Data sets were derived from two independent cross-sectional surveys of the veterinary profession (n = 8,829 and n = 1,796). Rasch analysis (n = 500) included response option thresholds ordering, tests of fit, differential item functioning, targeting, response dependency, and person separation index (PSI). Unidimensionality was evaluated by principal component analysis of residuals. The findings were validated across further subsamples from both data sets. The external construct validity of the Rasch-fitting item set was evaluated by associations with other measures of psychological health or psychosocial work characteristics.
Results
Data for the original 14 items deviated significantly from Rasch model expectations (chi-square = 558.2, df = 112, P = <0.001, PSI = 0.918). A unidimensional 7-item scale (Short WEMWBS, SWEMWBS) with acceptable fit to the model (chi-square = 58.8, df = 56, P = 0.104, PSI = 0.832) was derived by sequential removal of the most misfitting items. The external construct validity of SWEMWBS was supported.
Conclusions
SWEMWBS has robust interval-level measurement properties which support its suitability as an indicator of population mental health and well-being in this occupational group with elevated suicide risk.
Read more HERE.
