The Impact of Generating Initial Hypothesis Sets of Different Sizes on the Quality of the Initial Set, and the Resulting Time Efficiency and Final Judgment Accuracy

This study examines the impact of generating initial hypothesis sets of different sizes on the quality of the hypotheses generated (i.e., the ability to consider both the direction and accounts that are over‐ or understated). We also examine the time efficiency, information search effectiveness, and the final judgment accuracy, conditional on the quality of the initial hypothesis set. Sixty auditors performed an analytical procedures task where they were asked to generate and test either a specific number of initial hypotheses (one, three, or six), or any number of hypotheses desired in order to uncover an error seeded in the financial statements. The results indicate that the three hypotheses group initially generated hypotheses of the highest quality and maintained the hypothesis quality after efficiently searching information and generating additional causes. The one hypothesis group improved the quality of their hypotheses only after generating and testing several causes. However, auditors who generated six hypotheses or any number desired (as in audit practice) considered hypotheses of lower quality in the initial set, and did not improve the hypotheses quality after going through the information search stage. These results suggest that the size of the initial hypothesis set can lead to differences in the gains that accrue from the hypothesis generation and information search stages of diagnostic decisions.