Skip to main content
Photo of Dr Laszlo Horvath

Dr Laszlo Horvath

Research Fellow


01392 725952


My research is in the areas of political communication, media, and gender, using quantitative and computational methods.

I am based in the Exeter Q-Step Centre, between 2014 and 2018 as a PhD student and later as an Associate Lecturer. 

Since September 2018, I am a Postdoctoral Research Associate on an ESRC project investigating media narratives around Brexit alongside a team of ethnographers, to understand how citizens interact with traditional and social media in practice.

Research group links

Research interests

My PhD thesis explored an empowering or 'role model effect' linked to the increasing campaign and media presence of female politicians. The project brought together evidence from an eye tracking study of attentional bias to candidate gender, a survey experiment with experimentally manipulated gender balance as picture stimuli, and observational data from the British Election Study. Across all these approaches, I found that improving gender balance in politics increased citizens' political engagement especially female voters' self-confidence and political efficacy. My paper reporting some of these effects is now undergoing revisions before publication.
I am currently writing up the results of a new project on how political discussion emerges in online communities from nonpolitical chat. This is an inquiry into the language and meaning of politics looking at communities whose voices are not always represented in the public discourse, analysing text from, a parenting website, The Student Room, the most popular student forum, and Empty Closets, an LGBT forum. Using machine learning to tease out political content from informal chat, we classified up to a quarter of all comments across these forums as 'political', and found important differences in the dynamics of politicisation along topics including housing, crime, and trans rights on Mumsnet, and ethnicity, the environment, and veganism on The Student Room.
I am very excited about taking care of Unbias (warning: testing version), which is a piece of civic technology I am co-developing with researchers at the University of Exeter to flag the sexist coverage of female politicians in text, and highlight these in a smartphone application. The back-end of this software takes human coded examples of sexist attributes, first-person accounts of sexism from the blog Everydaysexism, and eventually user-input to generate predictions of sexist coverage. Here is my presentation of the connected research paper presented at the final dissemination event of ExpoNet (ESRC) at the Alan Turing Institute for Data Science.

I am currently a Postdoctoral Research Associate on an ESRC project covering media narratives around Brexit. This includes data collection efforts to get a full (and ongoing) coverage of Brexit and EU affairs in local and national media. One of our goals is to investigate the presence and dynamics of the media narratives around sovereignty, immigration, economy, generations, and history including 'empire nostalgia' in this corpus, using novel feature selection methods for classification, looking at the properties of casual language.

Research from my Industry and Doctoral Training Network (Marie Curie) has led to output on how to best use large-scale data sets from online Voting Advice Applications (VAA)—online information tools matching parties to voters on the basis of issue positions—for social research. This includes a paper on lessons learned from VAA usage during the United Kingdom European Union membership referendum, and current working papers using cross-country VAA data.

 Edit profile