I’m no expert on the sociological aspects of this, but do claim a bit of knowledge on statistics. And when a research finding contradicts everyday experience, i

I’m no expert on the sociological aspects of this, but do claim a bit of knowledge on statistics. And when a research finding contradicts everyday experience, it’s natural to be skeptical, so I looked at the paper in a bit of detail (https://www.pnas.org/doi/10.1073/pnas.2221311120 https://www.pnas.org/doi/10.1073/pnas.2221311120). While I do have other questions about the test setup, here’s the statistical analysis they did: They divided the students into quartile buckets based on prior knowledge on a subject, and then measured learning rates using some kind of online course system. What they found, in their own words, is that there’s only “small variability of student learning rate relative to variability in students’ initial knowledge”. In other words, and looking at their tables, there’s actually large variability in learning rate overall, but that’s almost perfectly correlated with initial performance. To be concrete, the improvement rate, using whatever measurement they use, is 2.56 for the top initial knowledge quartile and 1.70 for the bottom one. So top performers do indeed learn 50% faster. They hide this using lots of language about “little inter-quartile differences”, and then denounce all kinds of “problematic conceptualizations” that society has. When in reality, at the university level, this “initial knowledge” isn’t some independent random variable that you have to control for, but, as this study actually very clearly shows, the integral of the student’s learning rate over their lifetime.