The way you get better at something is with deliberate, difficult practice, and deep inquiry. When I was in high school preparing for the SAT I found this was the opposite of how most test prep worked.

I found one book--called something like SAT 1600, which was the max at the time--that talked about how test makers try to assess you, and how questions are constructed. The drill questions were very hard and came with an explanation of how the questions worked. Taking a practice test before and after the book my score went up by over 100 points. It got to the point where I could usually answer a math problem by guessing what the answer "should" be, based on how the question was built, and then just work backwards to verify the answer.

But most test prep was boring drills together with strategies like "use the process of elimination." Yeah, duh. That sort of brain-switched-off busywork is the common pattern in test prep and, more broadly, American education.

I agree. And isn't the idea of a test to find the people that are best a doing something. It doesn't really matter how much preparation goes into it.

So maybe someone that works really hard to be better than others at something is more valuable than someone who is naturally bright?

For example - a naturally bright lawyer isn't as good as a 'good' lawyer that tries really hard to be the best. Being good at a test shows that you'll be good at whatever the test is trying to find - no matter how this is achieved.

Expensive "test prep" also involves learning about scamming your way into a good score:

"Nationwide, only 2 percent of students who have taken the SAT over the past 10 years have done so untimed. Most of these students' diagnoses are presumably genuine. But in places like Greenwich, Conn., and certain zip codes of New York City and Los Angeles, the percentage of untimed test-taking is said to be close to 50 percent. "

http://www.slate.com/articles/health_and_science/medical_exa...

If I remember my days as a test-prep teacher, I believe one of the largest predictors of SAT score was simply the age of the student.

In other words, if you give a thousand students a free practice test, then sign them up for a course that summer, by the time they actually finish the course six months later they're already expected to have improved regardless of whether you teach them anything at all.

I hadn’t thought of this effect, but it’s kind of an ingenious way to market test prep. I think at some point, students’ scores stop improving, from a combination of having been exposed to all the background material necessary for the test to be accurate, and their brains finishing development (after which intelligence is largely stable), but up until then, students’ scores improve with age. You can still use a test like the SAT to compare younger students, but you have to norm it against students of the same age (I think it’s actually better to test students at a younger age because a test of the same difficulty has more ceiling room).

“Such tests are blind to a student’s high school record—instead, they are intended as an independent, objective measure of college “readiness”.”

Claiming that the SAT measures “college readiness” is complete and utter bullshit. The SAT measures intelligence and the College Board knows it. “College readiness” is a liberal euphemism. It actually turns out that if you try to test specific skills (I think the writing section kind of falls in this category) instead of overall intelligence, the results are less predictive for whatever purposes the test is being used for, since specific skills can be learned, whereas overall intelligence is largely stable and has broad effects. The fact that scores cannot be significantly improved by coaching is evidence that the SAT does in fact largely measure intelligence.