Academic journal article Canadian Journal of Behavioural Science

Response Latency Detection of Fakers on Personnel Tests

Academic journal article Canadian Journal of Behavioural Science

Response Latency Detection of Fakers on Personnel Tests

Article excerpt

Abstract

The ability of personnel test item response latencies to differentiate between individuals instructed to fake and those instructed to respond honestly was examined. Based on a general model of lying derived from schema theory, it was predicted that fakers should take relatively longer than honest respondents to admit to delinquent characteristics concerning themselves. Discriminant function analysis indicated that response latencies to items on standard personnel tests could significantly distinguish between fakers and honest test respondents in a personnel testing scenario. Results generalized across samples of students and unemployed job seekers and were interpreted as supporting a general model of lying. Implications are discussed.

Employee delinquency results in serious and costly organizational consequences. Consider these examples. Employee theft is the largest single source of loss attributable to crimes against business (American Management Association, 1977). Employee violence can disrupt the work flow in organizations, potentially endangering other employees, customers, and the public (Ones, Viswesvaran, & Schmidt, 1992). Worker drug use is empirically related to absenteeism and turnover (Normand, Salyards, & Mahoney, 1990), accidents, medical insurance costs, workers' compensation claims, as well as behaviours such as leaving work early, taking long lunch breaks, sleeping on the job, and taking work supplies home (Lehman & Simpson, 1992).

A potential solution to the problem of delinquent employees is to screen out at the pre-employment stage those job applicants likely to negatively affect organizational performance (Hogan & Hogan, 1989). Integrity tests have become a mechanism for attempting to achieve such a goal. Notwithstanding debates concerning the physiological basis, validity, and utility of the polygraph (Bashore & Rapp, 1993), the passage of legislation restricting the use of polygraphs for pre-employment screening has resulted in proliferating interest among employers in the use of self-report measures of integrity and personality (Murphy, 1993). For example, O'Bannon, Goldinger, and Appleby (1989) indicate that, each year in the United States, more than 2.5 million integrity tests are given now by more than 5,000 employers.

Although measures of integrity and personality can be valid in the hiring process (Hogan & Hogan, 1989; Hough, Eaton, Dunnette, Kamp, & McCloy, 1990) their apparent susceptibility to faking still evokes a great deal of skepticism regarding their usefulness (Ones, Viswesvaran, & Schmidt, 1993). In particular, the obvious content of many of these tests raises concerns as to how valid such tests can be under conditions where job applicants may be lying. An important, related issue is how to detect job applicants who are faking to a potential new employer about negative proclivities (McDaniel & Timm, 1990). A general index capable of identifying individual job applicants who are faking on a personnel test would be an asset for organizations using self-report measures in the hiring process. The present research focusses on the establishment of such an index for this identification of faking job candidates. In particular, the current investigations examine and confirm the potential of test item response times for indicating individual test respondents who are faking on a personnel test.

Research on the use of response latencies for detecting faking on psychological tests is to be noted for producing inconsistent and, sometimes, seemingly contradictory findings. For example, Hsu, Santelli, and Hsu (1989) have reported that faking results in faster test item latencies whereas, McDaniel and Timm (1990) have found that faking is associated with slower responding. Such discrepant results may be because latency data can be extraordinarily noisy. The raw response latency for a particular subject on a specific item has a multitude of factors influencing its value. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed

Oops!

An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.