Noah Smith is an "econ blogger" who mostly does econ stuff, but a week ago he decided to take a swipe at education. Yes, it's always super-awesome when economists try to talk about education, but this particular post caught my attention because Smith would like to explain why smaller classes are a snare and a delusion, and his argument highlights one of the major problems with education research over the past several decades.
Smith starts out commenting on the Nikole Hannah-Jones piece (I believe at this point there are three grown adults on the planet who have not expressed an opinion about the piece, and two of them are stuck under rocks) and then wades into a round of "do schools need more money" as a way of introducing his gripe, which is that when schools get more money, they spend it on teachers to try to reduce class size.
"I’ve been hearing all my life that smaller classes will improve the quality of education," he opines, then goes on to cite some contemporary praise for smaller class size. He cites a 2014 study and a 2013 program in Tennessee (Student/Teacher Achievement Ratio or STAR) and digs into a study of the entire country of Israel, where due to a rabbinical rule, class sizes were capped at 40 (like 40 students somehow qualifies as a small class in any year since about 1957).
The initial study showed "gains" but then a follow-up study showed no class size effect. Then he pulls up a couple of meta-analyses (a study of studies, for those of you who skipped this day in college). He quotes a study from Filges et. al.:
Overall, the evidence suggests at best a small effect on reading achievement. There is a negative, but statistically insignificant, effect on mathematics. For the non-STAR studies the primary study effect sizes for reading were close to zero but the weighted average was positive and statistically significant.
And more of the same, all looking at the effects of smaller class size on "student achievement."
Student achievement.
Here we go again. Hundreds of studies that accept Big Standardized Test scores as accurate, reliable, and valid proxies for the desired outcome of an education.
Do researchers use the proxy because they truly believe BS Tests are a valid proxy, because BS Tests are the easiest thing to find that comes in nice numbers, or because they know on some level that publishing a study that examines the effects of X on the scores from a single multiple-choice reading and math test given once a year would cause too many people to respond, "Well, who cares?" Or are they simply implicitly arguing that BS Test scores are, in fact, the whole point and object of education and nothing else matters?
Look, measuring education and student achievement and growth is really hard, because what we want for our children includes a whole lot of hard-to-measure stuff, not to mention things that don't really kick in or pay off until years later. Plus all the things that require a slow approach; can you measure how well a student can read and reflect and dive into a full novel by giving them a series of multiple choice questions about a 500-word excerpt? No, you cannot. But somehow many folks have come to accept that a tiny, not-very-valid measure of a tiny sliver of a tiny part of a whole human's education can be treated as a proxy for the whole big picture. And that leads to all sorts of bad conclusions.
This particular instance stands out because I don't think there's a single human being who remembers being in a K-12 classroom who does not understand that smaller class size is a plus. It guarantees students more attention, makes the teacher's job easier (both in teaching and management), and provides the student with a more human experience. There isn't a parent on the planet who, given the choice, would say "Don't put my child in the class with fifteen students-- I want her in the class with 45." There isn't a teacher on the planet who would say, "Yeah, could you just jam ten more students into each of my classes, because that will make my job easier." Need more? There's a whole organization (Class Size Matters) that can provide more.
You can line up studies that "prove" that class size doesn't matter from here to the moon and the eyerolls and muttered comments of "bullshit" will far outnumber the responses of, "Oh, well, I see. I guess I don't care how big the class is after all." Because no sensible human being would believe that class size doesn't matter. And when research is based on bad and incomplete data that causes it to arrive at literally unbelievable conclusions, it does not deserve to be taken seriously.
Smith goes on to make some valid points. Like good teachers are better than bad teachers, and there is not an infinite supply of good teachers. So if we hire a bunch more teachers in order to reduce class size, we will end up hiring less-than-great teachers. Fair enough. Though I will point out that putting a mediocre teacher in front of a small class gives you a better chance of success than putting them in front of a large class.
Maybe we could use the money to raise salaries and attract more good teachers rather than reduce class sizes, Smith suggests, and boy have we been here before. In 2015, Marguerite Roza and the Georgetown Edunomics Lab were trying to make a case for the Super Sardinemaster-- one great teacher in front of a couple hundred students. Way back in 2010, Bill Gates (who, I will remind you, has never been right about education) thought it sounded like a great idea. Reformsters like the idea because it's cheap. For a while, the Super Sardinemaster was imagined as a great teacher working over the internet with many students, but we ran a large scale experiment about that kind of teaching back in 2020, and mostly everyone hated it.
What many economists understand about education could be spread out on the head of a pin and still leave room for a flock of angels to hold a good-sized square dance. But they ought to understand that using bad data leads to bad results, and this is a particularly dramatic demonstration of that principle. Too much education research rests on a foundation of junk, no more reliably stable than a pile of sardines
No comments:
Post a Comment