Factors influencing working backwards among auditors in the application of nonstatistical decision aids.

Abstract

For more than a decade, research has demonstrated decision makers' unwillingness to rely on decision aids. However, extant research has most often studied the use of unfamiliar decision aids. For example, Kachelmeier and Messier (1990) and Messier, Kachelmeier, and Jensen (2001) provide evidence of auditors working backwards in the use of an unfamiliar decision aid. The current study builds on extant research by using process tracing software in an experimental setting to test whether auditors have learned from a decision aid used in practice and how learning affects decision aid use in relation to the use of an unfamiliar decision aid. Auditors used either a familiar decision aid, a modified familiar decision aid, an unfamiliar decision aid or a modified unfamiliar decision aid to compute a nonstatistical sample size. Results indicate that decision aid familiarity does not affect the auditors decision to work backwards, approximately 30 percent of auditors in all conditions worked backwards to alter the decision aid sample size. Further, working backwards has the affect of eliminating the significant, negative relationship that should exist between the number of controls tested and the substantive sample size. It also appears that auditors acquire task specific knowledge from general audit experience rather than task specific experience with the decision aid.