Identification Strategies in Survey Response Using Vignettes

In this paper we explore solutions to a particular type of heterogeneity in survey data which is manifest in the presence of individual-specific response scales. We consider this problem in the context of existing evidence on cross-country differences in subjective life satisfaction, and in particular the extent of cross-country comparability. In this instance observed responses are not directly comparable, and inference is compromised. We utilise two broad identification strategies to account for scale heterogeneity. Keeping the data fixed, we consider a number of estimators based on alternative generalisations of the ordered response model. We also examine a number of alternative approaches based on the use of additional information in the form of responses on one or more additional questions with the same response categories as the self-assessment question. These additional questions, referred to as anchoring vignettes, can under certain conditions, be used to correct for the resultant biases in model parameters.