This July saw us deliver our third in-person data journalism training for investigative reporters in southern Africa, as part of a pilot programme designed to better understand how to bring these important skills to newsrooms in the region.
Over the years we’ve delivered countless numbers of these workshops, to individual organisations, to students enrolled in academic programmes and to cohorts from multiple countries all at once. One thing we have learned over that time is that there is one issue which is all-too-common. Everyone loves the workshops and feels light bulbs going on when they see how to find, clean, analyse and visualise data - but on their return to the office it’s hard to find time to try new things. Deadlines remain deadlines, editors remain all-powerful and demanding, and data skills require time to practise and deploy which is rarely given.
The result? There is a huge gap between learning data skills and practicing them. It’s one thing to try out new things with prepared datasets, but when presented with a sheaf of bank statements that must be scanned, digitised, cleaned and analysed, it seems simpler to do things the old-fashioned way (by hand) than figure out what Google Sheets has to do with it.
The data journalism helpdesk
For the last 18 months we’ve been testing way to address this gap between learning and practice: the Data Journalism Helpdesk. Working with selected investigative newsrooms, we delivered our usual data storytelling training, but then invited participants to make use of our skills on stories they knew had a data angle, but couldn’t quite figure it out.
“We need more of these for newsrooms to remove some of the fear that people may have with using data as a source for stories.” - Workshop participant
We set aside a few hours a month to help with any questions along the full data storytelling pipeline, from finding data to creating visualisations. Over time, we have assisted with 23 separate stories: this includes some which were killed as a result of the data not standing up to scrutiny, and others which took days of work to produce detailed investigations.
What we learned
Over the course of the programme, we’ve learned several things. Not least is that our starting hypothesis was correct. Everyone who took part gave us very positive feedback about the trainings, but as much as journalists are enthusiastic about data skills, it’s a challenge to integrate them into workflows. That goes for big data work and small data too - creating simple bar charts for categorical data can be just as big a leap as using an investigative dashboard for forensic analysis or asking for help to access satellite imagery of forest fires.
“Every story can be a data story. People also generally do not have time to read through long written stories” - Participant
Here are some of our main findings.
- Getting hold of data is still the main problem. Sometimes because it doesn’t exist, but often because many important documents are still locked up in PDFs or poor quality scans. The tools for PDF scraping are still not intuitive enough, and the data that results often need cleaning.
- Cleaning data is as much an art as a science. A data journalist with a decade of experience can quickly see shortcuts for tidying spreadsheets and removing blank rows, combining lines and cross-referencing with original documents to avoid errors. These skills only come with practice, and non-specialists don’t have time to do that.
- Recognising a data story is also a difficult art to learn. It may be obvious that a straightforward table can be visualised as a chart, but what about survey answers or patterns of spending over time? Visual storytelling doesn’t come naturally to someone who has spent years honing their skills as a writer.
- You cannot build it and they will come. Despite having resources on hand to operate the data desk, it was underutilised for a good portion of the programme. Partly, this is because writers aren’t always aware of what can be done with the data in front of them (see the last point), partly it’s because they just forget the service is there. We attempted - with some success - to address this latter by putting more effort into community build and awareness through regular monthly lunchtime workshops
- The need for a local helpdesk is very much there. Because we had time and understood the local context for stories, we were able to assist in ways that simply weren’t possible via collaborations with larger organisations that support investigative work.
What’s next? This phase of the helpdesk is drawing to a close and we’d like to thank everyone who has participated. We’re going to take a bit of time to assess our findings further, and then - hopefully - return with something bigger, better and even more relevant.
