“Evaluating Indicators of Job Performance: Distributions and Types of Analyses”, 2016-11 (; similar):
Distributions of job performance indicators have historically been assumed to be normally distributed (Aguinis & 2014; 1983; 1947). Generally, any evidence to the contrary has been attributed to errors in the measurement of job performance (2008).
A few researchers have been skeptical of this assumption (1989; 1999; et al 1980); yet, only recently has research demonstrated that in certain specific situations job performance is exponentially distributed ( et al 2016; 2012). To date there have been few recommendations in the Industrial-Organizational Psychology literature about how to evaluate distributions of job performance to determine whether they fit an exponential curve. There also has not been substantial justification in the literature as to why distributions of job performance would be expected to be normally distributed versus exponentially distributed. Furthermore, recent research about job performance distributions has narrowly focused only on a few specific types of work and on a few specific indicators of performance. Thus, research concerning distributions of job performance indicators is, to date, of limited generalizability.
The current research attempts to close the gaps in the literature by identifying high fidelity methods and applying them to classify distributions of various indicators of job performance on a continuous spectrum from normal to exponential.
In this research, multiple types of indicators of performance (and indices computed from combinations of indicators) were found to produce exponential distributions. More specifically, managerial indicators of job performance were found to best fit a normal distribution whereas objective measures, as well as composite measures of performance consisting of objective and subjective indicators, were found to best fit an exponential distribution.
This study provides researchers and practitioners with new suggestions for classifying job performance distributions as well as new techniques for better differentiating between top and bottom performers.