Confirmation bias—one of the five commonly occurring judgment biases—has the potential to trip up auditors, particularly during the early stages of an audit. At that time, financial information is often highly aggregated and may be too ambiguous to allow the auditor to definitively identify the reason for a change in financial information.
As a result, an auditor’s initial hypothesis may not actually represent the true cause of the data fluctuation. Furthermore, the deeper one gets into investigating a particular hypothesis, the more difficult it becomes to consider other potential hypotheses. This is because once a potential explanation has been identified, it is common to seek evidence that supports the explanation and ignore evidence that disconfirms the explanation.
This is the behavior psychologists refer to as confirmation bias. As such, if the auditor generates an early hypothesis, he or she risks overlooking important contradictory evidence. This could result in a flawed evaluation of the data.
So what can be done? Auditors can take several simple and pragmatic steps to overcome this bias when performing analytical procedures. In fact, the following actions could lead to improved decision-making in other areas of the audit as well:
1. Take it all in: Don’t jump to conclusions. Treat the initial data-gathering stage as a fact-finding mission without trying to understand the specific causes of any identified fluctuations. That is, resist the temptation to immediately generate potential hypotheses, and instead wait until a more complete information set has been reviewed before considering reasons the data may differ from expectations.
2. Brainstorming: The rule of three. If possible, identify three potential causes for each unexpected data fluctuation that is identified. Why is three the magic number? Research has shown that auditors who develop three hypotheses are more likely to correctly identify misstatements when performing analytical procedures than those who develop just one hypothesis. From a probabilistic standpoint, the more plausible expectations brainstormed, the higher the likelihood that the underlying cause of the fluctuation will be identified. However, developing too many initial hypotheses may constrain the auditor’s ability to efficiently evaluate each potential explanation. Research published in the Journal of Accounting Research in 1999 revealed that auditors who develop three hypotheses are actually more efficient at, and just as effective at, identifying misstatements through the use of analytical procedures as those who develop more than three hypotheses.
3. Flag it. When identifying potential causes of a financial fluctuation, take note of the specific information that triggered the generation of the hypotheses. Present those data to a colleague to see whether he or she comes up with similar explanations. If the explanations are different, the colleague has assisted you in expanding your hypothesis set, thereby improving your chances of identifying the true explanation for the fluctuation. If the explanations are similar, the colleague has provided you with some validation of your existing set.
4. Prove yourself wrong. Once an initial set of hypotheses has been developed, the natural course of action is to seek out evidence that confirms these explanations. However, by simply accepting confirmatory evidence as support for a potential hypothesis, it is easy to ignore the fact that the evidence could also indicate a different explanation at the same time. In a similar fashion, it is also common to subconsciously ignore contradictory evidence. Unfortunately, both actions lead to a potential confirmation bias that may cause faulty judgments. So instead of searching for confirmatory evidence, try to disconfirm your initial suspicions by actively seeking out and considering contradictory information. Such an approach is likely to lead to stronger and more definitive conclusions.
5. Circle back. After identifying your initial hypotheses, the next step is to investigate the data further to determine which (if any) is the actual cause of the data fluctuation. While performing this investigation, additional information will invariably be analyzed to confirm or disconfirm these explanations. Don’t forget to “circle back” and consider new hypotheses when examining these new data. Remember that successful hypothesis generation during the performance of analytical procedures is an iterative process.
Also read "I’m not biased, am I?," Feb. 2015, page 26