Verbal lie-detection using the reality monitoring approach: an analysis of its effectiveness and moderating factors

A large body of research findings suggests that verbal cues to deception can boost deception-detection accuracy rates to levels significantly above chance. This thesis examines the effectiveness of, and influences on, one of the most popular and widely used verbal lie-detection approaches, Reality Monitoring (RM). The RM approach has advantages not only in terms of its underpinnings in memory research and theory, but also its ease with regard to practical application. The RM approach assumes that deceptive verbal accounts, because they are artificial and not based directly on actual experience, differ from truthful accounts according to a variety of criteria (truthful accounts contain more vivid, spatial, temporal and affective information, etc.). However, so far, as in many other areas of lie-detection research, research in the area of RM has lacked methodological standardisation; consequently, we know little of the potential effects of contextual and other moderating variables on RM measures. In view of this, the primary aims of the present thesis were a) to assess whether, if conveyed in a standard format, the RM approach has any value overall in distinguishing between truthful and deceptive accounts, and b) if it does, to investigate the circumstances under which it might give optimal results; i.e. to assess some of the main factors which may moderate its efficacy in this respect. To these ends, six experimental studies were conducted which looked at truthful and deceptive accounts (generated by participants in the laboratory using video and autobiographical sources) and considered the effects of a number of different moderators; these included, first and second-language effects, modality (i.e. written vs. spoken discourse), absence/presence of others, demand characteristics effects, scoring systems (rating scales vs. raw frequencies) and standardisation for account length. Overall, results indicated that in most studies there was evidence that total RM scores, as determined by the procedures applied here, successfully discriminated between truthful and deceptive accounts. However, results varied when RM criteria were considered individually, and when the influence of various moderators was assessed. For example, frequency measures of spatial and temporal information were found to be two of the most consistently effective diagnostic RM criteria. However, overall, RM was a more effective diagnostic tool before accounts were standardised for length; indeed, total RM scores failed to distinguish between truthful and deceptive accounts after accounts had been standardised for length. Also, the presence of others and modality (written or spoken) were two key moderator variables whose impact on total RM scores varied depending on whether or not the accounts were standardised for length. A number of other related variables were also considered; for example, truthful accounts were longer than deceptive accounts in both duration and length and the number of words produced per second was significantly greater for truthful accounts. Implications for research and practise are discussed; though perhaps most important in this respect was the finding that, despite the overall success of RM in discriminating truthful from deceptive accounts, RM criteria were not generally discriminating after standardisation for word-count or length. Moreover, a number of the moderators affected RM scores regardless of whether they were derived from truthful or deceptive accounts. This suggests that we may still be a long way from developing a method (such as the use of normative criteria) that could be used in the field to classify individual cases. Nevertheless, in the meantime, at the very least, the present results suggest that when judging the veracity of accounts using RM criteria, the scoring and other moderating variables identified in this thesis should be investigated systematically, and measured and applied consistently, if researchers wish to compare and replicate findings within and across studies