01 Oct 2005 - 00:56

Samuel Freedman on the achievement gap in NYC

eduwonk says The Achievement Gap in Elite Schools is must reading. Get there soon, because the TIMES keeps articles posted for 7 days & then they're gone.

update

OK, maybe it's not so hot after all.

Back to main page.

---

---

Comments

After entering a comment, users can login anonymously as KtmGuest (password: guest) when prompted.
Please consider registering as a regular user.
Look here for syntax help.

---

It is a dirty little secret that the so-called elite school districts fare just as poorly with the disadvantaged kids as the bad school districts. And if you think about it, it makes sense.

Most school districts use the same kinds of cognitively unsound curricula and pedagogy. If anything, the elite school districts are enamored with constructivism than most. If we know anything, it is than constructivism and its associated nonsense is an inefficient means of imparting knowledge. It is the disadvantaged kids that are most damaged by inefficient teaching.

This is because poor kids tend to be less smart than the rich kids. General intelligence correlates with socio-economic status. See Luninski, p. 171 Fig. 2. (i.e., poorer kids tend to be not as smart as richer kids). It is more difficult to teach kids who aren't smart. Accordingly, it is more difficult on average to teach poor kids (who happen to be not as smart and have less familial support). Thus, it is the poor kids that are most damaged by inefficent constructivism.

Since constructivism is rampant in both the elite and failing school districts, is it so surprising that both perform just as poorly with their most difficult learners? How much difference can a good teacher make if he is saddled with a bad curriculum? This should also put to rest the pernicious myth that money makes a difference. The elite districts are swimming in money.

Every scientifically valid study I've ever seen says that direct instruction is more effective than constructivism. Put simply, it is more efficient. This goes for poor kids, rich kids, dumb kids, and smart kids. Across the board.

In fact, the data shows that the elite school districts are no better than the failing ones in educating children. Any children. Once you factor out the socio-economic effects, most school districts operate in a narrow band of mediocrity. This is what is expected when an industry is beset with inefficiencies, like the education industry.

There are 501 school districts in Pennsylvania, where I live. Standard and Poor's makes available all the student performance data (reading and math proficiency) for the state, including socioeconomic data (median family income and parental education) and financial data (total instructional dollars spent). Using Excel, I graphed student performance versus family income, versus parental education, and versus instructional dollars spent.

For both the family income and parental education I got the almost the same correlation that Lubinski got -- about .4 which is a medium correlation. The data was closely clustered around the regression line. This tells us that socioeconomic status is probably the largest factor in determining student performance. (If the data points were more spread out from the regression line then we could conclude that other factors are also affecting performance, such as school effects.) Again, this is what we expect to see. An inefficient curriculum is going to have little effect on student performmance. The family effects will play a much bigger role.

In contrast, the graph of performance vs instructional dollars spent yielded an effect size of less than .1 which is close to random. You can spend all the money you want on an inefficient curriculum, it won't make much of a difference.

One question that remains is what kind of effect size would we get if schools actually taught the kids facts? E.D. Hirsch makes the case that we should see a .8 correlation (using the general information vs intellegence correlation line in the Lubinski graph) or twice as large an effect as the .4 socio-economic effect. Perhaps this is what explains the performance gap between Singapore and the US, at least in math.

This is a very off the cuff analysis and I'm no statistician. I wonder what a real statistician could do with this, especially with better data.

-- KDeRosa - 01 Oct 2005

---

The achievement gap can be eliminated in a few ways:

1. Make the test so hard that no one passes. political suicide.

2. Make the test so easy that everybody passes. This seems to be the way states are heading to comply with NCLB. Like TX and PA.

3. Teach the students better so that they all pass at a higher standard. This requires hard work, so don't expect it anytime soon. Plus, technically the gap will still be there because I know of no test in which the dull kids outperformed the smart kids.

See here for an explanation of the statistical chicanery.

-- KDeRosa - 01 Oct 2005

---

Make the test so easy that everybody passes. This seems to be the way states are heading to comply with NCLB. Like TX and PA.

Haven't read your first comment yet, but this option is the one I worry about.

-- CatherineJohnson - 01 Oct 2005

---

Then be prepared to be afraid, very afraid.

2005 Pennsylvania Achievement test (PSSA) % of 8th graders statewide who are advanced or proficient: 63%

2003 NAEP % of PA 8th graders who are proficient or advanced: 30%

And, we all know how difficult the NAEP test is.

Not the same cohort, but there is only a little variation between cohorts if you look at the longitudinal data.

-- KDeRosa - 01 Oct 2005

---

2005 Pennsylvania Achievement test (PSSA) % of 8th graders statewide who are advanced or proficient: 63%

Hey!

Success!

-- CatherineJohnson - 01 Oct 2005

---

wow

63% versus 30%

They should be ashamed.

-- CatherineJohnson - 01 Oct 2005

---

Hoxby says everything should be aligned to NAEP.

THEN (I'm saying this) every parent should have a copy of NAEP at home.

-- CatherineJohnson - 01 Oct 2005

---

This may be thinking 'too big,' but I feel parents need a whopping big national organization of their own that's equipped to give norm-referenced standardized tests outside the context of schools, government, etc.

We should be able to have our kids tested--easily, without a huge amount of rigmarole--outside the school system.

Lone Ranger left a Comment about the ITBS, which we can give our kids.....I found only one web site that will allow any parent to administer the test (the other wanted homeschooling credentials, etc.)

I may do it.

It's still a lot of rigmarole.

-- CatherineJohnson - 01 Oct 2005

---

It would be nice if the collegeboard people had an exam like the AP exam that could be administered at the end of each grade.

The best part of the AP exams, besides the rigor, is that it sets a goal. When the goal is not met prents know instantly that things have gone wrong and can then bitch to the school administration who then can act to improve things (hopefully) or the parents can step in. Your classic feedback loop.

As long as the collegeboard people do a good job setting the goals each year, we might get a more coherent curricula at least for the AP track people (which should include all kids in grade 1).

-- KDeRosa - 03 Oct 2005

---

this is a great idea. anyhow, it sure
sounds like an outfit i'd like to work for.

This may be thinking 'too big,' but I feel parents need a whopping big national organization of their own that's equipped to give norm-referenced standardized tests outside the context of schools, government, etc.

-- VlorbikDotCom - 03 Oct 2005

---

"Haven't read your first comment yet, but this option is the one I worry about."

It's bound to happen as NCLB is currently written. Otherwise a 100% pass rate is impossible to achieve.

-- CarolynJohnston - 03 Oct 2005