It’s almost Halloween, so I thought it was appropriate to write about something scary: statistics! (That was a joke, statisticians.) As a mathematician, I can get by in statistics, but I am not a native speaker. As someone who writes about math and science for a non-specialist audience, I think that statistics and an accurate portrayal of the statistics that support a scientific theory are extremely important parts of science writing. Later this week, I’ll be on a panel about statistics-based reporting, and I’ve been reading some statistics blogs to prepare for the session.
But should I be writing about those stats blogs on this math blog? After all, Statistics Is Not Math. This provocatively titled post by William M. Briggs says that while mathematics is useful in statistics, the purpose of statistics is to “compile evidence to use in quantifying uncertainty in (self-selected) hypotheses. How this evidence bears on those hypotheses may be best described mathematically, but why it does so cannot be.” Briggs believes that statistics classes should focus less on computations of statistical measures and more on what the equations mean—statistics education should be rooted more in philosophy than in math.
Appropriate for this blog or not, here are a few of the statistics blogs I read.
The first statistics blog I started reading was Statistical Modeling, Causal Inference, and Social Science by Andrew Gelman. His topics are all over the map, and he posts very frequently. I enjoyed his series about whether Harvard admits more Jewish students than it “should,” statistically speaking. In the first post, Gelman reported largely uncritically on an article by Ron Unz about the issue. In the next two posts about it, he changes his position based on further evidence and reflection and shares a very thorough critique of the article by Nurit Baytch. In general, I find his tone very confident, so it is nice to see that he is willing to rethink his positions and tell people when he does. Gelman often writes about studies he finds weak and sometimes posts responses from the researchers themselves. One of the most compelling posts of his I’ve read was about a study that almost got published. (The effect vanished.)
As a balance to Gelman’s Bayesian point of view, I also read Error Statistics Philosophy by D. G. Mayo, a “frequentist in exile.” It’s sometimes a bit beyond me, but I think she has a lot of interesting things to say about the strengths and weaknesses of frequentist and Bayesian statistics. The posts about Bayesian “comedy hours” are my favorites:
“Did you hear the one about the frequentist error statistical tester who inferred a hypothesis H passed a stringent test (with data x)?
The problem was, the epistemic probability in H was so low that H couldn’t be believed! Instead we believe its denial H’! So, she will infer hypotheses that are simply unbelievable!”
Simply Statistics by biostatistics professors Jeff Leek, Roger Peng, and Rafael Irizarry is new to me, although it probably shouldn’t be. Recently, they’ve been posting about an upcoming online “unconference” on the Future of Statistics happening tomorrow, October 30. The “two truths and one lie” credentials of the panelists look very impressive! I also enjoyed Irizarry’s post wondering why the best relief pitchers aren’t used when they’re most needed.
Hilda Bastian, who works for PubMed Health and PLOS Medicine, is one of my co-panelists this weekend. She writes Absolutely, Maybe on the Scientific American blog network. (Full disclosure: I also blog for Scientific American.) Absolutely, Maybe is a general-interest blog about “evidence and uncertainties about medicine and life.” It often goes into more depth and nuance than might be expected for a popular audience, but at a very understandable level. Her posts about the Hawthorne Effect and blemishes are particularly good. Bastian often illustrates her posts with cartoons she draws herself, and she has a statistics cartoon blog called Statistically-Funny.
Do you have a favorite statistics blog?