Impact of warning messages on the reliance level on decision aids under the framing effect

Abstract

Manuscript type: Research paper Research aims: This study empirically examines the role of warning messages on the reliance level on decision aids. We examine whether the offer of warning messages can lower the effect of framing so that it can influence the reliance level on decision aids. Design/Methodology/Approach: This study employs a 2 x 3 experimental design involving 65 undergraduates and 13 Accounting Profession Program professionals as samples. The dependent variable is the decision aid reliance which deploys a scenario that previously belonged to Gomaa, Hunton, Vaassen and Carree (2011). The framing effect is then manipulated as: (1) positive framing, and (2) negative framing while the warning messages are manipulated as (1) no warning, (2) weak warning, and (3) strong warning. Research findings: This study shows that the warning messages reduce the framing effect on decision aids. In particular, the findings show that in an audit context, both the weak and strong warning messages are successful in decreasing the framing effect bias on decision aids. These results suggest that warning messages can assist auditors in getting the best decisions in an audit process. Theoretical contribution/Originality: To the best of our knowledge, previous studies had only examined the debiasing effect of framing within a business context. In contrast, this study may be the first of its kind to test this debiasing technique within the auditing context. Practitioner/Policy implication: Warning messages can function as one of the best tools for auditors to produce the best decisions within an auditing process. Managers in public accountant firms can use the warning messages as a way to maintain their auditors or to employ the audit decision aids carefully. Research limitation/Implications: This study only measures the reliance level, and not the overreliance level because using a specific measure of overreliance can produce different results.