Salcedo, Jim Dean, Demy Kamboukos, Paul Espinosa, Jennifer Welch, and Jill Welch in particular. They have taught me so much about the meaning o f friendship, life, and love. It is dedicated to my mentors; Mr. de Santos, Dr. Warren, Terry Champlain, Brad Peterson, Ellis Gesten, Vicky Phares, Maria de Perczel, Charles Furman, Scott Van de Putte, and Lynn Dowell. I have been incredibly fortunate to w ork alongside such talented, supportive, and caring people. It is dedicated to children from high risk backgrounds, particularly children diagnosed with muscular dystrophy. They have been in my heart throughout and they are my inspiration. Above all, this paper is dedicated to my mother and father. Being lucky enough to watch two people from high risk backgrounds persist tirelessly to improve the lives of their children, of those in their community, and of each other would have a profound effect on anyone. Having them in my heart has made all the difference.

PAGE 3

Acknowledgements I truly thank my colleagues in the Pinellas County School system, especially Octavio Salcedo, Dr. Behrokh Ahmati, Jim Dean, and Dr. Steve Iachini for making this study possible. I gr atefully acknowledge the hundreds of teachers, thousands of students, and many other Pinellas School District personnel who have taken time from their busy schedules to provide the data included in this study. I sincerely thank Drs. Borman, Friedman, Phare s, and Spielberger for their professionalism and feedback throughout the dissertation process. Above all, I sincerely thank my mentor, Dr. Ellis Gesten, for all of the guidance and support he has given me through many projects over many years too numerable to mention.

PAGE 4

i Table of Contents List of Tables iv List of Figures vi Abstract vii Introduction 1 Systems Perspective 3 Connection to Parents 4 Connection to Peers 5 Connection to School 6 Developmental Perspective 7 Person Focused Perspective 9 Et hn icity and Socioeconomic Status 10 Gender 11 Hypotheses 13 Method 14 Participants 14 Measures 14 School Adjustment Survey 14 Connection to Parents Scale 15 School Discipline Records 15 Student Dropout Status 16 Demographic Data 16 Procedure 17 Data Ana lysis 17 Factor Analysis of the Student Adjustment Survey 17 Transition MANOVA 17 Cluster Analysis 18 Risk Group Chi S quare Analyses 18 Problem Behavior MANOVAs 19 Dropout Chi Square Analyses 19 Results 20 Factor Analysis of the Student Adjustment Survey 20 Transition MANOVA 22 Main Effects 22

PAGE 5

ii Interaction Effects 24 Cluster Analysis 24 R isk Group Chi Square Analyses 26 P roblem Behavior MANOVAs 27 Main Effects 28 Main Effects x Time 33 Two Way Interactions involving Ethnicity, SES, and Gender 36 Ethnicity x SES (x Time) 36 Ethnicity x Gender (x Time) 39 SES x Gender (x Time) 41 Two Way Interactions involving Risk 41 Risk x SES (x Time) 41 Risk x Ethnicity (x Time) 44 Risk x Gender 46 Three way Interactions involving Risk 48 Ethnicity x SES x Risk (x Time) 4 8 Ethnicity x Gender x Risk 50 SES x Gender x Risk 52 Ethnicity x SES x Gender x Risk 53 Dropout Chi Square Analyses 5 8 Discussion 60 Factor Analysis of the Student Adjustment Survey 60 Student Adjustment Survey and Connection to Parents Scale: Main Effec ts 61 Gender 61 Ethnicity and SES 62 The Middle School Transition 64 Student Adjustment Survey and Connection to Parents Scale: Interaction Effects 65 Gender x Time 65 SES x Time 66 Cluster Analysis 66 Problem Behavior MANOVAs 68 Main Effects of Gender, E thnicity, and Socioeconomic Status 68 Main Effects of Time 71 Interaction Effects 73 Main effects x Time 73 Ethnicity x SES (x Time) 74 Ethnicity x Gender (x Time) 75 SES x Gender (x Time) 76 Risk x SES (x Time) 77 Risk x Ethnicity (x Time) 78 Ethnicity x SES x Gender x Risk 78 Dropout Chi Square Analyses 80 Limitations 82 Conclusions 86

PAGE 6

iii References 90 Appendices 103 Appendix A: Conn ection to Parents Scale 104 Appendix B: Correlations 105 Appendix C: Facto r Analysis (with all loadings) 112 About the Aut hor End Page

PAGE 7

iv List of Tables Table 1 Student Adjustment Survey 21 Table 2 Student Adjustment Survey: Main Effects 23 Table 3 Student Adjustmen t Survey: Main Effect of TIME 23 Table 4 Student Adjustme nt Survey Interac tion Effects 25 Tab le 5 Cluster Analysis Results 26 Tab le 6 Gender x Risk Chi Square 26 Table 7 Ethnicity x Risk Chi Square 27 Table 8 SES x Risk Chi Square 27 Table 9 Problem Behavior Main Effects 5 th Grade 29 Table 10 Problem Behavior Main Effects 6 th Grade 29 Table 11 Problem Behavior Main Effects 7 th Grade 30 Table 12 Problem Behavior Main Effects 8 th Grade 30 Table 13 Problem Behavior Main Effects 9 th Grade 31 Table 14 Problem Behavior Main Effects 10 th Grade 31 Table 15 Problem Behavior Main Effect s 11 th Grade 32 Table 16 Mean Changes in Problem Behavior Ac ross Middle School Transition 32 Table 17 Main Effects x Time 2 Way Interac tions for Referrals 34 Table 18 Main Effects x Time 2 Way Interactions for Suspensions 35 Table 19 Ethnicity x SES (x Tim e) Interactions 37

PAGE 8

v Table 20 Ethnicity x Gender (x Time) Interactions 40 Table 21 SES x Gender (x Tine) Interactions 41 Table 22 Risk Group x SES (x Time) Interactions 43 Table 23 Risk x Et hnicity (x Time) Interactions 45 Table 24 Risk x Gender Interactio ns 47 Table 25 Ethnicity x SES x Risk (x Time) Interactions 49 Table 26 Ethnicity x Gender x Risk Interactions 51 Table 27 SES x Gender x Risk Interactions 53 Table 28 Gender x STATUS Chi Square 59 Table 29 Ethnicity x STATUS Chi Square 59 Table 30 SES x S TATUS Chi Square 59 Tabl e 31 Risk x STATUS Chi Square 60

PAGE 9

vi List of Figures Figure 1. Ethni city x Time (Total Referrals) 35 Figure 2a. Ethnicity x SES (5 th Grade) 38 Figure 2b. Ethnicity x SES (6 th Grade) 38 Figure 3. Ethnicity x SES x Gender x Risk for Total Referrals 7 th Grade 55 Figure 4. Ethnicity x SES x Gender x Risk for Classroom Referrals 7 th Grade 56 Figure 5. Ethnicity x SES x Gender x Risk for Total Referrals 8 th Grade 57 Figure 6. Ethnicity x SES x Gender x Risk for Classroom Referrals 8 th Grade 58

PAGE 10

vii Connections Count: Understanding Gender and Race Differences in School Based Problem Behavior during Adolescence Raymond C. Santa Lucia ABSTRACT Utilization of a large, diverse sample provided a rare opportunity to adva nce our understanding of gender, race, and socioeconomic differences in school based problem behavior. Yearly assessment of discipline referrals and suspensions received within the school context from 5 th through 11 th grade, as well as assessment of schoo l dropout, provided an opportunity to examine these issues through an extended prospective longitudinal design. Results highlight the middle school transition as a time when discipline referrals and suspensions increase markedly, while student reports of c onnections to others, motivation, and optimism decline sharply. Results indicate that boys, African American students, and students from low socioeconomic status backgrounds report lower levels of connections to others, motivation, and optimism in 5 th grad e. Boys, African American students, and students from low socioeconomic status backgrounds also receive more discipline referrals and suspensions from 5 th grade onward and are more likely to experience dropout. However, regardless of demographic group memb ership, students who report stronger connections to others, motivation, and optimism in 5 th grade receive fewer referrals and suspensions from 5 th through 11 th

PAGE 11

viii grade, and are much less likely to dropout of school than are students who report lower levels of connections, motivation, and optimism in 5 th grade. These results highlight the need to address students sharp declines in functioning across the middle school transition through both ecological and person centered prevention and school restructuring e fforts. Results also highlight the utility of movement away from a static, demographic based understanding of problem behavior toward a clearer understanding of person and environment factors that may underlie both between and within demographic group diff erences in outcomes. Placing emphasis upon factors that are potentially amenable through school based prevention efforts considerably increases the likelihood that all of our nations children are provided with equal opportunity to achieve their fullest po tential.

PAGE 12

1 Introduction The goal of the present study is to further our understanding of school based problem behavior in terms of students connections to parents, peers, and school during the early adolescent period. This study is unique in that reports of more than four thousand students permit examination of problem behavior outcomes in terms of gender, ethnicity, socioeconomic status, and risk status across the seven year timeframe of the study. Doing so provides an opportunity t o advance our understanding of critical issues concerning the role of gender, ethnicity, and socioeconomic status in relation to problem behavior outcomes. Connections to others across systems are highlighted as potentially key to the risk and protection o f students both across and within demographic groups. The middle school transition is highlighted as a potentially pivotal turning point in development. The nature of what students learn in school is highlighted as potentially encompassing much more than a cademics. With respect to the mental health of our Nations children, no less an authority than the former Surgeon General of the United States has stated, in no uncertain terms, that we are experiencing a health crisis in this country (United States Dep artment of Health and Human Services, 1999; p. 1). Epidemiologic studies estimate that between 12% and 30% of school aged children in the United States experience moderate to serious mental health problems that interfere with their daily functioning (Verhulst & Koot, 1992; Weist, 1997) Only about one in five of these children receive services to

PAGE 13

2 address their difficulties (United States Department of Health and Human Services, 1999) Of particular concern is that boys, Af rican American children, and children of low socioeconomic status are at heightened risk for both the expression of problem behaviors during childhood and relative increases in problem behavior beginning in early adolescence (Dryfoos, 1990) Ultimately our failure to prevent the development of problem behavior patterns and school dropout results in disproportionate representation of males, African American individuals, and individuals of low socioeconomic status in our nations criminal jus tice system (Council on Crime in America, 1996) The prevalence of children at risk for such severe outcomes underscore the importance and urgency of treating and preventing mental disorders and of promoting health in our society (United States Department of Health and Human Services, 1999; p.1). Seminal research conducted by Richard Jessor and his colleagues introduced the concept of a problem behavior syndrome in which multiple problem behaviors clustered together among adolescents (Jessor & Jessor, 1977) Through this work, Jessor and colleagues focused attention upon the co occurrence of problem behaviors, which later led to research cumulatively suggesting that the co occurrence of such behaviors may be linked to s imilar underlying risk factors (Dryfoos, 1990) In her landmark review, Dryfoos (1990) outlined separate domains of risk associated with the later expression of multiple problem behaviors including delinquency, substance abuse, teen pregna ncy, and school dropout. These risk domains included age, expectations for education and school grades, general behavior, peer influence, parental role, and neighborhood quality (Dryfoos, 1990) While understanding of the manner through wh ich these risk factors are

PAGE 14

3 associated with problem outcomes has improved since the time of Dryfoos (1990) review, several gaps in our understanding remain (Hinshaw & Park, 1999) Systems Perspective Writing the concluding chapter of the H andbook of Disruptive Behavior Disorders Hinshaw and Park (1999) cited the need to expand research incorporating assessment of functioning across multiple systems within longitudinal investigations of the development of problem behavior (H inshaw & Park, 1999; Richters, 1997) The present study integrates risk factors across multiple systems into a holistic longitudinal design. Risk factors incorporated in the present study are directly analogous to those that prior research has identified as the strongest and most consistent correlates of problem behavior (Dryfoos, 1990) These include parental role, peer influence, and expectations for school success (Dryfoos, 1990) Factors identified as associated with t he development of problem behavior subsequent to Dryfoos (1990) review including teacher child interaction (Davis, 2001; Hughes, Cavell, & Jackson, 1999; Hughes, Cavell, & Willson, 2001; Pianta, 1999) as well as bonding to school (Marcus & Sanders Reio, 2001; Najaka, Gottfredson, & Wilson, 2001) and motivation to achieve (Jimerson, Egeland, Sroufe, & Carlson, 2000; Najaka et al., 2001) were also integrated into the multisystemic framework of the present s tudy.. Age, general behavior, and neighborhood quality have also demonstrated the ability to predict multiple problem outcomes (Dryfoos, 1990) Age is included in the present study, insofar as the rise in problem behavior at early adolesc ence highlights the developmental significance of examining students connections and attitudes during this

PAGE 15

4 time. General behavior is not assessed as a predictor in the present study, which is designed to move beyond prediction of later behavior from earli er behavior toward a broader multisystemic perspective. Neighborhood quality is often, though by no means always, associated with the influence of poverty. Socioeconomic status is examined in relation to problem outcomes in the present study. Taken togethe r, risk factors examined in the present study are grounded solidly in the literature linking multiple domains of risk to the future expression of multiple problem behaviors. Connection to Parents Parent child relations have long been viewed as a significan t influence upon the expression and development of problem behavior (Maccoby & Martin, 1983) Meta analytic findings have provided support for the existence of this relationship framed within a developmental perspective (Ro thbaum & Weisz, 1994) Specifically, Rothbaum and Weisz (1994) found that cross sectional relationships between parenting variables and behavioral problems were stronger for older children (6 to 15.5 years) than for younger children (10.5 months to 5 year s). The strongest relationships found involved preadolescent boys and their mothers (Rothbaum & Weisz, 1994). Prior work also indicates that students who drop out of school are likely to have parents who are less involved in their lives than do students w ho stay in school (Alpert & Dunham, 1986; Ekstrom, Goertz, Pollack, & Rock, 1986; Hanson & Ginsburg, 1988; Rumberger, Ghatak, Poulos, Ritter, & Dornbusch, 1990) Dropouts are also more likely to have relationships with parents characterize d by less warmth, less communication, and higher levels of punitive forms of punishment (Bachman, Green, & Wirtanen, 1971; Ekstrom et al., 1986) and tend to have a more permissive parenting style (Rumberger et

PAGE 16

5 al., 1990) Taken together, parent child relations have consistently been implicated in the development of problem behavior. Connection to Peers Within the peer domain, studies have found that experiencing rejection by peers during childhood is associated with the l ater expression of problem behavior (Coie, Lochman, Terry, & Hyman, 1992; Kupersmidt, Coie, & Dodge, 1990; Parker & Asher, 1987; Tremblay, LeBlanc, & Schwartzman, 1988) The consensus of current studies indicates that rejection by prosocia l peers in childhood is associated with having behaviorally deviant friends beginning in early adolescence (Dishion, 1990; Dishion, Patterson, Stoolmiller, & Skinner, 1991) Evidence suggests that groups of children displaying problem beha vior are more likely to affiliate with peers outside of the school setting (Dryfoos, 1990) Prior work indicates that affiliation with deviant peers is associated with the development and progression of problem behavior through adolescence (Dishion, 1990; Dishion et al., 1991; Patterson, 1993; Vitaro, Tremblay, & Bukowski, 2001) Similarly, prior work has indicated that peer rejection in childhood is associated with the establishment of networks of behaviorally deviant pe ers, who are more likely to drop out of school (Kupersmidt et al., 1990; Parker & Asher, 1987) Taken together, this research presents a picture in which the development of problem behavior is associated with rejection by a majority of pee rs within the school environment throughout development. The consequence of which appears to be gradual detachment from school in favor of affiliation with others at heightened risk to engage in problem behavior.

PAGE 17

6 Connection to School In the present stu dy, connection to teachers, bonding to school, school expectancies, and academic motivation are examined in relation to the development of problem behavior. With respect to teacher child relations, existing evidence indicates that during childhood, engagem ent in problem behavior in the classroom setting is associated with poor teacher child bonding relations in which children feel less supported (Dodge, Coie, & Brakke, 1982; Marcus & Sanders Reio, 2001; Pianta, 1999) A growing body of work has suggested that childrens connection to teachers is associated with the progression of problem behavior expressed within the school setting through adolescence (Marcus & Sanders Reio, 2001; Pianta, 1999) Similarly, limited longitudin al evidence suggests that dropout is associated with poor relationships between teachers and students (Rutter, 1978; Rutter, Maughan, Mortimore, & Ouston, 1979; Werner, 1995) Student attitudes toward school have also been associated with the expression of school based problem behavior and school dropout. A recent meta analysis examined the association between changes in school based problem behavior and changes in bonding to school resulting from implementation of school based prevention efforts (Najaka et al., 2001) Bonding to school was operationalized as encompassing liking school, possessing motivation to achieve, and having expectations for success in school. Across studies, positive changes in bonding to school were accompanied by reductions in problem behavior within the school setting (Najaka et al., 2001) Similarly, studies have indicated that school dropouts like school less (Ekstrom et al., 1986; Marcus & Sanders Reio, 2001; Rum berger, 1987) are less motivated to achieve (Jimerson et al., 2000; Rumberger, 1987; Rumberger et al., 1990) and have lower educational aspirations

PAGE 18

7 (Hanson & Ginsburg, 1988; Rumberger, 1987; Wehlage & Rutter, 1986) than students who stay in school. Developmental Perspective From a developmental standpoint, early adolescence is known as the time when problem behaviors begin a steady increase in prevalence that continues through high school (Donovan & Jesso r, 1985) For most students in our nations schools, early adolescence is also coupled with the transition into middle school. The middle school transition has been identified as a key turning point in development (Carnegie Council on Adolescent Develop ment, 1989). Drastic changes in the structure of schooling, combined with the onset of adolescence have been associated with declines across parent, peer, and school systems (Eccles et al., 1993) The theoretical foundation for Eccles and her colleagues work rests upon their application of a stage environment fit perspective (Eccles et al., 1993) This approach is grounded in the person environment fit theory, which proposes negative consequences for individuals when they are in environments that do not fit well with their needs (Lewin, 1935) Drawing upon the early work of Lewin (1935), Hunt (1975) proposed a developmental variant of the person environment fit perspective and suggested applications of this model to students education. Specifically, from the standpoint of this model, teachers should provide structure consistent with students developmental level of maturity, while providing challenges that serve as opportunities for students to move toward higher levels of cognitive and social sophistication (Hunt, 1975) Eccles et al. (1993) suggested that qualities of the middle school classroom environment represent a mismatch with the developmental needs of early adolescents.

PAGE 19

8 These qua lities include evidence suggesting that compared to elementary school, junior high school classrooms are characterized by a greater emphasis on teacher control and discipline, by fewer opportunities for student decision making, by less positive teacher stu dent relationships, and by more competitive grading practices. Literature reviews have documented declines in motivation, teacher child relations, bonding to school, and academic expectations for success across the transition (Eccles et al. 1993; Midgley & Edelin, 1998; Roeser, Eccles, & Strobel, 1998) Further, power struggles associated with an increased need for autonomy have been associated with declines in the quality of parent child relations during early adolescence (Eccles et al., 1993). Early adolescence also appears to be the time when association with deviant peers solidifies for children on a developmental path toward future expression of problem behavior (Dishion, 1990) Declines across these domains coupled wi th increases in problem behavior outcomes strongly support the need for primary prevention efforts designed to address these normative declines in functioning found across the transition to middle school (Carnegie Council on Adolescent Deve lopment, 1989; Durlak & Wells, 1997; Felner et al., 2001; Felner, Jackson, & Kasak, 1997) Equally important is the need to develop a means of identifying students at the highest levels of risk to engage in elevated levels of problem behavior following th e transition to middle school. Ideally, it is important to identify these students while they are in elementary school and then to provide selective prevention efforts intended to decrease the likelihood that these students will engage in elevated levels o f problem behavior following the transition to middle school (Durlak & Wells, 1998)

PAGE 20

9 Person Focused Perspective In addition to calling for expansion of more holistic longitudinal examinations of problem behavior development, Hinshaw and Pa rks (1999) critique also centered upon the need to break set by employing research methods that provide alternatives to the predominant variable centered approach. Drawing upon Richters (1997) critique of variable focused research methods, Hinshaw and P ark (1999) argued strongly in favor of the use of person centered methods such as cluster analysis (Magnusson & Bergmann, 1988) Person focused methods are useful in that they identify more homogenous subgroups based upon underlying etiolo gy. Consequently, this approach is ideally suited to the task of identifying a subgroup of students at highest risk to engage in increased levels of problem behavior based upon their standing across multiple risk domains (Magnusson & Bergma nn, 1988) Consistent with this approach and the review of risk factors above, students reporting the poorest connections to parents, peers, and school in fifth grade would be expected to demonstrate the largest increases in problem behavior across the t ransition to middle school. These high risk students would also be expected to demonstrate elevated levels of disruptive behavior across middle school and high school. They would also be expected to drop out of school at higher rates than students who are not placed in the high risk group. These findings would support the need to direct selective prevention efforts toward this subgroup of students prior to and following entry into middle school (Durlak & Wells, 1998)

PAGE 21

10 Ethnicity and Soci oeconomic Status Importantly, a person focused approach may enhance our understanding of ethnic and socioeconomic status differences in problem behavior. African American students and those from low socioeconomic status backgrounds engage in higher level s of disruptive and delinquent behavior than do students from Caucasian and higher socioeconomic status backgrounds (Council on Crime in America, 1996; Elliot, 1994; Jones & Krisberg, 1994; United States Department of Health and Human Servi ces, 1999) African American students and those from low socioeconomic status backgrounds are also more likely to drop out of school (McLoyd, 1998; Tucker & Herman, 2002) While differences in problem behavior outcomes are clear, our unde rstanding of factors associated with these ethnic and socioeconomic differences is limited (Hinshaw & Park, 1999; Jeynes, 2002; Yung & Hammond, 1997) Existing research suggests that African American students and those from low socioecon omic status backgrounds encounter higher levels of risk factors across parent, peer, and school based systems. African American and low socioeconomic status students are more likely to experience punitive interactions with parents and are less likely to ex perience positive interactions (Borkowski, Ramey, & Bristol Power, 2002; McLoyd, 1998; Yung & Hammond, 1997) They are also more likely to perceive interactions with teachers as punitive and characterized by less reinforcement for successf ul performance (McLoyd, 1998; Polite, 1994) Low socioeconomic status and African American students are also more likely to feel detached from school (Hirschi, 1969; McLoyd, 1998; Steele, 1997) Limited research suggests t hat detachment from school may be associated with declines in attachment to prosocial peers in the school

PAGE 22

11 context (McAdoo & McAdoo, 1985) While socioeconomic and ethnic differences in motivation to succeed in school are unclear, low SES a nd African American ethnicity are both associated with increased levels of negative expectations for school success (Tucker & Herman, 2002) Taken together, these results suggest that African American and students from lower socioeconomic status backgrounds engage in higher rates of problem behavior and present at higher levels of risk across domains found to be associated with problem behavior. A consequence of reliance upon findings derived from samples of middle class, Caucasian childre n in longitudinal investigations of problem behavior is that our understanding of the how risk is related to problem behavior outcomes across and within ethnic groups is lacking (Loeber & Farrington, 1997; Loeber & Hay, 1997; Loeber & Stout hamer Loeber, 1998; Tucker & Herman, 2002; Yung & Hammond, 1997) A primary goal of the present study is to move toward integration of these two bodies of research. Through doing so we may better understand both between and within group differences in pro blem behavior based upon ethnicity and socioeconomic status. Gender There is broad consensus that boys engage in higher rates of disruptive behavior than do girls beginning in early childhood and continuing through adolescence at a rate of 4:1 (Giordano & Cernkovich, 1997) In contrast, dropout levels do not appear to differ as a function of gender (Ketterlinus & Lamb, 1994; Lerner & Galambos, 1998) Gender differences in rates of disruptive behavior have been used as ju stification for including only males in the majority of longitudinal studies examining disruptive behavior patterns (Giordano & Cernkovich, 1997; Loeber & Farrington, 1997) Consequently, our

PAGE 23

12 understanding of factors associated with the de velopment of problem behavior among girls is limited (Giordano & Cernkovich, 1997; Loeber & Stouthamer Loeber, 1998) Existing research suggests that boys encounter higher levels of risk factors across parent, peer, and school based domai ns relative to girls. Existing data suggest that parental relations with boys are characterized by higher levels of conflict (Rothbaum & Weisz, 1994) which in turn is associated with higher levels of problem behavior. However, longitudina l data examining these gender differences is needed (Giordano & Cernkovich, 1997; Loeber & Hay, 1997; Loeber & Stouthamer Loeber, 1998; Rothbaum & Weisz, 1994) Studies have also suggested that elementary school girls report having more su pportive relationships with their teachers (Davis, 2001; Hamre & Pianta, 2001; Wentzel, 2002) and liking of school (Murray & Greenberg, 2000) than do boys. Studies indicate that boys are at higher risk for peer rejection i n the elementary school years relative to girls (Dishion, 1990; Dishion et al., 1991) Research has also indicated that girls report higher levels of achievement motivation (Goodenow, 1993) and educational aspirations (Wentzel, 1997) during the elementary school years. These findings suggest that gender differences in rates of problem behavior may be associated in part with a higher ratio of boys being classified as high risk based upon their standing acro ss parent, peer, and school based systems in elementary school. Importantly, within group differences must also exist in levels of risk for both boys and girls. Through examination of within group differences, investigators may better understand the nature of factors associated with problem behavior outcomes for both boys and girls alike.

PAGE 24

13 Hypotheses In accord with the preceding review, students connections to parents, peers, teachers, and school, as well as motivation, were expected to decline across th e transition to middle school, while negative expectations were expected to rise (Eccles et al., 1993) Problem behavior was expected to increase across the middle school transition (Dryfoos, 1990) African American studen ts, boys, and students from low socioeconomic status backgrounds were expected to report lower levels of attachment to parents, peers, teachers, and school, as well as higher levels of negative expectations. Analyses of differences in motivation were explo ratory. African American students, boys and students from low socioeconomic status backgrounds were also expected to engage in higher rates of disruptive behavior and to dropout of school at higher rates than Caucasian students, girls, and students from hi gher socioeconomic status backgrounds. In accord with the person focused perspective of the present study, results of cluster analysis (Magnusson, 2000; Magnusson & Bergmann, 1988) were expected to place African American students, boys, and students from low socioeconomic status backgrounds in a high risk group or groups at higher rates relative to Caucasian students, girls, and students from higher socioeconomic status backgrounds. Students within higher risk groups were expected to enga ge in higher levels of disruptive behavior and dropout relative to students in lower risk groups. Despite higher rates of African American students, boys, and students from low socioeconomic status background in the high risk group, students from these dem ographics were expected to be represented in the average and low risk groups. These students were expected to engage in lower rates of future problem behavior

PAGE 25

14 relative to their high risk counterparts. This would indicate within demographic group variabilit y in terms of outcomes as a function of risk status. Method Participants Participants include 4,695 students for whom survey data were collected in both fifth grade and sixth grade. The sample is drawn from a large (101,000) geographically diverse school district in central Florida. The sample is 79.7% Caucasian (1806 boys / 1937 girls), 15.1% African American (359 boys, 350 girls), 2.8% Asian (54 boys / 76 girls), 2.3% Hispanic (48 boys / 58 girls), and 0.1% classified as Other (4 boys / 3 girls). Survey data were collected as part of a larger longitudinal study initiated and conducted by school district administrators and personnel. District personnel obtained passive parental consent and child assent. Five hundred children (9.6%) missing more than 20% o f their data are not included in the study. The special education population was underrepresented in this study (16.2%) relative to the percentage of special education students in the entire fifth grade class (22.1%) during the first year of the study. Und errepresented subgroups of special education students included students classified as EH (1.2% vs. 2.6%), SLD (11.6% vs. 12.5%), and SED (0.3% vs. 0.7%). Measures School Adjustment Survey The School Adjustment Survey is a self report scale consisting of 33 items assessing students motivation, achievement expectations, connection to school, connection to teachers, connection to peers, and connection to parents. Items were rationally selected by district personnel in accord with areas of interest to the di strict.

PAGE 26

15 Students are asked to state the degree to which they agree with each of the 33 statements on a five point scale ranging from (0) Strongly Disagree to (4) Strongly Agree. The Student Adjustment Survey was factor analyzed as part of the present s tudy. Results are presented in the results section below. Connection to Parents Scale The Connection to Parents scale is a self report scale consisting of six items (see Appendix A) assessing positive parent child interaction. District personnel rational ly selected scale items. Students are asked to state the degree to which they agree with each of the six statements on a five point scale ranging from (0) Strongly Disagree to (4) Strongly Agree. The mean of these six items was calculated for each part icipant. Alpha reliabilities for the present sample were .68 for fifth grade and .57 for sixth grade. School Discipline Records School discipline records for all students present in the study were obtained for each year from 1994 1995, when the students were in fifth grade, to the 2000 2001 school year. For each student present during each year of the study, the total number of school discipline referrals received was computed. The total number of violence related referrals and the total number of classro om related referrals was computed separately for each student present during each year of the study. Violence related referrals include battery against another student, battery against an adult, fighting, weapons possession, sexual battery, and the use of threats/intimidation against other students. The school district uses the following codes in conjunction with classroom related referrals: class disruption, lack of cooperation, use of profane or obscene language, disrespect/defiance/threats, and repeated misconduct. The total number of in school

PAGE 27

16 suspensions and out of school suspensions received by each student present during each year of the study was also calculated. Student Dropout Status The dropout status of each student was determined as of the con clusion of the 2000 2001 school year. Students will be classified as (1) Enrolled in High School, (2) Enrolled in Adult Education, (3) Dropout, or (4) Moved out of District. Although the district codes some students as having dropped out of school, there a re several codes used and situations in which students that have dropped out of school are not classified as such. For the purpose of this study, students were coded as having dropped out of school if they were listed as: (1) Dropout, (2) Did Not Enter, (3 ) Non Attendance, (4) Whereabouts Unknown, and (5) Other. Additionally, students who enrolled in adult education prior to graduation and were listed as inactive, meaning that they were not attending adult education classes were classified as having dropped out of school. Demographic Data Inclusion of gender and ethnicity were considered central to the present study, as was inclusion of an estimate of socioeconomic status (SES). Receipt of either free or reduced cost lunch in school in fifth grade was use d as a rough estimate of students socioeconomic status. Students were placed into groups in which they were either classified as receiving or not receiving free or reduced cost lunch. Incorporation of this variable in the present study was considered esse ntial due to sizable differences across ethnicities in rates of receiving free or reduced lunch. In the present sample, 7.4% of students received free lunch, while 29.7% of students received lunch at a reduced cost. Free and reduced lunch were received by significantly higher percentages of African

PAGE 28

17 American (79%), Hispanic (70.8%), and Asian (55.4%) students relative to Caucasian (27.5%) students (?(6) = 856.34, p < .001). Procedure School personnel conducted whole class administration of surveys during t he spring of the 1994 1995 and 1995 1996 school years. Scale items were read aloud to all participants. School district personnel coded survey data, and compiled discipline and dropout data. Data Analysis Factor Analysis of Student Adjustment Survey A principal components factor analysis with varimax rotation was performed for the 5 th grade administration of the School Adjustment Survey. A five factor solution consisting of Connection to Teachers, Connection to Peers, Connection to School, Motivation, a nd Negative Expectations was predicted. Other possible factor analytic solutions were explored as well. Approximate factor scoring was used. Sixth grade subscales were formed based upon the fifth grade solution chosen. As such, identical items were used to form subscales in both 5 th grade and 6 th grade. This permitted examination of stability or change in subscale means over time. Alpha reliabilities were calculated for all subscales. The mean of each subscale for each participant was computed. Transition M ANOVA A four way, 2 (Gender) x 2 (Ethnicity) x 2 (SES) x 2 (Time) MANOVA was conducted to examine if differences exist in mean levels of Connection to Parents, Connection to Peers, Connection to Teachers, Connection to School, Motivation, and

PAGE 29

18 Negative Exp ectations across levels of these four independent variables. Only Caucasian and African American participants were included in these analyses as concerns existed regarding the representativeness of Asian and Hispanic participants given limited sample size. To decrease the potential for Type I error in accord with the MANOVA design, effects for individual dependent variables were only examined if the omnibus F test for an effect was significant, indicating that the effect was significant for at least one of the six dependent variables investigated. Cluster Analysis Cluster analysis places participants into non overlapping subgroups who each share similar scores on a set of continuous variables. The clusters are characterized by their mean scores on each of t he variables used in the cluster analysis. For the present study, cluster analysis was used to identify fifth grade students at risk for adjustment problems following the transition to middle school in 1995 1996 through the 2000 2001 school year. Participa nts fifth grade scores on each of the Student Adjustment Survey subscales and the Connection to Parents scale were included in the analysis. An agglomerative hierarchical analysis using Wards method was used to form groups at different levels of risk bas ed on the scales used. Risk Group Chi Square Analyses Following the cluster analysis, three separate chi square analyses were used to determine if placement in risk groups varied by gender, ethnicity, and SES. Frequencies of participants classified as ( 1) High Risk, (2) Average Risk, and (3) Low Risk were compared against expected frequencies in each analysis.

PAGE 30

19 Problem Behavior MANOVAs Six separate MANOVAs were computed in the present study. One five way 2 (Gender) x 2 (Ethnicity) x 2 (SES) x 3 (Ri sk) x 2 (Time) MANOVA examined mean differences in total discipline referrals, violence referrals, classroom referrals, in school suspensions, and out of school suspensions received by this sample in fifth and sixth grade. This first MANOVA was followed by a series of five MANOVAs in which mean differences in these same dependent variables were examined within a 2 (Gender) x 2 (Ethnicity) x 2 (SES) x 3 (Risk) factorial design for each of the remaining five years of the study (grades 7 through 11). These ana lyses were performed separately for each year as potentially selective attrition in these latter years of the study would have resulted in missing data that would have excluded these students from a single analysis incorporating all seven years of the stud y. To control for Type I error, a conservative procedure was used to determine the significance of any single effect. To be considered significant, the omnibus F test for a given effect was first checked for significance at the .05 level. If significant, individual F tests for each of the five dependent variables were then examined for significance at the .05 level. If an effect was then significant for a particular dependent variable in a given year, conservative Bonferroni post hoc tests were then perfo rmed to determine whether individual means were significantly different at the .05 level, controlling for family wise error rate. Dropout Chi Square Analyses Four separate chi square analyses were used to examine differences in dropout status as of the completion of the 2000 2001 school year across levels of Gender,

PAGE 31

20 Ethnicity, SES, and Risk status. Frequencies of participants classified as (1) Enrolled in High School, (2) Enrolled in Adult Education, (3) Dropout, or (4) Moved out of District were compar ed against expected frequencies in each analysis. Results Factor Analysis of Student Adjustment Survey Multiple principal components and principal axis factor solutions of the fifth grade survey were examined to determine which solution best fit the data Consistent with prediction, the five factor principal components solution with varimax rotation with a .45 cutoff appeared to provide the best fit to the data (see Table 1). This solution accounted for 42% of the total variance. This solution was used to create subscales for both fifth and sixth grade administrations of the survey to allow for comparison of mean differences across time for subscales comprised of identical items. Factors, with alpha reliability coefficients for fifth and sixth graders resp ectively, were labeled Connection to Teachers (.78/.74), Connection to School (.78/.71), Connection to Peers (.69/ .11), Motivation (.55/.60), and Negative Expectations (.61/.57). The Connection to Peers subscale in sixth grade was the only subscale tha t did not manifest an acceptable level of internal reliability. Items with a positive tone (e.g. A student can be himself/herself and still be accepted by other students in this school) and negative tone (e.g. Making friends is very difficult in this sc hool) did not show a strong negative correlation.

PAGE 32

21 TABLE 1 Student Adjustment Survey a Factor Loading .65 .65 .64 .53 .53 .51 .48 Factor Loading .67 .66 .59 .58 .54 .51 Factor Loading .72 -.64 -.63 -.56 .46 Factor Loading .69 .63 .54 .46 Factor Loading .52 .51 .47 .46 Connection to Teachers 10. Most teachers like my friends and me. 7. I think my teachers care about me. 9. My teachers often get to know me well. 11. I care about what most of my teachers t hink about me. 12. Some teachers would choose me as one of their favorite students. 25. I feel that I can go to my teachers for advice or help with schoolwork. 26. I feel that I can go to my teachers for advice and help with non-school work. Connection to School 13. I like school. 17. I feel a sense of school spirit. 22. School is important to me. 21. I feel like I am learning a lot in school. 23. I believe I am learning important things in school. 20. Discipline is fair at this school. Connection to Peers 5. Most students include me in their activities. 16. Other kids in my class have more friends than I do. 2. Making friends is difficult at this school. 6. I always seem to be left out of important school activities. 4. A student can be him/herself and still be a part of this school. Motivation 32. Education is important for success in life. 34. I think I will go to college. 29. I try as hard as I can to do my best in school. 31. It bothers me when I don't do something well. Negative Expectancies 27. Most of my teachers don't really expect good work from me. 28. I don't care how well I do in school. 14. My teachers don't pay much attention to me. 18. I don't feel safe at school. Items that did not load on any factor include: 1. Students usually get along well with each other in this school. 3. I am in the wrong group to feel a part of this school. 8. Teachers are not usually available before class to talk with students. 15. I get a lot of encouragement at my school. 19. I have friends who are of different racial and ethnic backgrounds at this school. 24. I liked school more last year than I do this year. 33. I feel prepared for middle school. a.

PAGE 33

22 Transition MANOVA Main Effects School Adjustment Survey MANOVA results for main effects are highly consistent across all six dependent variables. For each variable, the main effects of Gender (see Table 2), Ethnicity (see Table 2), SES (see Table 2) and Time (see Table 3) were all significant with only one exception. The exception being that a significant main effect of Gender was not found for the Connection to Parents dependent variable. Wit h this exception, 5th grade students, Girls, Caucasian students, and Regular Lunch students reported higher levels of Connection to Teachers, Connection to School, Connection to Parents, and Motivation. Fifth grade students, Girls, and Regular Lunch studen ts also reported higher levels of Connection to Peers, while African American students reported higher levels of Connection to Peers than did Caucasian students. Fifth grade students, Girls, Caucasian students, and Regular Lunch students also reported sign ificantly lower mean Negative Expectations at the main effect level.

PAGE 34

23 TABLE 2 Student Adjustment SurveyMain Effects 2.62 (.64) 2.74 (.60) 9.30** 2.70 (.62) 2.55 (.63) 10.59** 2.73 (.61) 2.58 (.63) 12.24*** 2.64 (.65) 2.85 (.59) 29.55*** 2.77 (.63) 2.63 (.62) 5.07* 2.80 (.61) 2.65 (.65) 15.52*** 2.38 (.56) 2.49 (.52) 10.15** 2.43 (.55) 2.47 (.51) 16.05*** 2.47 (.55) 2.38 (.53) 22.20*** 2.87 (.52) 2.92 (.47) 4.13* 2.92 (.48) 2.79 (.54) 11.59** 2.95 (.45) 2.81 (.55) 14.90*** 1.56 (.59) 1.35 (.55) 55.63*** 1.43 (.56) 1.57 (.63) 7.16** 1.40 (.55) 1.53 (.61) 17.94*** 2.83 (.48) 2.86 (.46) ns 2.87 (.46) 2.71 (.49) 21.37*** 2.90 (.44) 2.75 (.51) 21.32*** Connection to Teachers Connection to School Connection to Peers Motivation Negative Expectations Connection to Parents M (SD) Boys M (SD) F Girls GENDER M (SD) Caucasian M (SD) F African-American ETHNICITY M (SD) Regular M (SD) F Free/Reduced SES TABLE 3 Student Adjustment SurveyMain Effect of TIME 2.78 2.58 2.78 2.71 2.70 2.17 3.29 2.50 .86 2.04 3.09 2.59 (.75) (.88) (.81) (.87) (.78) (.57) (.64) (.75) (.75) (.80) (.66) (.68) 63.56*** 7.16** 610.50*** 971.58*** 1820.76*** 405.57*** M (SD) F (TIME) 5th 6th Connection to Teachers 5th 6th Connection to School 5th 6th Connection to Peers 5th 6th Motivation 5th 6th Negative Expectations 5th 6th Connection to Parents

PAGE 35

24 Interaction Effects Omnibus F tests indicated significant interaction effects for Time x SES ( F(6,4439 = 10.42, p < .001) and Time x Gender ( F(6,4439) = 7.51, p < .001). Follow up F tests indicated that the significance of these interaction effects varied across dependent variables. The Time x SES interaction (see Table 4) was significant for Connection to Peers, Motivation, Negative Expectations, and Connecti on to Parents. Connection to Peers, Motivation and Connection to Parents declined more and Negative Expectations increased more for Regular Lunch students relative to students who received Free/Reduced Lunch. The Time x Gender interaction (see Table 4) was significant for students Connection to Teachers, Connection to School, and Motivation. Girls showed a larger decline relative to boys in Connection to Teachers, Connection to School, and Motivation across the transition to middle school. Cluster Analysi s Cluster solutions resulting in three through seven clusters were generated using the agglomerative method. All solutions were compared in terms of their ability to generate theoretically meaningful and homogenous subgroups demonstrating discriminant val idity in relation to the outcomes considered. Consistent with prediction, the three cluster solution best met these criteria (see Table 5). High, Average, and Low Risk subgroups were clearly differentiated across each of the six subscales entered into the analysis. Further, reduction of standard deviation values in the Average and Low Risk subgroups relative to the Overall variability present prior to formation of subgroups

PAGE 36

25 indicated that increased homogeneity was achieved in these subgroups relative to the overall sample. Finally, examination of three to seven cluster groups in relation to outcomes indicated strong discriminant validity for the three cluster solution across the range of outcomes examined, while increasing the number of clusters did not pro vide solid evidence of increased discriminant validity. TABLE 4 Student Adjustment Survey Interaction Effects 2.84 (.74) 2.68 (.76) ns 2.89 (.69) 2.66 (.80) 3.94* 2.63 (.87) 2.48 (.91) 2.59 (.88) 2.57 (.89) 2.83 (.79) 2.71 (.82) ns 2.96 (.71) 2.59 (.86) 27.48*** 2.78 (.84) 2.59 (.90) 2.74 (.86) 2.68 (.87) 2.77 (.79) 2.59 (.77) 21.27** 2.76 (.75) 2.64 (.82) ns 2.17 (.57) 2.17 (.57) 2.22 (.56) 2.12 (.57) 3.36 (.58) 3.16 (.73) 3.93* 3.38 (.59) 3.20 (.69) 28.07*** 2.53 (.71) 2.46 (.82) 2.47 (.72) 2.54 (.79) .76 (.70) 1.02 (.80) 11.41** .74 (.71) .98 (.78) ns 2.03 (.76) 2.05 (.87) 1.95 (.78) 2.14 (.81) 3.19 (.62) 2.93 (.69) 21.48*** 3.13 (.64) 3.05 (.67) ns 2.60 (.65) 2.57 (.73) 2.58 (.67) 2.60 (.70) 5th 6th Connection to Teachers 5th 6th Connection to School 5th 6th Connection to Peers 5th 6th Motivation 5th 6th Negative Expectations 5th 6th Connection to Parents M (SD) Regular M (SD) F (Time x SES) Free/Reduced SES M (SD) Girls M (SD) F (Time x Gender) Boys GENDER

PAGE 37

26 TABLE 5 Cluster Analysis Results 3.26 (.44) 2.52 (.60) 1.88 (.76) 2.78 (.75) 3.34 (.43) 2.51 (.60) 1.70 (.78) 2.79 (.80) 3.05 (.61) 2.50 (.75) 2.05 (.79) 2.70 (.78) 3.61 (.36) 3.22 (.53) 2.40 (.78) 3.29 (.64) .40 (.41) 1.09 (.66) 1.77 (.76) .86 (.75) 3.38 (.47) 3.00 (.57) 2.32 (.78) 3.09 (.66) Connection to Teachers Connection to School Connection to Peers Motivation Negative Expectations Connection to Parents M (SD) Low Risk 1 M (SD) Average Risk 2 M (SD) High Risk 3 RISK GROUP M (SD) TOTAL n = 2201 for the Low Risk group 1. n = 1867 for the Average Risk group 2. n = 627 for the High Risk group 3. Risk Group Chi Square Analyses Following the cluster analysis, chi square analyses were used to determine if placement in risk groups varied by gender (Table 6), ethnicity (Table 7), and SES (Table 8). Results indicated that boys, African American students, and students of low socioeconomic status were more likely to be placed in the high risk group relative to girls, Caucasian students, and students of higher socioeconomic status TABLE 6 Gender x Risk Chi-Square 1 375 1008 782 2165 286.4 866.1 1012.5 2165 214 773 1300 2287 302.6 914.9 1069.5 2287 589 1781 2082 4452 589 1781 2082 4452 Count Expected Count Count Expected Count Count Expected Count Boys Girls GENDER Total High Risk Average Risk Low Risk CLUSTER Total X(2) = 200.70, p < .001 1.

PAGE 38

27 TABLE 7 Ethnicity x Risk Chi-Square 1 456 1481 1806 3743 495.2 1497.4 1750.4 3743 133 300 276 709 93.8 283.6 331.6 709 589 1781 2082 4452 589 1781 2082 4452 Count Expected Count Count Expected Count Count Expected Count Caucasian African-American ETHNICITY Total High Risk Average Risk Low Risk CLUSTER Total X(2) = 31.68, p < .001 1. TABLE 8 SES x Risk Chi-Square 1 295 1105 1460 2860 378.4 1144.1 1337.5 2860 294 676 622 1592 210.6 636.9 744.5 1592 589 1781 2082 4452 589 1781 2082 4452 Count Expected Count Count Expected Count Count Expected Count Regular Free/Reduced SES Total High Risk Average Risk Low Risk CLUSTER Total X(2) = 82.50, p < .001 1. Problem Behavior MANOVAs A series of six MANOVAs were conducted to examine mean differences in students referrals and suspensions from fifth to eleventh grade using Risk Group, Ethnicity, SES, and Gender as independent variables. Results are divided into six sections. Main effects are presented first. Then two way interactions of main effects x time are presented. These are followed by two way interactions involving Ethnicity, SES, and Gender, including Ethnicity x SES, Ethnicity x Gender, and SES x Gender. These are followed by two way interactions involving Risk, including Risk x Ethnicity, Risk x SES, and Risk x Gender. These are followed by three way interactions involving Risk,

PAGE 39

28 including Ethnicity x SES x Risk, Ethni city x Gender x Risk, and SES x Gender x Risk. These are followed by four way interactions of Ethnicity x SES x Gender x Risk. Main Effects With few exceptions, main effects of Ethnicity, SES, Gender, and Risk were each significant across outcomes throu ghout the course of the study (see Tables 9 15). The only cases in which effects were not significant involved the effect of SES on Classroom Referrals in fifth grade, Ethnicity on Total Referrals in eleventh grade, Risk Group effects on Violence Referrals in fifth grade (post hoc not significant) and IS Suspensions in eleventh grade, and five of the eight main effects upon Violence Referrals in ninth and eleventh grades. With these exceptions, higher risk groups, students who received free and reduced lun ch, African American students, and boys received higher mean referrals and suspensions than lower risk groups, students who did not receive free or reduced lunch, Caucasian students, and girls respectively. The significant main effect of Time across the mi ddle school transition upon referrals and suspensions is presented in Table 16. These results indicate that total referrals, violence referrals, classroom referrals, in school suspensions, and out of school suspensions increase across the middle school tra nsition.

PAGE 40

29 TABLE 9 Problem Behavior Main Effects5th-Grade .11 (.61) 15.19*** .02 (.15) 4.13* a .05 (.37) 6.25** .01 (.12) 10.56*** .01 (.11) 4.47* a .26 (.98) .05 (.28) .10 (.51) .04 (.32) .04 (.26) .45 (1.33) .08 (.35) .15 (.60) .09 (.59) .06 (.36) .11 (.59) 38.22*** .02 (.17) 15.98*** .05 (.40) 3.03 .01 (.13) 14.28*** .01 (.11) 32.90*** .41 (1.25) .07 (.32) .13 (.57) .07 (.48) .06 (.34) .14 (.65) 45.50*** .03 (.19) 21.01*** .06 (.41) 11.07** .02 (.19) 18.45*** .01 (.14) 21.18*** .62 (1.62) .12 (.41) .18 (.69) .11 (.63) .11 (.45) .08 (.50) 52.93*** .01 (.10) 46.53*** .03 (.24) 19.69*** .01 (.10) 32.44*** .01 (.08) 33.06*** .36 (1.16) .07 (.32) .14 (.62) .06 (.43) .05 (.31) Low Risk Average Risk High Risk RISK GROUP Regular Free/Reduced SES Caucasian African-American ETHNICITY Girls Boys GENDER M (SD) F Total Referrals5th M (SD) F Violence Referrals5th M (SD) F Classroom Referrals5th M (SD) F IS Suspensions 5th M (SD) F OS Suspensions5th Average and High Risk means not significantly different a. TABLE 10 Problem Behavior Main Effects6th-Grade 1.06 (3.31) 27.05*** .10 (.46) 12.33*** .59 (2.28) 16.66*** .32 (1.22) 26.62*** .11 (.71) 14.06*** 2.17 (5.25) .23 (.73) 1.24 (3.55) .66 (1.78) .26 (1.05) 4.16 (7.50) .40 (.96) 2.36 (4.85) 1.44 (3.31) .49 (1.34) 1.06 (3.17) 87.48*** .10 (.43) 63.50*** .59 (2.21) 55.49*** .34 (1.24) 62.86*** .07 (.45) 87.14*** 3.45 (6.86) .35 (.93) 1.97 (4.54) 1.09 (2.61) .48 (1.45) 1.30 (3.60) 102.24*** .13 (.49) 46.01*** .72 (2.43) 81.43*** .41 (1.36) 80.62*** .10 (.57) 86.50*** 5.16 (8.57) .54 (1.17) 3.01 (5.75) 1.67 (3.34) .82 (1.92) 1.04 (3.46) 82.86*** .07 (.35) 87.80*** .56 (2.30) 70.97*** .35 (1.40) 56.99*** .11 (.69) 86.50*** 2.84 (6.02) .32 (.86) 1.63 (4.04) .88 (2.25) .33 (1.18) Low Risk Average Risk High Risk RISK GROUP Regular Free/Reduced SES Caucasian African-American ETHNICITY Girls Boys GENDER M (SD) F Total Referrals6th M (SD) F Violence Referrals6th M (SD) F Classroom Referrals6th M (SD) F IS Suspensions6th M (SD) F OS Suspensions6th

PAGE 41

30 TABLE 11 Problem Behavior Main Effects7th-Grade 1.62 (4.80) 12.12*** .10 (.48) 17.95*** .84 (3.21) 9.67*** .58 (1.81) 16.83*** .18 (.86) 9.54*** 2.70 (5.54) .19 (.57) 1.44 (3.64) .98 (2.31) .36 (1.18) 4.97 (8.04) .44 (.96) 2.90 (5.89) 1.87 (3.17) .73 (1.77) 1.50 (4.14) 58.01*** .10 (.43) 41.96*** .79 (2.87) 39.45*** .54 (1.67) 66.75*** .14 (.68) 89.91*** 4.29 (7.49) .32 (.82) 2.37 (5.10) 1.58 (2.96) .65 (1.66) 1.87 (4.73) 64.66*** .13 (.52) 23.26*** .97 (3.15) 52.09*** .70 (2.00) 35.98*** .20 (.81) 55.79*** 5.69 (8.58) .42 (.90) 3.27 (6.11) 1.98 (3.10) .96 (2.08) 1.45 (3.78) 72.09*** .08 (.37) 42.34*** .71 (2.43) 52.33*** .54 (1.59) 41.17*** .17 (.79) 45.21*** 3.59 (7.06) .29 (.77) 2.02 (4.90) 1.30 (2.76) .48 (1.43) Low Risk Average Risk High Risk RISK GROUP Regular Free/Reduced SES Caucasian African-American ETHNICITY Girls Boys GENDER M (SD) F Total Referrals7th M (SD) F Violence Referrals7th M (SD) F Classroom Referrals7th M (SD) F IS Suspensions7th M (SD) F OS Suspensions7th TABLE 12 Problem Behavior Main Effects8th-Grade 1.93 (4.90) 14.47*** .09 (.39) 6.65** 1.03 (3.30) 6.16** .71 (2.00) 11.31*** .23 (.96) 14.15*** 3.41 (6.49) .16 (.52) 1.81 (4.15) 1.26 (2.85) .44 (1.28) 5.07 (7.10) .27 (.67) 2.85 (4.78) 1.85 (2.92) .81 (1.80) 1.83 (4.34) 48.37*** .08 (.37) 14.05*** .97 (2.87) 36.01*** .69 (1.90) 46.71*** .17 (.82) 60.95*** 4.95 (7.78) .25 (.64) 2.68 (5.14) 1.78 (3.26) .78 (1.70) 2.27 (5.03) 50.74*** .10 (.40) 54.19*** 1.21 (3.35) 28.42*** .87 (2.27) 14.82*** .25 (.97) 54.24*** 6.24 (8.68) .37 (.76) 3.38 (5.66) 2.10 (3.35) 1.07 (2.02) 1.85 (4.33) 48.05*** .07 (.33) 39.76*** .85 (2.50) 50.05*** .69 (1.89) 25.61*** .23 (.85) 46.34*** 4.07 (7.16) .22 (.61) 2.34 (4.87) 1.48 (3.00) .56 (1.54) Low Risk Average Risk High Risk RISK GROUP Regular Free/Reduced SES Caucasian African-American ETHNICITY Girls Boys GENDER M (SD) F Total Referrals8th M (SD) F Violence Referrals8th M (SD) F Classroom Referrals8th M (SD) F IS Suspensions8th M (SD) F OS Suspensions8th

PAGE 42

31 TABLE 13 Problem Behavior Main Effects9th-Grade 1.99 (4.57) 14.18*** .04 (.21) 2.50 .75 (2.27) 11.84*** .52 (1.47) 19.43*** .25 (.88) 11.38*** 3.49 (6.38) .06 (.27) 1.42 (3.07) 1.08 (2.37) .42 (1.16) 5.04 (7.62) .09 (.37) 2.32 (4.16) 1.72 (3.08) .71 (1.58) 2.05 (4.63) 46.29*** .03 (.18) 23.11*** .78 (2.24) 43.68*** .58 (1.71) 35.27*** .22 (.78) 40.56*** 4.71 (7.39) .10 (.36) 2.03 (3.80) 1.48 (2.70) .67 (1.51) 2.58 (5.45) 8.07** .04 (.21) 24.87*** 1.02 (2.75) 9.65*** .74 (1.97) 14.80*** .30 (.95) 16.69*** 4.97 (7.37) .14 (.43) 2.22 (3.61) 1.66 (2.78) .78 (1.66) 2.37 (5.03) 7.11** .04 (.20) 3.07 .77 (2.19) 24.36*** .63 (1.62) 8.30** .29 (.92) 9.89** 3.62 (6.61) .07 (.31) 1.69 (3.51) 1.17 (2.57) .47 (1.28) Low Risk Average Risk High Risk RISK GROUP Regular Free/Reduced SES Caucasian African-American ETHNICITY Girls Boys GENDER M (SD) F Total Referrals9th M (SD) F Violence Referrals9th M (SD) F Classroom Referrals9th M (SD) F IS Suspensions9th M (SD) F OS Suspensions9th TABLE 14 Problem Behavior Main Effects10th-Grade 1.94 (4.35) 8.95*** .03 (.20) 8.10*** .64 (1.85) 6.96** .39 (1.20) 7.45** .21 (.90) 6.74** 3.27 (5.75) .05 (.25) 1.23 (2.94) .71 (1.57) .42 (1.20) 4.43 (7.08) .09 (.34) 1.87 (4.24) 1.06 (2.31) .64 (1.61) 2.11 (4.48) 28.21*** .02 (.17) 4.28* .67 (1.95) 40.51*** .42 (1.15) 23.87*** .22 (.82) 29.64*** 4.10 (6.68) .09 (.35) 1.74 (3.75) .98 (2.08) .61 (1.58) 2.36 (4.87) 12.95*** .03 (.19) 31.21*** .82 (2.41) 8.03** .47 (1.33) 22.74*** .26 (.93) 15.60*** 4.77 (7.11) .12 (.41) 2.02 (3.77) 1.23 (2.21) .78 (1.81) 2.07 (4.46) 15.21*** .03 (.20) 4.52* .66 (2.17) 13.92*** .42 (1.23) 9.37** .22 (.83) 19.25*** 3.49 (6.13) .06 (.28) 1.41 (3.15) .79 (1.79) .48 (1.38) Low Risk Average Risk High Risk RISK GROUP Regular Free/Reduced SES Caucasian African-American ETHNICITY Girls Boys GENDER M (SD) F Total Referrals10th M (SD) F Violence Referrals10th M (SD) F Classroom Referrals10th M (SD) F IS Suspensions10th M (SD) F OS Suspensions10th

PAGE 43

32 TABLE 15 Problem Behavior Main Effects11th-Grade 1.41 (2.97) 3.60* .02 (.16) 1.43 .45 (1.39) 3.34* .28 (.89) 2.50 .13 (.54) 4.45* 2.16 (4.11) .03 (.17) .80 (1.95) .51 (1.31) .25 (.77) 3.09 (5.76) .04 (.20) 1.21 (2.35) .66 (1.57) .44 (1.18) 1.58 (3.51) 18.12*** .01 (.12) 6.80** .51 (1.50) 19.87*** .31 (.92) 17.69*** .15 (.59) 26.10*** 2.56 (4.49) .05 (.24) 1.02 (2.20) .64 (1.55) .35 (.97) 1.69 (3.58) 3.53 .02 (.14) 2.56 .56 (1.60) 5.85* .34 (1.02) 8.83** .17 (.65) 3.99* 2.86 (4.96) .06 (.27) 1.17 (2.38) .77 (1.66) .40 (1.05) 1.15 (2.52) 42.26*** .02 (.15) < 1 .35 (1.17) 40.26*** .26 (.85) 19.20*** .11 (.48) 18.43*** 2.69 (4.81) .03 (.19) 1.02 (2.19) .58 (1.41) .32 (.92) Low Risk Average Risk High Risk RISK GROUP Regular Free/Reduced SES Caucasian African-American ETHNICITY Girls Boys GENDER M (SD) F Total Referrals11th M (SD) F Violence Referrals11th M (SD) F Classroom Referrals11th M (SD) F IS Suspensions11th M (SD) F OS Suspensions11th TABLE 16 Mean Changes in Problem Behavior across the Middle School Transition .22 1.91 .04 .19 .08 1.08 .03 .61 .03 .22 (.90) (4.96) (.24) (.66) (.47) (3.31) (.31) (1.88) (.22) (.96) 514.05*** 180.35*** 400.95*** 389.09*** 214.45*** M (SD) F 5th-Grade 6th-Grade Total Referrals 5th-Grade 6th-Grade Violence Referrals 5th-Grade 6th-Grade Classroom Referrals 5th-Grade 6th-Grade IS Suspensions 5th-Grade 6th-Grade OS Suspensions

PAGE 44

33 Main Effects x Time Results presented in Tables 17 and 18 indicate that mean differences between levels for each independent variable increase significan tly across the middle school transition. This effect is significant across each dependent variable. For example, as shown in Figure 1, the difference in Total Referrals between Caucasian and African American students increases from .48 (.62 .14) in 5 th g rade to 3.86 (5.16 1.30) in 6 th grade.

PAGE 45

34 TABLE 17 Main Effects x Time 2-Way Interactions for Referrals .11 (.61) 1.06 (3.31) 22.01*** .02 (.15) .10 (.46) 7.84*** .05 (.37) .59 (2.28) 14.94*** .26 (.98) 2.17 (5.25) .05 (.28) .23 (.73) .10 (.51) 1.24 (3.55) .45 (1.33) 4.16 (7.50) .08 a (.35) .40 (.96) .15 (.60) 2.36 (4.85) .11 (.59) 1.06 (3.17) 72.74*** .02 (.17) .10 (.43) 42.40*** .05 (.40) .59 (2.21) 54.71*** .41 (1.25) 3.45 (6.86) .07 (.32) .35 (.93) .13 (.57) 1.97 (4.54) .14 (.65) 1.30 (3.60) 84.79*** .03 (.19) .13 (.49) 26.98*** .06 (.41) .72 (2.43) 76.96*** .62 (1.62) 5.16 (8.57) .12 (.41) .54 (1.17) .18 (.69) 3.01 (5.75) .08 (.50) 1.04 (3.46) 64.86*** .01 (.10) .07 (.35) 47.00*** .03 (.24) .56 (2.30) 63.76*** .36 (1.16) 2.84 (6.02) .07 (.32) .32 (.86) .14 (.62) 1.63 (4.04) Low Average High RISK GROUP Regular Free/Reduced SES Caucasian African-American ETHNICITY Girls Boys GENDER M (SD) 5th M (SD) F 6th Total Referrals M (SD) 5th M (SD) F 6th Violence Referrals M (SD) 5th M (SD) F 6th Classroom Referrals Average and High Risk means not significantly different for 5th-grade Violence referrals a.

PAGE 46

35 FIGURE 1 Ethnicity x Time (Total Referrals) Grade 6th-Grade 5th-Grade Mean Total Referrals 6 5 4 3 2 1 0 ETHNICITY Caucasian African-American TABLE 18 Main Effects x Time 2-Way Interactions for Suspensions .01 (.12) .32 (1.22) 22.05*** .01 (.11) .11 (.71) 11.54*** .04 (.32) .66 (1.78) .04 (.26) .26 (1.05) .09 (.59) 1.44 (3.31) .06 a (.36) .49 (1.34) .01 (.13) .34 (1.24) 54.65*** .01 (.11) .07 (.45) 69.80*** .07 (.48) 1.09 (2.61) .06 (.34) .48 (1.45) .02 (.19) .41 (1.36) 70.05*** .01 (.14) .10 (.57) 73.84*** .11 (.63) 1.67 (3.34) .11 (.45) .82 (1.92) .01 (.10) .35 (1.40) 44.57*** .01 (.08) .11 (.69) 26.21*** .06 (.43) .88 (2.25) .05 (.31) .33 (1.18) Low Average High RISK GROUP Regular Free/Reduced SES Caucasian African-American ETHNICITY Girls Boys GENDER M (SD) 5th M (SD) F 6th IS Suspensions M (SD) 5th M (SD) F 6th OS Suspensions Average and High Risk means not significantly different for 5th-grade OS Suspensions a.

PAGE 47

36 Two Way Interactions involving Ethnicity, SES, and Gender Ethnicity x SES (x Time) Results presented in Table 19 indicate that mean differences in referral s and suspensions between Regular and Free/Reduced Lunch students differ across levels of Ethnicity. For each of the 17 significant Ethnicity x SES interactions, the mean difference between Regular and Free Reduced Lunch groups is larger for African Americ an students than for Caucasian students. The five significant Ethnicity x SES x Time interactions indicate that from 5th to 6th grade mean differences between Regular and Free/Reduced Lunch groups become larger for African American students relative to Ca ucasian students for each of the five outcomes. For example, as shown in Figures 2a and 2b, the difference in Total Referrals for those receiving Regular Lunch and those receiving Free/Reduced Lunch changes from .14 (.24 .10) in 5 th grade to 1.15 (2.13 0.98) for Caucasian students, and from .49 (.73 .24) to 3.41 (5.88 2.47) for African American students.

PAGE 48

37 TABLE 19 Ethnicity x SES (x Time) Interactions .10 (.56) .24 (.84) .24 (.95) .73 (1.74) 10.84** 22.89*** .98 (2.96) 2.13 (4.80) 2.47 (5.54) 5.88 (9.08) 27.18*** 1.40 (3.77) 3.13 (6.51) 3.34 (8.17) 6.31 (8.59) 4.84* 1.58 (3.54) 2.08 (3.67) 1.60 (2.80) 3.27 (5.42) 9.71** .02 (.16) .04 (.23) a .06 (.29) .13 (.44) 4.37* 20.06*** .10 (.41) .20 (.66) .19 (.71) .63 (1.24) 27.39*** .10 (.42) .23 (.72) .20 (.57) .48 (.96) 5.62* .05 (.38) .09 (.48) .12 (.65) .19 (.70) < 1 16.56*** .53 (2.07) 1.19 (3.15) 1.59 (3.89) 3.39 (6.10) 15.96*** .50 (1.48) .81 (1.95) .68 (1.88) 1.33 (2.50) 4.95* .01 (.12) .04 (.30) .05 (.29) .13 (.70) 3.31 18.96*** .31 (1.15) .66 (1.78) .81 (2.28) 1.89 (3.53) 21.16*** .52 (1.65) 1.19 (2.67) .93 (1.88) 2.26 (3.30) 7.40** .31 (.92) .48 (1.32) .41 (.93) .88 (1.82) 6.86** .01 (.11) .03 (.20) a .02 (.14) .13 (.50) 15.74*** 29.50*** .06 (.41) .22 (.85) .26 (.92) .96 (2.08) 37.62*** .13 (.62) .38 (1.17) .35 (1.39) 1.13 (2.20) 25.51*** .15 (.75) .52 (1.38) .47 (1.69) 1.22 (2.06) 5.77* .15 (.60) .25 (.78) .12 (.35) .49 (1.18) 12.06** 5th 6th 7th 11th Total 5th 6th 7th Violence 5th 6th 11th Classroom 5th 6th 7th 11th ISS 5th 6th 7th 8th 11th OSS M (SD) Regular M (SD) Free/Reduced Caucasian M (SD) Regular M (SD) F (Ethnicity x SES) F (Ethnicity x SES x Time) Free/Reduced African-American post hoc test of mean difference between Regular and Free/Reduced Lunch Caucasian students not significant a.

PAGE 49

38 Figure 2a Ethnicity x SES (5th-Grade) ETHNICITY African-American Caucasian Mean Total Referrals 6 5 4 3 2 1 0 SES Regular Free/Reduced Figure 2b Ethnicity x SES (6th-Grade) ETHNICITY African-American Caucasian Mean Total Referrals 6 5 4 3 2 1 0 SES Regular Free/Reduced

PAGE 50

39 Ethnicity x Gender (x Time) Results presented in Table 20 indicate that mean diffe rences in referrals and suspensions between Boys and Girls differ across levels of Ethnicity. For each of the nine significant Ethnicity x Gender interactions, Gender differences in means were larger for African American students than for Caucasian student s. The five significant Ethnicity x Gender x Time interactions indicate that from 5 th to 6 th grade mean differences between Boys and Girls become larger for African American students relative to Caucasian students.

PAGE 51

40 TABLE 20 Ethnicity x Gender (x Time) Interactions .04 (.28) .25 (.88) .33 (1.06) .91 (1.98) 5.72* 4.45* .58 (2.03) 2.07 (4.62) 3.61 (6.90) 6.67 (9.71) 6.13* .00 (.06) .05 (.26) .04 (.22) .19 (.53) 8.27** 5.20* .03 (.23) .22 (.65) .27 (.69) .80 (1.45) 11.19** .02 (.17) .11 (.56) .09 (.48) .26 (.84) < 1 6.67* .29 (1.36) 1.17 (3.15) 2.05 (4.68) 3.95 (6.50) 6.55* .00 (.08) .03 (.26) .03 (.20) .19 (.86) 9.39** 3.87* .19 (.84) .64 (1.72) 1.24 (2.81) 2.08 (3.74) 6.06* .00 (.02) .03 (.20) .03 (.20) .18 (.60) 7.93** 4.44* .03 (.25) .18 (.77) .55 (1.59) 1.08 (2.16) 7.22** 5th 6th Total 5th 6th Violence 5th 6th Classroom 5th 6th ISS 5th 6th OSS M (SD) Girls M (SD) Boys Caucasian M (SD) Girls M (SD) F (Ethnicity x Gender) F (Ethnicity x Gender x Time) Boys African-American

PAGE 52

41 S ES x Gender (x Time) Results presented in Table 21 below indicate that mean differences in referrals and suspensions between Boys and Girls differ across levels of SES. For Violence Referrals in 6th and 7th grade, Out of School Suspensions in 5th grade, a nd Classroom Referrals in 8th grade, Gender differences are larger among students who receive Free/Reduced Lunch. The significant SES x Gender x Time interaction for Violence Referrals indicates that from 5 th to 6 th grade the mean difference between Boys and Girls becomes larger among students receiving Free/Reduced Lunch than among students receiving Regular Lunch. TABLE 21 SES x Gender (x Time) Interactions .00 (.05) .04 (.24) .02 (.16) .13 (.43) 2.31 5.58* .03 (.22) .18 (.56) .15 (.50) .56 (1.19) 8.51** .03 (.22) .18 (.56) .17 (.53) .48 (1.02) 8.99** .00 (.03) .02 (.16) a .01 (.14) .11 (.46) 15.60*** na .49 (1.87) 1.49 (3.59) 1.54 (3.29) 3.82 (6.28) 7.55** na 5th 6th 7th Violence 5th OSS 8th Classroom M (SD) Girls M (SD) Boys Regular M (SD) Girls M (SD) F (SES x Gender) F (SES x Gender x Time) Boys Free/Reduced Means of Girls and Boys receiving Regular Lunch are not significantly different for OSS in 5th-grade a. Two Way Interactions involving Risk Risk x SES (x Time) Results presented in Table 22 indicate that mean differences in refer rals and suspensions between Regular and Free/Reduced Lunch students differ across levels of Risk. For each of the eight significant Risk x SES interactions, the mean difference

PAGE 53

42 between Regular and Free/Reduced lunch students increases significantly as Ris k increases. Significant Risk x SES x Time interactions indicate that from 5th to 6th grade the mean difference between Regular and Free/Reduced lunch students becomes larger across levels of Risk. For Total Referrals, Classroom Referrals, and In School Suspensions, the Risk x SES interaction was not significant in 5th Grade. However, by 6th grade the interaction was significant.

PAGE 54

43 TABLE 22 Risk Group x SES (x Time) Interactions .07 (.51) .21 (.79) .13 (.60) .49 (1.36) .24 (.82) .66 (1.66) 1.29 3.47* .56 (1.93) 2.23 (5.09) 1.35 (3.72) 3.52 (6.87) 2.43 (4.87) 5.89 (9.11) 3.75* 1.03 (3.95) 3.03 (6.16) 1.69 (3.71) 4.38 (7.38) 3.15 (5.80) 6.79 (9.46) 4.70** .04 (.37) .07 (.37) .06 (.42) .15 (.64) .12 (.44) .19 (.72) < 1 3.24* .31 (1.41) 1.25 (3.49) .75 (2.57) 2.03 (4.63) 1.39 (3.41) 3.34 (5.81) 3.08* .54 (2.84) 1.57 (3.87) .86 (2.36) 2.39 (4.96) 1.75 (4.27) 4.07 (6.97) 6.61** .00 (.09) .02 (.17) .01 (.14) .08 (.48) .02 (.24) .16 (.80) 2.49 5.89** .17 (.75) .69 (1.87) .44 (1.46) 1.02 (2.16) .78 (1.92) 2.10 (4.16) 6.86** .00 (.05) .02 (.18) a .02 (.16) .07 (.37) .00 (.06) .12 (.50) 4.50* 2.88 .03 (.33) .28 (1.17) .10 (.53) .51 (1.53) .15 (.63) .83 (1.72) 4.18* .08 (.55) .41 (1.30) .20 (.83) .63 (1.56) .24 (.67) 1.22 (2.31) 11.41*** 5th 6th 7th Total 5th 6th 7th Classroom 5th 6th ISS 5th 6th 7th OSS M (SD) Regular M (SD) Free/Reduced Low Risk M (SD) Regular M (SD) Free/Reduced Average Risk M (SD) Regular M (SD) F (Risk x SES) F (Risk x SES x Time) Free/Reduced High Risk post hoc test of mean difference between Regular and Free/Reduced Lunch students not significant a.

PAGE 55

44 Risk x Ethnicity (x Time) Results presented in Table 23 indicate that mean differences in referrals and suspensions between Caucasian and African American students differ across levels of Risk. For each of the four significant Risk x Ethnicity interactions, the mean difference between Caucasian and African American students increases significan tly as Risk increases. The significant Risk x Ethnicity x Time interaction for Out of School suspensions indicates that from 5th to 6th grade the mean difference between Caucasian and African American students becomes larger across levels of Risk.

PAGE 56

45 TABLE 23 Risk x Ethnicity (x Time) Interactions .08 (.53) .32 (.96) .16 (.70) .75 (1.74) .29 (.87) .98 (2.21) 4.77** 1.61 .70 (2.37) 3.43 (6.28) 1.52 (3.95) 5.40 (8.62) 2.97 (5.41) 8.23 (11.30) 2,62 .01 (.11) .02 (.20) a .02 (.21) .13 (.61) .04 (.31) .27 (1.09) 5.80** 2.46 .21 (.84) 1.05 (2.46) .47 (1.41) 1.62 (2.82) .97 (2.34) 3.03 (5.14) 3.82* .00 (.05) .05 (.26) .02 (.18) .13 (.49) .03 (.22) .17 (.63) < 1 3.12* .05 (.37) .51 (1.64) .13 (.66) .89 (1.97) .26 (.84) 1.28 (2.19) 3.46* 5th 6th Total 5th 6th ISS 5th 6th OSS M (SD) Caucasian M (SD) African-American Low Risk M (SD) Caucasian M (SD) African-American Average Risk M (SD) Caucasian M (SD) F (Risk x Ethnicity F (Risk x Ethnicity x Time) African-American High Risk post hoc test of mean difference between Caucasian and African-American students not significant a.

PAGE 57

46 Risk x Gender Results presented in Table 24 indicate that mean differences in referrals and suspensions between Boys and Girls differ across levels of Risk. For each of the four significant Risk x Gender interactions, the me an difference between Caucasian and African American students increased significantly as Risk increased.

PAGE 58

47 TABLE 24 Risk x Gender Interactions .04 (.31) .22 (.90) .12 (.68) .37 (1.14) .18 (.63) .60 (1.58) 4.29* .01 (.07) .04 (.22) .01 (.09) .09 (.36) .03 (.22) .10 (.40) 3.15* .00 1 (.05) .02 1 (.19) .01 (.16) .06 (.40) .01 (.12) .14 (.74) 7.40* .00 1 (.07) .02 1 (.15) .01 (.08) .07 (.34) .01 (.15) .09 (.44) 3.46* Total5th Violence5th ISS5th OSS5th M (SD) Girls M (SD) Boys Low Risk M (SD) Girls M (SD) Boys Average Risk M (SD) Girls M (SD) F Boys High Risk Means not significantly different 1.

PAGE 59

48 Three way Interactions involving Risk Ethnicity x SES x Risk (x Time) Significant Ethnicity x SES x Ris k interactions were found for In School Suspensions in 6 th grade, Classroom referrals in 7 th grade, and Out of School Suspensions in 7 th grade. Results presented in Table 25 indicate that in three cases, the Ethnicity x SES interaction was only significant at one or two, but not all three levels of Risk. For Classroom Referrals in 7th Grade, mean differences between Regular and Free/Reduced Lunch groups were only larger for African American students relative to Caucasian students at Average and High levels of Risk. For Out of School Suspensions in 7th Grade, mean differences between Regular and Free/Reduced Lunch groups were only larger for African American students relative to Caucasian students at the High Risk level. For In School Suspensions in 6th Grade mean differences between Regular and Free/Reduced Lunch groups were only larger for African American students relative to Caucasian students at Low and High levels of Risk. The Ethnicity x SES x Risk x Time interaction for In School Suspensions indicates that the Ethnicity x SES x Risk interaction for In School Suspensions became significant in 6 th grade, whereas it had not been significant in 5 th grade.

PAGE 60

49 TABLE 25 Ethnicity x SES x Risk (x Time) Interactions .00 (.09) .01 (.14) .00 (.00) .02 (.23) 1.23 3.16* .01 (.11) .05 (.35) .08 (.38) .14 (.65) .02 (.22) .08 (.41) .09 (.43) .31 (1.18) .16 (.69) .41 (1.19) .40 (1.49) 1.26 (2.67) a 3.75* .40 (1.35) .63 (1.54) 1.17 (2.69) 1.74 (2.84) .75 (1.83) 1.31 (2.92) 1.09 (2.88) 3.41 (5.41) a .46 (2.13) 1.02 (3.00) 2.05 (8.98) 2.60 (4.97) 3.26* na .81 (2.32) 1.63 (3.99) a 1.68 (2.80) 3.69 (6.08) a 1.72 (4.29) 3.18 (6.80) a 2.10 (4.15) 5.43 (7.04) a .07 (.47) .23 (.96) .25 (1.41) .75 (1.73) 4.62* na .18 (.76) .36 (1.05) .53 (1.58) 1.09 (2.10) .25 (.69) .76 (1.69) a .14 (.36) 1.92 (2.90) a Low Risk Average Risk High Risk 5th Low Risk Average Risk High Risk 6th ISS Low Risk Average Risk High Risk 7th Classroom Low Risk Average Risk High Risk 7th OSS M (SD) Regular M (SD) Free/Reduced Caucasian M (SD) Regular M (SD) F (Ethnicity x SES x Risk) F )Ethnicity x SES x Risk x Time) Free/Reduced African-American post hoc test of mean difference between Regular and Free/Reduced Lunch students IS significant (All others are NOT significant) a.

PAGE 61

50 Ethnicity x Gender x Risk Significant Ethnicity x Gender x Risk interactions were found for Violence Referrals in 7th grade and for Out of School Suspensions in 8th grade. Results presented in Table 26 indicate that the Ethnicity x Gender interaction differed across levels of Risk. For Violence Referrals in 7 th grade, gender diffe rences existed across all three levels of Risk for Caucasian students, while gender differences existed only at Average and High Risk levels for African American students. For Out of School Suspensions in 8 th grade, gender differences existed only at the L ow Risk level for Caucasian students, while gender differences existed only at the Average and High Risk levels for African American students.

PAGE 62

51 TABLE 26 Ethnicity x Gender x Risk Interactions 1 .02a (.16) .17b (.61) .21a (.69) .30a (.95) 3.53* .06a (.33) .19b (.53) .22a (.50) .64b (1.06) .16a (.53) .46b (1.05) .46a (.76) .89b (1.17) .06a (.35) .28b (1.13) .66a (1.63) .88a (1.79) 3.56* .21a (.86) .34a (1.05) .56a (1.21) 1.57b (2.46) .34a (.97) .69a (1.71) 1.14a (1.49) 1.96b (2.85) Low Risk Average Risk High Risk Violence7th Low Risk Average Risk High Risk OSS8th M (SD) Girls M (SD) Boys Caucasian M (SD) Girls M (SD) F (Ethnicity x Gender x Risk) Boys African-American Pairwise comparisons between Boys and Girls within Ethnicity across levels of Risk with different letters are significantly different 1.

PAGE 63

52 SES x Gender x Risk Significant SES x Gender x Risk interactions were found for Total R eferrals, Violence Referrals, Classroom Referrals, In School Suspensions, and Out of School Suspensions in 7th grade, and for Total Referrals and Classroom Referrals in 8 th grade. Results presented in Table 27 indicate that the SES x Gender interaction dif fered across levels of Risk. For Total Referrals in 7 th grade, gender differences for students receiving Regular lunch existed only at the Low and Average Risk levels, while for students receiving Free/Reduced lunch gender differences existed across all th ree levels of Risk. For Violence Referrals and for Classroom Referrals in 7 th grade, gender differences for students receiving Regular lunch existed only at the Low Risk level, while for students receiving Free/Reduced lunch, gender differences existed onl y at the Average and High Risk levels. For In School Suspensions in 7 th grade, gender differences existed across all three levels of Risk for both Regular and Free/Reduced lunch students. However, gender differences were larger among students receiving Fre e/Reduced lunch. For Out of School Suspensions in 7 th grade, gender differences did not exist at any level of Risk for Regular lunch students. However, gender differences at Average and High Risk levels for students receiving Free/Reduced Lunch. For Tota l Referrals and for Classroom Referrals in 8 th grade, gender differences for students receiving Regular lunch existed only at the Low Risk level, while for students receiving Free/Reduced lunch gender differences existed across all three levels of Risk.

PAGE 64

53 TABLE 27 SES x Gender x Risk Interactions .50 (1.73) 1.93 (5.96) 2.36 (4.96) 4.11 (7.60) 7.59** .95 (2.30) 2.23 (4.41) 2.64 (5.19) 5.80 (8.52) 2.05 (4.66) 3.70 (6.23) 1 4.54 (6.80) 8.20 (10.58) .01 (.12) .14 (.51) .13 (.51) .29 (.93) 1 3.16* .05 (.27) .16 (.48) 1 .15 (.48) .46 (.90) .13 (.50) .34 (.84) 1 .34 (.69) .78 (1.27) .18 (.81) 1.13 (4.47) 1.13 (2.98) 2.26 (4.92) 10.05*** .49 (1.56) 1.14 (2.78) 1 1.31 (3.43) 3.28 (5.78) 1.20 (3.65) 2.02 (4.54) 1 2.63 (4.94) 4.96 (7.87) .18 (.85) .67 (2.00) .85 (1.95) 1.52 (3.02) 3.16* .35 (1.12) .83 (2.09) .99 (2.20) 2.04 (3.31) .75 (1.61) 1.33 (2.51) 1.93 (2.86) 3.04 (4.13) .03 (.22) .17 (.84) 1 .33 (1.15) .56 (1.51) 1 3.73* .08 (.43) .28 (1.02) 1 .37 (1.22) .84 (1.77) .09 (.41) .32 (.76) 1 .75 (1.50) 1.51 (2.66) .68 (2.17) 2.03 (5.02) 2.88 (5.78) 5.05 (7.98) 4.11* 1.64 (4.13) 2.67 (5.02) 1 3.19 (5.11) 7.18 (9.92) 2.34 (4.33) 4.25 (6.16) 1 5.05 (6.58) 7.48 (8.70) .25 (1.02) 1.20 (3.73) 1.45 (3.56) 2.97 (5.71) 5.62** .80 (2.59) 1.48 (3.23) 1 1.39 (2.87) 4.02 (6.59) 1.34 (3.10) 2.42 (4.13) 1 2.24 (3.36) 4.54 (6.27) Low Risk Average Risk High Risk Total7th Low Risk Average Risk High Risk Violence7th Low Risk Average Risk High Risk Classroom7th Low Risk Average Risk High Risk ISS7th Low Risk Average Risk High Risk OSS7th Low Risk Average Risk High Risk Total8th Low Risk Average Risk High Risk Classroom8th M (SD) Girls M (SD) Boys Regular M (SD) Girls M (SD) F Boys Free/Reduced Pairwise comparisons between Boys and Girls within SES across levels of Risk not significantly different 1. Ethnicity x SES x Gender x Risk Four way Ethnicity x SES x Gender x Risk interactions (see Figures 3 6) were significant for Total Referrals ( F(2,4141) = 6.37, p < .01; F(2,3979) = 4.51, p < .05) and Classroom Referrals ( F(2,4141) = 8.19, p < .001; F(2,3979) = 4.96, p < .01) in 7th and 8th grades respectively. Post hoc pairwise comparisons were performed for Regular Free/Reduced Lunch pairs across Risk Levels within each of the four Ethnicity/Gender groups for each of the four significant dep endent variables.

PAGE 65

54 For both Total Referrals (Figure 3) and Classroom Referrals (Figure 4) in 7th grade, SES differences were significant for High Risk Caucasian Boys, Average Risk Caucasian Boys, and High Risk African American Boys. The SES discrepancy in mean Total Referrals was larger for African American than for Caucasian Boys. Highlighted is the highest bar representing High Risk, Low SES, African American Boys' mean of 11.25 Total Referrals received in the 7th grade. For both Total Referrals (Figure 5) and Classroom Referrals (Figure 6) in 8th grade, SES differences were significant for Average and Low Risk Caucasian Boys and for Average Risk African American Boys. Additionally, for Total Referrals in 8th grade only, SES differences were significant for Low Risk African American Girls.

PAGE 66

55 Figure 3 Ethnicity x SES x Gender x Risk Total Referrals in 7th-Grade Ethnicity/Gender Group AA Girls AA Boys C Girls C Boys Mean Total Referrals7th-Grade 12 10 8 6 4 2 0 Risk/SES Group High/Regular High/Free Average/Regular Average/Free Low/Regular Low/Free

PAGE 67

56 Figure 4 Ethnicity x SES x Gender x Risk Classroom Referrals in 7th-Grade Ethnicity/Gender Group AA Girls AA Boys C Girls C Boys Mean Class Referrals7th-Grade 8 6 4 2 0 Risk/SES Group High/Regular High/Free-Reduced Average/Regular Average/Free-Reduced Low/Regular Low/Free-Reduced

PAGE 68

57 Figure 5 Ethnicity x SES x Gender x Risk Total Referrals in 8th-Grade Ethnicity/Gender Group AA Girls AA Boys C Girls C Boys Mean Total Referrals8th-Grade 12 10 8 6 4 2 0 Risk/SES Group High/Regular High/Free-Reduced Average/Regular Average/Free-Reduced Low/Regular Low/Free-Reduced

PAGE 69

58 Figure 6 Ethnicity x SES x Gender x Risk Classroom Referrals in 8th-Grade Ethnicity/Gender Group AA Girls AA Boys C Girls C Boys Mean Class Referrals8th-Grade 7 6 5 4 3 2 1 0 Risk/SES Group High/Regular High/Free-Reduced Average/Regular Average/Free-Reduced Low/Regular Low/Free-Reduced Dropout Chi Square Analyses Results of chi square analyses presented in tables 2 8 31 indicate that being male, African American, low SES, and High Risk were each associated with significantly elevated levels of dropout. SES bore the strongest association with dropout (? = 185.65), while gender bore the weakest association with dropout ( ? = 14.87).

PAGE 70

59 TABLE 28 GENDER x STATUS Chi-Square 1 323 328 1381 133 2165 285.5 328.7 1429.2 121.6 2165.0 264 348 1558 117 2287 301.5 347.3 1509.8 128.4 2287.0 587 676 2939 250 4452 587.0 676.0 2939.0 250.0 4452.0 Count Expected Count Count Expected Count Count Expected Count Boys Girls GENDER Total Dropout Moved Present Adult Education STATUS Total X(3) = 14.87, p < .01 1. TABLE 29 ETHNICITY x STATUS Chi-Square 1 470 621 2471 181 3743 493.5 568.3 2471.0 210.2 3743.0 117 55 468 69 709 93.5 107.7 468.0 39.8 709.0 587 676 2939 250 4452 587.0 676.0 2939.0 250.0 4452.0 Count Expected Count Count Expected Count Count Expected Count Caucasian African-American ETHNICITY Total Dropout Moved Present Adult Education STATUS Total X(3) = 63.12, p < .001 1. TABLE 30 SES x STATUS Chi-Square 1 259 416 2068 117 2860 377.1 434.3 1888.0 160.6 2860.0 328 260 871 133 1592 209.9 241.7 1051.0 89.4 1592.0 587 676 2939 250 4452 587.0 676.0 2939.0 250.0 4452.0 Count Expected Count Count Expected Count Count Expected Count Regular Free/Reduced SES Total Dropout Moved Present Adult Education STATUS Total X(3) = 186.65, p < .001 1.

PAGE 71

60 TABLE 31 RISK x STATUS Chi-Square 1 135 97 311 46 589 77.7 89.4 388.8 33.1 589.0 250 277 1125 129 1781 234.8 270.4 1175.7 100.0 1781.0 202 302 1503 75 2082 274.5 316.1 1374.4 116.9 2082.0 587 676 2939 250 4452 587.0 676.0 2939.0 250.0 4452.0 Count Expected Count Count Expected Count Count Expected Count Count Expected Count High Risk Average Risk Low Risk CLUSTER Total Dropout Moved Present Adult Education STATUS Total X(3) = 122.18, p < .001 1. Discussion Factor Analysis of the Student Adjustment Survey Factor an alysis of the Student Adjustment Survey supported the proposed five factor solution. Factors included connection to teachers, connection to school, connection to peers, motivation, and negative expectancies. When combined with the connection to parents sca le utilized in the present study, incorporation of constructs assessing students connection to teachers, school, and peers, as well as motivation and negative expectations permitted examination of students self reported functioning across domains conside red central to the development of problem behavior (Dryfoos, 1990; Hirschi, 1969; Marcus & Sanders Reio, 2001; Najaka et al., 2001; Pianta, 1999; Rothbaum & Weisz, 1994; Vitaro et al., 2001) Internal reliability estimates for the Student Adjustment Survey factors and the Connection to Parents scale were generally acceptable. The presence of a wealth of validity data based upon these constructs indicates that a sufficient level of reliability was present. However, further refinement of the Student Adjustment Survey and the

PAGE 72

61 Connection to Parents scale is warranted given relatively low internal reliability estimates for connection to parents, connection to peers, motivation, and negative expectations, and the absence of test retest data at t his time. In particular, the connection to peers factor proved unreliable among sixth graders. Items with a positive tone (e.g. A student can be himself/herself and still be accepted by other students in this school) and negative tone (e.g. Making frien ds is very difficult in this school) did not show a strong negative correlation. This finding is consistent with research indicating that students in middle school are more likely to associate with subgroups of peers with whom they can be themselves, wh ile experiencing or perceiving rejection from a majority of peers (Dishion et al., 1991; Patterson, Forgatch, Yoerger, & Stoolmiller, 1998) Student Adjustment Survey and Connection to Parents Scale: Main Effects Findings derived from ex amination of mean differences across gender, ethnicity, socioeconomic status, and time in students connections to parents, peers, teachers, and school, as well as their levels of motivation and negative expectations were generally consistent with those fo und in prior work and the hypotheses of the present study. Gender Findings indicated that boys self reported connections with peers, school, and teachers in 5 th and 6 th grade, were generally lower than were girls reports. These results are consistent with prior studies indicating that girls report having more supportive relationships with their peers (Dishion, 1990; Dishion et al., 1991) and teachers (Davis, 2001; Hamre & Pianta, 2001; Wentzel, 2002) and report likin g school to a greater degree (Murray & Greenberg, 2000) than do boys. Heightened levels of motivation reported by girls in the present study are consistent with the work of Goodenow (1993), while

PAGE 73

62 heightened levels of negative expectations reported by boys are consistent with the work of Wentzel (1997) indicating that girls report heightened levels of educational aspirations than do boys. In contrast to these findings, the hypothesis that girls would report stronger connection to parents w as not supported by the findings of the present study. Prior research suggests that parental relations with boys are characterized by higher levels of conflict than are parental relations with girls (Rothbaum & Weisz, 1994) Based upon the se data, girls were expected to report higher levels of connection to parents in the present study. Differences were not found between boys and girls in their reports of connection to parents. Failure to find a significant gender difference is likely attri butable to issues associated with scale construction including item content and reliability. While the weight of data support the validity of the measure used in the present study, failure to obtain expected differences highlight the need for further scale refinement. Ethnicity and SES Findings indicated that African American students and those from low socioeconomic status backgrounds generally report higher levels of risk across domains relative to their Caucasian and higher socioeconomic status count erparts. These results are consistent with prior research suggesting that African American and low SES students are more likely to experience punitive interactions with parents and are less likely to experience positive interactions (Borkow ski et al., 2002; McLoyd, 1998; Yung & Hammond, 1997) Results are also consistent with work suggesting that African American students and those from low socioeconomic status backgrounds are more likely to perceive interactions with teachers as punitive a nd characterized by less reinforcement

PAGE 74

63 for successful performance (McLoyd, 1998; Polite, 1994) are more likely to feel detached from school (Hirschi, 1969; McLoyd, 1998; Steele, 1997) and are likely to experience increas ed levels of negative expectations for school success (Tucker & Herman, 2002) Although no a priori predictions were made based upon insufficient prior research, results of the present study indicated that African American students and st udents from lower socioeconomic status backgrounds reported lower levels of motivation than did their Caucasian and higher socioeconomic status counterparts. Given the lack of prior research in this area, these findings must be considered preliminary. Thes e findings are consistent with prior research suggesting that African American students and students from lower socioeconomic status backgrounds are more likely to feel detached from school and to experience increased levels of negative expectations for sc hool success (Hirschi, 1969; McLoyd, 1998; Steele, 1997; Tucker & Herman, 2002) However, relationships among these variables and student motivation are presently unclear. As such, further research in this area is warranted. Contrary to prediction, African American students reported higher levels of connections to peers relative to Caucasian students. This finding may reflect a somewhat stronger orientation toward peer involvement among African American students relative to Caucasian stud ents. However, lacking systematic prior research examining ethnic differences in connections to peers, this finding is considered preliminary. In contrast, consistent with prediction, students from lower socioeconomic status backgrounds reported lower leve ls of connection to peers. This finding is consistent with the hypothesis that students from lower socioeconomic status backgrounds would report

PAGE 75

64 lower levels of connection to peers as they are more likely to engage in disruptive behavior, which may serve t o alienate them from peers in the school setting (Dishion, 1990; Moffitt & Caspi, 2001; Patterson et al., 1998) The Middle School Transition The strongest findings concerning mean differences in Student Adjustment Survey factors and Conn ection to Parents involved changes across the middle school transition. Findings indicated that students reported declines in functioning across the middle school transition in each of the six domains assessed. From 5 th to 6 th grade, students mean self r eports of connections to parents, peers, teachers, and school declined. Students reports of their motivation declined as well, while reports of negative expectations increased dramatically. These declines are consistent with a body of research marking the middle school transition as a pivotal turning point in development (Carnegie Council on Adolescent Development, 1989; Eccles & Wigfield, 2002; Eccles et al., 1993; Goodenow, 1993; Lord, Eccles, & McCarthy, 1994; Midgley & Edelin, 1998; Wen tzel, 2002; Wentzel, 1997) These findings provide strong support for Eccles et al.s (1993) stage environment fit perspective. Eccles et al. (1993) suggested that qualities of the middle school classroom environment represent a mismatch with the develop mental needs of early adolescents. These qualities include evidence suggesting that compared to elementary school, junior high school classrooms are characterized by a greater emphasis on teacher control and discipline, by fewer opportunities for student d ecision making, by less positive teacher student relationships, and by more competitive grading practices. Sharp declines across domains in the present study support the hypothesis that the middle

PAGE 76

65 school environment may not effectively support the developm ental needs of early adolescents. Further, declines in students connections to parents across the middle school transition support work documenting declines in the quality of parent child relations during early adolescence (Eccles et al., 1993). Student Adjustment Survey and Connection to Parents Scale: Interaction Effects Gender x Time Findings indicated that declines in functioning across the middle school transition were particularly strong for girls. Self reports of connection to teachers, connection to school, and motivation each declined to a greater degree for girls than for boys. Whereas girls had enjoyed a stronger connection to teachers and school in elementary school, self reports in these domains were essentially equal for boys and girls in mi ddle school. This finding is consistent with research suggesting that the closer relationship that girls share with their teachers in elementary school dissipates across the transition to middle school (Goodenow, 1993; Lynch & Cicchetti, 19 97; Wentzel, 2002) Prior work has also suggested that declines in girls connection to teachers in middle school may be associated with disproportionate declines in motivation among girls relative to boys across the transition to middle school (Goodenow, 1993; Wentzel, 2002; Wentzel, 1997) While the relationship between girls connection to teachers and school with levels of motivation were not tested directly, disproportionate declines in each of these domains for girls relative to boys highlights gender differences in declines across these domains found in prior work (Goodenow, 1993; Wentzel, 2002; Wentzel, 1997)

PAGE 77

66 SES x Time Findings indicated that self reports of connections to parents and peers, as well as motivatio n declined to a greater degree for students not receiving free or reduced lunch relative to students receiving free or reduced lunch. Negative expectations for school success also increased to a greater degree across the transition for students not receivi ng free or reduced lunch. These findings are analogous to those found for gender differences. These findings are troubling in that equity is only achieved through a decreased sense of connectedness and an increased sense of academic disengagement from stud ents of higher socioeconomic status across the middle school transition. However, as prior research has not examined socioeconomic differences in changes in these domains across the transition to middle school, these results must be considered preliminary. Cluster Analysis Gender, ethnic, and socioeconomic differences found in the present study in students connections to parents, peers, teachers, and school, as well as their motivation and negative expectations have been in accord with prior research. Ho wever, there are considerable gaps in our understanding of how these differences may be associated with differences in problem behavior outcomes (Dryfoos, 1990; Giordano & Cernkovich, 1997; Tucker & Herman, 2002; Yung & Hammond, 1997) A c entral goal of the present study was to incorporate a person based perspective to examine the manner through which differences in levels of risk factors may be associated with differences in outcomes experienced across gender, ethnic, and socioeconomic gro ups.

PAGE 78

67 Prior research suggests that the expression of problem behavior is associated with cumulative risk (Rutter, 1978; Rutter & Sroufe, 2000; Simmons, Burgeson, Carlton Ford, & Blyth, 1987) where a number of risk factors aggregate to ori ent a child toward expression of problem behavior. A weakness of this research is that ethnicity and socioeconomic status in particular are either themselves used as risk factors (Dryfoos, 1990; Rutter, 1978; Rutter & Sroufe, 2000) or are not examined in studies of risk for the expression of problem behavior (Yung & Hammond, 1997) Through use of a person based perspective (Magnusson, 2000; Magnusson & Bergmann, 1988) the present study moved beyond a soci al address orientation to understanding the expression of problem behavior toward a person based understanding of why boys, African Americans, and students from lower socioeconomic status may be at risk to engage in higher rates of problem behavior. Clus ter analysis results identified groups of students reporting high, average, and low levels of risk across parent, peer, teacher, school, motivation, and negative expectation domains in 5 th grade. Consistent with the cumulative risk model, the high risk gro up reported elevated levels of risk across domains (Rutter, 1978; Rutter & Sroufe, 2000) Essentially, these are students without an arena of comfort as discussed in Simmons et al.s (1987) classic study of the middle school transition. Consistent with prediction, boys, African American students, and students from low socioeconomic status backgrounds were more likely to be in the high risk group relative to girls, Caucasian students, and students from higher socioeconomic status backgroun ds. The advantage of this approach is that risk status in the present study is based upon factors that are likely amenable to change, as opposed to demographic factors and social addresses (Lochman,

PAGE 79

68 1995; Lochman, 2000) Use of a person ba sed perspective permitted examination of problem behavior outcomes both between and within gender, ethnic, and socioeconomic groups. Problem Behavior MANOVAs Classifying students into high, average, and low risk groups based upon self reports of connecti ons to others, motivation, and expectations proved highly useful in terms of broadening our understanding of students who are at risk for engagement in problem behavior following the transition to middle school through eleventh grade. While the weight of d ata supported the existence of between group differences in outcomes based upon gender, ethnicity, and socioeconomic status, unique and interactive effects of risk status supported movement toward a model in which these differences are understood at least partly in terms of differences in students connections to others, motivation, and expectations. Main Effects of Gender, Ethnicity, and Socioeconomic Status With few exceptions, results indicated that referrals and suspensions received from fifth throu gh eleventh grade differed based upon gender, ethnicity, socioeconomic status, and risk status. As predicted based upon prior research, boys, African American students, and students from lower socioeconomic status backgrounds received higher mean levels of referrals and suspensions throughout the course of the study than did girls, Caucasian students, and students from higher socioeconomic status backgrounds (Giordano & Cernkovich, 1997; Loeber & Farrington, 1997; McLoyd, 1998; Tremblay, Mas se, Pagani, & Vitaro, 1996; Tucker & Herman, 2002) Importantly, risk status was

PAGE 80

69 associated with referral and suspension outcomes throughout the course of the study independent of the effects of gender, ethnicity, and socioeconomic status. The main effe ct of risk indicates that connections, motivation, and expectations do in fact count. The systematic main effects of risk across years indicate that regardless of demographic background, childrens reports of their connections, motivation, and expectations in fifth grade relate to mean levels of referrals and suspensions received each year from fifth through eleventh grade. The implications of these effects should not be understated. No matter which demographic group a student belongs to, from lower socioe conomic status, African American boys to higher socioeconomic status, Caucasian girls, students who are connected, motivated, and optimistic have more successful academic outcomes in terms of referrals and suspensions than those who are less connected, mot ivated, and optimistic. These results are critical from a screening and prevention standpoint. The vast majority of research and prevention trials focus upon reducing levels of problem behavior among boys (Loeber & Farrington, 1997; Loebe r et al., 2002; Loeber & Stouthamer Loeber, 1998) Doing so is supported by a vast body of research indicating that boys are more likely than girls to engage in physically aggressive forms of problem behavior (Loeber & Farrington, 1997; Lo eber & Hay, 1997; Loeber & Stouthamer Loeber, 1998) and the main effect of gender upon discipline referrals and suspensions throughout the course of the present study are consistent with these findings. However, to address the overwhelming focus upon boy s in research and prevention trials focused upon problem behavior, advances have been made in understanding qualitative differences in aggression in which girls are more likely to engage in relational forms of aggression

PAGE 81

70 (Crick, 1997; Crick & Rose, 2000) Results of the present study add to our understanding of the relationship between gender and problem behavior by indicating that within group differences exist in referrals and suspensions. While girls are less likely to receive referrals and suspensions than are boys, quantitative differences exist in mean levels of referrals and suspensions among boys as well as girls based upon levels of risk. Girls who are less connected, motivated and optimistic in fifth grade are more likely receive r eferrals and suspensions from fifth to eleventh grade than those who are less connected, motivated, and optimistic. The effect of risk status also extends research examining socioeconomic status and ethnicity in relation to problem behavior. The present study addressed a clear ethnicity and socioeconomic status paradox existing not only in problem behavior research, but research in general. The paradox being that we know African American students and students from low socioeconomic status backgrounds are in general at higher risk to experience a range of problem outcomes, while the vast majority of research is conducted using Caucasian, middle class samples (Tucker & Herman, 2002; Yung & Hammond, 1997) Paucity of research examining within group differences in problem behavior outcomes perpetuates a stereotyped view of functioning among African American and lower socioeconomic status students. Finding from the present study indicate that mean differences in referrals and suspensions exist w ithin ethnic and socioeconomic groups based upon risk status. While African American students and lower socioeconomic status students do clearly receive higher levels of referrals and suspensions from fifth to eleventh grade, mean levels of referrals and suspensions differ

PAGE 82

71 within groups based upon levels of risk associated with reports of connections, motivation, and expectations in fifth grade. Main Effect of Time Prior work has documented a steady increase in problem behavior outcomes beginning in earl y adolescence that continues through high school (Donovan & Jessor, 1985) For most students in our nations schools, early adolescence is also coupled with the transition into middle school. The middle school transition has been identifie d as a key turning point in development (Carnegie Council on Adolescent Development, 1989). The present study examined students referrals and suspensions during this critical time period. Findings indicated a drastic increase in mean levels of disciplin e referrals and suspensions received across the middle school transition. For example, students averaged one discipline referral per five students, or an average of 0.22 total discipline referrals in fifth grade. In sixth grade, students averaged almost tw o discipline referrals per student, or a mean of 1.91. In fifth grade, students averaged one out of school suspension per thirty three students, or an average of .03. In sixth grade, students averaged one out of school suspension per five students, or a me an of 0.22. These results are consistent with those suggesting that early adolescence marks the beginning of an increase in problem behavior (Donovan & Jessor, 1985) Findings are also consistent with research marking the middle school t ransition as a key turning point in development (Carnegie Council on Adolescent Development, 1989) While prior research has indicated that the middle school transition is associated with declines in student functioning across several area s (Eccles, Lord, & Roeser, 1996; Eccles et al., 1993) changes in levels of discipline referrals and suspensions received have not

PAGE 83

72 previously been highlighted. This may be due in part to the primary focus upon self esteem and other constru cts assessed through survey methods used in research examining the middle school transition (Eccles et al., 1996; Eccles et al., 1993) Results of the present study strongly indicate the need to incorporate assessment of actual discipline referrals and suspensions received by students prior to and following the middle school transition. Inclusion of these outcomes in the present study complements existing research documenting declines in functioning across several key domains. Importantly the sizable changes in referrals and suspensions received from fifth to sixth grade suggest that developmental factors are not solely responsible. Moving from one referral per five students in fifth grade to two referrals per student in sixth grade stro ngly suggests that there is a disconnect between elementary and middle schools in the way that problem behavior is addressed. Elementary schools in this school district are likely to have addressed problem behavior through means other than issuing discip line referrals and suspensions. In contrast, middle school policies were likely such that formal discipline referrals and use of suspensions were utilized more frequently as a means of addressing problem behavior. Based upon this disconnect, differences in policies, rather than differences in problem behavior may have accounted for low levels of referrals and suspensions in elementary school and the sharp rise in referrals and suspensions across the middle school transition. Both the magnitude of this disc onnect, if it does in fact exist, and the degree to which it may exist in other school districts is presently unclear. Consequently, research is necessary that compares levels of problem behavior exhibited in elementary and middle school environments in re lation to the number of discipline referrals and suspensions

PAGE 84

73 received. Through doing so, research will quantify the magnitude of the disconnect, if one exists, between behavior and formal discipline referrals received from elementary to middle school. Diff erent patterns of addressing school based problem behavior through time can then be evaluated in relation to future outcomes (Atkins et al., 2002; Costenbader & Markson, 1998; Raffaele Mendez, Knoff, & Ferron, 2002) Interaction Effects Main Effects x Time Findings indicated that increases in mean levels of referrals and suspensions across the middle school transition were particularly strong for boys, African American students, low socioeconomic status students, and students at higher levels of risk. These differences are also likely due to both a normative developmental increase in problem behavior and a change in the manner through which schools address discipline problems. Consistent with the developmental model of Moffitt and Caspi (2001), these data suggest that high risk students are more likely to show increases in problem behavior beginning in early adolescence. Further, students at average levels of risk are likely to form an adolescent starter group in which problem behavior is manifested in adolescence in the absence of high levels of both risk and problem behavior in childhood. These data also suggest that shifts in the manner through which problem behavior is addressed from elementary to middle school may have a particula rly strong effect upon groups exhibiting higher levels of problem behavior. If elementary schools in this school district addressed problem behavior through means other than issuing discipline referrals and suspensions then those students exhibiting the hi ghest levels of problem behavior in elementary school are those most likely to be protected from use of referrals and

PAGE 85

74 suspensions. Consequently, the group benefiting most from lack of referrals and suspensions in response to problem behavior in elementar y school is the group most likely to show the largest increases in referrals and suspensions when problem behavior is met with formal referrals and suspensions in middle school. This shift toward institutionalized punitive responses to students behavior likely creates a feedback loop consisting of problem behavior and institutional punishment. If punishment is provided in the absence of remediation, the goal for students then necessarily becomes to not get into trouble, or to avoid being caught and pun ished for offenses. To the degree to which these supports are absent in the middle school setting, students who engage in problem behavior are likely to adopt a view of schooling and perhaps institutions, laws, and society in general that is more adversari al than communal (Atkins et al., 2002; Costenbader & Markson, 1998; Raffaele Mendez et al., 2002) Findings of the present study suggest that this effect may be particularly strong for boys, African American students, students from lower s ocioeconomic status backgrounds, and those at higher levels of risk. Ethnicity x SES (x Time) While differences across the middle school transition were stronger for African American students and students from lower socioeconomic status, findings also i ndicated that the combination of ethnicity and socioeconomic status was associated with multiplicative increases in referrals and suspensions across the middle school transition. From fifth to sixth grade, mean differences between regular and free/reduced lunch groups in referrals and suspensions became larger for African American students relative to Caucasian students. This effect is illustrated by African American students from lower

PAGE 86

75 socioeconomic status backgrounds who change from receiving an average of 2.47 total referrals in fifth grade to an average of 5.88 total referrals in sixth grade. These findings indicate that the disconnect described above is magnified for African American students from lower socioeconomic status backgrounds (Atkins et al., 2002; Costenbader & Markson, 1998; Raffaele Mendez et al., 2002) Findings further indicated that the combination of ethnicity and socioeconomic status were associated with multiplicative increases in referrals and suspensions within year s of the study. Effects were significant in all but ninth and tenth grade. Low socioeconomic status was associated with higher elevations in mean referrals and suspensions for African American students relative to Caucasian students. These results are con sistent with prior studies indicating that students of low socioeconomic status and African American ethnicity are more likely to receive disciplinary actions throughout their school careers relative to their higher SES and Caucasian peers (Atkins et al., 2002; Costenbader & Markson, 1998) Ethnicity x Gender (x Time) Multiplicative effects were also found for the combination of ethnicity and gender. However, these findings were restricted to fifth and sixth grades only. Findings indic ated that stronger gender differences in referrals and suspensions existed for African American students relative to Caucasian students. African American boys received the highest numbers of referrals and suspensions in fifth and sixth grade. African Amer ican boys also received the largest increases in referrals and suspensions from fifth to sixth grade. These findings are consistent with prior work indicating the particular difficulties experienced by African American males in the school environment

PAGE 87

76 (Atkins et al., 2002; Costenbader & Markson, 1998; Raffaele Mendez et al., 2002) They also suggest that African American boys are most likely to become alienated from the school environment across the middle school transition as they experien ce an increase from a mean of .91 total referrals in fifth grade to a mean of 6.67 total referrals in sixth grade. SES x Gender (x Time) Findings indicated a multiplicative effect of socioeconomic status and gender as well. However, this effect was rest ricted primarily to violence referrals in fifth sixth and seventh grade. In each case, gender differences were stronger among lower socioeconomic status students than among higher socioeconomic status students, with boys receiving higher mean referrals Boys from lower socioeconomic status backgrounds also experienced the largest increase in mean violence referrals from fifth to sixth grade. These findings represent the only case in which there was a clear distinction based upon the type of referral. O therwise, throughout the course of the study, distinctions were not apparent between mean differences in violence, classroom, and total referrals. That this distinction was apparent in terms of violence referrals in fifth sixth and seventh grades is co nsistent with prior work indicating that boys from lower socioeconomic status backgrounds experience multiplicative levels of risk in terms of engagement in violent behavior (Loeber et al., 2002; McLoyd, 1998; Rutter et al., 1979; Sameroff & Chandler, 1975) Findings further indicate that these differences are associated with differential levels of punishment received in the school environment and that these differences are magnified across the middle school transition.

PAGE 88

77 Risk x SES (x T ime) While multiplicative effects involving gender, ethnicity, and socioeconomic status found in the present study support prior work examining differences in levels of problem behavior, multiplicative effects involving risk extend prior work by indicatin g that particular subgroups of students within levels of gender, ethnicity, and socioeconomic status are at multiplicative risk to experience elevated levels of referrals and suspensions. Multiplicative effects involving socioeconomic status indicated that mean differences in levels of total and classroom referrals, as well as in school and out of school suspensions from fifth to seventh grade between students receiving regular lunch and students receiving free or reduced lunch were larger at higher levels of risk. For example, in the low risk group, regular lunch students received 1.03 total referrals in seventh grade while free/reduced lunch students received 3.03 referrals. In contrast, among high risk students, those receiving regular lunch received 3.1 5 total referrals, while those receiving free or reduced lunch received an average of 6.79 total referrals. These findings indicated that the effect of socioeconomic status upon referrals and suspensions was strongest for high risk students. Conversely, the effect of socioeconomic status was weakest for low risk students who reported higher levels of connections and motivation, and lower levels of negative expectations in fifth grade. While risk status has an additive effect across all outcomes and years, a particularly strong multiplicative effect exists during the crucial middle school transition period for lower socioeconomic status students.

PAGE 89

78 Risk x Ethnicity (x Time) Similar results were obtained in terms of ethnicity. Findings indicated that mea n differences in total referrals, in school, and out of school suspensions become larger across levels of risk during fifth and sixth grade. For example, among the low risk group in sixth grade, Caucasian students received an average of 0.70 total referra ls and African American students received an average of 3.43 total referrals. In contrast, among the high risk students in sixth grade, Caucasian students received an average of 2.97 total referrals while African American students received an average of 8. 23 total referrals. Taken together, findings indicate that stronger connections to others, motivation, and optimism expressed in fifth grade are associated with substantially reduced mean levels of referrals and suspensions during the middle school trans ition period for African American students and students from lower socioeconomic status backgrounds. While main effects of risk were found across all years of the study for all students, these results indicate that these factors are associated with particu larly large reductions in levels of referrals and suspensions during this critical time period (Carnegie Council on Adolescent Development, 1989) for African American students and those from low socioeconomic backgrounds in particular. Ethnicity x SES x Gender x Risk Higher order multiplicative effects involving three and four way interactions were confined to the middle school years. Findings indicated that heightened levels of connections to others, motivation, and lower levels of neg ative expectations in fifth grade served a particularly protective effect for students who are members of two or more traditionally high risk groups, including boys, African American students, and students

PAGE 90

79 from lower socioeconomic status backgrounds. These findings involving higher order interactions did not generally conform to a clear pattern. However, findings involving total referrals and classroom referrals in seventh and eighth grade did conform to a noteworthy pattern. Multiplicative elevations in m ean levels of total referrals and classroom referrals were found among lower socioeconomic status African American and Caucasian boys in high risk and average risk groups. These findings are exemplified by an average of 11.25 total referrals received by lo wer socioeconomic status, African American boys in the high risk group in seventh grade. These findings are consistent with prior work indicating that African American boys receive the highest levels of disciplinary actions in our nations schools (Tucker & Herman, 2002) These findings extend prior work through demonstrating that effects are strongest among lower socioeconomic status, African American boys who reports lower levels of connections, motivation, and optimism during fifth grade While these multiplicative effects are most striking in terms of outcomes, the combination of main effects alone in the present study indicates that high risk, African American boys of low socioeconomic status received the highest levels of referrals a nd suspensions throughout the course of the study. For example, although not associated with a significant four way interaction, high risk, African American males of low socioeconomic status averaged approximately two out of school suspensions and four in school suspensions during each year of middle school. These means were clearly the highest among all subgroups in the study. The potential psychological effect of receiving, on the average, eleven discipline referrals, four in school suspensions, and two o ut of

PAGE 91

80 school suspensions in seventh grade alone for high risk African American boys of low socioeconomic status cannot be understated. These findings speak to the potential of middle school discipline policies to both reflect and encourage movement of av erage and high risk boys from low socioeconomic status backgrounds, and African American boys in particular, on a path toward eventual school dropout. These results may support work suggesting that many dropouts are in fact push outs (Raf faele Mendez et al., 2002) The push out perspective holds that students experiencing heightened levels of behavioral difficulties are in effect pushed out of the schooling process through disciplinary actions and suspensions that increase the likelihoo d that a student with behavioral difficulties will drop out of school (Raffaele Mendez et al., 2002) Students demonstrating problem behavior place higher demands upon teachers, administrators, and other school personnel than do students w ho do not engage in problem behavior. Consequently, investigators have argued that there are several practical benefits that may tempt administrators to utilize such practices (Raffaele Mendez et al., 2002) Dropout Chi Square Analyses Fi ndings indicated that students from lower socioeconomic status and those of African American ethnicity were more likely to drop out of school. These results were consistent with prior research indicating that students of low socioeconomic status and Africa n American ethnicity have higher rates of dropout relative to their higher socioeconomic status and Caucasian counterparts (McLoyd, 1998; Tucker & Herman, 2002) Findings also indicated that boys were more likely to dropout of school relat ive to girls. These findings contrast with prior research suggesting that gender differences do

PAGE 92

81 not exist in rates of dropout (Davis & Jordan, 1994; Dryfoos, 1990) This discrepancy may be associated with the manner through which dropout w as assessed in the present study. Investigators have lamented that dropout statistics provided by school districts often represent underestimates of the true prevalence of dropout (Doll, 1997; Doll & Hess, 2001) The design of the present study incorporated a more accurate assessment of dropout than generally provided by school districts. Consequently, the power to detect valid differences between groups was very likely increased. Given that the effect of gender upon dropout was small, it i s possible that the increased power and validity afforded by the design of the current study heightened the ability to detect gender differences. While findings associated with ethnicity, socioeconomic status, and gender replicated and extended prior work, the central focus of the present study involved the effect of risk status upon student dropout. Strong findings were obtained for risk status in which students reporting lower levels of connections to others, motivation, and optimism in fifth grade were m ore likely to dropout of school by the completion of eleventh grade. This finding represents a critical advance in terms of understanding factors associated with student dropout (Doll & Hess, 2001) Several review articles published in rec ent years have lamented the lack of progress made by research focused upon understanding the phenomenon of student dropout (Christenson, Sinclair, Lehr, & Godber, 2001; Doll & Hess, 2001; Rosenthal, 1998) These reviews and commentaries no te that traditional divisions between education and psychological research have hindered growth in understanding the nature of student dropout. They emphasize that dropout has generally been examined as if it were a

PAGE 93

82 secondary school issue linked primarily to the static demographic characteristics of the student (Doll, 1997; Doll & Hess, 2001) Recommendation have called for longitudinal research examining the precursors to dropout beginning in the elementary grades as well as a shift from e xamining dropout primarily in terms of demographic characteristics to inclusion of psychological and behavioral precursors to school dropout and completion that could be altered through intervention (Doll, 1997; Doll & Hess, 2001) The d esign and results of the present study provides a foundation for future research examining student dropout. Successful identification of a high risk group of students in elementary school, who then engage in the highest rates of dropout, supports the need to examine student dropout from a developmental perspective, with roots in the elementary school years (Doll, 1997; Doll & Hess, 2001) The need to examine early elementary school precursors to student dropout is not a new idea (Barclay, 1966; Barclay & Doll, 2001; Fitzsimmons, Cheever, Leonard, & Macunovich, 1969; Kuhlen & Collister, 1952) However, early prospective studies of high school dropout were not followed up with programmatic lines of inquiry into the nature of t he dropout problem grounded in development from elementary school onward (Barclay & Doll, 2001; Doll & Hess, 2001) As such, the present study serves to refocus examination of dropout as a developmental issue. Limitations There are severa l ways in which future studies can improve upon the design of the present study. Foremost is the importance of grounding future work in more precise measures. The measures used in the present study consistently demonstrated discriminant and predictive vali dity. This was true whether examined individually, or through their

PAGE 94

83 collective ability to differentiate students at different levels of risk whom subsequently engaged in different levels of future problem behavior. Clearly, though, the measures used can an d should be improved upon in future longitudinal studies of problem behavior. Future longitudinal studies should incorporate standardized measures that are either existing or developed to assess constructs central to the development of problem behavior (Dryfoos, 1990) A second limitation of the present study is that the measures used to identify student risk status were based entirely upon students self reports. Reports of connections to others, motivation, and expectations are influence d by the perceptions of the informant. Given generally low agreement found between raters using survey methods (Achenbach, McConaughy, & Howell, 1987) it is almost always optimal to incorporate measures from multiple sources. The influenc e of method bias was reduced in the present study by using outcomes that were not based upon student reports. However, future work intended to identify students at risk for school based problem behavior would be enhanced through use of multiple measures de rived from multiple informants or sources. A third limitation of the present study is that special education students were underrepresented. Underrepresentation of special education students shed light on the magnitude of difficulties faced by the entire population of students. The magnitude of problem behavior difficulties experienced by students in the present study reminds us that that the mental health needs of four fifths of our nations schoolchildren are not being met (Services, 1999 ) and that effective universal (Durlak & Wells, 1997) and selective (Durlak & Wells, 1998) prevention practices are necessary to address the difficulties that these students face. Nevertheless, future work must also focus upon the development of

PAGE 95

84 problem behavior among the most seriously disturbed students in our nations schools who have been placed in special education classes by the fifth grade. Future work is necessary to examine the degree to which findings of the pre sent study are generalizable to special education students. By definition, emotionally handicapped and severely emotionally disturbed students engage in higher levels of problem behavior than do regular education students. They are also more likely to drop out of school (Dryfoos, 1990) The degree to which the middle school transition affects special education students relative to regular education students is presently unclear. Also unclear is the degree to which variability within the spec ial education population with regard to students connections to others, motivation, and expectations is associated with discipline referrals, suspensions, and dropout through time. Obtaining a clearer understanding of these effects among special education students is necessary considering both the magnitude of their behavioral difficulties and the personnel and financial investment necessary to address their education needs. A fourth way in which future longitudinal work can improve upon the design of the present study is through examination of the precursors of childrens fifth grade reports. Examining the reports of fifth grade students in the present study was logical for several reasons. Prior research indicates that students face particular challenges associated with the middle school transition (Eccles et al., 1993) Research also indicates that the early adolescent period is marked by a rise in problem behavior (Donovan & Jessor, 1985) Fifth grade students can also provide reports in a large scale survey design that can be expected to have a degree of reliability and validity that minimizes the potential of a Type II error due to unreliability of measurement. However, future

PAGE 96

85 longitudinal research beginning early in e lementary school is necessary to more fully understand the precursors of students reports in fifth grade, which in turn are associated with future development of problem behavior. From an applied standpoint, lessons learned from such investigations can in form future selective prevention efforts (Durlak & Wells, 1998; Lochman, 1995) Future applied work in this area must consider both the statistical and clinical significance of results obtained. A limitation of the present study was that, due to the large number of participants, there were statistically significant effects that may not be clinically significant. For example, while statistically significant, the mean difference in motivation scale scores between boys (2.87) and girls (2.92) does not provide clinical support in favor of directing more resources intended for school motivation enhancement toward boys. In contrast, the mean difference in total referrals from 5 th grade (0.22) to 6 th grade (1.91) may have strong clinical implicati ons for students experiences in the school environment. A word of caution is necessary to not automatically disregard what may appear to be a small effect. Sometimes a small effect can have large clinical implications when the outcome is severe and the population under investigation is large. For example, the mean difference in out of school suspensions in 6 th grade between students in the low risk group (0.11) and students in the high risk group (0.49) may not appear large or clinically significant. How ever, in a population of 8098 fifth grade students, a mean of 0.11 would equal 890 out of school suspensions across the district, whereas a mean of 0.49 would equal 3968 out of school suspensions. The difference between having a district with all high risk 6 th grade students and a district of all low risk 6 th grade students would amount

PAGE 97

86 to more than 3000 out of school suspensions. That is clinically significant by any standard. Finally, as research advances utilizing complementary variable and person focu sed perspectives it is necessary to incorporate a broader range of independent variables hypothesized to be associated with the development of problem behavior. In this regard, indices of sociocultural and economic processes must be included in future inve stigations designed to more fully understand the manner through which SES and ethnicity are independently associated with the development of problem behavior. Further, incorporation of a broader range of outcome variables including measures of depression (Kovacs, 1992; Nolen Hoeksema & Girgus, 1994) relational aggression (Crick, 1997; Crick & Rose, 2000) and academic achievement is necessary to more fully understand heterotypic manifestations of pathology and resilience f rom childhood through adolescence. Conclusions Findings presented both supported and extended existing research concerned with understanding the expression of problem behavior. Findings supported existing research indicating that boys, African American s tudents, and students from lower socioeconomic status backgrounds generally experience higher levels of risk factors found to be associated with the development of problem behavior including connections to parents, peers, teachers, and school, as well as m otivation and negative expectations (Dryfoos, 1990) Findings also supported existing research indicating that boys, African American students, and students from lower socioeconomic status backgrounds engage in higher levels of problem beh avior relative to girls, Caucasian students, and students from higher

PAGE 98

87 socioeconomic status backgrounds (Dryfoos, 1990; Giordano & Cernkovich, 1997; McLoyd, 1998; Yung & Hammond, 1997) Findings presented extended this research in several key ways. Understanding within group differences in the expression of problem behavior served as the central focus of the present study (Magnusson, 2000; Magnusson & Bergmann, 1988) This focus came in response to calls to move beyond a st atic, demographic based understanding of problem behavior (Doll & Hess, 2001; Garcia Coll, Akerman, & Cicchetti, 2000; Tucker & Herman, 2002) Results indicated that considerable variability existed within groups in mean levels of problem behavior outcomes. Regardless of demographic status, stronger connections to others, motivation, and optimism expressed in fifth grade was associated with lower levels of discipline referrals and suspensions from fifth grade onward. Conversely, poorer conn ections, lower motivation, and heightened negative expectations were associated with higher mean levels of referrals and suspensions. In several instances the effect of risk status was particularly strong for boys, African American students, and students f rom lower socioeconomic status backgrounds. Risk status was also strongly associated with student dropout. These effects of cumulative and multiplicative risk found in the present study serve to advance our understanding of problem behavior outcomes thro ugh consideration of both between and within group differences in levels of risk and outcomes (Doll & Hess, 2001; Garcia Coll et al., 2000; Tucker & Herman, 2002) By grounding research in terms of factors amenable to change through preven tive intervention, the present study provides a template for future research examining gender, ethnic, and socioeconomic factors associated with problem behavior (Doll & Hess, 2001; Tucker & Herman, 2002)

PAGE 99

88 as well as prevention trials inte nded to diminish the incidence and prevalence of problem behavior (Lochman, 1995) Findings of the present study also served to extend prior work examining changes in student adjustment across the middle school transition (Carnegie Council on Adolescent Development, 1989; Eccles et al., 1993) Results supported prior work indicating that early adolescence is a time in which levels of problem behavior increase (Donovan & Jessor, 1985) Importantly, this re search is extended through presentation of drastic increases across the middle school transition in levels of referrals and suspensions received, as well as the particularly strong effect of the middle school transition upon referrals and suspensions recei ved by boys, high risk students, African American students, and students from lower socioeconomic status backgrounds. Evidence of a potential disconnect in students experience of school based responses to problem behavior provided critical information p otentially supporting the hypothesis that students may feel pushed out through the course of middle school (Raffaele Mendez et al., 2002) These findings provide strong support for the stage environment fit perspective of Eccles et al. (1993). Consistent with this model, findings suggest that the middle school environment may not provide structure or challenges consistent with middle school students developmental level of maturity (Eccles et al., 1993; Hunt, 1975) Com pared to elementary school, junior high school classrooms are characterized by a greater emphasis on teacher control and discipline, by fewer opportunities for student decision making, by less positive teacher student relationships, by more competitive gra ding practices, and by a reduced sense of teaching efficacy among junior high school teachers (Eccles et al.,

PAGE 100

89 1993; Hunt, 1975) Declines in connections to others, motivation, and expectations across the transition to middle school are li kely associated with this stage environment mismatch. These findings provide support for the necessity of ecological approaches to prevention work in schools (Cowen, 1997; Cowen & Work, 1988 ; Weissberg & Greenberg, 1997) ) Consistent with a transactional ecological model of preventive intervention, Felner and his colleagues (Felner et al., 1997) have achieved success through their School Transitional Environment Program (STEP). Through restructuring the m iddle school environment in accord with the recommendations of the Carnegie Task Force on Adolescent Development (1989), STEP has been associated with approximately 50% reductions in drop out rates and significant positive effects on school performance and attendance patterns. Increased implementation of ecological prevention approaches such as those incorporated into STEP are likely necessary to address the significant declines in functioning across the middle school transition by students in the present s tudy. Taken as a whole, these conclusions support movement toward a more holistic understanding of gender, ethnic, socioeconomic, ecological, and developmental factors associated with school based problem behavior. Findings support the importance of conne ctions, expectations, and motivation in students lives. Findings also support the necessity of advancing our understanding of connections between schooling and mental health (Roeser et al., 1998) Through doing so, the potential exists to considerably increase the likelihood that all of our nations children are provided with equal opportunity to achieve their fullest potential.

PAGE 101

90 References Achenbach, T., McConaughy, S., & Howell, C. (1987). Child/Adolescent behavior al and emotional problems: Implications of cross informant correlations for situational specificity. Psychological Bulletin, 101(2), 213 232. Alpert, G., & Dunham, R. (1986). Keeping academically marginal youths in school. Youth and Socie ty, 17 346 361. Atkins, M., McKay, M., Frazier, S., Jaksbsons, L., Arvanitis, P., Cunningham, T., Brown, C., & Lambrecht, L. (2002). Suspensions and detentions in an urban, low income school: Punishment or reward? Journal of Abnormal Child Psychology, 30 (4), 361 371. Bachman, J., Green, S., & Wirtanen, I. (1971). Youth in transition (Vol. 3). Ann Arbor, MI: University of Michigan, Institute for Survey Research. Barclay, J. (1966). Sociometric choices and teacher ratings as predictors of school dropout. Journal of School Psychology, 4, 40 44. Barclay, J., & Doll, B. (2001). Early prospective studies of the high school dropout. School Psychology Quarterly, 16 ( 4), 357 369. Borkowski, J., Ramey, S., & Bristol Power, M. (2002). Parenting and the child's wo rld: Influences on academic, intellectual, and social emotional development Mahwah, NJ: Erlbaum. Carnegie Council on Adolescent Development. (1989). Turning Points: Preparing American youth for the 21st century New York: Carnegie Corporation.

PAGE 102

91 Christens on, S., Sinclair, M., Lehr, C., & Godber, Y. (2001). Promoting successful school completion: Critical conceptual and methodological guidelines. School Psychology Quarterly, 16(4), 468 484. Coie, J., Lochman, J., Terry, R., & Hyman, C. (1992). Predicting e arly adolescent disorder from childhood aggression and peer relations. Journal of Consulting and Clinical Psychology, 60 783 792. Costenbader, V., & Markson, S. (1998). School suspension: A study with secondary school students. Journal of School Psycholo gy, 36(1), 59 82. Council on Crime in America (1996). The state of violent crime in America Washington DC: New Citizenship Project. Cowen, E. (1997). The coming of age of primary prevention: Comments on Durlak and Wells's meta analysis. American Journal of Community Psychology, 25(2), 153 167. Cowen, E., & Work, W. (1988). Resilient children, psychological wellness, and primary prevention. American Journal of Community Psychology, 16(4), 591 607. Crick, N. (1997). Engagement in gender normative versus nonnormative forms of aggression: Links to social psychological adjustment. Developmental Psychology, 33(4), 610 617. Crick, N., & Rose, A. (2000). Toward a gender balanced approach to the study of social emotional development: A look at relational aggres sion. In P. Miller & E. Kofsky Scholnick (Eds.), Toward a feminist developmental psychology New York: Routledge. Davis, H. (2001). The quality and impact of relationships between elementary school students and teachers. Contemporary Educational Psycholog y, 26, 431 453.

PAGE 103

92 Davis, J., & Jordan, W. (1994). The effects of school context, structure, and experiences on African American males in middle and high school. Journal of Negro Education, 63(4), 570 587. Dishion, T. (1990). The peer context of troublesome child and adolescent behavior. In P. Leone (Ed.), Understanding troubled and troubling youth (pp. 128 153). Newbury Park, CA: Sage. Dishion, T., Patterson, G., Stoolmiller, M., & Skinner, M. (1991). Family, school, and behavioral antecedents to early ado lescent involvement with antisocial peers. Developmental Psychology, 27 172 180. Dodge, K., Coie, J., & Brakke, N. (1982). Behavior patterns of socially rejected and neglected preadolescents: The roles of social approach and aggression. Journal of Abnorm al Child Psychology, 18, 389 409. Doll, B. (1997). Dropout prevention knowledge base: Preliminary findings (A report of the Interdivisional Task Force on School Dropout Prevention to the APA committee on children, youth and families). Washington, DC: Amer ican Psychological Association. Doll, B., & Hess, R. (2001). Through a new lens: Contemporary psychological perspectives on school completion and dropping out of school. School Psychology Quarterly, 16(4), 351 356. Donovan, J., & Jessor, R. (1985). Struc ture of problem behavior in adolescence and young adulthood. Journal of Counseling and Clinical Psychology, 53, 890 904. Dryfoos, J. (1990). Adolescents at risk: Prevalence and presentation Oxford: Oxford University Press.

PAGE 104

93 Durlak, J., & Wells, A. (1997) Primary prevention mental health programs for children and adolescents: A meta analytic review. American Journal of Community Psychology, 52(2), 115 152. Durlak, J., & Wells, A. (1998). Evaluation of indicated preventive intervention (secondary preventi on) mental health programs for children and adolescents. American Journal of Community Psychology, 26, 775 802. Eccles, J., & Wigfield, A. (2002). Motivational beliefs, values, and goals. Annual Review of Psychology, 53, 109 132. Eccles, J. S., Lord, S. E., & Roeser, R. W. (1996). Round holes, square pegs, rocky roads, and sore feet: The impact of stage environment fit on young adolescents' experiences in schools and families. In D. Cicchetti & S. Toth (Eds.), Adolescence: Opportunities and challenges (pp 47 92). Rochester, NY: University of Rochester Press. Eccles, J. S., Midgley, C., Wigfield, A., Buchanan, C., Reuman, D., Flanagan, C., & MacIver, D. (1993). Development during adolescence: The impact of stage environment fit on adolescents' experiences in schools and in families. American Psychologist, 48(2), 90 101. Ekstrom, R., Goertz, M., Pollack, J., & Rock, D. (1986). Who drops out of high school and why? Findings from a national study. In G. Natriello (Ed.), School dropouts patterns and policies (pp. 52 69). New York: Teachers College Press. Elliot, D. (1994). Serious violent offenders: Onset, developmental course, and termination The American Society of Criminology 1993 presidential address. Criminology, 32 1 21.

PAGE 105

94 Felner, R., Favazza, A., Shim M., Brand, S., Gu, K., & Noonan, N. (2001). Whole school improvement and restructuring as prevention and promotion: Lessons from STEP and the Project on High Performance Learning Communities. Journal of School Psychology, 39(2), 177 202. Felner, R., Jac kson, A., & Kasak, D. (1997). The impact of school reform for the middle years: Longitudinal study of a network engaged in Turning Points based comprehensive school transformation. Phi Delta Kappan, 78, 528 532. Felner, R., Jackson, A. W., Kasak, D., Mulh all, P., Brand, S., & Flowers, N. (1997). The impact of school reform for the middle grades: A longitudinal study of a network engaged in Turning Points based commprehensive school transformation. In R. Takanishi & D. Hamburg (Eds.), Preparing adolescents for the twenty first century: Challenges facing Europe and the United States (pp. 38 69). New York: Cambridge University. Fitzsimmons, S., Cheever, J., Leonard, E., & Macunovich, D. (1969). School failures: Now and tomorrow. Developmental Psychology, 1, 1 34 146. Garcia Coll, C., Akerman, A., & Cicchetti, D. (2000). Cultural influences on developmental processes and outcomes: Implications for the study of development and psychopathology. Developmental Psychopathology, 12, 333 356. Giordano, P., & Cernkovi ch, S. (1997). Gender and antisocial behavior. In D. Stoff, J. Breiling, & J. Maser (Eds.), Handbook of antisocial behavior (pp. 496 510). New York: Wiley.

PAGE 106

95 Goodenow, C. (1993). Classroom belonging among early adolescent students: Relationships to motivati on and achievement. Journal of Early Adolescence, 13(1), 21 43. Hamre, B., & Pianta, R. (2001). Early teacher child relationships and the trajectory of children's school outcomes through eighth grade. Child Development, 72 ( 2), 625 638. Hanson, S., & Gins burg, A. (1988). Gaining ground: Values and high school success. American Educational Research Journal, 25, 334 365. Hinshaw, S., & Park, T. (1999). Research problems and issues: Toward a more definitive science of disruptive behavior disorders. In H. Qua y & A. Hogan (Eds.), Handbook of disruptive behavior disorders (pp. 593 620). New York: Kluwer. Hirschi, T. (1969). Causes of delinquency Berkeley, CA: University of California Press. Hughes, J., Cavell, T., & Jackson, T. (1999). Influence of the teache r student relationship on childhood conduct problems: A prospective study. Journal of Clinical Child Psychology, 28 ( 2), 173 184. Hughes, J., Cavell, T., & Willson, V. (2001). Further support for the developmental significance of the quality of the teacher student relationship. Journal of School Psychology, 39(4), 289 301. Hunt, D. (1975). Person environment interaction: A challenge found wanting before it was tried. Review of Educational Research, 45, 209 230. Jessor, R., & Jessor, S. (1 977). Problem behavior and psychosocial development New York: Academic Press.

PAGE 107

96 Jeynes, W. (2002). The challenge of controlling for SES in social science and education research. Educational Psychology Review, 14(2), 205 221. Jimerson, S., Egeland, B., Sro ufe, L., & Carlson, B. (2000). A prospective longitudinal study of high school dropouts: Examining multiple predictors across development. Journal of School Psychology, 38(6), 525 549. Jones, M., & Krisberg, B. (1994). Images and reality: Juvenile crime, youth violence and public policy San Francisco: National Council on Crime and Delinquency. Ketterlinus, R., & Lamb, M. (Eds.). (1994). Adolescent problem behaviors: Issues and research Hillsdale, NJ: Erlbaum. Kovacs, M. (1992). Children's Depression In ventory manual North Tonawanda, NY: Multi Health Systems Inc. Kuhlen, R., & Collister, E. (1952). Sociometric status of sixth and ninth graders who fail to finish high school. Educational and Psychological Measurement, 12, 632 637. Kupersmidt, J., Coie, J., & Dodge, K. (1990). The role of poor peer relationships in the development of disorders. In S. Asher & J. Coie (Eds.), Peer rejection in childhood (pp. 274 305). New York: Cambridge University Press. Lerner, R., & Galambos, N. (1998). Adolescent deve lopment: Challenges and opportunities for research, programs, and policies. Annual Review of Psychology, 49, 413 446. Lewin, K. (1935). A dynamic theory of personality New York: McGraw Hill. Lochman, J. (1995). Screening of child behav ior problems for prevention programs at school entry. Journal of Consulting and Clinical Psychology, 63, 549 559.

PAGE 108

97 Lochman, J. (2000). Theory and empiricism in intervention research: A dialectic to be avoided. Journal of School Psychology, 38(4), 359 368. Loeber, R., & Farrington, D. (1997). Strategies and yields of longitudinal studies on antisocial behavior. In D. Stoff, J. Breiling, & J. Maser (Eds.), Handbook of antisocial behavior (pp. 125 139). New York: Wiley. Loeber, R., Farrington, D., Stouthamer Loeber, M., Moffitt, T., Caspi, A., & Lynam, D. (2002). Male mental health problems, psychopathy, and personality traits: Key findings from the first 14 years of the Pittsburgh Youth Study. Clinical Child and Family Psychology Review, 4(4), 273 297. Loeb er, R., & Hay, D. (1997). Key issues in the development of aggression and violence from childhood to early adulthood. Annual Review of Psychology, 48, 371 410. Loeber, R., & Stouthamer Loeber, M. (1998). Development of juvenile aggression and violence: So me misconceptions and controversies. American Psychologist, 53, 242 259. Lord, S. E., Eccles, J. S., & McCarthy, K. A. (1994). Surviving the junior high school transition: Family processes and self perceptions as protective and risk factors. Journal of Ea rly Adolescence, 14 ( 2), 162 199. Lynch, M., & Cicchetti, D. (1997). Children's relationships with adults and peers: An examination of elementary and junior high school students. Journal of School Psychology, 35, 81 99. Maccoby, E., & Martin, J. (1983). S ocialization in the context of the family: Parent child interaction. In P. Mussen (Ed.), Handbook of child psychology: Socialization, personality, and social development (Vol. 4, pp. 1 101). New York: Wiley.

PAGE 109

98 Magnusson, D. (2000). The individual as the org anizing principle in psychological inquiry: A holistic approach. In L. Bergman, R. Cairns, L. Nilsson, & L. Nystedt (Eds.), Developmental science and the holistic approach (pp. 33 47). Magnusson, D., & Bergmann, L. (1988). Individual and variable based approaches to longitudinal research on early risk factors. In M. Rutter (Ed.), Studies of psychosocial risk: The power of longitudinal data (pp. 45 61). New York: Cambridge University Press. Marcus, R., & Sanders Reio, J. (2001). The influence of attachme nt on school completion. School Psychology Quarterly, 16(4), 427 444. McAdoo, H., & McAdoo, J. (1985). Black children: Social, educational, and parental environments Beverly Hills, CA: Sage. McLoyd, V. (1998). Socioeconomic disadvantage and child devel opment. American Psychologist, 53(2), 185 204. Midgley, C., & Edelin, K. (1998). Middle school reform and early adolescent well being: The good news and the bad. Educational Psychologist, 33, 195 206. Moffitt, T., & Caspi, A. (2001). Childhood predictors differentiate life course persistent and adolescence limited antisocial pathways among males and females. Development and Psychopathology, 13, 355 375. Murray, C., & Greenberg, M. (2000). Children's relationship with teachers and bonds with school: An in vestigation of patterns and correlates in middle childhood. Journal of School Psychology, 38(5), 423 445.

PAGE 110

99 Najaka, S., Gottfredson, D., & Wilson, D. (2001). A meta analytic inquiry into the relationship between selected risk factors and problem behavior. P revention Science, 2, 257 271. Nolen Hoeksema, S., & Girgus, J. (1994). The emergence of gender differences in depression during adolescence. Psychological Bulletin, 115, 424 443. Parker, J., & Asher, S. (1987). Peer relations and later personal adjustme nt, are low accepted children at risk? Psychological Bulletin, 10, 357 389. Patterson, G. (1993). Orderly change in a stable world: The antisocial trait as a chimera. Journal of Consulting and Clinical Psychology, 61, 911 919. Patterson, G., Forgatch, M. Yoerger, K., & Stoolmiller, M. (1998). Variables that initiate and maintain an early onset trajectory for juvenile offending. Development and Psychopathology, 10, 531 547. Pianta, R. (1999). Enhancing relationships between children and teachers Washing ton DC: American Psychological Association. Polite, V. (1994). The method in the madness: African American males, avoidance schooling, and chaos theory. Journal of Negro Education, 63(4), 588 601. Raffaele Mendez, L., Knoff, H., & Ferron, J. (2002). Scho ol demographic variables and out of school suspension rates: A quantitative and qualitative analysis of a large, ethnically diverse school district. Psychology in the Schools, 39(3), 259 277. Richters, J. (1997). The Hubble hypothesis and the developmenta list's dilemma. Development and Psychopathology, 9, 193 229.

PAGE 111

100 Roeser, R. W., Eccles, J. S., & Strobel, K. R. (1998). Linking the study of schooling and mental health: Selected issues and empirical illustrations at the level of the individual. Educational P sychologist, 33(4), 153 176. Rosenthal, B. S. (1998). Non school correlates of dropout: An integrative review of the literature. Children & Youth Services Review, 20(5), 413 433. Rothbaum, F., & Weisz, J. R. (1994). Parental caregiving and child external izing behavior in nonclinical samples: A meta analysis. Psychological Bulletin, 116(1), 55 74. Rumberger, R. (1987). High school dropouts: A review of issues and evidence. Review of Educational Research, 57, 101 121. Rumberger, R., Ghatak, R., Poulos, G. Ritter, P., & Dornbusch, S. (1990). Family influences on dropout behavior in one California high school. Sociology of education, 63, 283 299. Rutter, M. (1978). Family, area, and school influences in the genesis of conduct disorders. In L. Hersov, M. Be rger, & D. Schaffer (Eds.), Aggression and antisocial behavior in childhood and adolescence (pp. 95 114). New York: Pergamon. Rutter, M., Maughan, B., Mortimore, P., & Ouston, J. (1979). Fifteen thousand hours: Secondary schools and their effects on child ren Cambridge, MA: Harvard University Press. Rutter, M., & Sroufe, L. (2000). Developmental psychopathology: Concepts and challenges. Development and Psychopathology, 12, 265 296.

PAGE 112

101 Sameroff, A., & Chandler, M. (1975). Reproductive risk and the contin uum of caretaking causality. In F. Horowitz, M. Hetherington, S. Scarr Salapatek, & G. Siegal (Eds.), Review of Child Development Research (Vol. 4, pp. 187 244). Chicago: University of Chicago Press. Simmons, R. G., Burgeson, R., Carlton Ford, S., & Blyth A. (1987). The impact of cumulative change in early adolescence. Child Development, 58(5), 1220 1234. Steele, C. (1997). A threat in the air: How stereotypes shape intellectual identity and performance. American Psychologist, 52, 613 629. Tremblay, R., LeBlanc, M., & Schwartzman, A. (1988). The predictive power of first grade peer and teacher ratings of behavior: Sex differences in antisocial behavior and personality at adolescence. Journal of Abnormal Child Psychology, 16, 571 583. Tremblay, R., Masse L., Pagani, L., & Vitaro, F. (1996). From childhood aggression to adolescent maladjustment: The Montreal prevention experiment. In R. Peters & R. McMahon (Eds.), Preventing childhood disorders, substance abuse and delinquency (pp. 268 298). Thousand Oaks CA: Sage. Tucker, C., & Herman, K. (2002). Using culturally sensitive theories and research to meet the academic needs of low income African American children. American Psychologist, 53(2), 762 773. United States Department of Health and Human Services (1999). Mental Health: A Report of the Surgeon General Rockville, MD: Author Verhulst, F., & Koot, H. (1992). Child psychiatric epidemiology: Concepts, methods and findings Newbury Park, CA: Sage.

PAGE 113

102 Vitaro, F., Tremblay, R., & Bukowski, W. (2001). Frien ds, friendships, and conduct disorders. In J. Hill & B. Maughan (Eds.), Conduct disorders in childhood and adolescence (pp. 346 378). Cambridge: Cambridge University Press. Wehlage, G., & Rutter, R. (1986). Dropping out: How much do schools contribute to the problem? In G. Natriello (Ed.), School dropout patterns and policies (pp. 70 88). New York: Teachers College Press. Weissberg, R., & Greenberg, M. (1997). School and community competence enhancement and prevention programs. In I. Sigal & K. Renninger (Eds.), Handbook of child psychology: Child psychology in practice (5 ed., Vol. 5, pp. 1 125). New York: John Wiley & Sons. Weist, M. (1997). Expanded school mental health services: A national movement in progress. In T. Ollendick & R. Pr inz (Eds.), Advances in clinical child psychology (Vol. 19, pp. 319 352). New York: Plenum. Wentzel, K. (2002). Are effective teachers like good parents? Teaching styles and student adjustment in early adolescence. Child Development, 73 ( 1), 287 301. Went zel, K. R. (1997). Student motivation in middle school: The role of perceived pedagogical caring. Journal of Educational Psychology, 89(3), 411 419. Werner, E. (1995). Resilience in development. Current Directions in Psychological Science, 4 ( 3), 81 84. Y ung, B., & Hammond, W. (1997). Antisocial behavior in minority groups: Epidemiological and cultural perspectives. In D. Stoff, J. Breiling, & J. Maser (Eds.), Handbook of antisocial behavior New York: Wiley.

PAGE 114

103 Appendices

PAGE 115

104 Appendix A: Connection to Parents Scale 1. When I do things I shouldnt do, an adult usually corrects me. 2. My parents give me help and encouragement when I need it 3. My parents/guardians know the parents/guardians of my close friends 4. My parents are proud of me 5. When I grow up and have a family, I hope it will be similar to my own 6. I am an important member of my family

PAGE 116

105 Appendix B: Correlations Student Adjustment Survey subscales and Parent Scale in 5 th and 6 th grade .64 ** .33 ** .31 ** .43 ** .44 ** .18 ** -.46 ** -.43 ** -.37 ** -.35 ** .42 ** .42 ** .34 ** .36 ** -.33 ** .15 ** .13 ** .11 ** .09 ** -.12 ** .13 ** .11 ** .14 ** .10 ** .10 ** -.12 ** .14 ** .58 ** .10 ** .10 ** .27 ** .05 ** -.09 ** .13 ** -.06 ** -.03 .00 -.03 .04 -.01 .00 .01 .37 ** .45 ** -.04 -.09 ** -.10 ** -.06 ** -.06 ** .09 ** -.07 ** .43 ** .41 ** -.17 ** .47 ** -.01 -.04 ** .02 -.02 -.01 .00 .35 ** .35 ** -.03 .38 ** .35 ** SCH PER MOT NEG PAR 5TH-GRADE TEA SCH PER MOT NEG PAR 6TH-GRADE TEA SCH PER MOT NEG PAR 5TH-GRADE TEA SCH PER MOT NEG 6TH-GRADE Correlation is significant at the 0.01 level (2-tailed). **. Correlation is significant at the 0.05 level (2-tailed). *. Discipline Referrals and Suspensions Within Years Discipline Referrals and Suspensions 5 th Grade .54 ** .78 ** .28 ** .56 ** .34 ** .38 ** .47 ** .54 ** .38 ** .18 ** VIO CLA ISS OSS 5TH GRADE TOT VIO CLA ISS 5TH GRADE Correlation is significant at the 0.01 level (2-tailed). **.

PAGE 117

106 Appendix B: Correlations (Continued) Discipline Referrals and Suspensions 6 th Grade .58 ** .95 ** .46 ** .87 ** .51 ** .82 ** .76 ** .54 ** .74 ** .59 ** VIO CLA ISS OSS 6TH GRADE TOT VIO CLA ISS 6TH GRADE Correlation is significant at the 0.01 level (2-tailed). **. Discipline Referrals and Suspensions 7 th Grade .55 ** .95 ** .45 ** .85 ** .46 ** .80 ** .75 ** .55 ** .73 ** .57 ** VIO CLA ISS OSS 7TH GRADE TOT VIO CLA ISS 7TH GRADE Correlation is significant at the 0.01 level (2-tailed). **. Discipline R eferrals and Suspensions 8 th Grade .43 ** .93 ** .35 ** .85 ** .35 ** .80 ** .75 ** .52 ** .71 ** .56 ** VIO CLA ISS OSS 8TH GRADE TOT VIO CLA ISS 8TH GRADE Correlation is significant at the 0.01 level (2-tailed). **.

PAGE 118

107 Appendix B: Correlations (Continued) Discipline Referrals and Suspensions 9 th Grade .31 ** .85 ** .26 ** .82 ** .26 ** .75 ** .75 ** .42 ** .66 ** .56 ** VIO CLA ISS OSS 9TH GRADE TOT VIO CLA ISS 9TH GRADE Correlation is significant at the 0.01 level (2-tailed). **. Discipline Referrals and Suspensions 10 th Grade .23 ** .81 ** .20 ** .76 ** .18 ** .77 ** .72 ** .33 ** .62 ** .55 ** VIO CLA ISS OSS 10TH GRADE TOT VIO CLA ISS 10TH GRADE Correlation is significant at the 0.01 level (2-tailed). **. Discipline Referrals and Suspensions 11 th Grade .20 ** .79 ** .16 ** .71 ** .14 ** .72 ** .68 ** .33 ** .57 ** .49 ** VIO CLA ISS OSS 11TH GRADE TOT VIO CLA ISS 11TH GRADE Correlation is significant at the 0.01 level (2-tailed). **.

PAGE 119

108 Ap pendix B: Correlations (Continued) Discipline Referrals and Suspensions Across Years Total Referrals from 5 th through 11 th Grade .38 ** .34 ** .68 ** .31 ** .54 ** .68 ** .22 ** .41 ** .47 ** .61 ** .18 ** .36 ** .43 ** .53 ** .68 ** .15 ** .30 ** .32 ** .42 ** .46 ** .61 ** 6TH GRADE 7TH GRADE 8TH GRADE 9TH GRADE 10TH GRADE 11TH GRADE 5TH GRADE 6TH GRADE 7TH GRADE 8TH GRADE 9TH GRADE 10TH GRADE TOTAL REFERRALS Correlation is significant at the 0.01 level (2-tailed). **. Violence Referrals from 5 th through 11 th Grade .26 ** .15 ** .47 ** .13 ** .32 ** .38 ** .17 ** .18 ** .20 ** .21 ** .07 ** .17 ** .20 ** .17 ** .19 ** .05 ** .08 ** .16 ** .08 ** .10 ** .14 ** 6TH GRADE 7TH GRADE 8TH GRADE 9TH GRADE 10TH GRADE 11TH GRADE 5TH GRADE 6TH GRADE 7TH GRADE 8TH GRADE 9TH GRADE 10TH GRADE VIOLENCE REFERRALS Correlation is significant at the 0.01 level (2-tailed). **. Classroom Referrals from 5 th through 11 th Grade .29 ** .24 ** .64 ** .24 ** .53 ** .65 ** .18 ** .41 ** .46 ** .58 ** .12 ** .35 ** .37 ** .51 ** .60 ** .12 ** .30 ** .30 ** .43 ** .50 ** .62 ** 6TH GRADE 7TH GRADE 8TH GRADE 9TH GRADE 10TH GRADE 11TH GRADE 5TH GRADE 6TH GRADE 7TH GRADE 8TH GRADE 9TH GRADE 10TH GRADE CLASSROOM REFERRALS Correlation is significant at the 0.01 level (2-tailed). **.

PAGE 120

109 Append ix B: Correlations (Continued) In School Suspensions from 5 th through 11 th Grade .23 ** .16 ** .53 ** .15 ** .41 ** .56 ** .11 ** .32 ** .41 ** .49 ** .08 ** .23 ** .36 ** .43 ** .55 ** .11 ** .18 ** .26 ** .30 ** .40 ** .51 ** 6TH GRADE 7TH GRADE 8TH GRADE 9TH GRADE 10TH GRADE 11TH GRADE 5TH GRADE 6TH GRADE 7TH GRADE 8TH GRADE 9TH GRADE 10TH GRADE IN SCHOOL SUSPENSIONS Correlation is significant at the 0.01 level (2-tailed). **. Out of School Suspensions from 5 th through 11 th Grade .36 ** .30 ** .59 ** .26 ** .47 ** .59 ** .14 ** .32 ** .38 ** .51 ** .04 ** .26 ** .34 ** .42 ** .54 ** .12 ** .20 ** .28 ** .29 ** .33 ** .45 ** 6TH GRADE 7TH GRADE 8TH GRADE 9TH GRADE 10TH GRADE 11TH GRADE 5TH GRADE 6TH GRADE 7TH GRADE 8TH GRADE 9TH GRADE 10TH GRADE OUT OF SCHOOL SUSPENSIONS Correlation is significant at the 0.01 level (2-tailed). **.

PAGE 121

110 Appendix B: Correlations (Continued) 5 th Grade SAS and Parent Scales with Referrals and Suspen sions -.10 ** -.11 ** -.05 ** -.13 ** .13 ** -.08 ** -.07 ** -.09 ** -.04 ** -.09 ** .08 ** -.05 ** -.06 ** -.05 ** -.02 -.10 ** .08 ** -.04 ** -.11 ** -.09 ** -.04 ** -.09 ** .06 ** -.05 ** -.06 ** -.08 ** -.03 -.11 ** .07 ** -.06 ** -.16 ** -.16 ** -.09 ** -.20 ** .18 ** -.15 ** -.12 ** -.10 ** -.07 ** -.15 ** .15 ** -.11 ** -.14 ** -.14 ** -.08 ** -.17 ** .15 ** -.14 ** -.15 ** -.14 ** -.08 ** -.18 ** .16 ** -.14 ** -.10 ** -.09 ** -.07 ** -.14 ** .13 ** -.11 ** -.16 ** -.14 ** -.06 ** -.19 ** .18 ** -.15 ** -.13 ** -.11 ** -.07 ** -.17 ** .16 ** -.13 ** -.14 ** -.14 ** -.06 ** -.17 ** .15 ** -.14 ** -.16 ** -.14 ** -.05 ** -.19 ** .17 ** -.14 ** -.11 ** -.11 ** -.05 ** -.16 ** .13 ** -.13 ** -.15 ** -.15 ** -.06 ** -.17 ** .17 ** -.16 ** -.10 ** -.10 ** -.05 ** -.11 ** .11 ** -.10 ** -.13 ** -.13 ** -.06 ** -.16 ** .15 ** -.14 ** -.14 ** -.13 ** -.04 ** -.14 ** .15 ** -.12 ** -.11 ** -.11 ** -.06 ** -.14 ** .13 ** -.13 ** -.14 ** -.14 ** -.04 ** -.14 ** .16 ** -.13 ** -.06 ** -.05 ** -.02 -.07 ** .06 ** -.05 ** -.13 ** -.14 ** -.05 ** -.15 ** .16 ** -.12 ** -.14 ** -.15 ** -.05 ** -.14 ** .16 ** -.12 ** -.11 ** -.10 ** -.04 -.12 ** .11 ** -.11 ** -.14 ** -.12 ** -.03 -.13 ** .17 ** -.10 ** -.06 ** -.05 ** -.02 -.06 ** .08 ** -.05 ** -.12 ** -.12 ** -.04 ** -.14 ** .15 ** -.13 ** -.12 ** -.12 ** -.03 -.13 ** .14 ** -.12 ** -.13 ** -.10 ** -.04 -.14 ** .13 ** -.10 ** -.14 ** -.14 ** -.05 ** -.08 ** .14 ** -.08 ** -.06 ** -.02 -.02 -.05 ** .05 ** -.04 -.13 ** -.15 ** -.05 ** -.09 ** .15 ** -.10 ** -.11 ** -.12 ** -.04 -.07 ** .12 ** -.08 ** -.13 ** -.14 ** -.05 ** -.08 ** .11 ** -.08 ** TOT VIO CLA ISS OSS 5TH GRADE TOT VIO CLA ISS OSS 6TH GRADE TOT VIO CLA ISS OSS 7TH GRADE TOT VIO CLA ISS OSS 8TH GRADE TOT VIO CLA ISS OSS 9TH GRADE TOT VIO CLA ISS OSS 10TH GRADE TOT VIO CLA ISS OSS 11TH GRADE TEA SCH PER MOT NEG PAR 5TH GRADE SAS and PARENT SCALES Correlation is significant at the 0.01 level (2-tailed). **. Correlation is significant at the 0.05 level (2-tailed). *.

PAGE 122

111 Appendix B: Correlations (Continued) 6 th Grade SAS and Parent Scales with Referrals and Suspensions -.04 ** -.06 ** -.01 -.02 .02 -.05 ** -.05 ** -.07 ** .00 -.05 ** .02 -.04 ** -.02 -.03 -.02 .01 .01 -.01 -.02 -.03 .00 .00 .01 .00 -.03 -.03 -.02 .00 .02 -.04 -.11 ** -.13 ** -.03 -.04 ** .04 -.05 ** -.07 ** -.10 ** -.01 -.04 ** .02 -.05 ** -.10 ** -.12 ** -.02 -.04 ** .03 -.05 ** -.09 ** -.12 ** -.01 -.04 ** .03 -.05 ** -.07 ** -.08 ** -.03 -.02 .03 -.03 -.10 ** -.12 ** -.01 -.04 ** .04 ** -.03 -.06 ** -.07 ** -.04 ** -.05 ** .02 -.03 -.10 ** -.11 ** .00 -.05 ** .02 -.03 -.09 ** -.11 ** -.01 -.03 .04 ** -.04 -.08 ** -.10 ** -.01 -.05 ** .01 -.04 ** -.09 ** -.10 ** .01 -.04 ** .05 ** -.02 -.06 ** -.08 ** -.02 -.05 ** .04 ** -.06 ** -.09 ** -.10 ** .01 -.04 .05 ** -.02 -.09 ** -.09 ** .00 -.04 ** .03 -.03 -.07 ** -.08 ** -.01 -.05 ** .03 -.03 -.07 ** -.08 ** -.02 -.02 .06 ** .01 -.02 -.04 ** .01 -.01 .04 .00 -.08 ** -.09 ** -.04 ** -.03 .06 ** -.01 -.07 ** -.08 ** -.03 -.03 .06 ** .01 -.03 -.05 ** -.03 -.01 .06 ** .01 -.07 ** -.09 ** .01 .00 .05 ** .00 -.04 -.04 -.02 -.02 .00 -.02 -.08 ** -.08 ** -.01 -.02 .05 ** -.01 -.07 ** -.09 ** -.01 -.02 .04 ** -.02 -.05 ** -.07 ** .00 -.02 .04 -.03 -.06 ** -.07 ** .00 .02 .07 ** -.01 -.03 -.02 .01 .00 .02 .00 -.06 ** -.08 ** .00 .00 .05 ** -.01 -.07 ** -.07 ** -.01 .02 .03 -.02 -.06 ** -.08 ** .00 .01 .03 -.01 TOT VIO CLA ISS OSS 5TH GRADE TOT VIO CLA ISS OSS 6TH GRADE TOT VIO CLA ISS OSS 7TH GRADE TOT VIO CLA ISS OSS 8TH GRADE TOT VIO CLA ISS OSS 9TH GRADE TOT VIO CLA ISS OSS 10TH GRADE TOT VIO CLA ISS OSS 11TH GRADE TEA SCH PER MOT NEG PAR 6TH GRADE SAS and PARENT SCALES Correlation is significant at the 0.01 level (2-tailed). **. Correlation is significant at the 0.05 level (2-tailed). *.

PAGE 123

112 Appendix C: Factor Analysis with all loadings (significant loadings highlighted) STUDENT ADJUSTMENT SURVEY FACTOR ANALYSIS .65 .22 .14 .19 -.10 .65 .21 .06 .13 -.23 .64 .03 .15 .16 -.10 .53 .25 -.01 .29 -.11 .53 .20 .17 .12 .00 .51 .30 .02 .17 -.27 .48 .30 -.01 .00 -.15 .16 .67 .05 .18 -.11 .25 .66 .13 .01 -.04 .11 .59 .01 .42 -.17 .22 .57 .06 .24 -.22 .16 .54 .03 .32 -.21 .27 .51 .11 .02 -.07 .41 .42 .29 .06 -.07 .15 .36 .35 .02 -.01 .18 .08 .72 .11 .05 -.03 .07 -.64 .04 .11 .02 -.09 -.63 -.06 .23 -.09 -.06 -.56 -.12 .32 .19 .32 .46 .08 .04 -.04 -.08 -.44 -.12 .38 .11 .12 .05 .69 -.14 .09 .05 .17 .63 -.10 .18 .23 -.01 .54 -.16 .25 .04 -.12 .45 .21 .10 .14 .31 .36 .03 .08 .13 .18 .32 .06 -.13 .03 -.06 -.31 .56 .00 -.11 -.02 -.44 .54 -.39 -.10 -.17 -.04 .51 -.03 -.21 -.26 -.07 .51 -.14 -.12 -.02 .07 .44 -.21 -.10 -.13 .14 .39 ITEMS 10. Most teachers like my friends and me. 7. I think my teachers care about me. 9. My teachers often get to know me well. 11. I care about what most of my teachers think about me. 12. Some teachers would choose me as one of their favorite students. 25. I feel that I can go to my teachers for advice or help with schoolwork. 26. I feel that I can go to my teachers for advice and help with non-school work. 13. I like school. 17. I feel a sense of school spirit. 22. School is important to me. 21. I feel like I am learning a lot in school. 23. I believe I am learning important things in school. 20. Discipline is fair at this school. 15. I get a lot of encouragement at my school. 1. Students usually get along well with each other in this school. 5. Most students include me in their activities. 16. Other kids in my class have more friends than I do. 2. Making friends is difficult at this school. 6. I always seem to be left out of important activities. 4. A student can be him/herself and still be a part of this school. 3. I am in the wrong group to feel a part of this school. 32. Education is important for success in life. 34. I think I will go to college. 29. I try as hard as I can to do my best in school. 31. It bothers me when I don't do something well. 33. I feel prepared for middle school. 19. I have friends who are of different racial and ethnic backgrounds at this school. 27. Most of my teachers don't really expect good work from me. 28. I don't care how well I do in school. 14. My teachers don't pay much attention to me. 18. I don't feel safe at school. 24. I liked school more last year than I do this year. 8. Teachers are not usually available before class to talk with students. TEACHER SCHOOL PEER MOTIVAITON NEG EXP Component

PAGE 124

About the Author Ray Santa Lucia received a Bachelors D egree in Psychology from Vassar College in 1995. While at Vassar, he gained clinical experience working with children at Bellevue Hospital in New York City. During the summer following his junior year he collaborated on a published research study through t he Yale Child Study Center. Upon graduation, Vassar College awarded him with the Margaret Floy Washburn Fellowship for excellence in the field of psychology. While in the Ph.D. program in clinical psychology at the University of South Florida, Mr. Santa L ucias clinical, research, and teaching activities focused upon enhancing mental health outcomes for high risk children and families. He completed his clinical internship at Visalia Youth Services in Visalia, California. He has taught Child Devel opment at USF. He has engaged in longitudinal research focused upon understanding factors associated with risk and resilience among children throughout his graduate training.