We report the results of a study involving 142 tax professionals designed to investigate the effects of decision aid design on information search (i.e., tax research) and confirmation bias. Results indicate that the participants exhibited confirmation bias when conducting tax research for clients. That is, participants showed a tendency to preferentially select information in support of their earlier recommendations to the client, even when the recommendation disagreed with the client's subsequent tax position. Results also indicate that while some decision aid features can reduce confirmation bias during tax research, others do not and may even enhance this bias. Specifically, a justification requirement decision aid reduced confirmation bias in terms of both the number and perceived importance of selected confirmatory cases, whereas a factor evaluation checklist decision aid either increased the bias (i.e., increased the perceived importance of cases) or had no effect on the bias (i.e., no effect on the number of cases). We suggest several decision aid design features for reducing confirmation bias in tax research.
