Wednesday, November 01, 2006

More musings on state IQ

Adjusting for the percentage of white children attending non-public schooling altered my IQ estimates modestly (they follow at the end of this post--the original estimates are here). The adjustment bumped up the correlation between McDaniel's estimates and my own from .962 to .964 and edged up the correlation between my estimates and that of the Vietnam Veterans from .52 to .53, as well as a miniscule increase in the relationship to Tickle's scores and the results from Project Talent (less than one one-hundreth in both cases).

But John S Bolton commented on something that had also given me pause--McDaniel gleaned his estimates from NAEP data on fourth graders from 1992 to 2005, and on eighth graders from 1990 to 2005. This might be inflating the scores of states that have experienced the greatest amount of demographic shifting over the last decade and a half. Presumably the Vietnam Vet and Project Talent scores would be similarly affected. Specifically, states that have seen the greatest influx of underachieving Hispanic immigrants should perform more poorly in my estimates made solely on 2005 data.

It's hard to discern much. Florida, New Mexico, and Texas fare better by McDaniel, but California, Arizona, Illinois, and Colorado do worse.

John also brought up the question of putative white racism. The relationship between one racial group's performance by state and the proportion of various racial groups in the state doesn't appear to be very robust across the board. But blacks in a state perform more poorly as the percentage of the state's population that is black increases (a state's black average score and the percentage of its population that is black are inversely correlated at a moderate .29). Is it irrational racism that is hindering blacks or is it that black underclass pathologies undercut already disadvantaged blacks even further?

Also of interest, black and white scores within a state trend in the same direction. The correlation between the average white and black score in a state is a firm .50 (excluding the aberration of DC, where stuffy whites do tremendously well). Better teaching methods increasing scholastic performance to the extent that's possible? The result of affluent, more gifted English lineage and blacks with a greater percentage of European ancestry concentrated in the Northeast?

1. Massachusetts -- 101.7
2. North Dakota -- 101.5
3. Vermont -- 101.5
4. South Dakota -- 101.3
5. Montana -- 101.3
6. New Hampshire -- 101.2
7. Minnesota -- 100.9
8. Wisconsin -- 100.6
9. Wyoming -- 100.5
10. Iowa -- 100.2
11. Maine -- 99.9
12. Idaho -- 99.9
13. Washington -- 99.9
14. Nebraska -- 99.8
15. Virginia -- 99.7
16. Kansas -- 99.7
17. Ohio -- 99.6
18. Colorado -- 99.4
19. New Jersey -- 99.4
20. Oregon -- 99.3
21. Delaware -- 99.1
22. Michigan -- 99.1
23. Utah -- 99.0
24. Connecticut -- 98.9
25. Pennsylvania -- 98.9
26. Missouri -- 98.7
27. Indiana -- 98.7
28. Alaska -- 98.7
29. Kentucky -- 98.3
30. New York -- 98.0
31. Illinois -- 97.9
32. South Carolina -- 97.7
33. North Carolina -- 97.6
34. Maryland -- 97.3
35. Texas -- 97.2
36. Rhode Island -- 97.0
37. Oklahoma -- 97.0
38. West Virginia -- 96.8
39. Tennessee -- 96.7
40. Arkansas -- 96.7
41. Georgia -- 96.5
42. Florida -- 96.2
43. Arizona -- 96.1
44. Nevada -- 95.1
45. Louisiana -- 94.9
46. California -- 94.8
47. New Mexico -- 94.6
48. Alabama -- 94.5
49. Hawaii -- 94.4
50. Mississippi -- 93.3
51. DC -- 87.9

(Human biodiversity2)


Steve Sailer said...

One interesting thing would be to look for states that do better or worse than you'd expect based on their overall demographics, which might say something about the quality of their schooling. For example, Texas looks pretty good to me in these numbers, so maybe Bush really did help the schooling? Or was a Texas advantage there already.

Hawaii seems pretty awful with whites and Asians performaing very badly.

I'm not exactly sure how to quantify this question, but it could be useful.


JSBolton said...

Maybe Texas, with its low-payout welfare system, and relatively unregulated business scene, attracts a higher average of Mexicans, and fewer low-end Central Americans in general.
It seems odd that the population shifts don't show as much as expected.
Maybe schools have been getting continually more efficient at excluding predictable low-scoring students from testing.

crush41 said...


Interesting observation regarding Texas. Indeed, Texas blacks, whites, and Hispanics all score higher than the blacks, whites, and Hispanics, respectively, in Arizona, Nevada, California, and New Mexico with the minor exception of New Mexico's blacks who have an estimated IQ less than one-tenth higher than that of Texas blacks.

This City Journal article by Jay Greene suggests in part what has always lurked in the back of my mind regarding NCLB--despite the emphasis it forces on the least endowed at the expense of the most gifted, at least it puts some pressure on school administrators and teachers to do something. I went through public schooling here in Johnson County, one of the wealthiest counties in the country (and so presumably enjoying one of the nation's better public school systems) and when I think back to my experiences, I'm reminded of such a stunning lack of discipline, especially in the non-honors courses, that little teaching was ever accomplished.

Of course the benefits of effective teaching, more discipline, and less intellectual profligacy in the classroom can only go so far, but it has to help to some extent.

I'll build a regression equation predicted scores based on racial composition and then point out where states 'should' fall given that racial composition and where they actually did with my estimates.


Could the Texas success be to some extent a legacy of Texas' historically large Hispanic population that came prior to the implementation of New Deal social programs, in that sharper, more entrepreneurial Hispanics were initially pulled into the state due to a lower tax burden (no state income tax, etc) and fewer state welfare benefits?

crush41 said...

That should read: "less than one-tenth of one IQ point higher than Texas' blacks".

Anonymous said...

I suggest comparing results to a map of ancestry by national origin. Most of midwest is largely german, Utah is largely English, Dakotas are largely scandinavian, etc.