Measuring Congressional Support for the President: Evaluating NOMINATE Scores

Assessing presidential influence in Congress requires accurate measures of legislators' baseline preferences. Common strategies extract preferences from roll‐call votes. One approach selects votes most likely to reveal uncorrupted, sincere preferences, and the other, NOMINATE, derives summaries of members' positions from large samples of votes. This article finds that NOMINATE scores do not accurately recover the relative positions of members in simulated legislatures. Errors increase with the number of policy dimensions and do not depend on agenda control or on vote trading. However useful NOMINATE is for summarizing voting patterns, this article cautions against this approach to measuring legislators' preferences.