Abstract: | Examined systematically the sample sizes necessary to provide adequate power in validation studies under various conditions of range restriction and criterion unreliability. For purposes of brevity, the examination was restricted to the validity parameter values (i.e., true validities) of .35 and .50. Results demonstrate that sample sizes required to produce adequate power in empirical validation studies are substantially larger than has typically been assumed. This finding leads to the conclusion that, from the viewpoint of sample-size requirements, criterion-related validity studies are "technically feasible" much less frequently than is commonly assumed. It is also shown that the "situational specificity" of employment test validities may be in large part a consequence of excessive faith in small-sample analyses, that is, belief in "the law of small numbers." (16 ref) (PsycINFO Database Record (c) 2010 APA, all rights reserved) |