Is predicted data a viable alternative to real data?

• Main Authors: FUJII, Tomoki, VAN DER WEIDE, Roy
• Format: text
• Language: English
• Published: Institutional Knowledge at Singapore Management University 2016
• Subjects: Prediction; Double sampling; Survey costs; Poverty; Income Distribution; Public Economics
• Online Access: https://ink.library.smu.edu.sg/soe_research/2296; https://ink.library.smu.edu.sg/context/soe_research/article/3295/viewcontent/Is_Predicted_Data_a_Viable_Alternative.pdf
• Institution: Singapore Management University

It is costly to collect the household- andindividual-level data that underlies official estimates of poverty and health. Forthis reason, developing countries often do not have the budget to update their estimatesof poverty and health regularly, even though these estimates are most neededthere. One way to reduce the financial burden is to substitute some of the realdata with predicted data. An approach referred to as double sampling collectsthe expensive outcome variable for a sub-sample only while collecting thecovariates used for prediction for the full sample. The objective of this studyis to determine if this would indeed allow for realizing meaningful reductionsin financial costs while preserving statistical precision. The study does thisusing analytical calculations that allow for considering a wide range of parametervalues that are plausible to real applications. The benefits of using double samplingare found to be modest. There are circumstances for which the gains can be moresubstantial, but the study conjectures that these denote the exceptions ratherthan the rule. The recommendation is to rely on real data whenever there is aneed for new data, and use the prediction estimator to leverage existing data.