Measuring household food security through surveys: Do the characteristics of the enumerators matter?

Motivation The accuracy of food security indicators, collected through enumerated surveys, is crucial for tailoring policies and programming. Nevertheless, there is little evidence on whether the characteristics of the enumerators conducting the interviews can affect food security data. Purpose The purpose of this article is to explore the so-called enumerator effect for an array of food security indicators that are commonly adopted for targeting, monitoring and evaluation in humanitarian and development assistance. Approach and methods We combine a household-level dataset collected in northern Uganda refugee-hosting districts with an enumerators' dataset. We estimate the enumerator effect and its determinants for a large set of food security indicators. Additionally, we test the size of the bias, namely the percentage of households that are misallocated to be food poor. Findings We find that the effect of the enumerator experience within the refugee context is the most relevant to the food indicators. Additionally, obtaining sensitive information on food shortage is more difficult when the interviewer comes from a similar sociocultural background as the interviewee. Moreover, we show that the enumerator effect leads to a bias in the food poor headcount, implying distortion in the targeting. Conclusion Data collections must acknowledge the enumerator effect. This has consequences that can ultimately affect the number of people reached by programmes and projects in both humanitarian and development contexts.