Measuring poverty in Tanzania: Comparison of diary and recall approaches to food consumption data collection

Abstract

Consumption data from household surveys continue to be the main source for poverty and inequality statistics in low- and middle-income countries. Although recent research has demonstrated that the choice of diary- versus recall-based methods for food consumption data collection can directly impact poverty measurement, the available evidence stems largely from small-scale, subnational survey experiments. This study uses data from a nationally representative randomized survey experiment in Tanzania to provide a comparative assessment of how household consumption and poverty measures may be impacted by relying on a 14-day food consumption diary versus two variants of a 7-day recall-based food consumption data collection. Both descriptive and regression analytical approaches were employed. The results reveal significant differences in food consumption expenditures across the diary and recall arms, and these differences result in some variations in total consumption expenditures as well. The results show further that while the diary method captures more diversity in food consumption items, the overall food consumption expenditure appears significantly lower than in the recall arms, even at different percentiles. Despite these disparities, the paper finds little statistically significant difference in poverty headcount between the diary and recall arms, even at different thresholds.