Identifying faking on self-report personality inventories: Relative merits of traditional lie scales, new lie scales, response patterns, and response times

Abstract

The use of personality tests throughout Canadian society is based on the assumption that their results are valid. However, research has shown that individuals can, and do, fake their responses on personality inventories. Individuals may fake good, emphasizing their positive characteristics, or fake bad, emphasizing negative characteristics, in order to obtain a desired outcome. Recent research has provided support for a congruence model of faking, which states that schema-consistent responses are provided more quickly than schema-inconsistent responses. Faking successfully, without being detected by validity indices, requires balancing favourable and unfavourable responses, regardless of the faking schema a participant adopts. This demand results in cognitive fatigue over time, producing increasingly unbalanced response patterns. Two studies were conducted to evaluate the efficacy of the congruence and cognitive overload models of faking in detecting instructed faking, and to examine whether these models or the newly developed Faking Response Strategy Scales provide added value in detecting faking relative to currently established gold-standard measures. Results showed that all of the self-report scales examined—whether traditional or new—were valid detectors of faking, which supports their ongoing use. However, results highlighted the weakness of the Impression Management subscale of the Balanced Inventory of Desirable Responding, the current gold-standard in the field, in providing added value relative to other scales. Response latency data supported the congruence model of faking, but results for the cognitive overload model were mixed: Study 1 data supported the cognitive overload model, but time constraints introduced in Study 2 seem to have caused random responding, rather than increasing cognitive overload as was intended. Results supported a multidimensional model of faking, and show that adding measures of response latency and response pattern can enhance the ability of traditional measures to detect faking. These findings have important theoretical and practical implications for methods of detecting faking and for the understanding of cognitive processes underlying faking.