2026/01/01 by Martin Paul Jr. Tabe-Ojong, Kibrom A. Abay, Gashaw T Abate +1 · 1 voice
Mathematics · Social Sciences · #Advanced Causal Inference Techniques #Survey Methodology and Nonresponse #Survey Sampling and Estimation Techniques
paper · doi:10.1093/jae/ejag001
openalex publication_date 2026/01/01 · openalex created_date 2026/03/07 · openalex updated_date 2026/06/19
Abstract Constructing per capita estimates (e.g., consumption), poverty rates, and other development indicators requires the listing of household members as well as their detailed demographic characteristics in household surveys using rosters. We designed an experiment to understand how respondents and enumerators manage such lengthy listing exercises in household surveys. We randomly assign respondents to ‘committed’ and ‘uncommitted’ household member listing rosters. If respondents are likely to reduce effort, they would decrease the number of household members reported in the uncommitted option, when they have the choice to list fewer members and save time. We find a 6–7% reduction in the number of household members listed when household rosters are enumerated with an uncommitted method, implying that 1 out of every 5 respondents reported an additional household member when asked using the committed household roster. Respondent/survey fatigue could explain the difference in the two measures of household size. Overall, our findings speak to the critical role of survey design in ensuring data quality and the important trade-off between survey length and data quality.