Composite vs. Multiple Criteria: A Review and Resolution of the Controversy

Abstract

THERE has been a controversy of long standing in industrial psychology between the advocates of the single composite criterion and those favoring multiple criteria. Earlier writers on this problem tended to favor the composite criterion concept (Toops, 1944;Brogden and Taylor, 1950; Nagel, 1953), and this doctrine remained essentially unchallenged until the second half of the fifties. Then, starting in 1956, a series of statements by a number of wellknown industrial psychologists appeared seriously questioning the utility of this concept and concluded that, in most cases, the use of multiple criteria was to be preferred to the use of the traditional composite (Ghiselli, 1956; Guion, 1961Guion, , 1965 Dunnette, 1963a Dunnette, , 1963b. To date, this controversy has not been satisfactorily resolved, as is obvious from the cautious and rather indecisive treatment the question receives in most standard texts in industrial psychology (Tiffin and McCormick, 1965, pp. 47-54, Blum and Naylor, 1968, pp. 184-193). The purpose of this paper is to set forth the basic arguments advanced for each position, to explore the assumptions underlying these arguments, and, finally, to offer a practical and conceptual resolution of the controversy.

Proactive Personality and Big Five Traits in Supervisors and Workgroup Members: Effects on Safety Climate and Safety Motivation

“…Schmidt & Kaplan, 1971), Ones and Viswesvaran (1996) state this is not necessarily the case. They argue that bandwidth and fidelity are not opposite ends of the same continuum; there is no reason that a broad measure cannot predict a narrow criterion.…”

“…In general, the best prediction is expected when the bandwidths of the predictor and the criterion match (Schmidt & Kaplan, 1971;Hogan & Roberts, 1996;Paunonen & Ashton, 2001a). Ones and Viswesvaran (1996) argue that most criteria used to predict work-related outcomes are based on broad samples of behavior gathered over time, for example, supervisor ratings of overall job performance.…”

Construction of a Job in General scale: A comparison of global, composite, and specific measures.

“…Composite measures, therefore, do not necessarily give the same results as asking for summary evaluations as in global scales (see Scarpello & Campbell, 1983;Schmidt & Kaplan, 1971;Smith et al, 1969). To summarize the arguments, composite scales may not be sufficient for estimating general satisfaction for several reasons:…”

“…The variety of content could even be an advantage in picking up valid general variance. (This is essentially the point made for criteria by Schmidt & Kaplan, 1971.) A combination of the items in the JDI subscales was therefore sought to form a composite scale.…”

Performance appraisal ratings: The content issue

“…This viewpoint appears to be based mw e on the selective appraisal of a narrow spectrum of studies than on an appraisal of all studies conducted in the field (Kavanagh, 1971). in any case, the difficulties and controversial issues inherent in ratings have been well documented (e.g., Barrett , 1966;Kavanagh, 1971;Ronan & Prien, 1971;Schmidt & Kaplan , 1971).…”