Arnold Kling  

What I'm Reading

Guilty as Charged... Morning Reading and Reactions...

I just finished Aubrey de Grey's Ending Aging. I think that anyone who likes reading science books for pleasure would enjoy it. I'll have more to say later.

I'm about halfway through Supercrunchers by Ian Ayres. It's about the rise of statistical decision-making. On p. 10:

We are in a historic moment of horse-versus-locomotive competition, where intuitive and experiential expertise is losing out time and time again to number crunching.

He sees physicians as occupying the horse-and-buggy niche. On p. 102:

The Super Crunching revolution is the rise of data-driven decision making...Most physicians (like just about every other decision maker we have and will encounter) still cling to the idea that diagnosis is an art where their expertise and intuition are paramount. But to a Super Cruncher, diagnosis is merely another species of prediction.

Skimming ahead, I see that he talks about the Direct Instruction reading method, which claims lots of statistical support. Years ago, I ordered Siegfried ("Zig") Engelmann's book about the method. Ideologically, I wanted to like it, because I agree with Ayres that educators are biased against using data to evaluate teaching methods. But reading the book made me so uncomfortable with the DI method and its robotic approach that it would take a ton of data to over-ride my prior that it's not really a good way to teach reading. Anyway, Ayres writes on p. 166:

In part, the struggle in education is a struggle over power. The education establishment and the teacher on the line want to keep their authority to decide what happens in the classroom. Engelmann and the mandate of "scientifically based" research are a direct threat to that power. Teachers in the classroom realize that their freedom and discretion to innovate is [sic] threatened. Under Direct Instruction, it is Zig who runs the show, who sets up the algorithm, who tests which script works best.

...In education, the struggle between the intuitivists and the Super Crunchers is ongoing, but in consumer lending the battle ended long ago.

And in mortgage lending, I would add. I've said before that I credit myself with a role in hastening the triumph of the Super Crunchers over the intuitivists in mortgage underwriting. I think that some of the rise in home prices over the past 15 years reflects that, as we made underwriting more accurate and less costly. Evidently, some mortgage underwriting Super Crunchers got overconfident in how far they could push the envelope, but still...

Later, I can see where Ayres comes to the defense of Larry Summers (p. 205-208).

His core claim, indeed his only claim, of innate difference was that the standard deviation of men's intelligence might be 20 percent greater than that of women...

I've recalculated Summers' estimates, using his same methodology, and his bottom-line characterization of the results, if anything, was understated. At three point five or four standard deviations above the mean, a 20 percent difference in standard deviations can easily translate into there being ten or twenty times as many men as women [in the highest reaches of math and science]

Finally, on p. 214 comes a statement I want to frame and put on the wall. He gives a standard Bayes' theorem problem and says,

In study after study, most physicians tend to estimate that the probability of cancer [in the example] is about 75 percent. Actually, this answer is about ten times too high. Most physicians don't know how to apply Bayes' equation.

Every year, I tell my statistics students an anecdote about the Harvard-trained doctor to whom I quoted Bayes' theorem in arguing that he was sending me for an unnecessary test. He had never heard of Bayes' theorem, and he thought I was nuts.

I never went back to him.

So, while doing this post I pretty much finished the book. Ending Aging took more time and, honestly, it gave my brain a better workout. But Ayres certainly was fun.

Comments and Sharing

COMMENTS (4 to date)
John Thacker writes:

There are people who are astoundingly good at intuitive behavior, and at things like teaching. It's probably true that they could outperform the expert systems. However, most people tend to overestimate their own skill at things like that at the same time, as studies show.

So in my opinion you end up with teaching methods and systems that would be better for the strong majority of teachers and a majority of teachers who feel that they're in the minority who do better improvising and handling it themselves.

Most of the popular intuitive educational methods do well when the best teachers do them, but fail in the hands of average teachers and below. Combine that with how the best teachers go to the wealthier, better schools, and IMO the educational establishment really ends up holding back poor kids a lot, pushing teaching methods that only work for the teachers that the poor kids don't get.

John Thacker writes:

With the US Open going on, the book could also discuss the rise of systems like Hawk-Eye to automate judging, slowing moving into the domain of umpires.

David Robinson writes:

You'll be glad to hear that even the introductory biology course at Harvard today teaches Bayes' theorem, so I have no idea how that doctor missed it (though biology teaching has historically been a bit of a mess up until not too long ago).

Shakespeare's Fool writes:


Does anyone have a link to or reference to (or
correction of) what I am (mis?)remembering as a
Samuel Johnson quote:

"The new way is to count, the old was was to guess."


Comments for this entry have been closed
Return to top