Rasch analysis of the Personal Wellbeing Index.

Abstract

The Personal Wellbeing Index (PWI) is used internationally to measure subjective well-being. While numerous studies have reported on the psychometric properties of the PWI using classic test theory, the current paper provides additional psychometric evaluation using the Rasch measurement model. The responses to the PWI items of 593 healthy adults living in Australia or Canada were analysed using Rasch analysis. Assessed were overall model fit was assessed, individual person fit and item fit, reliability, differential item functioning (DIF), unidimensionality, targeting, and response format. Results indicate excellent psychometric properties of the PWI, provided the religion/spirituality item is excluded. The seven-item PWI showed good model fit (overall item-trait interaction χ (2)=70.59, df=63, p=0.24), excellent person separation (PSI=.89), no item or person misfit, and no DIF for country or gender, and unidimensionality was supported. In addition, only very mild disordered thresholds were observed for the 'safety' item, indicating that overall, the 11-point response format was suitable for this population. It is recommended that the religion/spirituality item not be included in the aggregated or averaged total score in Western context and that caution should be used in comparison across different groups where the eight-item version has been used.