Auditors’ performance evaluations: An experimental analysis of the effects of initial impressions and task-specific experience on information later recalled

Abstract

Professional auditors at two firms completed a computerized task designed to test the effects of initial impressions and experience upon the information later recalled about a subordinate. Subjects were given either a positive, negative, or neutral initial impression with respect to the subordinate's technical competence. They then reviewed randomly presented behavioral descriptions, completed an unrelated “distracter” task, and completed a free recall task related to the subordinate from the original task. Subjects inheriting a negative initial impression exhibited a greater tendency toward recall of weakness-related information compared to those given no specific initial impression. Those given a positive initial impression did not exhibit a similar confirmatory tendency. Experience in performance evaluation was associated with greater recall of unfavorable information, but only for subjects given no initial impression.

Keywords:

• Performance evaluation
• Initial impressions
• Recall
• Task-related experience
• Auditors
• Public accounting

Acknowledgement

The authors wish to thank the participants of the American Accounting Association 2006 Annual Meeting, the 2006 Ohio Region Meeting and the 2006 Hawaii International Conference on Business for their comments on earlier versions of this paper.

Notes

3 Rothbart et al. (1979) suggest that disconfirming items, due either to their infrequent or surprising nature, may require more attention and may be processed at a deeper level.

4 It should be noted that the intent here is not to use experience as a pure “surrogate” for expertise in the sense referred to in much of the previous literature (e.g. see Davis & Solomon, 1989 for a discussion of the potential problems with such an approach). The task is more related to person perception than to the participants’ own performance of specific audit-related tasks. In this paper, there is no assumption that experience in performance evaluation leads to “expert” performance in the task of evaluating others, except that experienced individuals may exhibit different tendencies. The interest is in the relationship between experience and the amount and types of information participants remember about another person.

5 The initial impression is based on informal discussions with the senior on a previous audit to which the hypothetical senior was assigned. Anecdotal evidence suggests that impressions may be gained in an informal manner apart from viewing actual performance evaluations from previous audits. Furthermore, since auditors are often not assigned to a particular supervisor on a continuing basis, work history information may come from others who have previously worked with the subordinate (Hunt, 1995). These impressions gained from others have been shown to affect current judgments of others (Brierley & Gwilliam, 2001; Stolt, 1985).

6 The performance-related cues were based on a review of auditing textbooks and prior research on errors made by auditors (Burgstahler & Jiambalvo, 1986 and Bell & Wright, 1997). Also, to offer an unbiased test of memory for confirming and disconfirming behaviors, there is an equal number of confirming and disconfirming behaviors, all of which are reviewed by the evaluators. Hastie and Kumar (1979) found that an equal set size significantly reduced the advantage of incongruent behaviors over congruent behaviors.

7 The distracter task was an unrelated performance evaluation task with an entirely different scenario, designed to test for the effects of audit outcomes upon the rating of an auditor. The results are outside the scope of the current experiment and are reported in a separate paper.

8 This measure is similar to that used in Cloyd (1997). The Cloyd study was concerned with effectiveness in a tax research task and measured search effectiveness as the difference between the number of relevant and irrelevant tax citations used to support subjects’ tax positions. The measure used in this research allows for analysis both in terms of a confirmatory bias or negativity bias, the two potential biases of primary interest in this study.

9 The rating was added because of the performance evaluation context, but was not of interest to the current study, which focuses on recall of specific behaviors.

10 There were no significant differences between the two administrations in the number of strengths or weaknesses recalled (p > .6 in both cases), or the primary dependent variable of interest, i.e. the difference between the number of strengths and weaknesses recalled (p > .3). Therefore, neither the firm nor the difference in timeframe affected the primary responses of interest.

11 Neither months of audit experience (p > .8) nor the number of performance evaluations prepared (p > .6) are related to the initial impression treatment. Therefore, experience is not significantly different across treatments with respect to either of the two experience measures obtained.

12 Gender is not significantly different across initial impression treatments (Pearson χ² = .1.691, p > .4, two-tailed; d.f. = 2). In addition, gender was not significantly associated with the primary measure of recall (p > .7). Therefore, gender is excluded from subsequent analyses.