Freakonomics

Just finished reading Freakonomics; it's a really fast read, plus I've read so many excerpts from it in various sources that it felt more like re-reading! In its eminent readableness it reminded me of Blink, and whaddaya know, the author of Blink, Malcolm Gladwell, has a little blurb on Freakonomics' cover, "Prepare to be dazzled".

Anyway, well worth a read, shows how analyzing data can lead to some very surprising conclusions. The authors talk about how their results can overturn conventional wisdom (a phrase coined by John Kenneth Galbraith), which is "simple, convenient, comfortable, and comforting - though not necessarily true". Me, I find it more comforting to see the conclusions reached by data analysis, but I'm a geek.

However, I wonder about the use of regression analysis. I understand the basic concept, but it terms of how it really works, it might as well be magic to me. I remember reading Stephen Jay Gould's The Mismeasure of Man (what an exciting book, his humanist passion shone brightly through), in which, if I remember correctly, he talks about how measures of intelligence use flawed analyses (regression analysis? factor analysis? must reread!). Well, I followed along but ultimately I didn't have enough expertise to judge whether there were flaws in his argument. I had the same feeling of slight discomfort here - it sounded good, and made sense, but ultimately I took the authors'word for it.