گفتمان تیم، غنای رسانه و اثرات ناشناس ماندن در نشانه تقلب ممیزی طوفان مغزی را توضیح می دهد
|کد مقاله||سال انتشار||مقاله انگلیسی||ترجمه فارسی||تعداد کلمات|
|943||2011||18 صفحه PDF||سفارش دهید||محاسبه نشده|
Publisher : Elsevier - Science Direct (الزویر - ساینس دایرکت)
Journal : International Journal of Accounting Information Systems, Volume 12, Issue 3, September 2011, Pages 225–242
This article contributes by extending media richness (MRT) and media synchronicity theories (MST) to explore how media richness and anonymity influence team interactions and success in audit fraud brainstorming. Sixty-three, three-person teams, with 189 student participants from two Universities, identified fraud risk cues in a SAS 99 audit planning case. Participants were assigned to one of three conditions: electronic anonymity (EA; n = 18 teams), electronic identified (EI; n = 28 teams), or identified face-to-face (FtF; n = 17 teams). Compared with teams in the low media richness conditions, i.e., the EA and EI, discussions in FtF teams produced more and better dialog, which resulted in better identification of fraud risk cues. Additionally, compared with the discussions in the EA teams, FtF team discussions evidenced less narcissism and were more focused and inhibited. Mediation analyses of team interactions indicated that the quantity of dialog (team production) completely explains, fully mediates, the effects of media richness and anonymity on risk assessments. Contributions include extending MRT and MST, and using automated content analysis, to explicate the role of media richness, anonymity, and team interactions in explaining audit team fraud identification success. The concluding section identifies the sample, design, and method limitations, and, discusses the potential for group support technologies to enhance or detract from audit team processes, depending on task, context, and technology.
Detecting fraudulent financial reporting is among the external auditor's most critical and difficult responsibilities (Louwers, 2011, Wells, 1997, Wells, 2004 and Wells, 2005). Recent research describes the processes used by audit firms to conduct audit fraud brainstorming (Bellovary and Johnstone, 2007 and Brazel and Carpenter, 2010), explores the conditions in which individual auditors and audit teams successfully detect fraudulent financial reporting (e.g., Carpenter, 2007), and investigates the influence of technology on the detection of fraudulent reporting (e.g., Hunton and Gold, 2010). This study builds upon, and contributes to, this literature in multiple ways. First, SAS No. 99 – which mandates the use of brainstorming among audit teams – allows discretion in the methods and processes adopted for brainstorming sessions. International standards on auditing (ISA), specifically, International Federation of Accountants (IFAC) ISA 240 (IFAC, 2009a) and ISA 315 (2009b), require early-engagement, audit-team discussion of the possibility of material financial misstatements due to fraud or error. Available technologies increasingly support multiple aspects of audit team deliberations and decision-making (Bamber et al., 1996), thereby aiding team discussions. However, evidence suggests that communication technologies are not yet in common use among audit teams engaged in fraud brainstorming (Bellovary and Johnstone, 2007). In addition, variability in fraud brainstorming processes (Brazel and Carpenter, 2010) and differences in the structure and technology adopted influence team success (Hunton and Gold, 2010 and Lynch and Murthy, 2009). This article investigates conditions – enabled by recent advances in technology – which may improve, or impede, the success of audit team brainstorming. Second, we investigate these issues by extending theories, specifically, media synchronicity theory (MST) and media richness theory (MRT), that have received variable support in previous investigations (e.g., see Kahai and Cooper, 2003). One possible reason for the limited support for MST and MRT may be that much research has investigated these theories with relatively simple tasks that pose hypothetical questions (e.g., “How can we attract tourists to Tucson?”) whose solution does not require professional knowledge (cf. Carpenter, 2007). It is possible that, following the principles of MST and MRT, the beneficial effects of richer media increase with task complexity and task demands. This investigation applies MST and MRT to a complex task, using predictions that are consistent with the original intent and scope of the theory. Third, web-privacy concerns have created an expanding set of products that facilitate web and email anonymity (Tuna, 2010). One possibility – which must be considered in evaluating technologies for adoption by auditors in SAS 99 fraud brainstorming sessions – is whether auditor anonymity improves or impedes brainstorming outcomes (cf. DeZoort and Harrison, 2008a and DeZoort and Harrison, 2008b). This article manipulates, and tests the effects of, anonymity on brainstorming. Fourth, evidence from professional practice suggests that team processes influence brainstorming outcomes (Brazel and Carpenter, 2010). Unfortunately, until recently, the available technologies for investigating team processes were restricted to broad, aggregated metrics such as the time taken for deliberations and individual perceptions of team deliberations. Recent advances in automated content analysis, undertaken in this investigation, enable the detailed analysis of audit team processes. Finally, the detailed examination of team processes facilitates a more precise understanding of why and how team processes influence brainstorming outcomes. Coupling methodological advances in multiple mediation analysis, with detailed analyses of team processes, enables the determination of whether, and how, team processes explain, i.e., mediate, the effects of media richness and anonymity on audit fraud brainstorming success. The reported experiment manipulates two aspects of teams' communication processes. First, interactions in face-to-face (FtF) teams, i.e., using richer media, are compared with interactions among teams in two lower media richness conditions, i.e., electronic communication conditions in which participants are identified (EI) or anonymous (EA). Second, the experiment compares the effects and effectiveness of individual and team anonymity, i.e., teams in the EA condition, with that of the teams in the identified conditions, i.e., the EI and FtF. The investigation proceeds, initially, with a review of recent research investigating audit fraud brainstorming, followed by hypotheses based in, and extending, MST and MRT. Description of the experimental method and results follows, with a concluding section that explores the implications and limitations of the investigation.
نتیجه گیری انگلیسی
6.1. Summary Table 6 summarizes the hypotheses and results. The hypotheses related to media richness (H1–H4) and mediation (H7) are fully supported. Teams in the high media richness condition, FtF, identified more and better fraud risk factors than did low media richness, i.e., electronic, teams (H1). In addition, FtF teams produced more dialog (H2), which was more complex (H3) and descriptive (H4) than was dialog in the low media richness (i.e., EA and EI) teams. Differences in the quantity, and to a lesser extent — quality, of team discourse explain, i.e., mediate, the differences in fraud cue identification success. The quantity of team dialog fully, while the extent of narcissism partially, mediates (H7) the effect of team condition on fraud risk identification. These results suggest that the richer media available in FtF communication facilitated higher quality fraud risk brainstorming discourse, than in EA and EI communication, by increasing the amount of team discussion and, perhaps, also decreasing the extent of narcissism in this discussion. Two hypotheses explored the effects of anonymity on team processes. Results indicated that EA teams were more narcissistic (H5: i.e., self-absorbed) and less inhibited (H6) than were FtF teams, with EI teams falling between EA and FtF teams on these measures. Hence, in this task, anonymity produced less inhibited discussions but did not improve success in identifying fraud risk cues. Why might rich media, i.e., FtF, teams generate more discourse than e-chat teams despite approximately equal task time? And why might these results contrast with some previous results, showing that more team dialog occurs in e-chat, compared with FtF, teams? Evidence suggests that media richness and anonymity effects are contingent on task, implementation, and team effects (Kahai et al., 1998). For example, anonymity appears to facilitate group processes more in simple idea generation (versus decision) tasks, with greater (i.e., participant-level) anonymity, and when groups find initial common ground (rather than initially disagree) (Kahai et al., 1998). This article contributes an additional data point to this literature; the assigned task was complex and difficult, included the potential for disagreements among team members, contained socially sensitive topics (i.e., sex and drugs), and required that team members listen to others' contributions and engage in “sensemaking” (Weick, 1995) to discern the implications of the case cues for fraud risk. In a complex setting such as is explored herein, participants may engage in deeper, more meaningful discussions when they speak, i.e., in FtF dialog, than type, i.e., in e-chat dialog. In contrast, when producing dialog that is simple and uncontroversial (e.g., “how can we attract more tourists to Tucson?”), e-chat teams may produce more dialog since it is a simultaneous, i.e., parallel, communication channel and typing, if done in parallel, is faster than talking. Finally, in complex tasks, listening to others, and integrating cues, may be more important than the simple generation of large quantities of text without integration and reflection. Hence, task effects and demands may explain the higher productivity, and corresponding success, of the observed FtF, compared with e-chat, teams. 6.2. Implications — When might technology-facilitated anonymity help auditor judgment? EA teams were more self-focused (narcissistic) and less inhibited than were FtF teams. But the lessened inhibitions that resulted from anonymity did not improve EA team's ability to identify fraud risk cues. However, it is possible that anonymity may facilitate task performance in some accounting tasks, settings, and contexts that differ from those investigated herein. For example, in accounting and auditing tasks that demand creative solutions (Bryant et al., 2011) the dis-inhibition afforded by anonymity may increase the quantity of contributions (cf. Kahai et al., 1998). Future explorations could contribute by investigating subtasks within SAS 99 brainstorming that may benefit from anonymity. For example, we speculate that the domain of emergent frauds, within which conventional cues may be un-predictive, is one possible domain within which anonymity-enhanced dis-inhibition may provide benefits. Perhaps auditors who must identify jamais vu (e.g., Weick, 1993) fraud, i.e., frauds that have not been seen before and are therefore less susceptible to detection through standardized processes and procedures, may benefit from technology-created anonymity. For example, the Enron fraud brought together a unique nexus of market conditions (e.g., recent utility deregulation, particularly in California) and relations among stakeholders (e.g., Arthur Andersen's consulting engagement partner making all critical decisions regarding the Enron audit ( Seay and Bryan, 2002 and Williams, 2004)). Although speculative, it may be possible that in jamais vu audits, such as the 1999–2000 Enron audits, the valuing of idea production, i.e., “there are no bad ideas”, may help identify emerging, unconventional frauds. Organizational cultures with strong hierarchies may also benefit from anonymity. Hierarchies inhibit idea production, particularly when they include strong performance and accountability pressures. It seems likely that pressured staff auditors who work under autocratic supervisors are more likely to suppress ideas than are staff auditors who work in more relaxed and participative cultures. To the extent that auditors work in high pressure, organizational hierarchies, anonymity may improve risk cue assessments. One researcher with whom we spoke indicated that a research team had planned an experiment that included anonymous team contributions by audit staff, but that the audit firm with whom they were working would not approve this manipulation, saying that the firm would never allow anonymous staff contributions in practice. This anecdote suggests that resistance to anonymity may be rooted in accounting firm cultures. A second innovation that might facilitate uninhibited exchanges – which to our knowledge is unexplored in the existing auditing and accounting literature – is the use of a “devil's advocate” for assessing audit fraud risk (Schweiger and Sandberg, 1986, Schwenk, 1984, Schwenk, 1990, Valacich and Schwenk, 1995a and Valacich and Schwenk, 1995b). Using this approach, one individual or team would advocate for the existence of fraud in the audit while another individual or team would advocate for its absence. The use of this method has recently been applied to improve production quality at Toyota (Greimel and Rechtin, 2011) and as an aid in detecting lies (Leal et al., 2010). 6.3. Limitations Some evidence suggests that experienced-based differences exist in fraud detection knowledge and skill. For example, Carpenter (2007) compared fraud brainstorming effectiveness among audit staff, seniors, and managers. Results indicated that the fraud risk assessments of managers, and three-person audit teams (each of which included a manager, senior and staff member), were more accurate than those of individual seniors and staff. Surprisingly, Seow (2009) found that “low-knowledge” students had higher levels of fraud risk knowledge than did “high-knowledge” directors; there were no differences in students' and directors' success in identifying fraud risk cues (t(224) = 0.5352, p = 0.5931; calculated from Seow's Table 2). Similarly, data from the present study indicate no difference in fraud detection performance due to the number of team members who were M.Acc versus undergraduate students (F(2, 60) = 0.337, p = 0.715). The research objectives of the present investigation do not include testing accounting students' absolute level of ability to detect audit fraud. Instead, a primary goal was to understand better, the relative influence of group support technologies on the identification of fraud cues. Barring an interaction between technology and knowledge, for example, that students are more adept at using group support technologies than are professional auditors, the pattern of the observed results, though not the mean levels of detected fraud, should generalize to professional auditors.8 However, the continued investigation of the joint effects of accounting knowledge with decision and group support technology remains an important topic for future research. The research design of this investigation attempted to emulate the ecological hierarchy of audit teams through the use of assigned roles and differing case knowledge by role. The assignment of student participants to audit roles was consistent with aspects of the organization of audit teams. The use of student participants assigned to audit team roles lessens the likelihood of effects that may obtain from the organizational hierarchy of public accounting firms. However, there is evidence from classic psychological research (e.g., Milgram, 1963) that student participants, who are assigned to hierarchical roles, can readily assimilate into these roles, in some cases, to the point of engaging in criminal behaviors, e.g., physically beating student participants who were randomly assigned to a “prisoner” condition. 6.4. Contributions Teamwork is ubiquitous in auditing and accounting (Rich et al., 1997), yet, until very recently, few studies experimentally investigated audit teams. This study contributes by identifying how media richness and anonymity impact audit team fraud brainstorming success through the mediating influences of group processes. It extends theory by testing predictions, derived from, and extending, MST and MRT, in the complex and demanding task of audit fraud brainstorming. The results indicate that team processes differ in high versus low media richness, and identified versus anonymous, team conditions. Manipulating media richness and anonymity changed the quantity and quality of identified fraud risk cues and the nature of team interactions. This study differs from much of the preceding audit fraud brainstorming literature in that we jointly investigate the effects of media richness, anonymity, and team processes. In addition, this is among the first accounting studies to apply emerging multiple mediation and automated content analyses methods. The application of these methods allows a more complete investigation of the fundamental issue of causality — i.e., to specify why the observed results obtain. Both the quantity and quality of the linguistic content produced by teams influenced fraud risk identification. Accordingly, a focus on the content and process of team interactions would seem crucial to continuing recent advancements in audit fraud brainstorming research.