Saturday, February 14, 2015

State IQ estimates, blacks only (2013)

IQ estimates similarly arrived at through the utilization of 2013 NAEP results in math and reading among black 8th graders follow. The scores for both tests are on a 500 point scale, with a presumed standard deviation of 40. In the proceeding table, these are converted into IQ estimates with a mean of 98--corresponding to the national average NAEP scores of 283.62 for math and 266.02 for reading--and a standard deviation of 15. The math and reading scores are given equal weighting, except in the cases of Hawaii, Maine, and South Dakota, for which only reading, math, and reading scores are available, respectively. There are no estimates for Idaho, Montana, New Hampshire, Oregon, Utah, or Wyoming due to insufficient sample sizes in those states.

I dispense with the tradition of color-coding states in the table based on how their populations under consideration voted, for reasons made clear here:

1. New Jersey94.7
2. Hawaii94.5
3. Massachusetts94.2
4. Maine93.6
5. Maryland93.5
6. Washington93.1
7. North Dakota93.1
8. Texas92.8
9. Alaska92.4
10. South Dakota92.1
11. North Carolina91.5
12. West Virginia91.5
13. Florida91.3
14. Delaware91.2
15. Connecticut91.2
16. Vermont90.9
17. Virginia90.9
18. Georgia90.7
19. Arizona90.4
20. New York90.4
21. Ohio90.4
22. Rhode Island90.1
23. Pennsylvania90.0
24. Kansas89.9
25. Indiana89.9
26. Nevada89.8
27. Minnesota89.2
28. South Carolina89.1
29. Tennessee89.1
30. Kentucky89.1
31. Colorado89.1
32. Illinois89.0
33. Missouri88.6
34. California88.6
35. Louisiana88.5
36. District of Columbia88.3
37. New Mexico88.3
38. Iowa88.2
39. Nebraska87.9
40. Oklahoma87.8
41. Arkansas87.3
42. Michigan86.9
43. Mississippi86.2
44. Alabama85.8
45. Wisconsin85.3

A map of the same. The darker a state's shading, the higher its black IQ. Yellow indicates insufficient data:

Racial relations in the US military are generally regarded as being better than they are in the civilian world. This is potentially an important reason why that is the case. Hawaii's strong showing is more evidence of just how authentically black Barack Obama is!

The relatively high scores will jump out immediately to those who are familiar with the history of differences in results on IQ tests by race. The estimated national average IQ estimate for black Americans using the methodology described above is 90.1. Traditionally, US blacks have consistently scored around 85, and the gap between whites and blacks has held tenaciously at one standard deviation, so these estimates appear rather high. A few possibilities as to why:

- The black/white IQ gap may be narrowing. Flynn and Dickens argue that blacks have gained five or six points on whites over the last three decades, presumably spurred in part by better access to nutrition and healthcare. If accurate, that would put contemporary average African American IQ at 90 or 91, just as my estimates did.

- The rate of interracial marriage and procreation has increased over time. Richard Lynn marshals a good deal of evidence showing that racial 'hybrids' tend toward IQs that are an average of the two groups their parents represent. This should work to attenuate the gap by slightly lowering average white IQ and raising average black IQ a bit (the latter should rise more than the former falls due to sheer size).

- I built the formula for the estimates using Richard Lynn's data on international academic and IQ test results, so my equation is linear. Although Lynn believes that adjusting for attenuation yields a correlation between test scores and IQ of 1 (a perfect relationship), I'm inclined to assume that the correlations for IQ and math (.87) and IQ and science (.81), even adjusted for attenuation, are still not perfect. And because my formula is linear, moving further out from the averages if anything underestimates the magnitude of the true deviation. So depending on the true strength of the relationship, the estimates might be inflated by a couple of points (although I think this would only be on the order of a point or two).

- Differing rates of truancy by race may be artificially giving blacks a boost relative to whites. Assuming that children frequently absent from school generally come from more chaotic, less endowed households, it follows that students who are the least likely to show up on test day are among the poorest performers. And the absentee rate for black children (24%), defined as missing an average of three or more days per month, is greater than the rate for white striplings (19%).

- According to the late Philip Rushton, black adolescents develop faster than white adolescents (who in turn develop faster than Asians). Consequently, young blacks achieve greater parity with their white counterparts than they do as both groups age. Indeed, very early on black children may be more intelligent than whites and Asians due to relatively accelerated physical development. This meshes with the tentative assertion that the gap may have narrowed among children but doesn't appear to be shrinking among adults. To the extent that differing development levels affect average scores by race, IQ tests based on an average white IQ of 100 given to children will tend to underestimate the eventual cognitive capacity of Asians and overestimate it for blacks.

Parenthetically, the correlation between a state's white-black IQ gap and its propensity to vote for Romney in 2012 is .47 (p-value =  .0013). Unfairly, I can imagine Rush Limbaugh pointing this out as evidence that the Democrats benefit from keeping blacks down!


Anonymous said...

The average SDs of both math and reading scores are in the low 30s. You understate the gaps by assuming that the SDs are about 25 percent larger than they actually are. The actual SDs can be easily obtained from the NAEP Data Explorer (in the Edit Reports tab, click on Statistics Options, and then Build Reports).

Saint Louis said...

Re: Your point about racial mixing. I imagine it serves almost entirely to increase black IQ rather than lower white IQ since almost every mulatto is probably counted as black by the testers.

Audacious Epigone said...


37 for math and 34 for reading, so about a 10% overestimate. I'd never gone in that way, I've previously always used the state comparisons tool. Thanks, I will update both again.

JayMan said...

Great work!

"The relatively high scores will jump out immediately to those who are familiar with the history of differences in results on IQ tests by race. The estimated national average IQ estimate for black Americans using the methodology described above is 90.1. Traditionally, US blacks have consistently scored around 85, and the gap between whites and blacks has held tenaciously at one standard deviation, so these estimates appear rather high. A few possibilities as to why"

Or perhaps the true scores are these minus 5 points across the board?

But, always better to report what you find, and figure out what it up later.

Anonymous said...

New Jersey's high score is very surprising


Audacious Epigone said...


Yes, valid point and one I should've included. That parsimonious explanation might well be the correct one.

Audacious Epigone said...


Generally speaking, the fewer blacks a state has the better its blacks tend to do. Maryland is an exception, but it has a lot of suburban DC upper middle class blacks in it who are probably pulling up its numbers. New Jersey is a less obvious one.

TangoMan said...

A while back I was playing around with the NAEP database and saw some interesting results when I loaded the variables for the mothers' educational status and the students' eligibility for free lunch, and compared across race.

There is a lot of info buried in the cross-tabs.

Anonymous said...

group administered tests show less "gap" than individually administered tests because not everybody shows up to take it. those least likely to show up for all testing sessions tend to be those most apt to get low scores. figuring out who skipped the test can add insight into the scores.

Anonymous said...

so if it was not for whites, blacks would still be swinging in trees and chucking,blacks need whites ,whites DO NOT NEED BLACKS

Audacious Epigone said...


I'm looking through the variable definitions and I can't find how exactly how "parental education level" is defined. Is it the highest attainment of either parent, or just the mother's educational attainment?

Audacious Epigone said...

Never mind, got it. It's the highest level attained by either parent.

Anonymous said...

"so,blacks need whites ,whites DO NOT NEED BLACKS"

If that's a stunning revelation to to you(Whites don't need Blacks) then by your OWN theory you must be tired from swinging in all of those trees. Slow down. Take in the Stormfront and have a banana

Succantation said...

It's not necessarily about blacks or whites or asians. You should be more specific.

Geographically isolated / reproductively isolated populations with varying levels of genetic drift/rate of mutations/differential selection pressures lead to these results.

Maybe Black-African-heritage or latitude X, longitude Y, area N, width U-heritage type of humans would have more predictive IQ power, since not all whites nor all asians are equal with respect to one another on all continents.

Higher selection pressures, lower tolerances/safety nets, outbreeding and genetic engineering with CRISPR/gene therapy are some good solutions. You can also impose more burdens
through subtle political sabotage policies in the name of good by "politification of terms" or "dressing-up quick-do accounting under the public scheme mantra" via economic policies to reduce the fertility rate or survivalship of worser variant alleles that cost humanity rather than benefit it.