Overt integrity tests versus personality-based measures of delinquency: An empirical comparison

JOURNAL OF BUSINESS AND PSYCHOLOGY Volume 3, No. 3, Spring 1989

OVERT INTEGRITY TESTS VERSUS P E R S O N A L I T Y - B A S E D M E A S U R E S OF DELINQUENCY: AN EMPIRICAL COMPARISON Alan G. Frost Middle Tennessee State University

Fred M. Rafilson Illinois Institute of Technology

ABSTRACT: This study examined the ability of both an overt integrity test and a personality-based measure of delinquency to predict on-the-job theft and counterproductivity. A heterogeneous sample of 105 employees anonymously completed both tests and a criterion checklist that included a 9-item theft subscale and a 12-item counterproductivity subscale. Results indicated that the overt integrity test was a superior predictor of both on-the-job theft and drug abuse compared to the personality-based measure. The overt integrity test was also a slightly better predictor of general counterproductivity. These findings and their implications are discussed. In 1988, Sackett, Burris, and Caltahan identified the emergence of two distinct types of employee integrity tests t h a t are used for personnel selection. They labeled these tests "overt integrity tests" and "personality-based measures." This classification scheme closely parallels Cronbach's distinction between clear purpose and disguised purpose inventories (Cronbach, 1970). Overt integrity tests inquire directly about job applicants' attitudes toward theft. They were primarily developed to control both the prevalence and magnitude of employee theft and similar forms of dishonesty in the workplace (cf., Terris, 1985). On the other hand, personality-based tests were primarily designed to measure poor impulse control, lack of conscientiousness, and general delinquency in non-employee populations (cf., Gough, 1971). Sackett et al. (1988) suggested t h a t when a company is interested

Address all correspondence to Dr. Alan Frost, Psychology Department, Middle Tennessee State University, P.O. Box 507, Murfreesboro, TN 37132. 269

9 1989 Human ScmncesPress

270

JOURNAL OF BUSINESSAND PSYCHOLOGY

in predicting theft criteria, overt integrity tests appear to be more appropriate. Unfortunately, these authors also point out that " . . . the lack of head-to-head comparisons of tests prevents us from drawing strong conclusions about the relative merits of these different types of tests" (p. 6). While published research indicates that overt integrity tests do indeed predict employee theft (cf., McDaniel & Jones, 1988), no published study is available that shows a significant relationship between personality-based measures of delinquency and on-the-job theft criteria. Sackett et al. (1988) also suggest that the personality-based measures should be the most predictive o f " . . , an aggregate of counterproductive behaviors, or overall job performance" (pp. 5-6). However, there has been no research that tests this hypothesis. Sackett et al. (1988) do warn, however, that arguments like "personality-oriented measures must be superior because they are broader in coverage" may not be true for two reasons. Fi~rst, no empirical research proves that broad test content translates into superior prediction of job behaviors of interest. Second, some publishers of overt integrity tests use additional subscales in order to predict a wider range of counterproductive behaviors. These expanded test batteries include, but are not limited to, scales that assess the potential for on-the-job drug abuse, violence, and accident potential (cf., Terris, 1986). Common personality-based measures include the Personnel Reaction Blank (Gough, 1971), the PDI Employment Inventory (Paajanen, 1986), the reliability scale of the Hogan Personnel Selection Series (Hogan & Hogan, 1986), the responsibility scale of the California Psychological Inventory (Gough, 1986), and the Employee Reliability Inventory (Bay State Psychological Associates, 1986). Sackett et al. (1988) suggested that the conceptual underpinnings of all of these personalitybased tests are very similar, leading one to speculate that high correlations between the tests would be found if such research were conducted. In fact, one study showed that the PDI Employment Inventory correlated highly with the Personnel Reaction Blank (r = .56; Paajanen, 1986). This correlation supports the notion that personality-based measures are very similar. Ash (1987, 1988) reviewed the literature available on a number of personality-based measures, including the Personnel Reaction Blank and the PDI Employment Inventory. Ash pointed out that the Personnel Reaction Blank appears to be the model for the PDI Employment Inventory. Ash indicated that the use of personality-type questionnaires for the assessment of criminality is not new, and that a large number of personality tests are keyed for delinquency. However, the article indicates that the use of personality tests to assess criminality has yielded inconsistent results, especially in the employment context. Most of the non-polygraph based research on overt integrity tests

FRED M. RAFILSONAND ALAN G. FROST

271

has been conducted with the Personnel Selection Inventory or PSI (cf., Sackett & Harris, 1984). Terris (1979a) found that the PSI honesty scale accurately predicted theft admissions made in pre-employment polygraph examinations. Yet the validity of this scale is not limited to polygraph admissions. The honesty scale has also been found to predict the following criteria: (1) supervisors' ratings of employees' dishonesty (Jones & Terris, 1983), (2) applicants who are likely to get caught stealing once hired (Jones & Terris, 1981), (3) applicants who have a criminal history of armed robbery and similar offenses (Jones & Terris, 1985), and (4) applicants who are likely to make theft admissions in an anonymous testing situation (Terris, 1979b; Jones, 1980, 1981). Joy (1988) indicates that over 65 studies have been conducted with the PSI honesty scale. McDaniel and Jones (1988) summarized 23 studies in a meta-analysis and found that the average validity coefficient of the PSI honesty scale equals .50. The purpose of this study was to compare a leading overt integrity test to a leading personality-based measure of delinquency. The ability of these two tests to predict both theft and general counterproductivity was explored. It was specifically hypothesized that the overt integrity test would better predict theft criteria than the personality-based measure. This hypothesis was based on two factors. First, overt integrity tests were primarily designed to predict theft, reduce employee pilferage and hence control company shrinkage rates. Personality-based measures were originally developed to predict delinquency in the general adult population. Second, moderately high validity coefficients have been obtained between overt integrity tests and various theft criteria. No published studies exist that show that personality-based measures can predict theft. No hypothesis was forwarded as to which type of test would better predict general counterproductivity. Finally, the interrelationship between the two tests was examined.

METHOD

Employee Sample The heterogeneous sample consisted of 105 employees. The subjects were employed in a wide variety of business occupations at the time of the study. Types of industries represented included: manufacturing (11.4%), retail sales (21.9%), government (5.7%), restaurant (11.4%), construction (2.9%), health care (5.7%), and miscellaneous (41%). They averaged 30.7 hours (S.D. = 15) of employment per week. Their average age was 24.1 years (S.D. = 8). Approximately 52% of the sample were male, and 48% were female.

272

JOURNAL OF BUSINESSAND PSYCHOLOGY

Integrity Measures Overt Integrity Test. All employees completed the honesty scale from the Personnel Selection Inventory (PSI; London House, 1980). The PSI honesty scale has a Spearman-Brown split-half reliability equal to .95 (Terris, 1979a), and an average validity coefficient against various theft criteria equal to .50 (McDaniel & Jones, 1988). The PSI honesty scale was designed to measure job applicants' attitudes toward, perceptions of, and opinions about theft in the workplace. The honesty scale measures the following psychological constructs: (1) tolerance of others who steal, (2) estimates about the extent of theft by others, (3) acceptance of common rationalizations for theft, (4) ruminations about theft, and (5) admissions of theft-related activities. Higher scores mean more integrity. Personality-based Measure. The Personnel Reaction Blank (PRB; Gough, 1971, 1972) was one of the original personality-based measures of delinquency. Other personality-based measures seem to be patterned after the PRB (e.g., the PDI Employment Inventory). The PRB purports to measure dependability and conscientiousness. The PRB effectively discriminated between adult samples of delinquents and non-delinquents (Gough, 1971). Moreover, internal consistency coefficients have ranged from .65 to .95 depending on the type of sample studied (Gough, 1972). Higher scores m e a n more dependability.

Theft and Counterproductivity Criteria Employees anonymously completed a theft criterion checklist that included two subscales: (1) a 9-item theft subscale and (2) a 12-item counterproductivity subscale. The alpha reliability coefficients for both the theft subscale and the counterproductivity subscale equaled .82 and .72, respectively. Types of behavior assessed with the theft subscale include: (1) theft of money, property and merchandise from your employer; (2) overcharging and shortchanging customers for personal gain; (3) changing company records to get paid for work not actually done; (4) actively helping other employees steal from the company; and (5) selling merchandise to friends and relatives at a reduced price. Types of behavior assessed with the counterproductivity subscale include: (1) damaging company merchandise, property, or equipment while "horsing around" on the job; (2) engaging in alcohol and drug abuse during paid work hours; (3) being overly aggressive with customers, co-workers, and supervisors; (4) excessive tardiness and absenteeism; (5) faking an illness and calling in sick; and (6) doing slow and sloppy work on purpose. A 5-point Likert-type scale was used for all items. Employees used the following scale to rate how often during the last six (6) months they engaged in any of the relevant behaviors: 1 = very often (3 or more times

FRED M. RAFILSONAND ALAN G. FROST

273

per week); 2 = Often (1-2 times per week); 3 = Occasionally (1-2 times in one month); 4 = Seldom (Only once in 6 months); and 5 = Never (Not even once). Hence, hi ghe r scores on the criterion checklist means less counterproductivity.

Procedure The employees anonymously completed a questionnaire packet t h a t included the PSI honesty scale, the PRB, and the criterion checklist. The order of administering the PSI and the PRB was counterbalanced across subjects. The criterion checklist was always completed last. Subjects provided demographic data, yet t hey did not put their names on the questionnaire booklet. T hey anonymously completed the booklets on th eir own time. When t hey r e t u r n e d the booklets t hey put t hem in a box with other booklets in order to ensure anonymity. All data were statistically analyzed using the SPSS-PC software package for the behavioral sciences (Norusis, 1986).

RESULTS AND DISCUSSION The means, standard deviations, and intercorrelation m at ri x for all measures are summarized in Table 1. Inspection of this table reveals t h a t the PSI honesty scale and the PRB appear to be measuring differ-

Table 1 Descriptive a n d Inferential Statistics (N = 105)

Variables:

Mean

S.D.

2.

3.

4.

5.

1. PSI

33.70

13.10

0.25