Technical Documentation: 

Employment Test

Overview

This technical documentation is designed to meet various professional and legal guidelines. As such it contains language unfamiliar to many human resource and other professionals. We would be happy to answer any questions you have concerning this documentation. Questions on this report may be directed to Technical@workskillsfirst.com  or by calling 804.301.3036.

Test Content

The Employment Test consists of items that measure general mental ability, including vocabulary, reasoning, following instructions, number series, and analogies.  Employers seeking to determine the usefulness of the test for a specific job may review the test content to determine the appropriateness of the test for the job. Employers may also consider the reliability and validity evidence presented in this report.

The validity of cognitive ability tests to predict performance is around .51 (Hunter & Hunter, 1984; Schmidt & Hunter, 1998).

For selection – from Schmidt & Hunter (1998)

• .58 for professional – managerial jobs

• .56 for high –level complex technical jobs

• .51 for medium complexity jobs (62% of US jobs are considered medium complexity.  This category includes skilled blue collar and mid-level white-collar jobs – ex. upper level clerical and lower level administrative jobs)

• .40 for semi-skilled jobs

• .23 for completely unskilled jobs
  
   

For job-related learning/training – from Schmidt & Hunter (1998)

• Approximately .56 for the average predictive validity for performance in job-related training programs.

According to Schmidt and Hunter (1998) general mental ability (GMA) tests have the following relationships with these other measures.

This table shows that adding GMA to a test battery can add a large amount of accuracy to most commonly used selection methods.  For example, if an employer is currently using an unstructured interview with a validity of .38, adding a test of GMA can increase the validity to .55.  This represents a 17% increase in accuracy of prediction.

A difference of approximately one standard deviation has been repeatedly found between African Americans and White subgroups on cognitive ability testing (Hunter and Hunter, 1984; Gottfredson, 1986; Jensen, 1980; Schmitt, Clause, & Pulakos, 1996). 
 

According to Pulakos and Schmitt (1996), the difference between Hispanics and Whites on cognitive ability tests is just under one standard deviation.
 

According to the Principles (SIOP, 1987), [for cognitive ability tests] there is little evidence that there is differential prediction between the sexes or between racial groups.

In a study examining the ability to reduce the adverse impact caused by cognitive ability testing, Schmitt, Rogers, Chan, Sheppard and Jennings (1997) found that effect size differences between majority and minority groups remained high enough to often produce prima face cases of adverse impact even when three alternate predictors (personality, biodata, and structured interviews) were used along with cognitive ability testing. 

Schmitt et al. concluded that when possible to do so, the use of alternate predictors that have low between-group differences, high validity, and high intercorrelations will have the greatest effect of decreasing the overall adverse impact of a selection battery that includes cognitive ability testing.  This particular combination of alternate predictors will by itself decrease the likelihood of a selection battery producing adverse impact.  However, there is no guarantee that adverse impact will be completely removed. 

With regard to utility, the predictive ability of a set of predictors (multiple R) will be decreased if one uses a set of highly intercorrelated predictors.  Therefore, there are tradeoffs that must be made between predictive efficiency and decreasing levels of adverse impact.   In a weighted composite, the predictors with the best validity are often the ones that receive the most weight.  If these predictors also have large between-group differences, then the effect size between majority and minority groups will be increased. 

Schmitt, et al. (1997) actually found that a composite of alternate predictors that had low validity, high between-group differences, and low intercorrelations will produce greater differences between groups than will cognitive ability testing alone.

A score of 1-3 suggests that the candidate is suitable for jobs of low complexity or relatively unskilled jobs, such as janitor, nursing assistant, fast-food employee, material handler, data entry clerk, data copier/coder, or mechanic helper. 

A score of 4-6 suggests that the candidate is suitable for jobs of medium complexity or relatively skilled jobs, such as light industrial jobs, including machinist or electrician, or office jobs, including retail or sales associate, call center associate, bank clerk, laboratory technician, supervisor, or administrative assistant.

A score of 7-9 suggests that the candidate is suitable for highly skilled and/or professional/technical jobs, such as engineer, manager, IT analyst, financial analyst, accountant, or business analyst. 

Users of the Employment Test may wish to make their own hire recommendations based on the availability of applicants in the local applicant pool, the performance records of employees who were previously tested, and the adverse impact of the survey on their applicants. We recommend that users of the survey consult with an industrial organizational psychologist. Work Skills First, Inc. invites users of the the survey  to contact its staff with doctoral degrees in industrial/organizational psychology if the user has questions about the appropriate use of the test.

Gottfredson, L.S. (Ed.) (1986).  “The g factor in employment,” Special issue of the Journal of Vocational Behavior, 29, 293-450.

Hunter, J., & Hunter, R. (1984).  Validity and utility of alternative predictors of job performance.  Psychological Bulletin, 96, 72-98.

Jenson, A. (1980).  Bias in Mental Testing.  New York: Free Press.

Pulakos, E, & Schmitt, N.  (1996).  An evaluation of two strategies for reducing adverse impact and their effects on criterion-related validity.  Human Performance, 9, 241-258.

Schmidt, F., & Hunter, J.  (1998).  The validity and utility of selection methods in personnel psychology: Practical and theoretical implications of 85 years of research findings.  Psychological Bulletin, 124, 262-274.

Schmitt, N., Clause, C., &  Pulakos, E.  (1996).  Subgroup differences associates with different measures of some common job relevant constructs.  In C. L. Cooper & I. T. Robertson (Eds.), International Review of Industrial and Organizational Psychology (pp. 115-139). New York, Wiley

Schmitt, N., Rogers, W., Chan, D., Sheppard, L., & Jennings, D. (1997).  Adverse impact and predictive efficiency of various predictor combinations.  Journal of Applied Psychology, 82, 719-730. 

Copyright © 2004 by Work Skills First, Inc.