By request. I probably should look up earlier statements before writing this, but I hope I am consistent. Consider an education intervention and a set of tests that it must pass. The intervention could be “more spending” or “method X used in the classroom” or “longer school days” or “charter schools” or what have you.
1. It should show a meaningful difference under experimental conditions, meaning that selection bias is eliminated.
2. The difference should persist, rather than fade out. If you show a difference in first grade but by third grade or fifth grade the experimental group is on on the same level as the control group, then there is fade-out.
3. The results should be replicated. One experiment that works one time does not count.
4. The intervention should be scalable. The intervention does not depend on a uniquely gifted teacher.
The Null Hypothesis is that no intervention passes all four tests.
Does this Null Hypothesis then imply that differences in performance are of a biological determination? If not, I would venture to say that if schooling policy can’t fix it, the social and biological home-factors might be places to explore new options.
Last time you presented your null hypothesis, I brought up Bloom’s Two-Sigma Problem, which seems to have some replication. That is, one-on-one tutoring combined with mastery learning produces much better results than having a class work on the same material at the same time for a set length of time that’s enough for some of the students to get it — but it’s not always a two-sigma difference.
His Developing Talent in Young People is not so quantitative, but it finds a similar pattern, where prodigies didn’t start of with prodigious skills, but they were allowed to pursue their interests at their own accelerated pace.
I agree that there is probably some fruit left to pick in the realm of moving away from one-size-fits-all approaches, and towards increasing individualized tailoring ,personalization, and quasi-apprenticeship approaches.
Actually, radical personalization is my guess for newly pickable economic fruit in general and across many fields and even including government. It was not recently low-hanging, but new IT is like a growing ladder that extends our grasp.
Individualized instruction like tutoring or one-on-one coaching is really an extreme form of ‘tracking’, and it seems that abandoning tracking was a mistake.
It’s probably a result of the social inertia of industrial-era society and also egalitarian ideology and equal protection jurisprudence that makes us instinctively and reflexively think inside the box of one-size-fits-all approaches, with uniform curriculum in classrooms with high dispersion of human capital.
Still, I don’t think this approach would pass Kling’s test #4 of scalability. I’m assuming his notion of “scalability” includes “reasonably feasible and affordable”, though I could be wrong about that.
At least, not without a new technology and a cheap and effective automated substitute, and a political environment that would tolerate its use.
The great value of Judith Rich Harris’ work is she clarified the role of “peer effects.”
Peer effects are not just networking–they often rely on imitating “high status peers” which often means people a few years older, or people who are really good at something.
Among other things, schools create or foster peer groups.
Thus it’s not just genetic endowment, subject (or not) to epigenetics.
You don’t mention the *cost* of the intervention. Assuming the Null Hypothesis, we obviously don’t want *more spending* or any other costly intervention, since these would all fail to produce positive results. What we want is *less spending*–in fact, *zero spending*, giving us the same pedagogical results we are now getting, but for free!
So, what’s the real structure behind the hypothesis?
1. Current education system is optimal?
2. Current system is close to optimal up to noise?
3. The noise is so large that we can’t tell how close we are?
Thanks for clear Null Hypothesis phrasing!
I’m pretty sure tracking works for all kids living with their married biological parents, but that’s not really an option today.
Personalized AI tutor-bots are coming, and will succeed for those kids willing to do the work of learning. Partly by finding ways of keeping the kids interest/ attention, while they learn (eg fun/ play), but with measurable increases in test scores.
http://mindspark.in/ << from MR, (Alex, not Tyler) looks ready, finally, to be better than Null. Tho even if it IS better, it will take some time to validate and replicate and scale.
AI scales very rapidly, tho — sort of as fast as facebook or google, with HW mostly, once the AI software has been trained.
My kid was really into Prodigy(tm) and it was pretty impressive.
Why do we need #2?
The others seem sufficient. Requiring persistence seems contrary to experience. E.g. ask an adult, “can you fight?”, they respond, “I took a martial arts class in 3rd grade.” Do you think they can fight?
Point #2 is for those interventions like universal pre-school, where studies show Head Start really helps students, but then there is no significant difference by 3rd grade. The intervention is to give a permanent boost in academic results, to be built on.