« « Collaborating with computers to parse “big data”

Changes in the numbers of students majoring in programming and social sciences » »

Committing fact errors in visualizations

Posted by on Jan 18, 2010 in Blog, class, data visualizations, tufte | No Comments

At Medill, there’s a wonderful tradition called the “Medill F.”  Make a factual error of any sort, and you fail the assignment.  The sadistic part of me likes it — a journalist’s job is to tell the truth.  If you miss the mark, you’ve failed the public, and failed at your job for the day. Harsh but true. My only Medill F was writing down the wrong day for an exhibit closing. I’ll never forget that one.

Why do you care, you ask?  Because it’s an example of just how simple it is to make a factual error.  That’s something I’ve been thinking about a lot recently. 

Sure, it’s bad, immoral even, to make up sources or quotes, say you saw something you didn’t. The sin of omission, leaving out certain parts of a story, is a facutal error in its own right. I knew that.  But in one chapter of Edward Tufte’s “Visual Explanation,” titled “Explaining Magic: Pictorial Instructions and Disinformation Design,” we learn about the factual error of miscommunicating information.

I think the magic analogy is a strong one, because images and graphics are so powerful, it’s all too easy to use a sleight of hand and manipulate the information.  On the one hand, if you include everything, it becomes cluttered and it’s not easy for the user to discern what’s important.  Also, if you’re just dumping data into a visualization, you’re not really helping anyone, and I would argue that’s not journalism.  Dumping data into a visualization is different than the wonderful work being done with news apps like Politifact and Everyblock — even as they use lots of data, they are making sure to contextualize it. That’s what makes the idea of data-driven apps so appealing to me — I believe they are a very important part of where the future of journalism lies.

But from a journalistic standpoint, if you focus on certain pieces of information, you are making a sort of editorial judgment.  So, the lesson I take from this chapter is that it’s something I need to be careful of.  Don’t crowd out the important material to diminish its impact (Tufte’s example of this is a thick border around Surgeon General warnings on cigarette packages.)  Just as it’s an editorial judgment when you decide what story to report on, and where that story gets placed in your new product, similar judgments come into play with data viz.  And those are certainly not decisions to be taken lightly.

« « Collaborating with computers to parse “big data”

Changes in the numbers of students majoring in programming and social sciences » »