Lighthouse Reports heeft dit gerepost
Inspired by Big Local News and the The Associated Press shared data unit in the US and The Bureau of Investigative Journalism's Bureau Local and the BBC News' Shared Data unit, Lighthouse Reports had been seeking out an opportunity to put together an investigation around a high-value data set full of opportunities to tell important, localized accountability stories. We wanted to reveal not large scale corruption, but instead systemic inequality: to tell stories that focus not on a few high profile “bad guys” but rather on the victims of structural inequality and the policy, societal and economic failures in each country that created and perpetuates the status quo. "Brain waste” has been that opportunity for us. Read the first two parts of the series here: https://lnkd.in/g2Cf_MKn The EU Labour Force Survey unlocked potentially dozens of stories but also, as we quickly found out, could easily overwhelm our capacity, leading us producing literally hundreds of regressions that we had no capacity to evaluate. I was surprised about the lack of guidance available on how to pull together this kind of collaborative shared data project, so I have been sharing some lessons we are learning as we go. I’ll start off with what we learned about our data approach when we published Part 1 in April. It requires many, many ongoing conversations for data team members and reporting team members to speak the same language and reach a common understanding of what the data can and cannot tell us. We definitely had miscommunication and misunderstanding about what we were measuring, why and how! We learned from tried and tested social science techniques not only in terms of statistical approaches but also in terms of documentation, processing and transparency are key for both internal and external credibility. Without a track record of clear documentation, we would not have been granted access to the data by Eurostat in the first place. A variety of data perspectives is extremely important when conducting such experimental analysis. Without the insights and contributions of the data team of ElPais News and Financial Times, our findings would be less robust, ambitious and reliable. The findings are absolutely not “self-explanatory.” This relates to the first point. The more that data and traditional reporters define and shape the main investigative questions together, the less painful the process or reconciling the output of the analysis and the information that reporters need to understand the results and inform their reporting. Please check out our Github repo and methodology doc: https://lnkd.in/g9kgVwYD https://lnkd.in/guabW6qZ Next up, the lessons we learned about storytelling around data in Part 2, published in June.