spotlight

Clint Claessen

Visiting Student from the Department of Social Sciences, University of Basel

Clint's innovative work applies machine learning methods to political science, to produce insights into party leadership, unity and party discipline

Professor Ken Benoit

As the institutional cornerstone of data science activity at the London School of Economics and Political Science, the Data Science Institute (DSI) is proud and privileged to host Visiting Students from other leading institutions within the wider data science community.

In Lent Term of 2022, Clint Claessen joins the DSI from the Department of Social Sciences at the University of Basel. This edition of the data science spotlight showcases Clint's research on the leaders of political parties.

Clint Claessen

Clint Claessen

Doctoral Researcher in the Department of Social Sciences at the University of Basel.

Website | Email


Party leaders are arguably the most visible politicians in parliamentary democracies. They make speeches in parliament, appear in the media, and function as the face of election campaigns. Voters usually have ready-made opinions on party leaders and can easily give an evaluation of leaders like Boris Johnson, Keir Starmer, Nicola Sturgeon and Ed Davey. It is, therefore, unsurprising that party leaders have a great influence on their party’s brand. My research is focused on understanding party leaders better: What careers did they have before their ascent to leadership? How does this political experience influence their party’s success?

To answer these questions, I collected information on the political careers of all the post-war party leaders from four countries: Canada, Germany, the Netherlands and Switzerland. Through a sequence analysis, I developed a one-dimensional measure of political career experience that encompasses political roles held in the executive, the legislature and the party. The question is then to what extent this accumulated experience influences a party leader’s success. In other words, is extensive political experience in multiple political arenas an indicator that a party leader is better able to manoeuvre and to remain in office? Subsequently, does that increased ability help the party leader to attain their party’s goals?

These party goals are threefold: win an election and form a government while ensuring party unity. The assumption is that if a party leader manages to attain these three, they will be more likely to remain in office. To use Boris Johnson’s survival as premier as an example, he has achieved the first two goals but now has the hard task of ensuring party unity amidst the Partygate scandal. This third variable – party unity – is hard to define precisely and doing so requires more extensive operationalisation.

Thus, I am currently working on a measure of distance between party leaders and their party’s MPs based on parliamentary speeches. One could expect that a party leader aims at maintaining party unity in their parliamentary speeches to keep their party ‘on brand’. If the topics and language used by MPs are very distinct from those of the party leader, this brand is potentially damaged in the eyes of voters. The assumption is that it is, therefore, pertinent for a party leader to minimise the distance between their own speeches and those of other MPs. To measure this, I employ machine learning methods. I have trained a model that predicts party labels using speeches made in Parliament. My current findings indicate that party leaders’ speeches are generally more easily classified as being from their own party than those of other MPs. In other words, party leaders’ words are probably more on brand. This suggests that this method has potential to become a meaningful measure of party unity.

Although I am currently developing this spatial measure in the German context, the broader aim is to use it comparatively. One could then assess whether party leaders with more political experience are generally better at keeping their party more united through speeches in the legislative assembly or elsewhere. Obviously, however, some political systems allow more independence in parliamentary speeches than others. For instance, German or Dutch MPs are not as free to speak as independently as British or Canadian MPs. Consequently, this distance measure has to be normalised per political system and, perhaps, even per party. Indeed, who could imagine a German MP saying to a chancellor from their own party: “In the name of God, go!”?