Journal of Applied Psychology 1984, Vol 69, No 2, 322-333 Copynght 1984 by the Amencan Psychological ASSOC12t1on Inc Synthetic Validity: A Conceptual and Comparative Review Kevin W. Mossholder Department of Management, Auburn University Richard D. Arvey Industrial Relations Center, University of Mmnesota Synthetic vahd1ty is exammed with respect to its ongms and approaches taken to this concept Two programmatic approaches to synthetic vahd1ty, the J-coeffic1ent and Job Component Model, are discussed m comparative terms, with attent10n directed at the respective researcl:i evidence, statistical features, and relative strengths and weaknesses Other approaches to the concept are also reviewed, Illustratmg different means of attammg synthetic vahd1ty Issues of apphcat10n and future potential are h1ghhghted Personnel selection specialists have empha- sized the vahdat1on strategy of ident1fymg po- tential predictors and measures of job perfor- mance one JOb at a time and correlatmg them in some way. Although this strategy has served well, it has several hm1tat1ons (Gmon, l 976, Schmidt, Hunter, & Pearlman, 1981). For ex- ample, it may not be feasible for jobs with few people to study, the cost of Job-by-Job vah- dation may be prohibitive, or statistical prob- lems may lead to poor validity estimates. Per- sonnel selection specialists have therefore be- gun to consider alternatives (Tenopyr & Oeltjen, 1982). Synthetic validity is one al- ternative, though it is relatively poorly under- stood. The purposes of this article are (a) to define and explain the synthetic vahd1ty model, (b) to trace the history of the model, (c) to review and compare various approaches to the model, and (d) to summarize and suggest fu- ture trends and directions. Origins and Defimt10ns of Synthetic Vahd1ty Link ( 1920) suggested that traits reqmred on a given JOb could be described m terms of incumbents' performance on psychological tests. Tests could be taken by workers m a given JOb, mean scores compiled, and a profile established describmg the aptitudes required by the job. Applicants would be hired if their The authors thank Hubert S Feild for comments on preVJous draft on this article Requests for reprmts should be sent to Kevm W Moss- holder, Department of Management, School of Busmess, Auburn Umvers1ty, Auburn, Alabama 36849 322 profiles matched the required job profile. Re- search along these lines (e.g., Job Psychograph, Viteles, 1932) was seldom successful Usmg tests to describe JOb reqmrements imphes (a) the existence of umque traits that describe specific JOb reqmrements, (b) the discovery of tests that validly measure these traits, and (c) a correspondence between test score profiles and JOb performance. Ghiselh and Brown ( 1948) concluded that such condit10ns were unreahstic and mstead proposed that JOb and worker specifications be used to classify JObs mto occupational categories (e.g., D1ctwnary of Occupatwnal Tales, DOT). Emphasizmg worker aptitude reqmrements could encourage occupational testmg to verify apphcants' JOb fitness m an occupat10nal area. Segregatmg JObs by aptitude reqmrements and testing across these aptitudes foreshadowed synthetic vahd1ty logic Synthetic vahd1ty was mtroduced by Lawshe ( 1952) at a symposium on mdustrial psy- chology for small busmesses As opposed to situational vahd1ty-test validity based on ex- clusive use m one firm-and generalized va- lidity-test validity based on wide use and normative data-the term synthetic vallday was offered "to denote the inferrmg of validity in a specific situation" (p 32). An important difference between synthetic vahd1ty and gen- eralized validity is that the former reqmres greater exammation of the work s1tuat10n, whereas the latter mvolves global appraisals of job similarity Emphasizmg this difference, Baima ( 1959) refined the term to mean d1s- covermg vahdity m a specific s1tuat10n by an- alyzing Jobs mto thelf components, determm- ing test validity for these components, and combming these validities mto a whole.