Looking at the German election through Google Trends — an interview with Moritz Stefaner
Elections have never been more digital. Over the past few months, elections across the world have demonstrated the importance of digital tools for voters to make informed decisions on Election Day.
This is important to us at Google. Our mission is to organize the world‘s information and make it universally accessible and useful — and elections are moments when access to accurate information matters most. At the News Lab, we look at these moments as opportunities to empower journalists with the technology and data they need to keep the communities they serve informed.
For the German election on September 24th, Google and YouTube launched a digital tool kit to help the public, policy makers and journalists inform themselves online, quickly and easily, and participate in political debates. Google Trends offers insight into the candidate or parties Google users in Germany are most interested in, and the moments that dominate the election campaign. Our new Google Trends election hub highlights this Search interest in top political issues and parties, with embeddable graphics that surface what German Google users are most interested in during the election campaign.
2Q17, a unique data visualisation created by the renowned data designer Moritz Stefaner, depicts queries users in Germany are searching for in relation to the top candidates. This project is part of Google News Lab’s series of visual experiments to develop innovative and interactive storytelling formats covering unique moments of the news cycle.
We asked Moritz about his different data visualisations, the magic behind search interest, why he thinks Google Trends is a powerful tool for storytelling, and lessons learned.
Moritz, how did the idea for 2Q17 on the German election evolve?
Google Trends is a fascinating data source, and I’ve had great fun working with it in the past (for instance, for mapping food trends). So when the Google News Lab reached out to me to think about ideas for a joint project around the German elections, I was immediately interested. How can we capture the search interest of people related to the parties, the elections, and candidates? What do people care about? How quickly does attention shift? These are all super-interesting questions we can now answer empirically. So, together with my collaborators Dominikus Baur and Christian Laesser, I explored a few of those questions and the results are now online at http://2Q17.de
Tell us about the Google Trends data you are visualising? What is unique about it and how do you select the search terms for each candidate?
At the beginning of each my projects, there’s a always a long phase of data and concept explorations. What are the most interesting perspectives, aggregations, juxtapositions, and relations we can find? And what is the perfect form for the data to allow us to experience those effortlessly?
We were immediately interested in the rich content and topic structure. So, beyond just quantifying party and candidate attention, or focusing just on a few key terms, we were really interested what the various deep topic landscapes are behind people’s interest in the candidates.
After trying lots of different angles, we decided to slice the data set of top searched terms per day and candidate in three different of ways:
Daily tag clouds show us the topic profile of top searched terms that every top candidate is surrounded with:
The word clouds are embeddable: as a fully dynamic live widget or a static snapshot of a candidate profile on a specific day.
Candidate cards shed light on the development of the search interest, over the last 3 weeks — how does the attention fluctuate, and why.
Finally, the big timeline provides a long-term view. What were the big “attention peaks” since the beginning of the year?
This is actually my favorite view of the data, as it reveals so much of the big picture when it comes to search interest around the elections. We plan to extend it with more layers and annotations.
You are focussing on search terms related to the top 10 candidates of the seven most searched political parties in the run-up for the election in September. What surprised you?
The most interesting thing was to understand the mix of topics around the candidates. Search interest range from political topics to gossip (like Lindner’s hair), from short-lived memes to terms that have constant level of attention (such as “refugees” for Merkel). We actually used this finding to improve the design of the site: in the tag cloud, we now group persons, media, places, and gossip separately.
Other than that, I was pretty surprised how much attention all the local campaign events draw, so a lot of smaller towns and cities pop up when candidates campaign in those towns. Sometimes, the data also provides us with real puzzles: We saw searches for Cem Özdemir’s wife peak tremendously just on one specific day in March, and we really didn’t understand why — until we learned she was part of a question in a big quiz show here in Germany.
Google Trends are sometimes compared with election forecasts. Why are Google Trends interesting for data journalism and what are the limitations?
Search interest and polling data are obviously quite different. I might have interest in a topic about a candidate, although I might not vote for them. We see our site not as a tool to see “who’s ahead in the race” but much more as a launch pad for observations and investigations into individual topics, and a guide to formulating new questions. We are, ourselves, curious and excited what this live data source will now reveal over the next few weeks!
Google Trends is a fascinating tool to analyse and visualize the pulse of key moments in the news cycle. If you want to learn more about our Google Trends election hub or the 2Q17 project by Moritz Stefaner contact as at newslabsupport@google.com.
