Research on subjective wellbeing typically assumes that responses to survey questions are comparable across respondents and across time. However, if this assumption is violated, standard methods in empirical research may mislead. I address this concern with three contributions. First, I give a theoretical analysis of the extent and direction of bias that may result from violations of this assumption. Second, I propose to use respondents’ stated memories of their past wellbeing to estimate and thereby to correct for differentials in scale use. Third, using the proposed approach, I test whether wellbeing reports are intrapersonally comparable across time. Using BHPS data, I find that the direction in which explanatory variables affect latent satisfaction is typically the same as the direction in which scale use is affected. Unemployment and bereavement appear to have particularly strong effects on scale use. Although discussed in the context of life satisfaction scales, the proposed approach for anchoring response scales is applicable to a wide range of other subjectively reported constructs.
subjective wellbeing BHPS life satisfaction vignettes memories differential item functioning