Academic journal article Auditing: A Journal of Practice & Theory

Evaluation of Competing Hypotheses in Auditing

Academic journal article Auditing: A Journal of Practice & Theory

Evaluation of Competing Hypotheses in Auditing

Article excerpt

Key Words: Evaluation, Competing hypotheses, Auditing.

Data Availability: Contact the authors.

Auditors routinely evaluate competing hypotheses (e.g., when unexpected fluctuations are considered in conducting analytical procedures). In this paper evidence is presented on how auditors evaluate competing hypotheses. Hypothesis evaluation can affect audit efficiency, decision defensibility, judgment coherence and decision aid development. For instance, evaluating hypotheses as related entities allows evidence to play the dual role of directly (indirectly) implicating a particular hypothesis (the competing hypotheses). This strategy is efficient since it obviates the need to gather direct evidence about each competing hypothesis (i.e., confirming evidence about a hypothesis indirectly disconfirms the competing hypotheses). However, not gathering direct evidence about a competing hypothesis may be less defensible or may reduce confidence.

Hypothesis evaluation encompasses the cognitive relationship among hypotheses and the revision of hypotheses. Hypotheses could be evaluated as related cognitive entities or as separate cognitive entities (Van Wallendael and Hastie 1990).(1) With the former approach (hereafter, referred to as a "complementary" approach), hypotheses are treated as interrelated and evidence about a particular hypothesis affects the other hypotheses. Alternatively, each hypothesis could be evaluated as a separate cognitive entity. With this latter approach (hereafter, referred to as an "independent" approach), evidence is evaluated with respect to a particular hypothesis but not the competing hypotheses. Revision refers to changes in the probabilities of hypotheses upon receipt of additional evidence. A complementary revision is one in which an increase in the probability of a particular hypothesis is followed by a corresponding decrease in the probabilities of the competing hypotheses. The "one-hypothesis" syndrome refers to updating the probability of a hypothesis without a corresponding change in the probability of the remaining hypotheses.

Few studies have addressed the question of how auditors evaluate hypotheses. Further, two studies on this issue report conflicting findings. Heiman (1990) found that asking auditors to consider alternative hypotheses decreased their probability assessments of a target hypothesis. Presumably, auditors were aware that the hypotheses relate to each other; thus, the availability of a competing hypothesis is considered evidence against the target hypothesis. That is, the probability of any given hypothesis is discounted when other hypotheses are considered. Such a discounting strategy is consistent with a complementary evaluation. In a study that directly examined hypothesis revision, Asare and Wright (1993) found that auditors revised the probability of a single hypothesis and made no corresponding revisions in the probabilities o the other hypotheses. This resulted in a changing sum of the probabilities. Moreover, the sum of the probabilities of the hypotheses was supra-additive (i.e., exceeded one). Such hypothesis evaluation behavior is consistent with an independent approach.

The current study provides additional evidence and extends earlier research on how auditors evaluate competing hypotheses. Three issues, for which the alternative approaches yield different predictions, are delineated. The complementary approach suggests a negative association between the initial prior probability of the target (i.e., the most likely) hypothesis and the number of competing hypotheses. Moreover, no association is expected between the number of competing hypotheses and the summed probability ratings of the hypotheses. Lastly, probability revisions and summed probabilities are predicted to be complementary and additive (i.e., equals one), respectively. In contrast, under an independent approach, the initial prior probability of the target hypothesis is not expected to be associated with the number of competing hypotheses. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed


An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.