Data journalism is changing the way we report elections: here’s how

Whenever there’s an election in any democratic country, its — and sometimes the world’s media — goes tends to be put on full alert. That said, traditional reporting is generally limited to the political issues in the campaign and to winners and losers. Data journalism tools open a new perspective, offering the possibility for in-depth reporting and analysis.

As renowned professor of journalism Stephen Doig notes, data journalism is something like “social science done on deadline”. As is the case with social sciences, the tools of online journalism are generally freely available online and aren’t so difficult to use. While they can seem a little daunting, when paired with a little bit of training these tools are helping change the way we report on elections.

Twitter

Twitter is an important tool when it comes to interaction and discussion within any constituency. What the topics are and who plays a central role in the different debates are important questions that need to be asked.

It’s also an area where journalists can borrow from social sciences in the field of social network analysis. Journalists probably shouldn’t dig deep into the theory and the mathematics, but use simple programs for collecting data and analysis such as Gephi and NodeXL.

Both programs show you the network around, for instance, a particular candidate or a topic on Twitter, and help you analyse the central persons in those networks. This is a great help for finding interesting sources on Twitter.

Map

Mapping election results — illustrating who won and lost where and what is the turn out was in different areas — is an incredibly important part of election reporting. The maps tell the story!

Google Fusion Tables (FT) is a great help for putting up interactive maps online. You may however need a map of your different administrative levels. In this case, it’s worth having a look at DIVA-GIS for maps of your country. There is however a small problem: the maps are in shape format(shp) and can not be use in Google FT. Google uses an other format:Keyhole Markup Language (kml). Luckily there is a service available for converting shp to kml: shapescape.

Google makes nice maps for online viewing, but for print journalists, the resolution on their maps are too low. But you can take another direction: explore software for Geographical Information Systems(GIS). QGIS is a good choice for all platforms and produces nice printable maps. QGIS uses the shp format, so no conversion is needed. On top of this; once you have made the map you can save it in kml format; and use it directly in Google FT for online.

If meanwhile a newspaper wants to run its own map server to produce maps on its own website, QGIS has also a good map server and a nice web client, called Lizmap. Installing a web server is not for the weak at heart, and will most likely remain the reserve of geeks in the newsroom and the IT department.

Scraping

Making maps of winners and losers or the turn out is interesting, but analysing the relationships between variables is a lot more fun. Is there, for example, a relationship between average income, crime rate, unemployment and other social variable and the outcome of the elections?

For finding this kind social data, national bureaus for statistics could be a great help, alongside the databases of NGOs such as the Worldbank. I know most bureaus for statistics send out pdfs and not data in Excel(xls) format. If this is true of the particular one you’re dealing with then use Tabula for convergence or a free web service like pdftoexcel. If meanwhile, the data is on a web page then scraping is then the way to get your data, with Outwit-Hub the tool that probably makes it easiest.

Analysis
For analysing variables and their relationships, Excel is a good tool. Simple calculations for one variable is not difficult. Pivot tables in Excel are a great help for summarising data. Excel makes nice graphs that are fairly easy to use but anyone who wants to do some serious statistics, take a look at R project. You can either use R commander to make all the stats you need or dig in to R studio and learn the use of R. Swirl is a good interactive and multimedia introduction.

Visualising
Making graphics in Excel is okay and works fine for print publication but you simply cannot use it for making interactive online charts. Google Charts is a good alternative and has lots of interesting charting possibilities.

A more easy tool for visualizing is Data Wrapper. Just import your data and choose your graph type. The easiest tool for visualising though is Tableau Public; which offers an all-in-one solution. You can make tables, pivot tables, graphs and even maps all for free. All your work can be exported to the web too.

Survey
Vox pops are often used to show off the public’s opinion about a topic. They add colour to the story but they don’t prove anything. Opinion polls done by official agencies meanwhile are quite expensive and time-consuming. Fortunately, journalists can now take a more analytical approach.

Google Forms is a perfect tool for creating surveys around candidates, or a particular electoral issue. You create a questionnaire online with Google forms, send them by email to your sample group, and the results are stored in a spreadsheet format.

Data journalism presents new possibilities for enhancing the quality of reporting about elections. But there is a price to pay. Data journalism projects are time-consuming.

One possible solution to this is cooperation, not only within a newsroom, but also between different media and platforms. Collecting data together, for example, helps spread the cost. Creating a project, for different newsrooms and various platforms or media, is often the best way to go but defining the project’s objectives, time path, budget and training needs are essential for success.