Friday, August 21, 2009

Does Homeschooling Totally Rock?

Via Elusive Wapiti, apparently:

The Results

Overall the study showed significant advances in homeschool academic achievement as well as revealing that issues such as student gender, parents’ education level, and family income had little bearing on the results of homeschooled students.

National Average Percentile Scores

SubtestHomeschoolPublic School
Social Studies8450
  • a. Core is a combination of Reading, Language, and Math.

  • b. Composite is a combination of all subtests that the student took on the test.

There was little difference between the results of homeschooled boys and girls on core scores.

  • Boys—87th percentile

  • Girls—88th percentile

Household income had little impact on the results of homeschooled students.

  • $34,999 or less—85th percentile

  • $35,000–$49,999—86th percentile

  • $50,000–$69,999—86th percentile

  • $70,000 or more—89th percentile

The education level of the parents made a noticeable difference, but the homeschooled children of non-college educated parents still scored in the 83rd percentile, which is well above the national average.

  • Neither parent has a college degree—83rd percentile

  • One parent has a college degree—86th percentile

  • Both parents have a college degree—90th percentile

Whether either parent was a certified teacher did not matter.

  • Certified (i.e., either parent ever certified)—87th percentile

  • Not certified (i.e., neither parent ever certified)—88th percentile

Parental spending on home education made little difference.

  • Spent $600 or more on the student—89th percentile

  • Spent under $600 on the student—86th percentile

The extent of government regulation on homeschoolers did not affect the results.

  • Low state regulation—87th percentile

  • Medium state regulation—88th percentile

  • High state regulation—87th percentile

The study's significant blind spot is the failure to control for race -- although who can blame them? Why kick that hornet's nest when there's nothing in it for them?

Another useful control would have been marital status. Homeschooling families are almost always two-parent families, so the relevant comparison would have been with other two-parent families. There is no indication that this kind of data on the marital status of the public school control group was available to the researchers. This probably inflated the homeschool performance advantage.

On the other side, as near as I could tell, the categories the researchers did consider only applied to homeschooling families, not to the "control" group of non-homeschooled children. But this only understates the homeschool advantage. Had non-college-educated homeschool families been compared only with non-college-educated public school families, for instance, I would expect their percentile advantage would have been even more than it already was.

I'm a big fan of homeschooling, obviously, but the advantage shown here is so large, and the variation is so small, that it seems pretty obvious that there must be some control category that should have been applied, but wasn't. I just don't know what it is.

Still, though, it's heartening to see team homeschool doing so well.


Elusive Wapiti said...

Phi, thanks for the link!

"The study's significant blind spot is the failure to control for race"

I've never seen or heard of a non NAM homeschooler.

"Homeschooling families are almost always two-parent families...This probably inflated the homeschool performance advantage. "

No doubt. I wonder what proportion of the performance data is due to the efficiencies/advantage of homeschooling, and what proportion is simply a function of having a dad in the home?

Here's an anecdote for you. My PEW is withdrawing S1 and S2 from a Catholic private school and enrolling them in a public school against my will. As a response, I proposed to my attorney that the decree be modified to permit 60/40 visitation (in my favor), and Mrs Wapiti (who is a Cal-credentialed sped teacher) and I homeschool S1 and S2.

You know what he said?

"The judge will laugh you out of court for so radical an idea".

It's a shame when it is seen as radical to homeschool your children when the advantages are obvious.

Trumwill said...

The smallness of the variation is startling. I would have expected it to be larger if only because the rationale for upper-class people to homeschool seems to be more achievement-based and for lower-class to be more value-based. Goes to show what they say about making assumptions.

The only hesitation I have about it is that it seems likely that a lot of parents that homeschool, regardless of their own class, are also more likely to have their children perform well in public schools. Self-selection and all that. Take the same parents and the same kids and put them in public schools and I think the gap would close considerably.

As far as public policy is concerned, this doesn't matter. What matters is that kids that are homeschooled perform well across-the-board. As far as Clancy and I having to make a homeschooling-or-not decision for ourselves, though, knowing how these kids would do in public (or non-elite private) school would be helpful. Unfortunately, self-selection in cases like this is pretty hard to control for.

Anonymous said...

How about controlling for parental IQ?

- Thursday

Brandon Berg said...

Headline: Study Proves Advantages of Class Size Reduction.

Φ said...

It's a shame when it is seen as radical to homeschool your children when the advantages are obvious.

Radical in two ways: As in "unusual" (relatively speaking) and as in "subversive" (to the educational cartel and its Leftist indoctrination).

Interestingly, I've always had one or two other homeschool dads among the techie types I work with.

[S]elf-selection in cases like this is pretty hard to control for.

This is an excellent point. I expect many if not most parents self-select the educational arrangements that optimize their children's performance.

How about controlling for parental IQ?

Parental IQ, by itself, would be a proxy for student IQ. It also lies at the heart of the racial differences in the study's samples. It would be interesting to learn that parental/student IQ of homeschool families lies at the 80th or so percentile of public school families independent of race.

Study Proves Advantages of Class Size Reduction.

It would also have been interesting to control for family size. My impression is that large families are overrepresented among homeschoolers.

Φ said...

An additional thought, inspired by Brandon's comment: it may be that the mechanism by which small class-size, and especially homeschooling, improves educational outcomes is that individualized instruction allows students to learn at their optimum pace. We have long recognized that teachers tend to teach to the median student's ability to assimiliate knowledge. Thus, the above-average children in a class are learning at a slower rate than they would otherwise be capable.

Similarly, while some below average children rise to the challenge of the class material by putting forth extra effort, many become overwhelmed. They find themselves listening to lectures on material D and E, while not yet having learned material A, B, and C, even though this material was necessary to the apprehension of D and E. What starts as a small skills deficit snowballs into a large one.

I've been in both of those categories in my academic career.

Obviously, tracking alleviates this difficulty, but not entirely. The problem undermines the case for the industrial model of education we have embraced since at least John Dewey.

Brandon Berg said...

A corollary of this, if you're correct, is that the benefits of class size reduction are contingent to a considerable degree on the use of tracking.

Anonymous said...

I agree with Thursday, the best possible control would be by parental IQ. Controlling by the student's IQ would be unfair to homeschooling, because IQ is not entirely genetic---only from 40-80% so according to the best current research. So homeschooling enriching the child's environment MIGHT be associated with a higher measured IQ for the child relative to what he would have had if he had been a public school kid. The ideal would be if we could control by both parent's SAT scores, normed according to the date taken.
Another question that would be interesting to investigate would be this: Do homeschool kids under or overperform their IQ scores?