To interpret elections, social scientists and media pundits often ask: how much did particular groups, or voting blocs, contribute to a candidate’s vote total? The default tool for studying voter behavior — regressions of vote choice on voter characteristics — is useful for evaluating correlates and determinants of vote choice, but is incapable of assessing how many votes a group contributes. Accounting for votes also requires knowing a group’s size and its turnout rate. We introduce a set of tools for estimating how many votes a bloc gives to a candidate, how voting bloc patterns differ from prior elections, and how support changes under counterfactuals. We apply these tools to study U.S. presidential elections, demonstrating that there is little evidence to show Black and Latino voters are shifting toward Republicans in recent elections and that Donald Trump’s support was concentrated among voters with moderate attitudes toward racial outgroups.