Well, actually it’s snowing. Grrr. And I didn’t really wash my car. But I wanted to point you to a fun new web site and encourage you to participate:
It’s a variation of what I have often thought would be a cool use of “Everybody Votes” or whatever that thing is on the Wii: ask a bunch of random questions and then run some correlations. Post hoc ergo propter hoc is everybody’s favorite logical fallacy, of course, and even the best of us sometimes fall into the correlation/causality trap. My graduate alma mater is huge on theory, hence we are trained to value deductive reasoning and look down on inductive (in fact, I finally learned to keep them straight by associating “inductive” with “sinful.”) But, of course, we all use both, at some point in our research. And there are whole disciplines (epidemiology, e.g.) built on induction. (It’s a topic for another day (hopefully soon) but I sometimes wonder that in our quest for “data driven decision making” in higher ed., esp. around questions of student success, retention, etc., we commit one or more of these errors. On the other hand, the alternative in this area doesn’t seem to be well-founded deduction, but instead, mostly unprovable musings based on anecdote or “common knowledge.”)
Interestingly, I got to this via the Freakonomics web site. I really want to like Freakonomics; Levitt is a U of C guy, I liked the first book and thought it was mostly right and interesting, and am generally supportive of their style of reasoning. But in the end, I find them tiresome. I am subscribed to their podcast, but rarely actually listen to it. I worry that, in trying to create an industry, they are trying too hard, overreaching, and thus casting doubts on their truly valuable findings (or at least insights). And promoting “Correlation” might do the same thing: not everyone is going to draw the same line between reasonable and ridiculous conclusions to draw from correlations.
But play along anyway!