Simon Hegelich

University of Siegen

Siegen, NRW
57068 |  Visit Personal Website

Search Google Scholar
Search for Google Scholar Profile

The focus of my research lies on political data science. I use machine learning to find deeper insights in theories of the policy process, especially punctuated equilibrium theory. In addition, I try to enhance computer science based approaches with concepts from political science about causality.

The Communication Behavior of German MPs on Twitter: Preaching to the Converted and Attacking Opponents (with Morteza Shahrezaye), in: European Policy Analysis (EPA), 1/2, 155-174.
Abstract: What effect does the communication of politicians on Twitter have? Is it reinforcing existent ideologies because users get messages of politicians mostly from their own ideological cluster? Or is Twitter exposing the users to cross ideological content as well? We argue that both is the case. We show that politicians use the different communication channels Twitter provides to distinguish between communication within their own ideological cluster in order to organize support and across these clusters to argue against their opponents. Considering German general elections as case study, we present empirical tests that politicians – more than other politically interested users – use Twitter mostly to provide information but with significant differences between parties. We furthermore show that politicians use the whole spectrum of communication channels provided by Twitter. Finally, there is empirical evidence of different qualities of the communicated content: Measured by sentiment analysis the communication with members of the same party is less harsh than the communication with political rivals. These two effects of communication on Twitter together might lead to stronger polarization in political discourses
Point Predictions and the Punctuated Equilibrium Theory: A Data Mining Approach (with Cornelia Fraune and David Knollmann), in: Policy Studies Journal (PSJ), 43/2, 228-256.
Abstract: In Punctuated Equilibrium Theory (PET), information processing under the constraints of limited attention and bounded rationality leads to stick-slip dynamics in policy outcomes. Empirical work in this field often focuses on the macro level. Using the case of nuclear energy policy in the United States as proof of concept, we demonstrate how decisive budget changes in a specific policy subsystem can be linked to attention of Congress and the president. We utilize a mixed-methods data-mining approach: Maximum likelihood estimation is used to analyze the distribution of the nuclear energy RD&D budget. Then attention data of both Congress and the president are structured by means of cluster analysis and principal component analysis. Finally, these data are used in a generalized linear model to predict specific budget shifts. The article is designed as a proof of concept: In the case of nuclear energy policy, we are able to predict budget shifts without violating the assumptions of PET. More importantly: we can demonstrate that attention is not only affecting the final policy outcome but also the corridor of the possible.

Substantive Focus:
Science and Technology Policy PRIMARY
Social Policy SECONDARY

Theoretical Focus:
Policy Process Theory PRIMARY