Are Voter Rolls Suitable Sampling Frames for Household Surveys? Evidence from India

Abstract

We examine the use of publicly available voter rolls for household survey sampling as an alternative to household listings or field-based sampling methods. Using voter rolls for sampling can save most of the cost of constructing a sampling frame relative to a household listing, but there is limited evidence about their accuracy and completeness. We conducted a household listing in 13 polling stations in India comprising 2,416 households across four states and compared the listing to voter rolls for the same polling stations. We show that voter rolls include 91% of the households found in the household listing. We conduct simulations to show that sampling from voter rolls produces estimates of household-level economic variables with minimal bias. These results suggest that voter rolls may be suitable for constructing household sampling frames, particularly in rural India.

Sampling and Categorization of Households for Research in Urban India

“…For example, in Delhi (Janaagraha, 2015), 49% of citizens spoken to on the field were omitted from the voter list from the Polling Part where they were living, while 23% of citizens on the voter list were found to have shifted from their listed address, were disenfranchised, deceased or listed more than once. Likewise, an ID Insight working paper by Joshi et al (2020) notes that voter lists may be more suitable as sampling frames in rural areas than urban ones. The informal nature of many settlements in urban India, as well as rapid urbanization and greater transience of citizens, may be contributing to this.…”