The Psychometric Equivalence of the Personal Wellbeing Index for Normally Functioning and Homeostatically Defeated Australian Adults

Understanding subjective wellbeing (SWB) at the population level has major implications for governments and policy makers concerned with enhancing the life quality of citizens. The Personal Wellbeing Index (PWI) is a measure of SWB with theoretical and empirical credentials. Homeostasis theory offers an explanation for the nature of SWB data, including the distribution of scores, maintenance and change over time. According to this theory, under normal conditions, the dominant constituent of SWB is Homeostatically Protected Mood (HPMood), which is held within a genetically determined range of values around a set-point. However, in extreme circumstances (e.g., financial hardship, chronic illness), HPMood may dissociate from SWB, as cognitive/emotional reactions to the cause of homeostatic challenge assume control over SWB. This study investigates two groups as people scoring in the positive range for SWB and people who are likely to be experiencing homeostatic defeat/challenge. We test whether the reduced influence of HPMood on SWB due to homeostatic defeat has implications for the validity of SWB measurement. Participants were 45,192 adults (52 % female), with a mean age of 48.88 years (SD = 17.35 years), who participated in the first 23 surveys of the Australian Unity Wellbeing Index over the years 2001–2010. Multiple regression analysis, multiple group confirmatory factor analysis, and Rasch modelling techniques were used to evaluate the psychometric performance of the PWI across the two groups. Results show that while the PWI functioned as intended for the normal group, SWB in the challenged group was lower across all PWI domains, more variable, and the domain scores lacked the strength of inter-correlation observed in the normal, comparison group. These changes are consistent with predictions based on homeostasis theory and one major implication of the findings is that SWB measures may not function equivalently across the entire spectrum of possible domain satisfaction scores.