If only journalists displayed this much facility with stats. XKCD comic by Randall Munroe.

A year of layoffs and anaemic ad buys has given journalists an excuse to turn inwards like never before. By now, even folks outside the industry must be sick of hearing about the Future of Journalism — my own fervent hope is to never read another article about social media for reporters. But I do think that an instinct for self-improvement is useful, so I’m going to add something else to the agenda — call it a Margin of Error manifesto. I’d like to talk about statistical literacy.

I know that it’s a bit predictable, even self-serving, to argue that everyone should have a skill that you have already developed. And I hate to add another job description to a list that is rapidly becoming unmanageable. Increasingly, it seems that we are all expected to be programmers and photographers, designers and copy editors, fact checkers and fundraisers. I’m not sure what to make of this job ooze, as an economist — we know something about the value of specialization — but so long as we’re all marginally employed and learning new skills, let’s try to pick up this indispensable one.

Craig Silverman has been saying smart things about journalism since before it was fashionable, and he is forceful about the importance of numeracy. But my point is not just about knowing how to add. When I say that we need to be literate, I mean that journalists should be comfortable enough with statistical methods to skim an academic paper, poll, or piece of market research and know whether the numbers say what our sources claim. We should have a clear hold on what it means to control for something, and when and why something is statistically significant. We should be able to compare contradictory studies. A couple rigorous university-level courses in social science methods would do the trick, more or less, but reading a few textbooks might be even better.

Unfortunately, most journalists don’t have a single course in methods, or any math past high school. We are, for better or worse, an industry of math-phobic English majors. And yet we report on statistics almost daily. Even crime reporters have to throw in the occasional paragraph on whether the murder rate is going up or down, and lifestyle columnists just love to write about neuroscience studies.

A few things happen when you have to report on something that you barely understand. You are forced to trust the researchers absolutely — whenever private companies release their own research, that’s a risk. You introduce errors by paraphrasing. You fail to communicate which results are suspect and which are nearly indisputable. Everything is reported as a breakthrough, because no researcher will admit to an incremental result. You can’t put contradictory studies in context, which drives readers nuts — who can remember whether red wine is good or bad for you, anyway? What are we supposed to eat, if everything causes cancer?

My sense is that political reporters develop some expertise on polling, and health reporters are getting serious about understanding drug and diet trials. A select group of journalists are making data what they do, and the New York Times has been doing some beautiful work with online infographics. But if you type “study” into Google News and scroll down, there is plenty of depressing nonsense to be found.

In August, the New York Times declared this the golden age of statistics, quoting Google’s resident math geek Hal Varian. Thanks to the internet, more data is more available than ever before. Statistics is only going to become more important. We can only hope that we will be able to cover it. Without statistical literacy, we will just be writing fiction.