Using survey questions to measure preferences : lessons from an experimental validation in Kenya

Can a short survey instrument reliably measure a range of fundamental economic preferences across diverse settings? We focus on survey questions that systematically predict behavior in incentivized experimental tasks among German university students (Becker et al. 2016) and were implemented among representative samples across the globe (Falk et al. 2018). This paper presents results of an experimental validation conducted among low-income individuals in Nairobi, Kenya. We find that quantitative survey measures - hypothetical versions of experimental tasks - of time preference, attitude to risk and altruism are good predictors of choices in incentivized experiments, suggesting these measures are broadly experimentally valid. At the same time, we find that qualitative questions - self-assessments - do not correlate with the experimental measures of preferences in the Kenyan sample. Thus, caution is needed before treating self-assessments as proxies of preferences in new contexts.