Genetic Instrumental Variable (GIV) regression: Explaining socioeconomic and health outcomes in non-experimental data

  • Thomas A. DiPrete, Columbia University
  • An IDIES Bi-Monthly Seminar
  • When: March 28, 2018, 16:00
  • Where: Arellano Theater, Levering Hall
    3400 N Charles Street
    Baltimore, MD 21218


Tom DiPrete

Thomas A. DiPrete is Giddings Professor of Sociology, co-director of the Institute for Social and Economic Research and Policy (ISERP), co-director of the Center for the Study of Wealth and Inequality at Columbia University, and a faculty member of the Columbia Population Research Center. DiPrete holds a B.S. degree from the Massachusetts Institute of Technology, and a Ph.D. from Columbia University. He has been on the faculty of the University of Chicago, Duke University, and the University of Wisconsin–Madison as well as Columbia. DiPrete’s research interests include social stratification, demography, education, economic sociology, and quantitative methodology. A specialist in comparative research, DiPrete has held research appointments at the Max Planck Institute for Human Development in Berlin, the Social Science Research Center – Berlin, the German Institute for Economic Research in Berlin, the VU University Amsterdam, the Netherlands Institute for Advanced Study in the Humanities and Social Sciences, and the University of Amsterdam. His recent and ongoing projects include the study of gender differences in educational performance, educational attainment, and fields of study, the determinants of college persistence and dropout in the U.S., a comparative study of how educational expansion and the structure of linkages between education and the labor market contribute to earnings inequality in several industrialized countries, and the study of how social comparison processes affect the compensation of corporate executives.


In non-experimental data, uncontrolled confounding variables are a source of bias in efforts to estimate the effect of an exposure on an outcome. Instrumental variable (IV) regression is a potential solution, but valid instruments are scarce. Existing literature proposes to use genes related to the exposure as instruments (i.e. Mendelian Randomization – MR). However, this approach is problematic due to possible pleiotropic effects of genes that can violate the assumptions of IV regression and undermine the ability of MR to correct for endogeneity bias from environmental sources as well. An alternative approach, GIV regression provides more accurate estimates for the causal effect of the exposure and gene-environment interactions involving the exposure under conditions of pleiotropy. As a valuable byproduct, GIV regression also provides accurate estimates of the chip heritability of the outcome variable. GIV regression uses polygenic scores (PGS) for the exposure and the outcome of interest, both of which can be constructed from genome-wide association study (GWAS) results. By splitting the GWAS sample for the outcome into non-overlapping subsamples, we obtain multiple indicators of the outcome PGS that can be used as instruments for each other, and, in combination with other methods such as sibling fixed effects, can address endogeneity bias from both pleiotropy and the environment. In two empirical applications, we demonstrate that our approach produces reasonable estimates of the chip heritability of educational attainment (EA) and demonstrates that both OLS and MR provide upwardly biased estimates of the causal effect body height on EA.

IDIES Bi-Monthly Seminar

View All Events