I investigate how auditors integrate information technology (IT) specialist input into internal control over financial reporting (ICFR) issue classifications. Given the ill-structured nature of evaluating ICFR issues and the impact of these issues on audit quality, combining knowledge from different perspectives is likely beneficial. Drawing on social identity theory, I predict and find that a weaker one-team identity between auditors and IT specialists yields benefits. Auditors with a weaker versus stronger team identity place more weight on IT specialist input for IT-related issues and differentially weight higher and lower quality input for non-IT issues. I also find that more severe ICFR issues drive the predicted results. My study provides insight into how team identity influences auditor integration of input from specialists. The implications of my study are of interest to researchers, regulators, and practitioners, especially as recent firm initiatives encourage a one-team view for auditors and IT specialists.

Auditors are required to identify internal control over financial reporting (ICFR) issues and classify them by severity (Public Company Accounting Oversight Board [PCAOB] 2007). The severity classification of ICFR issues (i.e., material weakness versus significant deficiency versus control deficiency) impacts planned audit procedures (PCAOB 2007), which determine audit quality. Evaluating ICFR issues is difficult due to the ill-structured nature of the task and limited guidance in standards (Earley, Hoffman, and Joe 2008). Regulators are concerned with the number of ICFR-related audit deficiencies noted in inspection reports (International Forum of Independent Audit Regulators [IFIAR] 2017) and cite ineffective coordination and communication between auditors and information technology (IT) specialists as a root cause of ICFR-related audit deficiencies (PCAOB 2012). Given these concerns, I experimentally investigate how a key aspect of the audit team environment, strength of team identity, influences auditor integration of input from IT specialists in the context of evaluating ICFR issues.

ICFR issues can be either IT-related or non-IT depending on the presence or absence of IT-related elements. IT specialists are a credible source of input (i.e., advice, recommendations) for both types of issues. Their work focuses on not only IT areas, but also on risk and control areas in general. In fact, IT specialists often view themselves as risk and control experts, rather than just IT experts (Bauer and Estep 2019). Despite IT specialists being a credible source for IT-related and non-IT issues, auditor integration of input from IT specialists may differ across these issue types. The advice-taking literature provides evidence that individuals attempt to consider both their own and the advisor's expertise when evaluating advice (Harvey and Fischer 1997). Auditors likely view IT-related (non-IT) issues as outside (within) their own domain of expertise and feel less (more) comfortable evaluating these issues without input from IT specialists. I expect ICFR issue type to interact with the strength of auditors' team identity with IT specialists (i.e., the extent to which auditors view the IT specialists as part of their team) in influencing how auditors use IT specialists' input.

Team identity strength will likely influence auditors' recognition of IT specialists' relative expertise. Individuals extend their own attitudes, beliefs, and attributes to in-group members (i.e., those with whom one holds a relevant, salient identity) and view in-group members as trustworthy (Wilder 1984; Haslam and Ellemers 2005). Differences in attitudes, beliefs, and attributes are assumed and salient for out-group members (Brewer and Gardner 1996; Zellmer-Bruhn, Maloney, Bhappu, and Salvador 2008). For auditors who have a weaker team identity, I expect that the emphasis on differences in personal attributes between the auditor and IT specialist will increase the prominence of the IT specialist's expertise. For auditors with a stronger team identity, I expect that the emphasis on similarities in personal attributes between the auditor and IT specialist will decrease the prominence of the IT specialist's expertise. This disparity in expertise recognition based on strength of team identity will translate into input weighting differences across ICFR issue types. Specifically, for IT-related issues, I predict that auditors with a weaker versus stronger team identity will weight input from the IT specialist more heavily because the IT specialist's expertise will be more apparent to them. For non-IT issues, I predict the opposite: auditors with a stronger team identity will weight input more heavily because they project their own attributes onto and have higher trust in the IT specialist.

The quality of the input received can qualify the previous prediction, especially for non-IT issues. Auditors can more readily assess the quality of input for non-IT issues than for IT-related issues. However, auditors with a stronger team identity will rely on their trust in the IT specialist and heavily weight input received regardless of input quality. In contrast, auditors with a weaker team identity will not have this trust to rely on, and they will be more likely to critically evaluate input received and recognize differences in input quality. Thus, I predict a trust heuristic pattern of input weight for non-IT issues (Kadous, Leiby, and Peecher 2013): auditors with a weaker team identity will place heavier weight on higher versus lower quality input, but auditors with a stronger team identity will weight input relatively heavily regardless of quality. Given that I expect auditors to view IT-related issues as outside their domain of expertise and to feel uncomfortable evaluating these issues without IT specialist input, I do not have expectations about their ability or willingness to differentiate levels of input quality for IT-related issues.

I test my predictions in an experiment in which experienced auditor participants complete four client cases with ICFR issues adapted from Earley et al. (2008). I manipulate two variables between participants: team identity strength with the IT specialist (stronger versus weaker) and input quality in terms of whether the input reflects the true severity of the issue (higher versus lower). Two client cases contain ICFR issues that are IT-related and two client cases contain ICFR issues that are not IT-related, with a more and less severe case for each type. As reporting consequences differ across levels of ICFR issue severity, I include more and less severe cases to examine whether my predictions hold across these levels.

Participants provide initial severity ratings and classification judgments for the ICFR issues in each case. They then read a description of an IT specialist who will be providing input on each case, with wording to induce either a stronger or weaker team identity with the IT specialist. Participants receive the IT specialist input in the form of rating and classification recommendations. This input either matches (higher quality) or does not match (lower quality) the designed severity level of the case. Participants then make their final judgments. The main dependent variable, weight of input, measures the extent to which auditors revised their judgments in the direction of the IT specialist's input (Tost, Gino, and Larrick 2012).

As predicted, I find a significant interaction of team identity and issue type for weight of input. For IT-related issues, auditors with a weaker versus stronger team identity weight the IT specialist's input more heavily, whereas for non-IT issues, auditors with a stronger versus weaker team identity weight input from an IT specialist more heavily. Further, for non-IT issues, auditors with a stronger team identity weight input relatively heavily regardless of input quality, consistent with a trust heuristic. Auditors with a weaker team identity place heavier weight on higher (lower) quality input for the more (less) severe non-IT issue. Because the higher (lower) quality input for the more (less) severe issue indicates that the issue is more severe, these results provide evidence of auditor conservatism. This conservatism is functional from an audit effectiveness perspective, as it only results in inappropriate weighting for the less severe issue.

My study provides important contributions to auditing research and practice. I identify an aspect of the audit team environment—audit team identity strength—that moderates the likelihood of auditors benefiting from an internal firm specialist's expertise. Firms are encouraging auditors and internal firm specialists to embrace a one-team view consistent with a stronger team identity (Ernst & Young [EY] 2013; Bauer and Estep 2019). While ample evidence indicates that a stronger team identity has benefits (e.g., improved team learning; Van Der Vegt and Bunderson 2005), my study highlights a benefit to weaker team identity: weaker team identity emphasizes the IT specialist's expertise, resulting in heavier weight of the specialist's input on IT-related issues. Thus, my study cautions that firms' one-team approach may not resolve current problems with auditors' use of IT specialists' work (PCAOB 2012; IFIAR 2017) and may even create new problems. My findings motivate future research on conditions under which better input integration can occur with stronger team identities. The use of decision aids or better communication in the planning stages of the audit (Bauer and Estep 2019) may improve input integration while capitalizing on the benefits of a strong team identity.

My study builds on advice-taking research by examining input weighting in the auditor-IT specialist context for ICFR issues. This setting contains key aspects that differentiate my study from prior work. Input is provided: (1) by an individual from a different domain of expertise, and (2) for different types of issues (IT-related and non-IT). Examining advice-taking in a setting with these aspects is important because cross-functional audit teams comprised of auditors and internal firm specialists with various domains of expertise (e.g., IT, tax, valuation) are critical to the performance of audits in complex business environments (Griffith, Hammersley, and Kadous 2015; Boritz, Kochetova-Kozloski, Robinson, and Wong 2017; Griffith 2018, 2020; Bauer and Estep 2019). Further, my predictions of differing input weighting patterns across issue types are based on my expectations that auditors see IT-related issues as more outside their domain of expertise than non-IT issues and feel less comfortable evaluating these issues without IT specialist input. Future research can investigate whether similar situations exist in other auditor-specialist areas, where internal firm specialists could serve as a source of input across different types of issues.

Section II of this paper provides theory and hypotheses development. Section III describes the experimental design. Section IV discusses results, and Section V concludes.

### IT Specialist Input on ICFR Issues

Internal control over financial reporting (ICFR) issues indicate failures in internal control and can have serious implications for financial reporting quality. The classification of an ICFR issue as a control deficiency, significant deficiency, or material weakness impacts audit quality. The level of severity of the ICFR issue determines the degree to which the planned audit procedures need to be altered and the nature of the disclosure of the ICFR issue (PCAOB 2007). Insufficient adjustments to planned audit procedures in light of ICFR issues can result in insufficient evidence to support audit conclusions. That said, correctly classifying an ICFR issue is difficult (Earley et al. 2008).1 Regulators have expressed concern that auditors have difficulty assessing ICFR deficiencies, resulting in underreporting of material weaknesses (Croteau 2013; Franzel 2014). Further, inspections frequently identify deficiencies in the auditing of ICFR (PCAOB 2012; IFIAR 2017). Given these concerns, a better understanding of auditor decision processes for evaluating ICFR issues is needed. In this study, I investigate how auditors integrate IT specialist input on ICFR issues.2

ICFR issues can be either IT-related or non-IT, depending on whether the deficient controls and processes have IT elements. Input from IT specialists on IT-related issues is especially valuable, as auditors likely view IT-related issues as falling outside their own domain of expertise and feel uncomfortable evaluating these issues without IT specialist input. Prior research also provides evidence that auditors have difficulty evaluating IT-related risks and control deficiencies. For example, auditors are susceptible to management persuasion tactics for IT, but not manual control issues (Wolfe, Mauldin, and Diaz 2009). Auditors with lower self-assessed IT expertise assess control risk lower than auditors with higher self-assessed IT expertise (Brazel and Agoglia 2007). Auditors often lack even a base level of training in IT areas and may not understand the purpose and scope of IT specialists' work and, thus, what level of reliance is appropriate (PCAOB 2012; Bauer and Estep 2019). In fact, PCAOB inspection reports cite specific instances of audit teams failing to sufficiently evaluate the severity and impact of IT-related ICFR issues (e.g., PCAOB 2013, 2015).

IT specialists' work is not limited to IT-related risks and controls. IT specialists sometimes perform the work for entire control processes on audits (Bauer and Estep 2019) and help clients design control procedures (both IT and non-IT) on advisory engagements. IT specialists provide input on all types of control issues, as they typically review the list of identified control deficiencies and related classifications before signing off on the audit file. Thus, IT specialists are a credible source of input on both IT-related and non-IT ICFR issues. However, I expect that auditors' perceptions of the strength of team identity with IT specialists, or the extent to which auditors view the IT specialists as part of the audit team, will influence the integration of IT specialist input on ICFR issues.3

### Auditors' Team Identity with IT Specialists

Team identity is a particularly salient and influential social identity in the integrated audit context, as lower-level identities (e.g., department, workgroup) can have more influence on behavior than higher-level identities (e.g., professional and organizational) (Ashforth and Mael 1989; Ashforth and Johnson 2001). A number of factors related to group formation can influence the strength of a social identity for an individual (e.g., similarity, interpersonal interaction, liking, proximity; Ashforth and Mael 1989). In the audit context, despite firms pushing for a one-team view in which audit specialists are included in the audit team (EY 2013), Bauer and Estep (2019) provide evidence that the extent to which auditors see IT specialists as part of the audit team (i.e., auditors' strength of team identity with IT specialists) varies across audit engagements.4 IT specialists are frequently included on audit engagements (Bauer and Estep 2019) and auditors and IT specialists interact throughout the year on these engagements, including being on-site at the client together. However, IT specialists have varying educational backgrounds and often are not trained as accountants (Brazel and Agoglia 2007; Bauer and Estep 2019). Further, prior studies show that auditors dispute the value of IT specialists and, in fact, often view IT specialists as “budget busters” (Curtis, Jenkins, Bedard, and Deis 2009; Bauer and Estep 2019). The quality of relationships between auditors and IT specialists has improved in recent years, and auditors are better able to see the value in IT specialists and their input on audits; however, the two still struggle to work together effectively (Bauer and Estep 2019).

While prior research often identifies benefits of stronger (versus weaker) team identities (e.g., valuing input more, Dovidio, Gaertner, and Validzic [1998] and Van Der Vegt and Bunderson [2005]; greater coordination, Towry [2003]; reduction in self-serving bias, King [2002]), a stronger team identity does not always lead to better outcomes. A stronger social identity can cause members to place too much trust in information from an in-group member (Dukerich, Kramer, and Parks 1998). Kadous et al. (2013) find that auditors weight advice from a stronger social bond peer similarly regardless of whether strong or weak justification (i.e., higher or lower input quality) accompanies the advice; they refer to this tendency as a trust heuristic. When a weaker social bond exists, the weight that auditors put on advice depends on advice quality. In a survey-based field study, Levin and Cross (2004) find that weaker social ties are associated with the receipt of useful knowledge after controlling for competence- and benevolence-based trust. Given that a stronger or weaker team identity is not always superior, I build my hypotheses around two key aspects that differentiate my study from prior work: (1) the individual providing input is from a different domain of expertise, and (2) the ICFR setting allows examination of input weighting on different types of issues (IT-related and non-IT).

### Weight of Input: Influence of Team Identity Depends on Issue Type

While IT specialists are a credible source of input for both IT-related and non-IT issues, auditors will view IT-related (non-IT) issues as outside (within) their own domain of expertise and feel less (more) comfortable evaluating these issues without input from IT specialists. In fact, one might expect auditors to defer to IT specialist judgment on IT-related issues, given IT specialists' expertise in this area, and incorporate a token amount of input on non-IT issues. However, while the advice-taking literature shows that individuals take both their own and the advisor's expertise into account, assessing and recognizing an advisor's differential expertise can be difficult (Libby, Trotman, and Zimmer 1987; Harvey and Fischer 1997; Bonaccio and Dalal 2006). Auditors likely lack sufficient information to assess the true level of an IT specialist's expertise (Brazel and Agoglia 2007). I expect that the strength of auditors' team identity with an IT specialist will influence their perceptions of the IT specialist's relative expertise.

Social identities influence the perceptions of in- and out-group members (Tajfel 1978; Ashforth and Mael 1989). Individuals see in-group members as similar to themselves and out-group members as different from themselves, even without evidence to support these perceived similarities and differences (Wilder 1984). Said differently, the mere inclusion of another individual in one's in-group results in assumed and salient similarities in attitudes, beliefs, and other attributes with oneself, and the mere categorization of someone in an out-group produces assumed and salient differences in personal attributes (Hensley and Duval 1976; Ashforth and Mael 1989; Brewer and Gardner 1996; Zellmer-Bruhn et al. 2008).

In the context of the current study, I expect that the salience of differences (similarities) with the IT specialist depends on auditors' team identity: auditors with a weaker (stronger) team identity will emphasize (deemphasize) the relative IT expertise of the specialist. This heightened awareness of the IT specialist's expertise by auditors with a weaker versus stronger team identity will result in different weighting of input across types of ICFR issues. For IT-related issues, auditors with a weaker team identity will weight input from the IT specialist more heavily than auditors with a stronger team identity. For non-IT issues, I expect the opposite: auditors with a stronger team identity who assume that the IT specialist is more similar to themselves will more heavily weight input from the IT specialist than those with a weaker team identity. Thus, I predict that issue type and team identity will interact to influence auditor weight on IT specialist input, as stated in the following hypothesis:

H1:

When auditors receive input from an IT specialist on IT-related ICFR issues, auditors with a weaker team identity will weight input more heavily than auditors with a stronger team identity. For input received on non-IT ICFR issues, auditors with a stronger team identity will weight input more heavily than auditors with a weaker team identity.

### Weight of Input: The Role of Input Quality

Evaluating ICFR issues is difficult. Moreover, IT specialists vary in ability, as well as knowledge and training (Brazel and Agoglia 2007). Thus, the quality of IT specialists' input will vary, and input provided may not reflect the true severity of the ICFR issue being evaluated. I expect that the quality of the input will qualify the H1 prediction for non-IT issues, but I do not have this expectation for IT-related issues.5 Auditors should be able to critically evaluate input received on non-IT issues, but this evaluation requires cognitive effort. Research in psychology and auditing shows that individuals simplify cognitive processing by relying on other heuristic cues, such as team identity (Tversky and Kahneman 1974; Kadous et al. 2013).

I expect a similar pattern of input weight to occur for non-IT ICFR issues, although the mechanism underlying the propensity to trust differs. Consistent with H1 and social identity theory, I expect that auditors with a stronger team identity will have a propensity to trust due to projecting their own attributes onto the IT specialist and the general tendency to view in-group members as trustworthy (Haslam and Ellemers 2005). Despite auditors likely having the knowledge to evaluate the quality of the input received on non-IT ICFR issues, the presence of trust in the specialist for auditors with a stronger team identity will allow them to rely on this trust and weight input relatively heavily regardless of input quality. The absence of trust in the specialist for auditors with a weaker team identity will motivate them to critically evaluate the input received and, using their relevant knowledge, place heavier weight on higher versus lower quality input. Thus, for non-IT issues, I expect a trust heuristic pattern to emerge for auditors weighting input from IT specialists, as described in the following hypothesis:

H2:

When auditors receive input from an IT specialist on non-IT ICFR issues, auditors with a weaker team identity will weight higher versus lower quality input more heavily, but auditors with a stronger team identity will weight input relatively heavily regardless of input quality.

### ICFR Issue Severity: Will Weight of Input Patterns Differ?

It is important to consider whether the predicted patterns of input weighting will hold across levels of ICFR issue severity because reporting consequences differ across these levels. Control deficiencies (the least severe level) are documented in the workpapers and discussed with management, while significant deficiencies must also be disclosed to the audit committee, and material weaknesses (the most severe level) are disclosed publicly as these issues require an adverse ICFR opinion (PCAOB 2007).

My predictions may hold across severity levels if auditors approach the evaluation of all ICFR issues with a similar attitude and desire for accuracy, given the potential for negative consequences of misclassifying ICFR issues. However, the negative consequences of misclassifying an ICFR issue as less severe than it actually is (e.g., classifying a material weakness as a significant deficiency or control deficiency) differ from the negative consequences of misclassifying an ICFR issue as more severe than it actually is (e.g., classifying a control deficiency as a significant deficiency or material weakness). The former reduces audit effectiveness and can result in regulator action, litigation, and/or reputation effects. The latter reduces audit efficiency and can result in lower profit. Auditors are likely more concerned about effectiveness than efficiency. Further, auditor conservatism, or the tendency to give more attention to, and be more influenced by, negative information or outcomes (Smith and Kida 1991), suggests that auditors may anchor on input that signals the issue is more severe, regardless of the quality of the input. For example, auditors may heavily weight input from an IT specialist that indicates an ICFR issue is very severe (e.g., input indicating the issue is a material weakness), even if the input is of poor quality and the actual severity of the issue is low (e.g., a control deficiency). Given the potential for asymmetric reactions, I investigate whether my predictions hold across levels of ICFR issue severity.

### Participants

Practicing auditors were recruited for participation through the Center for Audit Quality (CAQ) and American Accounting Association (AAA) Access to Audit Personnel program. A total of 101 participants from the eight CAQ member firms completed the experimental instrument via the online survey tool Qualtrics.7 Of the 101 participants, 79 are Big 4 auditors and 22 are non-Big 4 auditors. Participants are senior-level auditors with an average of 46 months of experience.8 Seventy-seven participants are CPAs, none are Certified Information Systems Auditors (CISAs; a common certification for IT specialists), and participants spend 35 percent of their time, on average, on Sarbanes-Oxley (SOX) related work.9

My study employs a mixed design, whereby I manipulate team identity and input quality at two levels each between participants. All participants complete four client cases with ICFR issues adapted from Earley et al. (2008). I instruct participants to evaluate each case independently.10 Two cases contain no IT components and two cases contain IT components, with a more and less severe case for each type. Refer to Appendix A for case descriptions.

Auditors assume the role of the lead auditor on the client engagements and perform initial ICFR issue judgments for the four cases. The materials inform participants that the lead IT specialist on the engagements has reviewed the client cases and present a description of the IT specialist. The between-participants manipulation of stronger versus weaker team identity takes place within this description and is explained in a later section. I reinforce the manipulation via an open-ended question, asking participants to consider what it would be like to work with this IT specialist and to enter any thoughts that come to mind.

After completing this open-ended question, participants are asked to provide final judgments for each client case, this time receiving input from the IT specialist. I manipulate input quality (higher versus lower) between participants by varying the input provided by the IT specialist; the details of this manipulation are provided in a later section.

For each case, participants are reminded of their initial rating, receive IT specialist input, and make final judgments. Next, participants answer task-related questions, including a team identity manipulation check, a rating of the quality of the IT specialist input, and assessments of the competence, trustworthiness, and objectivity of the IT specialist. I also collect additional post-test measures and professional background information.

### Independent Variables: Within Participants

#### Issue Type

As mentioned previously, two of the four cases completed by participants contain IT components and two cases do not (i.e., two cases are IT-related and two cases are non-IT; see Appendix A). My predictions for the IT-related versus non-IT issues are based on auditors seeing IT-related issues as more outside their domain of expertise than non-IT issues and feeling less comfortable evaluating the issue without help or input from an IT specialist. To verify that the cases capture these aspects, I conducted a separate survey with 15 auditor participants with an average of four years of experience. I asked the auditors three questions related to each case (all on 11-point scales with endpoints of 0 = Not at all and 10 = Very): (1) “To what extent is this issue IT-related?” (2) “How comfortable would a typical auditor be evaluating this issue without receiving input from an IT specialist?” and (3) “To what extent would a typical auditor view this issue as outside his/her domain of expertise?”11

Results (untabulated) are consistent with my expectations. Mean ratings are significantly higher for the IT-related versus non-IT cases for questions (1) and (3), indicating that auditors recognize the IT-related cases as more IT-related (6.42 versus 2.29; t14 = 6.07, p < 0.001) and further outside an auditor's domain of expertise than the non-IT cases (3.20 versus 1.64; t13 = 5.33, p < 0.001). Mean ratings are significantly lower for the IT-related versus non-IT cases for question (2), indicating that auditors would be less comfortable evaluating the IT-related versus non-IT cases without input from an IT specialist (5.55 versus 8.11; t13 = −5.70, p < 0.001). To allow participants to directly compare the cases, I asked them to rank order the four cases from 1 to 4—most IT-related to least IT-related. On average, the IT-related versus non-IT cases were ranked as significantly more IT-related (2.04 versus 2.96; t12 = −2.52, p = 0.027). These results indicate a successful manipulation of issue type.

#### Severity

Participants complete a more and less severe case for each issue type. Earley et al. (2008) designed the more (less) severe cases to fall between a significant deficiency and material weakness (control deficiency and significant deficiency). See Appendix A for case details.

### Independent Variables: Between Participants

#### Team Identity

Prior empirical research on team identity has typically taken one of two approaches to investigate the team identity construct: (1) measuring team identity in a field study (e.g., Van Der Vegt and Bunderson 2005), or (2) manipulating team identity in a lab experiment using students (e.g., Brewer and Gardner 1996; Dovidio et al. 1998; Kane 2010). Manipulations can be visual (e.g., shared color of name tags), linguistic (e.g., use of them versus us), and/or include a common fate (e.g., experimental payout determined by team versus individual actions). I chose to manipulate, rather than measure, team identity in order to allow for strong causal inferences. I could not use prior manipulations because my participants are working professionals completing the experimental materials during firm training on their own computers. Therefore, I designed the following team identity manipulation to elicit stronger versus weaker perceptions of team identity by activating team-related and integration concepts.

I induce a stronger versus weaker team identity between participants through three pieces of information about the IT specialist. First, the IT specialist particularly enjoys (does not particularly enjoy) thinking about financial statement audit issues as compared to advisory/consulting issues for the stronger (weaker) condition. Second, the IT specialist appears to want to be a helpful (only technically a) member of the team and tries to sit with (sits away from) the audit team while on-site for the stronger (weaker) condition. Finally, the participant views the IT specialist as a core (just obligatory) member of the team for stronger (weaker) team identity.12 These pieces of information map into factors associated with the strength of team identity: attitude similarity (Hensley and Duval 1976), implied proximity and interaction frequency, and perceptions of team membership (Hogg and Turner 1985; Ashforth and Mael 1989).13

#### Input Quality

I manipulate input quality between participants by varying the input provided by the IT specialist to match or not match the designed severity level of the case, i.e., the input is of higher or lower quality, respectively (see Appendix B). Participants receive input from the IT specialist for each case that includes both a rating (corresponding to an 11-point scale labeled “Control deficiency” [1], “Significant deficiency” [6], and “Material weakness” [11]) and a classification of the issue.14 For the two more severe cases, participants in the higher (lower) quality condition receive from the IT specialist a rating of 8.5 (3.5) and a classification that lies between a significant deficiency and material weakness (control deficiency and significant deficiency). For the two less severe cases, I reverse these input ratings and classifications; participants in the higher (lower) quality condition receive a rating of 3.5 (8.5) and a classification that lies between a control deficiency and significant deficiency (significant deficiency and material weakness).15 Refer to Appendix C for an overview of the experimental design.

### Dependent Variable

Participants perform two initial classification assessments for each case: (1) rating ICFR issue severity on an 11-point scale labeled “Control deficiency” [1], “Significant deficiency” [6], and “Material weakness” [11], and (2) classifying the ICFR issue using a forced-choice scale (Control deficiency, Significant deficiency, or Material weakness). Elicited final judgments include the same severity rating and ICFR issue classification judgments as in the initial judgments, as well as the rationale for judgments (open-ended). The weight placed by auditors on the IT specialist's input in their final ICFR issue severity ratings is the key dependent variable. Following prior literature, weight of input is calculated using the formula below (Yaniv 2004; Bonaccio and Dalal 2006; Kadous et al. 2013):
$$\def\upalpha{\unicode[Times]{x3B1}}$$$$\def\upbeta{\unicode[Times]{x3B2}}$$$$\def\upgamma{\unicode[Times]{x3B3}}$$$$\def\updelta{\unicode[Times]{x3B4}}$$$$\def\upvarepsilon{\unicode[Times]{x3B5}}$$$$\def\upzeta{\unicode[Times]{x3B6}}$$$$\def\upeta{\unicode[Times]{x3B7}}$$$$\def\uptheta{\unicode[Times]{x3B8}}$$$$\def\upiota{\unicode[Times]{x3B9}}$$$$\def\upkappa{\unicode[Times]{x3BA}}$$$$\def\uplambda{\unicode[Times]{x3BB}}$$$$\def\upmu{\unicode[Times]{x3BC}}$$$$\def\upnu{\unicode[Times]{x3BD}}$$$$\def\upxi{\unicode[Times]{x3BE}}$$$$\def\upomicron{\unicode[Times]{x3BF}}$$$$\def\uppi{\unicode[Times]{x3C0}}$$$$\def\uprho{\unicode[Times]{x3C1}}$$$$\def\upsigma{\unicode[Times]{x3C3}}$$$$\def\uptau{\unicode[Times]{x3C4}}$$$$\def\upupsilon{\unicode[Times]{x3C5}}$$$$\def\upphi{\unicode[Times]{x3C6}}$$$$\def\upchi{\unicode[Times]{x3C7}}$$$$\def\uppsy{\unicode[Times]{x3C8}}$$$$\def\upomega{\unicode[Times]{x3C9}}$$$$\def\bialpha{\boldsymbol{\alpha}}$$$$\def\bibeta{\boldsymbol{\beta}}$$$$\def\bigamma{\boldsymbol{\gamma}}$$$$\def\bidelta{\boldsymbol{\delta}}$$$$\def\bivarepsilon{\boldsymbol{\varepsilon}}$$$$\def\bizeta{\boldsymbol{\zeta}}$$$$\def\bieta{\boldsymbol{\eta}}$$$$\def\bitheta{\boldsymbol{\theta}}$$$$\def\biiota{\boldsymbol{\iota}}$$$$\def\bikappa{\boldsymbol{\kappa}}$$$$\def\bilambda{\boldsymbol{\lambda}}$$$$\def\bimu{\boldsymbol{\mu}}$$$$\def\binu{\boldsymbol{\nu}}$$$$\def\bixi{\boldsymbol{\xi}}$$$$\def\biomicron{\boldsymbol{\micron}}$$$$\def\bipi{\boldsymbol{\pi}}$$$$\def\birho{\boldsymbol{\rho}}$$$$\def\bisigma{\boldsymbol{\sigma}}$$$$\def\bitau{\boldsymbol{\tau}}$$$$\def\biupsilon{\boldsymbol{\upsilon}}$$$$\def\biphi{\boldsymbol{\phi}}$$$$\def\bichi{\boldsymbol{\chi}}$$$$\def\bipsy{\boldsymbol{\psy}}$$$$\def\biomega{\boldsymbol{\omega}}$$$$\def\bupalpha{\bf{\alpha}}$$$$\def\bupbeta{\bf{\beta}}$$$$\def\bupgamma{\bf{\gamma}}$$$$\def\bupdelta{\bf{\delta}}$$$$\def\bupvarepsilon{\bf{\varepsilon}}$$$$\def\bupzeta{\bf{\zeta}}$$$$\def\bupeta{\bf{\eta}}$$$$\def\buptheta{\bf{\theta}}$$$$\def\bupiota{\bf{\iota}}$$$$\def\bupkappa{\bf{\kappa}}$$$$\def\buplambda{\bf{\lambda}}$$$$\def\bupmu{\bf{\mu}}$$$$\def\bupnu{\bf{\nu}}$$$$\def\bupxi{\bf{\xi}}$$$$\def\bupomicron{\bf{\micron}}$$$$\def\buppi{\bf{\pi}}$$$$\def\buprho{\bf{\rho}}$$$$\def\bupsigma{\bf{\sigma}}$$$$\def\buptau{\bf{\tau}}$$$$\def\bupupsilon{\bf{\upsilon}}$$$$\def\bupphi{\bf{\phi}}$$$$\def\bupchi{\bf{\chi}}$$$$\def\buppsy{\bf{\psy}}$$$$\def\bupomega{\bf{\omega}}$$$$\def\bGamma{\bf{\Gamma}}$$$$\def\bDelta{\bf{\Delta}}$$$$\def\bTheta{\bf{\Theta}}$$$$\def\bLambda{\bf{\Lambda}}$$$$\def\bXi{\bf{\Xi}}$$$$\def\bPi{\bf{\Pi}}$$$$\def\bSigma{\bf{\Sigma}}$$$$\def\bPhi{\bf{\Phi}}$$$$\def\bPsi{\bf{\Psi}}$$$$\def\bOmega{\bf{\Omega}}$$$${{\left| {Final\,rating - Initial\,rating} \right|} \over {\left| {IT\,specialist\,rating - Initial\,rating} \right|}}$$

Weight of input provides an appropriate test of theory as it measures the extent to which auditors revised their initial ratings in the direction of the IT specialist's rating (Tost et al. 2012). Since participants receive different IT specialist ratings across conditions and cases (see Appendices B and C), it is important to use a measure that incorporates the input received, as well as the initial and final ratings. Weight of input can be thought of as a percentage shift measure (Harvey and Fischer 1997), where a value of 0 indicates no change in the auditors' rating between the initial and final judgment (i.e., no integration of the IT specialist's input), while a weight of input of 1 indicates that the auditor matched (or shifted 100 percent to) the IT specialist's input. A value of 0.50 indicates some integration of the IT specialist's input as the auditor assigned equal weight (or 50 percent) to his or her initial rating and to the IT specialist's rating.

While well-established, the weight of input measure is subject to limitations and requires careful review of the data (Yaniv 2004; Bonaccio and Dalal 2006; Gino and Moore 2007). First, the weight of input measure does not differentiate between judgments moving away from versus toward input. For example, in one observation, the initial rating is 9, input is 3.5, and the final rating is 10.5; the weight of input value is 0.27, suggesting higher weighting even though the individual is moving away from input. Moving away from input occurs for 23 observations (∼5 percent), and I adjust these weight of input values to zero. Second, weight of input has a lower bound of zero, but no upper bound, and when a final judgment exceeds the input received, the weight of input value is greater than 1. For example, for one observation, the initial rating is 6, input is 8.5, and the final judgment is 11, and a weight of input of 2. This outcome occurs for 19 observations (∼5 percent) and I truncate these weight of input values to 1.16 Third, the weight of input cannot be calculated for four observations (< 1 percent). In three instances, the participant's initial rating equals the IT specialist rating; in one instance, an initial rating was not provided.17 The number of undefined and adjusted observations does not significantly differ across between-participants conditions (Stronger Team Identity/Higher Quality Input: 11; Stronger Team Identity/Lower Quality Input: 12; Weaker Team Identity/Higher Quality Input: 12; Weaker Team Identity/Lower Quality Input: 11, p = 0.989) or within-participants cases (More severe, IT-related: 10; More severe, Non-IT: 9; Less severe, IT-related: 10; Less severe, Non-IT: 17, p = 0.294) using Fisher's exact test.

### Manipulation and Recall Checks

To test the effectiveness of my team identity manipulation, I use the Inclusion of Other in the Self scale (A. Aron, E. Aron, and Smollan 1992), a validated measure of identity (Tropp and Wright 2001) that has been used in prior identity-related audit research (Bauer 2015). I present participants with images of two overlapping circles labeled “Self” and “IT specialist.” Seven variations are presented, coded 1 through 7, ranging from no overlap (weakest identity—coded as 1) to near-complete overlap (strongest identity—coded as 7). Participants select the image that best describes how “your personal attributes, qualities, and values align or overlap with the attributes, qualities, and values of the IT specialist who provided the input to you.” Mean ratings are significantly higher for the stronger versus weaker team identity condition (4.57 versus 3.64, t94 = 3.56, p < 0.001, one-tailed).18 I also ask participants a team-related recall check: “In terms of being a member of the team, you view the IT specialist who provided the input as” rated on an 11-point scale with endpoints of −5—More obligatory than core and 5—More core than obligatory. Mean ratings are significantly higher for the stronger versus weaker team identity condition (1.85 versus −0.17, t99 = 4.71, p < 0.001, one-tailed).

To verify the within-participants case severity manipulation, I estimate a repeated measures ANOVA on the initial issue ratings (untabulated). Only a main effect of Severity is significant (F1, 96 = 13.34, p < 0.001), and the mean ratings are higher for the two more severe cases than for the less severe cases (More severe, IT-related: 6.43 and More severe, Non-IT: 6.75 versus Less Severe, IT-related: 5.86 and Less severe, Non-IT: 5.67).

### Tests of H1–H2

Table 1, Panel A provides descriptive statistics for the weight of input employed across conditions. Figure 1 graphically depicts the means by case. I test H1–H2 based on the repeated measures ANOVA reported in Table 1, Panel B.

TABLE 1

Weight of Input from IT Specialist

#### Weight of Input: Results Graph by Case

FIGURE 1
FIGURE 1

Panel A: IT-Related Cases

Panel A: IT-Related Cases

Close modal

Panel B: Non-IT Cases

Input Quality and Team Identity are manipulated at two levels each between participants: higher versus lower and stronger versus weaker, respectively. The dependent variable, weight of input, is calculated via the following formula: |Final rating − Initial rating| / |IT specialist rating − Initial rating|, where participants provide initial and final rating judgments on an 11-point scale with labels of “Control deficiency” [1], “Significant deficiency” [6], and “Material weakness” [11].

Panel B: Non-IT Cases

Input Quality and Team Identity are manipulated at two levels each between participants: higher versus lower and stronger versus weaker, respectively. The dependent variable, weight of input, is calculated via the following formula: |Final rating − Initial rating| / |IT specialist rating − Initial rating|, where participants provide initial and final rating judgments on an 11-point scale with labels of “Control deficiency” [1], “Significant deficiency” [6], and “Material weakness” [11].

Close modal

#### Tests of H1

H1 predicts an interaction between Team Identity and Issue Type. Per Table 1, Panel A, for IT-related issues, the mean weight of input is higher for weaker versus stronger team identity conditions (0.37 versus 0.30). For non-IT issues, the mean weight of input is lower for weaker versus stronger team identity conditions (0.31 versus 0.40). This pattern is significant as indicated by the Issue TypeTeam Identity interaction in Table 1, Panel B (p = 0.013). When considering the weight of input across Issue Type, it is surprising that auditors with a stronger team identity weight input from IT specialists more heavily for non-IT issues than IT-related issues (0.40 versus 0.30). Common advice-taking strategies include averaging and shifting a token amount of weight (e.g., 20 to 30 percent) to an advisor's input (Bonaccio and Dalal 2006). Stronger team identity auditors may have employed closer to an averaging approach for input utilization on non-IT issues and placed a token amount of weight on input for IT-related issues. I leave further exploration of this finding to future research.

I further examine H1 by investigating the effect of team identity by case. For the more severe IT-related case, as expected, auditors in the weaker versus stronger team identity conditions weight input from the IT specialist more heavily (0.41 versus 0.28; p = 0.031, one-tailed; see Table 2, Panel A). H1 is not supported for the less severe IT-related case, as weight of input does not differ across team identity conditions (M = 0.33 for both conditions; p = 0.498, one-tailed; see Table 2, Panel B). For the non-IT cases, mean weight of input is directionally lower for weaker versus stronger team identity auditors in both the more (0.33 versus 0.38) and less severe (0.29 versus 0.43) cases, consistent with H1. However, the difference between the identity conditions is significant for the less severe case (p = 0.038, one-tailed; see Table 3, Panel B) and not the more severe case (p = 0.180, one-tailed; see Table 3, Panel A).19

TABLE 2

Weight of Input: IT-Related Cases

TABLE 3

Weight of Input: Non-IT Cases

I also perform simple effects tests of the between-participants conditions for the IT-related cases (see Table 2). For the more severe IT-related case (see Table 2, Panel A), when comparing weight of input across identity conditions within each level of the input quality, only the Weaker Team Identity/Higher Quality Input versus Stronger Team Identity/Higher Quality Input difference is significant (0.52 versus 0.34; p = 0.046); this result is consistent with H1. While I do not make predictions regarding the effect of input quality for IT-related issues, I report these simple effects for completeness. Interestingly, auditors in the weaker team identity condition weight higher versus lower input quality significantly heavier (0.52 versus 0.29, p = 0.014), but auditors in the stronger team identity condition do not (0.34 versus 0.22, p = 0.238). For the less severe IT-related case (see Table 2, Panel B), no simple effects are significant.

#### Tests of H2

H2 predicts that for non-IT ICFR issues, auditors in the weaker team identity conditions will weight higher versus lower quality input more heavily, but auditors in the stronger team identity conditions will weight input relatively heavily regardless of input quality. Thus, the pattern I predict is consistent with a traditional ordinal interaction, wherein the weight of input for the Weaker Team Identity/Lower Quality Input condition is lower than the other three conditions. As a first step in testing H2, I evaluate the visual fit for the non-IT cases (Guggenmos, Piercey, and Agoglia 2018); see Figure 1, Panel B. The pattern for the more severe non-IT case is consistent with my prediction, while the pattern for the less severe case is not. Therefore, I evaluate these cases separately.

For the more severe non-IT case (see Table 3, Panel A), I estimate a contrast with weights (−3, +1, +1, +1) corresponding to: Weaker Team Identity/Lower Quality Input, Weaker Team Identity/Higher Quality Input, Stronger Team Identity/Lower Quality Input, Stronger Team Identity/Higher Quality Input, respectively. The contrast is significant (p < 0.001, one-tailed). Simple effects tests are also consistent with H2 for the more severe non-IT case; refer to Table 3, Panel A. Auditors with a weaker team identity weight input significantly heavier when receiving higher versus lower quality input (0.48 versus 0.16; p < 0.001, one-tailed). Auditors with a stronger identity do not differentially weight input across input quality conditions (Higher: 0.42 versus Lower: 0.33; p = 0.357). H2 is supported for the more severe non-IT case.

For the less severe non-IT case, the pattern of results (see Figure 1, Panel B) does not support H2. While there is evidence of a trust heuristic pattern (heavier weighting with a stronger team identity regardless of input quality and differential weighting with a weaker team identity across input quality levels; Kadous et al. 2013), auditors with a weaker team identity weight lower rather than higher quality input more heavily in this case, inconsistent with my prediction. Recall from the design of the input quality manipulation that for less severe cases, lower quality input indicates that the issue is more severe than higher quality input. Heavier weight on the lower quality input is consistent with conservatism. This conservative tendency is significant across all four cases, as indicated by the two-way interaction of SeverityInput Quality in the repeated measures ANOVA reported in Table 1, Panel B (p < 0.001). For the more (less) severe cases, auditors place heavier weight on input in the higher (lower) quality input condition.

Given that the pattern for the less severe non-IT issue is not consistent with H2, I do not test for the planned contrast and only examine the simple effects, which also provide evidence of the trust heuristic (see Table 3, Panel B). Auditors in the stronger team identity condition do not differentiate between input quality levels (Higher: 0.39 versus Lower: 0.47; p = 0.447). Auditors with a weaker team identity receiving lower quality input weight that input significantly heavier than weaker team identity auditors receiving higher quality input (0.44 versus 0.16; p = 0.008).

#### Summary of Findings

Overall, I find support for my predictions, although only for the more severe ICFR issues. For the more severe IT-related issue, I find that weaker versus stronger team identity auditors weight input from an IT specialist more heavily. For the non-IT issues, I find evidence consistent with the trust heuristic pattern across both levels of severity (Kadous et al. 2013). Auditors in the weaker, but not stronger, team identity condition differentially weight input across input quality levels for both of the more severe cases and the less severe non-IT case. The differential weighting observed is consistent with auditor conservatism (placing heavier weight on input that indicates an issue is more severe), but is functional from an effectiveness perspective as it only results in an inappropriate weighting scheme for the less severe issue.

### Supplemental Tests

#### ICFR Issue Classification

The tests described above provide direct evidence for my predictions about how auditors weight input received from IT specialists. In this section, I report tests using change in ICFR classification as an alternate dependent variable.20 The only new information provided to participants between the initial and final judgments is about or comes from the IT specialist, so changes in classifications are likely based primarily on the input from the IT specialist.

For this analysis, I use an indicator variable representing a change in classification where 1 indicates that the final classification differs from the initial classification, and 0 indicates no difference. Refer to Table 4, Panels A and B for the percentage of responses in which participants changed their classification across between-participants conditions by case.21 I estimate a logistic regression, clustering on participant, with change in classification as the dependent variable to test my predictions; see Table 4, Panel C.22 In support of H1, a marginally significant Team IdentityIssue Type interaction is present (p = 0.057); auditors with a weaker versus stronger team identity are more likely to change the classification when the issue is IT related, and less likely when it is non-IT (see Table 4, Panel D). For H2, only the pattern of results for the more severe non-IT case is consistent with the predicted pattern (see Table 4, Panels A and B). In support of H2, for the more severe non-IT case, auditors with a weaker team identity receiving lower quality input are less likely than the other three conditions to change the classification (p = 0.092, one-tailed; see Table 4, Panel E).23

TABLE 4

Change in ICFR Issue Classification

The results for the alternate dependent variable of change in classification are consistent with the results from the primary dependent variable, weight of input. For H1, the same pattern is observed for the Team IdentityIssue Type interaction (refer to [e] in Table 1, Panel A for the weight of input means). However, the significance of the interaction is weaker for change in classification, which is not surprising as change in classification is a less precise measure of auditors' input integration than weight of input. For H2, across both dependent variables, only the results for the more severe non-IT issue are consistent with my prediction. The same pattern of results is observed for this case across both change in classification and weight of input (refer to [c] in Table 1, Panel A for weight of input means). Similar to the H1 results, the test for H2 is weaker for change in classification than weight of input. Overall, the analyses using change in classification as an alternate dependent variable provide additional support for my findings.

#### Final Judgment Rationale

Following the completion of the final rating and classification in each case, participants were asked to provide the rationale for their judgments. Coding these responses allows me to examine whether the weaker team identity increased the salience of the IT specialists' expertise. I expect that auditors with a weaker team identity with the IT specialist will be more likely to mention the IT specialist in their rationales. An independent coder with five years of Big 4 audit experience and I coded responses; both coders were blind to experimental conditions. Initial intercoder agreement was 92.96 percent, with a Kappa of 0.76 (p < 0.001 indicating agreement better than chance). We reconciled all items of disagreement. Results of coding (untabulated) are consistent with theory. Auditors with a weaker team identity more frequently mention the IT specialist in their responses than auditors with a stronger team identity (χ2(1) = 5.86, p = 0.015).

#### Perceptions of IT Specialist and Input Received

I collected measures related to participants' perceptions of the IT specialist and the input received. I elicited overall quality of the IT specialist input, the competence, objectivity, and trustworthiness of the IT specialist providing the input, and whether the participant would want to work with the IT specialist again (all on 11-point scales). I estimate separate ANOVAs (untabulated) for each dependent variable and find a significant main effect of team identity on all five measures (all p < 0.013).24 Perceptions of the IT specialist and the IT specialist's input are significantly more favorable for the stronger versus weaker team identity condition. These results provide support for the increased level of trust when a stronger team identity is present. Further, these findings are interesting in light of the fact that auditors with a stronger versus weaker team identity with an IT specialist do not weight input significantly heavier overall.

Audit teams made up of auditors and specialists work together to complete the required audit procedures and determine appropriate audit conclusions. To improve our understanding of the audit team environment (Trotman, Bauer, and Humphreys 2015), I investigate how auditors integrate input received from IT specialists. Drawing on social identity theory, I examine the extent to which auditors weight input received from an IT specialist on ICFR issue judgments and how the strength of team identity influences this weight. I predict and find that auditors who have a weaker team identity with the IT specialist providing input weight the input received more heavily than auditors with a stronger team identity for IT-related issues. For non-IT issues, auditors with a weaker team identity place heavier weight on higher (versus lower) quality input, and auditors with a stronger team identity weight input relatively heavily regardless of input quality. Additionally, I find that my predictions hold for the more, but not less, severe issues.

Regulators and recent research highlight concerns regarding ineffective communication and coordination between auditors and IT specialists. My study investigates one aspect of these issues: the way in which auditors integrate input from IT specialists. Thus, I bridge research on audit specialists (Griffith 2018, 2020; Griffith et al. 2015; Bauer and Estep 2019; Boritz et al. 2017) and advice-taking in auditing (Ng and Shankar 2010; Kadous et al. 2013).

I find evidence of benefits to auditors having a weaker team identity with internal firm specialists; that is, auditors weight input from the specialist more heavily (compared to auditors with a stronger team identity) when the issue falls within the specialist's domain of expertise. Auditors with a weaker team identity appropriately differentiate between levels of input quality for more severe issues, while auditors with a stronger team identity with the IT specialist are less sensitive to levels of input quality. However, I also find evidence of a tendency for conservatism. This tendency is functional from an effectiveness perspective; it only results in an inappropriate weighting scheme for the less severe issues.

A substantial body of literature supports the benefits of a stronger team identity, such as increased cooperation and information sharing (e.g., Dovidio et al. 1998; Kane et al. 2005). Encouraging a weaker team identity between auditors and IT specialists may sacrifice these benefits. Future research can identify ways to capitalize on the benefits of stronger team identity and the input integration benefits of a weaker team identity identified in my study. Further, future research should investigate the implications of team identity for interactions between auditors and other internal firm specialists, such as tax or valuation, especially as research highlights ongoing issues with valuation specialists (Griffith et al. 2015; Griffith 2020).

My study also suggests possible benefits of audit firms increasing focus on advisory (e.g., consulting) services, as audit firms have expanded their advisory practices considerably in recent years (Rapoport 2012; The Economist 2012). Specialists often serve dual roles in audit firms today: acting as consultants on advisory engagements and as auditors on integrated and financial statement audit engagements. Audit firms claim that specialists' expertise improves audit quality because they bring knowledge gained from serving on advisory engagements to audits (Deloitte 2013; PricewaterhouseCoopers [PwC] 2013). However, realizing the benefits of specialists' knowledge depends on whether auditors integrate the information and input they receive. While firms are pushing auditors and specialists to adopt a one-team perspective (EY 2013; Bauer and Estep 2019), the dual role of specialists as auditors and consultants may decrease the extent to which auditors identify with and view specialists as part of the audit team. Therefore, auditors could be more likely to incorporate input on matters directly related to the specialists' area of expertise and realize these knowledge spillovers when specialists serve dual roles.

The implications, as well as limitations, of my study provide fruitful avenues for future research. While my experimental design allows control to make causal inferences, the lack of interaction between auditors and IT specialists could alter the impact of stronger and weaker team identities in the way auditors and IT specialists work together. Further, the participants made their ICFR assessment judgments in isolation, whereas in the audit environment, these decisions would be made in a hierarchical team setting (e.g., Chen, Trotman, and Zhou 2015). Future research should investigate whether these interpersonal interactions influence the way in which audit specialist input is integrated into audit team decisions.

I focus on auditors' perceptions of team identity with IT specialists; future research should investigate how IT (or other) specialists' perceptions of team identity influence input provided to auditors. Other relevant social identities also exist, such as client, organizational, and professional identities (Bamber and Iyer 2002, 2007; Suddaby, Gendron, and Lam 2009; Bhattacharjee and Brown 2018). Depending on the strength and salience of these identities (Bauer 2015), interactions could occur that influence how auditors weight input from specialists, especially when the input is consistent or inconsistent with client preferences. Finally, contextual features of the audit environment, such as time pressure, budget pressure, or client importance, could influence how willing auditors are to listen to audit specialists. Future research can investigate these potential interacting variables to further improve our understanding of how auditors incorporate input from audit specialists.

Aron,
A.,
Aron
E.,
and
Smollan
D.
1992
.
Inclusion of other in the self scale and the structure of interpersonal closeness
.
Journal of Personality and Social Psychology
63
(
4
):
596
612
.
Ashforth,
B.,
and
Johnson
S.
2001
.
Which hat to wear? The relative salience of multiple identities in organizational contexts
.
In
Social Identity Processes in Organizational Contexts
,
edited by
Hogg,
M. A.,
and
Terry
D. J.,
31
48
.
:
Psychology Press
.
Ashforth,
B.,
and
Mael
F.
1989
.
Social identity theory and the organization
.
14
(
1
):
20
39
.
Bamber,
E.,
and
Iyer
V.
2002
.
Big 5 auditors' professional and organizational identification: Consistency or conflict?
Auditing: A Journal of Practice & Theory
21
(
2
):
21
38
.
Bamber,
E.,
and
Iyer
V.
2007
.
Auditors' identification with their clients and its effect on auditors' objectivity
.
Auditing: A Journal of Practice & Theory
26
(
2
):
1
24
.
Bauer,
T.
2015
.
The effects of client identity strength and professional identity salience on auditor judgments
.
The Accounting Review
90
(
1
):
95
114
.
Bauer,
T.,
and
Estep
C.
2019
.
One team or two? Investigating relationship quality between auditors and IT specialists: Implications for audit team identity and the audit process
.
Contemporary Accounting Research
36
(
4
):
2142
2177
.
Bhattacharjee,
S.,
and
Brown
J. O.
2018
.
The impact of management alumni affiliation and persuasion tactics on auditors' internal control judgments
.
The Accounting Review
93
(
2
):
97
115
.
Bonaccio,
S.,
and
Dalal
R.
2006
.
Advice taking and decision-making: An integrative literature review, and implications for the organizational sciences
.
Organizational Behavior and Human Decision Processes
101
(
2
):
127
151
.
Boritz,
J.,
Kochetova-Kozloski
N.,
Robinson
L.,
and
Wong
C.
2017
.
Auditors' and specialists' views about the use of specialists during an audit
.
Working paper, University of Waterloo.
Brazel,
J.,
and
Agoglia
C.
2007
.
An examination of auditor planning judgments in a complex accounting information system environment
.
Contemporary Accounting Research
24
(
4
):
1059
1083
.
Brewer,
M.,
and
Gardner
W.
1996
.
Who is this “we”? Levels of collective identity and self representations
.
Journal of Personality and Social Psychology
71
(
1
):
83
93
.
Chen,
C.,
Trotman
K.,
and
Zhou
F.
2015
.
Nominal versus interacting electronic fraud brainstorming in hierarchical audit teams
.
The Accounting Review
90
(
1
):
175
198
.
Croteau,
B.
2013
.
Audit policy and current auditing and internal control matters
.
Remarks before the 2013 National Conference on Current SEC and PCAOB Developments, December 9. Available at: https://www.sec.gov/news/speech/2013-spch120913btc
Curtis,
M.,
Jenkins
J.,
Bedard
J.,
and
Deis
D.
2009
.
Auditors' training and proficiency in information systems: A research synthesis
.
Journal of Information Systems
23
(
1
):
79
96
.
Deloitte.
2013
.
Audit Quality: Our Responsibility, Our Commitment
.
New York, NY: Deloitte.
Dovidio,
J.,
Gaertner
S.,
and
Validzic
A.
1998
.
Intergroup bias: Status, differentiation, and a common in-group identity
.
Journal of Personality and Social Psychology
75
(
1
):
109
120
.
Dukerich,
J.,
Kramer
R.,
and
Parks
J.
1998
.
The dark side of organizational identification
.
In
Identity in Organizations: Building Theory through Conversations
,
edited by
Whetten,
D. A.,
and
Godfrey
P. C.,
245
256
.
Thousand Oaks, CA
:
Sage
.
Earley,
C.,
Hoffman
V.,
and
Joe
J.
2008
.
Reducing management's influence on auditors' judgments: An experimental investigation of SOX 404 assessments
.
The Accounting Review
83
(
6
):
1461
1485
.
Ernst & Young (EY).
2013
.
Vision 2020: Our plan for our people, our clients and our future
.
Franzel,
J.
2014
.
Effective audits of internal control in the current “perfect storm.” Speech given at the Institute of Internal Auditors 2014 General Audit Management Conference, March 26. Available at: https://pcaobus.org/News/Speech/Pages/03262014_IIA.aspx.
Gino,
F.,
and
Moore
D.
2007
.
.
Journal of Behavioral Decision Making
20
(
1
):
21
35
.
Griffith,
E.
2018
.
When do auditors use specialists' work to improve problem representations of and judgments about complex estimates?
The Accounting Review
93
(
4
):
177
202
.
Griffith,
E.
2020
.
Auditors, specialists, and professional jurisdiction in audits of fair values
.
Contemporary Accounting Research
37
(
1
):
245
276
.
Griffith,
E.,
Hammersley
J.,
and
K.
2015
.
Audits of complex estimates as verification of management numbers: How institutional pressures shape practice
.
Contemporary Accounting Research
32
(
3
):
833
863
.
Guggenmos,
R.,
Piercey
M.,
and
Agoglia
C.
2018
.
Custom contrast testing: Current trends and a new approach
.
The Accounting Review
93
(
5
):
223
244
.
Harvey,
N.,
and
Fischer
I.
1997
.
Taking advice: Accepting help, improving judgment, and sharing responsibility
.
Organizational Behavior and Human Decision Processes
70
(
2
):
117
133
.
Haslam,
S.,
and
Ellemers
N.
2005
.
Social identity in industrial and organizational psychology: Concepts, controversies and contributions
.
In
International Review of Industrial and Organizational Psychology
,
edited by
Hodgkinson,
G. P.,
and
Ford
J. K.,
39
118
.
Chichester, U.K
.:
John Wiley & Sons, Ltd
.
Hensley,
V.,
and
Duval
S.
1976
.
Some perceptual determinants of perceived similarity, liking, and correctness
.
Journal of Personality and Social Psychology
34
(
2
):
159
168
.
Hogg,
M.,
and
Turner
J.
1985
.
Interpersonal attraction, social identification and psychological group formation
.
European Journal of Social Psychology
15
(
1
):
51
66
.
International Forum of Independent Audit Regulators (IFIAR).
2017
.
Report on 2016 Survey of Inspection Findings
.
K.,
Leiby
J.,
and
Peecher
M.
2013
.
How do auditors weight informal contrary advice? The joint influence of advisor social bond and advice justifiability
.
The Accounting Review
88
(
6
):
2061
2087
.
Kane,
A.
2010
.
Unlocking knowledge transfer potential: Knowledge demonstrability and superordinate social identity
.
Organization Science
21
(
3
):
643
660
.
Kane,
A.,
Argote
L.,
and
Levine
J.
2005
.
Knowledge transfer between groups via personnel rotation: Effects of social identity and knowledge quality
.
Organizational Behavior and Human Decision Processes
96
(
1
):
56
71
.
King,
R.
2002
.
An experimental investigation of self-serving biases in an auditing trust game: The effect of group affiliation
.
The Accounting Review
77
(
2
):
265
284
.
Levin,
D.,
and
Cross
R.
2004
.
The strength of weak ties you can trust: The mediating role of trust in effective knowledge transfer
.
Management Science
50
(
11
):
1477
1490
.
Libby,
R.,
Trotman
K.,
and
Zimmer
I.
1987
.
Member variation, recognition of expertise, and group performance
.
Journal of Applied Psychology
72
(
1
):
81
87
.
Ng,
T.,
and
Shankar
P.
2010
.
Effects of technical department's advice, quality assessment standards, and client justifications on auditors' propensity to accept client-preferred accounting methods
.
The Accounting Review
85
(
5
):
1743
1761
.
Peduzzi,
P.,
Concato
J.,
Kemper
E.,
Holford
T.,
and
Feinstein
A.
1996
.
A simulation study of the number of events per variable in logistic regression analysis
.
Journal of Clinical Epidemiology
49
(
12
):
1373
1379
.
PricewaterhouseCoopers (PwC).
2013
.
Our Focus on Audit Quality: 2013 Report
.
New York, NY: PwC.
Public Company Accounting Oversight Board (PCAOB).
2007
.
An Audit of Internal Control over Financial Reporting that is Integrated with an Audit of Financial Statements. Auditing Standard (AS) 2201. Washington, DC: PCAOB.
Public Company Accounting Oversight Board (PCAOB).
2012
.
Observations from 2010 Inspections of Domestic Annually Inspected Firms Regarding Deficiencies in Audits of Internal Control over Financial Reporting. PCAOB Release No. 2012-006
.
Washington, DC: PCAOB.
Public Company Accounting Oversight Board (PCAOB).
2013
.
Report on 2012 Inspection of KPMG LLP. PCAOB Release No. 104-2013-147
.
Washington, DC: PCAOB.
Public Company Accounting Oversight Board (PCAOB).
2015
.
Report on 2014 Inspection of Ernst & Young LLP. PCAOB Release No. 104-2015-121
.
Washington, DC: PCAOB.
Rapoport,
M.
2012
.
Eyebrows go up as auditors branch out. Wall Street Journal (December): Available at: https://www.wsj.com/articles/SB10001424127887324705104578149222319470606
Ricchiute,
D.
1999
.
The effect of audit seniors' decisions on working paper documentation and on partners' decisions
.
Accounting, Organizations and Society
24
(
2
):
155
171
.
Sidanius,
J.,
Pratto
F.,
and
Mitchell
M.
1994
.
In-group identification, social dominance orientation, and differential intergroup social allocation
.
Journal of Social Psychology
134
(
2
):
151
167
.
Smith,
J.,
and
Kida
T.
1991
.
Heuristics and biases: Expertise and task realism in auditing
.
Psychological Bulletin
109
(
3
):
472
489
.
Suddaby,
R.,
Gendron
Y.,
and
Lam
H.
2009
.
The organizational context of professionalism in accounting
.
Accounting, Organizations and Society
34
(
3–4
):
409
427
.
Tajfel,
H.
1978
.
Social categorization, social identity and social comparison: Studies in the social psychology of intergroup relations
.
In
Differentiation Between Social Groups
,
edited by
Tajfel,
H.,
61
76
.
London, U.K
.:
.
The Economist.
2012
.
The Big Four accounting firms: Shape shifters. The Economist (September). Available at: https://www.economist.com/node/21563726
Tost,
L.,
Gino
F.,
and
Larrick
R.
2012
.
Power, competitiveness, and advice taking: Why the powerful don't listen
.
Organizational Behavior and Human Decision Processes
117
(
1
):
53
65
.
Towry,
K.
2003
.
Control in a teamwork environment—The impact of social ties on the effectiveness of mutual monitoring contracts
.
The Accounting Review
78
(
4
):
1069
1095
.
Tropp,
L.,
and
Wright
S.
2001
.
Ingroup identification as the inclusion of ingroup in the self
.
Personality and Social Psychology Bulletin
27
(
5
):
585
600
.
Trotman,
K.,
Bauer
T.,
and
Humphreys
K.
2015
.
Group judgment and decision making in auditing: Past and future research
.
Accounting, Organizations and Society
47
:
56
72
.
Tversky,
A.,
and
Kahneman
D.
1974
.
Judgment under uncertainty: Heuristics and biases
.
Science
185
(
4157
):
1124
1131
.
Van Der Vegt,
G.,
and
Bunderson
J. S.
2005
.
Learning and performance in multidisciplinary teams: The importance of collective team identification
.
48
(
3
):
532
547
.
Vittinghoff,
E.,
and
McCulloch
C.
2007
.
Relaxing the rule of ten events per variable in logistic and Cox regression
.
American Journal of Epidemiology
165
(
6
):
710
718
.
Wilder,
D.
1984
.
Empirical contributions: Predictions of belief homogeneity and similarity following social categorization
.
British Journal of Social Psychology
23
(
4
):
323
333
.
Wolfe,
C.,
Mauldin
E.,
and
Diaz
M.
2009
.
Concede or deny: Do management persuasion tactics affect auditor evaluation of internal control deviations?
The Accounting Review
84
(
6
):
2013
2037
.
Yaniv,
I.
2004
.
Receiving other people's advice: Influence and benefit
.
Organizational Behavior and Human Decision Processes
93
(
1
):
1
13
.
Zellmer-Bruhn,
M.,
Maloney
M.,
Bhappu
A.,
and
R.
2008
.
When and how do differences matter? An exploration of perceived similarity in teams
.
Organizational Behavior and Human Decision Processes
107
(
1
):
41
59
.
APPENDIX A

Client Case Descriptions

### APPENDIX C

#### Experimental Design

• Participants were randomly assigned to one of four between-participants conditions:

• Weaker Team Identity/Lower Quality Input (n = 23)

• Weaker Team Identity/Higher Quality Input (n = 27)

• Stronger Team Identity/Lower Quality Input (n = 25)

• Stronger Team Identity/Higher Quality Input (n = 26)

Refer to Appendix B for the details of the between-participants manipulations.

• All participants completed four cases (details provided in Appendix A):

• More severe, IT-related case

• More severe, Non-IT case

• Less severe, IT-related case

• Less severe, Non-IT case

The order of completing the cases was randomized across participants. Participants received input from an IT specialist on each case and the input received depended on: (1) whether the participant was in a higher or lower input quality condition, and (2) the severity level of the case. The table below indicates the input ratings received (corresponding to an 11-point scale with labels of “Control deficiency” [1], “Significant deficiency” [6], and “Material weakness” [11]). As indicated, all participants receive input of 3.5 twice and input of 8.5 twice.

1

While Auditing Standard (AS) 2201 (PCAOB 2007) provides guidance for classifying identified ICFR deficiencies, this guidance is subjective and can be difficult to apply. For example, a material weakness, the most severe classification, exists if “there is a reasonable possibility that a material misstatement of the company's annual or interim financial statements will not be prevented or detected on a timely basis” (PCAOB 2007, A7). A significant deficiency, the next most severe classification, is “a deficiency … that is less severe than a material weakness, yet important enough to merit attention by those responsible for oversight of the company's financial reporting” (PCAOB 2007, A11). AS 2201 also outlines indicators of material weaknesses in ICFR (e.g., identification of fraud, restatement of previously issued financial statements; PCAOB 2007, 69).

2

I focus on internal firm IT specialists, and I use the term “input” to represent information, advice, or recommendations that auditors receive from these IT specialists. My definition of input is similar to “advice” in the psychology and auditing literature and points to the ownership auditors ultimately have over the audit process.

3

While auditors rely on IT specialists to perform tests in the specialized areas of IT and controls, auditors determine final judgments and decisions on audit engagements, and “own” the workpapers and audit file (Bauer and Estep 2019; Griffith 2020). Unlike in formal consultation, where auditors must follow recommendations provided, auditors receiving input from IT (and other) specialists have no obligation to follow those recommendations (Boritz et al. 2017; Bauer and Estep 2019).

4

Bauer and Estep (2019) provide evidence that perceptions of team identity strength sometimes overlap and sometimes differ between auditors and IT specialists. My interest lies in examining how auditors' perceptions of team identity impact their judgments.

5

I do not make predictions about the effects of input quality for IT-related issues. Auditors vary in their level of IT knowledge and their ability to evaluate input for IT-related issues (Brazel and Agoglia 2007). Given that I expect that auditors view IT-related issues as more outside their domain of expertise than non-IT issues and feel less comfortable evaluating these issues without IT specialist input, I do not have ex ante expectations about their ability or willingness to differentiate levels of input quality for IT-related issues.

6

Studies outside of accounting (Kane, Argote, and Levine 2005; Kane 2010) suggest that those with a stronger team identity will be more likely to discern levels of input quality. These studies expect and find that groups receiving knowledge from an in-group member have greater motivation to more thoroughly consider the knowledge provided. However, the decision context of these studies differs from my study. The knowledge being considered in these studies relates to an interdependent production task, not to an individual judgment. The task in both Kane et al. (2005) and Kane (2010) involved assembly line production of origami sailboats by groups. The discount rate judgment task in Kadous et al. (2013) provides a decision context similar to that of my study.

7

The eight participating firms reviewed the instrument prior to data collection to confirm that the materials were realistic and understandable to the auditor participants. I also obtained the necessary institutional approval to perform human-subjects research prior to collecting data.

8

These participants are appropriate, as senior-level auditors are most likely to be interacting with IT specialists on audit engagements. Further, senior-level auditors often make initial assessments and higher-level auditors are influenced by recommendations made by subordinates (Earley et al. 2008; Ricchiute 1999).

9

Total months of experience in public accounting is higher for those in the stronger versus weaker team identity condition (F1,96 = 5.36, p = 0.023). Percentage of time spent on auditing (consulting) is marginally lower (higher) for those in the higher versus lower input quality condition, with F1,97 = 2.78, p = 0.099 (F1,97 = 3.04, p = 0.085). Inferences reported in Section IV are unchanged when controlling for these variables. Reported p-values are two-tailed unless otherwise noted.

10

The order of completing client cases was randomized across participants.

11

As in the main experiment, the order of cases was randomized across participants.

12

The words “helpful,” “technically,” “core,” and “obligatory” appear in all conditions to keep word choice consistent. Also, sharing a social identity can increase perceptions of competence (Sidanius, Pratto, and Mitchell 1994). I attempt to control for this response by including a statement in both conditions regarding the competence of the IT specialist on both audit and advisory/consulting engagements, indicating that the IT specialist splits time between the two. See Appendix B for the detailed conditions.

13

Team identity is a complex construct, and the three pieces of information are collectively intended to manipulate the strength of auditors' team identity with the IT specialist. While I have no reason to believe that any of the aspects of my manipulation would interact differently with the other independent variables of interest or have a different effect on the dependent measure, I leave it to future research to examine whether and how the various aspects can have a differential impact on auditors' strength of team identity with specialists.

14

While I do not make any ex ante predictions about input quality for the IT-related cases, I manipulate input quality for all cases for a fully crossed design.

15

An 8.5 (3.5) on the ICFR issue severity rating scale corresponds to an ICFR issue at the midpoint of a significant deficiency and a material weakness (a control deficiency and a significant deficiency). I hold constant justification for the provided rating in each case across the input quality conditions.

16

Results of the main tests of hypotheses are inferentially the same without the upper bound adjustments. For two of these observations, participants move in the direction of advice and to an extreme value past it. For example, for one of these observations, the initial rating is 10.1, input is 8.5, and the final rating is 3.8, yielding a weight of input of 3.94. For consistency in adjustment approach, I adjust the weight of input value to 1 for these two observations for the main analyses; however, if these two observations are dropped or changed to zero, results of the main tests of hypotheses are also inferentially the same.

17

Given four undefined values of weight of input, total n is 400 (as there are 101 participants who completed four cases each). However, I use a repeated measures ANOVA for my main analyses, which requires a weight of input value for all four cases for each participant, or else the participant is dropped from analyses. Thus, all observations for the four participants who have one undefined value each are dropped from analyses, resulting in a total n of 388. Reported results are inferentially the same when estimating separate ANOVAs for each case on the between-participants conditions that include the dropped weight of input observations.

18

Five participants did not respond to this question. I retain these participants in all other tests.

19

Recall that the H1 prediction for non-IT cases is qualified by the interaction of team identity and input quality predicted in H2; I further explore the non-IT cases in the following section.

20

I am interested in auditors' judgment process, as my theory and predictions relate to how auditors incorporate input received from IT specialists. Therefore, I do not predict or test differences in outcomes (i.e., final ICFR issue classifications), as auditors received different input across conditions and cases (see Appendix C). While weighting the same input similarly would likely not result in classification differences, weighting the same input differently, weighting different input similarly, or even weighting different input differently could all translate into differences in final issue classifications.

21

Of the 103 changes, all but seven move in the direction of the input received from the IT specialist, and these seven observations do not differ across between-participants conditions (Fisher's exact p = 0.486) or within-participants cases (Fisher's exact p = 1.00).

22

Given potential concerns about the reliability of logistic regression models with lower events per variable (Peduzzi, Concato, Kemper, Holford, and Feinstein 1996; Vittinghoff and McCulloch 2007), I ran the logistic regression for change in classification collapsing across Severity. The direction and significance of relevant effects remain the same as those reported in Table 4, Panel C, providing comfort over the reliability of these results.

23

Results are inferentially the same using implied classification changes based on the initial and final severity ratings provided.

24

Reported results are robust to controlling for these factors.

## Author notes

This paper is based on my dissertation, completed at the University of Illinois at Urbana–Champaign. I am grateful to my dissertation committee: Dolores Albarracin, Tim Bauer (director of research), Mark Peecher (chair), Ken Trotman, and Michael Williamson. This paper has also been improved by helpful comments from Jacqueline S. Hammersley (editor), two anonymous reviewers, Elizabeth Altiero, Jasmijn Bol, Willie Choi, Keith Czerney, Paul Demere, Brooke Elliott, Brian Gale, Brent Garza, Emily Griffith, Erin Hamilton, Lynn Hannan, Paige Harrell, Sean Hillison, Vicky Hoffman, Kathryn Kadous, Jeremy Lill, Tracie Majors, Don Moser, Kathy Rupar, Lauren Reid, Amanda Winn, Dan Zhou; doctoral seminar participants at Emory University, workshop participants at Emory University, Rutgers, The State University of New Jersey, Tulane University, University of Florida, University of Illinois at Urbana–Champaign, University of Nebraska–Lincoln, and University of Pittsburgh; and conference participants at the 2017 International Symposium on Audit Research and the 2018 Auditing Section Midyear Meeting. I greatly appreciate the support of the Center for Audit Quality and American Accounting Association Auditing Section Access to Auditors program and the time and effort of the audit firm participants. I thank the American Institute of Certified Public Accountants Accounting Doctoral Scholars Program and the Deloitte Foundation Doctoral Fellowship for funding while pursuing my doctoral degree.

The views expressed in this article and its content are those of the author alone and not those of the Center for Audit Quality.

Cassandra Estep, Emory University, Goizueta Business School, Department of Accounting, Atlanta, GA, USA.

Editor's note: Accepted by Jacqueline S. Hammersley, under the Senior Editorship of Mary E. Barth.