Detecting Framing Effects in Financial Statements*

Abstract. In this study, we address the question of what kinds of cognitive representations auditors use in a situation of potential financial statement fraud. We divide the problem of detecting fraud into two parts: detecting the frame management has constructed to mask the fraud, and detecting the fraud. We examine two ways proposed by Kahneman and Tversky (1986) for detecting a frame: (1) use of multiple representations that provide alternative interpretations of data in the financial statements; and (2) use of a procedure that transforms financial statement data into a standard representation. Twenty‐four audit partners served as participants in the study. Each partner conducted a simulation of a concurring partner review. All auditors reviewed four cases in which management had created a misleading description of the company (a frame) and a financial statement fraud. The results support Kahneman and Tversky's proposal that frames can be detected by transforming a problem into a standard representation. Auditors who used a standard representation successfully detected management's frame, aggregated the items, and detected fraud in all four cases. Auditors who used a standard representation followed a procedure specified by generally accepted auditing standards (Canadian Institute of Chartered Accountants 1988, American Institute of Certified Public Accountants 1984) for aggregating items. Auditors who used multiple representations detected management's frame on all four cases. These auditors, however, did not use the aggregation procedure specified by auditing standards and failed to detect the fraud on all four cases.