For the first time in over a year this past Thursday I taught a class without the use of Zoom. Dear God, it was fantastic. If I get lucky it may happen again on Friday. Soon by you.
This is from “Who Gets In?” by Rebecca Zwick, overall a sort of dull book, but with a provocative take on grades vs. standardized-test scores in college admissions:
“One way to think about this is in terms of signal-to-noise ratio. In engineering, this ratio compares the strength ofa “signal” (often an electrical signal) to the level of background “noise.” In the case of grades and test scores, we can think of the signal as “academic skills.” The noise is anything that interferes with our assessment of the signal (measurement error in psychometric terms). Because they are affected by many nonacademic factors, including the teacher’s stringency and the student’s attitudes and classroom behavior, grades are relatively noisy measures. SAT and ACT scores are less noisy (more reliable, psychometrically speaking) in that they are less susceptible to these sorts of extraneous influences. This in itself does not mean tests measure “the right thing” or that they are more appropriate for use in admissions. What it does mean is that they are more consistent and precise. All other things being equal, a less precise measure will yield smaller group differences. In other words, because grades are noiser than test scores, they will tend to blur any distinctions among groups in academic performance.”
But maybe its consistency and precision that is in itself a source of unfairness? Is it fair to precisely measure an individual student, given that the test itself is (as any measure of academic skills must be) an imperfect one?
Suppose you were a music executive trying to evaluate the quality of a pop song. On the one hand you have an algorithm that consistently produces a single number that does an OK (not great) job predicting how many streams the song will get. And then you have a panel of experts that have are less consistent in their evaluations – more error prone – but on average performs just as well as the algorithm. You’d want the algorithm and the panel, wouldn’t you?
It’s interesting to read about selective college admissions. But as Zwick notes at the very start of this book, “Getting into college in America is remarkably easy.” Most community colleges take pretty much everyone, and she cites tates that over 50% of colleges admit more than 3/4 of applicants.
I read things saying that more poor kids need to go to college, I agree. Others say sending everyone to college would devalue the financial benefits of a college degree and not necessarily improve the economy, and I agree with that too. People write that you shouldn’t need a college degree to survive in the wealthiest country in the world, and that sounds smart also. I don’t know how it all fits together, but something’s not right.
Also I have read this thing and that on whether elite college actually provides benefits to wealthy students who attend. (All seem to agree that it increases earnings for poorer students.) Very confused on all this.
Early Family Math // Galois Theory textbook (excellent first chapter) // LEGO microscope // Proofs of the Pythagorean Theorem? (pdf from MAP) // Lost Cities is a good game // Modeh