The relation between consensus and accuracy in low-to-moderate accuracy tasks: An auditing example

This study examines the relation between consensus and accuracy using an er ror frequency estimation task for which auditors' overall accuracy is known to be low to moderate. We also investigate whether experience moderates th e relation between consensus and accuracy for the three industries examined : manufacturing, natural resources, and banking. We find that accuracy is p ositively related to consensus for ail auditors in manufacturing and for au ditors with more than 12 (36) months of experience in natural resources (ba nking). For banking and natural resources, we provide evidence that auditor s with little experience in these industries use a heuristic consistent wit h manufacturing error frequencies as an "educated guess" for the specialize d industries' error frequencies. This heuristic leads to consensus among au ditors, but results in low accuracy. The results are important for auditing practice and research since reliance on high consensus as a surrogate for accuracy may prove inefficient or, worse, ineffective. The results also dem onstrate the need for further investigation of the determinants of audit kn owledge and performance across multiple industries and tasks.