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

<p>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. <i>M</i>ore risk and ambiguity averse farmers are less likely to adopt fertiliser microdosing indicating the importance of insurance and strategies to promote learning.</p>