Counting Every Vote
Category: Best data-driven reporting (small and large newsrooms)
Country/area: South Africa
Organisation size: Small
Publication date: 4 Dec 2019
Credit: Athandiwe Saba, Jacques Coetzee
M&G Data Desk analysed voting results for the past 20 years ensuring every vote and area was part of the discussion leading up to May 8.
Three months prior to the National Elections in May we extracted, compiled and analysed election data from the past 20 years yielding stories about how the youth wouldn’t vote, how one smaller party would cause an upset and the decline of the ruling party.
We were innovative in approaching the coverage of these elections focusing on using new technologies and directly engaging voters who desperately wanted to be heard.
The M&G Data Desk was innovative in how it approached the elections using technologies that fit into our minuscule budget yet producing work that can rival any newsroom and can be used by our readers and society to make the best decisions.
The 2019 National Elections were the most hotly contested and this year instead of focusing on the politics of parties and personalities, the M&G Data Desk felt compelled to focus on the voters and what were the most important issues to them. But we didn’t only want to focus on anecdotal evidence. The Data Desk created databases with elections records spanning over 20 years of how voters would either turn out or not. It was imperative to also analyse the progress made in key areas where voters had protested for instance and where voter turnout had been the lowest. The Data Desk engaged the electorate about issues that were important to millions and would dictate which party would make strides and which would fall to the wayside.
We questioned the data we had gathered and ask questions such as What do Women Want? The data showed clearly that more women than men registered to vote and possibly turned out at the polls yet their interests were not proportionately represented by political parties. We also focused on why certain areas all over the country did not turn out to vote – Where voters won’t come out to vote – and which areas were using their democratic right to chop and change who they wanted as their representatives. These are all issues affecting the electorate which have not been adequately covered by the media.
We used Python to extract the data from the dozens of pages elections data had been compiled in. The data was in various formats and had to be cleaned in CSV. The data was then analysed using Excel, Tableu and
Once we had the data prepared in various CSV worksheets and the data had sent us to tell the stories in particular areas we used Flourish to create interactive storytelling methods to better deliver the content.
Graphics were created with DataWrapper and Flourish.
We then created the Live Results Page using PYTHON by connecting it to the Electoral Commission of South Africa’s API, ensuring zero lag time between results allowing voters and readers to engage and ask pertinent questions about their preferred party’s. This allowed for on-time reader engagement to direct the content our readers needed.
This platform had more than 72 000 unique views over three days, this is unheard of in a newsroom.
From our analysis, we first noted that we were able to correctly predict how a right-wing political party would gain a substantial number of votes in the 2019 National Election.
Our data was credited and discussed across the media spectrum.
The stories and interactive graphics were viewed and shared hundreds of times on all platforms. Most importantly the Data Desk created content backed by technological and scientific means to open conversation on television and radio shows.
What was the hardest part of this project?
We learned that the lack of resources in newsrooms does not mean the inferior quality of journalism. There are numerous avenues a team like ours can tap into to ensure citizens are better informed and their stories are told, backed by intense research and data analysis.
Partnerships are one of the most important aspects of ensuring journalism continues to thrive and governance and government are held to account.
On the other hand, the Data Desk spent weeks creating the Live Results Map which viewers spent more than 8 minutes at any given time on yet we did not turn this into a commercial success and this is what newsrooms need. We learned a very difficult lesson here and going forward more team members must be brought in to flesh out how best such digital success can be turned into commercial success.
In the six months of its establishment, the Data Desk has changed how the M&G newsroom and other platforms approaches stories and specifically this election. The team has produced data-driven stories that can be used by our readers to make better decisions, stories that drive the national debate and bring facts and figures to the national discourse.
What can others learn from this project?
Data journalism is not simply about the numbers and graphics. It’s about journalism and how the numbers affect people. Though we had a large amount of data, it did not mean much unless citizens were engaged and understood what it all means. This is what we did with this election. We used publicly available data to find how the elections would turn out. We were directed by the data showing us where the voiceless were.
The 2019 National Elections were the most hotly contested and instead of focusing on the politics of parties and personalities, the M&G Data Desk felt compelled to focus on the voters and what were the most important issues to them. But we didn’t only want to focus on anecdotal evidence. The Data Desk created databases with elections records spanning over 20 years of how voters would either turn out or not. It was imperative to also analyse the progress made in key areas where voters had protested for instance and where voter turnout had been the lowest. The Data Desk engaged the electorate about issues that were important to millions and would dictate which party would make strides and which would fall to the wayside.