Statistical Pictionary

Posted by Milos Sugovic

I randomly (pun intended) came across this rant by Andrew Gelman, a professor of statistics at Columbia University, on the use, misuse and abuse of graphs. Now if you’re too lazy to read the entire post, this pretty much sums it up: 

Graphs are gimmicks, substituting fancy displays for careful analysis and rigorous reasoning. It's basically a tradeoff: the snazzier your display, the more you can get away with a crappy underlying analysis. Conversely, a good analysis doesn't need a fancy graph to sell itself…[I]n writing this piece right here have I realized the real problem, which is not so much that graphs are imprecise, or hard to read, or even that they encourage us to evaluate research by its "production values" (as embodied in fancy models in graphs) rather than its fundamental contributions, but rather that graphs are inherently a way of implying results that are often not statistically significant. 

Given his credentials, I may not be in a position to challenge his point of view, but I can’t help but disagree with him on a few things.  

Yes, graphs can, if used incorrectly (or unintentionally), be misleading. Robust quantitative research must not be based on pictorial examinations (unless there’s a legitimate reason), especially in the case of academic research. 

But isn’t there some merit in presenting a statistically sound analysis in a “snazzier” fashion? Is there no value in efficient communication between the presenter and the recipient of the research and analysis? Graphs summarize and simplify. If backed by a robust analysis, what’s the harm in implying what you’ve been able to “prove” in the first place? I wonder, has he ever heard of the old saying “A picture is worth a thousand words”?