As predictive validity is so generally established using bivariate correlations or multiple regressions job performance must usually be reduced to a single variable. This is in spite of considerable evidence that job performance is a multidimensional construct composed of several facets. Additionally, the predictive validity of ability tests declines with increasing time from testing, suggesting that job performance may best be viewed as a developmental construct. When both of these premises are true, it is possible that different developmental trends in job performance facets are obscured by considering only univariate performance. The purpose of this paper is to explore how to predict the development of different dimensions of job performance over time. We refer to this situation as validation for multivariate dynamic criteria. Viewing predictive validity in this way necessitates a process model for the domain of job performance. This paper develops a theoretical process model that draws together concepts from socialization and the literature on skill acquisition (Ackerman, 1987; Murphy, 1989) and presents a multivariate multilevel model for multiple performance facets measured at 4 times. The model was fitted to data from a 12-week long police training academy using the SAS and R systems. Cognitive ability and Big 5 personality traits are used to predict slopes and intercepts for the performance criteria. Changes in the criteria are mapped onto changes in identity and motivation mech- anisms. The pattern of findings is generally consistent with corresponsive development and socialization effects.
【 预 览 】
附件列表
Files
Size
Format
View
Multivariate dynamic criteria: a process model of job performance
PDF
download
878987797
028-85220240
