Friday, November 17, 2006

State NAEP performance with race considered

Comparing scholastic achievement across states is inherently flawed if the demographic composition of the student body isn't taken into account. That flaw isn't limited to the educational realm, of course--it holds for criminality, income, and a host of other social measures. It also skews international comparisons.

Considering racial characteristics isn't enough to make perfect comparisons between states, but it goes a long way in predicting crime rates (the correlation between violent crime and race as defined by the five categories of non-Hispanic white, black, Hispanic, Asian, and Native American is .85). For estimated average IQ by state, race correlates at .85 as well. In other words, the racial composition of a state 'explains' 72% of its performance.

The correlation is vigorous but not perfect. So how do the various states fare given their respective compositions? I built a regression equation taking into account only said demographic data by state to come up with estimated state scores, then compared those to the actual results. Following is that comparison--a positive number indicates that the state scored X IQ points higher than would be expected by racial composition alone, a negative number indicates a score X points lower than would be expected:

1) Virginia -- 2.57
2) Massachusetts -- 2.51
3) Delaware -- 2.25
4) New Jersey -- 2.20
5) Maryland -- 1.60
6) South Carolina -- 1.55
7) Minnesota -- 1.48
8) Texas -- 1.38
9) Colorado -- 1.17
10) New York -- 1.15
11) Wisconsin -- .90
12) North Dakota -- .84
13) North Carolina -- .78
14) South Dakota -- .77
15) Montana -- .77
16) Georgia -- .73
17) Illinois -- .69
18) Conneticut -- .65
19) Ohio -- .57
20) Michigan -- .57
21) Washington -- .55
22) Alaska -- .50
23) Kansas -- .40
24) Nebraska -- .29
25) Vermont -- .10
26) New Hampshire -- .07
27) Pennsylvania -- .02
28) Florida -- (.13)
29) Idaho -- (.16)
30) Missouri -- (.23)
31) Wyoming -- (.25)
32) Louisiana -- (.25)
33) Oregon -- (.27)
34) Iowa -- (.27)
35) California -- (.63)
36) Hawaii -- (.71)
37) Arizona -- (.94)
38) Indiana -- (.95)
39) Oklahoma -- (1.08)
40) Maine -- (1.13)
41) Utah -- (1.21)
42) New Mexico -- (1.21)
43) Tennessee -- (1.26)
44) Arkansas -- (1.45)
45) Kentucky -- (1.47)
46) Mississippi -- (1.58)
47) Rhode Island -- (1.75)
48) DC -- (1.93)
49) Alabama -- (2.09)
50) Nevada -- (2.25)
51) West Virginia -- (4.09)

Erudite Massachusetts performs well even when its propitious demographic composition is weighed against it. Virginia and Colorado, states that Steve notes tend to be the pride of the red states, both do well. What about the Bush Texas miracle? Does the threat of accountability have a modestly positive effect on average test scores? The Northeast generally does pretty well, with the exception of Rhode Island. The destitute Appalachian state of West Virginia rounds out the bottom.

The military's test battery not only helps create the racial harmony and solid performance that it is renowned for, it also leads to army brats a notch above their civilian peers. The DoDEA (the school system for the children of military personnel overseas), if counted as an individual state, is 3.35 points higher than predicted, besting the rest of the country.

5 comments:

JSBolton said...

The unsurpassed results of the DoDEA in comparison to states, tells where the problem lies.
It has to be the children of the parents with IQ's below 90, or whatever the cut-off for foreign-stationed military with school-age children may be.

faq said...

I wonder what the relationship is between expenditures and results. It should be easy to find out. Race is probably a much better predictor than dollars, yet how many times is THAT brought up as a way to improve test scores?

crush41 said...

Faq,

When I get back home I will. But I've messed around with the numbers before, and expenditures and performance show no statistically sound relationship at all. Adjusted for race, money helps a bit for all groups except blacks, although this appears to be mostly the effect of higher costs (and standards) of living.

Steve Sailer said...

I wonder what's going on in South Carolina? Most of the other top states have lots of colleges and/or jobs to bring in bright people, but South Carolina might be doing well just on good schooling techniques.

crush41 said...

South Carolina does particularly well with its black population, although its Hispanics and whites both perform better than expected, especially for the South. That probably has something to do with Fort Jackson (DoDEA operates mostly outside the US).