ABSTRACT
In a recent discussion of measurement models in climate research, James (1982) recommended that a decision of whether to aggregate individual’s climate scores should be a function of the magnitude of an intraclass correlation estimate of interrater reliability. This recommendation was based on the following rationale: (a) the basic unit of theory (unit of analysis) for climate is an individual’s perception of his or her psychological climate (James & Sells, 1981; Jones & James, 1979; Joyce & Slocum, 1979; Schneider, 1983); (b) a composition theory relating psychological climate scores to aggregate psychological climate scores (e.g., organizational climate scores) may be established if the perceptions of psychological climate are shared among the individuals on whom the aggregate is computed (Roberts, Hulin, & Rousseau, 1978); and (c) the typical design employed in climate studies is a random effects, one-way analysis of variance (ANOVA), from which, given reasonable satisfaction of assumptions, it is possible to estimate interrater reliability (perceptual agreement, degree perceptions are shared) by the intraclass correlation equation for the reliability of a single rating or measurement (referred to here as ICC(1)–cf. Bartko, 1976; Ebel, 1951; Shrout & Fleiss, 1979; Winer, 1971).
