Since we already know the answer, I guess I'll just expose my ignorance: what % of test score variance is explained by the home environment?
And while we are on the topic, what % of variance is caused by: IQ, non-IQ ethnic factors, school quality, teacher quality, non-home environment, etc? I.e., what are the principal components of educational outcomes?
Also, what are the "wrong questions" this article is "designed" to get people to ask?
Both of the sources I linked to are a good place to start in terms of understanding the numbers. Another good source is the Coleman Report, which is probably the most important education study ever done: http://www.icpsr.umich.edu/icpsrweb/ICPSR/studies/06389
The data was later reanalyzed using more modern statistical techniques, so I'm not sure if the link has the new analysis or the original analysis. Both come to basically the same conclusions though. The most famous finding of the study was probably that within-school effects are much greater than between-school effects, meaning that what track your kid is placed into inside the school has a much greater impact on their education than if they go to a 'good' or 'bad' school. The interesting thing is that this study was commissioned in order to find evidence that black schools were performing much worse than white schools in order to justify integration, but the study actually found the opposite so the government essentially tried to bury it. But it was actually the largest study ever conducted at the time, and I think it's still one of the top five largest to this day.
Forgive me, but I lack a copy of the expensive proprietary software necessary to open your second link. Could you just tell us what % of variance is explained by home environment?
(I'm also having a hard time finding it in your 1969 paper, but I admit I only skimmed the first 15 pages or so.)
The percent variance is different depending on what you want to know the variance of. It's higher for reading than for math, because kids in high-SES families spend much more time reading at home than they spend doing math at home. And even for literacy it depends on what you're measuring, e.g. it's different for vocabulary knowledge than for reading comprehension.
The reason I linked to the 1969 paper is that it was the first of its kind, so the easiest way to find the more recent research is to look for papers that cite it. A quick glance at their research shows that upwards of 80% of the variance in their reading metrics was due to differences in how students spent the four summer vacations. This isn't taking into account either the initial gap (100% from home differences) or the fact that home differences effect progress made during the year. Basically the amount of variance due to home differences is definitely over 90%, but there isn't enough information from this study to put an exact number on it.
Basically the only two things that have changed since 1969 are that A) poor people are much poorer and B) schools for poor people are worse. I don't know offhand how these two factors balance each other out, but I'd guess that overall the numbers are pretty similar.
Interesting - so if 80-90% of variance is due to home environment, it seems like schools are just a minor concern (assuming the remaining 10-20% is all due to schools). Improving schools by 10% only improves students by 1-2%! Unless such improvements are cheap, they are almost certainly not cost effective.
It also means that international comparisons of student performance are almost certainly worthless - we aren't comparing school performance, we are simply comparing student quality.
Incidentally, you are wrong on your changes since 1969. The poor have gotten vastly richer since then. In 1969, the bottom 10% of the country didn't even have flush toilets.
"Improving schools by 10% only improves students by 1-2%"
The fact that 90% of the gap between high-SES students and low-SES students is attributable to home factors doesn't tell us what percent of their total knowledge students gain at school verse at home.
And remember that percent is very different for high-SES and low-SES students-- the study says that high-SES students learn 3.5-4x times faster in school than at home. Whereas low-SES students learn 16x times faster in school than at home. And the vast majority of students are low- or middle-SES.
So while improving schools by 10% probably wouldn't be a good way to close the gap between high-SES and low-SES students, it might still be cost effective at increasing their total knowledge and ability.
Also, when I said the poor have gotten poorer, what I meant was that the gap between the rich and poor has gotten larger, because again that's what this study is about.
The fact that 90% of the gap between high-SES students and low-SES students is attributable to home factors doesn't tell us what percent of their total knowledge students gain at school verse at home.
Sorry, I guess I misunderstood you then. I asked "what % of test score variance is caused by the home environment", and you said "80-90%". So is it fair to say that we really don't know what factors predict student achievement, all we really have is a possible explanation of one particular delta?
In that case, the story we are discussing is perhaps more useful than you initially thought.
Also, when I said the poor have gotten poorer, what I meant was that the gap between the rich and poor has gotten larger, because again that's what this study is about.
This is also unclear. In terms of dollars, sure, but not necessarily in terms of living conditions. In 1970, the rich had servants while the poor had outhouses. In 2010, the rich have iPhones while the poor have Droids (sometimes even a dumb phone without data).
I think the point was to get an actual number. Pointing at a mountain of evidence and saying to go find the number yourself doesn't advance the conversation.
And while we are on the topic, what % of variance is caused by: IQ, non-IQ ethnic factors, school quality, teacher quality, non-home environment, etc? I.e., what are the principal components of educational outcomes?
Also, what are the "wrong questions" this article is "designed" to get people to ask?