Tips for avoiding blunders with numbers in the news

These are some tips for avoiding blunders when writing stories with numbers and statistics. I came across with them while reading handbooks on Data Journalism, so this is a summary of the best of them:

• When requesting a database, ask for all the variables and records included in it, not only those which could answer the questions for your immediate story. This can prevent you from reaching wrong conclusions as well as provide you new ideas for follow-up stories.
• Steve Doig, from the Walter Cronkite School of Journalism of Arizona State University, gives one interesting piece of advice: always ask the agency giving you data if there are any undocumented elements in the data, whether it is newly-created codes that haven’t been included in the data dictionary, changes in the file layout, or anything else. After that, always examine the results of your analysis and ask “Does this make sense?”
• Do not assume that the background notes will explain what the data is about with enough clarity. If key doubts emerge from the data, try to solve them with the press officers before publishing the story.
• Clean the database up before starting to extract data and conclusions.
• Explore the data and play with it, trying to come up with stories which would be consistent with them.
• Do not add your own interpretations, beliefs, or preconceptions to data, that will pollute your story.
• Do not comply only with the data, but request interviews to obtain more information and check your hypothesis. According to Michael Blastland, freelance journalist, “different sources provide new angles, new ideas, and richer understanding”.
• Put the number in context. Is that number really big? In comparison with what?
• Check your sources. Data must look very interesting, but its reliability might be weakened if it comes from a source with interests in getting that story published.
• Data Journalism can make great headlines, but you should choose them carefully for them to match with your data. Do not exaggerate the results of your investigation. Do not mislead. Avoid sensationalism.
• Choose the best structure to tell your data story. It can be a comparison with the past, it can highlight a proportion, it can be divided by categories (political parties, football teams…), etc.
• Correlation and causation is not the same thing. This is specially important in my field, Science Journalism, but equally crucial in data stories.

---