I happened to come across this comment from Jrc a few years back:
A lot of your posts about journalism and writing are almost like literary criticism – close reading for context, picking out particular sentences or clauses that reveal something fundamental about the viewpoint from which they originate.
I wonder if there isn’t something closely related between that kind of skill and good statistical analysis: nuance, context, one or two funny little coefficient-movements/word-choices that you could easily look past (or read over), but, to an alert reader/analyst, might reveal something fundamental about the relation between the model (or idea) and the real world.
Maybe it makes more sense the other way: bad writers write some stuff and keep what sounds good (regardless of its actual truth or falsity or depth of insight) the same way bad analysts run some regressions and keep the ones that are statistically significant.
Interesting point! Close reading is indeed a skill. I thought of that after doing the work that led to this post and this article. It all looks clear once it’s all done, but when you first come at a paper, it’s all smooth surfaces and it can be hard to figure out how to get into it.
That said, there are lots of ways of doing literary criticism and lots of ways of doing statistical analysis, and I think some people can do well without having that close-reading skill.
I have always felt that good statisticians are able to flip back and forth between the forest and the trees.
This is usually true of anyone who’s good at anything complicated. You have to pitch your ideas at the forest level, then execute at the level of trees. I first got this lesson concretely in a NY Times Magazine piece about Newt Gingrich, who was explaining how he pitched ideas.
And one way of doing literary criticism using statistics is known as “distant reading.” :)