Foreign-Sponsorship Effects in Developing-World Surveys: Evidence from a Field Experiment in Lebanon

Abstract

This research note assesses how foreign sponsorship of surveys in the developing world affects participation rates, descriptive representation, and nonresponse bias. It analyzes a survey experiment conducted in Lebanon that varies its putative sponsor to include foreign universities and governments. It finds that sponsorship by foreign governments increases refusal rates, but sponsorship by universities does not; in addition, subjects assigned to the government conditions drop out systematically. University samples provide unbiased estimates of public opinion, but government samples misread the population as more Western friendly than it actually is. Weighting observations to account for compositional imbalances between conditions reduces but does not eliminate this bias.