Classroom effects were mostly nonsignificant

From Estimating classroom-level influences on literacy and numeracy: A twin study. by Katrina L. Grasby, Callie W. Little, Brian Byrne, William L. Coventry, Richard K. Olson, Sally Larsen, and Stefan Samuelsson. From the Abstract.

Classroom-level influences on literacy skills in kindergarten through Grade 2, and on literacy and numeracy skills in Grades 3, 5, 7, and 9, were examined by comparing the similarity of twins who shared or did not share classrooms with each other. We analyzed two samples using structural equation modeling adapted for twin data. The first, Study 1, was of Australia-wide tests of literacy and numeracy, with 1,098; 1,080; 790, and 812 complete twin pairs contributing data for Grades 3, 5, 7, and 9, respectively. The second, Study 2, was of literacy tests from 753 twin pairs from kindergarten through Grade 2, which included a sample of United States and Australian students and was a reanalysis and extension of Byrne et al. (2010). Classroom effects were mostly nonsignificant; they accounted for only 2–3% of variance in achievement when averaged over tests and grades. Although the averaged effects may represent a lower-bound figure for classroom effects, and the design cannot detect classroom influences limited to individual students, the results are at odds with claims in public discourse of substantial classroom-level influences, which are mostly portrayed as teacher effects.

I am not disputing their findings because they are common and frequently replicated.

Its just that I really don't want genes and randomness to be key determinants of cognitive outcomes but that is what most the evidence suggests. I want parents and teachers to matter more than we can prove they do. And I don't know how to reconcile my championed evidence-based decision-making when it collides with my deeply held emotional beliefs.

If I set aside my biases and assume that these findings are correct, then we are left with a massive disconnect. We pour inordinate amounts of money into schools and education training and ever-evolving new teaching practices. And they don't make a lick of difference if we are to believe these results.

IQ, personality traits, strong supportive cultural incentives for learning, a safe environment. Do that, and, this research suggests, you've done all you need to do to get the best results.

I agree that our educational institutions are hide-bound and more like sinecure-protecting guilds than paragons of education, but I do believe individual teachers can make a world of difference to individual children at particular moments in time. It just doesn't easily show up in the data.