Eliciting Risk Attitudes in the Field: Surveys or Experimental Methods? An Empirical Comparison in Rural Niger

We compare several risk preference elicitation methods - including incentivised, non-incentivised, and framed methods as well as a traditional Likert survey question - in a developing country and empirically test how well consequent measures of risk attitudes predict risk taking behaviour. We find that Likert scale and non-incentivised framed survey questions are not sufficient substitutes for costlier incentivised methods in rural Niger. Instead, the incentivised framed question works best while a simplified incentivised lottery question works almost as well. More risk and ambiguity averse farmers are less likely to adopt fertiliser microdosing indicating the importance of insurance and strategies to promote learning.