I’m a doctoral candidate in computational social science at the Graduate School of Decision Sciences and a research associate at the Center for Data and Methods at the University of Konstanz in Germany.
In my dissertation, I analyze social media users' consumption awareness. In particular, I focus on the accuracy of Twitter users' statements about their online activity and the political heterogeneity of their Twitter network. Further, I analyze the normative perceptions of biased news diets. My work utilizes online surveys in addition to analyses of social media data. My dissertation is chaired by Karsten Donnay, Peter Selb and Pablo Barberá. I received both my BA (Major in Sociology, Minor in Statistics) and my MSc (Social and Economic Data Analysis) from the University of Konstanz.
I mainly use Python to retrieve the data used in my dissertation, while I employ SQL-databases accessed via MariaDB to maintain the huge datasets social media data might generate. In order to finally make sense of the data, I utilize R to employ models and visualize my results. Additionally, I use the EFS software to create surveys and experiments, which I complemented in one project with an additional website programmed in Python Flask. To guarantee everything runs smoothly, I utilize the remote Linux servers hosted by BWScope and the University of Konstanz. As I work with sensible data, I also became a semi-expert on data privacy and the regulations of the EU-GDPR.
In my spare time, I enjoy playing the guitar, painting, dancing Lindy Hop and other social dances, and spending time outside. As I live in Konstanz, the mountains are close and offer plenty of opportunities to go skiing or hiking.
My dissertation consists of three main projects and several side projects. Below I outline the scope and the purpose of several projects.
In this project, I utilize the results of a survey conducted by YouGov and provided by Pablo Barberá and colleagues in order to evaluate Twitter user's awareness of the political alignment of their Twitter friends. I complimented the survey data with data retrieved from the Twitter API in an effort that included the retrieval of Twitter profile information of 246.693 Twitter users. We find that Twitter users' perceptions are biased by a False Consensus Effect and that political knowledge helps identify the most frequent political group among one's Twitter friends. This project is the first to analyze the perceptions about the political constitution of one's Twitter friends and employs a novel approach by utilizing the procedure of Pablo Barberá's Tweetscores.
In this project, I employed a survey via several platforms (Respondi, JobBoy, Twitter Ads) in order to compare the quality of the self-reported Twitter activity. I compared survey responses to questions regarding one's Twitter activity with observed activity. Results suggest that people generally misreport their own activity and these results have implications for the treatment of survey responses in general. Additionally, I find a significant impact of intrinsic motivation for the quality of survey responses and I discuss the potential remedies to this problem of misreporting.
In this project, I analyze the ability to detect political slant in Twitter networks and the normative evaluations of these slants by distributing a vignette experiment to Amazon Mechanical Turk workers.
In a joint project with Marius Sältzer from the University of Mannheim I am working towards a classifier that enables the classification of German Twitter users' political ideology. The work includes the analysis of more than 8 Mio Followers of German political elites and news outlets and encompasses a methodology that is similar to the procedure of Pablo Barberá's Tweetscores.
Below you can find the most recent version of my CV.
Phone: +49 7531 88 3225