Requested or Required: The Changing Composition of Roll Call Votes in the European Parliament

Roll call votes are frequently used to study the behavior of legislators. However, in most legislatures, the recording of votes is partial. Researchers' ability to generalize from roll call votes to all votes depends on the similarity between recorded and unrecorded votes. Focusing on the European Parliament (EP), we describe the pattern of the recording of votes from 2004 to 2019, highlighting the difference in recording of roll calls on part vs whole votes. Roll call on whole votes have become more prevalent due to rule change requiring roll calls on a growing share of final votes. In contrast, roll calls on part votes must still be requested. Comparing the effect of increasing the roll call prevalence across these different type of votes, we find no effect of roll call prevalence on final votes' vote margins, loyalty-scores, and unity scores. In contrast, for part votes, we do find that an increase in roll call prevalence correlates with changing vote margins, loyalty-scores and frequency of centain coalitions forming. As a result, differences in observed patterns of roll call voting over time may be due to different patterns in the recording of votes as well as behaviour changes. The former set of differences should be accounted for before concluding on the latter.