Post-decision inference editing of supportive and counterindicative signals among external auditors in a going concern judgment

There is evidence that the extent of common judgment biases – e.g., hindsight, anchoring and adjustment, representativeness, and confirmatory strategies – often is reduced when experts perform job-related tasks as compared to students performing generic tasks or professional tasks for which they lack experience: that specialized judgment heuristics may be used for tasks within an expert's domain, whereas general cognitive heuristics may be used in situations where the individual lacks information and/or expertise. Our focus here is on post-decision inference restructuring/editing, the unconscious revision of what one saw and thought in the pre-decision stage as a result of outcome feedback/lack of feedback. We found that professional auditors, in retrospect, were likely to edit crucial decision making signals, but that the extent of the post-decision editing is a function of the task and the presence/absence of feedback. Professional auditors were provided with relevant financial and non-financial information (based on an actual company), and asked to isolate the most prominent signals (positive and negative) concerning the company's viability. The participants were then asked to give the company either an unmodified (supporting the company's viability) or a going concern (raising questions about the company's survival) audit report, and to rate the importance of key aspects with respect to the chosen alternative. After an intermission, the participants in the experiment group were given information about the actual outcome, whereas the participants in the control group received no feedback. The participants were than asked to replicate their aspect ratings, and to assess the probability that the company actually received a going concern modification/an unmodified audit report. The results indicate that outcome feedback/lack of feedback had an effect on post-decision restructuring of signals. Lack of feedback among professional participants seems to invite greater post-decision adjustments than access to feedback, in particular among participants who chose a going concern report. This relatively less common, more consequential choice, invited drastic post-decision editing of the second and third most supportive aspects, as well as of the three counterindicative aspects. We believe that this kind of post-decision inference editing has less to do with traditional hindsight than with a strengthening predisposition with roots in the pre-decision phase.