What we learned training South Africa's new generation of data journalists

by Adi Eyal
Oct 30, 2018 in Data Journalism
Lines of code

When Code for South Africa launched Africa’s first data journalism academy in February this year, we were trying to address a problem facing local newsrooms. We saw that South African newsrooms lacked the internal technical skills to start wrangling datasets and using them to find stories or to complement investigations. That rare hybrid that blends editorial judgement and data analysis is seldom found in South African media organizations.

The reasons for this are many and complex, but it is partly due to the lack of trained data journalists. And so we launched the academy to fill that gap. The program begins with an intensive two-week training, starting with data acquisition (freedom of information requests, web scraping, Google-fu, etc.), and moving to preparation (data cleaning) and analysis.

This two-week training period is followed by four weeks working full-time in our newsroom practicing these skills. Journalists are expected to regularly produce data-driven stories, which are then sent to their own newsrooms for publication. This way, the journalists don’t have to take a hiatus from their newsrooms when attending our data academy—instead, they remain productive.

Our rationale for this is that two weeks aren’t enough to internalize these new skills. Consider what would happen if after a two-week training, these journalists would return to a newsroom where they are the only data journalists. Given their limited experience in wielding their new skills, they are likely to quickly abandon their spreadsheets for their notebooks and pencils, as they are pressured by news editors who don’t understand what spreadsheets can offer and don’t know the wealth of stories that are locked up in boring datasets.

That’s why Code for South Africa’s data journalism academy aims to support journalists as they master the ins and outs of data. Our goal is to provide scaffolding during the learning process, reducing support as the journalists become more confident in their ability to tell data-driven stories.

We are just about to begin the training of our third cohort of journalists and are currently reflecting on our experiences, model and future direction. Unsurprisingly, we encountered a number of challenges, some of which were expected, some not.

The challenges

1. The difficulty some journalists experience in mastering the skills they have been taught. The two-week initial training moves at a fast pace in order to cover as much material as possible, and some journalists felt they were not given sufficient time to absorb everything. We’re refining the curriculum to address this issue. We are also starting to tweak which tools we teach and incorporating far more time for practical work during the initial two weeks. 

That part is easy. Inculcating an appreciation for data and an understanding of why it can enhance existing journalism skills is hard. Data journalism, as I see it, is not merely a spreadsheet and a bag full of data visualization tools. We prefer the term data-driven storytelling. We want our graduates to have a deeper appreciation for how this new paradigm can enhance storytelling skills.

2. Getting data stories published in journalists’ newsrooms. There is no guarantee that visualizations produced by the academy can actually be ingested by a news organization’s content management system (CMS). Many of these cannot include any artifacts except for text and images. Most of the time, our development team has had to engage with the newsrooms’ CMS developers to shoehorn the work produced by the journalist into their news websites.

3. Protecting what happens to a story once it is filed. Often, the newsroom sausage machine receives a beautifully crafted story and mangles it beyond recognition when it is published. Graphics are re-arranged, important analysis chopped and the layout stripped bare. These inflexible workflows prevent the publication of great-looking stories. Still, we think that the more stories we push through the system, the better the process becomes. Content management systems improve, layout editors understand how to incorporate non-standard layouts and the entire production becomes much more tolerant of this new approach to storytelling.

While challenging, there is a clear path towards addressing these problems. Unfortunately, there are others to which we do not yet have answers.

Mainly, does it make good business sense to produce more data journalism? How do news organizations benefit from these expensive and slower-to-produce stories?

When we first set out, our overarching goal was to produce systemic change in the industry. Our approach was to address the skills shortage by training the next generation of data journalists. Unfortunately, answering the “so what?” question is hard. Will these journalists be able to produce harder-hitting, more engaging stories? Will these stories draw larger numbers of readers and attract more lucrative advertising streams?

I believe we already have some examples of stories that help justify the investment in data-driven storytelling. These include the award-winning living wage story and calculator, as well as these stories on the lack of toilets in parts of a Cape Town township. This work contributed to a research tool that is being used by civil society organizations when advocating for the provision of permanent sanitation facilities in townships.

These stories and tools have the potential to help drive policy changes, or at least contribute substantially to a national dialogue on issues of poverty and inequality in South Africa.

As for financial sustainability, we are exploring these options too. What is clear to us is that attempting to help shape the future of journalism in the local market is a complex problem with multiple challenges that must be tackled together. Addressing them in isolation is unlikely to result in the widespread change we are hoping for.

Adi Eyal is the founder of Code for South AfricaLearn more about his work as an ICFJ Knight Fellow here. 

Main image CC-licensed from Flickr via Kjetil Korslien.