The order of your questions changes what your survey measures.

We ran a randomized experiment in Burundi — 751 households, 4 survey versions, same questions — and found that asking about child nutrition practices before attitudes increases how consistent respondents appear across both. By 30% of a standard deviation. Not because anything changed in their lives. Because of where the question sat in the questionnaire.

Three takeaways for anyone designing KAP surveys in development contexts:

1. Your questionnaire is not neutral. The sequence you choose shapes the data you get — before a single respondent reads the first question.
2. Attitude-behavior gaps may be artifacts. If your survey asks attitudes first and practices second (as most do), you may be overestimating inconsistency.
3. Spacing between modules does not fix this. We tested it. No measurable effect.

Full paper: https://lnkd.in/evbsmsuR

With @Marco Giani, @Daniela Mejía Tejada, Laura Quintero, and @Philip Verwimp

#DevelopmentEconomics #SurveyDesign #EvidenceForPolicy #FieldResearch #KAP