The moderating effect of problem characteristics on the expertise-decision-making link.

Abstract

The goal of this research was to demonstrate that there are benefits to expertise when solving complex problems that have ambiguous characteristics. This research effort was motivated by the expertise-performance paradox (Camerer and Johnson 1991), a counter-intuitive finding that despite greater domain-specific knowledge of experts relative to novices, they are little if at all better in terms of the quality of decision outputs. However, much of the previous research required integrating a small number of clearly defined inputs to reach a pre-specified objective. In such environments--which are hardly representative of real-world managerial problems--there may be insufficient latitude for experts to demonstrate their superior problem-solving ability. To achieve our objective, we explored how providing or withholding a decision aid (which helps to identify inputs) and varying the level of noise inherent in externally available information (which affects the ability to evaluate inputs) moderates the expertise-decision making link. Eight hypotheses were proposed which were tested by running a 2 x 2 x 2 between-subjects experiment (expert/novice by decision aid/no aid by low versus high input noise). Relative to novices, we found that experts were better at selecting diagnostic inputs and evaluating non-quantified inputs. As a result, they made more accurate and tightly clustered decisions. They were also more confident in their decisions, but their confidence level exceeded what was justified by their greater accuracy. The benefits of expertise were more pronounced when solving less structured problems. Without the aid, novices' decisions degraded considerably: there was greater variability in their judgments; they were more prone to extreme errors; their mean error increased; and the calibration of their judgments decreased. Experts, in contrast, were less frequently affected by the aid manipulation. We therefore conclude that experts outperform novices when solving complex problems with ambiguous characteristics because they can impose a more meaningful structure onto such problems which reduces a greater portion of the problem's uncertainty. These findings provide important evidence regarding expert-novice performance differences by advancing our knowledge of when experts outperform novices and why they are able to do so.