Since then many professionals with more time and more money have had a look at the datasets. This essay reviews a number of them. Key finding: "Every study found something different, and it isn’t even close." In other words any positive effects were, at best, at the edge of detectability.
A longitudinal study in progress won't be ready for a few more years, but the interim results aren't promising. Tiny effects--maybe. Sometimes. When the Moon is full. It is frustrating because sometimes you seem to get good results, but in the big picture they wash out. Maybe that's just statistical fluctuations, or maybe there's some detail in the broad notion of "teaching" that we're missing.
I once asked whether the typical school structure might not be optimal for students with certain disadvantages. One problem often correlated with modern poverty is lack of home discipline. A school structure that relies on well-disciplined students won't work so well when they aren't. Some of our children did well with a fair bit of flexibility in scheduling, while others needed "a schedule so tight it squeaked." One teaching size does not fit all. What if we establish classes for the discipline-deprived?
Of course you have to be careful that lazy administrators(*) don't indulge in warehousing. And this kind of class is not a good place to put the violent, who probably can't be handled in school at all.(**)
(*) We only hear about the lazy, loony, crooked, or invertebrate. Presumably there are plenty of others that don't make the news. But we have to keep an eye on them just in case.
(**) I'm not suggesting that this is a good model, though I have heard of classes that might have benefited from the approach.