A methodological comparison of the empirical assessment of adaptation to disability: Can time heal all wounds?

Chronically disabled patients generally report a higher quality of life in their impaired health state compared to members of the general public when imagining the experience of the same health state. The difference in the patient’s experience and the public’s ideation is often attributed to adaptation. This master thesis studies adaptation to chronic disability in both self-perceived health and life satisfaction in a large longitudinal data set. Moreover, it examines what model specification is best suited to measuring adaptation in panel data. I select over 5000 respondents of the Survey of Health, Ageing and Retirement in Europe (SHARE) who develop a chronic illness and disabilities during the span of the 6 waves of data collection. In order to examine the effect of time since the onset of disability on self-perceived health and life satisfaction, a fixed effects ordered probit model and a linear fixed effects model are recommended in the literature. I propose a fixed effects ordered logit model because the dependent variable is measured on an ordinal scale and the proposed parameterization of the fixed effects in the ordered probit model is prone to misspecification. In order to assess how different model specifications affect the effect associated with adaptation, I also analyze the fixed effects probit and linear models. Self-perceived health significantly decreases when the disabilities occur, but life satisfaction remains the same. Supportive evidence for adaptation in the life satisfaction analysis was found, but not in that for self-perceived health. It is possible that the effect of adaptation to chronic disability in selfperceived health can only be found after a longer duration than that measured here. The model features that appear to affect the measured adaptation process the most are added dynamics in the form of a lagged dependent variable and the assumed measurement scale of the dependent variable with associated estimation methods. The difference in outcome between the analysis with self-perceived health and that with life satisfaction can be explained by the contextualization of the response variables, where the question on self-perceived health is more focused on health limitations and the question on life satisfaction on general well-being.