I hesitate to say this before having read the article (it's not out yet), but I think I figured out what's going on. This article in was published last Wednesday in EdWeek and describes a forthcoming journal article in which the author claims that smaller classes do not reduce achievement gaps. Meanwhile, another researcher who looked at the same data says that they do. I didn't have time to really think it through the first time I read it, but I think I see the difference now.
Both researchers look at data from the Tennessee STAR project, which is the only randomized trial of class size (students were randomly assigned to a class with either 13-17 or 22-26 students for a few years) and took place about 20 years ago. The author of this article, Spyros Konstantopoulos, says that the gap between the high and achievers was higher within small classes than it was in large classes. In other words, higher achieving students benefited more from smaller classes, so smaller class sizes do not impact achievement gaps.
Meanwhile, another researcher, Alan Kreuger, says that his research shows that lower-performing students and African-American students benefited more, and that, therefore, smaller classes reduce achievement gaps.
What? They're both looking at the same data. How can they reach different conclusions? Barring dishonesty or highly-technical formulas, here's what I think it is:
I think the answer is simple. In the newer article, he looks at the gap between the high and low achievers within each class, not the average score for each class. Meanwhile, the older study looks at the gap between the higher and lower performers across the sample. So the average score for lower performing classes could rise more than the average score for higher performing classes (many, if not most, classes are not extremely diverse -- they're within schools that are in wealthy suburbs or poor inner-city neighborhoods), meaning that the gap between high and low performing classes shrinks while the gap between the high and low performers within these classes actually increases. In other words, the lowest scoring students overall gained more than the highest performing students, but the highest-scoring kids within each class gained more than the lowest scoring kids within each class. So, in short, both of them are right.
If they're both right, to whom should we listen? Does reducing class size work or not? Well, you can argue it both ways. On the one hand, the society-wide "achievement gap" is really what we care about but, on the other, it seems that reducing class size doesn't reduce this in quite the way we'd anticipate. It seems that using the newest study to imply that smaller classes don't reduce the achievement gap is misleading, but that the way these changes affect distributions of achievement within classes offers food for thought. Of course, it would be nice to have a second study, one that's less than twenty years old, to compare to these findings.