Detecting deception in written statements

Abstract

PurposeThe purpose of this paper is to review the most recently published and most comprehensively designed study of scientific content analysis (SCAN), a widely used but rarely researched method of content analysis for detecting deception in written statements.Design/methodology/approachThe author reviewed the design, data, and findings of the study and performed statistical re‐analysis of the raw data. Prior citations and critiques of the study by scholars also were reviewed.FindingsThe design and data analysis of the British Home Office study are in part deficient. The design includes features of a quasi‐experimental study which were operationalized poorly, and the data aggregation and analysis produced an incomplete and problematic interpretation of the raw data. Prior reviewers of the Home Office study erred in part in their understanding of the study's findings and deficiencies.Research limitations/implicationsThe research was limited to the raw data of the study, but even so, the results justify additional research on SCAN. In particular, studies should be designed that control for variables such as length of law enforcement career, educational level, number of written statements taken during the career, measures of verbal intelligence, and pre‐SCAN‐training ability to detect deception in statements. The design and analysis flaws of the Home Office study which are identified in the paper would also need to be avoided.Originality/valueThis is the first published paper to identify the full range of design and data analysis deficiencies of the Home Office study and to argue that its data nonetheless support the recommendation that SCAN be researched more thoroughly.