Just swapping the order of questions on a survey can change the results by 30%—which could waste millions in charity money.
March 20, 2026
Original Paper
The Questionnaire as Treatment: Question Order and 2 Measurement Bias in Development Surveys
SSRN · 6441737
The Takeaway
Researchers found that if you ask parents about their behaviors before asking about their attitudes, they subconsciously anchor their answers to appear more consistent. This 'order effect' suggests that high-stakes data used by organizations like UNICEF might be reflecting the design of the questionnaire rather than the actual parenting realities or needs of the poor.
From the abstract
KAP (knowledge-attitude-practice) surveys are the primary tool for identifying parenting program targets in low-income countries, yet question order is rarely treated as a consequential design parameter. We treat the questionnaire itself as the treatment: a 2x2 randomized factorial experiment embedded in a UNICEF household survey in Burundi (N = 751 households with children under six) varies the order and spacing of attitude and behavior modules. Placing behavioral questions before attitudinal q