The impact of survey characteristics on the measurement of food consumption

- We study the impact of survey design on food consumption measurement.
- Recall interviews report higher caloric consumption compared to diaries.
- Surveys focusing on food acquisition report more calories than others.
- Measurement errors show a higher impact than survey characteristics.

Survey characteristics affect the quality of the measurement of food consumption within households; thus, it is important to identify best practices for designing surveys that collect food data. This paper analyses the impact of survey characteristics on the measurement of food consumption from a sample of 81 national surveys. Results highlight regularities that can inform best practices in designing surveys and promoting the use of the data for multiple purposes. Surveys focused on food acquisition collect higher food quantities compared to those that target food consumption. Surveys based on recall interviews collect higher food quantities compared to those based on diaries, but the difference decreases with long reference periods. The use of standard units of measurement as well as the consideration of partakers in meals and of seasonality generates significant differences in the survey results. The impact of the different survey characteristics carries substantive implications when food consumption data are employed for assessing food security conditions. The results are part of a wider work program aimed at improving the quality of household survey data. More evidence is needed, ideally through coordinated sets of analyses and experiments in different contexts. Additionally, survey characteristics must be complemented by effective field work in order to generate high quality data. Towards this end, statistical capacity development is crucial to promote better data and more evidence-based decision making.