|USFDC Home | USF Electronic Theses and Dissertations||| RSS|
This item is only available as the following downloads:
Auditors Performance in Computer-Mediated Fr aud Assessment Brainstorming Sessions: An Investigation of the Effects of Anonymity and Creativity Training by Antoinette L. Lynch A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy School of Accountancy College of Business Administration University of South Florida Major Professor: Uday Murthy, Ph.D. Stephanie Bryant, Ph.D. Rosann Collins, Ph.D. Gary Holstrum, Ph.D. Jacqueline Reck, Ph.D. Date of Approval: June 1, 2004 Keywords: creativity techniques, audit planning, group simulator; Jabri associa tive/bisociative scales, teams Copyright 2004, Antoinette L. Lynch
DEDICATIONS This dissertation is dedicated to my precious daughters, Jaleesa and Shauntia Lynch. Each day, you gave me the motivation to continue to put my best foot forward. Jaleesa, I have seen you eliminate distractions so that you can do your best. I admire you for your dedication to your family and to your education. Shauntia, I admire you for your courage to challenge situations with which you are not comfortable and to seize opportunities. I love you for your independence and your creativity. You both are stars in many ways. I love you dearly and thank you for making this journey one that we have all come to manage and love. To my personal angel and God sent mom, Betty Copling, you have always demonstrated to me that women are strong and that I can achieve anything as long as I put God first in my life. I thank you for the many days of prayer for me and my family. I thank you for calling me countless times a day to see how I was doing. To my dad, Lee Copling, you have always said that I never ask for much, but I always knew that whatever I needed, you would be right there supporting me. Thank you for providing comfort and security in my life. To my sisters, Maleka and Leenette Copling. Maleka, you are truly a wonderful woman who has never been afraid to speak her mind, and who's beautiful smile can lighten any day with joy. Thank you for your Saturday morning calls. I love you for that and so much more. L eenette, you are currently experiencing everything that I have experienced while in college. You are so much more and will achieve so much more. I love you for everything that you are. The bot h of you have always been supportive and have demonstrated that distance could never hinder th e love and support we have for each other. Vanita Nottingham, you have helped me acknow ledge the need to make some changes in my life. When I was too busy to spend time with my daughters, you were there reminding me that family comes first. You were very supportive in helping me manage my home and school. Being a single parent is difficult, and I am glad that we were graced by your wisdom and love.
ACKNOWLEDGEMENTS To all my financial supporters who helped me pay the bills and raise two lovely daughters, while managing a difficult school load, I appreciate you for eliminating financial burdens so that I could focus on school and my family. To my mentor, Dr. Tanya Benford, who listene d to my complaints, who fussed me out for going to bed early. Thank you for all those mome nts you kept it real and gave me pep talks and stayed up late hours with me. You are more than a mentor; you are a friend and a role model. If only my brain was half the size of yours. Thank you for being a part of my life. To all my professors at the University of South Florida who accepted nothing but the best, and who constantly reminded me of how proud th ey were of me. Thank you for teaching me how to be a professor, not only in research but also in teaching. Thank you Dr. Murthy for your undivided attention and guidance. Thank you fo r teaching me how to interpret and write programming language. You have a lot of patience. Thank you Dr. Reck for being my advocate in all situations and making sure everything was in order. Thank you Dr. Collins for allowing me to just walk into your office and have long-hour discussions on methodology issues. Thank you Dr. Bryant for advice on life in general and letting me know it is okay to just be myself and live life to its fullest. Thank you Dr. Holstrum for your b eautiful smile and your cheerful hello that told me you were proud of me. And last but not least thank you Dr. Engle for allowing me to walk into your office for advice from A to Z. Thank you for being both a friend and a mentor.
i TABLE OF CONTENTS List of Tables................................................................................................................. .................iv List of Figures................................................................................................................ .................vi List of Exhibits............................................................................................................... ................vii Chapter 1: Introduction........................................................................................................ ............1 1.1. Issues and the Need for Research......................................................................................1 1.2. Purpose and Research Questions.......................................................................................5 1.3. Motivation................................................................................................................ .........5 Chapter 2: Literature Review and Hypotheses................................................................................9 2.1 Introduction............................................................................................................... ........9 2.2 Sas No. 99 and the Role of Independent Auditors............................................................9 2.3 Fraud Risk Assessment...................................................................................................10 2.3.1 Fraudulent Financial Reporting................................................................................11 2.3.2 Misappropriation of Assets.......................................................................................12 2.4 Interaction Mode........................................................................................................... ..12 2.5 Paradigm-Modifying Creativity Technique....................................................................13 2.6 Brainstorming Effectiveness...........................................................................................15 2.7 The Effect of Interaction Mode On Brainstorming Effectiveness...................................16 2.7.1 Evaluation In General:..............................................................................................16 2.7.2 Gss and Evaluation Apprehension............................................................................18 2.8 The Effect of Paradigm-Modifying Creativity On Brainstorming Effectiveness............20 2.9 The Effect of Interaction Mode and Paradigm-Modifying Creativity Technique On Brainstorming Effectiveness...........................................................................................23 Chapter 3: Method.............................................................................................................. ...........26 3.1 Introduction............................................................................................................... ......26 3.2 Research Design............................................................................................................ ..26 3.3 Task....................................................................................................................... ..........27 3.4 Participants............................................................................................................... .......28 3.4.1 Audit Interns............................................................................................................ .28 3.4.2 Internal Auditors.......................................................................................................2 9 3.5 Pilo t Study................................................................................................................ .......29 3.6 Covariates................................................................................................................. .......30 3.6.1 Intrinsic and Extrinsic Motivation............................................................................30 3.6.2 Creative Person.........................................................................................................3 1
ii 3.7 Experiment Materi als and Procedures.............................................................................33 3.8 Treatments/Independent Variables..................................................................................37 3.8.1 Interaction Mode Treatment.....................................................................................37 3.8.2 Paradigm-Modifying Creative Technique Training Treatment................................38 3.9 GSS Technology............................................................................................................. 39 3.10 Dependent Variable and Data Collection........................................................................41 Chapter 4: Results............................................................................................................. .............45 4.1 Descriptiv e Statistics..................................................................................................... ..45 4.2 Correlati on Matrices....................................................................................................... .47 4.3 Effectiveness of Training................................................................................................51 4.3.1 Tea Quantity............................................................................................................. 52 4.3.2 Tea Novelty.............................................................................................................. 53 4.3.3 Audit Interns and Tea Task......................................................................................54 188.8.131.52 Multivariate Normal Distribution Assumption.........................................................54 184.108.40.206 Equal Variance-Covariance Assumption..................................................................55 4.3.4 Tea Task Results for Audit Interns...........................................................................56 220.127.116.11 Multivariate Normal Distribution Assumption.........................................................56 18.104.22.168 Equal Variance-Covariance Assumption..................................................................57 22.214.171.124 Tea Task Results for Internal Auditors.....................................................................57 4.3.5 Summary of Training Effectiveness..............................................................58 4.4 Fraud Quantity............................................................................................................. ...58 4.5 Fraud Novelty.............................................................................................................. ....59 4.6 Fraud U sefulness........................................................................................................... ..60 4.7 Effect of Covariates and Other Measured Variables.......................................................61 4.7.1 Evaluation Apprehension:........................................................................................62 4.7.2 Social Presence.........................................................................................................6 5 4.7.3 Task Complexity......................................................................................................67 4.7.4 Intrinsic Motivation and Extrinsic Motivation.........................................................69 4.7.5 Creative Person.........................................................................................................7 0 4.8 Manipulation Checks......................................................................................................71 4.8.1 Interaction Mode......................................................................................................71 4.8.2 Paradigm-Modifying Creativity Training.................................................................72 4.9 Test of Hypotheses H1 Though H3.................................................................................72 4.9.1 Power Analysis.........................................................................................................72 4.9.2 Audit Interns and Assumptions of Manova..............................................................74 126.96.36.199 Nature of Distribution...............................................................................................74 188.8.131.52 Equality of Variance-Covariance Matrices..............................................................75 184.108.40.206 Fraud Task Test Results for Audit Interns................................................................75 4.9.3 Internal Auditors and Assumptions of Manova........................................................79 220.127.116.11 Nature of the Distribution.........................................................................................79 18.104.22.168 Equality of Variance-Covariance Matrices..............................................................80 22.214.171.124 Fraud Task Test Results for Internal Auditors.........................................................80 4.10 Additional Analysis....................................................................................................... ..81 4.10.1 Manipulation Check Questions Revisited................................................................81 4.11 Post Hoc Analysis......................................................................................................... ..86
iii Chapter 5: Summary............................................................................................................. .........91 5.1 Discussion of the Results.................................................................................................. ...91 5.2 Contributions.............................................................................................................. .........93 5.3 Limitations................................................................................................................ ...........94 5.4 Future Resear ch............................................................................................................ .......97 Refere nces..................................................................................................................... .................99 Exhibits....................................................................................................................... .................109 Appendix A: Research Materialssection 1 Consent Form and Log On Screen........................114 Appendix B Instructions To Raters...........................................................................................1 39 Appendix C Information On Coders and Raters........................................................................142 About the Author............................................................................................................... .End Page
iv LIST OF TABLES Table 1 Research Design Layout............................................................................................... ..27 Table 2 Procedures for Participants.......................................................................................... ...35 Table 3 Participant Demographics............................................................................................. .46 Table 4 Number of Participants in Each Treatment Condition for the Fraud Task.....................47 Table 5 Correlation Matrix for Audit Interns..............................................................................49 Table 6 Correlation Matrix for Internal Auditors........................................................................50 Table 7 Tea Quantity D escriptive Statistics................................................................................52 Table 8 Tea Novelty D escriptive Statistics.................................................................................54 Table 9 Fraud Quantity Descriptive Statistics.............................................................................58 Table 10 Fraud Novelty Descriptive Statistics............................................................................59 Table 11 Fraud Usefulness Descriptive Statistics.......................................................................61 Table 12 Cronbachs Alpha of Measured Items..........................................................................62 Table 13 Evaluation Apprehension Descri ptive Statistics Audit Interns..................................64 Table 14 Evaluation Apprehension Descrip tive Statistics Internal Auditors...........................65 Table 15 Social Presence Descriptive Statistics Audit Interns.................................................66 Table 16 Social Presence Descriptiv e Statistics Internal Auditors...........................................67 Table 17 Mean Task Complexity Descriptive Statistics Audit Interns.....................................68 Table 18 Mean Task Complexity Descriptive Statistics Internal Auditors..............................68 Table 19 Intrinsic Motivation Descriptive Statistics Audit Interns..........................................69 Table 20 Intrinsic Motivation Descriptive Statistics Internal Auditors....................................70
v Table 21 Cronbachs Alpha of Measured Items Problem-Solving Scale.................................70 Table 22 Analysis of Covariance for Fraud Quan tity, Fraud Novelty, and Fraud Usefulness for Audit Interns.................................................................................................................. ........76 Table 23 Summary of Findings................................................................................................. ..81 Table 24 ANCOVA for Fraud Quantity, Fraud Novelty, and Fraud Usefulness........................83 Table 25 ANCOVA for Fraud Quantity, Fr aud Novelty, and Fraud Usefulness........................84 Table 26 Summary of Findings Reported Alpha Level.............................................................85 Table 27 ANCOVA for Fraud Quantity, Fr aud Novelty, and Fraud Usefulness........................88 Table 28 ANCOVA for Fraud Quantity, Fr aud Novelty, and Fraud Usefulness........................89
vi LIST OF FIGURES Figure 1. The Four-Ps Model of Creativity....................................................................................14 Figure 2: Research Model....................................................................................................... .......16 Figure 3. An Insight Model of Creativity......................................................................................2 1 Figure 4. Fraud Quantity Plot Main Effect of Interaction Mode Audit Interns........................77 Figure 5. Fraud Novelty Plot Main Effect of Interaction Mode Audit Interns.........................77 Figure 6. Fraud Novelty Plot Main Effect of Creativity Training Audit Interns......................78 Figure 7. Fraud Usefulness Plot Main Effect of Creativity Traini ng Audit Interns.................78
vii LIST OF EXHIBITS Exhibit 1: Four Phases of an Independent Audit.......................................................................1110 Exhibit 2: Summary of Hypotheses and Research Questions......................................................111 Exhibit 3: Characteristics of Adaptors and Innovators................................................................113
viii AUDITORS PERFORMANCE IN COMPUTER-MEDIATED FRAUD ASSESSMENT BRAINSTORMING SESSI ONS: AN INVESTIGATION OF THE EFFECTS OF ANONYMITY A ND CREATIVITY TRAINING Antoinette L. Lynch ABSTRACT In the wake of recent corporate accounting scandals, auditors are encouraged to improve their method of fraud detection. Although Stat ement on Auditing Standards (SAS) No. 99 does not change the responsibility of the auditor for detecting fraud, it does provide new procedural requirements for assessing fraud risk, such as brainstorming among key team members about the potential for fraud. Using audit interns and internal auditors, this study empirically examines two interventions hypothesized to improve the quality of ideas generated by audit interns and internal auditors. In the first intervention, auditors use a computer-based group support system to brainstorm either non-anonymously or anonymous ly. For the second intervention, auditors were either trained to use a paradigm-modifying creativity technique or not trained. Additionally, it is hypothesized that the creativity training will have the greatest impact on brainstorming effectiveness when auditors brainstorm anonym ously. However, the results suggest that audit interns working non-anonymously generated the greatest number of fraud ideas and also the greatest number of original ideas. Audit intern s who received training on a paradigm-modifying creativity training technique ge nerated the greatest number of unique ideas and received, on average, the highest usefuln ess to the audit process score.
1 CHAPTER 1: INTRODUCTION 1.1. Issues and the Need for Research There has been considerable public criticism of the attest function performed by auditors of publicly held corporations (Hilzenrath 2002; Johnson and Masters 2003; Pulliam et al. 2003; Thornburgh 2004; Wyatt; Zeff 2003). When performi ng external audits, auditors are responsible for providing reasonable assurance that a compan ys financial statements are free of material fraud and errors. In 1997, in an effort to addr ess concerns of both the profession and the public, the AICPAs Auditing Standards Board (ASB) i ssued Statement on Auditing Standards (SAS) No. 82: Consideration of Fraud in a Financial Statement Audit, which was designed to assist auditors in fraud detection. Relying on academ ic research, and recommendations from the Panel on Audit Effectiveness, the ASBs Fraud Task Force, and various stakeholders, the ASB concluded that SAS No. 82 fell short of its inte nded goal of enhancing auditors performance in considering material fraud in financial statements In an effort to address perceived deficiencies of SAS No. 82, the ASB issued SAS No. 99: Cons ideration of Fraud in a Financial Statement Audit, in 2002 (AICPA 2002a). One of the requirements of SAS No. 99 is that the auditors consideration of fraud must involve the exchange of ideas or brainstormi ng among the audit team members, including the auditor with final responsibility for the audit, about how and where they believe the entitys financial statements might be susceptible to mate rial misstatement due to fraud, how management could perpetrate and conceal fraudulent financial reporting, and how assets of the entity could be misappropriated (AICPA 2002b, paragraph 14). However, SAS No. 99 provides limited guidance on who should attend the brainstorming session, indicating that key members of the audit team should participate, making no reference as to whether staff auditors should be included
2 or excluded. Importantly, SAS No. 99 does not provide any guidance regarding effective brainstorming methods. Prior to the issuance of SAS No. 99, the asse ssment of fraud risk was often performed by one of the key audit personnel, utilizing practice aids such as check-off sheets and expert systems (Hirst et al. 1996; Shelton et al. 2001; Solom on 1987). Depending on the size of the engagement, key personnel on the audit team would include, at a minimum, one or more on-site supervisory auditors (senior auditors), a manager, and the pa rtner in charge of the engagement (Rich et al. 1997). Although SAS No. 99 is silent on the possibil ity of including staff auditors, often it is the staff auditor who first encounters potential audit problems and interacts with employees who may be attempting to conceal fraud (Rabinowitz 1996). St aff auditors are the eyes and ears of the audit team and represent the audit firms frontline pe rsonnel. Staff auditors obtain audit evidence, and based on that evidence, reach conclusions th at are subsequently evaluated by supervisory team members. According to Ashton and Kennedy (2002, p. 221), judgments of staff auditors often determine the type and extent of docu mentation in audit work papers and serve as preliminary inputs for senior auditors judgments and choices. Thus, it can be argued that participation in the brainstorming session by staff auditors could sensitize them to the possibility of fraud as they gather audit evidence. The presence of superiors or more experienced auditors could impact a staff auditors ability to effectively brainstorm about possible fra udulent misstatements that materially affect the entitys financial statements. A drive theory of social facilitation (Zajonc 1965) and prior research in psychology suggests that, under certain conditions the mere presence of superiors inhibits the productivity of junior members in a brainstormi ng session (Cottrell et al. 1968; Zajonc 1965). For instance, working with senior team members may co nvey to the more junior members of the team that they are accountable for their ideas, or that their ideas must meet with the approval of the senior members (Agarwal 2000). This phenomenon, referred to as evaluation apprehension, could
3 inhibit the ability of staff auditors to provid e candid (and possibly valuable) input to the fraud brainstorming session. The purpose of this research is to explore whether two interventions improve the effectiveness of ideas generated by auditors involved in fraud brainstorming sessions mandated by SAS No. 99. Specifically, this dissertation examines the question: How does interaction mode and creativity training impact idea generation of staff auditors in a fraud brainstorming session? Nagasundaram and Bostrom (1995) suggest that organizations empower employees through nonhierarchical teams in order to tap in to the creative ideas of the entire workforce. Although hierarchical audit teams represen t a long-established aspect of the auditing environment, which is unlikely to change, group support systems (GSS), deployed by most large auditing firms (e.g., Lotus Notes1), provide an opportunity to simulate a nonhierarchical setting for the purpose of brainstorming. GSS facilitat es the communication between team members who may be located in the same or different loca tions, and who may interact synchronously or asynchronously (Bamber et al. 1998; Bamber et al. 1996; Pinsonneault et al. 1989). Additionally, GSS has many features such as anonymity, parallel communication, e-mail, and group memory that maximize positive group processes, such as allowing more information to be communicated among group members, and minimizes negative group processes such as information overload (Bamber et al. 1998; Bamber et al. 1996; Pins onneault et al. 1989). Group support systems permit anonymous interaction brainstorming sessions, by masking the identities of team members for the duration of the session. Since knowledge of the identities of individual audit team members could cause inhibition during the brainstorming session, enabling anonymous contributions to the session should free staff auditors to provide their candid ideas without fear of senior disapproval (Pinsonneault et al. 1998). 1 Accounting firms are using collaborative software, su ch as Lotus Notes to facilitate knowledge sharing
4 The first intervention investigated in this research is the interaction mode in the brainstorming session, specifically whether staff a uditors brainstorming performance is superior when the interaction mode is anonymous rather than non-anonymous. If anonymous interaction in the brainstorming session is shown to result in more effective ideas, the findings would lend support to the use of a GSS that permits anon ymous interaction for SAS No. 99 brainstorming sessions. An added benefit of using GSS is that these technologies permit team members to interact regardless of their physical location. In todays global environm ent, audit teams may be geographically dispersed, especially on audits of large multinational corporations. Thus, there may be occasions when it is not feasible or cost-effective for key engagement personnel to brainstorm at the same time and in the same location. Auditors must be creative and unpredictabl e in their fraud detection methodologies. For instance, auditors rarely ask for unlimited access to clients records, but instead rely on clients to provide requested documents. The assessment of known fraud cases by the National Commission on Fraudulent Financial Reporting (FCFFR) (popularly known as the Treadway Commission) found that creative revenue rec ognition methods were adopted by high-tech companies. For example, high-tech companies in flated earnings using creative methods known as sham sales and conditional sales (Beasley et al. 2000). These creative methods suggest that fraud perpetrators are familiar with standard aud it procedures and go out of their way to avoid detection. Therefore, auditors need to think outside the box, or to think creatively about how fraud perpetrators can conceal fraud. There is considerable evidence in the litera ture that creativity training techniques can enhance the degree of creativity of an individuals output. Thus, the second intervention investigated in the research is whether the use of a creativity training technique results in the generation of more innovative ideas during SAS No. 99 fraud brainstorming sessions. If proven effective, such creativity training techniques re present a relatively low-cost intervention that
5 auditing firms can employ in order to improve the effectiveness of the fraud brainstorming sessions mandated by SAS No. 99. In other areas, an increasing number of companies are using electronic communication media to solicit innovative ideas from employees. Companies are forming task teams that use brainstorming techniques to generate ideas for new business initiatives. For example, of the five top busi ness initiatives at Royal Dutch/Shell Group, four initiatives came as a result of analyzing ideas ge nerated by employees. Proctor and Gamble has 33 new initiatives that came as a result of a brai nstorming task force (Stepanek 1999). Creativity can be used to look for new ways to solve ol d problems and to solve complex problems (Amabile 1996). 1.2. Purpose and Research Questions The purpose of this study is to investigate how interaction mode, when using a group support system and training on a paradigm-modify ing creativity technique, can impact staff auditors ability to generate innovative ideas in fraud brainstorming, sessions mandated by SAS No. 99. The research questions are: (1) Does interaction mode using a GSS affect the quantity, utility, and novelty/rarity of ideas generated by staff auditors? (2) Does training in a paradigm-modifying creativity technique improve the quantity, utility, and novelty/rarity of ideas generated by staff auditors? (3) Do interaction mode and creativity training jointly affect the quantity, utility, and novelty/rarity of ideas generated by staff auditors? 1.3. Motivation Fraud prevention is a high priority in th e accounting profession, and to the country in general, as evidenced by President Bushs discussion about fraud in his 2002 State of the Union address, the Sarbanes-Oxley Act of 2002, and the recent release of SAS No. 99 (Bamber 2002; 1998; Whittington 2002). The importance of fraud ri sk assessment cannot be over emphasized. It
6 is one of the few tasks that when mishandled can jeopardize an auditors career, the success of the accounting firm, and the reputation of the a udit profession (Palmrose 1987). Fraud-related conclusions reached during the initial planning task will impact decisions about the next stage of the audit, the field work. Examples of decisions a ffect include the level of expertise needed for the audit, and the timing and extent of audit test s (Anderson 1977). It is important to note that field work is conducted primarily by staff auditors, underscoring the importance of the need for them to be involved in planning stage fraud brainstorming sessions as required by SAS No. 99. The GSS literature on anonymity reveals a dive rsity of opinions on whether anonymity is an important feature for electronic brainstorming. The notion that GSS-anonymity is useful is supported by the research of Connolly et al. (1990) and Sosik et al. (1999). Other research have been unsupportive of GSS-anonymity (Jessup et al. 1991; Valacich et al. 1992). Cooper et al. (1998) suggest that one reason for the mixed results is that GSS research on anonymity tends to have low statistical power caused by small sample size, where many studies have only five to 12 groups per treatment. Pinsonneault and Heppel ( 1998) argue that the mixed results in prior research on the impact of anonymity on idea generation are caused by a weakness in the manipulation of evaluation apprehension. Laborator y environments using student subjects fail to simulate corporate America, where power and job status are salient. The authors provide a compelling need for future anonymity research in a direction that considers situational variables, such as hierarchical structures, computer-based communications, and the use of actual employees (Pinsonneault et al. 1998). Although several researchers have called attention to the need for an empirical examination of the impact of the hierarchical audit team structure on performance (Bamber 2002; Bamber et al. 1996; Murthy 2002; Solomon 1987), r esearch in this area is limited. Jamal and Tan (2001) created three member teams by pairing an audit manager with a top senior and a mediocre senior. The authors main goal was to determin e if members of the team could predict the
7 preferences of other team members evaluation of a high/low ambiguity task. However, the researchers did not explore how ha ving multi-level team participants (more than 2 hierarchical layers) affected judgments. Solomon (1982) co mpared the specification of prior probability distributions (PPDs) by audit teams to the specification of PPDs by individual auditors. Staff, senior, and management auditors were randomly assigned to an individual, nominal/interacting group, or interacting/nominal group treatment. Participants assigned to the group treatments worked as a three-person team, consisting of a sta ff auditor, senior auditor, and a manager; two staff auditors and a senior; or two seniors and a manager. Solomons (1982) focus was primarily on the performance differences between groups and individuals rather than on how alternative team compositions affected the behavior of i ndividual team members. Johnson (1994) also used a three-person team composition in a memory task involving audit work paper reviews. Unlike Solomon (1982), Johnson (1994) did not set out to ensure teams consisted of multiple levels of expertise, but instead randomly assigned staff a uditors, seniors, and managers to conditions. Thus, prior research has shed little light on the impact of hierarchical team composition on idea generation during fraud brainstorming sessions. The complexity of fraud assessment has increased commensurate with the level of creativity and innovation in the commission of fra udulent activities. Fraud perpetrators have employed unique methods that may not be cons idered during traditional (non-brainstorming) fraud risk assessments. For example, an investment advisor, who failed to register with the SEC, used online chat rooms to increase stock prices. This was stock held in the personal investment portfolio of the advisor. However, instead of leav ing a paper trail for auditors or being restricted to the companys internal system for communicati ng to clients, the advisor relied on chat room sessions to commit fraud (Danner 2000). A former executive of Symbol Technologies was accused of committing securities fraud by persuading distributors to purchase scanners that the distributors did not need. In return, Symbol promised distributors that any unsold scanners would
8 be re-purchased. This practice is known as channe l stuffing. The executives illegal and clever method allowed Symbol to inflate reported sales (Berenson 2003). When considering the potential for fraud, in orde r to be effective, auditors must think just as creatively and unconventionally as fraud perpetra tors. Thus, training auditors in a creativity techniqueone that allows them to expand thei r boundaries to look at the situation from a different anglecould improve auditors capab ility to detect fraud and could significantly improve the professions fraud detection success ra te. Creativity training techniques, specifically paradigm-modifying2 techniques have been shown to be effective in the information systems literature (Garfield et al. 2001; Hender et al. 2002; Satzinger et al. 1999), and should, therefore, improve auditors brainstorming effectiveness. The remainder of this dissertation is organized as follows: Section II provides a review of prior literature and develops a research fram ework and the hypotheses. Section III presents the research design and methodology. Section IV presents the results of the study. This dissertation concludes with Section V, a discussion of r esults, contributions, limitations and potential implications of the findings. 2Paradigm-modifying techniques are those tec hniques that tend to generate ideas that are revolutionaryideas that redefine the problem and the belief system of the existing paradigm (Garfield et al. 2001).
9 CHAPTER 2: LITERATURE RE VIEW AND HYPOTHESES 2.1 Introduction The literature review for this dissertation pr ovides an overview of the factors and events leading to the changes in fraud related proce dures dictated by SAS No. 99 and the relevant theoretical constructs (interaction mode, pa radigm-modifying creativity training, and brainstorming effectiveness). Additionally, it sy nthesizes relevant prior research in auditing, management information systems (MIS), and ps ychology relevant to the research model and hypotheses proposed in this study. Extant audit lite rature examined in this dissertation focuses on fraud risk assessment in various contexts, and the impact of interaction mode. The MIS literature reviewed in this section includes studies that examine the impact of anonymity, computermediated communications and paradigm-modifyin g techniques on brainstorming effectiveness (the quantity, utility, an d novelty of the ideas generated in brainstorming sessions). The relevant psychology literature also includes research on f actors that impact brainstorming effectiveness. These factors include social facilitation and evaluation apprehension. 2.2 SAS No. 99 and the Role of Independent Auditors External auditors are responsible for pr oviding reasonable assurance that financial statements are prepared in accordance with generally accepted accounting principles (GAAP). Along with company management and directors, auditors are responsible for the integrity of the companys financial reporting (AICPA 2002b). The audit process comprises four phases, as shown in Exhibit 1. Although SAS No. 99 states th at brainstorming can be conducted throughout the audit process (AICPA 2002b), the standard requi res that brainstorming be conducted during Phase I of the audit process. Phase I is the audit planning phase where auditors gather information about the business, such as information regardi ng the entitys industry and its competitors. During
10 the final phase, Phase IV, auditors issue an audit report that includes an opinion on the financial statements. Users of financial reports (i.e., stockholders, the government, etc.) rely on the auditors opinion as to whether the financial stat ements, prepared by management, are free of material misstatements due to errors (uni ntentional misstatements) and fraud (intentional misstatements). According to SAS No. 47: Audit Risk and Materiality in Conducting an Audit (AICPA 1983), auditors have the same responsibility for fraud detection as for error detection. In order to provide reasonable assurance that material fraud does not exist when conducting a financial statement audit, the aud itor is required to comply with SAS No. 99, which was issued in October 2002 by the Auditing Standards Board (ASB). SAS No. 99 resulted from a long history of the auditing professions effort to clarify the auditors role in fraud detection, and it supe rseded SAS No. 82 (AICPA 2002a; Nieschwietz et al. 2000). One of the new requirements of SAS No. 99 is for audit team members to exchange ideas about ways an entitys financial statements may be materially misstated due to fraud associated with fraudulent financial reporting, and fraud associ ated with misappropriation of assets. For the first time, auditors are required to brainstorm ; however, SAS No. 99 provides little guidance as to how to conduct the brainstorming session, indicating only that key members of the audit team should participate in the session. 2.3 Fraud Risk Assessment Extant literature has directed our attention to ward auditors inability to detect fraud or properly analyze fraud-risk factors (Bell et al. 2000; Erickson et al. 2000; Hackenbrack 1992; Nieschwietz et al. 2000; Palmrose 1987; Pincus 1989) For example, Pincus (1989) examined the use of red flag indicators as a method for examining audit fraud risk. Pincus (1989) research was motivated by the increased use of red flag i ndicators as a method for assessing fraud risk. Using in-charge auditors from a large CPA firm, Pincus (1989) assigned auditors to either a fraud or no fraud case, and to either the use of a red flag indicator questionnaire or no questionnaire.
11 Auditors responses were measured on comp rehensiveness, uniformity, and fraud risk assessment. This study found that although questi onnaire users considered a more comprehensive set of fraud indicators and exhibited a high degree of uniformity, the participants who did not rely on a questionnaire performed better at assessing fra ud risk than those participants who used a questionnaire. The use of only a red flag qu estionnaire to assess fraud risk may have limited auditors thinking to a restricted set of risks, discouraging them from thinking beyond the information presented to them. When exchanging ideas or brainstorming, SAS No. 99 requires the audit team to consider two types of fraud: fraudulent financial repor ting and misappropriation of assets (AICPA 2002). In order to comply with SAS No. 99, the team must exchange ideas about how management could perpetrate and conceal fraudulent financ ial reporting (AICPA 2002, paragraph 6). For misappropriation of assets, the audit team must exchange ideas about how assets of the entity could be misappropriated (AICPA 2002, pa ragraph 6). The ideas generated during the brainstorming sessions are used by auditors to assess the risk of material misstatements due to fraud. 2.3.1 Fraudulent Financial Reporting The National Commission on Fraudulent Fina ncial Reporting defines fraudulent financial reporting as intentional or reckless conduct, whether by act or omission, that results in materially misleading financial statements (NCFFR 1987, p. 8). This can be due to a failure to disclose significant information, overstating earnings, inflating assets, or inappropriate accounting procedures (Beasley and Salterio 2001, Dechow et al. 1996). SAS No. 99 states that fraudulent financial reporting may be accomplished by: Manipulation, falsification, or alteratio n of accounting records or supporting documents from which financial statements are prepared;
12 Misrepresentation in or intentional omission from the financial statements of events, transactions, or other significant information; Intentional misapplication of accounting principles relating to amounts, classification, manner of presentation, or disclos ure (AICPA 2002, paragraph 6). 2.3.2 Misappropriation of Assets Misappropriation of assets occurs when one or a group of individuals commit fraud for financial gain (Romney and St einbart 2002). SAS No. 99 states that misappropriation of assets may be accomplished by larceny or skimming of assets (i.e., cash, inventory, receivables) or fraudulent disbursements. Fraudulent disbursements include billing schemes, payroll schemes, expense reimbursement schemes, and check tampering. 2.4 Interaction Mode Interaction mode is how teams interact/communicate. Teams are typically described as consisting of individuals with di stributed knowledge with one team leader who is responsible for making final team decisions (Hedlund et al. 1998; Taggar et al. 1999; Phillips 2001; Phillips 2002). Solomon (1987) describes coacting teams as those whose members work concurrently to solve a problem or to perform a task, but implies that coacting teams can consist of members with various job titles or levels of power. The focus of this study is on hierarchical audit teams where power is distributed, with both novices and mo re expert auditors on the audit team. Face-to-face, GSS-anonymous, and GSS-non-anonymous are the three ways in which interaction mode can be operationalized (M urthy 2002). Without anonymity, individuals, particularly low status participants, may withhol d ideas due to negative evaluation or may feel pressured to conform to the group majority or sen ior participants views (Dennis et al. 2001, p. 169). Prior research has defined anonymity as a multidimensional concept, arguing that lack of identification is one of several elements needed to operationalize the degree team members feel
13 liberated from being evaluated (Nunamaker et al. 1991a; Pinsonneault and Heppel 1998). For example, in addition to lack of identification, individual team members need to feel secure in their proximal distance from other team members (i.e., team members in the next cubical versus team members geographically dispersed). In this st udy, the levels of interaction mode are defined as GSS-anonymous (team members know the composition of their team, but are unaware of the author of each comment) and GSS-non-anonym ous (team members know the composition of their team, and are aware of the author of each comment)3. 2.5 Paradigm-Modifying Creativity Technique Creativity is a complex, dynamic phenomenon in that it is comprised of four interactive components: the creative product, creative process, creative person, and creative environment (see Figure 1) (Rhodes 1961; Rothenberg and Hausman 1976; Couger 1995). Each component can be described independently, but must inter act to operate functionally (Rhodes 1961; Fellers and Bostrom 1993). For example, the creative envi ronment can be one that is constructive or destructive to creativity (Rhodes 1961). The crea tive person component, which encompasses an individuals innate creativity, is treated as a cova riate and discussed under Section 3.5, while the creative process (paradigm-modifying creativity t echnique training) and environment (interaction mode) are manipulated, as explained below. Finall y, the creative product is the outcome variable and is discussed in Section 2.6. 3 The definitions of GSS-anonymous and GSS-non-anonym ous are similar to the definitions used by Karan et al. (1996).
14 FIGURE 1. THE FOUR-PS MODEL OF CREATIVITY The creative process component is how the cr eative product (ideas generated) comes into being. It is the thought process of the individual while creating ideas (Amabile 1983). Koester (1964) described the process as the displacement of attention to something not previously noted, which was irrelevant in the old and is relevant in the new context; the discovery of hidden analogies as a result (Koestler 1964, p. 119). Over 20 creativity techniques are available that influence an individuals thought process (Van Gundy 1988; Couger 1996). Most techniques fall into two categories, analytical or intuitive (VanGundy 1988). Analytical techniques are paradigmpreserving. Paradigm-preserving ideas support or extend the existing paradigm; they are evolutionary in that they adapt elements of the existing paradigm (Garfield 2001, p. 323). An example of an analytical technique that is para digm-preserving is force field analysis. Individuals using the force field analysis technique generate id eas that are stimulated by what is perceived as being weaknesses and strengths of a problem, thus preserving thought patterns similar to those used in traditional problem-solving methods (Couger 1996). However, prior research suggests that traditional problem-solving methods have not been effective for fraud risk assessment (Palmrose 1987; Pincus 1989; Hackenbrack 1992; Be ll and Carcello 2000; Erickson et al. 2000; Nieschwietz et al. 2000). Process Product Person Ado p ted from Cou g er ( 1995 )
15 Intuitive techniques use either unrelated (i.e., guided fantasy) or related (i.e., brainstorming) stimuli. Intuitive techniques that rely on unrelated stimuli are more likely to produce novel, paradigm-modifying ideas than tech niques that rely on related ideas. Paradigmmodifying ideas are revolutionary in that they redefine the problem or its elements (Garfield et al. 2001, p. 323). Although there are additional intuitive techniques such as analogies, wishful thinking, and wildest idea, brainstorming is the most common intuitive technique applied in research studies (Satzinger et al. 1999; Garfield et al. 2001; Hender et al. 2002). A thorough literature review revealed only three studies th at specifically examined different creativity techniques (Satzinger et al. 1999; Garfield et al 2001; Hender et al. 2002). These studies indicate that intuitive techniques that use unrelat ed stimuli lead to more novel ideas. 2.6 Brainstorming Effectiveness As previously stated, Rothenberg and Haus man (1976) and Rhodes (1961) describe several components of creativity, one of which is the creative product (Rhodes 1961; Rothenberg and Hausman 1976; Couger 1995). Effective brainsto rming is the generation of ideas that are considered useful, novel, and appropriate (Amabile 1983; Eisenberger et al. 1999; Garfield et al. 2001). In this study, the creative product consis ts of the ideas generated during brainstorming sessions. One of the purposes of brainstorming is to allow the organization to get input from all members of a team, rather than just from the more vocal members of the team. What is produced or observable from this effort is the product. One measure of brainstorming session effectiveness is the number of ideas generated by each particip ant. The utility of ideas is a measure of how useful the idea is for the audit planning process. N ovelty is a score of rarity or uniqueness; ideas mentioned by fewer participants are more novel that those mentioned by more participants. The research model is shown in Figure 2, and considers the many dimensions of creativity. Interaction mode represents the environment, crea tivity training is the process, and brainstorming
16 effectiveness, which is predicted to be a func tion of interaction mode, creativity training, and their interaction, is the product. FIGURE 2: RESEARCH MODEL 2.7 The Effect of Interaction Mode on Brainstorming Effectiveness (link 1) 2.7.1 Evaluation in General: Prior research findings in psychology on the impact of expected evaluation on creativity are mixed. Several theorists have maintained that external evaluation must be minimized in order to foster creativity (Osborn 1963; White and Owen 1970). Osborn (1963) maintains that when the environment is playful and nonjudgmental, indi viduals are comfortable suggesting ideas to a team. Similarly, Bartis et al. (1988), using a br ainstorming technique, found that creativity was greater for those participants not being evaluate d than for those participants who were in the experimenter-evaluation condition. Conversel y, Gagne and Zuckerman (1999) found that participants performing a brainstorming task work ed harder when co-participants, as well as the experimenter, could evaluate performance. Specifi cally, as the evaluation potential increased, so did performance. Creativity Training ParadigmModifying Creativity Training No Training Interaction Mode Anonymous Non-Anonymous H2 H3 H1 Brainstorming Effectiveness: Quantity Novelty Utility
17 Shalley (1995) conducted two studies to investigate the effect of coacting group members, expected evaluation, and goal setting on individual creativity and productivity while working on a complex-heuristic task. Results of study 1 revealed that creativity was highest for individuals who worked alone and productivity was highest for individuals who expected no evaluation. However, contrary to what was pred icted, Shalley (1995) found insignificant mean differences in productivity between individuals working alone and coacting group members, and insignificant differences in overall creativity be tween no expected evaluation and expected evaluation. Shalley (1995) conducted her second st udy to address the impact of creativity goal setting and to address the limitations of study 1. St udy 2 revealed that when individuals worked alone and were told to be creative in a no-evalua tion environment, they had the highest level of creativity. Productivity was low when individuals worked alone or were assigned a creativity goal. Evaluations that are more passive and gene rally less intentional than individuals being explicitly told that their performance would be ev aluated are referred to as social facilitation or social inhibition (Amabile 1996). An example of passive valuation would be working in the presence of others. Findings as to whether perform ance is enhanced when working alone or in another persons presence are mixed (Forsyth 1990) Triplett (1897) is well cited for the first study to indicate that the presence of others mo tivates individuals. Triplett (1897) observed that the speed for bicyclists, in the company of othe r competing bicyclists, was significantly faster than those bicyclists who raced alone. Zajonc (1965), using the work of Triplett (1897) and Allport (1924), proposed a drive theory of so cial facilitation. According to Zajonc (1965), whether performance is enhanced or increased when working on a team or in the mere presence of others depends on whether the task is an easy, we ll-learned task or a challenging, difficult task. Cottrells (1968) conceptualization of soci al facilitation is that the potential to be evaluated is an antecedent to the increased ge neral arousal produced by the mere presence of
18 others (Zajonc 1965; Cottrell et al. 1968; Gagne and Zuckerman 1999). Evaluation apprehension posits that arousal is not only caused by the me re presence of others, but by those others who have the potential to evaluate ones performan ce (Cottrell et al. 1968; Henchy and Glass 1968; Bond and Titus 1983). Cottrell et al. (1968) was the first study to challenge the notion that the mere presence of others is responsible for audience effects on performance (Platania and Moran 2001). Cottrell et al.s (1968) results were similar to those obtained by Zajonc (1965) in that the presence of interested spectators increased arousal. However, Cottrell et al.s (1968) results also indicated that when the audience is not observing and not interested, the arousal response is not significantly different relative to those who pe rformed the task alone. Evaluation apprehension suggests that arousal is caused by individuals within the environment that have the potential to evaluate ones performance (Cottrell et al. 1968; Henchy and Glass 1968; Bond and Titus 1983). 2.7.2 GSS and Evaluation Apprehension The benefits of computer-med iated groups have been extensively investigated. Typically labeled as GSS, these systems have built-in feat ures such as anonymity, parallel communication, and group memory, to minimize communication barriers (Pinsonneault and Kraemer 1989; Bamber et al. 1996; Bamber et al. 1998). Prio r GSS research has suggested that anonymity reduces evaluation apprehension because individuals can generate ideas without fear of criticism (Nunamaker et al. 1997). Conversely, in non-a nonymous computer-mediated groups, evaluation apprehension has the potential to impair crea tivity and the production of good ideas. The potential to be evaluated is reduced through anonymous computer-mediated groups, allowing individuals to express unique ideas, free of being criticized by peer or superior team members (Barki and Pinsonneault 2001; Dennis et al. 2001). Collaros and Anderson (1969) manipulated th e level of evaluation apprehension through interaction mode. Teams either in cluded all experts or one expert (unidentified), while the control group did not have any member identified as an expert. The authors found that participants in the
19 control group, with no mention of expertise, fe lt the least amount of evaluation apprehension, and, on average, had the highest rating score on practicality and origina lity of ideas. The one expert group mean score on creativity was significantly higher than the all experts group. Diehl and Strobe (1987) manipulated high and low evaluation apprehension through the belief that performance would be evaluated by judg es and peers, respectively. The main effect of this manipulation on productivity (the generation of nonr edundant ideas) was significant. In other words, high evaluation apprehension led to significantly fewer nonredundant ideas than low evaluation apprehension. Cooper et al. (1998) ex amined the effect of anonymity on generating controversial ideas when the topic is more controve rsial or less controversial. Individuals working under GSS-anonymous cond itions produced more controversia l comments than other treatment groups and GSS-anonymous groups produced more nonredundant ideas than individuals of nonanonymous groups. Examination of the mean scores on perceived evaluation apprehension supported the notion that anonymity reduces eval uation apprehension for both noncontroversial and controversial topics. Jessup et al. (1990) found that anonymous group members communicated more effectively than non-anonymous group members. Sp ecifically, the authors stated that the data suggest that anonymous groups are more critical and probing and more likely to embellish an idea (Jessup et al. 1990, p. 318). In a simila r study, Jessup and Tansik (1991) manipulated evaluation apprehension (anonymous vs. non-anonymous ). As predicted, the main effects of both anonymity and group proximity were significant on generating comments. While the GSS literature on anonymity h as shown mixed results (Pinsonneault and Heppel 1998; Dennis et al. 2001; Murthy 2002), the advantage of anonymity remains a strong argument in recent literature. Vitharana and Ra mamurthy (2003) look ed into a software development teams ability to identify flaws in the software. The authors argue that anonymity may be beneficial for software inspection team s, whose members are typically peers but have
20 explicit hierarchical differences. Using a complex software inspection task that involved correctly identifying seeded errors, the authors found that anonymity enhanced software inspection. Those in a three-person anonymous gr oup could neither identify other team members nor trace which member identified a software defect. The three-person non-anonymous groups were less effective. The above discussion leads to the first research hypothesis, stat ed below in alternate form: H1: In a computer-mediated brainstorming session among members of a hierarchical audit team, auditors interacting anonymously will be more effective at brainstorming than audito rs interacting non-anonymously. H1a: In a computer-mediated brainstorming session among members of a hierarchical audit team, auditors interacting anonymously will generate a greater quantity of fraud ideas than auditors interacting nonanonymously. H1b: In a computer-mediated brainstorming session among members of a hierarchical audit team, auditors interacting anonymously will generate more novel fraud ideas than auditors interacting nonanonymously. H1c: In a computer-mediated brainstorming session among members of a hierarchical audit team, auditors interacting anonymously will generate more useful fraud ideas than auditors interacting nonanonymously. 2.8 The Effect of Paradigm-Modifying Creativity on Brainstorming Effectiveness A substantial body of literature suggests th at individuals think with a narrow set of solutions when trying to solve complex problem s rather than thinking creatively (Tversky and Kahneman 1974; Connolly et al. 1993; Hender et al. 2002). The detection of fraud is a complex task, and the objective of creativity techniques is to develop a new way of looking at complex problems and to develop ideas that would not be accomplished through traditional problemsolving approaches (VanGundy 1988; Couger 1996; Lowe et al. 2002). As shown in figure 3, Barlow (2000) graphically depicts the notion of using creativity as an approach for providing insight to the real problem.
21 FIGURE 3. AN INSIGHT MODEL OF CREATIVITY Adopted from Barlow (Barlow 2000) Guided fantasy helps participants step out of their current frame of thought into a fantasy frame where they are asked temporarily to suspend disbelief. They are then asked to generate ideas by relating their fantasies to th e problem (Nagasundaram and Bostrom 1995, p. 95). Guided fantasy is a form of symbolic play where pretending takes place (Piaget 1962; Bateson 1976). Russ et al. (1999) contends that pret end play is the most important type of stimuli for creativity. According to Dansky (1999), an activit y is playful to the extent that an individual is intrinsically motivated, self-directed, and free fro m external rules or constraints, and the link between the means and ends is loose and flexible. Satzinger et al. (1999) studied whether the type of social interaction would impact the ideas generated by individuals. Social interacti on was the information participants were exposed to via group memory. Group memory exposed pa rticipants to either a paradigm-preserving technique (force field analysis) or a paradigm-modifying technique (guided fantasy). Those participants exposed to paradigm-modifying ideas tended to generate additional paradigmTHE REAL PROBLEM (INTUITIVELY PERCEIVED) The new viewpoint The old viewpoint Insight Shift Old ideas Newly available ideas
22 modifying ideas to add to group memory. Likewise, those participants exposed to paradigmpreserving ideas tended to generate additional paradigm-preserving ideas to add to group memory. Even when individuals had creative styles different than their respective treatment, their creative style was influenced by the ty pe of creative technique they used. In a similar, more recent study, Garfield et al. (2001) evaluated the effect creativity techniques have on an individuals creative outpu t. They were interested in whether ideas generated by participants woul d be paradigm-preserving or paradigm-modifying ideas based on the creative technique used, the type of ideas generated from phantom team members, and participants measured personality type and creative style. They concluded that while individual characteristics were important, the number of paradigm-modifying ideas were significantly greater for those individuals using an intuitive technique, guided fantasy, than those who used an analytical technique, force field analysis. Hender et al. (2002) manipulated the type of stimuli received by participants. Undergraduate participants asked to generate ideas to improve a restaurants ability to maintain customers were randomly assigned to either electronic brainstorming alone (no stimuli), a creative technique known as assumption reversals (related stimuli)4, or a creative technique known as analogies (unrelated stimuli)5. Similar to the findings of Satzinger et al. (1999) and Garfield et al. (2001), participants exposed to unrelated stimuli produced significantly more creative ideas (measured on originality and paradi gm relatedness) than those participants who received no stimuli or a related stimuli. The above discussion leads to the second research hypothesis, stated below in alternate form: 4 Assumption reversal idea generation technique is when individuals write down all the assumptions they know about the problem. The listed assumptions are then reversed in any way possible. Participants use the reversed list of assumptions as a stimulus for generating ideas. 5 Analogies idea generation technique involves generating a list of analogies or problems that are similar in concept. Subjectively, the individual or group selects one or more of the analogies and provides additional detailed information about the analogy while dismissing thoughts about the initial problem. These details or unrelated stimuli are then forced back to the original problem to assist with the generation of ideas for the original problem.
23 H2: In a computer-mediated brainstorming session among members of a hierarchical audit team, auditors receiving training in a paradigm-modifying creativity technique will be more effective at brainstorming than auditors receiving no creativity training. H2a: In a computer-mediated brainstorming session among members of a hierarchical audit team, auditors in a paradigm-modifying creativity technique will generate a greater quantity of fraud ideas than auditors receiving no creativity training. H2b: In a computer-mediated brainstorming session among members of a hierarchical audit team, auditors in a paradigm-modifying creativity technique will generate more novel than auditors receiving no creativity training. H2c: In a computer-mediated brainstorming session among members of a hierarchical audit team, auditors in a paradigm-modifying creativity technique will generate more useful than auditors receiving no creativity training. 2.9 The Effect of Interaction Mode and Para digm-Modifying Creativity Technique on Brainstorming Effectiveness (link 3) As previously mentioned, the environment can be constructive or destructive to creativity (Rhodes 1961). Power distribution, accountability, a nd job status are organizational environments that can constrain creativity (Nunamaker et al. 1991a; Amabile 1996; Couger 1996). Research has demonstrated that accountability, which is inhere nt in a non-anonymous hierarchical audit team, impacts judgment and decision-making (Hoffman and Patton 1997; Rich et al. 1997; Turner 2001; Wilks 2002). The positive aspects of accountability notwithstanding, placing auditors in a frame where they must constantly think about defending their ideas, would have a detrimental effect on their brainstorming effectiveness. Brai nstorming requires playfulness, relaxation, and no criticism. Thus, it is unclear whether a profession th at is held highly accountable to internal and external stakeholders can create an environmen t that is conducive to creative thinking. As noted previously, guided fantasy is an intuitive technique shown to produce or stimulate the production of paradigm-modifying id eas that would otherwise not be generated through the use of analytical techniques or without the assistance of stimuli (Satzinger et al. 1999;
24 Garfield et al. 2001; Hender et al. 2002). By rem oving individuals from their existing paradigm, guided fantasy, psychologically, should remove individuals from environmental constraints (Dansky 1980; Amabile 1996). Amabile (1996) sugg ests that individuals have the ability to reduce the salience of extrinsic goals by the way in which they engage in the task or by removing themselves from those constraints. As previ ously discussed, the Garfield et al. (2001) and Satzinger et al. (1999) studies demonstrate how ideas are more novel when individuals are provided a stimulus designed to free individuals from their traditional paradigm. Further, contributing ideas under condi tions of anonymity reduces evaluation apprehension and enhances team communicati on (Jessup et al. 1990; Wilson and Jessup 1995; Vitharana and Ramamurthy 2003). While junior audit team members could provide valuable input and fresh thinking, they are likely to be apprehensive about providing their ideas candidly when they are interacting with their superior s on the audit team. Similar to guided fantasy, anonymity is designed to remove individuals from environmental factors that may inhibit performance. Through anonymity, individual team me mbers are free of social inhibition and other external constraints. Thus, anonymity and guided-fantasy have complementary effects on performance. Consequently, the combined effect s of anonymous interaction and training in a paradigm-modifying creativity technique should resu lt in the greatest brainstorming effectiveness. This expectation leads to the following interaction hypothesis: H3: The effect of creativity training on brainstorming effectiveness in a computer-mediated brainstorming session will be greater when the interaction mode is anonymous than when it is non-anonymous. H3a: The effect of creativity training on the quantity of fraud ideas generated in a computer-mediated br ainstorming session will be greater when the interaction mode is anonymous than when it is nonanonymous. H3b: The effect of creativity training on the novelty of fraud ideas generated in a computer-mediated br ainstorming session will be greater when the interaction mode is anonymous than when it is nonanonymous.
25 H3c: The effect of creativity training on the usefulness of fraud ideas generated in a computer-mediated br ainstorming session will be greater when the interaction mode is anonymous than when it is nonanonymous. The research hypotheses and the related results are summarized in Exhibit 2.
26 CHAPTER 3: METHOD 3.1 Introduction A 2 x 2 factorial design is employed to in vestigate experimentally whether interaction mode and the use of a creativity training i ndependently and jointly affect brainstorming effectiveness (H1, H2, and H3) (see research model in Figure 2). An important aspect of this study is that it uses a realistic aud it task, internal auditors, and interns6 training to be practicing auditors for one of the Big-4 accounting firms. The task used in this experiment is designed to simulate the task required by SAS No. 99 during the planning stage of the audit cycle. 3.2 Research Design The independent variables manipulated in the 2 x 2 between subjects design are (a) interaction mode: GSS-non-anon ymous (authors of comments made by other team members identified by name and rank) or GSS-anonymous (authors of comments made by other team members identified by team member number only), and (b) paradigm modifying creativity training (guided fantasy training or no creativity tr aining). The experimental design is depicted in Table 1. 6 For the remainder of the dissertation, these pa rticipants are referred to as audit interns
27 TABLE 1 RESEARCH DESIGN LAYOUT 2 x 2 Factorial Research Design Factor 1: Interaction Mode Non-Anonymous Anonymous No Factor 2: Paradigmmodifying creativity training Yes Factor 1: Interaction mode Level 1: GSS-non-anonymous authors of comments made by other team members identified by name and rank Level 2: GSS-anonymous authors of comments made by other team members identified by team member number only Factor 2: Paradigm-modifying creativity training (Creativity Training) Level 1: Creativity Training guided fantasy training Level 2: No Creativity Training 3.3 Task Participants completed two tasks, a training task and the actual experimental task, a fraud task. The training task involved the generation of ideas on how to use excess tea bags. This task was adapted from Garfield et al. (2001). Particip ants were given 7 minutes to brainstorm about how to use excess capacity of tea bags. They we re told that they were employed by a company that was producing an excessive amount of tea bags. Their task was to come up with as many ideas as possible on how to use excess tea bags. Th e purpose of the training task was to (1) familiarize participants with the GSS interface and (2) to assess whether participants would respond to creativity training using a task and tec hnique that had been successfully employed in prior IS research. After completing the tea task, participants were then introduced to the misappropriation of assets task. Participants read a misappropriation of assets case adapted from Strand et al. (2002). The case was on a lumber company similar to Home Depot, and included key accounting personnel such as the controller, the chief accountan t, accounts payable clerk, and so forth. After
28 reading the case, participants were provided 15 minutes to brainstorm about how employees of Lakeview Lumber might commit fraud. 3.4 Participants Three groups of participants were recruited for this study: (1) audit interns from a Big Four CPA firm, (2) internal auditors who were r ecruited from the Institute of Internal Auditors, and (3) staff auditors from one of the Big Four CPA firms and a smaller regional CPA firm. A total of 191 auditors participated in this study: 77 audit interns, 90 internal auditors, and 24 staff auditors7. Box plot tests, used to check for outlie rs, resulted in dropping the data for four participants from the tea training task and dr opping the data for 12 participants from the misappropriation of assets task. All subsequent analys is includes 163 participants for the tea task (74 audit interns and 89 internal auditors) and 155 participants for the fraud task (70 audit interns and 85 internal auditors).8 All participants who complete d the study were paid $15 each. 3.4.1 Audit Interns Junior staff auditors had college degr ees and some practical experience on audit engagements. The audit interns participating in this study have many attributes in common with staff auditors. According to senior personnel from the participating CPA firm, the typical audit intern has completed at least 12 units of acc ounting (the two introductory courses and two intermediate courses or their equivalent). The interview and selection process for interns is the same as the process and selection criteria used for full-time audit hires. Once on the job, and after training, interns are assigned to engagements for th e remainder of their internship. At the time of 7 For staff auditors, multivariate test s, prior to deleting outliers, revealed insignificant differences on Fraud Quantity, Fraud Novelty, and Fraud Usefulness. When outlie rs were deleted, the small sample sizes in each cell (e.g., only two participants were in the no anonymity/no training cell) were too small to support statistical testing (see Table 4). Therefore, sta ff auditors were dropped from further analysis. 8 There was no a priori reason for expecting differe nces among the three groups of participants; thus, separate hypotheses for each group were not pr oposed. However, once data were collected, the demographics (see Table 3) revealed a clear difference between the audit interns and the internal auditors across all demographic measures.
29 this study, interns were attending their second day of training, which exposed them to the accounting firms culture and the knowledge base that is necessary for conducting client service engagements. During and at the end of the in ternship, the audit intern undergoes formal evaluation and is usually considered for fulltime employment. Consequently, audit interns experience the same kinds of pressures to perform well, as do staff auditors. 3.4.2 Internal Auditors Under Standard for the Professional Practi ce of Internal Auditing 1210.A2, internal auditors have a professional res ponsibility relating to fraud while performing normal internal audit responsibilities and in fraud investigations. Further, in light of recent fraud cases, the internal auditor is being asked to become more of a partner and consultant to the external auditor. Internal auditors are in-house experts within the clients environment and may be called upon to brainstorm with external auditors about the po ssibility of fraud in their organization. The task and treatments outlined in this study remain the same for internal auditors. Internal auditors were reminded about responsibilities for investigating fraud and working with external auditors. 3.5 Pilot Study A pilot study was conducted using graduate accounting students enrolled in a contemporary auditing graduate course and undergraduate students enrolled in an internal auditing course. The topic of SAS No. 99 and fra ud brainstorming was covered by the instructor in both courses. The purpose of the pilot study was to ensure that the computerized application, created for the purpose of this study, worked as desired and that the experimental manipulations had the intended effect. Although the participant pool in the pilot study was not large enough to enable formal testing of the hypotheses, the pilot data revealed support for the primary hypotheses regarding the effects of anonymity and creativity training on brainstorming effectiveness. However, it should be noted that although student participants in the pilot study
30 were told to assume they were actual auditors participating in a brainstorming session, it is unlikely that they experienced the pressures that actual audit firm employees would experience. In particular, the pilot study participants were not expecting to be evaluated by senior audit firm personnel as actual audit firm employees would be. Some modifications to the computerized application were made subsequent to the running of the pilot study sessions. 3.6 Covariates As discussed below, prior r esearch has shown that intrinsi c motivation and creative ability can influence brainstorming effectiveness. Therefore, these variables, along with prior brainstorming and fraud detection experiences, are measured and included in the analysis as covariates. 3.6.1 Intrinsic and Extrinsic Motivation Shalley (1995, p. 484) defines intrinsic mo tivation as the inner-directed interest in a task. In order to be intrinsically motivated, i ndividuals must be both interested in the problem and motivated to find a solution. The notion is that in order for individuals to be creative, they need to be motivated to work hard to break down obstacles to cr eativity. When individuals are interested in a task and find the task enjoyable, they are intrinsically motivated. When individuals are primarily motivated to complete a task by goals imposed on them, they are extrinsically motivated (Condry and Chambers 1978; McGraw 1978; Amabile 1983). Expected evaluation is a form of extrinsic motivation and can have detrimental effects on creativity (McGraw 1978; Amabile 1983). Extrinsic motivation can stem from trying to get an award, meet a deadline, or obtain the approval of others or a positive ev aluation from a supervisor, whereas intrinsic motivation comes from within (Condry and Chambers 1978; McGraw 1978; Amabile 1983). Individuals are intrinsically motivated to the extent that they enjoy accomplishing the task without being told or paid to do so. In order for intrinsic motivation to occur, the individual must
31 feel free from strong external control, be engaged in a playful activity rather than work, have a sense of competence in completing the task, and be curious or stimulated by the task (Osborn 1957; Amabile 1983; Couger 1995). Research investigating external evaluation on performance has revealed that this form of extrinsic motivation can have a detrimental eff ect on performance (Cottrell et al. 1968; Shalley and Oldham 1985). In a fraud assessment task, it may be difficult to promote an environment that fosters intrinsic motivation because of several ex trinsic constraints. As recent events have demonstrated, claims of failure to detect fraud may result in a collapse of the stock market, an increase in audit oversight, additional accounting ru les, and/or the accounting firm going out of business (Plitch 2003). These extrinsic motivating fact ors should affect how auditors perceive the fraud assessment task. Extrinsic motivation inhibits an individuals ability to take risk and focus on the task (Amabile 1996). 3.6.2 Creative Person An individuals innate creativity is another dimension of creativity that likely correlates with performance on the experimental task (R hodes 1961). Some individuals are a constant source of creativity in the workplace (Mumford and Simonton 1997). Guildford defines creativity from the individual perspective as the abilities that are most characteristic of creative people. Creative abilities determine whether the individual can exhibit creative behavior to a noteworthy degree (Guilford 1950, p. 444). Research in this area examines individual traits (i.e., personality type, intellect, and habits) associated with the creative product. Several instruments exist for measuring crea tivity traits: the Torrance Tests of Creative Thinking (TTCT; (Torrance 1974)), Guilfords Unusual Uses Test (Guilford 1950), Goughs Creative Personality Scale for the Adjective Checklist (Gough 1979), Kirton Adaption-lnnovation Inventory (KAI; (Kirton 1976)), and Consensual Assessment Technique (CAT; (Amabile 1982)). Kirtons (1976) idea of a creative person is one who has either an adaptive-creative style or
32 innovative-creative style. While both styles are ch aracteristic of a creative person, individuals categorized as adaptive are more likely to fo rm paradigm-preserving ideas, while innovative individuals are more likely to form para digm-modifying ideas (Nagasundaram and Bostrom 1995). Jabri (1991) discusses the limitations of existing measures. For example, Jabri (1991) criticizes KAIs consolidation of scores that f actor on three different dimensions (fluency, efficiency, and rule), into one single score. He argues that valuable information is lost by combining the three dimensions, and thus, provides misleading results. Additionally, KAI is costly to access and administer. Jabris (1991) theory is that individuals have a preferred style for solving problems. Individuals either generally solve problems intuitiv ely or in a logical, systematic manner (Jabri 1991). The systematic approach occurs when an individual follows step-by-step procedures and prefers to stay within the guide lines of rules and problems. The systematic approach is likely to lead to a traditional approach to solving probl ems, generating conventional solutions (Scott and Bruce 1994). The intuitive approach occurs for t hose who tend to retrieve and use information across paradigm boundaries to solve problems, not restricting themselves to established rules and traditional boundaries (Isaksen 1987; Scott and Br uce 1994). Individuals who approach problemsolving intuitively are likely to generate more novel, paradigm-modifying ideas (Isaksen 1987; Scott and Bruce 1994). Synonymous with systematic and intuitive approaches to problem-solving are associative and bisociative thinking (Scott and Bruce 1994). A ssociative thinking is based on habit or set routines that could be expressed in words or by symbols. This is contrasted with bisociative thinking which occurs when two matrices of thought are combined resulting in a nonhabitual thought which is only made known by judgment, decision, or action (Jabri 1991). With these definitions in mind and addressing the limitations of existing measures of problem-solving, Jabri
33 (1991) developed and validated an instrument that consists of two independent subscales: associative thinker or bisociative thinker. Scott and Bruce (1994) used the associat ive/bisociative scales to measure problemsolving style in a model that predicted intuitiv e problem-solving style and systematic problemsolving style would have a direct influence on an individuals innovative behavior. While intuitive problem-solving style had insignificant results, systematic problem-solving style had a significant negative influence on innovative behavi or. The reported Cronbachs alpha was .90 for the associative scale and .91 for the bisociative scale. Shalley and Perry-Smith (2001) adapted five items of Jabris (1991) instrument to measure an individuals creativity ability. The Cronbachs alpha was .73 for the five items. Jabris (1991) subscales, which were used by Scott and Bruce (1994), are used in this study to control for an individuals problem-solv ing style. Because individuals problem-solving style is likely to impact their brainstorming effectiveness, it was necess ary to account for the effect of this potential covariate on the dependent variable. Exhibit 3 shows the 19 items used to measure an individuals most dominate problem-solving style. 3.7 Experiment Materials and Procedures For the audit interns, the experimenter atte nded a training workshop held by a major CPA firm. Prior to participating in the experiment audit interns attended a session held by the CPA firm that provided them with a general overvie w of the audit process and fraud. The experiment for audit interns was administered onsite, in a controlled area designated specifically for the experiment. Approximately three to four computers were set up at each table. The GSS developed for the purpose of the study is Intern et-based and was accessed using a Web browser (i.e., Internet Explorer).
34 Internal auditors used the same GSS system, but were contacted by e-mail and signed up for a time to participate in the study. Expe rimental procedures for all participants took approximately 1 hour and are outlined in Table 2. Participants accessed the GSS system develope d for the study over the Internet, using a web brower. One way to create a virtual team envi ronment and give individual participants the impression that they are participating in a brainstorming session along with other auditors is by creating a simulator that feeds phantom ideas in to the system, as if those ideas are coming from other members (when in reality the ideas are being retrieved from a database). A script that retrieves ideas from a database table and inserts th em into the participants idea log window at random intervals was used to create the illusion th at additional individuals, other than the actual participants, are a part of the team. Each particip ant engaged in electronic brainstorming, which is similar to a chat room or virtual meeting place. The general procedures for all participants were as follows: Using an Internet-enabled computer, participants were instructed to go to the studys Website address. At this point, participants were prompted to enter a user id and password provided by the experimenter. Once participants gained access to the system, they we re presented with a screen containing informed consent information with an option to click on an agree button to proceed with the study or a disagree button to abort the study. All partic ipants selected the agree option, choosing to follow through with the study. Participants were told that the purpose of th e study is to understand the impact that SAS No. 99 has on an auditors ability to assess fraud. Next, participants were asked to enter their first and last name and select the auditing firm for which they work from a drop down list. Internal auditors selected other from the drop down me nu. At this point, the system randomly assigned participants to one of four treatment conditi ons (guided fantasy training with non-anonymous interaction, guided fantasy training with anonym ous interaction, no training with non-anonymous
35 interaction, or no training with anonymous inter action). Next, participants were instructed to respond to pre-experimental questions that consis ted of demographics, the associative/bisociative subscales developed by Jabri (1991) and a measu re of evaluation apprehension to capture participants perception about how they interact when others are present. TABLE 2 PROCEDURES FOR PARTICIPANTS9 Training/ Non-Anonymous Team Training/ Anonymous Team No Training/ Non-Anonymous Team No Training/ Anonymous Team Complete pre-study questionnaire Complete pre-study questionnaire Complete pre-study questionnaire Complete pre-study questionnaire Introduction to Study Introduction to Study Introduction to Study Introduction to Study Tea Task Problem Tea Task Problem Tea Task Problem Tea Task Problem Guided Fantasy-Brazil Gu ided Fantasy-Brazil Brainstorm: Tea Task Brainstorm: Tea Task Brainstorm: Tea Task Brainstorm: Tea Task Employee Fraud Case Employee Fraud Case Employee Fraud Case Employee Fraud Case Guided Fantasy-Inspector Gadget Guided Fantasy-Inspector Gadget Brainstorm: Fraud Task Brainstorm: Fraud Task Brainstorm: Fraud Task Brainstorm: Fraud Task Complete post-study questionnaire Complete post-study questionnaire Complete post-study questionnaire Complete post-study questionnaire 9 Format of table, adopted from Yip-Ow and Tan (2000).
36 After the pre-experiment questions, part icipants saw a screen that states, Please wait while the rest of your team logs on After randomly waiting for approximately 30, 45, or 90 seconds, the screen stated, All team members are now logged on. Please proceed. The purpose of this screen was to enha nce the illusion that participants would be interacting with real independent/ext ernal auditors at other locations. Next, participants were told about the task acco rding to their randomly assigned treatment condition. All participants were first traine d to use the Internet-based brainstorming system through the use of a tea bag machine under-utilization problem. Participants in the creativity training treatment, in additi on, received training on the guided fantasy creativity technique. Participants were told th at they were a part of a four-person team. However, unknown to the participants, each electronic brainstorming session only consists of the actual par ticipant in the study, while th e other three team members appeared to be a part of the team through a program designed to create the illusion that other team members (hereafter referred to as phantom members) existed and were providing input to help the team accomp lish its brainstorming assignment. After training, all participants performed th e following steps sequentially: (1) read the misappropriation of assets case, (2) brainstormed about potential fraud committed by employees mentioned in the case, and (3) completed post-expe riment questions, such as manipulation checks and intrinsic motivation. Although the risk of communication among participants between experimental sessions was minimized by isolati ng audit interns who had completed the study from those who had not as yet participated, and by having internal auditors from different companies, it was impossible to prevent such communication. To minimize the potential contamination of the results of the study due to such communication, all participants were
37 debriefed simultaneously via e-mail once all sessions were completed, rather than after each experimental session. Participants were instructed not to ask qu estions of other team members or comment on others ideas, and to simply offer their own ideas and read ideas put forth by other team members. To eliminate the potential of extraneous variabl es affecting the outcome, all participants received the same pre-scripted ideas. These pre-scripted ideas for the fraud case were derived from ideas generated by graduate students w ho participated in the pilot study and senior auditors and managers of two local CPA firms. These pre-scr ipted ideas from the phantom auditors included a mixture of creative and non-creative ideas, and were programmed to appear on the screen at random intervals. In the non-anonymous treatme nt condition, each comment was tagged with a phantom name that is gender ne utral and a job title (e.g., Pat G ., Manager). In Weisband et al. (1995), tagging comments with a name and title made student participants keenly aware of status differences. In the current study, in the a nonymous treatment condition, each idea was tagged only with a team member number (e.g., Team Member 1), with no indication of the idea authors name or job title. 3.8 Treatments/Independent Variables 3.8.1 Interaction Mode Treatment Interaction mode was operationalized as non-anonymous and anonymous. The nonanonymous interaction mode and the anonymous in teraction mode were expected to induce high and low evaluation apprehension, respectively. Conditions were mode led after Collaros and Anderson (1969), who manipulated inhibition throug h the manipulation of perceived expertise. In both the non-anonymous and the anonymous treatments participants were told that they were on an audit engagement team with three superior team members: senior manager, manager, and senior auditor, who are experts in the area of fraud detection. For the non-anonymous treatment, participants saw the name and rank of the team member making each comment. Further,
38 participants saw their comment tagged with their firs t and last name initial in bold letters. For the anonymous treatment, participants were told that their identity and the identity of others will be concealed and will remain anonymous. Additionally, they were told that their log-in name cannot be traced back to the ideas they submit. All par ticipants were told that the study is designed strictly to determine the effectiveness of SAS No. 99. 3.8.2 Paradigm-Modifying Creative Technique Training Treatment Participants were randomly assigned to eith er guided fantasy training, which is an intuitive creativity technique, or to an unstructu red brainstorming group, in which participants were not trained on a creativity technique, but simply told to brainstorm. The actual instructions for each treatment are outlined in Appendix A, Section 3. Guided fantasy stimuli is a short paragraph intended to be unrelate d to the problem, freeing the individual of external pressure, and expanding the individuals thinking boundaries or imagination. Task-unrelated stimuli increases brainstormi ng effectiveness through the use of concepts unrelated to the problem statement. These concepts are thought to promote paradigm-modifying ideas that would otherwise not be considered us ing task-related stimuli (Satzinger et al. 1999; Garfield et al. 2001; Hender et al. 2002). The no training guided fantasy treatment only receives task-related stimuli (which is the brainstorm ing of ideas), while the guided fantasy training treatment receives both task-related and task-unrelated stimuli. Participants in the guided fantasy training treatment were exposed to two different taskunrelated stimuli, a Brazil stimulus during the tea task training phase and an Inspector Gadget stimulus during the potential fraud brai nstorming phase. For the Brazil stimuli (see Appendix A, Section 3), participants were asked to imagine or fantasize about a vacation in Brazil. The Brazil scenario includes embedded c oncepts such as night life, mosquitoes, and a beach scene. Participants were th en instructed to use these concep ts to assist in generating ideas. For example, knowing that mosquitoes and other insects are a problem during the Brazil vacation
39 may generate an idea to use empty tea bags to screen out mosquitoes. Unrelated concepts in the Inspector Gadget Scenario (see Appendix A, Section 3) include security guards, backdoor entrance, and mechanical monsters. The one secur ity guard may be viewed as being analogous to the internal auditor, who commits fraud. Auditors are encouraged to set aside prior beliefs during the brainstorming phase of the audit (AICPA 2002), thus considering internal factors should not exclude internal auditors. The backdoor entrance by Dr. Claw and his goons could trigger an idea about programmers who could leave a backdoor into the system to allow him/her unlimited access to the firms system, or could include add itional scripts in the code that would transfer minute amounts of each transaction into an account to which he/she has access. The statues may symbolize software programs, initially dormant, th at have the potential to corrupt data. Finally, the Inspector Gadget scenario includes gadge ts such as helping hands, telescopic eyes and neck. In maintaining professional skepticism, a uditors should look beyond the surface, relying on paper and computer trails. The helping hands should trigger ideas about relying on inside informants, the audit committee, and internal auditors. To eliminate timing differences in treatments, only one of the creativity training treatments was administered during each site visit. For those in the guided fantasy training treatment, brainstorming took place after providing participants with the stated problem and an Inspector Gadget scenario/stimuli. Participants first read the case, and then received the Inspector Gadget unrelated stimuli prior to brainstorming about fraud. Participants in th e unstructured brainstorming group were not instructed to use a particular brainstorming tec hnique and were expected to use their natural (instinctive) brainstorming method. The no-tr aining brainstorming group follows the same procedures as the guided fantasy training group, without any training on the unrelated stimuli. 3.9 GSS Technology Using information technologies already deployed in most large public accounting firms (e.g., Lotus Notes), audit teams can use GSS to transcend time and space boundaries. GSS assists
40 audit teams in collaborating within or acro ss boundaries to accomplish tasks (Saunders 2000). One of the primary characteristics of GSSs is parallel communication, which enables team members to brainstorm simultaneously to produ ce a pool of ideas (Bamber et al. 1996). The pool of ideas is created through simultaneous info rmation exchanged among members and becomes a stimuli for the generation of additional ideas relate d to the problem statement (Dennis et al. 1998; Hender et al. 2002). Both the training and no traini ng treatment groups receive related stimuli or pool ideas, thought to promote additional ideas. The G SS literature typically refers to this concept as synergy: Good ideas spur more good ideas, and member utterances may contain task-related stimuli that elicit new ideas from other memb ers (Barki and Pinsonneault 2001, p. 164). The specific type of GSS technology used in th is study is chat, where team members work at the same time (synchronous), but different loca tions (dispersed). Additionally, a key feature of GSS technology is used, that is, some groups are anonymous, while others are not. Because phantom team members are created, one of the ma in goals is to have pa rticipants believe that they are part of a four-member team, when in act uality, they are the only real participant on the team. Social presence theory posits that the use of different communication media, such as GSS, can affect the extent to which factors about th e environment and other team members are salient (Short et al. 1976). The effect of others bei ng present can be achieved by creating a credible illusion that others are working on the same team through the use of GSS. The Internet-based GSS system created for this study was designed to transmit information about the number of individuals on the team, generated ideas, and, for the nonanonymous condition, the implied level of experience of the other team members based on their title. However, lack of verbal (i.e., brief utte rances such as yes, ummm) and visual cues (i.e., physical appearance) can lower social pr esence, which in turn can reduce evaluation apprehension (Short et al. 1976; Nunamaker et al. 1991b; Sia et al. 2002). Communication media transmit information, such as facial expressions a nd hand gestures, in different ways. The extent
41 to which a non-anonymous inte raction mode induces evaluation apprehension depends on the extent to which characteristics about other te am members are presented by the medium. The weight given to these transmitted characteristics is determined by the individual, making social presence a subjective measure of the medium used which in turn influences an individuals behavior (Short et al. 1976). The removal of visual and verbal cues causes the communication media used in this study to have a low degree of social presence. However, the GSS in this study was designed to increase social presence for the non-anonymous team by attaching the job title along with a gender-neutral name to each idea ge nerated by phantom te am members (i.e., Pat G, senior manager). Also, for the non-anonymous tr eatment, the participants first name, initial of last name, and position title was attached to each comment submitted (i.e., Dana S, junior auditor). Position titles in the audit environment symbolize authority and expertise (citation in support of this notion?). Finally, comments from phantom team members were submitted in a format similar to that found in chat sessions. For example, instead of submitting comments that were grammatically correct, some comments were submitted with typographical errors. 3.10 Dependent Variable and Data Collection The dependent variables are Fraud Quantit y, Fraud Usefulness, and Fraud Novelty, referred to collectively as brainstorming effectiven ess. During the training stage, which involves a tea task, brainstorming effectiveness is measured in terms of the quantity and novelty of ideas relating to the use of tea bags, whereas for th e actual fraud case brainstorming effectiveness is measured in terms of the quantity, novelty, and utility of fraud ideas generated by each participant. The software utilized in this expe riment captured and stored the ideas entered by participants. In brainstorming, one school of tho ught is that quantity breeds quality (Osborn 1963; White and Owen 1970). Osborn (1963) argues that it is important to generate as many ideas as possible. The generation of one idea leads to other ideas. It is a way of generating possible hypotheses, where typically the high quality id eas are the last 50 ideas generated during a
42 brainstorming session (Osborn 1963; White a nd Owen 1970). Also, generating many ideas provides alternatives and reassurance that every po ssible idea has been explored, regardless of its utility or usefulness (Osborn 1963; White and Owen 1970). During the brainstorming session, participants are encouraged to build off of othe rs ideas, create novel ideas, and generate as many ideas as possible. Thus, the quantity component of brainstorming effectiveness is measured by counting the number of non-redundant ideas per task type (tea and fraud). The utility component of brainstorming effectiveness is defined as the ex tent to which raters believe the idea would be used in the audit planning process. The novelty of ideas was determined based on whether an idea is rarely mentioned by other participants. Ideas produced by Participant A that were rarely mentioned by other participants we re deemed novel for Participant A. First, to determine the quantity of ideas, two coders were used. The qualifications of the coders can be ascertained by reviewing their curricu lum vitae, which are included in Appendix C. The coders were first asked to code each idea as an identified control weakness, fraud idea, or a comment. The coders also identified redundant id eas. The coders were blind to the hypotheses of this study and independently c oded 1,528 tea ideas and 1,648 fra ud ideas. Cohens Kappa interrater reliability analysis was .692 for tea id eas and .707 for fraud ideas. Both values are statistically significant, indicating that the rate rs coded in a similar manner. After this initial assessment, coders resolved any disagreements, until they reach 100 percent agreement. Nonideas (comments and identified control weaknesses) per individual were eliminated to determine the quantity of ideas. Second, once the non-redundant ideas were identified, this lis t of ideas was submitted to two audit managers from a local CPA firm (see attached resumes) who rated the utility of each idea in the audit planning process. Raters were instru cted to rate the extent to which they believed the idea would be used in the audit planning pro cess, using a 3-point Likert-scale, where 1= not useful and 3= very useful. Manage rs are responsible for reviewing staff auditors work and should
43 be in the best position to rate overall utility of an idea (See Appendix B for instructions to raters). The raters were blind to the hypotheses of this study and independently coded 98 fraud ideas that were either a unique idea (not similar to other ideas) or a representative of other ideas (similar to other ideas). For those ideas that were representa tive of other ideas, the same usefulness score rating applied to the representative was also app lied to all similar items. Cohens Kappa interrater reliability analysis was less than adequate (. 500, p-value < .01). The coefficient implies that at least one-half of the variance may be due to random error (Kline 1998). After this initial assessment, raters resolved a ny disagreements, until they reach 100 percent agreement. An average utility score was calculated for each par ticipant. Raters submitted a brief biography (see Appendix C) to demonstrate their qualifications as a rater and for publication in this study. Third, using the original quantity list, as dete rmined by the coders, an idea rarity score was generated to gauge the novelty of ideas by count ing the number of times each participants idea was listed by other participants, per task. Next, the reciprocal or multiplicative inverse of each idea was computed. For example, if an idea was lis ted three times, that is by three participants across all treatments, then the reciprocal would be 1/3 or .333, which is the rarity score for that idea ascribed to all three participants. Those sco res approaching 1 are indicative of the least common ideas (high on originality), while those sco res closest to zero are indicative of the most common ideas (low on originality). A novelty score for each participant was obtained by summing the rarity scores of each idea submitted by that participant (minus any ideas that was coded as redundant within an individual participant). The Consensual Assessment Technique (CAT ) (Amabile 1996) is employed in the current study to evaluate the utility of products (ideas) generated by each participant. The technique involves using judges to evaluate each idea, and assessing the inter-judge reliability. Amabile (1996) states that three requirements are nece ssary in order for the task to be appropriate for the consensual assessment techni que. First, the task must be one that leads to some product or
44 response that can be observed by judges. This requirement was met by having participants generate observable ideas about employee fraud. S econd, the task must be open-ended to allow flexibility and generation of novel ideas. The misappropriation of assets case was adapted and modified from its original form by providing ge neral background information about the company, being careful not to include obvious red flags that would indicate fraud. Presenting case material to participants without clearly labeling or iden tifying fraud indicators, allowed participants flexibility in generating ideas vers us being confined to red flags already established by the profession. Third, the task should not depend heavily on specific skills (i.e., drawing ability or verbal fluency). However, Amabile (1996) states th at if the task is heavily skill-specific, then participant selection must be based on a pro cess that ensures a uniform level of baseline performance to help reduce extreme performance di fferences between individuals. In this study, although they were not required to be expert fraud examiners, participants needed to be knowledgeable about fraud, such as understanding what fraud is, how it can occur, and what effect fraud has on the financial statements or the audit opinion. In general, junior auditors have general fraud knowledge acquired th rough college courses and in-house fraud training (Bedard et al. 1993).
45 CHAPTER 4: RESULTS 4.1 Descriptive Statistics Demographic data regarding participants are shown in Table 3. On average, across all the groups, participants had less than two years of exte rnal audit experience. As would be expected, a significant number of internal auditors and aud it interns lacked extern al auditing experience. Male auditors represented a smaller portion of th e sample (42%, n=76) than female auditors (58%, n=103). The majority of the participants were between the ages of 20-24. There were 133 participants between the ages of 20 and 35. Most of the participants had previous brainstorming experience (88% for internal auditors, 96% for a udit interns, and 88% for staff auditors). While audit interns lacked fraud experience, a majority of the internal auditors had worked on an engagement where fraud was either suspected, detected, or both. Additionally, descriptive statistics indicate that these three populations (audit interns, internal auditors, and staff auditors) have characteristics that prevent them from being homogeneous, or grouped together as one population. The number of participants in each tr eatment condition is displayed in Table 4.
46 TABLE 3 PARTICIPANT DEMOGRAPHICS (all participants, n= 179) Demographic Information Items Internal Auditors (n=85) Audit interns (n=70) Staff Auditors (n=24) Years of Internal Auditing Experience: Mean 5.14 SD 5.18 Min 0 Max 28 Years of External Auditing Experience: Mean 1.51 .01 1.61 SD 2.33 .12 1.20 Min 0 0 0 Max 10 1 4 Gender: Female 50 38 15 Male 35 32 9 Age 20-24 2 65 7 25-29 13 4 13 30-34 24 1 4 35-39 12 40-44 17 45-49 10 50 or more 7 On approximately how many audit engagements have you worked in your auditing career?: Mean 44.51 .04 14.79 SD 32.22 .266 11.25 Min 1 0 0 Max 99 2 40 Have you worked on an audit engagement where fraud was suspected? Yes: 58 (68%) Yes: 0 Yes: 4 (16%) Have you worked on an audit engagement where fraud was detected? Yes: 43 (51%) Yes: 0 Yes: 2 (8%) Have you ever brainstormed (i.e., hastily write down thoughts) with others (in a group setting, in any context)? Yes: 75 (88%) Yes: 67 (96%) Yes: 21 (88%) Highest Level Education: Bachelors degree 57 (67%) 7 (10%) 10 (42%) Masters degree 27 (39%) 1 (1.4%) 14 (58%) Ph.D. 1 (1%) Who had training related to SAS #99 63 (80%) 13 (19%) 12 (50%)
47 TABLE 4 NUMBER OF PARTICIP ANTS IN EACH TREATMENT CONDITION FOR THE FRAUD TASK Panel A: Audit Interns Participants per Treatment Interaction Mode No Anonymity Anonymity Count Count Creativity Training No Training 17 18 Training 19 16 Panel B: Internal Auditors Participants per Treatment Interaction Mode No Anonymity Anonymity Count Count No Training 26 21 Creativity Training Training 18 20 Panel C: Staff Auditors Participants per Treatment Interaction Mode No Anonymity Anonymity Count Count No Training 6 7 Creativity Training Training 6 5 4.2 Correlation Matrices The Pearsons correlation coefficient (Pearson r ) is presented in Tables 5 and 6 for two reasons. First, multivariate analysis of variance (MANOVA) creates a combined variate of all the dependent variables and controls for experime nt-wide error rate. Based on theory discussed previously, and the significance of the Pearson r the use of MANOVA to combine the set of dependent variables and determine their effect, if any, across treatment groups is supported. Second, the Pearson r and its associated significance level are used to determine whether the
48 continuous covariates and other measures (intrinsic motivation, creative ability, evaluation apprehension, and social presence) are correlate d with the dependent variables. In selecting covariates to place in the model, it is important that covariates are highly correlated with the dependent variables without being highly correlate d with the independent variables (Hair et al. 1998). Table 5 shows the correlation matrix for audit interns. The correlation matrix for the tea task (Panel A) shows a significant correlation be tween the two dependent variables, Tea Quantity and Tea Novelty (Pearson r = .783, p-value < .01). The covariate, mean pre-evaluation apprehension, is negatively correlated with Tea Quantity (Pearson r = -.296, p-value < .05) and Tea Novelty (Pearson r = -.261, p-value < .05). Finally, mean post-evaluation apprehension is negatively correlated with Tea Quantity only (Pear son r = -.244, p-value < .05). The correlation matrix for the fraud task (Panel B) shows that Fraud Quantity has a significant positive correlation with both Fraud Nove lty (Pearson r = .606, p-value < .01) and Fraud Usefulness (Pearson r = .411, p-value < .01). However, Fraud Novelty and Fraud Usefulness are not significantly correlated (Pearson r = .163, p-value=.177). Table 6 presents the correlation matrix for inte rnal auditors. The correlation matrix for the tea task (Panel A) shows a significant correlation between the two dependent variables for the tea task, Tea Quantity and Tea Novelty (.690, p-valu e < .05), and between the covariate mean score of intrinsic motivation and Tea Quantity (Pear son r=.303, p-value <.01) and Tea Novelty (Pearson r = .247, p-value < .05). The correlation matrix for the misappropriation of assets task (Panel B) shows correlation between the dependent variables, Fraud Quantity, Fraud Novelty, and Fraud Useful, however, covariates are not significan tly correlated with either of the dependent variables.
49 TABLE 5 CORRELATION MA TRIX FOR AUDIT INTERNS Pearson Correlation Coefficient (Sig. 2-tailed) Panel A: Audit interns (n=74) Tea Task Tea Quantity Tea Novelty Mean Pre EA Mean Post EA Mean Social Presence Mean Intrinsic Motivation Associative Principal Component Bisociative Principal Component Tea Quantity 1 .783(**) -.296(*) -.244(*) .044 .137 -.051 .014 Tea Novelty 1 -.261(*) -.111 -.111 -.097 -.065 .095 Mean Pre EAa 1 .293(*) -.100 -.168 .000 -.292(*) Mean Post EAb 1 .018 -.274(*) -.051 -.271(*) Mean Social Presence 1 .269(*) -.123 .035 Mean Intrinsic Motivation 1 .284(*) .213 Associative Principal Component 1 .182 Bisociative Principal Component 1 Correlation is significant at the 0. 05 level (2-tailed). ** Correlation is significant at the 0.01 level (2-tailed). aThe mean score of evaluation apprehension prior to the experiment; bThe mean score of evaluation apprehension after the experiment. Panel B: Audit interns (n=70) Fraud Task Fraud Quantity Fraud Novelty Fraud Usefulness Mean Pre EA Mean Post EA Mean Social Presence Mean Intrinsic Motivation Associative Principal Component Bisociative Principal Component Fraud Quantity 1 .606(**) .411(**) -.201 -.295(*) -.254(*) -.095 -.079 -.015 Fraud Novelty 1 .163 -.191 -.183 -.174 -.064 -.243(*) .190 Fraud Usefulness 1 -.049 -.074 -.307(**) .009 .113 -.051 Mean Pre EAa 1 .300(*) -.096 -.181 -.004 -.282(*) Mean Post EAb 1 .029 -.287(*) -.055 -.253(*) Mean Social Presence 1 .257(*) -.117 .051 Mean Intrinsic Motivation 1 .292(*) .194 Associative Principal Component 1 .181 Bisociative Principal Component 1 Correlation is significant at the 0. 05 level (2-tailed). ** Correlation is significant at the 0.01 level (2-tailed). aThe mean score of evaluation apprehension prior to the experiment; bThe mean score of evaluation apprehension after the experiment.
50 TABLE 6 CORRELATION MATR IX FOR INTERNA L AUDITORS Pearson Correlation Coefficient (Sig. 2-tailed) Panel A: Internal Audito rs (n=89) Tea Task Tea Quantity Tea Novelty Mean Pre EA Mean Post EA Mean Social Presence Mean Intrinsic Motivation Associative Principal Component ( Bisociative Principal Component Tea Quantity 1 .690(**) -.124 -.025 .042 .303(**) -.023 .111 Tea Novelty 1 .010 .152 -.091 .247(*) .092 .059 Mean Pre EAa 1 .238(*) -.077 -.074 .017 -.292(**) Mean Post EAb 1 -.406(**) -.286(**) -.031 -.005 Mean Social Presence 1 .347(**) .083 .116 Mean Intrinsic Motivation 1 .276(*) .036 Associative PC 1 .035 Bisociative PC 1 Correlation is significant at the 0. 05 level (2-tailed). ** Correlation is significant at the 0.01 level (2-tailed). aThe mean score of evaluation apprehension prior to the experiment; bThe mean score of evaluation apprehension after the experiment. Panel B: Internal Auditors (n=85) Fraud Task Fraud Quantity Fraud Novelty Fraud Usefulness Mean Pre EA Mean Post EA Mean Social Presence Mean Intrinsic Motivation Associative Principal Component Bisociative Principal Component Fraud Quantity 1 .665(**) .336(**) -.084 .014 -.072 -.059 -.199 .138 Fraud Novelty 1 .185 -.001 .083 -.129 .060 -.164 .139 Fraud Usefulness 1 .192 .192 .049 -.047 -.103 -.011 Mean Pre EAa 1 .234(*) -.076 -.103 .002 -.310(**) Mean Post EAb 1 -.384(**) -.296(**) -.037 -.017 Mean Social Presence 1 .337(**) .081 .109 Mean Intrinsic Motivation 1 .287(**) .062 Associative PC 1 .062 Bisociative PC 1 ** Correlation is significant at the 0.01 level (2-tailed). Correlation is significant at the 0.05 level (2-tailed). aThe mean score of evaluation apprehension prior to the experiment. bThe mean score of evaluation apprehension after the experiment.
51 4.3 Effectiveness of Training The purpose of this section is to assess the eff ectiveness of the tea bag training task, which was used to train participants according to thei r treatment before engaging in the actual task, misappropriation of assets task. First, the mean, st andard deviation, and cell size is reported for Tea Quantity and Tea Novelty, grouped by treat ments. Second, the MANOVA assumptions for audit interns are discussed, followe d by the related multivariate test results. Next, the assumptions are reviewed again for internal auditors and the related MANOVA results are reviewed. For all multivariate analysis tests throughout this study, the F-statistic and p-value of the omnibus test (Pillais Trace, Wilks Lambda, Hotellings Trace, and Roys Largest Root) were identical for each measured variable. Only th e Pillais Trace statistic is reported for each analysis. While, Wilks lambda is the most widely used, Pillais is reported since it is the most robust with respect to violations of the normality and homogene ity of variances assumptions (Bray and Maxwell 1985). The assumption of independent observations is satisfied for both the audit interns and the internal auditors. Since the experiment is a between subjects design, participants were not measured on the same variable more than once dur ing the study. The value of dependent variables per participant does not influence the value of depe ndent variables for other participants. Further, participants were randomly assigned to each treatment group and the data is not of a time-series nature. MANOVA and parametric tests, in general, center around the assumption of equal variance and covariance and a normal distribution. Thus, in cases where the assumptions were violated, a two-independent-samples nonparametric test using Mann-Whitney, available through SPSS, was conducted. Mann-Whitney does not assume a normal distribution. Similar to ANOVA, the purpose of this test is to determine whether the values of a variable differ across treatment groups. However, the problem with this nonparametric test is that only one independent variable at two levels (i.e. no anonymity and anonymity) can be examined, discounting the contributions of other
52 independent variables and covariates to the model. SPSS refers to the nonparametric test as a twoindependent-samples test. In all cases, the Mann-Whit ney test statistic supported the results of the parametric statistics. 4.3.1 Tea Quantity (Number of Ideas Generated) Tea Quantity was measured during the 7-minute training exercise that involved generating ideas about how to use excess tea bags. Du ring this tea task, the chat application stored all ideas generated by participants. Two coders first identified each idea as either an idea or a comment. The coders also identified redundant ideas. Non-ideas were eliminated and only the non-redundant ideas, as determined by the coders, were counted per participant. The descriptive data for both audit interns and internal auditors ar e shown in Table 7. For the audit interns (Panel A) that did not have anonymity, the highest ove rall mean occur when they did receive training (=5.05). However, for internal auditors (Panel B) that did not have anonymity, the highest overall mean occur when there was no training (=5.57). TABLE 7 TEA QUANTITY (N UMBER OF IDEAS GENERATED) DESCRIPTIVE STATISTICS (mean, standard deviation, range, n) Panel A: Audit interns (n=74) Creativity Training No Training Training No Anonymity 3.89 (2.21) 0 7 19 5.05 (1.99) 3 10 19 4.47 (2.15) 38 Interaction Mode Anonymity 4.16 (1.80) 1 7 19 3.76 (2.08) 1 6 17 3.97 (1.92) 36 4.03 (1.99) 38 4.44 (2.10) 36
53 Panel B: Internal Auditors (n=89) Creativity Training No Training Training No Anonymity 5.57 (2.20) 1 9 28 5.20 (2.75) 1 12 20 5.42 (2.42) 48 Interaction Mode Anonymity 3.48 (2.16) 21 0 9 5.45 (3.07) 1 12 20 4.44 (2.79) 41 4.67 (2.40) 49 5.32 (2.88) 40 4.3.2 Tea Novelty The descriptive statistics for Tea Novelty ar e presented in Table 8, showing the average novelty score for participants in each condition. Tea Novelty was measured by how frequently the same idea was mentioned, thus each idea receive d a novelty score ranging from 0 to 1, with scores close to 0 indicating ideas that were not ve ry novel (i.e., mentioned by most participants) and scores approaching 1 indicating more novel ideas (i.e., mentioned by very few other participants). As with the number of ideas for th e tea task, the highest overall mean Novelty score occurred for audit interns who were trained and did not have anonymity (=1.561). However, for internal auditors (Panel B), the highest mean occurred when they received training and the interaction was anonymous (=1.4766, s.d.=.972, n= 20), which is inconsistent with the internal auditors results for Tea Quantity.
54 TABLE 8 TEA NOVELTY DE SCRIPTIVE STATISTICS (mean, standard deviation, range, n) Panel A: Audit interns (n=74) Creativity Training No Training Training No Anonymity 1.54 (1.26) 0 3.68 19 1.56 (.97) .42 3.76 19 1.51 (1.11) 38 Interaction Mode Anonymity 1.30 (.74) .25 2.66 19 1.12 (.66) .08 2.06 17 1.21 (.70) 36 1.38 (1.02) 38 1.35 (.86) 36 Panel B: Internal Auditors (n=89) Creativity Training No Training Training No Anonymity 1.45 (1.18) .07 4.24 28 1.32 (1.04) .10 4.37 20 1.40 (1.11) 48 Interaction Mode Anonymity .87 (.68) .0 2.49 21 1.48 (.97) .04 3.35 20 1.17 (.88) 41 1.20 (1.03) 49 1.40 (1.00) 40 4.3.3 Audit Interns and Tea Task 126.96.36.199 Multivariate Normal Distribution Assumption To determine whether the assumption of multivariate normality of the set of dependent variables (Tea Quantity and Tea Novelty) acr oss treatment groups was satisfied, several univariate normality tests were performed using SPSS. The shape of the distributed data was reviewed for proximity to a normal distribu tion, the degree of skewness and kurtosis was
55 analyzed using the rule of thumb by Hair et al. (1998), and the box plots and stem and leaf plots were reviewed for extreme outliers. For moderate sa mple sizes, multivariate analysis is robust to departures from normality when it is due to skewness and/or kurtosis, but not outliers (Hair et al. 1998). Three extreme outliers were examined and deleted after reviewing all treatment cells. In the no training, no anonymity treatment group, the Kolmogorov-Smirnov (KS-statistic) for Tea Quantity was insignificant, but significant for Tea Novelty (KS-statistic p-value for Tea Quantity=.137 and .016 for Tea Novelty). There were no extreme outliers in this treatment group. Two extreme outliers were identified in the tr aining/anonymity treatment group. After their deletion, KS-statistic became insignificant (KS-st atistic p-value for Tea Quantity=.200 and .200 for Tea Novelty). For the no training/anonym ity treatment group, the KS-statistic for Tea Quantity was significant, while the KS-statistic fo r Tea Novelty was insignificant (KS-statistic pvalue for Tea Quantity=.033, and for Tea Nove lty=.200). Finally, the training/no anonymity treatment group, the KS-statistic improved overa ll for both dependent variables, but remained significant for Tea Novelty after the deletion of one extreme outlier (KS-statistic p-value for Tea Quantity=.184 and for Tea Novelty=.035). A multivariate normality test was also perfo rmed using principal components analysis. As explained previously, principal component analys is takes the data of the original variable to form one or more principal components that account for a portion of the variance of the original variable (Hair et al. 1998). Assessing the multivar iate normality of the principal component is essentially assessing the multivariate normality of the original data (Tea Quantity and Tea Novelty) (Johnson 1998). 188.8.131.52 Equal Variance-Covariance Assumption The Levenes test of equality of error va riances revealed that both Tea Quantity and Tea Novelty violated the assumption of equal varian ce-covariance across treatment groups (F=2.417, p-value=.074 for Tea Quantity and F=5.795, pvalue=.001 for Tea Novelty). The principal
56 component compiled for these dependent variabl es was checked for equality of error variances across treatment groups. The principal com ponent score supported the univariate findings (F=4.479, p-value=.006). According to Hair et al (1998), MANCOVA is robust to violations of this assumption when the cell sizes are approximately equal, that is, the largest cell size (training/no anonymity=19) divided by the smalle st cell size (training/anonymity=17) is less than 1.5 (19/17= 1.12). 4.3.4 Tea Task Results for Audit interns The multivariate analysis test statistics s howed a significant mean difference across the interaction mode treatment (Pillais Trace=.097, F= 3.341, p-value=.042) and its interaction with training (Pillais Trace=.142, F=5.147, p-value=. 009), after controlling for mean post-evaluation apprehension (Pillais Trace=.125, F=4.14, p-value= .016). The univariate analysis showed that the main effect of interaction mode and the in teraction term were significant for Tea Quantity (F=6.787, p-value < .05 and F=8.444, p-value < 01, respectively). Because of the significance of the interaction term, the significance of the main effect of interaction mode on Tea Quantity cannot be interpreted without k nowing if training was received. For the interaction term, the effect of training is more likely to increase Tea Quantity when no anonymity is provided (=5.05). However, when anonymity is available, the effect of training on Tea Quantity can be detrimental (=3.76). 184.108.40.206 Multivariate Normal Distribution Assumption Similar procedures discussed previously for examining the normality assumption for audit interns were applied thr oughout this study. For internal auditors, only one outlier was deleted, and the assumption of normality was gene rally satisfied throughout each treatment group. In the no training/no anonymity treatment group, KS-statistic p-value= .200 for Tea Quantity and .099 for Tea Novelty. In the training/anonymity treatment group, KS-statistic p-value=.082 for
57 Tea Quantity and .200 for Tea Novelty. For th e no training/anonymity treatment group, KSstatistic p-value=.179 for Tea Quantity and .200 for Tea Novelty. In the training/no anonymity treatment group KS-statistic p-value=.952 for Tea Quantity and .887 for Tea Novelty. MANCOVA is fairly robust to departures from norma lity that are not due to outliers (Hair et al. 1998). 220.127.116.11 Equal Variance-Covariance Assumption For the equal variance-covariance assumption, both the univariate test of this assumption (F=2.070, p-value=.11 for Tea Quantity and F= 1.747, p-value=.164 for Tea Novelty) and the multivariate test using a principal component (F-1 .515, =.217) indicated that the test of equality of variance-covariance was satisfied. 18.104.22.168 Tea Task Results for Internal Auditors The multivariate tests of MANOVA showed insignificant mean differences for all variables, except the interaction term. Subse quent ANOVA test statistics revealed that the interaction term was significant on Tea Quantity (F =4.668, p-value < .05). Tea Quantity is likely to be highest under conditions of no training and no anonymity for internal auditors. However, with no training and anonymity, Tea Quantity is likel y to be lower than Tea Quantity in any other treatment group. 4.3.5 Summary of Training Effectiveness The results of this section suggest that creativ ity training for the studys participants was effective, but only when participants lacked a nonymity. The results are thus indicative of an interaction between creativity training and mode of interaction, at least for the tea bag task used for training purposes.
58 4.4 Fraud Quantity The descriptive data for Fraud Quantity for bot h audit interns and internal auditors are shown in Table 9. For audit interns (Panel A), th e highest overall mean is again found when there was training, but no anonymity (=3.89). For inte rnal auditors, consistent with the Tea Quantity results, the highest overall mean occur in the no training/no anonymity intervention (=5.27). TABLE 9 FRAUD QUANTITY DESCRIPTIVE STATISTICS (mean, standard deviation, range, n) Panel A: Audit interns (n=70) Creativity Training No Training Training No Anonymity 2.88 (1.87) 0 7 17 3.89 (2.16) 1 9 19 3.42 (2.06) 36 Interaction Mode Anonymity 3.11 (1.91) 0 6 18 2.69 (1.352) 0 5 16 2.91 (1.66) 34 3.00 (1.86) 35 3.34 (1.91) 35 Panel B: Internal Auditors (n=85) Creativity Training No Training Training No Anonymity 5.27 ( 2.88) 1 11 26 4.06 ( 2.31) 1 10 18 4.77 ( 2.70) 44 Interaction Mode Anonymity 3.76 ( 2.45) 0 8 21 4.95 ( 2.70) 1 11 20 4.34 ( 2.61) 41 4.60 ( 2.77) 47 4.53 ( 2.53) 38
59 4.5 Fraud Novelty Descriptive statistics for Fraud Novelty ar e quite different when comparing the audit interns and the internal auditors (Table 10). Fo r audit interns (Panel A), similar to previously discussed dependent variables, the highest overall mean was found in training, but not anonymity (=.732). As with the novelty of tea task ideas, the highest overall mean Novelty score for fraud ideas occurred for internal auditors who were trained and who had anonymity (=.532). TABLE 10 FRAUD NOVELTY DESCRIPTIVE STATISTICS (mean, standard deviation, range, n) Panel A: Audit interns (n=70) Creativity Training No Training Training No Anonymity .33 (.29) .0 1.02 17 .73 (.67) .08 2.15 19 .54 (.56) 36 Interaction Mode Anonymity .32 (.22) .0 .72 18 .21 (.144) .0 .58 16 .27 (.19) 34 .33 (.25) 35 .49 (.56) 35 Panel B: Internal Auditors (n=85) Creativity Training No Training Training No Anonymity .51 ( .44) .02 1.63 26 .30 ( .24) .0 1.02 18 .42 ( .39) 44 Interaction Mode Anonymity .37 ( 38) .0 1.25 21 .53 ( .43) .03 1.49 20 .45 ( .41) 41 .45 ( .42) 47 .42 ( .37) 38
60 4.6 Fraud Usefulness Two audit managers from a local CPA firm ra ted the usefulness of each idea to the audit planning process on a 3-point scale (1=not useful, 2=useful, and 3=very useful). The descriptive statistics are presented in Table 11 for both audit in terns (Panel A) and internal auditors (Panel B). For audit interns, the highest overall mean on Fraud Usefulness was found when training was provided with anonymity (=2.49). This was th e first dependent variable where the highest overall mean was reported under conditi ons of anonymity for audit interns. For internal auditors, the highest overall mean occurred for particip ants who were trained and did not receive anonymity (=2.61). The highest mean being reported in the training/no anonymity cell was the first occurrence for internal auditors.
61 TABLE 11 FRAUD USEFULNESS DESCRIPTIVE STATISTICS (mean, standard deviation, range, n) Panel A: Audit interns (n=70) Creativity Training No Training Training No Anonymity 1.87 (1.04) 0 3 17 2.39 (.33) 2 3 19 2.14 (.78) 36 Interaction Mode Anonymity 2.24 (.79) 0 3 18 2.49 (.77) 0 3 16 2.36 (.78) 34 2.06 (.92) 35 2.43 (.57) 35 Panel B: Internal Auditors (n=85) Creativity Training No Training Training No Anonymity 2.54 ( .37) 2 3 26 2.61 ( .41) 2 3 18 2.57 ( .38 ) 44 Interaction Mode Anonymity 2.33 ( .92) 0 3 21 2.51 ( .53) 1 3 20 2.42 ( .75 ) 41 2.45 ( .67 ) 47 2.55 ( .47 ) 38 4.7 Effect of Covariates and Other Measured Variables In order to assess the reliability of measures, the Cronbachs alpha ( ), mean ( ), and standard deviation are reported for the measure s of preand post-experimental measure of evaluation apprehension and extrinsic and intrinsi c motivation. All measures, except a measure of extrinsic motivation, show an acceptable measure of internal consistency ( > .70) (Hair al. 1998). A low alpha means that the inter-item consis tency or reliability is low and the opposite is true for a high alpha. The reliability estimates we re based on the number of participants included
62 in the fraud task. Table 12 presents the measures for preand postevaluation apprehension, and intrinsic and extrinsic motivation. Assessing relia bility is ascertaining the degree of confidence that can be placed in the scores (Pedhazur and Schmelkin 1991). For extrinsic motivation, the reliability scores were relatively low, but remaining measures were reliable ( >.70). TABLE 12 CRONBACHS ALPHA OF MEASURED ITEMS Scale (Measured Items) Audit interns (n=70) Internal Auditors (n=85) Pre-Evaluation Apprehension .848 (4) .897 (4) Post-Evaluation Apprehension .931 (4) .920 (4) Extrinsic Motivation .365 (4) .487 (4) Intrinsic Motivation .862 (5)a .860 (5)a aAlthough participants answered 6 it ems for intrinsic motivation, reli ability statistics indicated that Cronbachs alpha would increase from .792 to .862 for a udit interns if the first intrinsic motivation item was deleted and from .713 to .860 for internal auditors. *Number in ( ), represent N of items. 4.7.1 Evaluation Apprehension: Evaluation apprehension questionnaire items were completed by participants before the experiment (pre-evaluation apprehension) and after the experiment (post-evaluation apprehension). In measuring pre-evaluation apprehen sion, participants were asked four questions using a 7-point scale: (1) Usually in a group, I am reluctant to offer an idea for fear of criticism from other members, (2) Usually in a group, I feel inhibited in offering an idea due to the presence of others who have more experience with brainstorming, (3) Usually in a group, if I offer an idea that is 'way out,' I get discour aged if I sense a certain disapproval from team members, and (4) I tend to withhold ideas, for fear of possible disapproval from other members. The pre-evaluation apprehension measurement was reliable for both audit interns and
63 internal auditors. The post-questionnaire items, asked similar questions but in a past tense: (1) I was reluctant to offer an idea for fear of critic ism from other members, (2) I was inhibited in offering an idea due to the presence of others, (3) Although no overt criticism was expressed, I was reluctant to offer an idea that was 'way out ,' for fear of disapproval from members, and (4) I withheld ideas for fear of possible disapprova l from other members. Table 13 and 14 reports the mean scores of preand post-evaluation a pprehension for the audit interns and internal auditors, respectively. A difference score was computed (mean post-evaluation apprehension score minus mean pre-evaluation apprehension score) for both audit interns and internal auditors. In most cases, both for audit interns and internal auditors, eval uation apprehension decreased after participating in the study. However, as expected, there w as a significantly larger decrease in evaluation apprehension when audit interns brainstormed anonymously (= -1.324) than when they brainstormed non-anonymously (= -.278) (F=10.83, p-value < .01). Likewise for internal auditors, the larger decrease in evaluation appreh ension occurred when par ticipants brainstormed anonymously (=-1.421) than when they brains tormed non-anonymously (=-.890), and this difference was marginally significant (F=3.829, p-value < .10). An additional ANOVA test (F=2.814, p-value < .10), using the difference score as the dependent variab le and the participant group as the independent variable, revealed that, marginally, the largest decrease in evaluation apprehension significantly occurred for internal aud itors (= -1.149) rather than for audit interns (= -.786).
64 TABLE 13 EVALUATION APPREHENSI ON DESCRIPTIVE STATISTICS (mean, standard deviation, n) audit interns (n=70) Panel A: Mean Pre-E valuation Apprehension Panel B: Mean Post-Evaluation Apprehension Creativity Training No Training Training No Anonymity 3.24 (1.07) 17 2.49 (.75) 19 2.84 (.98) 36 Anonymity 3.25 (1.33) 18 3.14 (.96) 16 3.20 (1.15) 34 Interaction Mode 3.24 (1.19) 35 2.79 (.90) 35 Creativity Training No Training Training No Anonymity 2.66 (1.26) 17 2.47 (1.55) 19 2.56 (1.40) 36 Anonymity 1.86 (1.26) 18 1.89 (.96) 16 1.88 (1.11) 34 Interaction Mode 2.25 (1.30) 35 2.21 (1.33) 35
65 TABLE 14 EVALUATION APPREHENSI ON DESCRIPTIVE STATISTICS (mean, standard deviation, n) internal auditors (n=85) Panel A: Mean Pre-E valuation Apprehension Creativity Training No Training Training No Anonymity 2.47 (1.17) 25 2.25 (1.18) 18 2.38 (1.16) 43 Interaction Mode Anonymity 2.62 (1.29) 21 2.76 (1.12) 20 2.69 (1.20) 41 2.54 (1.21) 46 2.52 (1.16) 38 Panel B: Mean Post-Evaluation Apprehension Creativity Training No Training Training No Anonymity 1.56 (1.10) 26 1.38 (.83) 18 1.48 (.99) 44 Interaction Mode Anonymity 1.24 (.45) 21 1.30 (.57) 20 1.27 (.51) 41 1.41 (.88) 47 1.34 (.70) 38 4.7.2 Social Presence Social presence was measured to get a sense of experimental realism for participants. Participants brainstormed in a computer mediat ed environment without the physical appearance of superiors. As previously discussed, the comm unication media used can impact the extent to which factors about other team members are salient This is the first known study in auditing that
66 used phantom members to create the illusion of distributed power within an audit team. Thus, it is important to determine to what extent partic ipants believed they were actually communicating with other auditors. After participating in the st udy, participants answered two questions on a 7 point scale where 1=strongly disagree and 7=strongly agree: (1) The session was functionally equivalent to a scenario where I was in the same r oom with my team, each at a different computer terminal, and (2) The session worked as well as it would have if the team was in the same room. The two items were positively correlated fo r audit interns (Pearson r =.693, p-value < .01) and for internal auditors (Pearson r = .432, p-va lue < .01). Using the mean social presence score as the dependent variable and the participant group as the independent variable, ANOVA results (F=17.519, p-value < .01) revealed that social presence for internal auditors (=4.747) was significantly higher than social presence for audit interns (=3.650). However, among audit interns and internal auditors, social presence was not significantly different for those who had anonymity, than for those who did not have anonymity (F=.434, p > .10; F=.276, p > .10). Table 15 and Table 16 report the descriptive statistics for audit interns and internal auditors, respectively. TABLE 15 SOCIAL PRESENCE DESCRIPTIVE STATISTICS (mean, standard deviation, n) audit interns (n=70) Creativity Training No Training Training No Anonymity 3.68 (1.70) 17 3.87 (1.42) 19 3.78 (1.54) 36 Interaction Mode Anonymity 2.92 (1.73) 18 4.19 (1.68) 16 3.51 (1.80) 34 3.29 (1.73) 35 4.01 (1.53) 35
67 TABLE 16 SOCIAL PRESENCE DESCRIPTIVE STATISTICS (mean, standard deviation, n) internal auditors (n=85) Creativity Training No Training Training No Anonymity 4.73 (1.78) 26 4.56 (1.41) 18 4.66 (1.62) 44 Interaction Mode Anonymity 4.86 (1.59) 21 4.83 (1.59) 20 4.84 (1.57) 41 4.79 (1.68) 47 4.70 (1.50) 38 4.7.3 Task Complexity Zajonc (1965) and Amabile ( 1983) have argued that task complexity is necessary in order to inhibit an individuals performance when others are present. Thus, to determine if participants perceived the fraud task to be complex, particip ants were asked to indicate their response to a complexity task question on a 7-point scale adapted from Pinsker (2002). Specifically, the questioned stated, I thought that the experimental task was very easy, where 1=strongly disagree and 7=strongly agree. ANOVA results f ound a marginally significant mean difference on interaction mode (F=3.198, p-value < .10). Those audit interns provided no anonymity (=3.972, n=36) were somewhat more likely to consider the fraud task complex than those who were anonymous (=4.529, n=34). Similar resu lts were found on the ANOVA test for internal auditors (F=3.514, p-value < .10). Internal auditors in the non-anonymous treatment group (=5.022, n=44) were also somewhat more likely to rate the task as complex than those internal auditors in the anonymity treatment group (=5.610, n=41). According to Zajonc (1965) performance is inhibited when individuals are worki ng on a task that they perceive to be difficult, in the presence of others. Under the Yerkes-Dodson theory (1908), a certain amount of pressure,
68 performance can be enhanced. The findings of th is study are consistent with Yerkes-Dodson (1908), but contradict that of Zajoncs (1965). Table 17 and Table 18 report the descriptive statistics for audit interns and in ternal auditors, respectively. TABLE 17 MEAN TASK COMPLEXIT Y DESCRIPTIVE STATISTICS (mean, standard deviation, n) audit interns (n=70) Creativity Training No Training Training No Anonymity 3.94 (1.30) 17 4.00 (1.25) 19 3.97 (1.25) 36 Interaction Mode Anonymity 4.50 (1.65) 18 4.56 (.96) 16 4.53 (1.35) 34 4.23 (1.50) 35 4.26 (1.15) 35 TABLE 18 MEAN TASK COMPLEXIT Y DESCRIPTIVE STATISTICS (mean, standard deviation, n) internal auditors (n=85) Creativity Training No Training Training No Anonymity 5.00 (1.72) 26 5.06 (1.47) 18 5.02 (1.61) 44 Interaction Mode Anonymity 5.48 (1.21) 21 5.75 (1.29) 41 5.61 (1.24) 41 5.21 (1.52) 47 5.42 (1.41) 85
69 4.7.4 Intrinsic Motivation and Extrinsic Motivation After completing the study, participants were asked to rate their level of intrinsic and extrinsic motivation toward the task using items adapted from Amabile (1979) and Conti et al. (2001), and modified for the purpose of this experiment (see Appendix A, Section 4). Internal auditors had a significantly higher level of intr insic motivation than audit interns (F=25.062, pvalue < .01). Two additional ANOVA tests were conducted using, first, audit interns as the population and then internal auditors as the population. For audit interns who had training, the level of intrinsic motivation was significantly high er than for those with no training (F=4.157, pvalue < .05). The level of intrinsic motivation w as greatest for internal auditors who had training and interacted anonymously (F=6.096, p-value < .01). The measure of extrinsic motivation was not reliable ( = .365 for audit interns and .487 for internal auditors). Table 19 and Table 20 report the descriptive statistics for audit inte rns and internal auditors, respectively. TABLE 19 INTRINSIC MOTIVATION DESCRIPTIVE STATISTICS (mean, standard deviation, n) audit interns (n=70) Creativity Training No Training Training No Anonymity 4.51 (.90) 17 4.92 (.95) 19 4.72 (.93) 36 Interaction Mode Anonymity 4.67 (.86) 18 5.24 (1.29) 16 4.94 (1.11) 34 4.59 (.87) 35 5.06 (1.11) 35
70 TABLE 20 INTRINSIC MOTIVATION DESCRIPTIVE STATISTICS (mean, standard deviation, n) internal auditors (n=85) Creativity Training No Training Training No Anonymity 5.74 (.91) 26 5.33 (.95) 18 5.57 (.94) 44 Interaction Mode Anonymity 5.31 (.97) 21 5.99 (.74) 20 5.64 (.92) 41 5.55 (.95) 47 5.68 (.90) 38 4.7.5 Creative Person Each participate completed Jabris associa tive/bisociative measurement prior to the experiment. The Cronbachs alpha for the cu rrent study is reported in Table 21. TABLE 21 CRONBACHS ALPHA OF MEASURED ITEMSPROBLEMSOLVING SCALE Scale (Measured Items) Audit interns (n=70) Internal Auditors (n=85) Associative Scale .882 (10) .825 (10) Bisociative Scale .744 (9) .752 (9) The objective of using principal component analysis is to reduce the measures of associative and bisociative scales to one princi pal component for each. Specifically, instead of using ten variables to explain associative thinki ng and nine variables to explain bisociative thinking, a principal component for each dimensi on was calculated. Principal component analysis allows the researcher to find linear combinati ons of Xs so that all principal components are uncorrelated and account for maximum variance in the X's. One of the most advantageous aspects of principal component analysis is that it solves the problem of multicollinearity without dropping variables and losing information. The caveat in using principal component analysis is
71 that, although the principal component defines th e true dimensionality of the data, the principal component may not be meaningful (Johnson 19 98). The principal component analysis was performed using the correlation matrix which can be applied when the measurement scale is consistent across items. For audit interns, th e total variance for the ten associative items was eigenvalue=5.034 (50 percent of the total variance of the original data). The total variance for the nine bisociative items was eigenvalue=3.080 (34% of the total variance of the original data). For internal auditors, the total variance for the ten items accounted for by the principal component was eigenvalue=3.995 (40 percent of the total varian ce of the original data). The total variance for the nine bisociative items was eigenvalue=3.096 (34% of the total variance of the original data). With the exception of associative principal com ponent being negatively associated with Fraud Novelty (Pearsons r = -.243, p-value < .05) for au dit interns, both the asso ciative and bisociative principle components were not significantly asso ciated with other dependent variables. Additionally, both principal components were not significantly effective in subsequent model analyses. 4.8 Manipulation Checks 4.8.1 Interaction Mode One dichotomous measure was used to de termine if participants recognized the interaction mode (anonymous or non-anonymous). Pa rticipants were asked to respond Yes, Dont Know, or No on a 3-point scale to Were you told that you were in an anonymous group, where your team members could not determine which ideas you submitted? Participants response was satisfactory, where 80 percent of the audit inte rns responded and 83 percent of the internal auditors responded correctly.
72 4.8.2 Paradigm-Modifying Creativity Training To assess whether participants understood th at they received brainstorming training, participants were asked if they received trai ning on a brainstorming technique that involved fantasizing, using a dichotomous response measure of yes or no. Participants response was less than satisfactory, where 71 percent of the audit interns responded correctly and 67 percent of the internal auditors responded correctly. 4.9 Test of Hypotheses H1 though H3 The research model includes two categorical variables (interaction modeanonymous or non-anonymous, and creativity trainingyes or no) and one continuous depe ndent variable with three dimensions (quantity, u tility, and novelty). To test each hypothesis, the statistical significance of the MANOVA model was evaluated using multivariate statistics (i.e., Wilks Lambda, Hotelling T2, Pillais statistic). When the overall MANOVA was significant, a series of ANOVAs were performed to draw c onclusions about the hypotheses. In a manner similar to the tea task, this sec tion is divided into several subsections. First, for audit interns, the MANOVA assumptions ar e discussed, followed by the related MANCOVA results. Next, for internal auditors, the as sumptions of MANOVA are discussed followed by related MANOVA results. As discussed in Section 4.10, the participants in this study were randomly assigned to each treatment group and partic ipated in the study independently of others, allowing all observations to be independent of each. Thus, scores by a participant on the dependent measures do not influence the scores of other participants. 4.9.1 Power Analysis Power was analyzed and reported for situations in which the researcher found insignificant mean differences. Analyzing power should provid e some indication as to whether the lack of
73 significance was due to a low sample size or a lo w effect size. SPSS was used to calculate the partial eta square, the noncentrality parameter, and the observed power. The ANCOVA models, at a significance level of = .05, sample size of 70, and medium effect size, for Fraud Quantity, Fraud Novelty, and Fraud Usefulness were analyzed for power. First starting with audit interns and the fraud t ask, the overall observed power for Fraud Quantity was .935 (partial eta squared =.240, noncentrality parameter =20.23), which is considered high power. However, the main effect of creativity training on Fraud Quantity was not significant (pvalue = .257). The observed power was low at .204 (partial eta squared=.020, noncentrality parameter =1.311). Finally, the interaction term was not supported (p-value = .232), with an observed power of .221 (partial eta squared = .022, noncentrality parameter = 1.455). Both observed power levels indicate that there was less than a 22 percent chance that a significant difference would have been found using the treatments. The overall observed power for Fraud Usefulness was high at .886 (partial eta squared = .212, noncentrality parameter = 17.214). The ma in effect of interaction mode on Fraud Usefulness was insignificant (p-value =.285). The observed power was .186 (partial eta squared = .018, noncentrality parameter = 1.16), indicating low power of detecting a significant difference given the sample and the effect size. The interaction term was also insignificant for Fraud Usefulness (p=.816). The power r esult was low at .056 (partial eta squared = .001, noncentrality parameter = .055). Finally, the overall observed pow er for Fraud Novelty was high at .991 (partial eta squared = .316, noncentrality parameter = 29.56) For internal auditors, the ANOVA models, at a significance level of = .05, sample size of 85, and a medium effect size, for Fraud Quan tity, Fraud Novelty, and Fraud Usefulness were analyzed for power. Given that the ANOVA model fo r each dependent variable was insignificant, power analysis was conducted at the model leve l. The overall observed power was low for all dependent variables. Specifically, for Fraud Quantity, the power was .418 (partial eta
74 squared = .058, noncentrality parameter = 4.96). For Fraud Novelty, the power was .423 (partial eta squared = .58, noncentrality parameter = 5.02), and for Fraud Usefulness, the power was .212 (partial eta squared = .028, noncentrality parameter = 2.350). 4.9.2 Audit interns and Assumptions of MANOVA 22.214.171.124 Nature of Distribution The assumption under MANOVA is that all va riables are multivariate normal. However, there is no direct test of multivariate normality (Hair et al. 1998). Instead, the researcher must rely on univariate normality, where each dependent variab le is reviewed individually across treatment groups. To test for univariate normality, the histogram was visually examined, the stem and leaf plots were examined for extreme observations and box plots were examined for outliers. Additionally, Kolmogorov-Smirnov (K S-statistic) with Lilliefors correction was the statistical test used to test for normality. Fraud Novelty Quantity, and Usefulness were reviewed for normality across treatment groups. In the no traini ng/no anonymity group, all but one dependent variable violated the normality assumption (KSstatistic p-value=.082, p-value=.200, and pvalue=.074 for Fraud Quantity, Fraud Novelty, an d Fraud Usefulness, respectively). The box plot and stem and leaf plots for the three dependent variables revealed two extreme outliers. After deleting these two outliers, the box plot did not id entify additional outliers for this treatment group. For the no training/anonymity treatment group, four outliers identified by the stem and leaf and box plots were deleted. The KS-statistic p-value was .166 for Fraud Quantity, .200 for Fraud Novelty, and .031 for Fraud Usefulness. All depe ndent distributions met the univariate normality statistical test except Fraud Usefulness. One a dditional outlier was deleted from the training/no anonymity treatment group.
75 126.96.36.199 Equality of Variance-Covariance Matrices Using only the population of audit interns, the assumption of equality of variancecovariance was violated for the dependent vari ables Fraud Novelty and Usefulness (F=.8.400, pvalue=.000 for Fraud Novelty a nd F=3.446, p-value=.022 for Fr aud Usefulness). A subsequent multivariate Levene test revealed insignifican t results (F=1.759, p-value=.164), thus the assumption of equal variance-covariance is satis fied. Additionally, because the cell sizes are approximately equal, MANOVA is robust to departures from this assumption. 188.8.131.52 Fraud Task Test Results for Audit interns H1 predicted that brainstorming effectiv eness in a computer-mediated brainstorming session among members of a hierarchical audit team would be higher for members interacting anonymously compared to members interacting non-anonymously. Table 22 presents the results of the subsequent ANCOVA test. A MANCOVA w as run controlling for the mean score of postevaluation apprehension and the mean score of social presence, and with the independent variables and their interaction. The dependent va riables were Fraud Quantit y, Fraud Novelty, and Fraud Usefulness. After controlling for the mean score of evaluation apprehension and the mean score of social presence, the main effect of interaction mode (Pillais Trace=.200 F=5.160, pvalue < .01) was significant. A univariate F-test found a significant effect on both Fraud Quantity (F=4.492, p-value < .05) and Fraud Novelty (F=12.999, p-value < .01). Although there was a significant mean difference, it was not in the h ypothesized direction. Both Fraud Quantity and Fraud Novelty were significantly higher for memb ers interacting non-anonymously compared to members interacting anonymously. Thus, H1 is unsupported. H2 predicted that brainstorming effectives in a computer-mediated brainstorming session among members of a hierarchical audit team would be higher for auditors receiving training in a paradigm-modifying creativity technique compared to staff auditors receiving no training. Table 22 presents the statistical results. After controlli ng for the mean score of evaluation apprehension
76 and the mean score of social presence, the main effect of creativity training (Pillais Trace=.154, F=3.753, p-value < .05) was significant. A univa riate F-test found a moderate effect on Fraud Novelty (F=3.556, p-value < .10) and a significan t effect on Fraud Usefulness (F=8.177, p-value < .01). As predicted, both Fraud Novelty and Fraud Usefulness were significantly higher for members who were trained to use the creativity technique compared to members who were not trained. Thus, both H2b and H2c were supported. H3 predicted that the effect of creativity training on brainstorming effectiveness will be greater when the interaction mode is anon ymous. H3 was unsupported (see Table 22). The multivariate tests resulted in the interaction te rm not adding to the model (Pillais Trace=.079, F=1.782, p-value = .160). TABLE 22 ANALYSIS OF COVARIANCE FOR FRAUD QUANTITY, FRAUD NOVELTY, AND FRAUD USEFULNESS FOR AUDIT INTERNS DF Fraud Quantity F-Statistic Fraud Novelty F-Statistic Fraud Usefulness F-Statistic Main Effect Interaction Mode 1 4.492** 12.999*** 1.164 Training 1 1.311 3.556* 8.177*** Interaction Training x Interaction Mode 1 1.455 5.516 .055 Covariate Mean Evaluation Apprehensionb 1 9.185*** 6.141** .042 Mean Social Presencec 1 5.678 3.524 9.905 Error 64 2.896a .143a .521a ***p<.01, **p<.05, *p<.10 aMean square error bMean evaluation apprehension re presents the mean score on fo ur post-experimental items.
77 FIGURE 4 FRAUD QUANTITY PLOT FOR MAIN EFFECT OF INTERACTION MODE-AUDIT INTERNS 2.5 3 3.5 4 No AnonymityAnonymityMean Fraud Quantity FIGURE 5 FRAUD NOVELTY PLOT FO R MAIN EFFECT OF INTERACTION MODE AUDIT INTERNS 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 No AnonymityAnonymityMean Fraud Novelty
78 FIGURE 6 FRAUD NOVELTY PLOT FOR MAIN EFFECT OF CREATIVITY TRAINING AUDIT INTERNS 0.25 0.3 0.35 0.4 0.45 0.5 0.55 No TrainingTrainingMean Fraud Novelty FIGURE 7 FRAUD USEFULNESS PLOT FOR MAIN EFFECT OF CREATIVITY TRAINING AUDIT INTERNS 1.5 2 2.5 3 No TrainingTrainingMean Fraud Usefulness
79 4.9.3 Internal Auditors and Assumptions of MANOVA 184.108.40.206 Nature of the Distribution To satisfy the assumption of normal distribution of dependent variables across treatments, a total of five extreme outliers were deleted. Three outliers were deleted from the no training/no anonymity treatment group, with resulting KS-s tatistic=.136, p-value=.200 for Fraud Quantity, KS-statistic=.158, p-value=.092 for Fraud Novelty, and KS-statistic=.152, p-value=.127 for Fraud Usefulness. Thus, all dependent variables were normal except Fraud Novelty. Although a review of the histogram for Fraud Novelty appeared to look positively skewed, the skewness-statistic was 1.124 which is acceptable (Hair et al. 1998). Ku rtosis-statistic for Fraud Novelty was .428. No extreme variables were identified in the training/anonymity treatment group, although Fraud Usefulness did not meet the normality assumption according to the KS-statistic (KSstatistic=.244, p-value=.003 for Fraud Useful ness, KS-statistic=.157, p-value=.200 for Fraud Novelty, and KS-statistic=.143, p-value=.200 for Fr aud Quantity). For the no training/anonymity treatment, only Fraud Quantity was normally distri buted (KS-statistic=.110, p-value=.200), while Fraud Novelty (KS-statistic=.197, p-value=. 032) and Fraud Usefulness (KS-statistic=.234, pvalue=.004) were not normally distributed accordi ng to the tests of normality. After deleting one additional extreme outlier found in the training/no anonymity treatment, the KS-statistic for Fraud Quantity remained significant (KS-statistic=.202, p-value=.051), while the KS-statistic for Fraud Novelty (KS-statistic=.154, p-value=.200) and Fraud Usefulness (KS-statistic=.174, pvalue=.156) became insignificant. In most cases, the normal distribution was satisfied. MANOVA is fairly robust to departures of normality when the reasons for violations are not due to outliers (Hair et al. 1998).
80 220.127.116.11 Equality of Variance-Covariance Matrices To test the assumption of equality of variance-covariance matrices across treatment groups, both a univariate analysis and a multivariate analysis using the principal component were conducted. Although the univariate Levene test resulted in only Fraud Novelty meeting this assumption (F=.575, p-value=.633), using the principal component, the dependent variables met the assumption of equal variance-covariance across treatment groups (F=1.778, p-value=.158). 18.104.22.168 Fraud Task Test Results for Internal Auditors H1 through H3 were unsupported. The multiv ariate main effects for training (Pillais Trace=.013, F=.352, p-value=.788), interaction m ode (Pillais Trace=.031, F=.841, p-value=.475) and their interaction (Pillais Trace=.066, F=1.847, p-value=.146) were insignificant in the model. When the multivariate tests indicate that the va riables do not add to the model, subsequent univariate analysis cannot be interpreted. Hypoth esis testing results are summarized in Table 23.
81 TABLE 23 SUMMARY OF FINDINGS REPORTED ALPHA LEVEL Format of table, adopted fr om Venkatesh et al. (2003) Hypothesis Number Dependent Variable Independent Variables Covariates Supported Unsupported Directionally1 Reference H1a Fraud Quantity IM2 Post Evaluation Apprehension Social Presence .05 Table 22 Figure 4 H1b Fraud Novelty IM Post Evaluation Apprehension Social Presence .01 Table 22 Figure 5 H1c Fraud Usefulness IM Social Presence H2a Fraud Quantity CT Post Evaluation Apprehension Social Presence H2b Fraud Novelty CT Post Evaluation Apprehension Social Presence .10 Table 22 Figure 6 H2v Fraud Usefulness CT Social Presence .01 Table 22 Figure 7 H3a Fraud Quantity IM x CT Post Evaluation Apprehension Social Presence H3b Fraud Novelty IM x CT Post Evaluation Apprehension Social Presence H3v Fraud Usefulness IM x CT Social Presence 1Although a significant difference in the means, it was not in the predicted direction 2IM=Interaction Mode; 3CT=Creativity Training 4.10 Additional Analysis 4.10.1 Manipulation Check Questions Revisited The interaction mode was intended to aff ect the degree of evaluation apprehension. Specifically, participants in the anonymity c ondition should have experienced less evaluation apprehension than those who did not have anonym ity. Thus, to establish the effectiveness of the anonymity treatment, multivariate tests were conducted using only those participants who answered the interaction mode manipulati on check question correctly. Tables 26 and 27 summarize the analysis for both audit interns and internal auditors.
82 For audit interns who answered the interac tion mode correctly (n=56), the multivariate analysis test showed significance on the mean score of social presence (Pillais Trace=.158, F=3.070, p-value < .05), the main effect of traini ng (Pillais Trace= .192, F=3.873, p-value < .05) and interaction mode (Pillais Trace= .272, F= 6. 093, p-value < .01), and the interaction term (Pillais Trace= .179, F= 3.551, p-value < .05). Fo r audit interns (Panel A), after controlling for the mean score of social presence, the main eff ect of training was significant on Fraud Novelty, the main effect of interaction mode was signi ficant on Fraud Quantity and Fraud Novelty, and their interaction was significant on Fraud Novelt y. The quantity of fraud ideas was significantly greater for audit interns who lacked anonymity than for those who did not. Fraud Novelty was significantly higher for those who received traini ng and had anonymity. Multivariate test were insignificant for internal auditors (n=71). Table 24 summarizes these findings. Compared to the initial analysis, the findings are slightly different. For audit interns (Table 24, Panel A), the main effect of interaction mode remained significant for Fraud Quantity. For Fraud Novelty, although the main effect of interaction mode remained significant, the interaction term was also significant for Fraud Novelty. Thus, Fraud Novelty was significantly higher for those who received training and had anonymity. Although the main effect of creativity training remained significant for Fraud Novelty, the interaction term (interaction mode x creativity training) was significant for Fraud Novelty. Thus, the main effect of creativity trai ning is not interpreted without knowing the type of interaction mode. Creativity training did not remain significant for Fraud Usefulness, however. Thus, under these conditions of using only pa rticipants who accurately responded to the manipulation for interaction mode, H1 through H3 are unsupported for both audit interns and internal auditors.
83 TABLE 24 ANCOVA FOR FRAUD QUANTITY, FRAUD NOVELTY, AND FRAUD USEFULNESS Excluding Participants who Failed Interaction Mode Manipulation Check Panel A: Audit Interns DF Fraud Quantity F-Statistic Fraud Novelty F-Statistic Fraud Usefulness F-Statistic Main Effect Interaction Mode 1 4.182** 17.628*** .438 Training 1 2.019 7.753*** 4.411 Interaction Training x Interaction Mode 1 1.854 10.754*** .011 Covariate Social Presence 1 4.569** 5.381** 5.039 Error 51 3.495a .115a .564a ***p<.01, **p<.05 aMean square error Panel B: Internal Auditors DF Fraud Quantity F-Statistic Fraud Novelty F-Statistic Fraud Usefulness F-Statistic Main Effect Interaction Mode 1 .093 .390 1.649 Training 1 .337 .003 .370 Interaction Training x Interaction Mode 1 4.624 4.146 .027 Error 67 7.080a .161 a .410 a aMean square error The results for the creativity manipulation question are in Table 25 and Table 26. For those audit interns who answered the creativity tr aining manipulation check correctly (n=49), multivariate tests revealed that after cont rolling for the mean score of post evaluation apprehension, both the main effect of interac tion mode (Pillais Trace= .215, F=3.836, p-value < .05) and the interaction term (Pillais Trace= 142, F= 2.326, p-value < .10) were significant. Fraud Quantity and Fraud Novelty were significan tly different on interaction mode (F=4.966, pvalue < .05 and F=9.567, p-value < .01, respec tively) and on the interaction term (F=3.827, pvalue < .10, F=7.023, p-value < .05, respectively). In the original analysis, the interaction term was insignificant for Fraud Quantity and Fraud Nove lty. Thus, for the current analysis, excluding those who failed the creativity training manipula tion check question, when audit interns received
84 training and were not anonymous, they were more likely to generate the greatest number of ideas and the most novel ideas. Under these conditions, the main effect of training is insignificant and the main effect of interaction mode cannot be in terpreted due to the inte raction term. For audit interns, H1 through H3 are unsupported. For internal auditors (n=56), MANOVA multivar iate tests showed a moderately significant mean difference on both interaction mode (Pillias Trace=.125, F=2.379, p-value < .10) and the interaction term (Pillais Trace=.128, F=2.449, p-value < .10). Subsequent univariate tests revealed that Fraud Usefulness was significantly different on interaction mode (F=6.265, p-value < .05). Contrary to the predicted relationship, internal auditors who di d not receive anonymity (=2.636), generated significantly higher usef ul ideas than those who received anonymity (=2.186). Additionally, Fraud Quantity was signifi cantly higher for internal auditors who were not trained and were non-anonymous. In the origin al analysis, insignificant results were found across all treatments for each dependent vari able. Thus, although H1 through H3 remain unsupported for internal auditors; significant mean differences were found in the opposite direction. TABLE 25 ANCOVA FOR FRAUD QUANTITY, FRAUD NOVELTY, AND FRAUD USEFULNESS Excluding Participants who Failed Crea tivity Training Manipulation Check Panel A: Audit Interns DF Fraud Quantity F-Statistic Fraud Novelty F-Statistic Fraud Usefulness F-Statistic Main Effect Interaction Mode 1 4.996** 9.567*** .949 Training 1 .926 1.796 2.479 Interaction Training x Interaction Mode 1 3.827* 7.023** .450 Covariate Post Evaluation Apprehensionb 1 8.786*** 4.210** .559 Error 44 3.827a .130a .636a ***p<.01, **p<.05, *p<.10 aMean square error bMean evaluation apprehension re presents the mean score on four post-experimental items.
85 Panel B: Internal Auditors DF Fraud Quantity F-Statistic Fraud Novelty F-Statistic Fraud Usefulness F-Statistic Main Effect Training 1 1.186 .380 4.563 Interaction Mode 1 2.199 .114 6.265** Interaction Training x Interaction Mode 1 5.692** 5.665 2.024 Error 52 6.586a .152a .408a ***p<.01, **p<.05, *p<.10 aMean square error TABLE 26 SUMMARY OF FINDINGS REPORTED ALPHA LEVEL Format of table, adopted fr om Venkatesh et al. (2003) Panel A: Excluding Participants who Failed Int eraction Mode Manipulati on Check (Audit Interns only) Hypothesis Number Dependent Variable Independent Variables Covariates Supported Unsupported Directionally1 H1a Fraud Quantity IM2 Social Presence .05 H1b Fraud Novelty IM Social Presence H1c Fraud Usefulness IM H2a Fraud Quantity CT Social Presence H2b Fraud Novelty CT Social Presence H2v Fraud Usefulness CT H3a Fraud Quantity IM x CT Social Presence H3b Fraud Novelty IM x CT Social Presence .01 H3v Fraud Usefulness IM x CT 1Although a significant difference in the means, it was not in the predicted direction 2IM=Interaction Mode; 3CT=Creativity Training
86 Panel B: Excluding Participants who Failed Creativi ty Training Manipulation Check Audit Interns and Internal Auditors (note: alpha for Internal Auditors is shown in parentheses.) Hypothesis Number Dependent Variable Independent Variables Covariates Supported Unsupported Directionally1 H1a Fraud Quantity IM2 Post Evaluation Apprehension H1b Fraud Novelty IM Post Evaluation Apprehension H1c Fraud Usefulness IM (.05) H2a Fraud Quantity CT Post Evaluation Apprehension H2b Fraud Novelty CT Post Evaluation Apprehension H2v Fraud Usefulness CT H3a Fraud Quantity IM x CT Post Evaluation Apprehension .10 (.05) H3b Fraud Novelty IM x CT Post Evaluation Apprehension .05 H3v Fraud Usefulness IM x CT 1Although a significant difference in the means, it was not in the predicted direction 2IM=Interaction Mode; 3CT=Creativity Training Alpha level in parentheses are for internal auditors 4.11 Post Hoc Analysis The analysis thus far has fo cused separately on the two distinctly different pools of participants, i.e., audit interns and internal auditors. Post hoc analyses were conducted to address the question of how audit interns and internal auditors compared in terms of brainstorming performance. A 2 x 2 x 2 factorial design was employed using the factors interaction mode (anonymous or non anonymous), creativity training (yes or no), and population (audit interns or internal auditors), with task complexity, th e mean score of social presence, and associative problem-solving style included as covariates. The results of all the multivariate tests (Wilks Lambda, Pillais Trace, etc.) had the same F-valu e and were significant for creativity training (Pillais Trace=.051, F=2.552, p < .10), population (audit interns and internal auditors) (Pillais Trace=.068, F=3.441, p < .05), a two-way interac tion between interacti on mode and population (Pillais Trace=.077, F=3.962, p < .01) and a thr ee-way interaction between interaction mode, creativity training, and population (Pillais Trace = .071, F=3.596, p < .05),.
87 Subsequent univariate analyses are shown in Table 27. After controlling for the mean score of social presence, when participants received creativity training, they had a higher fraud usefulness score than those participants who did not receive creativity training (F=6.064, p-value <.05). Although the main effect of population was significantly different for Fraud Quantity and Fraud Usefulness, so was its twoand three-way interaction with other independent variables. Thus, the effect of population alone cannot be interpreted. Examining the two-way interaction between interaction mode and population, for Fraud Usefulness, the highest overall mean occurred for internal auditors who lacked anonymity (F=4.793, p-value < .05). The three-way interaction (interaction mode x creativity training x population) was significant for Fraud Quantity and Fraud Novelty. For Fraud Quantity, the highest overall outcome occurred for internal auditors who lack ed training and anonymity. For Fraud Novelty, the highest overall outcome occurred for audit in terns who were trained and lacked anonymity. The implications of these findings are summarized in Chapter 5.
88 TABLE 27 ANCOVA RESULTS FOR FR AUD QUANTITY, FRAUD NOVELTY, AND FRAUD USEFULNESS Including Only Audit Intern s and Internal Auditors DF Fraud Quantity F-Statistic Fraud Novelty F-Statistic Fraud Usefulness F-Statistic Main Effect Interaction Mode 1 3.406 3.772 .012 Training 1 .240 1.409 6.064** Population 1 6.525** .054 4.793** Interaction Interaction Mode x Training 1 .906 .048 .040 Interaction Mode x Population 1 .170 7.274*** 2.817* Training x Population 1 .969 3.470 2.241 Interaction Mode x Training x Population 1 5.707** 10.467*** .458 Covariate Task Complexity 1 26.257*** 7.449*** 2.138 Social Presence 1 6.453** 6.871*** 4.358** Associative Problem-Solving Style 1 5.343** 7.186*** .010 Error 144 4.425a .141a .447a ***p<.01, **p<.05, *p<.10, aMean square error Table 28 combines all participants that were or iginally considered for this study, audit interns, internal auditors, and staff auditors. Th e multivariate tests, after controlling for gender, task complexity, social presence, and associative pr oblem-solving style, revealed the main effect of training (Pillais Trace = .065, F = 3.251, p-value < .05), the main effect of population (Pillais Trace = .068, F = 3.415, p-value < .05), the twoway interaction between interaction mode and population (Pillais Trace = .090, F= 4.637, p-valu e < .01), and the three-way interaction among interaction mode, training, and pop ulation (Pillais Trace = .074, F = 3.778, p-value < .05) to be significant to the model. The subsequent univariate analysis revealed that after controlling for gender, task complexity, social presence, and associative pr oblem-solving, the three-way interaction was significant on Fraud Quantity (F = 5.739, p-value < .05) and Fraud Novelty (F = 11.074, p < .01). For Fraud Quantity, the highest overall outcome occurred for internal auditors who lacked training and anonymity. For Fraud Novelty, the highest overall outcome occurred for audit interns
89 who received training and lacked anonymity. Add itionally, after controlling for gender and social presence, the two-way interaction between inte raction mode and population was significant on Fraud Usefulness (F= 3.260, p-value < .10). For Fraud Usefulness, the highest overall outcome occurred for internal auditors who lacked a nonymity. These findings are consistent with the findings in the main analysis and the findings displayed in Table 27. TABLE 28 ANCOVA RESULTS FOR FR AUD QUANTITY, FRAUD NOVELTY, AND FRAUD USEFULNESS Including All Participants (Audit Interns, In ternal Auditors, and Staff Auditors, n=179) DF Fraud Quantity F-Statistic Fraud Novelty F-Statistic Fraud Usefulness F-Statistic Main Effect Interaction Mode 1 3.264 3.443 .000 Training 1 .111 .460 8.923*** Population 1 6.568** .080 4.757** Interaction Interaction Mode x Training 1 .886 .065 .029 Interaction Mode x Population 1 .191 8.012*** 3.260* Training x Population 1 .897 3.145* 2.718 Interaction Mode x Training x Population 1 5.739** 11.074*** .419 Covariate Gender 1 .481 5.808** 6.537** Task Complexity 1 25.918*** 7.282*** 2.460 Social Presence 1 6.072** 5.955** 5.503** Associative Problem-Solving Style 1 4.849** 5.765** .176 Error 143 4.441a .136a .431a ***p<.01, **p<.05, *p<.10, aMean square error Another analysis was conducted to determine if significant mean differences exist across treatments for staff auditors only. Multivariate te sts results showed insignificant differences prior to and after deleting extreme outliers. Finally, bo th audit interns and staff auditors are employed as external auditors, thus the possibility of combining the two samples was considered. Multivariate tests, after controlling for task co mplexity and associative problem-solving, showed Pillais Trace=.144, F=3.25, p-value < .05. Th e significant mean difference was found on Fraud Novelty (F = 4.203, p-value < .05) and Fraud Quantity (F = 8.805, p-value < .01). Similar to the
90 findings using only audit interns, although the main effect of creative training was insignificant when the brainstorming was not anonymous, both Fraud Quantity and Fraud Novelty were significantly higher compared to participants brainstorming with anonymity. After deleting outliers, treatment cells contained five or fewer st aff auditor participants. Further, the results of using staff auditors alone were insignificant. T hus, it appears that the brainstorming performance of the audit interns is driving the results when audit interns are combined with staff auditors.
91 CHAPTER 5: SUMMARY 5.1 Discussion of the Results This study sought to examine factors that could impact brainstorming effectiveness among members of a hierarchically structured audit t eam. Using audit interns and internal auditors, brainstorming effectiveness, defined as Fraud Quantity, Fraud Novelty, and Fraud Usefulness, was predicted to be affected by interacti on mode (anonymity or no anonymity), paradigmmodifying creativity technique training (guided fantasy training or no training), and their joint effect. H1 hypothesized that Fraud Quantity, Fra ud Novelty, and Fraud Usefulness would be higher for participants who brainstormed anon ymously. This hypothesis was not supported for either audit interns or internal auditors. Contra ry to H1, this study found that the quantity and novelty of ideas generated were greatest for aud it interns who brainstormed without anonymity, although there was no statistical difference between anonymity or a lack of anonymity for the internal auditors. The results of this study may suggest that under certain corporate environments, anonymity is not best for novices. For instance, in one situation you have audit interns, who are clearly part of a corporate culture, where constant pressure to move up or out is prevalent. In another situation, you have intern al auditors who are in an environment where long-term stability in one position is highly likely as long as they are competent in their du ties as an auditor. Thus, given the results of this study, heightened ev aluation apprehension through lack of anonymity could have induced, for audit interns, a p erformance-related pressure to do well. This suggestion is consistent with pr ior accounting literature that h as shown that auditors judgment and decision-making are influenced by the potential to be evaluated (Koonce et al. 1995; Rich et al. 1997). This conclusion is also consistent w ith the Yerkes-Dodson principle which posits an
92 inverted U relationship between pressure a nd performance such that pressure initially increases performance but eventually leads to a declin e in performance (Yerkes and Dodson 1908). The second hypothesis, H2, posited that the qua ntity, novelty, and usefulness of fraud ideas generated would be higher for participants who received creativity training than for those who did not. Creativity training did not significantly incr ease the total number of ideas generated by audit interns or internal auditors. While there was no difference in the total number of novel ideas or usefulness score for internal auditors, th e number of novel ideas and useful ideas was significantly higher for audit interns who received creativity training than for those who did not. The results of this study suggest that this e ffect was dominated by the non-anonymous treatment group who received creativity training, however, the results did not show significant differences for the interaction term. The lack of results for in ternal auditors may be simply due to the limited amount of training they received. Perhaps the training time or the training technique was not sufficient to modify the internal auditors mental schema. While internal auditors have always been responsible for safeguarding corporate assets, audit interns have yet to develop a mental schema for detecting fraud, and thus, may have b een more receptive to the training technique. H3 stated that the effect of training on fraud quantity, novelty, and usefulness would be higher for individuals working anonymously than non-anonymously. This hypothesis was unsupported for both audit interns and internal a uditors. Although, Table 10, Panel A, show that the highest mean occurred for a udit interns in the non-anonymous treatment group who received creativity training, statistically significant findi ngs showed that training alone was sufficient to impact performance for audit interns irrespectiv e of the interaction mode (anonymity or nonanonymity). However, this was not the case for internal auditors since neither training nor interaction mode impacted the outcomes. Internal auditors are more experienced and have a preestablished taxonomy for considering fraud, and it may be that the limited amount of training in this study was insufficient to modify their para digm. This conjecture is supported by research on
93 expertise, which says that experts typically ha ve an established taxonom y and employ heuristic reasoning, which means that experts, through e xperience, develop an intuitive method for solving problems. However, for the audit interns w ho do not have a pre-established taxonomy, the training helped them think creatively about fraud, which improved their fraud brainstorming performance. Although the creativity literature suggests that brainstorming sessions should be free of environmental pressures (Osborn 1957), the results of this study suggest that, under certain conditions, environmental pressures may enha nce performance. Also, given that fraud perpetrators have been known to employ creativ e techniques, auditors way of thinking about fraud must be unpredictable to avoid familiarizati on and predictability of audit procedures. The initial brainstorming session sets the tone of the a udit and affects the audit plan and the level of fraud skepticism. Auditors must not be content w ith the way they currently approach the audit process. 5.2 Contributions SAS No. 99 mandates brainstorming as part of overall fraud risk assessment. This study provides initial evidence regarding factors that may impact the effectiveness of brainstorming sessions designed to more accurately assess the risk of fraud related to the audit engagement. One contribution is that for junior members of a hierarchical team and/or organization, where the norm is to either be promoted or to leave the organization, no anonymity serves to increase evaluation awareness, as opposed to evaluati on apprehension, and improves performance. However, this is not necessarily true when an i ndividuals reputation has already been established or when an individuals job status is not affected by an up or out promotion policy, as in the case of internal auditors. Whether the lack of resu lts for internal auditors is due to level of expertise, team/firm structure, or some other unidentified artifact is unknown and is an area for future research.
94 A second contribution is that this study provides support for the use of creativity training to improve the brainstorming effectiveness for novice staff auditors. Whether the technique can be used effectively for internal auditors or more sen ior external auditors remains an open question. Creativity training for junior auditors or novices is necessary given that they are the eyes and ears of the audit team. Thus, their observations and fee dback to senior audit team members is vital to the audit process. Lack of an effect of creativity training for the internal auditors may be due to their level of expertise and reliance on heuristics, or may be due to the amount of time allocated for training. For internal auditors, the limited amount of training may have been insufficient to modify a pre-established paradigm or way of thin king. Whether creativity training can be useful for experts, and/or the conditions under which it is found to be useful for expert auditors, is a question for future research. 5.3 Limitations This study is subject to a number of limita tions. First, multiple se ssions were conducted. Thus, there was the slight potential that earlier pa rticipants communicated with later participants about the true nature of the study. To minimize this internal validity threat, individuals were debriefed only after all subjects participated in the study. Additionally, during the beginning of each session, participants had the opportunity to state whether they have discussed particulars of the study with previous participants. The inter-rater reliability for the Fraud Usefuln ess indicated that the audit managers were inconsistent in their rating (Cohens Kappa= .500). The Fraud Usefulness instrument was developed specifically for this study, and thus, lacked prior empirical support. Although Fraud Usefulness was defined for the audit managers, th e extent to which the audit managers relied on the specified definition or some other form of usefulness was not determined. Future research should determine what is meant by the term use fulness. How this term is defined is important because the definition will impact which ideas are co nsidered in the audit process. Also, because
95 the scale lacked empirical support, the researcher was faced with the decision of what type of scale would be best to rate Fraud Usefulness. Future research applying a measure of usefulness should look cautiously at the appropriate way to measure the concept, that is, once the term usefulness has been clearly defined. Given the low power of the statistical test, ther e is a chance that due to the effect size (the relationship between the variables), the sample si ze, or both, that this study failed to find a significant mean difference on a variable, when one exists. This is known as a Type II error. The lower the power, the higher the likelihood of a Type II error (Keppel and Wickens 2004). Type II error for this study could have been controlled without increasing the risk of a Type I error by increasing the number of participants in each cell and by increasing the effect size through alternate methods of making the creativity traini ng and anonymous interaction treatments more salient. Future research should consider both of these options for increasing power. Auditors rarely encounter fraud (Palmrose 1987; Pincus 1989; Hackenbrack 1992; Bell and Carcello 2000; Erickson et al. 2000; Nieschwietz et al. 2000). Thus, results are limited to the performance of audit interns and cannot be genera lized to those individuals who have had actual fraud experience. The use of audit interns as su rrogates for staff auditors warrants discussion, particularly for external validity purposes. Intern al validity speaks to the experimental realism of the study, while external validity refers to gene ralization of findings to a targeted population and setting. Internal validity is a necessary conditi on for external validity (Pedhazur and Schmelkin 1991). External validity, at the expense of internal validity, may have been jeopardized with the use of audit interns instead of staff auditors. Evaluation apprehension of junior auditors would be difficult to study in a controlled environment. As experienced by Schultz and Hooks (1998), including staff auditors in the current study was both difficult and costly. Staff auditors were rarely available in large numbers at one location, since they were dispersed to various field locations. The use of audit interns allowed
96 evaluation apprehension to be salient with the u se of phantom team members who were superior to the audit interns. Given that evaluation appreh ension was a necessary component of this study, audit interns enhance the realism, which is an im portant aspect of internal validity. For example, names of phantom team members were likely to be more believable to audit interns than by staff auditors. According to Gibb ins (1984), staff auditors or auditors in general are likely to be aware of the fact that they are participating in an experiment, and thus unlikely to be affected by deception techniques in experiments. The experi mental outcome relied heavily on the deception by having participants be lieve they were part of an actual brainstorming session with superiors. Also, given that the topic of brainstorming among audit team members is a new research area in the accounting profession, the use of audit interns provides insight into the performance of junior auditors when evaluation apprehension is likely to be present. Thus, although external validity may be limited, the use of audit interns as surrogates for staff auditors was necessary in order to achieve experimental realism, thus enha ncing internal validity. Schultz and Hooks (1998) make a compelling argument that can be applied to this study. First, few, if any, studies have reviewed the audit team in a hierarchical structur e where it is important to simulate a hierarchical audit team structure and explore the performan ce of the junior member. Additionally, although audit interns had never encountered fraud, they were likely to be familiar with ways employees could misappropriate company assets. It was obvious that audit interns had acquired the conceptual meaning of fraud. Audit interns genera ted ideas similar to those of internal auditors. Thus, it is hoped that the findings for audit intern s, who were carefully recruited by one of the Big Four international CPA firms, can help us gain insight on the brainstorming performance of newly hired staff auditors, who are often exposed to their first audit immediately after being hired. These findings associated with audit intern s cannot be generalized to staff auditors who have a college degree, on-the-job training certifications, and auditing experience.
97 5.4 Future Research Overall, although creativity training can be beneficial, a lack of anonymity may be beneficial to the brainstorming process when te am members are constantly auditioning for the next level. It is possible that evaluation awareness occurs on the upside, and evaluation apprehension occurs on the downside. Unlike partic ipants used in psychology literature, auditing is a unique profession in that there are consequen ces for not doing a job well the first time. Thus, future research can determine if the findings on an onymity hold true for a ll levels of external audit positions. Anonymity may be unnecessary for al l external auditors, where failure to do well and receive recognition for accomplishments come at a high price. Phantom members were used in the current st udy, and thus, participants did not receive feedback on their ideas. Results may have been different, especially given that audit interns experienced a higher degree of evaluation apprehensi on than internal auditors, if negative versus positive feedback was provided from superior team members. While it would be difficult to create a believable simulation using phantom me mbers to provide negative/positive feedback, future research could employ confederates playin g the role of senior auditors providing either positive or negative feedback during the brainstorming session. Internal auditors have typically been assi gned the task of safeguarding company assets. Thus, it is natural to expect internal auditors to have more experience at generating ideas about employee fraud than audit interns. Given this level of expertise for internal auditors, it is possible that the creativity training, in general, will not affect brainstorming performance, or perhaps the particular creativity training used in this study cannot effectively be applied to experts who have a well-trained methodology for analyzing fraud. Futu re research should determine whether or not a creative tool that is effective for training novices is also an effective tool for training experts. In other words, some creativity training techniques may be more effective at modifying a preexisting taxonomy or paradigm than other creativity training techniques.
98 Future research could also determine the most appropriate amount of time to allocate to the brainstorming session. Each idea in the current study was time-stamped, and thus it may be informative to further analyze the data to dete rmine when the best brainstorming performance generally occurred for each individual. Anothe r approach would be to manipulate the time allocated for each brainstorming session and determine if the amount of time is important to quantity, quality, and usefulness of fraud ideas. Additionally, this study used a misappropriation of assets case. Future research can examine the same factors using a fraudulent financ ial reporting case. Because this type of fraud is likely to involve revenue recognition matters and internal control overrides, the results may be different for junior auditors, who typically do not deal directly with management and for internal auditors, who may have limited exposur e to fraudulent financial reporting. The Yerkes-Dodson theory posits that various levels of pressure have a positive effect on performance. However, over time, increased leve ls of pressure will become overwhelming and cause performance to suffer. The team structure in this study was held constant. However, future research should consider studying performance unde r increased levels of pressure. For example, during the brainstorming session, it would be inter esting to examine the effect of bringing in another senior auditor all of a suddenwould such an intervention improve or inhibit brainstorming effectiveness? Would such incr eased pressure cause brainstorming performance to decline, per the Yerkes-Dodson principle. These and other research questions are worthy of investigation in this line of research that seeks to shed light on the most effective and e fficient methods of improving auditors ability to detect fraud.
99 REFERENCES Agarwal, R. 2000. Individual acceptance of information technologies. Framing the Domains of IT Management Lithographing, Inc. : 85-104. AICPA 1983. Statement on Auditing Standards (SAS) No. 47: Audit Risk and Materiality in Conducting an Audit American Institute of Certified Public Accountants, New York. AICPA 2002a. Exposure Draft: Proposed St atement of Auditing Standards American Institute of Certified Pub lic Accountants, New York. AICPA 2002b. Statement on Auditing Standards (SAS) No. 99: Consideration of Fraud in a Financial Statement Audit American Institute of Certified Public Accountants, New Jersey. Allport, F. H. 1924. Social Psychology Houghton Mifflin, Boston. Amabile, T. 1996. Creativity in context Westview Press, Boulder, Colo. Amabile, T. M. 1979. Effects of Extern al Evaluation on Artistic Creativity. Journal of Personality and Social Psychology 37 (2): 221-233. Amabile, T. M. 1982. Social-psychology of creativity a consensual assessment technique. Journal of Personality and Social Psychology 43 (5): 997-1013. Amabile, T. M. 1983. The Social Psychology of Creativity Springer-Verlag,New York. Anderson, R. J. 1977. The External Audit Pitman, Toronto. Ashton, R. H. and J. Kennedy. 2002. Eliminating recency with self-r eview: The case of auditors' 'going concern' judgments. Journal of Behavioral Decision Making 15 (3): 221-231. Bamber, E. M. 2002. The Auditor's Report 25: 1-26. Bamber, E. M., M. C. Hill, R. T. Watson. 1998. Audit groups and group support systems: A framework and propositions for future research. Journal of Information Systems 12 (2): 45-74.
100 Bamber, E. M., R. T. Watson, M. C. Hill. 1996. The effects of group support system technology on audit group decision making. Auditing-A Journal of Practice & Theory 15 (1): 122-134. Barki, H. and A. Pinsonneault. 2001. Small group brainstorming and idea quality Is electronic brainstorming the most effective approach?" Small Group Research 32 (2): 158-205. Barlow, C. M. 2000. Deliberate insight in team creativity. Journal of Creative Behavior 34 (2): 101-117. Bartis, S., K. Szymanski, S. G. Harkins. 1988. Evaluation and performance A 2-edged knife. Personality and Social Psychology Bulletin 14 (2): 242-251. Bateson, G. 1976. The Psychology of Play : A Book of Readings Arno Press, New York. Beasley, M. S., J. V. Carcello, D. R. Hermanson, P. D. Lapides. 2000. Fraudulent financial reporting: Consideration of industry traits and corporate governance mechanisms. Accounting Horizons 14 (4): 441-454. Bedard, J., M. T. H. Chi, L. E. Graham J. Shanteau. 1993. Expertise in auditing: Discussion. Auditing-A Journal of Practice & Theory 12 (Supplement): 21-45. Bell, T. B. and J. V. Carcello. 2000. A d ecision aid for assessing the likelihood of fraudulent financial reporting. Auditing-A Journal of Practice & Theory 19 (1): 169-184. Berenson, A. 2003. Ex-executive pleads guilty to fraud on sales figures. The New York Times Bond, C. F. and L. J. Titus. 1983. Social Fa cilitation a Meta-Analysis of 241 Studies. Psychological Bulletin 94 (2): 265-292. Bray, J. H. and S. E. Maxwell. 1985. Multivariate analysis of variance Beverly Hills, Sage Publications. Collaros, P. A. and L. R. Anderson. 1969. Eff ect of perceived expertness upon creativity of members of brainstorming groups. Journal of Applied Psychology 53 (2P1): 159. Condry, J. and J. Chambers. 1978. Intrinsic mo tivation and the processing of learning. The Hidden Costs of Reward: New Perspectives on the Psychology of Human Motivation John Wiley & Sons, New York : 61-84.
101 Connolly, T., L. M. Jessup, J. S. Valacic h. 1990. Effects of anonymity and evaluative tone on idea generation in computer-mediated groups. Management Science 36 (6): 689-703. Connolly, T., R. L. Routhieaux, S. K. Sc hneider. 1993. On the effectiveness of group brainstorming Test of one unde rlying cognitive mechanism. Small Group Research 24 (4): 490-503. Cooper, W. H., R. B. Gallupe, S. Pollar d, J. Cadsby. 1998. Some liberating effects of anonymous electronic brainstorming. Small Group Research 29 (2): 147-178. Cottrell, N. B., D. L. Wack, G. J. Sekerak, R. H. Rittle. 1968. Social facilitation of dominant responses by presence of an audience and mere presence of others. Journal of Personality and Social Psychology 9 (3): 245-&. Couger, J. D. 1995. Creative Problem Solving a nd Opportunity Finding Boyd & Fraser Publishing Company, Hinsdale. Couger, J. D. 1996. Creativity and Innovation Boyd & Fraser Publishing Company, Danvers. Danner, P. 2000. Old fraud, new medium. Broward Daily Business Review Ft. Lauderdale : B9. Dansky, J. 1999. Play. Encyclopedia of Creativity Academic Press, San Diego. 2: 393407. Dennis, A. R., K. M. Hilmer, et al. 1998. Information exchange and use in GSS and verbal group decision making: E ffects of minority influence. Journal of Management Information Systems 14 (3): 61-68. Dennis, A. R., B. H. Wixom, R. J. Vandenberg. 2001. Understanding fit and appropriation effects in group suppo rt systems via meta-analysis. MIS Quarterly 25 (2): 167-193. Diehl, M. and W. Stroebe. 1987. Productivity lo ss in brainstorming groups Toward the solution of a riddle. Journal of Personality and Social Psychology 53 (3): 497-509. Eisenberger, R., F. Haskins, P. Gambleton. 1999. Promised reward a nd creativity: Effects of prior experience. Journal of Experiment al Social Psychology 35 (3): 308-325. Erickson, M., B. W. Mayhew, W. L. Feli x. 2000. Why do audits fail? Evidence from lincoln savings and loan. Journal of Accounting Research 38 (1): 165-194. Forsyth, D. R. 1990. Group Dynamics Brooks/Cole Pub. Co., Pacific Grove, Calif.,
102 Gagne, M. and M. Zuckerman. 1999. Performa nce and learning goal orientations as moderators of social loafi ng and social facilitation. Small Group Research 30 (5): 524-541. Garfield, M. J., N. J. Taylor, A. R. Denni s, J. W. Satzinger. 2001. Research report: modifying paradigms Individual differen ces, creativity techniques, and exposure to ideas in group idea generation. Information Systems Research 12 (3): 322-333. Gibbins, M. 1984. Propositions about the psycho logy of professional judgment in public accounting. Journal of Accounting Research 22 (1): 103-125. Gough, H. G. 1979. Creative personality sc ale for the adjective check list. Journal of Personality and Social Psychology 37 (8): 1398-1405. Guilford, J. P. 1950. Creativity. American Psychologist 5 : 444-454. Hackenbrack, K. 1992. Implications of seemingl y irrelevant evidence in audit judgment. Journal of Accounting Research 30 (1): 126-136. Hair, J. F., R. E. Anderson, R. L. Tatham, W. C. Black. 1998. Multivariate Data Analysis Prentice Hall, Englewood Cliffs, NJ. Hedlund, J., D. R. Iigen, J. R. Hollenb eck. 1998. Decision accuracy in computermediated versus face-to-face decision-making teams. Organizational Behavior and Human Decision Processes 76 (1): 30-47. Henchy, T. and D. C. Glass. 1968. Evaluation apprehension and soci al facilitation of dominant and subordinate responses. Journal of Personality and Social Psychology 10 (4): 446-&. Hender, J. M., D. L. Dean, T. L. Rodgers, J. F. Nunamaker. 2002. An examination of the impact of stimuli type and GSS struct ure on creativity: Brainstorming versus nonbrainstorming techniques in a GSS environment. Journal of Management Information Systems 18 (4): 59-85. Hilzenrath, D. S. 2002. Pitt wants special audi ts to find fraud; SEC readies rules for accounting. The Washington Post Washington, D.C. : E.01. Hirst, D. E. and L. Koonce. 1996. Audit anal ytical procedures: A field investigation. Contemporary Accounting Research 13 (2): 457-486. Isaksen, S. G. 1987. Introduction An orientation to the frontiers of creativity research. Frontiers of creativity research: Beyond the basics Bearly Limited, Buffalo : 131.
103 Jabri, M. M. 1991. The development of con ceptually independent subscales in the measurement of modes of problem-solving. Educational and Psychological Measurement 51 (4): 975-983. Jamal, K. and H. T. Tan. 2001. Can auditors predict the choices made by other auditors?" Journal of Accounting Research 39 (3): 583-597. Jessup, L. M., T. Connolly, J. Galegher. 1990. The effects of anonymity on GDSS groupProcess with an idea-generating task. MIS Quarterly 14 (3): 313-321. Jessup, L. M. and D. A. Tansik. 1991. Deci sion-making in an automated environment the effects of enonymity and proximity with a group decision support system. Decision Sciences 22 (2): 266-279. Johnson, C. and B. A. Masters. 2003. Prosecu tors, regulators step up pace of auditor probes; in effort to avert collapse of more accounting firms, officials shift focus to people who commit misdeeds. The Washington Post Washington, D.C. : E.01. Johnson, D. E. 1998. Applied Multivariate Methods for Data Analysts Duxbury, Belmont, CA. Johnson, E. N. 1994. Auditor memory for audit evidence Effects of group assistance, time-delay, and memory task. Auditing-A Journal of Practice & Theory 13 (1): 3656. Karan, V., D. S. Kerr, U. S. Murthy, A. S. Vinze. 1996. Information technology support for collaborative decision making in audi ting: An experimental investigation. Decision Support Systems 16 (3): 181-194. Keppel, G., and T. Wickens. 2004. Design and Analysis: A Researcher's Handbook Prentice Hall, Englewood Cliffs, NJ. Kirton, M. 1976. Adaptors and innovat ors Description and measure. Journal of Applied Psychology 61 (5): 622-629. Kline, R. B. 1998. Principles and Practice of St ructural Equation Modeling New York, Guilford Press. Koestler, A. 1964. The Act of Creation New York, Macmillan. Koonce, L., U. Anderson, C. Marchant. 1995. Justification of decisions in auditing. Journal of Accounting Research 33 (2): 369-384.
104 Kozloski, T. M. 2002. The influence of simila rity and risk on knowledge transfer from prior fraud risk assessments. Accounting and Tax Drexel University, Philadelphia : 192. Lamont, I. 2000. The coolest kind of collaboration network world. Framingham 17 (46): 117-120. Lowe, D. J., P. M. J. Reckers, S. M. Wh itecotton. 2002. The effects of decision-aid use and reliability on jurors' eval uations of auditor liability. The Accounting Review 77 (1): 185-202. McGraw, K. O. 1978. Detrimental effect on performance. The Hidden Costs of Reward: New Perspectives on the Psychology of Human Motivation John Wiley & Sons, New York : 31-60. Mumford, M. D. and D. K. Simonton. 1997. Creativity in the workplace: People, problems, and structures. Journal of Creative Behavior 31 (1): 1-6. Murthy, U. S. 2002. Group support systems rese arch in accounting: A theory-based framework and directions for future research. Researching Accounting as an Information Systems Discipline American Accounting A ssociation, Sarasota, FL : 326. Nagasundaram, M. and R. P. Bostrom. 1995. Th e structuring of crea tive processes using gss: a framework for research. Journal of Management Information Systems 11 (3): 89-116. NCFFR. 1987. Report of the National Commissi on on Fraudulent Financial Reporting. American Institute of Certified Public Accountants, New York. Nieschwietz, R. J., J. J. Schultz, M. F. Zimbelman. 2000. Empirical research on external auditors' detection of financial statement fraud. Journal of Accounting Literature 19 : 190-246. Nunamaker, J., R. O. Briggs, D. D. Mitt leman, D. R. Vogel, P. A. Balthazard. 1997. Lessons from a dozen years of group suppor t systems research: a discussion of lab and field findings. Journal of Management Information Systems 13 (3): 163 207. Nunamaker, J. F., A. R. Dennis, J. S. Valacich, D. R. Vogel 1991a. Information technology for negotiating groups Generating options for mutual gain. Management Science 37 (10): 1325-1346. Nunamaker, J. F., A. R. Dennis, J. S. Valacich, D. R. Vogel, J. F. George. 1991b. Electronic meeting systems to support group work. Communications of the ACM 34 (7): 40-61.
105 Nunamaker, J. F., A. R. Dennis, J. S. Va lacich, D. R. Vogel, J. F. George. 1993. Group support systems research at arizona: experience from the lab and field. Group Support Systems: New Perspectives Macmillan Publishing Company, New York : 125-145. Osborn, A. F. 1957. Applied imagination; principles and procedures of creative problemsolving Scribner, New York. Osborn, A. F. 1963. Applied imagination Scribner, New York. Palmrose, Z. V. 1987. Litigation and independent auditors The role of business failures and management fraud. Auditing-A Journal of Practice & Theory 6 (2): 90-103. Pedhazur, E. J. and L. P. Schmelkin. 1991. Measurement, Design, and Analysis Lawrence Erlbaum Associates, Inc., Hillsdale, NJ, Phillips, J. M. 2001. The role of decision in fluence and team performance in member self-efficacy, withdrawal, satisfaction with the leader, and willingness to return. Organizational Behavior and Hu man Decision Processes 84 (1): 122-147. Phillips, J. M. 2002. Antecedents and consequen ces of procedural justice perceptions in hierarchical decision-making teams. Small Group Research 33 (1): 32-64. Piaget, J. 1962. Play, Dreams, and Imitation in Childhood Norton, New York. Pincus, K. V. 1989. The efficacy of a red flag s questionnaire for assessing the possibility of fraud. Accounting, Organizations and Society 14 : 153163. Pinsker, R. E. 2002. The impact of con tinuous business reporting on stock price valuation, volatility and variation. University of South Florida Pinsonneault, A. and N. Heppel. 1998. Anonym ity in group support systems research: A new conceptualization, measure, and contingency framework. Journal of Management Information Systems 14 (3): 89-108. Pinsonneault, A. and K. L. Kraemer. 1989. The impact of technological support on groups: an assessment of the empirical research. Decision Support Systems 5 (Special Issue on Group Decision Support Systems): 197-216. Platania, J. and G. P. Moran. 2001. Social faci litation as a function of the mere presence of others. Journal of Social Psychology 141 (2): 190-197. Plitch, P. 2003. When market scandals erupt, regulation can come in a flood. Wall Street Journal
106 Pulliam, S. P. and J. Bandler. 2003. Leading the news: KPMG is likely to face fraud charge --SEC set to file civil action ag ainst accounting firm for role in xerox audits. Wall Street Journal New York, N.Y. : A.3. Rhodes, M. 1961. An analysis of creativity. Phi Delta Kappan 42 (7): 305-310. Rich, J. S., I. Solomon, K. T. Trot man. 1997. The audit review process: A characterization from the persuasion perspective. Accounting Organizations and Society 22 (5): 481-505. Romney, M. B. and P. J. Steinbart. 2002. Accounting Information Systems Pearson Education, Inc., Upper Saddle River, NJ. Rothenberg, A. and C. Hausman. 1976. The Creativity Question Duke University Press Durham, N.C. Russ, S. W., A. L. Robins, B. A. Christia no. 1999. Pretend play: Longitudinal prediction of creativity and affect in fantasy in children. Creativity Research Journal 12 (2): 129-139. Santanen, E. L. 2002. Directed brainstorm ing and the Cognitive Network Model of Creativity: An empirical investigation of c ognitive factors related to the formation of creative solutions using an elec tronic brainstorming environment. Management Information Systems University of Arizona. Satzinger, J. W., M. J. Garfield, M. Nagasundaram. 1999. The creative process: The effects of group memory on i ndividual idea generation. Journal of Management Information Systems 15 (4): 143-160. Saunders, C. S. 2000. Virtual teams: Piecing together the puzzle. Framing the Domains of IT Management Pinnaflex Education Resour ces, Inc., Cincinnati, OH : 29-50. Schultz, J. J. and K. L. Hooks. 1998. The eff ect of relationship and reward on report of wrongdoing. Auditing-A Journal of Practice & Theory 17 (2): 15-35. Scott, S. G. and R. A. Bruce. 1994. Determin ants of innovative behavior A path model of individual innovation in the workplace. Academy of Management Journal 37 (3): 580-607. Shalley, C. E. 1995. Effects of coaction, expected evaluation, and goal-setting on creativity and productivity. Academy of Management Journal 38 (2): 483-503.
107 Shalley, C. E. and G. R. Ol dham. 1985. Effects of goal difficulty and expected external evaluation on intrinsic motivation A laboratory study. Academy of Management Journal 28 (3): 628-640. Shalley, C. E. and J. E. Perry-Smith. 2001. E ffects of social-psychological factors on creative performance: The role of in formational and controlling expected evaluation and modeling experience. Organizational Behavior and Human Decision Processes 84 (1): 1-22. Shelton, S. W., O. R. Whittington, D. La ndsittel. 2001. Auditing firms' fraud risk assessment practices. Accounting Horizons 15 (1): 19-33. Short, J., E. Williams, B. Christie. 1976. The social psychology of telecommunications Wiley, New York. Sia, C. L., B. C. Y. Tan, K. K. Wei. 2002. Group polarization and computer-mediated communication: Effects of communication cues, social presence, and anonymity. Information Systems Research 13 (1): 70-90. Solomon, I. 1982. Probability assessment by i ndividual auditors and audit teams: An empirical investigation. Journal of Accounting Research 20 (2): 689-709. Solomon, I. 1987. Multi-auditor judg ment/decision making research. Journal of Accounting Literature 6 : 1-25. Sosik, J. J., S. S. Kahai, B. J. Avolio. 1999. Leadership style, anonym ity, and creativity in group decision support systems: The mediating role of optimal flow. Journal of Creative Behavior 33 (4): 227-256. Stepanek, M. 1999. Using the net for brainstorming. Business Week Strand, C. A., S. T. Welch, S. A. Holms, S. L. Judd. 2002. Developing student abilities to recognize risk factors: a series of scenarios. Issues in Accounting Education 17 (1): 57-67. Taggar, S., R. Hackett, S. Saha. 1999. Leader ship emergence in autonomous work teams: Antecedents and outcomes. Personnel Psychology 52 (4): 899-926. Torrance, E. P. 1974. Torrance tests of creative thinking Scholastic Testing Service m Bensenville, IL. Triplett, N. 1897. The dyanamogenic fact ors in peacemaking and competition. American Journal of Psychology 9 : 507-533.
108 Tversky, A. and D. Kahneman. 1974. Judgment u nder uncertainty: Heuristics and biases. Science 185 (4157): 1124-1131. Valacich, J. S., A. R. Dennis, J. F. N unamaker. 1992. Group-size and anonymity effects on computer-mediated idea generation. Small Group Research 23 (1): 49-73. VanGundy, A. 1988. Techniques of Structured Problem Solving Van Nostrand Reinhold Company Inc., New York. Venkatesh, V., M. G. Morris, G. B. Davi s, F. D. Davis. 2003. User acceptance of information technology: Toward a unified view. MIS Quarterly 27 (3): 425-478. Vitharana, O. and K. Ramamurthy. 2003. Comp uter-mediated group support, anonymity, and the software inspection proces s: An empirical investigation. IEEE Transactions on Software Engineering 29 (2): 167-180. Weisband, S. P., S. K. Schneider, T. C onnolly. 1995. Computer-mediated communication and social information Status salience and status differences. Academy of Management Journal 38 (4): 1124-1151. White, K. and D. Owen. 1970. Locus of evalua tion for classroom work and development of creative potential. Psychology in the Schools 7 (3): 292-295. Whittington, R. 2002. ASB Update as of September 15, 2002. The Auditor's Report 25 (4): 1-17. Yerkes, R. M. and J. D. Dodson. 1908. Relation of strength of stimul us to rapidity of habit-formation. Journal of Comparative Neurology and Psychology 18 : 459-482. Yip-Ow, J. and H. T. Tan. 2000. Effects of the preparer's justification on the reviewer's hypothesis generation and judgment in analytical procedures. Accounting Organizations and Society 25 (2): 203-215. Zajonc, R. B. 1965. Social facilitation. Science 149 (3681): 269-&.
110 EXHIBIT 1: FOUR PHASES OF AN INDEPENDENT AUDIT Phase II Perform tests of controls and substantive tests of transactions Phase III Perform analytical procedures and tests of details of balances Phase IV Complete the audit and issue an audit report Primary Focus of Proposal Phase II Perform tests of controls and substantive tests of transactions Phase I Plan and design an audit approach
111 EXHIBIT 2: SUMMARY OF HYPOTHES ES AND RESEARCH QUESTIONS Research Questions: Research Question 1: Does interaction mode us ing a GSS affect the quantity, utility, and novelty of ideas generated by staff auditors? Research Question 2: Does training in a paradi gm-modifying creativity technique improve the quantity, utility, and novelty of id eas generated by staff auditors? Research Question 3: Do interaction mode and cr eativity training jointly affect the quantity, utility, and novelty of ideas generated by staff auditors? Hypotheses: H1: In a computer-mediated brainstorming session among members of a hierarchical audit team, auditors interacting anonymously will be more effective at brainstorming than audito rs interacting non-anonymously. H1a: In a computer-mediated brainstorming session among members of a hierarchical audit team, auditors interacting anonymously will gene rate more fraud ideas than auditors interacting non-anonymously. H1b: In a computer-mediated brainstorming session among members of a hierarchical audit team, auditors interacting anonymously will ge nerate more novel fraud ideas than auditors interacting non-anonymously. H1c: In a computer-mediated brainstorming session among members of a hierarchical audit team, auditors interacting anonymously will generate more useful fraud ideas than auditors interacting non-anonymously. H2: In a computer-mediated brainstorming session among members of a hierarchical audit team, auditors receiving training in a paradigmmodifying creativity technique will be more effective at brainstorming than auditors receiving no creativity training. H2a: In a computer-mediated brainstorming session among members of a hierarchical audit team, auditors in a paradigmmodifying creativity technique will generate more fraud ideas than auditors receiving no creativity training. H2b: In a computer-mediated brainstorming session among members of a hierarchical audit team, auditors in a paradigmmodifying creativity technique will generate more novel than auditors receiving no creativity training.
112 H2c: In a computer-mediated brainstorming session among members of a hierarchical audit team, auditors in a paradigmmodifying creativity technique will generate more useful than auditors receiving no creativity training. H3: The effect of creativity training on brainstorming effectiveness in a computer-mediated brainstorming session will be greater when the interaction mode is anonymous than when it is non-anonymous. H3a: The effect of creativity training on the number of fraud ideas generated in a computer-mediate d brainstorming session will be greater when the interaction mode is anonymous than when it is non-anonymous. H3b: The effect of creativity training on the novelty of fraud ideas generated in a computer-mediate d brainstorming session will be greater when the interaction mode is anonymous than when it is non-anonymous. H3c: The effect of creativity training on the usefulness of fraud ideas generated in a computer-mediated br ainstorming session will be greater when the interaction mode is anon ymous than when it is non-anonymous.
113 EXHIBIT 3: CHARACTERISTICS OF ADAPTORS AND INNOVATORS TAKEN FROM (JABRI 1991) ITEMS FOR INDEPENDENT SUBSCALES: ASSOCIATIVE AND BISOCIATIVE Associative Bisociative Adhering to the commonly established rules of my area of work. Being confronted with a maze of ideas which may, or may not, lead me somewhere. Following well-trodden ways and generally accepted methods for solving problems. Pursuing a problem, particularly if it takes me into areas I dont know much about. Being methodical and consistent in the way I tackle problems. Linking ideas which stem from more than one area of investigation. Paying strict regard to the sequence of steps needed for the competition of a job. Being fully occupied with what appear to be novel methods of solution. Adhering to the well-known techniques, methods and procedures of my area of work. Making unusual connections abut ideas even if they are trivial. Being strict on the production of results, as and when required. Searching for novel approaches not required at the time. Accepting readily the usua l and generally proven methods of solution. Struggling to make connections between apparently unrelated ideas. Being precise and exact about production of results and reports. Spending time tracing relationships between disparate areas of work. Adhering carefully to the standards of my area of work. Being caught up by more than one concept, method or solution. Being fully aware beforehand of the sequence of steps required in solving problems.
114 APPENDIX A: RESEARCH MATERIALS
115 SECTION 1 CONSENT FO RM AND LOG ON SCREEN General Introduction and Consent Form Let me start by thanking you. We need your he lp, and we appreciate you taking the time to participate in this study. Your efforts will guide us as we consider the effectiveness of SAS No. 99: Consideration of Fraud in a Financial Statement Audit Specifically, the purpose of this study is to determine how well indivi duals brainstorm and assess fraud risk in accordance with SAS No. 99. Please take a moment to read and sign the participant consent form below. The following information is being presented to help you decide whether or not you want to take part in a minimal risk research study. Please read this carefully. If you do not understand anything, ask the person in charge of the study. Title of Study: A Study Examining the Effectiven ess of SAS No. 99: Consideration of Fraud in a Financial Statement Audit Principal Investigator: Antoinette Lynch, Univer sity of South Florida (email@example.com; 813974-6863) Study Location(s): This is an Internet (Web-based) study. You are being asked to participate because you are an auditor. The purposes of this research study are (1) to obta in future auditors ideas for potential material misstatements due to fraud in a financial statem ent audit and (2) to determine the performance of individuals in a virtual environment. You will be asked to brainstorm and respond to a series of questions on different screens in this Web-based study. The entire study will take a pproximately 60 minutes to complete. You will receive $15 for your participation. By taking part in this research study, you will help increase the overall knowledge of the relative effectiveness of the brainstorming requirement for SAS No. 99. This study will help the aud iting profession understand the impor tance of using brainstorming techniques to consider solutions for complex problems. There are no known risks involved in taking part in this research study. By taking part in this research study, you will help increase the overall knowledge of the relative effectiveness of the brainstorming requirement fo r SAS No. 99. This study will help students to understand the importance of using brainstorming techniques to consider solutions for complex problems. There are no known risks involved in taking part in this research study. Your responses will be kept confidential to the ex tent of the law. It is possible because you are responding online that unauthorized individuals could gain access to your responses. Authorized research personnel, employees of the Department of Health and Human Services, and the Institutional Review Boards at the University of South Florida may inspect the records from this research project.
116 The results of this study may be published. However, the data obtained from you will be combined with data from others in the publica tion. The published results will not include your name or any other information that would persona lly identify you or your firm in any way. Your responses will be coded with a unique identifie r and will be stored in a database on a secure server located in the College of Business Administ ration at the University of South Florida. Only the Principal Investigator and a doctoral stude nt will have access to the database on the secure server. Your decision to participate in this research st udy is completely voluntary. You are free to participate in this research study or to withdraw at any time. You are free to refuse to answer any questions that make you feel uncomfortable. I understand that this research study has been revi ewed and approved by the University of South Floridas Institutional Review Board. For r esearch-related problems or questions regarding subjects rights, I can contact the Division of Research Compliance of the University of South Florida at 813-974-5638. Questions and Contacts If you have any questions about this research study, contact Ms. Antoinette Lynch at 813-9746863 or Dr. Uday Murthy at 813-974-6523. If you have questions about your rights as a pe rson who is taking part in a research study, you may contact the Division of Research Compliance of the University of South Florida at 813-9745638. Consent to Take Part in This Research Study By answering the questions, you agree that: I have fully read or have had read and explai ned to me this informed consent form describing this research project. I have had the opportunity to question one of th e persons in charge of this research and have received satisfactory answers. I understand that I am being asked to participate in research. I understand the risks and benefits, and I freely give my consent to participate in th e research project outlined in this form, under the conditions indicated in it. Investigator Statement: I certify that participants have been shown an on line information sheet via the Internet that has been approved by the University of South Florid as Institutional Review Board and that explains the nature, demands, risks, and be nefits involved in participating in this study. I further certify that a phone number has been provided in the event of additional questions. _________________________ _________________________ _______________ Signature of Investigator Printed Name of Investigator Date
117 Log On Screen Firm: Please select your firm... First name: Last name: Log on! R eset
118 SECTION 2 DEMOGRAPHICS, EVALUATION APPREHENSION, AND JABRI Demographics for External Auditors (adapted from Kozloski (2002)) First, we will need you to provide some basic demographic data. 1. How many years of external auditing experience do you have? 2. Which of the following classifications best represents your current position? Intern Junior Auditor Senior Auditor Manager Senior Manager Partner 3. What is your gender? male female 4. What is your e-mail address? 5. What is your age? 20-24 25-29 30-34 35-39 40-44 45-49 50 or more 6. Are you a Certified Public Accountant, Cer tified Management Accountant, or Certified Fraud Examiner? (Please check all that apply.): 10CPA CMA CFE CISA CIA None of the above 7. What is the highest level of education that you have earned? Bachelors degree Masters degree Ph.D. or DBA 8. What year did you obtain the degree listed above? 9. Have you ever brainstormed (i.e., hastily write down thoughts) with others (in a group setting, in any context)? Yes No 10. Have you ever been trained to use a brainstorming technique? Yes No 11 Within the last 12 months, how often have you brainstormed in a group setting to consider fraud in a clients financial statements? 12. On approximately how many audit engagements have you worked in your auditing career? 10 CPA: Certified Public Account ant; CMA: Certified Management Accountant; CFE: Certified Fraud Examiner; CISA: Cer tified Information Systems Auditor; CIA: Certified Internal Auditor
119 13. On approximately how many audit engagements have you worked where you were responsible for performing or s upervising planning procedures? 14. On approximately how many audit engagements have you worked where you were responsible for performing or supervis ing procedures relating to SAS No. 99, The Consideration of Fraud in a Financial Statement Audit? 15. Have you worked on an audit enga gement where fraud was suspected? Yes No 16. Have you worked on an audit engagement where fraud was detected? Yes No 17. Please briefly describe any training you have had related to the consideration of fraud or the detection of fraud. Please break this trai ning down into the following categories listed below. Please also indicate the length of said training (e.g., 4 CPE hours or day, as the case may be). 17a. Training relating to SAS No. 99 (or SAS No. 82), The Consideration of Fraud in a Financial Statement Audit 17b. Other fraud related training (please select all that apply): 1 CPE course in fraud Multiple CPE courses in fraud 1 fraud workshop (non CPE) multiple fraud workshops (non CPE) in-house fraud training co llege-level course(s) in fraud
120 Pre-Measure of Evaluation Apprehension 18. Please respond to the following questions using the 7-point scale provided. Answer questions from a work-related context. Click under the number that indicates the best representation of your judgment: 18a. Usually in a group, I am reluctant to offer an idea for fear of criticism from other members 1--------------2--------------3--------------4--------------5--------------6--------------7 Strongly Neutral Strongly Disagree Agree 18b. Usually in a group, I feel inhibited in offe ring an idea due to the presence of others who have more experience with brainstorming. 1--------------2--------------3--------------4--------------5--------------6--------------7 Strongly Neutral Strongly Disagree Agree 18c. Usually in a group, if I offer an idea that is way out, I get discouraged if I sense a certain disapproval from team members. 1--------------2--------------3--------------4--------------5--------------6--------------7 Strongly Neutral Strongly Disagree Agree 18d. I tend to withhold ideas, for fear of possible disapproval from other members. 1--------------2--------------3--------------4--------------5--------------6--------------7 Strongly Neutral Strongly Disagree Agree
121 Demographics for Internal Auditors First, we will need you to provide some basic demographic data. 1. How many years of internal auditing experience do you have? 2. How many years of external auditing experience do you have? 3. Years with your current company Less than 3 years 3 to 5 years 6 to 10 years 11 to 15 years 16 to 25 years 25 or more years 4. What is your gender? male female 5. What is your e-mail address? 6. What is your age? 20-24 25-29 30-34 35-39 40-44 45-49 50 or more 7. Are you a Certified Public Accountant, Cer tified Management Accountant, or Certified Fraud Examiner? (Please check all that apply.): 11CPA CMA CFE CISA CIA None of the above 8. What is the highest level of education that you have earned? Bachelors degree Masters degree Ph.D. or DBA 9. What year did you obtain the degree listed above? 10. Please select the industry of your company (select one)12: Agriculture Health Care/Medical Real Estate Banking/Securities Insurance Retail/Wholesale/Trade Business Services Legal Telecommunications Computer/Software Services Manufacturing Transportation Construction Professional Services Utilities Education Public Accounting Food Services Public Adm/Government Other 11. Please check your company/organizations size (employees): 11 CPA: Certified Public Account ant; CMA: Certified Management Accountant; CFE: Certified Fraud Examiner; CISA: Cer tified Information Systems Auditor; CIA: Certified Internal Auditor 12 Questions 9 through 11 adapted from http://www.businessfinance mag.com/survey/2003.cfm)
122 0 to 200 201 to 500 501 to 1,000 1,001 to 5,000 5,000 to 10,000 higher than 10,000 12. Have you ever brainstormed (i.e., hastily write down thoughts) with others (in a group setting, in any context)? Yes No 13. Have you ever been trained to use a brainstorming technique? Yes No 14 Within the last 12 months, how often ha ve you brainstormed in a group setting to consider fraud in a client s financial statements? 15. On approximately how many audit engagements have you worked in your auditing career? 16. Have you worked on an audit enga gement where fraud was suspected? Yes No 17. Have you worked on an audit engagement where fraud was detected? Yes No 18. Please briefly describe any training you have had related to the consideration of fraud or the detection of fraud. Please break this trai ning down into the following categories listed below. Please also indicate the length of said training (e.g., 4 CPE hours or day, as the case may be). 18a. Training relating to SAS No. 99 (or SAS No. 82), The Consideration of fraud in a Financial Statement Audit 18b. Other fraud related training (please select all that apply): 1 CPE course in fraud Multiple CPE courses in fraud 1 fraud workshop (non CPE) multip le fraud workshops (non CPE) in-house fraud training co llege-level course(s) in fraud
123 Pre-Measure of Evaluation Apprehension 19. Please respond to the following questions using the 7-point scale provided. Answer questions from a work-related context. Click under the number that indicates the best representation of your judgment: 19a. Usually in a group, I am reluctant to offer an idea for fear of criticism from other members 1--------------2--------------3--------------4--------------5--------------6--------------7 Strongly Neutral Strongly Disagree Agree 19b. Usually in a group, I feel inhibited in offe ring an idea due to the presence of others who have more experience with brainstorming. 1--------------2--------------3--------------4--------------5--------------6--------------7 Strongly Neutral Strongly Disagree Agree 19c. Usually in a group, if I offer an idea that is way out, I get discouraged if I sense a certain disapproval from team members. 1--------------2--------------3--------------4--------------5--------------6--------------7 Strongly Neutral Strongly Disagree Agree 19d. I tend to withhold ideas, for fear of possible disapproval from other members. 1--------------2--------------3--------------4--------------5--------------6--------------7 Strongly Neutral Strongly Disagree Agree
124 Jabris Measure of Problem-Solving Style These questions explore problem-solving style. Remember, there are no correct or incorrect answers. Please answer the following questions on the 7-point scale that rang es from unlikely to enjoy to likely to enjoy. Click under the number that i ndicates the best representation of your judgment. 1. Adhering to the commonly established rules of my area of work. 1--------------2--------------3--------------4--------------5--------------6--------------7 Unlikely Neutral Likely to Enjoy to Enjoy 2. Being confronted with a maze of ideas which may, or may not, lead me somewhere. 1--------------2--------------3--------------4--------------5--------------6--------------7 Unlikely Neutral Likely to Enjoy to Enjoy 3. Following well-trodden ways and generall y accepted methods for solving problems. 1--------------2--------------3--------------4--------------5--------------6--------------7 Unlikely Neutral Likely to Enjoy to Enjoy 4. Pursuing a problem, particularly if it ta kes me into areas I dont know much about. 1--------------2--------------3--------------4--------------5--------------6--------------7 Unlikely Neutral Likely to Enjoy to Enjoy 5. Being methodical and consistent in the way I tackle problems. 1--------------2--------------3--------------4--------------5--------------6--------------7 Unlikely Neutral Likely to Enjoy to Enjoy 6. Linking ideas which stem from more than one area of investigation. 1--------------2--------------3--------------4--------------5--------------6--------------7 Unlikely Neutral Likely to Enjoy to Enjoy 7. Paying strict regard to the sequence of steps needed for the completion of a job. 1--------------2--------------3--------------4--------------5--------------6--------------7 Unlikely Neutral Likely to Enjoy to Enjoy
125 8. Being fully occupied with what app ear to be novel methods of solution. 1--------------2--------------3--------------4--------------5--------------6--------------7 Unlikely Neutral Likely to Enjoy to Enjoy 9. Adhering to the well-known techniques, me thods and procedures of my area of work. 1--------------2--------------3--------------4--------------5--------------6--------------7 Unlikely Neutral Likely to Enjoy to Enjoy 10. Making unusual connections abut ideas even if they are trivial. 1--------------2--------------3--------------4--------------5--------------6--------------7 Unlikely Neutral Likely to Enjoy to Enjoy 11. Being strict on the production of results, as and when required. 1--------------2--------------3--------------4--------------5--------------6--------------7 Unlikely Neutral Likely to Enjoy to Enjoy 12. Searching for novel approaches not required at the time. 1--------------2--------------3--------------4--------------5--------------6--------------7 Unlikely Neutral Likely to Enjoy to Enjoy 13. Accepting readily the usual and gene rally proven methods of solution. 1--------------2--------------3--------------4--------------5--------------6--------------7 Unlikely Neutral Likely to Enjoy to Enjoy 14. Struggling to make connections between apparently unrelated ideas. 1--------------2--------------3--------------4--------------5--------------6--------------7 Unlikely Neutral Likely to Enjoy to Enjoy 15. Being precise and exact about production of results and reports. 1--------------2--------------3--------------4--------------5--------------6--------------7 Unlikely Neutral Likely to Enjoy to Enjoy
126 16. Spending time tracing relationships between disparate areas of work. 1--------------2--------------3--------------4--------------5--------------6--------------7 Unlikely Neutral Likely to Enjoy to Enjoy 17. Adhering carefully to the standards of my area of work. 1--------------2--------------3--------------4--------------5--------------6--------------7 Unlikely Neutral Likely to Enjoy to Enjoy 18. Being caught up by more than one concept, method or solution. 1--------------2--------------3--------------4--------------5--------------6--------------7 Unlikely Neutral Likely to Enjoy to Enjoy 19. Being fully aware beforehand of the sequ ence of steps required in solving problems. 1--------------2--------------3--------------4--------------5--------------6--------------7 Unlikely Neutral Likely to Enjoy to Enjoy
127 SECTION 3 MANIPULA TION OF TREATMENTS Task Introduction for Undergraduate an d Graduate Students Only (Pilot Test) You have been selected to participate in a brains torming exercise. You will be acting the role of a newly hired auditor, who is working for a Big 4 accounting firm. The purpose of this study is to test the effectiveness of SAS No. 99, which requir es auditors to brainstorm about the possibility of fraud. Thus, today, you will work together with a team of external auditors and exchange ideas about fraud in a financial statement audit. We th ink you will find this fun and a good way to get experience working in a virtual environment. T oday, you will work with experts of a nationally known Big 4 accounting firm: a senior auditor, mana ger, and senior manager. Thus, your team will consist of you and these three team members. Task Introduction for Internal Auditors Only You have been selected to participate in a brainsto rming exercise. As an internal auditor, you will be asked to brainstorm with a group of extern al auditors about a financial division of a hypothetical company. Under Standard for the Professional Practice of Inte rnal Auditing 1210.A2, internal auditors have a professional responsibility relating to fraud while performing normal internal audit responsibilities and in fraud investigations. Furthe r, in light of recent fraud cases, the internal auditor is being asked to become more of a pa rtner and consultant to the external auditor. The purpose of this study is to test the effec tiveness of SAS No. 99, which requires auditors to brainstorm about the possibility of fraud. Thus, today, you will work together with a team of external auditors and exchange ideas about fraud in a financial statement audit. We think you will find this fun and a good way to get experience wo rking in a virtual environment. Today, you will work with experts of a nationally known Big 4 accounting firm: a senior auditor, manager, and senior manager. Thus, your team will consist of you and these three team members. Task Introduction for GSS-Non-An onymous Interaction Mode Only You have been selected to participate in a brains torming exercise. The purpose of this study is to test the effectiveness of SAS No. 99, which requir es auditors to brainstorm about the possibility of fraud. Thus, today, you will work together as a team of four auditors and exchange ideas about fraud in a financial statement audit. We thi nk you will find this fun and a good way to get experience working in a virtual environment. Toda y, you will work with experts that were handpicked by a national representative of your firm: a senior auditor, manager, and senior manager. Thus, your team will consist of you and these three team members. You will begin by practicing with a task to get you acquainted with the software. The goal is to come up with as many ideas as possible to solv e the problem. No idea is too wild. Research shows that the more solutions you generate, th e more likely you are to generate good solutions. Brainstorming is a way to generate a lo t of solutions in a very short time.
128 Here are general brainstorming rules that appl y since you are brainstorming with other team members who are considered experts in your field. (1) Generate ideas that would be used in the audit planning process. (2) It is possible that someone will come up with an idea similar to yours. (3) Criticism is ruled out. Adverse judgment of ideas must be withheld until later. (4) Free-wheeling is welcomed. The wilder the id ea, the better; it is easier to tame down than to think up. (5) Quantity is wanted. The greater the number of ideas, the more the likelihood of useful ideas. (6) Combination and improvement are sought. In addition to contributing ideas of your own, you should suggest how ideas of others can be turned into better ideas; or how more ideas can be joined into still another idea. Task Introduction for GSS-Anonymous Interaction Mode Only You have been selected to participate in a brains torming exercise. The purpose of this study is to test the effectiveness of SAS No. 99, which requir es auditors to brainstorm about the possibility of fraud. Thus, today, you will work together as a team of four auditors and exchange ideas about fraud in a financial statement audit. We thi nk you will find this fun and a good way to get experience working in a virtual environment. Toda y, you will work with experts that were handpicked from a national representative of your firm: a senior auditor, manager, and senior manager. Thus, your team will consist of you and these three team members. The ideas of all team members will be anonymous This means that your team members will not be able to trace ideas to you. Your log-on na me and identification information is in no way tied to your comments. Likewise, you will not be ab le to determine if the idea was generated by the senior auditor, manager, or senior manager. You will begin by practicing with a task to get you acquainted with the software. The goal is to come up with as many ideas as possible to solv e the problem. No idea is too wild. Research shows that the more solutions you generate, th e more likely you are to generate good solutions. Brainstorming is a way to generate a lo t of solutions in a very short time. Here are general brainstorming rules that appl y since you are brainstorming with other team members. (1) Criticism is ruled out. Adverse judgment of ideas must be withheld until later. (2) Free-wheeling is welcomed. The wilder the id ea, the better; it is easier to tame down than to think up. (3) Quantity is wanted. The greater the number of ideas, the more the likelihood of useful ideas.
129 (4) Combination and improvement are sought. In addition to contributing ideas of your own, you should suggest how ideas of others can be turned into better ideas; or how more ideas can be joined into still another idea. Training Task for Unstructured Brainstorming Group Tea bag machine task For this task you are asked to brainstorm about: How to use excess capacity of tea bags. You work for a company that makes tea bags. Th e tea bag machines are currently producing tea bags over the expected capacity. The company wo uld like for you to come up with ways to use the excess tea bags. Remember, the goal is to come up with as many ideas as possible to solve the problem. No idea is too wild, criticism is ruled out, and quantity is wanted. Training Task for Guided Fantasy Training Group Tea bag machine task For this task you are asked to brainstorm about: How to use excess capacity of tea bags. You work for a company that makes tea bags. Th e tea bag machines are currently producing tea bags over the expected capacity. The company wo uld like for you to come up with ways to use the excess tea bags. Remember, the goal is to come up with as many ideas as possible to solve the problem. No idea is too wild, criticism is ruled out, and quantity is wanted. Guided Fantasy is a popular brainstorming techni que that is used to help individuals think outside the box. This activity w ill help you generate different id eas about using tea bags. You will want to read the following scenario at a slow pace. [ (Participants first name, captured by log on screen) please read the following scenario at a slow pace, and then use th e scenario to fantasize. Now, we will guide you into a fantasy. Sit comfor tably, close your eyes and take a few moments to relax. Become aware of your breathing and how it flows in and out. Once you are completely relaxed, read the following.
130 Destination: Brazil You have just won a dream vacation to Brazil. Y our vacation will take you from the night life of Rio de Janeiro, to the beautiful white beaches of Brazil. You and 3 of your friends will have passes to a fashionable and trendy nightclub. In th is nightclub, you will see the latest fashion wear and the movers and shakers of the Latin world, no outfit is too unique. Your hotel is on the beach, where you can have your breakfast served on your por ch and head to the beach for a day of fun in the sun (dont forget your sunscreen). After 4 da ys in the city and on the beaches you will be taken to the tropical rain forests. Here you w ill see hundreds of species of animals and flora. The mosquitoes and other bugs will be biting so rememb er to protect yourself. While there, you will have a chance to scale the great forest canopy an d experience the life of the rain forest. Enjoy your trip, bon voyage! Now, you are ready to begin the actual brainstorming session. Remember, you are in Brazil and your goal is to brainstorm about how to use excess capacity of the tea bags!!! Example of Brainstorming Simulator No n-Anonymous Team Interaction Treatment Chat log window TEAM MEMBER TEXT OF IDEA Participants First Name and Last Name Initial. Junior Auditor To keep mosquitoes away Pat S. Senior Auditor strain vegetables Chris T. Senior Manager To wash jewelry Dana P. Manager stuff pillow Chris T. Senior Manager could be used ofr instant coffee Participants First Name and Last Name Initial. Junior Auditor To wash delicate items Example of Brainstorming Simulator Anonymous Team Interaction Treatment Chat log window TEAM MEMBER TEXT OF IDEA Team Member 1 stuff pillow Team Member 4 To wash delicate items Team Member 3 could be used ofr instant coffee Missappropriation of Assets Case Co nsistent Across All Treatments
131 Chat system training completed... The actual task works exactly the same way. Yo u are required to read the following case about an audit client and then make an assessm ent about the likelihood of fraud. Lakeview Lumber, Inc. Case Information: (In order to protect the companys privacy, names have been changed) Here is the actual case.... For this task you are asked to brainsto rm about: How employees of Lakeview might commit fraud. Lakeview Lumber, Inc. is located in the city of Lakeview. Lakeview Lumber sells between 30,000 and 35,000 different kinds of building materials, lawn and garden products, and home improvement supplies to retail customers, as well as to contractors and other building professionals. Retail customers are required to pay in cash or by a major credit card at the time of their purchase. However, the vast majority of contractors and build ing professionals have established credit accounts and are billed on a m onthly basis. Lakeview Lumbers main competitors are The Home Depot, Inc. and Eagle Hardware & Garden. THE KEY ACCOUNTING PLAYERS Joe Metros, Controller of Lakeview Lumber, Inc., is responsible for the firms accounting activities. Joe was recently hired and had been the Deputy Director of a finance department in a nearby town for the past five years. A repor ter from the Daily Observer interviewed Joe for a feature article in the business section. Joe talk ed about his family and the many civic activities that he supported, both financially and by volunteering his time. He also discussed his vision for the future of the Accounting Department and id entified a number of short-term and long-term goals. Initially, Joe wants to implement a number of changes designed to improve the efficiency and effectiveness of departmental operations. He plans to eliminate a number of accounts that are rarely used. He also hopes that financial info rmation can be provided more quickly when requested by auditors and department heads. Joe is especially concerned about the extent of employee turnover. Five of the seven department employees have held their current positions less than one year, and training costs can be rather significant. Joe has been told that the previous Controller, Crystal Smith, was very controlling and task-oriented, and that this may have caused employees to seek employment elsewhere. In a ddition to Joe, the Accounting Department includes the following personnel: Libby Jones, Chief Accountant She manages and maintains the General Ledger. Libby is also responsible for general office management and day-to-day operations in the department. She earned a degree in accounting from the local univers ity and has worked for the department for 15 years. Libby is 37; her husband owns a local hardware store. Marsee Weston, Senior Accountant She is responsible for monitoring property, plant, and equipment. She also maintains all records of fixe d/real assets. Marsee has been employed by the department for eight months. She is 39; her hus band teaches mathematics at the local high school.
132 Scott Smyth, Senior Accountant He is the Cash Manager; main tains bank relations; manages all investments; performs all wire transfers; and reconciles all bank accounts. Scott is 32 and has been employed by the department for seven months. Scotts wife is a sales associate at one of the local automobile dealers. Cathy Elgin, Staff Accountan t. She maintains all records pert aining to credit accounts; invoices those contractors and building professionals w ho owe money on their credit accounts; maintains control of all Petty Cash Funds; accounts for a ll daily deposits from departments within the company; and is also the secondary payroll clerk. Cathy is 27 and has been employed by the department for almost nine months. Her husband is employed by the U.S. Postal Service. Bob Thomas, Accounts Payable Clerk He processes all payments to suppliers with names beginning with A through L. Bob is 36 and has wo rked in the department for almost two years. He is single and has lived in town his entire life except for the five years he served in the U.S. Navy. Nora Stewart, Accounts Payable Clerk She processes all payments to suppliers with names beginning with M through Z. Nora is 20, and has been employed by the department for six months. She is single and lives in an apartment complex near the university campus. Chuck Sanchez, Payroll Clerk He processes all bi-weekly and monthly payrolls and maintains all payroll records. Chuck is 31, recently divorced and has been working in the department for ten months. Chuck lives in an older neighborhood with his 7-year-old son.
133 Guided Fantasy Treatment Group Inspector Gadget Fantasy 13 Now, you will be guided into an Inspector Gadge t fantasy. As previously stated, this Guided Fantasy technique is designed to help you think outside the box, just like the Brazil scenario. Remember to read the scenario at a slow pace. Come up with as many creative ideas as possible. Sit comfortably, close your eyes and take a few moments to relax. Become aware of your breathing and how it flows in and out. Once you are completely relaxed, read the following... [Participants first name, captured by log on screen] you are Inspector Gadget on a special assignment at the Museum of Modern Art in Ma nhattan. Your assignment is to prevent the theft of the museums inventory. At closing time, you send the museums security guards on their way, except for one who is to watch the doors for you. You intend to spend the night in the museum to protect the artwork. Dr. Claw and two of his goons, meanwhile, have backed a tractor trailer up to the rear of the museum, and are getting r eady to do some dirty work. Dr. Claw activates two of his mechanical monsters who are inside the museum posing as enormous statues. As these monsters distract you, Inspector Gadget, the trusted security guard opens the rear door of the museum to allow Dr. Claw and his goons to enter. Go Gadget! Go! You are going to need every techno-trick up your cyber-sleeve to defeat the ruthless Claw.14 You have your helihat that allows you to fly from room to room; your helping hands just in case you need an extra pair; your telescopic legs to raise you up so that you can see beyond normal distances. The same can be done with your telescopic neck. Okay, Inspector Gadget, use your techno-tricks, special gadgets, squirt guns, roller skates, and magna glass to solve this mystery. Now, you are ready to begin the actual brainsto rming session. Remember, the goal is to think about the Lakeview Lumber case, brainstorm a bout how employees of Lakeview Lumber, Inc. might commit fraud, and remain in your Inspector Gadget mode. Go Gadget! Go!!! 13 http://www.geocities.com/Hollywood/Screen/7219/ 14 The concept of this story is adapted from http://www.angelfire.com/80s/inspectorgadget/first_season/art_heisthtml
134 On the following screen, you are to brainstorm about the possible ways in which fraud might be committed by Lakeview Lumbers employees. Remember, no idea is too wild, no idea will be criticized by anyone, and more ideas are better. Again, time is important. You will have 15 minutes to complete this task. It is important that the 15 minutes be used as efficiently as possible. Brainstorming Session for fraud
135 SECTION 4 POST-EXPERIMENTAL QU ESTIONNAIRE CONSISTENT ACROSS TREATMENTS Post-Study Questionnaire Please respond to the following questions using th e 7-point scale provided. Please click under the number that indicates the best representation of your judgment: Intrinsic Motivation For me, the brainstorming activity was motivated more by intrinsic factors (my own interest) than by extrinsic factors (e.g., others working in the gr oup, the instructions that were provided to me). 1--------------2--------------3--------------4--------------5--------------6--------------7 Strongly Neutral Strongly Disagree Agree I got a lot of pleasure out of brainstorming about employee fraud. 1--------------2--------------3--------------4--------------5--------------6--------------7 Strongly Neutral Strongly Disagree Agree I enjoyed the opportunity to participate in this study. 1--------------2--------------3--------------4--------------5--------------6--------------7 Strongly Neutral Strongly Disagree Agree I achieved new insights through brainstorming about employee fraud. 1--------------2--------------3--------------4--------------5--------------6--------------7 Strongly Neutral Strongly Disagree Agree I derived satisfaction from brainstorming about employee fraud. 1--------------2--------------3--------------4--------------5--------------6--------------7 Strongly Neutral Strongly Disagree Agree I enjoyed being involved with other team members during the brainstorming activity. 1--------------2--------------3--------------4--------------5--------------6--------------7 Strongly Neutral Strongly Disagree Agree
136 Extrinsic Motivation How much did you think about impressing other team members while generating ideas? 1--------------2--------------3--------------4--------------5--------------6--------------7 Very Little Neutral A Whole Lot How much did you want to generate ideas that were comparative or better than other team members ideas? 1--------------2--------------3--------------4--------------5--------------6--------------7 Very Little Neutral A Whole Lot I completed this study because it was something I felt I had to do. 1--------------2--------------3--------------4--------------5--------------6--------------7 Strongly Neutral Strongly Disagree Agree I did not take the task seriously because there was no monetary or other tangible benefit for performing well. 1--------------2--------------3--------------4--------------5--------------6--------------7 Strongly Neutral Strongly Disagree Agree
137 Interaction Mode Manipulation Check Please select the statement below that best describes the team you brainstormed with: Members of my group included a senior auditor, manager, senior manager. I cannot recall the ranking status of my group members. Please select the statement below that best describes the team you brainstormed with: Each team members idea was tagged with his/her first name, last name initial, and job title. Each team members idea was anony mous and tagged as Team Member 1, 2, 3, or 4. Anonymity: Were you told that you were in an anonymous group, where your team members could not determine which ideas you submitted? 1= yes, 2= dont know; 3=no Evaluation Apprehension Manipulation Check Please respond to the following questions using th e 7-point scale provided. Please click under the number that indicates the best representation of your judgment: I was reluctant to offer an idea for f ear of criticism from other members. 1--------------2--------------3--------------4--------------5--------------6--------------7 Strongly Neutral Strongly Disagree Agree I was inhibited in offering an idea due to the presence of others. 1--------------2--------------3--------------4--------------5--------------6--------------7 Strongly Neutral Strongly Disagree Agree Although no overt criticism was expressed, I was reluct ant to offer an idea that was way out, for fear of disapproval from members. 1--------------2--------------3--------------4--------------5--------------6--------------7 Strongly Neutral Strongly Disagree Agree I withheld ideas for fear of possible disapproval from other members. 1--------------2--------------3--------------4--------------5--------------6--------------7 Strongly Neutral Strongly Disagree Agree
138 I was aware of the position of each person. 1--------------2--------------3--------------4--------------5--------------6--------------7 Strongly Neutral Strongly Disagree Agree I was mindful of the job titles/rank of my team members. 1--------------2--------------3--------------4--------------5--------------6--------------7 Strongly Neutral Strongly Disagree Agree Guided Fantasy Manipulation Check Did you receive training on a brainstorming technique that involved fantasizing? Yes No Task Complexity Overall, how would you rate the difficulty of the brainstorming task for the employee fraud case you had to do in this study? Compared to the tasks I usually work on, I tho ught that the experimental task (brainstorming about ways employees could commit fraud) was very easy 1--------------2--------------3--------------4--------------5--------------6--------------7 Strongly Neutral Strongly Disagree Agree Social Presence The session was functionally equivalent to a scenar io where I was in the same room with my team, each at a different computer terminal. 1--------------2--------------3--------------4--------------5--------------6--------------7 Strongly Neutral Strongly Disagree Agree The session worked as well as it would have if the team was in the same room. 1--------------2--------------3--------------4--------------5--------------6--------------7 Strongly Neutral Strongly Disagree Agree Debriefing Thank you for your participation!!! To preserve the integrity of this research, please do not discuss this study with your colleagues. When the study is complete, we will send you an e-mail providing additional information regarding the purpose of this study. If you have any questions, please feel free to contact Ms. Antoinette Lynch at firstname.lastname@example.org or Dr. Murthy at email@example.com
139 APPENDIX B INSTRUCTIONS TO RATERS
140 Rating Instructions (adapted from Santanen (2002)) Please read these instructions: Thank you for agreeing to assist with this research project. You have been asked to participate due to your experience in the area of fraud. Your participati on will consist of scoring solutions that were generated in response to a fraud case involving misappropriation of assets. It is estimated that scoring the solutions in th is file will take you approximately 2 hours. This file contains three worksheets (each is a di fferent tab across the bottom of the spreadsheet). 1. The first sheet contains these INSTRUCTIONS. 2. The second sheet contains the background information provided for the CASE and the instructions provided to participants. 3. The third sheet contains the pool of SOLUTIONS to the fraud task generated by audit interns and internal auditors during a 15-minute period. Additionally, this sheet contains one measure: Review each idea and rate the extent to which you believe you would use or consider the idea in the audit planning process of Lakeview Lumber (see case on 2nd sheet). Do not worry about related cost. Rate each idea on a scale of 1 to 3, where 1= not useful; 2=moderately useful; and 3=very useful. Additional procedures for scoring the solutions c ontained in the third worksheet are as follows: 1. Please take a moment to read through the case and instructions contained on the CASE worksheet. 2. Before you score any of the IDEAS, please take a moment to familiarize yourself with the ideas by reading a random sampling of them (perhaps 10 to 20 ideas). For example, read several ideas from the top of the list, read some from the mi ddle of the list, and then read some closer to the end of the list. There are approximately 98 ideas in total. 3. These ideas have been generated by auditors who may not have much experience in the area of fraud. Please rate these ideas RELATIVE TO ONE ANOTHER rather than rating them against some absolute standard that may exist for fraud in general. The rating scale for usefulness is to flow along a range of 1 to 3 such that a value of 1 means not useful and 3 represents very useful. The general aim is to rate the ideas relative to each other using your experience and judgment as a guide. The ideas have been numbered for your convenience. 4. In order to score the solutions relative to one another, please score AT LEAST ONE solution as 1 (not useful) and score at lease one solution as 3 (very useful). It is entirely possible, though not required, that multiple solutions may receive a score of 1 and multiple solutions may receive a score of 3. Please use your own subjective j udgment in making these assessments. Remember, junior auditors generated these solutions. 5. If you have any questions about this procedure, please contact Antoinette at the following email address: firstname.lastname@example.org or 813-974-6863. 6. It is clear that some of the subjects gave this task a more serious effort than others. For the sake of making comparative judgments, it was neces sary for all the solutions to be included in this data set for scoring. As this data set cont ains data from different experimental treatments, some of these differences may be a result of the particular technique that was used with each
141 treatment. This is why we need your assistance. 7. Thank you for your help!!! Your tim e is greatly appreciated! Thank you for your participati on in this research project.
142 APPENDIX C INFORMATION ON CODERS AND RATERS
143 VERONDA WILLIS, CPA ( CODER ) University of Colorado Campus Box 419 Boulder, CO 80309-0419 E-mail: Veronda.Willis@Colorado.EDU Ms. Veronda Willis, CPA, is an accounting Ph.D candidate (minor in econometrics) at the University of Colorado. She received both her Bachelor of Science Degree and a Masters Degree in Professional Accounting from the University of Texas at Austin. She was held many positions from 1995 to 2000, including accounting manager at Enron Capital & Trade Resources. She also worked for PricewatershouseCoopers from 1990 to 1994. Ms. Willis has many accolades and her vita is available from the author, upon request. MARGARITA MARIA LENK ( CODER ) Associate Professor, Departments of Computer Information Systems (CIS) and Accounting 208 Rockwell Hall, Colorado State University, Fort Collins, CO 80523 (970) 491-4983 FAX (970) 491-5205 E-mail: Margarita.Lenk@colostate.edu Dr. Margarita Lenk, CMA, is an associate professor at Colorado State University. She received her Ph.D. in 1991 from the University of South Carolina, her MACC degree from the University of North Carolina at Chapel Hill, and her BSBA degree from the University of Central Florida. Dr. Lenk has an overwhelming number of referred publications, presentations, and books that she has authored. Her vita is av ailable from the author, upon request.
144 DANIEL J. JOHNSON ( RATER ) Senior Audit Manager, Kirkland, Russ, Murphy, & Tapp www.KRMTCPA.com Mr. Johnson received his Bachelor of Scien ce Degree in Accounting and Finance from Augustana College, Rock Island, Illinois. He sp ent 10 years with Arthur Andersen, LLP in Tampa and Chicago. He has extensive experience in all aspects of audit, review, transaction due diligence, and benefit plan services. He served several publicly traded companies and is well versed in SEC reporting issues. Mr. Johnsons re levant industry experi ence includes work with real estate, hospitality, construction, timeshare, ma nufacturing, retail, distribution, and financial service clients. BOB BATZ ( RATER ) Audit Manager, Kirkland, Russ, Murphy, & Tapp www.KRMTCPA.com Mr. Batz received his Bachelor of Science Degree in Accounting from the University of South Florida. He has extensive experience with all aspects of audit, review and compilation services. Mr. Batz plays a significant role in fi rm-wide training and recruiting efforts. His relevant industry experience includes automotive, restaurants, construction, credit counseling, manufacturing, and distribution. He also has s ubstantial knowledge of th e reporting requirements of employee benefit plans.
ABOUT THE AUTHOR Dr. Antoinette LaBarbara Lynch was born and raised in Hampton, Virginia. She has two lovely daughters, Jaleesa and Shauntia Lynch. In 2000, she relocated to Tampa, Florida, to pursue a Doctorate of Philosophy in A ccounting Information Systems. Prior to pursing her Ph.D., Dr. Lynch gradua ted from Christopher Newport University with a BSBA in Accounting. She also completed 2 years of master-level coursework at The College of William and Mary. She has worked for the Univers ity of South Florida, NASA Langley Research Center, Eason, Lawson, and Westphal, P.C., and the Air Force Audit Agency. She is a member of several accounting organizations and recipient of several scholarships and awards. Her research interests include behavioral research in areas of improving auditors abilit y to detect fraud and understanding the impact of information technol ogy on the audit process and individual behavior. Dr. Lynch believes that we all face challe nges in our life, and how you handle those challenges will affect the road you travel. She be lieves that goals are not possible if you do not rely on the wisdom of others, and if you are not grateful for that wisdom. Seeking knowledge and sharing knowledge is the key to personal developmen t. Everyone should take time to self analyze themselves, and not spend energy on criticizing choices of others. The world is a very dynamic universe, and how we view the world and our pos ition in the world, will impact all that we know and do.
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 001474317
007 cr mnu|||uuuuu
008 040712s2004 flua sbm s000|0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0000362
Lynch, Antoinette L.
Auditors' performance in computer-mediated fraud assessment brainstorming sessions
h [electronic resource] :
an investigation of the effects of anonymity and creativity training /
by Antoinette L. Lynch.
[Tampa, Fla.] :
University of South Florida,
Thesis (Ph.D.)--University of South Florida, 2004.
Includes bibliographical references.
Text (Electronic thesis) in PDF format.
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 156 pages.
ABSTRACT: In the wake of recent corporate accounting scandals, auditors are encouraged to improve their method of fraud detection. Although Statement on Auditing Standards (SAS) No. 99 does not change the responsibility of the auditor for detecting fraud, it does provide new procedural requirements for assessing fraud risk, such as brainstorming among key team members about the potential for fraud. Using audit interns and internal auditors, this study empirically examines two interventions hypothesized to improve the quality of ideas generated by audit interns and internal auditors. In the first intervention, auditors use a computer-based group support system to brainstorm either non-anonymously or anonymously. For the second intervention, auditors were either trained to use a paradigm-modifying creativity technique or not trained. Additionally, it is hypothesized that the creativity training will have the greatest impact on brainstorming effectiveness when auditors brainstorm anonymously. However, the results suggest that audit interns working non-anonymously generated the greatest number of fraud ideas and also the greatest number of original ideas. Audit interns who received training on a paradigm-modifying creativity training technique generated the greatest number of unique ideas and received, on average, the highest usefulness to the audit process score.
Adviser: Murthy, Uday
Jabri associative/bisociative scales.
x Business Administration
t USF Electronic Theses and Dissertations.