Abstract
This study investigated the validity of two measures of work engagement (the Utrecht Work Engagement Scale (UWES) and the May, Gilson and Harter scale) that have emerged in the academic literature. Data were collected using surveys with 139 employees in the Auckland-based call centers of two finance organizations, to assess the validity of the two measures. Some evidence for convergent, discriminant and predictive validity was found for both scales, although neither showed discriminant validity with regard to job satisfaction. Overall, the three factors of the UWES (vigor, dedication and absorption) performed slightly better across analyses than the three factors from the May, Gilson and Harter (Citation2004) measure (cognitive, emotional and physical). There are some important differences between the two scales, raising questions about how we should be measuring work engagement. The current use of different descriptions and measures means that findings will be specific to each of these. This limits generalizability across studies, which will both slow theoretical progress and reduce the ability of science to contribute to practice.
Notes
1. Of the three other engagement measures published in academic journals, Rothbard's (Citation2001) scale focuses on attention and absorption, which are both similar to the UWES absorption dimension. Saks (Citation2006) measured job and organization engagement, with job engagement similar to the UWES absorption dimension, and organization engagement having some overlap with the UWES vigor dimension. Recently, Rich, LePine and Crawford (Citation2010) have developed a measure of engagement from Kahn's model, which partly draws from Rothbard's (Citation2001) engagement measure for the cognitive or absorption aspect.
2. Note that the two-item cognitive scale showed acceptable reliability.
3. Following the advice of one reviewer, we conducted exploratory factor analyses. These were conducted in MPlus version 6.11 since exploratory analyses are simpler with this program. The analyses compared one-, two- and three-factor models. For both the UWES and the May et al. scales, a three-factor model gave the best fit to the data (UWES: χ2 = 219.48, df = 102, p = 0.00, TLI = 0.870; CFI = 0.914, SRMR = 0.048; May: χ2 = 29.514, df = 25, p = 0.243; TLI = 0.959, CFI = 0.981, SRMR = 0.036).