Audit EvidenceAudit evidence refers to the information obtained by an auditor in arriving at the conclusions on which an audit opinion is based. Decisions on the quantity and quality of the audit evidence to be obtained in any audit and the procedures for obtaining this evidence are central to the overall audit process. Guidance on these decisions is provided in the following auditing standards: International 500; Australian 502; Canada 5300; UK 400; USA 326. The International Standard on Auditing 500 (ISA 500) states that audit evidence is obtained from an appropriate mix of tests of controls and substantive procedures or entirely from substantive procedures. Substantive procedures include (1) tests of details of transactions and balances, and (2) analysis of significant ratios and trends including the resulting investigation of unusual items (referred to as analytical procedures“, which i have discussed. If haven’t, you may want to read it: Analytical Procedures in Auditing).



How to Obtain Audit Evidence

ISA 500 suggests the following procedures for obtaining audit evidence: inspection, observation, inquiry and confirmation, computation and analytical procedures.

Let’s recall the substantive test procedure explained in my previous post. Read on…


Inspection consists of examining records, documents, or tangible assets. Documentary audit evidence may consist of evidence created and held by third parties (e.g., securities or stock certificates held at a bank), evidence created by third parties and held by the client (e.g., listing of securities held by the bank and sent to the client, and evidence created and held by the client) (e.g., internal records detailing purchases and sales of investments). Inspection of tangible assets such as inventory and property plant and equipment is a direct way of verifying that an asset exists.

Observation involves the auditor looking at a process or procedure being performed by others, usually client staff. For example, an auditor could observe client staff clocking on or observe from a tour of the factory that some plant appears obsolete.

Inquiry involves seeking information from client staff or other knowledgeable individuals independent of the client. For example, client staff could be asked about reasons for a drop in sales revenue. Inquiries may be made of individuals outside the organization about a variety of items including movements in real estate values in a particular area, expected sales of various products, new technological developments that may affect the client, etc. Corroborative evidence should be obtained from the client.

Confirmation refers to the receipt of a written response from an independent third party to corroborate information contained in the accounts.

Computation involves checking the arithmetical accuracy of source documents and accounting records. For example, the auditor may check such items as depreciation calculations, a listing of accounts receivable, invoice totals, and the sales journal.

Analytical review procedures consist of the analysis of significant ratios and trends including the investigation of fluctuations and relationships that are inconsistent with expectations. Analytical review procedures include:

  • comparison of the financial information with information for the comparable prior period(s);
  • comparison of the financial information with anticipated results (e.g., budgets and forecasts);
  • comparison of the financial information with similar information based on industry averages, results of competitors or other operations in the same industry;
  • study of the relationships among elements of financial information that would be expected to conform to a predictable pattern based on the entity’s experience (e.g., gross margin percentages);
  • study of the relationships between the financial information and relevant nonfinancial information (e.g., payroll costs to number of employees).


Analytical review procedures vary from simple comparisons to reasonably complex regression analysis and time series modeling. In recent years there has been an increased emphasis on analytical review as part of substantive testing. Research has shown that analytical review can be a very effective means of detecting errors (e.g., Hylas & Ashton, 1982; Wright & Aston, 1989).


The Role of Judgment

The professional and academic auditing literature has recognized for a number of decades the importance and pervasiveness of judgment in auditing. More recently, professional standards have discussed the need for auditor’s judgments in an extensive number of areas. Some examples of auditor judgments include establishing materiality, identifying important audit objectives and assertions, assessing the inherent risk environment, evaluating internal controls, developing an audit strategy, selection and evaluation of analytical review procedures, evaluating the results of audit testing and determining whether the going concern basis is appropriate (Bamber et al., 1995). With respect to audit evidence Bamber et al. suggest the following questions need to be considered:

  • What audit procedures should be performed?
  • In what areas should audit effort be focused, and on what issues?
  • What should be the nature, timing and extent of audit procedures?
  • How should the evidence obtained from a variety of sources be aggregated? 
  • What audit conclusions may be drawn?


Sufficient Appropriate Evidence

Auditors are required to obtain sufficient appropriate audit evidence in order to draw reasonable conclusions on which to base their audit opinions. The judgments that will be examined here are those judgments needed to be made by the auditor in order to obtain sufficient appropriate audit evidence.

In deciding what appropriate audit evidence is, the auditor’s judgment will be influenced by such factors as mentioned below:  

  • the auditor’s assessment of the nature and level of inherent risk at both the financial report level and the account balance or class of transactions level;
  • the nature of the internal control structure and the assessment of control risk;
  • the materiality of the item being examined;
  • the experience gained during previous audits;
  • the results of audit procedures, including fraud, other illegal acts and errors which may have been found; and
  • the source and reliability of information available.


Let’s discuss in more detail on each of the above factor. Read on…


First, a number of studies have examined what factors are important to auditors in assessing the level of inherent risk. In general, it has been found that auditors perceive variables pertaining to characteristics of management and history of errors to be major determinants of the assessment of inherent risk (e.g., Monroe et al., 1993). While auditing standards suggest that the auditor’s assessment of inherent risk should affect the level of testing, early studies have found that changes in inherent risk do not have an effect on the amount of actual or planned audit testing (Mock & Wright, 1993).

Second, auditors’ judgments with respect to internal control systems are one of the most frequently studied issues in auditing. Most of these studies have extended Ashton (1974) who examined auditors’ internal control judgments over payroll. Ashton found that auditors placed greatest significance on the segregation of duties cues and that the extent of consensus between auditor judgments on the strength of the internal control systems was relatively high compared to the consensus found in studies of other types of expert judges, e.g., stockbrokers and radiologists. The Ashton (1974) study has been extensively replicated and extended by subsequent research. Trotman & Wood (1991) identify 17 studies that examine consensus in internal control judgments. They conclude that while there is considerable variation in consensus between studies, overall the results show higher mean consensus than is typically reported in non audit studies. This level of consensus was increased by use of the review process (Trotman & Yetton, 1985). Many of these studies also examined cue usage, judgment stability over time, and self-insight into cue usage. In all of the studies that examined cue usage, a policy-capturing model explained almost all of the variance in subjects’ judgments. Normally about four to six cues were found to be significant and there were considerable individual differences in cue usage across auditors (Solomon & Shields, 1994). Bonner (1990) found that differences in consensus in cue weighting between experienced and inexperienced auditors was greater for analytical risk assessments (expert task) than control risk assessments (novice task). While earlier auditing studies found almost no evidence of configural cue usage, more recent studies by Brown & Solomon (1990; 1991) provided evidence of configural cue processing in situations where the nature of the configural processing was specified ex ante based on the context of professional evaluations. Studies that have also examined judgment stability over time have found that stability has been higher than consensus across subjects which is consistent with studies in other areas of judgment/ decision-making. Auditors have also demonstrated a reasonably high level of self-insight in connection with their use of information.

In forming an opinion about an account balance, class of transactions, or a control, the auditor generally does not examine all of the information available. Instead, the auditor uses judgmental or statistical sampling procedures. While there has been reasonably high levels of consensus on internal control evaluations, this has not been the case for sample sizes. A study by Mock & Turner (1981), for example, found large differences in recommended sample sizes by experienced auditors working on detailed work papers.


Third, there have been a number of studies that have examined materiality. When planning an audit, auditors need to consider what factors could generate material misstatements in the financial accounts. Most studies have found that the percentage effect of an item on profit is the most important factor in deciding if an item is material (Messier, 1983; Chewing et al., 1989). Other factors found to be significant included earnings trend and total assets. Studies have also found differences between users of accounts and auditors as well as large differences in materiality thresholds between auditors (e.g. Chewing et al., 1989).

Fourth, prior knowledge of clients gained during previous audits can also affect the extent of testing needed. For example, Houghton & Fogarty (1991) carried out a survey to determine the characteristics of auditor-detected errors and whether areas in which errors occur could be determined during the planning process. They found that 73 percent of errors examined were or could have been identified during the planning stage based on prior knowledge of the client.

The fifth factor is the results of audit procedures, including fraud, illegal acts, and errors which have been found. The usage of a checklist of potential fraud indicators has become an important part of the risk assessment process in practice (see Gushing et al., 1995). With respect to other errors, Bedard & Biggs (1991) show that auditors can improve hypothesis generation of possible errors by examining patterns of discrepancies.

The final factor relates to the source and reliability of information available. In contrast to non auditing studies, research that has examined auditors’ sensitivity to the reliability of the source of the evidence has generally found that auditors are relatively sensitive to the reliability of evidence (e.g. Hirst, 1994).


Auditors Subject to Heuristics and Biases

Early studies of probabilistic judgments by auditors suggest that auditors are also subject to many of the heuristics and biases found in the psychology literature, such as representativeness and anchoring and adjustment. This research has recently been extended in a number of areas directly related to auditor’s collection and evaluation of evidence. For example:

  • Auditing researches have used the Hogarth & Einhorn’s (1992) belief-adjustment model, which assumes that belief-adjustment follows an anchoring and adjustment process, to study audit evidence which is obtained in a sequential manner.
  • Several studies (e.g. Ashton & Ashton, 1988; Tubbs, 1990) have found that auditors place more weight on evidence received most recently. This is known as a “recency effect“. Such an effect has major implications for audit practice, since this suggests that the order in which auditors receive and evaluate evidence may have a substantial impact on decision-making.


That is, two auditors may receive exactly the same evidence but in varying order and subsequently arrive at different conclusions, thus potentially reducing audit effectiveness or efficiency.

Biases may not only occur in processing evidence but also in the search for evidence. A number of studies (e.g. Kida, 1984) have investigated evidence-search strategies used by auditors. Auditors often explicitly or implicitly formulate hypotheses to explain certain factors (e.g., a change in key ratios during preliminary analytical review) and then search for evidence to test the hypothesis. Kida (1984) examined whether the hypothesis-testing strategies employed by auditors affect their search for evidence. Kida noted that audit tasks require auditors to sift through a number of pieces of information, some of which can provide confirming evidence and some disconfirming. The overwhelming conclusion from the psychology literature is that individuals preferentially collect evidence that tends to confirm rather than disconfirm their hypothesis. Although Kida found limited support for the existence of confirmatory strategies, the effect was less powerful than found in many psychological studies. Subsequent research has provided very little confirmation of the presence of confirmatory strategies in the information search and recall process of auditors.

There has also been audit research to suggest that the search for confirming and disconfirming evidence related to a hypotheses can be influenced by an already-generated hypothesis. This is known as “interference effects” and has been found in studies examining analytical procedures (e.g., Libby, 1985; Libby & Frederick, 1990; Heiman, 1990).