Advances in Accounting Behavioral Research

Researchers are continually trying to find reliable fraud indicators (e.g., Beasley, 1996) and some are working on building fraud prediction models (e.g., Spathis, 2002) to aid auditors in fraud detection. With this same goal of predicting fraud in mind, the purpose of this study is to explore the potential of qualitative fraud risk indicators. Content analysis is used in analyzing the Management's Discussion and Analysis (MDA) section of the annual report to identify potential indicators of deception to increase the likelihood of fraud detection in a timelier manner than current quantitative models.

By examining asynchronous communication contained in annual reports for companies required by the SEC to restate their financial statements, patterns of key linguistic characteristics were identified and compared to those used by companies not required to restate. Findings evidence significant differences on several dimensions. Using language cues for detection of deception has the advantage over quantitative methods of providing a more timely method of determining deception. Quantitative models often cannot detect deception until the effects are validated by financial impairment.

Implications of the findings suggest that qualitative methods of deception detection may provide an earlier, and thus more useful, method of the detection of fraud. The results of this study should provide stakeholders with a set of indicators to aid in identifying misstated information. This approach is also one that can be generalized to other written documents used to predict fraudulent communication.