What you said

1. Take time to understand your data

It’s tempting to want to dive into a new dataset straight away, but doing so risks producing a misleading or inaccurate story. First, advised scientific journalist and founder of formicablu, Elisabetta Tola, you need to understand the science behind the data, including the models, variables and measures used by scientists to collect the data.

To this end, Dianne M. Finch suggested that you interview your data as you would a human source: Who produced the dataset and why? How was the data collection project funded? What is the margin of error? Where are the outliers--and are they data entry errors or actual outliers?

Done correctly, this process will ensure that your journalism meets standards of fairness and accuracy. For example, Jamie Smith Hopkins from The Center for Public Integrity told us about a situation where their checks identified that a manufacturing plant was accidentally listed among America’s 100 biggest polluters due to an emissions reporting error.

2. Reach out to scientists

As part of your workflow to understand the context and format of your data, don’t be afraid to reach out to scientists and researchers working in the field.

Scientific data explained Elisabetta Tola, "are produced with highly specialised software, and the outputs are often in formats which are not the ones we use in data journalism".

For example, the Texas Observer’s Naveena Sadasivam told us about how hazard classifications, provided by the National Inventory of Dams, have misled reporters to assume that a 'high' hazard dam is in poor condition and is more likely to fail. But, in actuality, this classification is simply a measure of the level of damage if a dam fails.

Asking scientists and advocates for their input can also help to focus your reporting, said Aleszu Bajak from Northeastern. "If it’s going to be overwhelming, they’ll tell you. If it’s interesting and achievable, they may help."

But, Elisabetta Tola warned, do not treat a scientist’s insights as infallible. Look for conflicts of interests, biases, and the possibility of misinterpretation. Above all: "it is crucial not to base an entire story on a single source, albeit an expert one, and integrate it with others to test and verify."

3. Avoid scary data tactics

If you’re looking to convey a big picture message about an environmental issue, ironically the advice from Global Forest Watch is to narrow things down and make them relevant.

"Often people feel overwhelmed or even powerless when learning about complex, global environmental issues. Bring it back by personalising stories or even including a call to action so that your readers can relate."

This finding is echoed in media effect research. Fenja De Silva-Schmidt, from the University of Hamburg, said that stories which frame climate change in terms of ‘catastrophe‘ and ‘guilt‘ are often demotivating for audiences, rather than helping them understand and mitigate climate change in their daily lives.

To this end, Gianna-Carina Grün, Head of Data Journalism at DW, shared an example where they used travel costs to personalise CO2 emissions data.

Although it’s a great way to connect with readers, Gianna warned that hyper-personal projects, such as The New York Times hometown interactive, can present their own set of challenges. Many readily available datasets don’t hold data on a very granular level, and most teams lack the resources to research adequate datasets, one country at a time. So, her tip is to personalise by scaling down to a region or even just a few countries.

4. Consider the constraints of visualisation

The nature of geospatial data can also impose limitations on how it can be visualised. Timo Franz, from Dumpark, said that they often face technical constraints, such as loading times and computational limitations, when mapping large and detailed datasets.

When producing their dot map of plastic pollution in the world's oceans, for instance, this meant that they had to reduce the number of overall dots, ending up with each dot representing 20kg rather than more intuitive 1kg or 10kg representations.

Similarly, Timo noted that mapping around the Earth’s poles can also be problematic, as many tools use projections that aren’t well suited to it.

As a more general visualisation tip, Federica Fragapane, creator of Carbon Dioxide Emissions, explained the importance of providing your audience with different levels of visual analysis. In her piece, she complemented the main informative layer, annual country-level carbon dioxide emissions between 1992 and 2012, with colour shading to illustrate how the emission situation has evolved throughout the years.

5. Sometimes you have to take matters into your own hands

Fiona Macleod, Editor at Oxpeckers Investigative Environmental Journalism, told us that many government bodies and civic organisations claim their data is open source, but negotiating access can prove tricky and time-consuming.

In many instances, they’ve had to use South Africa’s freedom of information framework, or try to find ways to scrape the data from the web.

"Sourcing, cleaning, sorting, analysing and applying the data in this way may be counter-intuitive and time-consuming, but in some instances it's the only way to get the job done," she said.

Our next conversation

We're excited for an ‘ask me anything' with the author of the Data Journalism Handbook 2's social media data chapter, Buzzfeed’s Lam Thuy Vo, next week.

Until next time,

Madolyn from the EJC Data team