This paper is relevant for audit practice by highlighting and informing the audit profession about the potential behavioral challenges to the application of professional skepticism when using ADA. Possible mitigation methods provided by academic literature to audit practice are also discussed in this paper.

Audit firms around the globe have invested heavily into a variety of audit technologies (e.g., Alles and Gray 2016; Deloitte 2016; KPMG 2016, 2019; EY 2018; PwC 2020; Eilifsen et al. 2020; Austin et al. 2021). Of these technological developments, audit data analytics (ADA) in particular are receiving increasing attention in auditing because they facilitate the incorporation of larger and more complex datasets as well as more diverse data sources into audit testing (e.g., FRC 2017). This paper adopts the ADA definition by the AICPA (2015, 2017) as being “the science and art of discovering and analyzing patterns, identifying anomalies, and extracting other useful information in data underlying or related to the subject matter of an audit through analysis, modeling, and visualization for the purpose of planning or performing the audit” (AICPA 2015, p. 92; 2017, p. 1).

ADA are expected to facilitate more effective and efficient auditor judgment (e.g., AICPA 2017). Instead of incorporating only a limited number of data sources, ADA enable auditors to simultaneously compare data from a wider variety of sources (e.g., prior year balances, budgets, industry data, data from related accounts, and non-financial measures) as suggested by auditing standards such as ISA 520 (IAASB 2018). Hence, the use of ADA will enable auditors to potentially gain deeper insights into their clients’ data and obtain a better understanding of the clients’ business and environment compared to traditional methods (e.g., Brown-Liburd et al. 2015; Cao et al. 2015; Krahel and Titera 2015; Yoon et al. 2015; FRC 2017; IAASB 2020c, 2021; Austin et al. 2021; PCAOB 2021a). The use of ADA is also expected to improve audit efficiency. Using ADA potentially reduces the audit time and cognitive effort needed for processing the information (e.g., Anderson et al. 2020). Therefore, the speed at which auditors can identify inconsistencies, anomalies, or red flags indicating higher risk of material misstatements will potentially increase substantially (e.g., AICPA 2017).

Of particular interest in the current paper, ADA are expected to improve the appropriate application of auditors’ professional skepticism, ultimately improving audit quality. Auditing standards such as ISQM 1 (IAASB 2020a) and ISA 220 (IAASB 2020b) define professional skepticism as an attitude of a questioning mind and a critical assessment of audit evidence. Using ADA potentially improves auditors’ judgment quality because it helps the auditor to efficiently identify patterns and inconsistencies, thereby motivating more application of professional skepticism and increasing the likelihood of identifying material misstatements (IAASB 2020c).

Besides potential benefits, research shows that ADA may also bring challenges to the appropriate application of professional skepticism (e.g., Appelbaum et al. 2017; Rose et al. 2017; Barr-Pulliam et al. 2020; Holmstrom 2020; Austin et al. 2021; Commerford et al. 2021; Holt and Loraas 2021). Being alert to the potential behavioral challenges and exploring approaches of mitigating those challenges are critical for audit practice to fully utilize the benefits of ADA. Therefore, this paper’s objectives are to discuss the potential behavioral challenges created by ADA to professional skepticism as identified in the literature, review possible approaches of mitigating those challenges, and suggest future research opportunities. Namely, I discuss the potential behavioral challenges from the following five perspectives:

First, forming an appropriate attitude about ADA is important since both under-reliance and overreliance on ADA may impede the appropriate exercise of professional skepticism (section 2). Second, auditors should be aware of how the data characteristics (e.g., data reliability and data relevance) may influence their interpretations of audit evidence obtained from ADA, potentially distorting the appropriate exercise of professional skepticism (section 3). Research further identifies that the anomalies identified by ADA may influence auditors’ application of professional skepticism (e.g., the larger number of anomalies, false positives, and false negatives) (section 4). Next, auditors need to adopt appropriate mindsets when using ADA since mindsets can influence their judgment quality (section 5). Finally, the social contexts and interactions between auditors and other stakeholders in the ADA journey may also influence the exercise of professional skepticism (section 6). I conclude the paper with a summary and discussion in section 7.

Attitude is usually defined as “an evaluative integration of cognitions and affects experienced in relation to an object” (Crano and Prislin 2006, p. 347). Auditors’ forming of attitudes during an audit has the potential to influence their professional skepticism (e.g., Nolder and Kadous 2018). In fact, auditing standards define professional skepticism as a skeptical attitude including a questioning mind and a critical assessment of audit evidence (IAASB 2020a, 2020b). Therefore, taking appropriate attitudes in the auditing process is critical to auditors’ skeptical judgment quality and hence predicts their subsequent skeptical intentions and behaviors, ultimately determining audit quality (e.g., Nolder and Kadous 2018). Auditors’ use of ADA adds complexity to this requirement since auditors also need to form an appropriate attitude toward ADA in their cognitive and evaluative responses to the audit evidence generated by ADA.

One example of a potentially inappropriate attitude in this regard is auditors’ exhibition of algorithm aversion in their use of ADA. Specifically, advanced ADA algorithms (e.g., artificial intelligence) employ machine learning techniques to integrate, model, and analyze large and diverse datasets, thereby assisting auditors with some challenging tasks, such as evaluating complex accounting estimates, assessing fraud risk, and estimating the financial distress related to going concern opinions (e.g., Murphy 2017; Gepp et al. 2018; Commerford et al. 2021). However, prior research suggests that auditors may suffer from an aversion to algorithms and as a result they may under-rely on the evidence provided by ADA (e.g., Önkal et al. 2009; Commerford et al. 2021). Algorithm aversion implies that humans perceive algorithms as incapable when the task involves a high level of subjectivity (Castelo et al. 2019; Yeomans et al. 2019). Commerford et al. (2021) find that auditors exhibit algorithm aversion when artificial intelligence is used in auditing of complex accounting estimates. Specifically, auditors receiving evidence that contradicts management’s position from an artificial intelligence system, compared with evidence from a human specialist, potentially propose smaller audit adjustments to management’s complex accounting estimates. As such, auditors’ susceptibility to algorithm aversion may even lead to an increased reliance on management’s evidence. Therefore, algorithm aversion may undermine auditors’ questioning mind and their subsequent skeptical behavior (e.g., Commerford et al. 2021).

While under-reliance on ADA can be problematic, it is equally important for auditors to avoid overreliance on ADA (IAASB 2021). Although there is little research on algorithm appreciation (Logg et al. 2019) in the auditing literature, regulators and standard-setters express concerns about the possibility of auditors’ overreliance on audit technologies and its potential damage to the critical assessment of audit evidence from ADA (IAASB 2021).

As a result, it is critical for auditors to adopt an appropriately balanced attitude toward ADA. The auditing literature has identified some possible solutions to increase auditors’ reliance on audit technologies in general or audit evidence from ADA, such as increasing transparency of those audit technologies (Holmstrom 2020), and increasing auditors’ control over and hence perceived ownership of those tools (Dietvorst et al. 2018; Holmstrom 2020).

Further research could examine behavioral interventions that can be used to enhance auditors’ appropriate reliance on ADA. Specifically, future research may explore what features can be added to current ADA tools to prime auditors’ appropriate level of reliance when using those tools.

In this section, I discuss how auditors’ perceptions of the reliability and relevance of data inputs processed by ADA may influence their evaluations of audit evidence obtained from ADA and hence influence their application of professional skepticism.

In this section, I discuss how the anomalies identified by ADA may influence auditor skepticism. The topics discussed in the current section are the larger number of anomalies, false positives and false negatives.

A mindset refers to “a set of judgmental criteria and cognitive processes and procedures that produce a disposition or readiness to respond in a certain manner” (Griffith et al. 2015a, p. 54). Mindsets can influence auditors’ judgment quality and application of professional skepticism (e.g., Griffith et al. 2015a; Brewster and Bucaro 2020; Saiewitz and Wang 2020). Therefore, adopting inappropriate mindsets may diminish the benefits of ADA to audit practice. Although there is limited research on mindset during ADA use (Cao et al. 2021), implications can be drawn from research in related auditing areas (e.g., Griffith et al. 2015a; Plumlee et al. 2015; Backof et al. 2018; Austin et al. 2020; Saiewitz and Wang 2020).

Mindsets potentially influence auditors’ ADA adoption decisions. Cao et al. (2021) find that auditors with growth mindsets, compared to auditors with fixed mindsets, are more likely to adopt ADA when inspection risks are higher. It is often observed that people with a growth mindset believe that abilities and intelligence can be developed if they expend more effort to learn, and hence they usually attribute their successes (failures) to (lack of) effort. In contrast, those with a fixed mindset believe that abilities and intelligence are fixed attribute that cannot be developed by devoting effort, and therefore, they usually attribute their successes (failures) to (lack of) abilities or intelligence (Dweck and Leggett 1988; Elliott and Dweck 1988; Dweck 2008). In a high-inspection-risk environment, auditors with a growth mindset are more likely to adopt an ADA because they recognize the learning potential. In contrast, auditors with a fixed mindset tend to be concerned about the potential negative implications of performance failures, which decreases their motivation to adopt ADA. Although Cao et al. (2021) do not examine the effect of growth versus fixed mindsets on professional skepticism, it can be expected that auditors with a growth mindset, compared with a fixed mindset, are more likely to investigate the anomalies identified by ADA because they are likely to regard investigations as good opportunities to improve their knowledge and expertise.

In addition to ADA adoption decisions, future research could examine how mindsets influence auditors’ assessment and evaluation of contradictory evidence from ADA. Griffith et al. (2015a) find that auditors in a deliberative mindset, compared to in an implemental mindset, make higher-quality judgments in complex accounting estimate audits. According to the mindset theory of action phases (Gollwitzer 2012), when making goal decisions, decision makers tend to adopt a deliberative mindset to broadly process available information in an open-minded and balanced manner, which “facilitates a broad consideration of the pros and cons of various alternatives” (Griffith et al. 2015a, p. 55). Once the goal is determined, decision makers turn to an implemental mindset, which “facilitates planning how, rather than whether, to execute a task or reach a goal” (Griffith et al. 2015a, p. 55). Therefore, auditors employing a deliberative mindset, relative to an implemental mindset, are less likely to exhibit bias in their information evaluation, hence motivating their examination on contradictory information and identification of inconsistencies (Griffith et al. 2015a). Similar implications can be obtained from similar studies, such as Austin et al. (2020) and Saiewitz and Wang (2020). Overall, research implies that employing mindsets which improve auditors’ examination on contradictory evidence is critical to improve their skeptical judgment and application of professional skepticism.

Future research could more examine whether the implications from prior literature can be generalized to ADA practice. Particularly, future research should explore whether there are certain types of mindsets specifically related to ADA so that auditors can employ to improve professional skepticism when using ADA.

In this section, I discuss how the contextual environment around auditors and their interactions with key stakeholders in the ADA journey potentially influence their judgment quality and motivation for skeptical behavior when using ADA. The potential factors discussed in the current section include tone at the top, the work of data specialists, the audit committee’s attitude, sophistication of the client’s information technology (IT) systems, and regulations.

Many stakeholders believe that the time has come for auditors to embrace technology (e.g., Alles and Gray 2016; Deloitte 2016; KPMG 2016, 2019; EY 2018; Eilifsen et al. 2020; PwC 2020; Austin et al. 2021). Although the adoption and use of ADA brings many potential benefits to audit practice, it also creates many challenges. This paper discusses the behavioral challenges to the appropriate application of professional skepticism from five perspectives: auditors’ attitudes toward ADA, data characteristics, anomalies identified by ADA, auditors’ mindsets, and social contexts and interactions involved in ADA practice. First, an inappropriate attitude toward ADA may negatively influence auditors’ assessment of audit evidence from ADA and therefore their application of professional skepticism. Second, auditors’ evaluation of audit evidence from ADA may be influenced by the perceived reliability (e.g., sources and structure) and relevance of data inputs. Since auditors’ exercise of professional skepticism can be negatively influenced by unreliable and irrelevant data inputs, judgment and effort may be necessary to conduct data evaluation and screening before analysis. Third, auditors’ judgment can also be influenced by the anomalies identified by ADA, such as the larger number of anomalies, false positives, and false negatives. Further, adopting appropriate mindsets is also expected to be critical to enhance the exercise of professional skepticism when using ADA. Last but not least, auditors should be aware that their social contexts and interactions with others (e.g., tone at the top, the work of data specialists, the audit committee’s attitude, sophistication of the client’s IT systems, and regulations) in their ADA journey can also influence their exercise of professional skepticism.

Future research linking ADA and auditors’ application of professional skepticism from those five perspectives has been recommended in this paper. For example, further research could examine potential behavioral interventions to prime auditors’ appropriate reliance on ADA. Future research is also recommended to investigate the potential effects of data characteristics on auditors’ assessment of evidence obtained from ADA. Researchers could continue to explore measures to mitigate the potential negative influences of ADA (e.g., the large number of anomalies and false positives) on auditors’ application of professional skepticism. Future research aiming to investigate appropriate mindsets that auditors should adopt when using ADA is also encouraged. Finally, research can explore social, contextual and environmental factors that motivate auditors’ better ADA practice and application of professional skepticism.

Overall, this paper informs academia, the audit profession, standard-setters, and regulators about the potential challenges to the appropriate application of professional skepticism when using ADA so that stakeholders can be alert to and prepared for those potential issues. Concluding, multiple efforts are needed to solve those challenges in auditors’ ADA journey and motivate the appropriate application of professional skepticism.