New Approaches to Estimating Immigrant Documentation Status in Survey Data

PWP-CCPR-2020-003

Abstract

Approximately a quarter of the 43 million immigrants living in the United States are thought to be undocumented. Yet, the lack of accurate population-level information about undocumented immigrants provides fertile ground for public misconceptions, political and media hype, and false claims. The goal is to determine how well descriptions of the undocumented population are likely to mirror the reality of undocumented immigrants’ lives in the US. We compare (1) the distribution of the population by documentation status and (2) distributions of the characteristics of undocumented and documented immigrants produced by two methods. The first method (the “decomposition method”) is a commonly used strategy used in previous work and the second method is an alternative, independent method developed in this article. We used the Survey of Income and Program Participation (SIPP) and the Los Angeles Family and Neighborhood Survey (LAFANS). The existing decomposition method works reasonably well if the data contains information on whether respondents are naturalized citizens or and lawful permanent residents. However, when these variables are missing or problematic, the decomposition method produces biased results. The actual undocumented population in the US may be even more socioeconomically disadvantaged than studies based on existing decomposition methods indicate. This article evaluates methods to conduct reasonably accurate nationally representative, policy relevant research on the lives of undocumented immigrants without potentially jeopardizing members of this vulnerable population.