Auditor independence became a focus of regulatory and accounting research due to a number of large corporate collapses involving fraud in the early 2000s. The focus of research was the reduction of auditor bias, and attention to improving the auditors’ objective position as a means to enhance the identification of fraud and mitigate the risks of issuing an inappropriate audit opinion or being implicated in fraud cases.
Audit team members are required, under the International Auditing Standards, to exercise professional scepticism throughout the audit process. In Australia, this requirement is mandatory under the Legislative Instruments Act 2003, and auditors have been held accountable in the courts for failure to exercise an appropriate degree of professional scepticism following audit failure.
Two aspects under International Auditing Standard ASA200 (AUASB, 2013b) are associated with professional scepticism: A questioning attitude, and critical assessment of evidence. Interestingly, a professional level of that attitude is not defined within the Standard, so it is not clear how auditors can judge whether they are exercising an appropriate standard of scepticism.
This research explored whether a ‘professional’ standard, or level, of scepticism could be identified, and developed the notion of level to distinguish a range of ‘professional scepticism’ from ranges of lay scepticism and less objective attitudes. The levels are then used to explore whether external auditors exercise the appropriate professional scepticism, as mandated by the International Auditing Standards. This extends prior research, which has attempted to measure auditor scepticism focused on the attitude factor only, and resulted in relative measures of scepticism that compared auditors with other auditors rather than comparing with a ‘professional’ level or benchmark.
In this research, a conceptual model of professional scepticism is developed, illustrating the interrelationships between the attitude and critical assessment aspects. The literature informing the model supports the view that the attitude factor reflects an inherent trait, while critical assessment is a learned skill, developed over time, and influenced by training and/or experience. Together, these factors represent a default level of scepticism, being a default mindframe in the absence of prompts to increase vigilance, and distinct from a state that may be achievable by such an increase in vigilance.
Three established questionnaires were combined to address the trait and skill aspects of the model: Wrightsman’s (1991) Interpersonal Trust Scale, Budner’s (1962) Intolerance of Ambiguity Scale, and The Curiosity and Exploration Inventory-II (Kashdan et al, 2009). These generic inventories were chosen to facilitate participation by non-auditors, enabling exploration of a broad range of sceptical and non-sceptical attitudes and skills, increasing the range of observations across the model continuum for the purpose of establishing scepticism levels.
The resulting questionnaire was administered online to potential participants via social media. This method of recruitment enabled participation by subjects across the world, appropriate to the modern international educational, commercial and investment scenarios. Clusters of relevant participant groups were encouraged by advertising the survey to the memberships of qualified fraud examiner and auditor groups’ online discussion boards, and the social media construct was utilised to snowball recruitment to a reasonably diverse cross-section of adult internet users without accounting, auditing or fraud training. The non-specialist participants could be expected to have a direct or indirect interest in the truth and fairness of financial statements, as business decision makers, employees, investors and/or consumers. Participation was voluntary, anonymous and without reward.
Of 298 completed surveys, 7 participants were excluded, resulting in 291 usable responses. After preliminary analysis, the survey instrument was analysed and reduced to a 17-question instrument with a Cronbach’s Alpha of 0.910 and two confirmed factors.
Each participant’s data was scored in two ways: Firstly, a Bias Score indicated a degree of personal bias as well as the direction of bias, toward either trust or distrust, where a score of zero represents the lowest level of subjective or dispositional bias and the maximum bias score is +/-26. This score was derived solely from the items in Wrightsman’s (1991) Interpersonal Trust Scale. Secondly, each participant’s data was received a Scepticism Score, derived from all items in the composite scale, converted to a percentage, where 100% represents the highest level of professional scepticism. The objective of the dual-scoring process is to offer clarity in instances that an auditor’s Scepticism Score is lower than the professional level benchmark: The directional Bias Score may be useful for deciding upon relevant skill development activities for those whose scores suggest a high level of trust in others (at the positive sign end) which could result in increased audit risk or, conversely, a high level of distrust (at the negative sign end) which could result in audit inefficiencies.
Two standard deviations of the mean of scores of qualified fraud examiners (n=41), whose work is centred on use of admissible, relevant and reliable evidence for prosecution of fraud cases, were utilised to establish a boundary to the ‘professional’ range of scepticism, by virtue of the role of evidence in their work. This top level of scepticism was expected to capture 95% of those participants whose work entailed objective evidence-based judgments. Two further ranges were established: A middle range, between the professional benchmark and the mid-score of 50%, expected to capture those participants whose judgements are more likely informed by evidence than subjectivity; and the lower range, below 50%, expected to capture those participants whose judgements are more likely informed by subjectivity than evidence.
An unexpected Sophisticated Layperson (SL) group emerged from the exploration: Qualified accountants without external audit experience or fraud training (n=52), employed in corporate roles. This group’s mean Bias score was substantially less trusting than all other participant groups’, and the range of scores within the group was quite narrow. In contrast, the scores of the Non-sophisticated Layperson (NSL) group (n=153) spread very much further along the scepticism continuum in both trust and distrust directions, as anticipated.
Testing revealed that 51% of external auditors (n=45) in the study returned default scepticism scores within the ‘professional’ range, with the average mean score attributing to the auditors slightly less trusting attitudes than fraud examiners. Auditors scored .0947747 points, 95% CI [.140962, .048587], above the professional benchmark, set at two standard deviations of the mean of Qualified Fraud Examiners’ scores. This difference is statistically significant, t(44) = 4.135, p < .001, and statistically moderate at d = 0.6165. Further, Auditor participants were, on average, less biased than all other participants, and the mean Auditor score diverges only 3% (1/26) from the optimal objectivity point.
Overall, the results of this research suggest that a small majority of external auditor participants in the study do exhibit a professional level of default scepticism.
Other observations arising from the data suggest that future research may be warranted to explore whether accounting training is more effective than fraud training for the purpose of mitigating trust bias and increasing the efficacy of evidence-based judgments, and to understand whether the high scores achieved by trained Accountants working in corporate roles are attributable to taking a preparer perspective, rather than a user perspective, when evaluating information.