Lying person-to-person about life events: a cognitive framework for lie detection.

Undetected lies of prospective or current employees cost business billions of dollars annually. The ability to detect these lies would be of immense benefit. Several recent reports have called for research on new, theoretically based methods of lie detection. In response, we tested the Activation-Decision-Construction Model of lying (Walczyk, Roper, Seeman, & Humphreys, 2003) according to which response time is a cue to deception. Participants were tested person-to-person. In Experiment 1, half lied to questions probing recent episodic memories. The other half answered honestly. Liar-truth teller response time differences were observed between subjects. Discriminant analyses demonstrated the value of response time for uncovering deceit. Those highest in social skills were the quickest liars. In Experiment 2, lying was shown to take longer than truth telling within subjects, and within-subject response time standard deviations were shown to be converging cues to deception. Based on these data and the ADCM, a Time-Restricted Integrity Confirmation (Tri-Con) framework for lie detection is proposed that might one day provide cost effective lie detection for business.

**********

The Costs to Business of Undetected Lies

The costs of applicant dishonesty and employee misconduct are difficult to calculate (Robinson & Bennet, 1995). Estimates range from $6 billion to as much as $200 billion annually in the United States (Berry & Lilly, 2003; Lipman & McGraw, 1988). A May 2001 study conducted by Avert, Inc., a company specializing in background checks, found that 24% of the 1.8 million job applicants whose credentials were checked misrepresented their qualifications. The most common inaccuracies were falsifications concerning employment experience, job titles, or education (Ollian, 2003). Moreover, the United States Chamber of Commerce has estimated that approximately one-third of business failures each year can be traced to employee theft and other personnel crimes. The cost of employee theft has been estimated to be slightly more than $15 billion annually (Hollinger & Davis, 2002).

Given these staggering costs, exposing applicant dishonesty and employee misconduct is crucial. Historically, organizations have used paper-and-pencil integrity tests and the polygraph. Even so, the polygraph, the most common method of lie detection, has been and remains controversial (Iacono & Lykken, 1997). For instance, the Employee Polygraph Protection Act of 1988 (PL 100-347, 102 Stat 646) states that employers "engaged in production for commerce cannot ... have an employee or prospective employee submit to a polygraph test" as a condition of employment. In recent years experts have argued that the polygraph is not theoretically based, is poorly standardized, is susceptible to countermeasures, and is subjectively scored (CRSER 2003; DePaulo et al., 2003; Lykken, 1998). An alternative method of lie detection that could answer these criticisms would be useful to personnel departments. The remainder of this article proposes a theory and provides some empirical support for a method of lie detection based on response time.

Response Time and Deception

Faking when applying for a job. Holden, Kroner, Fekken, and Popham (1992) proposed a cognitive model of what impression managers (fakers) do when responding falsely to items on a personality inventory such as when trying to secure a job for which they are not suited. According to their model, honest respondents compare individual items to self-schemata. For instance, applicants reporting how strongly they agree with a personality test item such as "My workplace is always clean and orderly" (p. 158, Schmitt & Chan, 1998) who possess well-integrated schemata of their conscientiousness will respond quickly. Applicants possessing well-integrated conscientiousness schemata who are doubtful that the item describes them will be delayed in responding as they consider numerous paths in memory. Honest applicants for whom conscientiousness schemata do not exist, perhaps because they have never reflected on the issue, will have to consider a disorganized collection of data in memory before arriving at a decision, a slow process.