A Note on Empirical Studies of Life-Satisfaction: Unhappy with Semiparametrics?

Abstract

This article deals with the modeling of life-satisfaction, and estimating the impact of age on it. We investigate how findings and the interpretation of empirical studies hinge on the respectively assumed model. Assuming a specific model comprises various hypothesis made on the data generating process, like indicator selection, measurement, or functional form specifications. In this study we focus on the latter two issues. In particular, we show how different response behaviors (optimistic, pessimistic, extreme averse, etc.) lead to seemingly contradictory conclusions if the researcher does not address them adequately. In fact, we show that one can reproduce any shape found in the literature simply by modifying the way respondents rank life satisfaction on a bounded scale.