In most parliaments, registration of how each member cast her vote is not the norm. Instead of this roll call voting, as it is called, each proposal is decided by members indicating their position by either a show of hands, getting up from their seat, or by shouting ?yea? or ?nay? when asked. It is then in the hands of the president of the parliament to decide the outcome. However, it is often possible for a group of legislators to request the recording of how each and every legislator voted. This act is commonly known as a roll call request. However, in parliaments whose rules of procedures open for this possibility, there is large variation in both the proportion of roll call requests as well as the actual number of such requests.
In StREP, we seek to uncover the strategic game behind the requests of roll call votes. While we seek a understanding of the phenomena in general, the initial empirical focus is on the EP. The reason for this is twofold. First, the EP is an important legislator in its own right, as EU legislation is not only affecting the lives of more than the 500 million people of the EU member states. A substantive proportion of the votes are not taken by roll call. Although existing research has pointed towards possible strategic reasons for requesting roll calls on some votes but not others, these strategic accounts have not been subjected to systematic statistical testing in any large extent.
We have developed theoretical based statistical estimators suitable for addressing this question. We are in the final stages of developing an R-package with a common user-interface for these models, and are finalizing testing under which conditions our estimators perform better than existing models. We are collecting data that enables systematic testing of strategic models of roll call requests. We have obtained clearance to collect these data from NSD, Norwegian Center for Research Data. These results will subsequently lay the foundation for the second part of the project. Here, we incorporate the insight regarding roll call requests when developing and testing new models of voting behavior, coalition formation and voting outcomes. In the third part of the project, we will explore applications of these models and methods to settings beyond the EP.
What are the effects of roll call requests on voting behaviour, coalition formation and voting outcomes in the European Parliament (EP)? The standard approach for addressing such strategic questions empirically in the context of EU legislative politics is to derive comparative statics given an appropriate solution concept such as Nash equilibrium and subject these implications to empirical tests . In contrast, we rely upon Quantal Response Equilibrium (QRE), a solution concept that allows for the possibility that actors may make mistakes. We apply this logic to roll-call vote requests in the EP to investigate why political groups make roll call requests, and how, and to what extent they take into account their expectations of what other political groups will do when deciding whether or not to request a roll call on a particular vote.
Having established why political groups make roll call requests, the project turns the investigation to the effect of roll call requests on voting behavior. We will extend the standard item-response model by adding more co-variates, possible in a hierarchical setting as votes are nested in proposals. Furthermore, as we have data on all votes, we will treat these as non-ignorable missing data and add a selection stage to the models in order to both account for missing votes on individual legislators as well as predict voting participation and behavior of all members in electronic votes where information about the vote split, but not how individual members votes, is available. This will in turn allow for assessing the range of the possible magnitude of the effect of roll call requests that is consistent with the data under varying theoretical assumptions, thereby answering the second secondary research question of what the effect of roll call request is on voting behavior.