Validity - What is it?

Validity is the benchmark of any selection procedure's usefulness. The validity of a test indicates:

1. A test measures what it purports to measure.
2. Inferences made from test scores are supported by research

Several validation strategies are outlined in the Uniform Guidelines on Employee Selection Procedures, including content validity, criterion-related validity, construct validity, and transportability.

Content Validity refers to the relationship of test content to job content. Content validity is achieved by a combination of job analysis, evaluation of test content for job relevance, and evaluation of statistical properties of the test.

Criterion-Related Validity is established by showing how strongly a particular hiring method (or predictor) relates to job performance. The resulting coefficient indicates the strength of that relationship. Validity coefficients can range between 0.0 and 1.0. Industrial psychologists have determined that, in most cases, a validity coefficient above .20 is acceptable and a coefficient above .30 is good depending on sample size.

Construct Validity involves showing that tests are an acceptable measure of an attribute, such as mechanical aptitude, that is also found to be important to job performance. This demonstration usually requires criterion-related validation.

Transportability involves using the results of a criterion-related validation study which has been conducted elsewhere to support the validity of a test in a new setting. The procedure involves careful documentation to show that jobs in the new setting are substantially similar to the jobs studied in the original validation study.

The strongest case for test validity will draw on several lines of validity evidence to document validity. Additionally, a case can be made for VALIDITY GENERALIZATION. Until the late 1970s and early 1980s, professional standards and practices were based on the belief that validities for cognitive ability tests were situationally specific. That meant that a test that was valid for one job might not be valid for another job because the characteristics of the situation tended to moderate the relationship between the test and criterion measure. However, n 1977, landmark research provided a different explanation for the observed differences in validity coefficients (Schmidt & Hunter, 1977). Schmidt and Hunter showed that the variability in validity results may not be due to real differences between settings but to artifactual sources. Specifically, three sources of artificial error were shown to account for almost all of the variation in validities across studies: sampling error, differences among studies in the amount of test range restriction, and differences among studies in the reliability of criterion measures. Thus, in many cases the situational specificity hypothesis can be shown to be false. A set of quantitative procedures, called meta-analysis, can be used to demonstrate that true validities are consistent from situation to situation and that validity is generalizable.

CM Consults is a human resources consulting firm that specializes in the development and implemention of valid and non-discriminatory employee hiring, promotion, and training procedures. We can help you establish the validity of your selection and promotion procedures.