The Measurement and Profile Analysis of Cognitive Multidimensional Workplace Identification

Trevor Thomas Coppins, Western University

Abstract

Workplace identification literature has assumed a multidimensional lens since its beginnings, but it has yet to fully adopt this scope conceptually or in research methodology. The present research sought to address these gaps through the creation of a new multidimensional measure of workplace identification and latent profile analyses, which simultaneously examine how different identifications of differing strengths can be packaged together to predict outcomes. In study 1, the cognitive multidimensional workplace identification (CMWI) scale was developed and pilot tested for factor structure and reliability (n = 184). Results indicated that the CMWI scale had excellent fit to the anticipated 6-factor structure (organizational, team, coworker, supervisor, occupation, and career identification), and outperformed the common keyword replacement Ashforth & Mael (1992) identification scale, which failed to exhibit satisfactory fit indices on any model examined. In study 2 (n = 484), latent profile analyses indicated a 5-profile solution was optimal: fully identified, disidentified, career-focused, moderate-negatively identified, and supervisor-distant predominantly identified profiles. These profiles differed significantly on all measures of work attitudes (e.g., commitment, turnover intentions), psychological health (i.e., burnout), and behaviour (i.e., citizenship behaviour). In study 3, measurement invariance of the CMWI and profile similarity tests were conducted across a sample of participants from North America (United States and Canada) and the United Kingdom. The CMWI exhibited complete invariance with the exploratory structural equation model and high reliability. The latent profile similarity tests exhibited mixed results. The United Kingdom sample had a 4-profile optimal solution which replicated the fully identified, career-focused, and disidentified profiles with a 4^th unique profile, relationally-focused, that was hypothesized but not found in study 2. When the 5-profile structure was imposed four out of five profiles replicated across samples. Profile similarity tests indicate complete similarity for a parsimonious 3-profile solution. With the exception of dispersion similarity, the 5-profile solution exhibited all forms of similarity across samples. Theoretical implications and future research directions of person-centered identification research are discussed.