Subjective well-being and social desirability

Abstract

Survey measures of depression are increasingly used by economics researchers to provide a nuanced account of well-being. I show that levels of depression reported using such measures are significantly understated and levels of happiness significantly overstated in survey interviews conducted using a response mode that does not allow for anonymous reporting compared to a mode that does in three longitudinal surveys widely used in economics research. I exploit randomized assignment to survey mode, as well as panel methods, to show that this reflects the causal effect of survey mode, not selection. The difference in reported depression and happiness between modes is comparable to the difference between individuals in the 25th and 75th income percentiles. This finding suggests perceptions of social desirability may substantially bias measures of subjective well-being.