Programs and policies designed to improve the lives of the poor often rely on an understanding of peoples’ livelihoods and welfare that is based on information collected during long, multi-topic household surveys. These survey efforts are invaluable tools for development practitioners, but choices made by those designing these surveys can impact the information collected in ways that are difficult to predict. Increasingly researchers are incorporating experiments into survey work in an effort to better understand these methodological issues. For example, researchers can vary the order of questions to determine impacts of respondent fatigue, or test different versions of the same question to analyze how responses change.

In this webinar, we review three such experiments and reflect upon what we have learned about choices in survey design and what questions still remain. The experiments include an analysis of how respondent fatigue related to repeated questions affects labor measurement in rural Ghana, how the placement of the dietary module affects dietary diversity estimates in a phone survey in Ethiopia, and finally how different ways of defining reference periods changes responses in a food consumption module, also in Ethiopia.