Overall Research Project Description This project aims to understand how political scientists typically formulate hypotheses and test null hypotheses using panel data (i.e., multiple units observed over multiple time periods). Panel data are frequently used to test theories in a variety of sub-fields of political science including international relations, comparative politics, American politics, and public policy. Recent advances in political methodology suggest that statistical models such as linear regression and fixed effects regression estimate different quantities of interest than what social scientists believed for many years. This project seeks to understand how political scientists typically formulate hypotheses in panel data settings to understand the theoretical comparisons that political scientists view as reasonable. The project also aims to compare these hypotheses to the null hypothesis tests that are actually implemented in quantitative analysis. To accomplish these goals, we will collect data on hypotheses and empirical tests that use panel data and appear in the top journals in Political Science (APSR, AJPS, and JOP), in addition to sub-field journals such as International Organization, Comparitive Politics, Journal of Conflict Resolution, and more. This data will be used to quantify how political scientists formulate hypotheses and the extent to which their empirical approaches match theoretically relevant hypotheses, in an attempt to understand how large the gap between theory and empirical tests is in practice. This project aims to produce a publicly available data set, in addition to academic output (e.g., peer-reviewed publications), and public-facing output.