Scientific Content Analysis and Deception Detection: A Systematic Review of the Empirical Literature

Background: Scientific Content Analysis (SCAN) is a verbal credibility assessment method used by law enforcement agencies worldwide to identify deception in written statements. Despite its widespread practitioner adoption, the empirical validation of SCAN remains contested. This systematic review synthesizes the peer-reviewed literature on SCAN and related content-based deception detection methods identified through a PRISMA-guided search of Scopus, Web of Science, and Google Scholar. Methods: A three-database search retrieved 591 Google Scholar results, 12 Scopus records, and 13 Web of Science records. After removing duplicates, non-peer-reviewed sources, books, theses, reviews, and irrelevant items, 37 empirical and applied studies were retained for analysis. Literature reviews and meta-analyses were used exclusively for theoretical framing. Results: The empirical record consistently shows that SCAN in its classical form performs at or near chance level for distinguishing truthful from deceptive statements. Inter-rater reliability is low except for pronoun use. Practitioners nonetheless report high confidence in SCAN. Adjacent methods, including Criteria-Based Content Analysis (CBCA), Reality Monitoring (RM), the Verifiability Approach (VA), and automated Natural Language Processing (NLP) tools, show more robust validity, though effect sizes remain modest. Cross-linguistic and cross-cultural applications reveal that several core verbal deception indicators are partially transferable across languages, with necessary contextual adaptation. Conclusions: SCAN should not be used as a standalone veracity assessment tool in criminal proceedings. However, specific criteria, notably pronoun use, denial structure, and missing information, retain investigative utility when used as interview preparation tools rather than as lie detectors. Future research should prioritise standardised rubrics, cross-validation procedures, and real-crime corpora.