We examine the interaction between internal control assessments and substantive testing in a model of fraud detection. The purpose of our study is to examine a two‐stage model of the auditor‐manager interaction in which the auditor assesses the “likelihood” or possibility of fraud in the first stage and conducts substantive tests in the second stage. We examine the allocation of audit resources across these two distinct facets of the audit. We find that, regardless of the auditor's allocation, the probability of undetected fraud remains the same, but the allocation of some audit resources to internal control assessment may provide cost savings for the auditor.

This paper considers the interpretation of evidence from social experiments when persons randomized out of a program being evaluated have good substitutes for it, and when persons randomized into a program drop out to pursue better alternatives. Using data from an experimental evaluation of a classroom training program, we document the empirical importance of control group substitution and treatment group dropping out. Evidence that one program is ineffective relative to close substitutes is not evidence that the type of service provided by all of the programs is ineffective, although that is the way experimental evidence is often interpreted.