Decomposing Response Errors in Food Consumption Measurement : Implications for Survey Design from a Survey Experiment in Tanzania

Friedman, Jed ; Beegle, Kathleen ; De Weerdt, Joachim ; Gibson, John

World Bank, Washington, DC

关键词: LIVING STANDARDS;    PER CAPITA CONSUMPTION;    HOUSEHOLD SIZE;    HOUSEHOLD SURVEY;    DURABLE GOODS;   

DOI  :  10.1596/1813-9450-7646
RP-ID  :  WPS7646

学科分类：社会科学、人文和艺术（综合）

来源: World Bank Open Knowledge Repository

【 摘 要 】

There is wide variation in howconsumption is measured in household surveys both acrosscountries and over time. This variation may confound welfarecomparisons in part because these alternative survey designsproduce consumption estimates that are differentiallyinfluenced by contrasting types of survey response error.Although previous studies have documented the extent of neterror in alternative survey designs, little is known aboutthe relative influence of the different response errors thatunderpin a survey estimate. This study leverages a recentrandomized food consumption survey experiment in Tanzania toshed light on the relative influence of these various errortypes. The observed deviation of measured householdconsumption from a benchmark is decomposed intoitem-specific consumption incidence and consumption value soas to investigate effects related to (a) the omission of anyconsumption and then (b) the error in value reportingconditional on positive consumption. The results show thatvarious survey designs exhibit widely differing errordecompositions, and hence a simple summary comparison of thetotal recorded consumption across surveys will obscurespecific error patterns and inhibit the lessons for improvedconsumption survey design. In light of these findings, therelative performance of common survey designs is discussed,and design lessons are drawn to enhance the accuracy ofitem-specific consumption reporting and, consequently, themeasures of total household food consumption.