Adjusting for Confounding with Text Matching∗

We identify situations in which conditioning on text can address confounding in observational studies. We argue that a matching approach is particularly well-suited to this task, but existing matching methods are ill-equipped to handle high-dimensional text data. Our proposed solution is to estimate a low-dimensional summary of the text covariates and condition on this summary via matching. We propose several specific methods for doing so and weigh their merits. We illustrate the importance of conditioning on text to address confounding with three applications: the effects of censorship on Chinese bloggers, the effect of perceptions of author gender on citation counts in academia, and the effect of Usama bin Laden’s death on the popularity of his writings. ∗We thank the following for helpful comments and suggestions on this work: David Blei, Naoki Egami, Chris Felton, James Fowler, Justin Grimmer, Erin Hartman, Chad Hazlett, Seth Hill, Kosuke Imai, Rebecca Johnson, Gary King, Adeline Lo, Will Lowe, Chris Lucas, Ian Lundberg, Walter Mebane, David Mimno, Jennifer Pan, Marc Ratkovick, Matt Salganik, Caroline Tolbert, Simone Zhang, audiences at the Princeton Text Analysis Workshop, Princeton Politics Methods Workshop, Microsoft Research, Text as Data Conference, the Political Methodology Society and the Visions in Methodology conference, and some tremendously helpful anonymous reviewers. We especially thank Dustin Tingley for numerous insightful conversations on the connections between STM and causal inference. Dan Maliniak, Ryan Powers, and Barbara Walter graciously supplied data and replication code for the gender and citations study. Crimson Hexagon provided data for the study of blogs in China. The JSTOR Data for Research program provided academic journal data for the international relations application. This research supported, in part, by The Eunice Kennedy Shriver National Institute of Child Health & Human Development under grant P2-CHD047879 to the Office of Population Research at Princeton University. The research was also supported by grants from the National Science Foundation RIDIR program, award numbers 1738411 and 1738288. †Assistant Professor, Department of Political Science, University of California, San Diego, Social Sciences Building 301, 9500 Gilman Drive, #0521, La Jolla, CA 92093-0521, 360-921-3540, [email protected], MargaretRoberts.net ‡Assistant Professor, Department of Sociology, Princeton University, 149 Wallace Hall, Princeton, NJ 08544, 757-636-0956, [email protected], brandonstewart.org §Associate Professor, Department of Political Science, Massachusetts Institute for Technology, 77 Massachusetts Avenue, E53 Room 455, Cambridge, MA, 02139, 617-254-4823, [email protected], http://www.mit.edu/ rnielsen/research.htm ¶The first draft of this paper was circulated in the summer of 2015 under the title “Matching Methods for High-Dimensional Data with Applications to Text”