What if we took within-person performance variability seriously?

摘要

Extract: Efforts to understand what seems to be an unacceptably weak relationship between actual performance and rated performance have focused exclusively on the rater side of the model, not on the performance side. For instance, the Murphy (2008) model shows error only for ratings. Therefore, efforts to remedy the situation have also focused exclusively on raters: adjust the relationships of poor-quality ratings to other variables for attenuation because of unreliability, improve the raters by training, clarify the rating task by providing a better format, or enhance rater motivation to be honest in recording what they really think. A strong implicit assumption underlies all these approaches: that an employee’s job performance is stable and that there is some true level of performance available to be observed and rated, if raters were just capable or motivated to do so. But what if part of the problem is that performance is not entirely stable over the short term? I will first establish that this is the case, then draw out some implications of true performance fluctuation for the relationship between performance and performance ratings.