The Long and Short-Form Warwick-Edinburgh Mental Well-Being Scale: A Reliability Generalization Meta-Analysis

Abstract

Score reliability is an essential property of the measurement process with implications for validity of scores and subsequent inferences. Using a reliability generalization (RG) approach, score reliability estimates produced by the Warwick-Edinburgh Mental Well-being Scale (WEMWBS) and the Short Warwick Edinburgh Mental Well-being Scale (SWEMWBS) were synthesized then evaluated. More specifically, the following questions were addressed: (1) What is the typical reliability coefficient for scores generated by using the WEMWBS/SWEMWBS? and (2) What (if any) factors impact score reliability estimates across studies using the WEMWBS/SWEMWBS? This study used non-experimental research design with archival data. Guided by systematic inclusion/exclusion criteria, electronic database searches identified 294 published articles reporting estimates of score reliability for either the WEMWBS or SWEMWBS. Sample, design and instrument characteristics were extracted then coded examined to address factors that may impact score reliability for both instruments. Across all published studies, mean score reliability estimates for the WEMWBS and SWEMWBS were 0.89 and 0.81 respectively across published studies. Moderator analyses revealed select sample (e.g., Age) and instrument characteristics (e.g., standard deviation of scores) that influenced score reliability. Limited error of measurement was evident based on average score reliability estimates for the WEMWBS/SWEMWBS. Further, interpretation of the moderator analyses demonstrated scale scores are quite robust to test administration in different samples, instrument versions and study designs.