How Representative are Snowball Samples? Using the Ethnosurvey to Study Guatemala-U.S. Migration

This analysis draws on binational data from an ethnosurvey conducted in Guatemala and in the United States in Providence, Rhode Island, to develop a refinement of the weighting scheme that the Mexican Migration Project uses. The alternative weighting procedure distinguishes between temporary and settled migrants by using a question on household location in the Guatemala questionnaire that is not used in the MMP. Demographic characteristics and integration experiences of the most recent U.S. trip are used to assess the composition and representativeness of the U.S. sample. Using a composite index of migrant integration to compare the impact of alternative U.S. sample weights on point estimates, I find that although the U.S. sample is broadly representative across a range of background characteristics, the MMP sample weighting procedure biases estimates of migrant integration downward.