Germans are getting ready to vote on 26 September. With Chancellor Angela Merkel stepping down and headlines dominated by climate change, covid-19 and migration, these elections are shaping up to be unlike any we have seen in recent years.
Public debate will be more dependent on online platforms than ever before. While they enable a more vibrant and diverse debate, social media also hold the potential for misuse.
We will be keeping a close eye on this in the lead up to the election, providing regular analysis and special reports as the campaign progresses. We are also partnering with Berlin daily Tagesspiegel to showcase online trends and patterns.
Our aim is to ask the right questions and provide readers with the information and insights they need to ensure that the 2021 German election is open and fair.
What have we done so far?
We are working to capture the broad election discourse amongst key actors and the public.
Our team is collecting posts mentioning #BTW21 and #Bundestagswahl (the hashtags most often used to refer to the 2021 elections) in various formats, posts by top candidate accounts, posts by top political party accounts at the federal and state (Länder) levels, and posts about the candidates and parties themselves. We will continue to monitor more areas as the political landscape and election evolve.
What can you expect?
Our data-driven analyses will consider the election discourse, and its discontents, along several lines:
- The tone of the discourse surrounding certain candidates and issues, up to and including whether hate speech is present in this discourse, particularly targeting female candidates.
- The presence of disinformation or manipulated media and how these spread naturally or through coordinated inauthentic networks (i.e. ‘bots’) … or how they die out.
- How the agendas of the political parties and candidates make it into the general election debate, or conversely how these actors take elements from the general election discourse and incorporate them into their own agenda – even if that agenda was originally driven by another party.
DRI will present this information in deep-dive articles as well as in quickly digestible infographics. We will cover general trends, such as the overall social media activity of parties, politicians and influencers. We will also zoom in on issues of the moment, such as the birth and growth of a hashtag, new ingenuine content, or the success of a party in capturing our attention.
How are we doing it?
At the DRI and Tagesspiegel offices in Berlin, our team is examining large amounts of social media data. You, as the reader, will only see the final product of these efforts – colourful graphs and our summary analysis.
Here is a peek at our process, which follows three key steps: data collection, data processing, and data analysis.
a) The first step of data collection is to decide what to search for. Selecting a bad search criterion can lead you to miss important things or see entirely different conversations. Our search criteria starts by looking at the general election hashtags and main candidates and parties. From these data, we then see what the citizens and the politicians of Germany are talking about and use these insights to expand our search over time. For social media data collection, we use first most the official APIs (Application Programming Interfaces) provided by the platforms and, secondly, when necessary and permitted by the platforms' terms of service, publicly available open-source programmes (e.g. Python libraries, GitHub repositories).
b) Between collection and analysis lies the unloved task of data processing. Raw data is of little value, and indeed the release of raw, context-free data is a tactic of those wishing to obscure information and rarely increases public knowledge and transparency. Therefore, data must be first cleaned and standardized to ensure we are not comparing apples to oranges.
c) For our analysis, we will draw on statistical and machine learning software packages to identify patterns we cannot observe naturally and put them in their political context, while still keeping our findings explainable. Python is the programming language that unites most members of the team and it and its associated libraries, especially for natural language processing, are our core tools.
A more elaborate description of our methodology can be found on the DRI website soon. See our Digital Democracy Monitor Toolkit to learn more about our approach more generally.
Election Monitor – Germany 2021 is implemented by Democracy Reporting International in collaboration with Tagesspiegel with funding by Stiftung Mercator.