The Substantive Validity of Work Performance Measures: Implications for Relationships Among Work Behavior Dimensions and Construct-Related Validity

Date

2012-10-19

Journal Title

Journal ISSN

Volume Title

Publisher

Abstract

Performance measurement and criterion theory are critical topics in the fields of I/O psychology, yet scholars continue to note several issues with the criterion, including empirically redundant behaviors, construct and measure proliferation, and definitions that conflict. These interconnected problems hinder the advancement of criterion measurement and theory. The goal of this study was to empirically examine the issues of theory/construct clarity and measurement as they exist regarding work performance behaviors.

This study's first objective was to clarify definitions of core performance behaviors, particularly to resolve issues of construct proliferation and conceptual conflict. Universal definitions of four core criterion constructs (i.e., task performance, citizenship performance, counterproductive work behavior, and withdrawal) were developed that integrated existing definitions of similar behaviors. Each definition reflects a parsimonious conceptualization of existing performance behaviors, which serves to clarify existing, and at times divergent, criterion conceptualizations. Importantly, these integrated definitions represent commonly-held definitions of the constructs and replace the largely discrepant accumulation of definitions.

The second objective was to determine whether existing items assumed to measure the four core work performance behaviors were judged by raters to represent their respective constructs. The results showed that of the 851 items examined, over half were judged to not represent their respective constructs which, importantly, replicated previous research. Additionally, the results highlight items that match their respective construct definition and contain minimal overlap with non-posited constructs.

Finally, the third objective was to determine the implications of using the problematic items for both the empirical relationships among work performance behaviors and evidence of construct-related validity. The results provided preliminary evidence that while nomological networks are minimally affected, relationships among some work performance dimensions are significantly affected when problematic items are removed from measures of performance constructs.

This dissertation demonstrated the need for more attention to the construct labels placed on the behaviors described in work performance items, as there are potentially adverse consequences for theory and measurement. Ultimately, the results of this study showed that work performance behaviors/items have often been assigned incorrect construct labels which, subsequently, may cast considerable doubt on the theoretical and empirical understanding of the criterion domain.

Description

Citation