Benjamin Bloom, the guy who coined Bloom's Taxonomy, wrote an article called The Two Sigma Problem. Thanks to Shaun Doherty for the article.
U of Chicago studies showed that tutoring created a ginormo, 2 standard deviation improvement in learning. "Good teaching," which they call "mastery teaching," created a 1 standard deviation improvement. The baseline was regular ol' teaching, I suppose.
Bloom wondered in 1984: can we devise teaching (i.e., a whole group at a time) methods that generate the same level of learning as tutoring?
We're obsessed with the same question today. Bloom's team thinks if you combine a bunch of stuff, you can get there. Like a drug cocktail. But so far as I can tell, they don't have evidence. They just argue that the results of 3 or 4 different methods are cumulative/additive. Sometimes stuff is additive. Sometimes it's not.
Another way to consider the problem: what if, instead of a system where we try to make all teachers "mastery" level, we provide a lot more tutoring?
Tutoring is too expensive. That's the usual response.
But shouldn't the question be: How much learning is generated per dollar?
A median kid in a typical American school probably gets 1,000 hours per year of class time (including some that's probably "mastery" level).
And while there's no good data on this that I know of, that same kid also gets about...10 hours per year of tutoring? That's 20 minutes per school week over the year.
Is that really the right ratio? 1,000 hours of class time and 10 hours of tutoring time?