This article provides a comprehensive analysis of Eurosceptic contestation within the legislative arena of the European Parliament (EP) from 2009 to 2019. Under what conditions do Eurosceptics vote differently from their Europhile peers? The literatures on European integration, party competition and policy types lead us to expect variation in Eurosceptic contestation across policy areas. Drawing on roll-call votes in the EP, we introduce two new measures of such contestation: Eurosceptic dissent, that is, the extent to which Eurosceptics diverge from the Europhile plurality, and integration polarization, that is, the extent to which Eurosceptics and Europhiles oppose each other as cohesive camps. Our two indicators show that Eurosceptic contestation is particularly pronounced when the EP votes on cultural, distributive and constituent issues. When voting on redistributive policies, in contrast, dissent and polarization are curbed by national and ideological diversity.
Improved multilevel regression with post-stratification through machine learning (autoMrP)
with Lucas Leemann
and Reto Wüest 2022 in The Journal of PoliticsarticleR-package
Multilevel regression with post-stratification (MrP) has quickly become the gold standard for small area estimation. While the first MrP models did not include context-level information, current applications almost always make use of such data. When using MrP, researchers are faced with three problems: how to select features, how to specify the functional form, and how to regularize the model parameters. These problems are especially important with regard to features included at the context level. We propose a systematic approach to estimating MrP models that addresses these issues by employing a number of machine learning techniques. We illustrate our approach based on 89 items from public opinion surveys in the US and demonstrate that our approach outperforms a standard MrP model, in which the choice of context-level variables has been informed by a rich tradition of public opinion research.
Who gains legislative influence in early agreement negotiations (trilogues) between the European Parliament and the Council of the European Union? Practitioners from both institutions suspect that it is the other side. Meanwhile, critics point at trilogues' lack of transparency. This article proposes that legislative power and institutional transparency are inversely related: Opacity makes an actor more influential in political negotiations. The argument is tested on a matched sample of legislative files from the 1999–2009 period. The findings suggest that the European Parliament became more influential in early agreement negotiations — where it became opaque vis-à-vis the Council. In such negotiations, the relative influence of the European Parliament substantially increased; by contrast, the European Parliament did not gain influence in negotiations where it remained transparent.
Data innovation for international development: An overview of natural language processing for qualitative data analysis
2017 in IEEE Proceedings of the International Conference on the Frontiers and Advances in Data Science (FADS)
with Anna Hanchararticle
Availability, collection and access to quantitative data, as well as its limitations, often make qualitative data the resource upon which development programs heavily rely. Both traditional interview data and social media analysis can provide rich contextual information and are essential for research, appraisal, monitoring and evaluation. These data may be difficult to process and analyze both systematically and at scale. This, in turn, limits the ability of timely data driven decision-making which is essential in fast evolving complex social systems. In this paper, we discuss the potential of using natural language processing to systematize analysis of qualitative data, and to inform quick decision-making in the development context. We illustrate this with interview data generated in a format of micro-narratives for the UNDP Fragments of Impact project.
Citizens who support a party which enters government are systematically more satisfied with democracy compared to voters who supported a party which ends up in the opposition. This relationship is labelled as the “winner-loser gap,” but we lack firm causal evidence of this gap. We provide a causal estimate of the effects of voting for a winning or losing party by leveraging data from surveys fielded before and after new government formations were announced in contexts of very high uncertainty. Using a regression discontinuity design comparing citizens' levels of satisfaction with democracy just before and just after their electoral status (winner or loser) was revealed, we find that the impact of winning or losing is indistinguishable from zero. We conclude by discussing the implications of our findings.
Regularized regression in Multilevel Regression and Poststratification: A Practical Guide and New Developments, ed. Rohan Alexander, Andrew Gelman and Lauren Kennedy.
with Lucas Leemann
and Reto Wüest conditionally accepted at Cambridge University Press.
Revise and Resubmit
What Unites the Right in the European Parliament
with Bjørn Høyland invited to revise and resubmit
This letter analyzes coalition patterns among right-wing groups in the
European Parliament (EP), exploring conditions that lead these groups to
unite on policy issues. Using voting records from the 9th EP (2019-2024)
and topic data from the Legislative Observatory, we identify three
coalition types: the Grand Right (all right-leaning groups), the
Far Right (excluding the center-right European People's Party,
EPP), and the Center Right+ (EPP plus at least one far-right
group). Right-wing unity is common and right-wing potential substantial,
especially on EU Integration, climate, security, and cultural issues.
For 2024-2029, the EPP faces a strategic choice: ally with far-right
groups on cultural, climate, and security topics, challenging the
EP's progressive stance, or unite the far-right on economic issues,
normalizing right-wing influence in EU policy.
If all votes were recorded
with Bjørn Høyland invited to revise and resubmit
Voting records are essential for the study of legislative behavior, but recording of individual-level vote choices is often partial. While roll calls are taken by default on some votes, they need to be requested on others. This partial recording causes a selection problem. We rely on an external shock to the procedure of recording votes which required, for a time, roll calls on votes that would otherwise only be recorded upon request. Using a difference in differences design, we compare roll calls that were always recorded with roll calls that needed to be requested under the regular regime but were recorded during a temporary remote voting regime. Our results show that the requirement to request roll calls makes parliaments appear more competitive and legislators more loyal. Contrary to recent contributions, our findings demonstrate bias in two basic quantities of interest: vote splits and legislator loyalty scores. One reason for this is that when all votes were recorded, political group leaderships are likely to instruct their rank and file to vote a particular way on the votes that were recorded under the remote voting regime but would not have seen a request under the regular regime. Our findings are thus evidence for the use of roll call requests as a disciplining tool.
Legislation against the odds: Overcoming ideological gridlock in EU decision-making
with Lukas Obholzer
and Christine Reh invited to revise and resubmit
Negotiations in bicameral settings face the risk of gridlock when veto players' preferences diverge on the ideological (left-right) dimension. In the European Union, a high proportion of legislation appears to be gridlocked yet is nevertheless adopted. We argue that ideological gridlock is resolved through “generous concessions.” These concessions are made to uphold supranational cooperation, to capitalize on log-rolls, and under “permissive consensus” at the domestic level. We draw on novel data of amendments and repeals of existing EU legislation (1999–2014) and demonstrate that homogeneous elite-level preferences on continued cooperation and the possibility for issue-linkage help resolve gridlock. However, the potential for compromise has decreased substantially with rising levels of Euroskepticism. We contribute to established debates about legislative bargaining and demonstrate that polarized publics constrain elites' ability to reach compromise, diminishing polities' problem-solving capacity.
Working Papers
What does European Union Politics Look Like: Ideology, Values, Both?
with Bjørn Høyland
How has the political space in European Union politics evolved and has left-right politics been replaced by a value based pro-EU - anti-system dimension? In this paper, we address this question. Based on rollcall data for all European Parliaments from 1979 to 2024, we estimate dynamic ideal points in six months intervals for all Members of the European Parliament (MEPs). We compare our method to the standard procedure where ideal points are estimated in two dimensions, separately for each legislative term and, hence, not comparable over time. Contrary to conventional wisdom, we demonstrate that over 90 percent of variation in voting behavior is explained by one dominant ideological dimension – a trend that persists even with an influx of Euroskeptic MEPs, the politicization of EU politics and throughout periods affected by Brexit and the Covid pandemic. Furthermore, we compare our results to votes from the U.S. House of Representatives for the same time period and demonstrate that the first dimension explains a similar amount of variation. We contribute to the literature on the dimensionality of politics in the EU and in addition, we provide the longest time-series of European Parliament legislator ideal points and uncertainty estimates. Contrary to the standard procedure, our ideal points are comparable over time and, therefore, more useful for scholars of legislative politics.
A Comparison of Deep Hierarchical Models and Machine Learning for Multilevel Regression with Poststratification (autoMrP)
with Lucas Leemann
, Florian Schaffner
, and Reto Wüest
In this paper, we present a comprehensive evaluation of modeling approaches for small-area estimation in political science, focusing on Multilevel Regression and Poststratification(MrP) models and recent extensions. Leveraging a dataset compiled by Buttice and Highton (2013), which comprises responses to 89 survey items from at least 25,000 participants,we systematically assess the predictive performance of various models under different conditions. Our analysis considers the impact on predictive accuracy of model complexity, sample size, and the number of context-level features. Contrary to recent claims, we find that while deep interaction models may outperform machine learning counterparts under specific circumstances, machine learning ensembles, particularly autoMrP and SRP, generally exhibit superior performance across diverse scenarios. We acknowledge that deep hierarchical models can enhance existing machine learning ensembles and show under which circumstances this is the case. We also make this new variant of an MrP machine learning model available through CRAN. Overall, our study offers valuable insights into model selection for small-area estimation tasks that can guide researchers working on specific applications.
Patterns of roll-call vote requests in the European Parliament
with Bjørn Høyland
Studies of roll-call votes provide us with a unique insight into individual-level legislative behavior in EU politics. However, critiques warn against generalizing from roll-calls because they are requested strategically, for example, to signal a policy position or to enforce party group cohesion. What are the patterns of roll-call vote requests and do they suggest strategic behavior? To address this question, we leverage newly collected data on all roll-call vote requests in the European Parliament for the last two-and-a-half legislatures, from 2009 until today. We sketch the patterns of roll-call requests over time, across procedures, policy areas, and actors to evaluate the degree of (un)representativeness of roll-call votes along these dimensions. We contribute to a rich literature on legislative behavior in the European Union that is based on roll-call voting and that assumes that roll-calls are a representative sample of all voting in the European Parliament. In addition, we add to the scholarly debate on the representativeness of roll-call votes with a more comprehensive large-scale study than has been carried out on EP legislation to date.
The evolution of eurosceptic contestation in the European Parliament
with Lukas Obholzer
We provide a comprehensive analysis of change in eurosceptic contestation in the European Parliament from 1994 to 2019. Using novel preference data, we test whether increased euroscepticism is driven by replacement or preference change. Replacement takes place when Members of the European Parliament are replaced by more eurosceptic representatives. In contrast, when preferences of re-elected representatives become more eurosceptic over time, we refer to within-individual preference change. We disaggregate by policy and furthermore, we test whether more “extrme” members of eurosceptic parties are more likely to be re-elected in subsequent elections. Methodologically, we employ a Bayesian IRT model to estimate preferences that are comparable over legislative periods. Re-elected representatives' priors are set to the estimate from the previous term. This approach allows comparison over time and provides a measure of preference change: The difference between prior and posterior. We contribute to the growing literature on euroscepticism in the European Union by providing insights into the question whether increased scepticism takes place “top-down” where legislators become more eurosceptic or “bottom-up” where representatives are replaced by more eurosceptic legislators.
Substantive representation of euroscepticism: Contesting European integration within the European Parliament
with Lukas Obholzer
While some eurosceptic members of the European Parliament (MEPs) engage constructively with legislative work, others opt for blanket opposition. This is reflective of distinct legislative roles of MEPs and results in differences in the representation of eurosceptic voters. We ask whether and under what conditions eurosceptics' behaviour within the European Parliament is shaped by the potential for public attention. Combining novel data on voting behaviour in committee-level as well as in plenary-level roll calls, we leverage the difference in public attention that the two arenas attract. Our analysis draws on variation in attendance and voting behaviour patterns across levels to test competing explanations of how and to what extent eurosceptic legislators contest European integration when in the “public eye” in plenary as opposed to more secluded everyday work in committees. The findings demonstrate that MEPs are mindful of the distinct principals and audiences to whom they pander. The paper maps the behaviour and contributes to a growing literature on legislative behaviour of eurosceptics.
