Reproducibility package for Measuring Diets At Scale: How Respondent Selection Protocols Shape Dietary Diversity And Quality Measurement In Surveys

This study examines the effects of different respondent selection approaches on the measurement of dietary diversity and quality in household surveys. It uses recently implemented large-scale, nationally representative surveys from Ethiopia and Nigeria that included the Global Diet Quality Questionnaire (DQQ). The study evaluates whether interviewing one randomly chosen adult per household provides results as reliable as interviewing all adults, given the cost and logistical challenges of collecting data from every household member. The analysis considers alternative methods for respondent selection, such as random selection with and without replacement, the use of proxy respondents, and interviewing the household head. Findings show that proxy and non-responses systematically differ from self-responses, introducing bias into the estimates. Similarly, household heads differ from the broader adult population in several key demographic and socioeconomic variables. The study highlights that, when appropriately calibrated and executed in the field, random respondent selection with replacement can produce robust indicators of dietary diversity and quality. These findings offer valuable guidance to national statistical offices and survey practitioners seeking cost-effective yet reliable approaches to individual-level dietary data collection in low- and middle-income countries.