A review of criterion-related validities of personality constructs indicated that six constructs are useful predictors of important job-related criteria. An inventory was developed to measure the 6 constructs. In addition, 4 response validity scales were developed to measure accuracy of self-description. These scales were administered in three contexts: a concurrent criterion-related validity study, a faking experiment, and an applicant setting. Sample sizes were 9,188,245, and 125, respectively. Results showed that (a) validities were in the .20s (uncorrected for unreliability or restriction in range) against targeted criterion constructs, (b) respondents successfully distorted their self-descriptions when instructed to do so, (c) response validity scales were responsive to different types of distortion, (d) applicants' responses did not reflect evidence of distortion, and (e) validities remained stable regardless of possible distortion by respondents in either unusually positive or negative directions. Recent reviews of criterion-related validity studies show that personality scales, when organized according to predictor construct, correlate significantly with a

The standard deviation of performance quality measured in dollars, SD$, is critical to calculating the utility of personnel decisions. A popular technique for obtaining SD$ calls for supervisor estimates of the dollar value of performance at different levels. In many cases supervisors can base their estimates on the cost of contracting out the various levels of performance. Estimation problems can arise, however, in contexts where contracting out is not possible, such as in government organizations without private industry counterparts. Estimation problems may also exist where individual salary is only a small percentage of the value of the performance to the organization or of the equipment operated. This paper presents two strategies for estimating the value of performance and for determining SD$ by considering the changes in the numbers and performance levels of system units which lead to increased aggregate performance. One hundred U.S. Army tank commanders provided data about their jobs for these two strategies as well as for the supervisor estimation and salary percentage strategies. The new strategies appear to provide more appropriate and acceptable values of SD$ for those complex, expensive systems where dollar values of performance are less easily estimated.

No abstract

This research explored possible between-assessor differences in the validity of assessment center ratings Thirteen Army recruiters were trained as assessors, they then evaluated 450 soldiers on four different assessment exercises These soldiers subsequently entered the Army's recruiter school, where training performance scores were gathered. A unit-weighted composite of the assessment ratings correlated .32 with overall training performance. Regarding differences between assessors in assessment validities, a nonparametnc randomization test revealed that these differences were at a level not greater than chance However, there were significant differences between assessors in mean assessment ratings Possible reasons for the validity finding are that the randomization test lacks statistical power to detect actual differences in this setting, that assessor training smoothes out differences in abilities related to accurate assessment, and/or that assessment exercises uniformly provide all assessors with good (and equal) opportunity to view behavior relevant to criterion performance. Additional research possibilities in this area are discussed m light of results obtained here and methodological issues raised in the article.