Why is a physics professor surprised by this? You will get exactly the same "step" plot directly from the binomial distribution if you assume independent trials and a linear dependency between the SAT band and the probability of getting an A in a course.
If someone's probability of getting an A in a single course is 0.7 or above, the probability of getting at least 8 As out of 16 courses rapidly converges to 1, similar result holds for those with probability below 0.3.
Nothing new has been shown here - can't believe they actually wrote a paper.
It's not helpful to simply call the commenter out on not reading the article. First, it's unproductive, and second, it's merely conjecture. Perhaps you both read it but interpreted it differently? It would instead be more productive to point out why you think the person didn't read the article, thereby furthering discussion.
Notice that to get a "step" plot it is necessary to assume that probability of getting an A is strongly correlated with the SAT score. The paper says that for humanities there is no strong correlation.
On the other hand, the "result" for physics is an elementary consequence of repeated trials (binomial distribution)
I think independent trials is not an obviously good assumption. Any properly designed major will have later courses build on earlier ones, so one might have predicted that performance in the first course was more predictive of GPA than SAT score (though this isn't the only model that would lead to such a prediction). I, for one, am surprised that it appears to be so independent.
I should also point out that what is really relevant to success in a college course in physics or mathematics, even more than SAT score or some other IQ proxy, is specific subject matter background in science and mathematics. Whole countries do MUCH better than the United States in providing good background in science and mathematics to pupils in K-12 education,
so much so that the United States median achievement level is only at the bottom quartile level for those countries, or (from another point of view) the United States top quartile level of achievement is only at the median level for those countries. I think United States policy-makers ought to devote a lot more time to thinking about how to learn from those good examples.
Some United States mathematicians have taken care to look at how badly mathematics is taught to elementary pupils in the United States
> I should also point out that what is really relevant to success in a college course in physics or mathematics, even more than SAT score or some other IQ proxy, is specific subject matter background in science and mathematics.
But this is not meaningful without controlling for IQ. What allows you to achieve "subject matter background in science and mathematics"?
> the United States median achievement level is only at the bottom quartile level for those countries
And this is not meaningful without controlling for ethnicity. Disaggregated results show that Americans of European descent are competitive with Europeans, Americans of Asian descent competitive with East Asians, and so on.
What allows you to achieve "subject matter background in science and mathematics"?
Ask a question, get an answer. What allows you to achieve subject matter background in science and mathematics is good teaching of those subjects. That's the point of the international educational achievement comparisons. They show, as do studies of high-ability United States students,
that what limits educational achievement (and thus preparation for college studies) of many United States students is not limits in their capacity to learn, but limits in the curriculum they were taught. The United States school system doesn't step up to INCREASE the ability of the national populace the way many other school systems do better.
After edit: If you actually want to learn something serious about ethnicity in its many manifestations around the world, start here
I thought that it was much more interesting that they had hardly any data for students with a SAT-m under 600, which makes the conclusions slightly less convincing.
That said, there is a clear wall in the data they have so far. It would definitely be interesting to look into this further and see if that is the case, and also investigate more fields (I didn't see anywhere in the paper a complete listing of all the subjects they compared, which was a pity). I know I would not be surprised if a similar pattern was found in Computer Science!
For curiosity's sake, how many percent of the students taking a typical college course in math or physics typically receive the grade A? (As you might guess, I'm not American.)
At big competitive public schools where there's no grade inflation and grades are held to a curve, like Berkeley for example, perhaps 15% receive A's, 40% receive B's, 35% receive C's etc
has an incoming class of students whose SAT and ACT score interquartile ranges are at about the national median for the college 25th percentile. There are not a lot of conspicuously high-scoring students at that college, so the study submitted here has the issue of restriction of range.
It varies by school and professor. I know several big schools suffer from grade inflation
http://en.wikipedia.org/wiki/Grade_inflation#Grade_Inflation...
In general, I think its usually a average of a B, at least at Northwestern University, but other people might have different experiences.
Thanks. For the record, the average grades for the courses at my university tend to be Cs (in the case of the big introductory courses, the average is usually closer to a D than a B), with about 10% of the students receiving As. This discrepancy is quite unintuitive; (intuitively) you'd expect that the grade "A" means about the same thing in two different university systems that use letter grades.
My experience above is from discussions with professors in higher level classes. Im not too sure if that applies to intro levels.
I think that once you reach a certain level of difficultly the grade does not really reflect how well you learn the material, but more of how ridiculously smart everyone else in the class is (see that most of the test averages come out to about 50% or less).
As always, it's important to note that correlation does not imply causation. Do students with higher SAT-M scores make better physicists, or do physics programs reward students who are better test-takers?
If someone's probability of getting an A in a single course is 0.7 or above, the probability of getting at least 8 As out of 16 courses rapidly converges to 1, similar result holds for those with probability below 0.3.
Nothing new has been shown here - can't believe they actually wrote a paper.