Migration Status Gradients in Immigrant Poverty: A Comparison of Imputation Methods

Area of Science:

• Sociology
• Demography
• Public Policy

Background:

• Research on migration status and its societal effects has grown significantly.
• Limited representative data and variable imputation methods have hindered accurate analysis.
• Existing methods for determining migration status often yield inconsistent estimates.

Purpose of the Study:

• To introduce a novel, data-driven indicator for migration status.
• To compare the accuracy of this new indicator against existing self-report and imputation methods.
• To assess the impact of different migration status measures on poverty gradient analysis.

Main Methods:

• Constructed a new migration status indicator by linking two federal surveys with the Social Security Administration's Numident file.
• Utilized administrative records for a comprehensive dataset of U.S. citizens and legal residents.
• Employed statistical models to predict poverty using the new indicator and compared results with self-reported and logic-based measures.

Main Results:

• The new administrative indicator produced poverty estimates comparable to self-reported migration status in federal surveys.
• Both administrative and survey-based measures showed divergent poverty gradients compared to logic-based imputation methods.
• Evidence suggests potential bias in logic-based algorithms for imputing migration status.

Conclusions:

• Administrative record linkages offer a promising approach for improving migration status research.
• The developed indicator provides a more reliable measure for analyzing socioeconomic disparities related to migration.
• Findings highlight the need for caution when using certain logic-based imputation methods for migration status.