3.1 The audit expectation gap

The audit expectation gap relates to the difference between what users expect from the auditor and the financial statement audit and the statutory requirement for an audit. The issuance of an unqualified audit opinion in the auditor's report demonstrates that the auditor concludes the audited financial statements provide a true and fair view of the financial position and performance of the company and comply with the relevant financial reporting framework. Prior research has found the users of financial statements (such as investors, bankers, and financial analysts) often perceive that auditors provide an absolute level of assurance when they read such information. In other words, they perceive auditors guarantee the accuracy of the audited financial statements, leading to naïve and unreasonable expectations (e.g., Epstein & Geiger, 1994).

However, the IAASB merely requires the auditor to obtain a reasonable level of assurance since inherent audit limitations exist to prevent auditors from achieving absolute assurance (IAASB, 2009a). McEnroe and Martens (2001) performed a survey of audit partners and investors on a number of different attributes of auditor's responsibilities. They found many significant differences on what an auditor should do or judge before issuing a qualified opinion, for example, fraud detection. Investors' assigned much greater responsibility on auditors than would be reasonable according to what is required by the auditing standards. Anderson, Maletta, and Wright (2003) also observed that users assign greater responsibilities to auditors. For instance, they perceive that it is the auditor's responsibility to prepare and present the financial report. However, that is clearly the management's responsibility. Users may erroneously associate audits with a guarantee of the absence of fraud and a recommendation for them to make investment in the respective firm. These types of differences in perceptions make up the audit expectation gap.

A recent survey by the Association of Chartered Certified Accountants (ACCA) of 11,000 members of the public from around the world support these findings. They show that users continue to expect more from auditors compared to what they actually do (ACCA, 2019). They describe this component of the expectation gap as the knowledge gap and note that the introduction of KAMs by the IAASB is an important step in reducing this gap. Overall, prior research shows that users generally expect far more than the audit currently delivers. Therefore, if KAMs are to reduce the audit expectation gap, they need to reduce users' perceptions on what the audit provides.

3.2 Audit report formats

Chong and Pflugrath (2008, p. 223) note that the primary strategy of regulators to reduce the expectations gap is to “provide additional explanations, clarifications and condensed summaries of the auditors' role and the audit process in the audit report” (e.g., SAS 58, SAS 600, and AUP 3). There are distinct but related possibilities in which changes to the format of the audit report may be useful to minimize the expectation gap as outlined in Mock et al. (2013). As audit reports are often criticized for adopting standardized wording, the first possibility is to depart from that, allowing auditors to decide what shall be said to satisfy the need of users in the audit report. The second possibility is to maintain the standardized wording and expand the audit report. This is to enable reporting of significant management and auditor's judgments involved in the audit, the inherent limitations of the financial report, and the scope, nature, and extent of the audit work.

It can be seen from standards implemented over the last two decades that standard setters have paid more attention to the second possibility (i.e., ISA 700, IAASB, 2009b). However, users may lose interest in the auditor's report as it becomes more standardized (Mock et al., 2013). Users' interest is important since this strategy depends on users reading the lengthy description of the responsibilities of auditors and the audit process.

In the 1980s in response to the findings of the Cohen Commission, a new audit report standard SAS 58 was released in the United States. Research that investigated the effect of these changes to the format, wording, and content of the audit report in SAS 58 concluded that it did not significantly reduce the expectation gap. It was found that this expanded audit report made no difference in users' perceptions (Bailey, Bylinski, & Shields, 1983). Monroe and Woodliff (1994) examined the impact of a then-new Australian reporting standard—AUP 3. This standard provided greater details of management's responsibilities and a better explanation of the audit objectives and procedures in the scope section of the auditor's report. They reported that users' perceived responsibilities of auditors and management altered significantly. However, new and widening differences occurred in other areas, thus the authors concluded that an expanded audit report has a limited impact on the expectation gap overall.

Chong and Pflugrath (2008) in a between-subjects design compared the SAS 58 report; the ISA 700 plain language report (by removing jargon and presenting the audit report using a language that is easily understood by users) at the beginning of the financial statements; and the ISA 700 plain language report at the end of the financial statements. The differences between shareholders' and auditors' perceptions were still evident for the expanded format compare to short format (audit-opinion only), although the plain language expanded report with the audit opinion at the start, the differences decreased but not significantly. In another experiment on the revised ISA 700, Gold, Groneworld, and Pott (2012) manipulated an auditor's report with explanations as mandated by ISA 700 compared to an opinion only audit report. They found that more explanations do not result in a narrower expectation gap, and the audit opinion alone would be enough to signal sufficient relevant information.

It is clear that the auditing research literature shows the various changes in formats of the audit report over the years have made little impact on users' perceptions and failed to reduce the audit expectations gap as the gap exists and continues to persist.

3.3 Key audit matters

The IAASB and the PCAOB have implemented several significant changes to the auditor's report. The IAASB were quite explicit that an important motivation for the changes is an attempt to reduce the expectation gap between auditors and users (IAASB, 2011). Among those changes, there is the requirement for auditors to identify and disclose KAMs in a separate section of the auditor's report. KAMs discuss areas of the audit that were of most significance or required significant auditor attention in performing the audit (ISA 701, 2015). The proposed auditing standards also require the auditor to describe the identified matters and to explain why the matters were considered to be of most significance in the audit, or in simple terms, appropriate for disclosure.

The audit report has often been criticized as a standardized boilerplate because it follows specific wording and format requirements set in the auditing standards (e.g., ISA 700), which can reduce its communicative value (Gray, Turner, Coram, & Mock, 2011; Mock et al., 2013). A KAM paragraph responds to investor demands for more company-specific information that can be relevant and useful for their decision-making. Sirois, Bédard, and Bera (2018) in evaluating users' information acquisition using eye-tracking technology find that KAMs do direct the attention of users to disclosures related to KAMs communicated in the auditor's report. However, communication of several matters in the auditor's report reduce attention to other parts of the financial statements, that is, a substitution effect.

Much of the behavioral research on KAMs so far performed has focused on evaluating auditor liability. In an experiment conducted by Brasel et al. (2016) using lay jurors as participants, they find auditors who disclosed a KAM related to the undetected misstatement were judged to be less liable than auditors who did not. Brown et al. (2018) report that KAMs have the effect of reducing negligence assessment and did not have an adverse effect on increasing damages. Kachelmeier et al. (2019), who use MBA students as participants, observe that users' perceive less auditor liability for a misstatement in a financial statement area disclosed in the auditor's report as a KAM compared to a financial statement area not disclosed as a KAM. Overall, these studies suggest that KAMs do not increase auditors' liability and the likely explanation for these findings is that KAM disclosures are regarded as a partial disclaimer of auditor responsibility.

In contrast to these findings, Gimbar et al. (2016) find that that jurors perceive that KAMs constrain auditors' control over financial reporting outcomes so they perceive lower liability for auditors, however, this is only in a scenario where there are precise accounting standards. Backof et al. (2018) report that jurors perceived auditors to be more negligent when KAM disclosures relate to an undetected misstatement. However, they find that clarifying the limitations of reasonable assurance provided some protection from increased liability.

The focus of previous research on perceived auditor liability (while important) is more at the extreme end of reactions to the provision of KAM disclosures. KAMs are providing information to users of financial statements and so it is reasonable that they may affect users' perceptions on responsibility and reliability of the audit information provided without necessarily thinking about legal liability.

Prior research has repeatedly indicated the existence of an expectation gap between auditors and users in relation to the auditor's responsibilities and the reliability of the audited financial statements. Various institutional changes have been made to narrow the gap, the proposal of disclosing KAMs in the audit report is one such attempt.

One of the reasons for the implementation of KAMs was to reduce the expectation gap by improving the communicative value of the audit report. However, the prior research on previous unsuccessful efforts to reduce the expectation gap by changing the audit report (see Mock et al., 2013) and recent mixed findings on the effect on auditor liability from disclosing KAMs (Backof et al., 2018; Brasel et al., 2016; Brown et al., 2018; Gimbar et al., 2016; Kachelmeier et al., 2019) suggest that the effect of this type of disclosure may not affect the perceived responsibility of auditors and perceived reliability of the financial statements—and thereby affect the audit expectation gap.

In this study we first explore whether the inclusion of KAMs changes users’ perceptions on a couple of important aspects of the audit expectation gap as outlined in the following research questions:

3.4 Precision of accounting standards

An important issue that may affect the value of KAMs in the audit report relates to the nature of the disclosure made by the auditor. A significant issue that affects accounting disclosures is the precision of the accounting standards and this may be a relevant issue in determining user perceptions of auditors' responsibility (Gimbar et al., 2016). The precision of accounting standards concerns whether they relate to subjective or objective matters and the effect of this on the perception of users about auditors' responsibility.

A subjective KAM is governed by imprecise accounting standards, which are more associated with principles-based standards, while an objective KAM is governed by precise accounting standards, which are more associated with rule-based standards. Significant professional judgment and estimates are required in the application of imprecise standards as they provide less-detailed guidance to practitioners and auditors. Almost all companies are required to prepare their financial report as set out by the IASB, whose standards are often regarded as imprecise (generally principles-based). In contrast, the precise standard is basically a list of detailed rules that must be complied with when preparing financial statements. The most prominent example of this is in the United States where generally accepted accounting principles (GAAP) are frequently referred to as rules-based standards. Gimbar et al. (2016) showed that critical audit matters relating to different types of accounting standards had an impact on users' perceptions of auditor liability.

Disclosure of KAMs implies that the auditor has identified and considered the matters as requiring a significant amount of professional judgment or posing difficulty in obtaining and evaluating evidence (PCAOB, 2017). Users are likely to view the client's compliance with the precise standards as evidence that the auditor intended and did fulfill their responsibility to perform a high-quality audit (Kadous & Mercer, 2016). It is expected users will assess auditors as having less responsibility when a KAM disclosure relates to a precise standard (objective KAM) because they are perceived to have limited control over the underlying accounting treatment. The perception is consistent with Jamal and Tan's (2010) argument that the ability of auditors to negotiate with clients to implement an alternative accounting policy is constrained when they disagree with a proposed accounting policy that complies with a precise standard. By contrast, imprecise standards emphasize the importance of professional judgment and auditor skill (Carmona & Trombetta, 2008).

Practitioners may have different interpretations toward the same standard after considering a variety of contingent factors. Imprecise standards remove the auditor's constraint on negotiation with clients under precise standards and auditors are likely to have more control. This can enable them to require clients to make adjustments in the financial report. By disclosing a KAM on an imprecise standard (subjective KAM), it indicates that the auditor was aware of “the heightened risk associated with the accounting treatment” (Gimbar et al., 2016, p. 1633). Under these circumstances, users can perceive the auditor should have performed necessary audit procedures to assure relevant accounts are fairly presented. We present this as a research question as follows:

Jamal and Tan (2010) argue that precise standards constrain the ability of auditors to negotiate with clients when there is a dispute about a proposed accounting treatment that conforms to a precise standard. This is reasonable to the extent that the client may argue its accounting treatment is consistent with the relevant standards and there is no need to make an adjustment. Auditors are not able to give a modified opinion merely because of this disagreement. Instead, they could disclose the issue as a KAM in the audit report. The disputed issue may be determined to be a KAM because it is of a higher assessed risk of material misstatement (ISA 701, para 9(a)).

However, imprecise standards remove this constraint, enhancing the negotiating power of auditors (Gimbar et al., 2016). The auditor is more able to require the client to alter the accounting treatment. Clients are more likely to amend their accounting treatments if auditors provide alternative interpretations toward relevant standards because auditors can express a modified opinion if a disagreement occurs. The main objective of disclosing the disputed issue as a KAM in this situation instead is to simply indicate significant difficulty encountered in the audit process as per para 9 (b) in ISA 701 (e.g., time spent for the auditor to negotiate with the client).

However, earlier research on this topic generally came to a different conclusion that less-precise standards gave auditors a chance to acquiesce to their clients' demands because there is no bright line to constrain aggressive reporting (Hackenbrack & Nelson, 1996; Ng & Tan, 2003; Trompeter, 1994). If this is the way users' perceive auditors behave where there are less-precise standards, we expect that this manifests in their perceptions more in terms of the reliability of the audit rather than the auditor's responsibility. Given the mixed findings on this we pose our final research question as follows:

4.1 Participants

Recall that, the objective of this study is to evaluate whether the changes implemented by KAMs have reduced the audit expectation gap. The expectation gap is made up of an information gap and a communication gap (Mock et al., 2013).² Central to both of these gaps are the financial statement users as presented in the model developed by Mock et al. (2013). Some prior expectation gap studies have sought the views of auditors (e.g., Chong & Pflugrath, 2008; Gold et al., 2012). However, we have focused on users because the expectation gap is the difference between what users expect from the auditor and the financial statement audit, and the statutory requirement for an audit (IAASB, 2011). The statutory requirements clearly outline what an audit is meant to provide. Therefore we think that by focusing on how users' perceptions change and whether their perceptions move closer to what is defined in the auditing standards helps answer the question on whether KAMs have reduced the audit expectation gap.

We obtained financial statement users with a range of levels of financial sophistication to represent a variety of user groups to participate in this experiment. The participants were obtained from Amazon Mechanical Turk (AMT), which is an online web-based platform for recruiting and paying workers to perform tasks (Berinsky, Huber, & Lenz, 2012).³ The first group selected were users who were working in the areas of accounting and finance (excluding auditors) to enable an assessment of more sophisticated investors. The second group selected were graduates that in AMT are defined as those holding a bachelor's degree to represent nonsophisticated investors. Compared to the average person in society, graduates should have some knowledge of financial reporting practices and a number of them have been involved in some levels of investing activities. They have the ability to make reasonably informed assessments, but are not experts so they are an appropriate representation of nonsophisticated investors.

In running the second experiment we excluded workers whose job function was accounting or finance.⁴ Both of these two groups were located in the United States. The recruitment of participants was in two stages to ensure there was no overlap. A possible concern was that the two participant groups may overlap because those working in the accounting and finance industry probably also have a bachelor's degree. The two groups were analyzed together because as a total group they all fit within the definition of users of financial statements. However, an advantage of obtaining these two different groups is that obtaining a range of differing levels of task expertise and diverse backgrounds allows us to make conclusions more generally on users' perceptions of KAMs.

The total sample contains responses from 240 participants, which includes 120 from each group. Each participant within each group was randomly assigned to an experimental condition. In terms of the overall group, the average participant was 40 years old and had 7 years of work experience. Further, 52% of the sample were females and 8% were full-time students. Participants with CPA or ACCA qualifications comprised 12% of the total sample, a percentage that is consistent with previous studies that have used users of financial reports (refer to Hasan, Roebuck, & Simnett, 2003).

Participants were also asked questions relating to audit reports. They were asked: first, how often they read the audit report when considering whether to invest in a company (11-point scale with 0 never and 10 always) and the mean response was 7.37; second, they were asked their familiarity with the audit report (11-point scale with 0 not at all familiar and 10 very familiar) and the mean response was 6.58; and finally, they were asked their understanding of the report (11-point scale with 0 not at all easy to understand and 10 very easy to understand) and the mean response was 7.46. Overall, the participant group had a reasonable knowledge of audit reports based on their self-reported evaluations.

The significant differences between the two groups were as expected relating to work experience, professional accounting qualifications, and use, familiarity, and understandability of the audit report.

4.2 Experiment task design

A 3 × 1 between-subjects experiment was conducted that involved manipulation of audit reports with three variables: no KAMs, subjective KAM, and objective KAM. The experimental instrument started with instructions for participants and some background information about a hypothetical stock-listed airline company (see Appendix). Providing too much financial information on the company could have diluted participants' attentiveness to the audit report (consistent with Kachelmeier et al. (2019)), and it was not necessary for the objectives of this study. All participants then received a copy of the auditor's report on the financial statements that began with the standard wording for an unqualified audit report adopted in current practice.

The between-subjects manipulations was as follows. The first was no provision of a KAM so no extra information was provided beyond the standard audit report. The second was the provision of a subjective KAM and some details of the relevant accounting standard with a principles-based focus. Finally, the third manipulation was the provision of an objective KAM, which was also provided with some details of the relevant accounting standard with a rules-based focus. These KAMs were adopted from actual KAMs from a couple of audit reports.

The relevant accounting standard used in the experimental materials related to revenue and this is a very common KAM issue. According to KPMG's Auditor's Report Snapshot (KPMG, 2017), it is reported that 20% of the entities they observed reported a KAM on revenue in their audit report. The company in the case study was an airline company and from perusal of audit reports from a number of airlines it became apparent that the accounting treatment for revenue recognition for customer loyalty programs (CLP) is particularly highlighted in most of the top airlines' audit reports. The likely explanation is that although International Financial Reporting Interpretations Committee (IFRIC) 13 has been released in an attempt to standardize alleged widespread and divergent accounting practice in accounting for customer loyalty programs since July 1, 2008 (Chapple, Moerman, & Rudkin, 2010), some problems still remain unresolved. For example, IFRIC 13 requires using fair value accounting for revenue recognition, which is the amount for which the award credits could be sold separately (IFRIC 13, IASB, 2008). However, no clarification has been given regarding how to measure this amount, and thus the measurement is ultimately determined by the reporting entity itself. This means that this topic is an ideal one to manipulate the accounting standard type (imprecise vs precise) in an experiment without violating requirements set out in IFRIC 13.

Within the audit report, we manipulated KAM disclosure at three levels: without KAM, subjective KAM, and objective KAM. In the without KAM condition, participants received the standard audit report as is currently mandated by the U.S. auditing standards.⁵ Although the study was undertaken in the United States, we decided to call the additional disclosures KAMs even though they will be called critical audit matters (CAMs) when they are required to be disclosed in that jurisdiction. The reasoning for this is that CAMs had not been implemented at the time of undertaking this experiment. CAMs were implemented in the United States first for large accelerated filers for fiscal years ending on or after June 30, 2019. It is therefore unlikely that the participants have seen either a KAM or CAM and the probability that they have been following the development of these new auditing standards is low. Therefore, for the purposes of developing this experiment and evaluating users' perceptions we decided to use the current international terminology, which is KAMs.

In both KAM conditions, the audit report included a separate paragraph titled Key Audit Matters, described as “those matters that, in our professional judgement, were of most significance in our audit of the Financial Report of the current period” (para. 8).⁶ The KAM in the subjective condition stated that the company's unredeemed frequent flyer revenue was measured by applying a forward-looking model and this measurement involved significant management's judgment and estimates. The KAM in the objective condition stated that the unredeemed frequent flyer revenue's measurement was calculated by actuarial specialists based on historical redemption trends, and this measure may not reflect changes in the environment in the future.

The given accounting standard excerpt precision was also manipulated across different experimental treatments. In the subjective condition, the accounting standard excerpt merely stated that the unredeemed revenue was measured by reference to their fair value and it especially emphasizes that judgment was required to select and apply an estimation technique that should be most appropriate in the circumstances. Alternatively, in the objective condition, the excerpt of accounting standard explicitly required that actuarial specialists should be employed to make the estimate and their estimate shall be based on historical redemption trends. The wording of these manipulations is presented in the Appendix.

4.3 Dependent variables

Following the case study information, participants were asked to respond to a series of questions relating to the two expectation-gap constructs (discussed below) on 11-point Likert-type scales. The order of questions was randomized, and the questions were followed by demographic questions and manipulation checks. The demographic questions relate to the readership of, familiarity with, and understandability of the audit reports that are used as proxies for the financial statement users' auditing/accounting background.

The audit expectation gap is normally measured based on multiple-item constructs in prior research (i.e., auditor responsibility factor, financial statement reliability factor, and management responsibility factor). For the purpose of this study, we focus on the first two of these constructs because they are where KAMs would most likely have affected perceptions relating to the audit expectation gap.

The questions in this study to measure the audit expectation gap were drawn from the types of questions used in recent expectation-gap studies (e.g., Chong & Pflugrath, 2008; Gold et al., 2012), which originated from the work of Libby (1979). In the broadest sense, the questions allow us to understand what the public expects from an audit, with a focus on the responsibilities of auditors and the reliability of the financial statements. All of the questions in our study can be framed using Porter's (1993) definition of the audit expectation gap that encompasses two elements: a performance gap and a reasonableness gap.

The performance gap is defined as the gap between what society can reasonably expect auditors to accomplish and what they are perceived to achieve, which is associated with auditors' existing duties (Porter, 1993). In the Appendix, a number of questions in the instrument captures this gap. These questions relate to the extent they agree the audited financial statements comply with accepted accounting practice, whether they agree the financial statements are free from material misstatement, and their confidence level on the amounts of unredeemed frequent flyer revenue.

The reasonableness gap is defined as the gap that exists between what society expects auditors to achieve and what they can reasonably be expected to accomplish (Porter, 1993). These questions are also presented in the Appendix and asked users: the level of assurance the auditor provides; who should take the responsibilities of preparing and presenting the financial statements, and making accounting estimates and judgments; who is responsible if the particular revenue account is materially misstated; and to what extent they agree the auditee is free from fraud.

4.4 Experimental procedures

Each participant was given a link to an online case study hosted by Qualtrics, which is a web-based tool for creating and conducting online surveys. Using the outside website helped us to obtain informed consent and implement additional screening procedures. The survey's recruiting narrative identified this study as research in accounting about the informative value of an audit report without mentioning anything related to the KAM proposal. Three experimental treatments (unqualified auditor's report without KAMs, or one subjective KAM, or one objective KAM) were distributed randomly to the participants. The task took an average of 15 minutes to complete.

5.1 Manipulation checks

Due to the nature of collecting data using the AMT platform, careful checking was done to ensure the quality of the data. All participants answered two manipulation check questions after they had completed the experimental survey. The first manipulation check related to the effectiveness of the manipulation of the presence versus absence of the KAM in the auditor's report. Participants were asked the extent to which they agreed with the statement, “The auditor's report provided in the materials explicitly described matters that involved especially challenging, subjective, or complex auditor judgement in the audit of the Financial Report” (11-point scale with 0 strongly disagree and 10 strongly agree). Matters that explain the especially challenging, subjective, or complex concerns pertaining to the audit are called critical audit matters (CAM) in the United States. The frameworks for determining a CAM/KAM are substantially similar, so the definition of CAM was adopted to design this manipulation check question. The rationale for this choice was because the definition of KAM was given to the participants in the experiment and therefore it had been explicitly defined and presented to them.

The second manipulation check was to verify the effectiveness of the manipulation of the precision of the accounting standard. Participants were next asked to what extent they agree with the statement, “the accounting standard excerpt in the materials provide clear guidance on how to account for award credits in customer loyalty programmes” (11-point scale with 0 strongly disagree to 10 strongly agree). This question should only be answered if an accounting standard excerpt was provided in the material (i.e., in either Subjective KAM or Objective KAM condition). All experimental questions except this second manipulation check question forced participants to answer the questions before they could progress to the next page of the survey.⁷ It was expected participants would skip this question if no accounting standard excerpt is provided (i.e., No KAM condition). However, 29 out of the 83 (34.94 percent) participants who received an audit report with no KAM disclosed (and therefore no accounting standard manipulation) answered this question. Forty-eight out of the 157 (30.57 percent) participants did not answer this question when they were provided with an audit report with KAM disclosed (with an accounting standard manipulation) (26 from the Subjective KAM condition and the remaining 22 are from Objective KAM). These failures indicated a lack of attention of participants in reviewing the experimental materials, so they were deleted from the main analysis of the results.

To further ensure that the results reflect responses from participants who adequately comprehended the experimental material, the reported analyses excluded the following participants. First, two participants who answered 9 or 10 for the first manipulation check question in the No KAM condition; second, two participants who answered 0 or 1 for the first manipulation check question in the With KAM condition; third, 19 participants who answered 9 or 10 for the second manipulation check question in the Subjective KAM condition; and finally, two participants who answered 0 or 1 for the second manipulation check question in the Objective KAM condition.

After removal of those who did not pay sufficient attention to the case materials and those who failed the manipulation checks, a total of 138 participants were included in the primary analyses. While this is a high number of removals, it is not unusual when using AMT and it provides confidence that the group we have analyzed provided appropriate attention to the experimental materials. For participants, the overall mean response for the first manipulation check question was 4.12 for the No KAM condition, 7.20 for the With KAM condition. The means in the two different conditions are significantly different at a 1 percent significance level. The manipulation check was also successful when conducted separately for each group (i.e., sophisticated users versus nonsophisticated users). In terms of the second manipulation check question, the overall mean response was 6.00 for the Subjective KAM condition and 7.33 for the Objective KAM condition. The means are significantly different (p < 0.005), indicating a successful manipulation. The manipulation check was also successful when conducted separately for each group (largest p < 0.01).

5.2 Main findings

To evaluate whether there were any significant effects, MANOVAs (multivariate analysis of variance) on perceived audit responsibility and reliability were performed. Descriptive statistics of perceived measures of audit responsibility and reliability are presented in Table 1. A MANOVA is only presented in this article that compares perceptions of reliability between subjective and objective KAMs to address RQ3 in Table 2.

In examining whether inclusion of a KAM affects users' perceived audit responsibility, in the overall model it was found that the four variables associated with auditor responsibility were not affected by whether there was a KAM or not (F = 0.27, p = 0.90). Further, none of the four individual variables were found to be significant. Therefore, in evaluating RQ1, we do not find any evidence that KAMs affect users' perceived audit responsibility compared to when no KAMs are included in the audit report.

Research Question 2 related to whether inclusion of a KAM affects users' perceived audit reliability; in the overall model it was found that the four variables associated with audit reliability were not affected by whether there was a KAM or not (F = 1.75, p = 0.14) and none of the four individual variables were found to be significant. Therefore, for RQ2 we also do not find any evidence of KAMs making a difference to users' perceptions of the reliability of the audit.

Research Question 3 explores whether users perceive auditors to have less responsibility when an objective KAM is disclosed compared to a subjective KAM. In the overall model we find that the four variables associated with auditor responsibility are not affected by whether the KAM is subjective or objective (F = 0.60, p = 0.66). Therefore, we cannot conclude on any differences in addressing RQ3.

Finally, RQ4 examines whether there are differences between a subjective or objective KAM in the perceived reliability of the audit. A MANOVA was performed to test for an overall effect on the reliability measures and then univariate analyses were done for each of the variables as show in Table 2. In the overall model presented in Table 2, Panel A, we find that the four variables associated with perceived reliability are affected by whether an objective KAM compared to a subjective KAM was provided. The provision of an objective KAM is associated with an increase in the perceptions of reliability of the audited financial statements, which suggests that this disclosure is actually increasing the audit expectation gap.

In examining the significance of the individual variables, this is reported in Table 2, Panel B. Two of the four reliability variables are significant. First, Question 5, which related to the extent users perceived the entity was free from fraud, there was a mean response of 8.96 with an objective KAM and 7.31 with a subjective KAM (10 is strongly agree) as can be seen in Table 1, and this is significant (F = 12.39, p = 0.00). Second, Question 8, which related to the extent that fair value disclosure was credible, there was a mean response of 9.06 with an objective KAM and 7.97 with a subjective KAM (10 is strongly agree) and this is significant (F = 7.07, p = 0.01). The other two variables also provided mean response differences in the same direction but were not significant, although one was marginally significant.