> That sounds good in theory but research seems to suggest that homeschoolers overwhelmingly perform better on every single metric.

Being homeschooled and unschooled myself, I got to see that first hand in other families. But I don't think it actually means that homeschooling/unschooling is universally better: plenty of people try it, find it doesn't work for them, and quit.

My best guess is it increases the variance of results more than the average, with some kids doing very well on it, and some very poorly. While that's good for society, it doesn't mean one can entirely replace the other.