ABSTRACT
This chapter discusses the criterion-related and construct validity both require empirical evidence for their support. There are two subcategories of criterion-related validity: predictive and concurrent. Predictive validity refers to how test performance correlates with some future performance on a criterion and thus involves the problem of prediction. A pencil-and-paper test of knowledge of computers represents a concurrent validity problem where the simultaneous criterion is actual skills in computer use. Evaluating the criterion-related validity of a test requires examining the magnitude of the correlation coefficient, which is referred to as the validity coefficient. The linkages maybe correlation based or experimentally based hypothesis testing specifying between group differences. Lewis Terman et al. gave the tests to many thousands of participants and studied the nature of the distribution of the scores, correlations between the scores and other criteria, and the internal consistency of the subtests across different populations.
