*increase*in the amount of resources diverted to children.

But does money actually do anything? CATO's David Salsibury points out that private schools spend half the amount of money per student as public schools do, yet they see consistently better results. Part of this can be attributed to special needs children, who cost roughly twice as much as average students and make up around 13% of the public student body. The profligate spending on special needs is another issue entirely, but it does not come close to explaining the entire gap. Top-heavy administration is another qualm. But according to NCES data, these costs only represent 5% of the total public school budget, compared to almost 70% going to "instructional services." We could stand to cut out some of the middle-management fodder, but that's relatively innocuous as well.

So where is the explanation? The mind police might not like the suggestion, but it appears to reside largely in the children themselves. Looking at the four chief demographic categories in the US (white, black, Hispanic, and Asian) shows that there are wide disparaties between children's performance based on racial categories. More interestingly, the way these groups appear to benefit from added education dollars varies substantially.

Using 2003 NCES eighth grade reading scores for each state and DC the data was compiled in Excel (by a state's overall average as well as by demographic categories) alongisde the state's expenditure per student and average teacher's salary according to NEA statistics. Finally, there is a cost of living index provided by state governments (there were two states that did not have data available for CoL so I estimated them by using income per capita by state and averaged it to that of the surrounding states). The findings follow.

Not surprisingly, neither expenditures per student nor teacher's salary had anything to do with test performance for all groups without a cost of living adjustment. The r-squared value was a risible .001 (.20 is considered a low correlation, .40 is moderate, and .60 is high) and the significance factor was .87 (essentially meaning that there was only a 13% chance that these two variables might have anything to do with test scores--generally the significance factor must be less than .05 or sometimes more liberally .10 to lend any credence to the validity of a possible correlation).

When CoL is taken into consideration, the picture becomes slightly more clear, although there is still nothing solid. The significance factor comes in at .12 and the r-squared at .08.

Thus, the right has ample evidence here that increased expenditures on students and teacher's salaries do nothing to increase test scores. The left could be correct, however, that more money might help. That is, it may help at least some groups. But given the cultural Marxism (ie political correctness) that infests public dialogue, the left is unlikely to take that route anytime soon. We will bravely forge ahead, nonetheless!

Taking white students only, the picture changes drastically. Adjusted for cost of living (as all the subsequent analyses are), expenditures per student show an r-squared value of .20 and an astounding significance factor of .001 (99.9% chance that the results are accurately correlated to spending). For roughly every $7 spent per student, white test scores increase by a point on average. Teacher's salaries initially seem to be marginally beneficial as well with an r-squared of .07 and a significance factor of .06 ($50 more per teacher "brings" a one point test increase). When the two variables are taken together, however, virtually all the benefit is shown to come from student expenditures. A plausible reason for this may be that districts that spend more money per student are likely to see some of that increase directed towards teacher's salaries even though the increased salaries apparently provide little benefit in and of themselves.

Now to Asians, who are in many ways beyond education very similar to whites. Because data were only available for 23 of the 50 states on Asians due to numerical paucity, the standard error for the category was 23 (compared to 4 for whites). Basically, these data are not as reliable because the sample size is considerably smaller. Nonetheless, strong indicators emerge. For student expenditures, the r-squared was .13 but the significance factor was .10, putting it at the outer-edges of a 90% confidence interval. Teacher's salaries showed an r-squared of .16 and a significance factor of .06. Given the small sample size, it is reasonable to assume that teacher's salaries are correlated with Asian student performance, albeit only modestly.

Blacks: For student expenditures the r-squared was a negligible .04 with no signficance (.20). Teacher salaries showed slightly less meaninglessness, although it appears money again does nothing beneficial for blacks. The r-squared was .06 and the significance factor was .14. Taken together, there was even more confusion: the significance rises to .25.

Hispanics: student expenditures had an r-squared of .01 and a significance factor of .54. With teacher's salaries r-squared became .09 but the significance became an interesting .07. With a sample size of only 37 and a standard error of 10, that catches attention. Roughly a $22 up in teacher's salaries buys a one point increase in test scores for Hispanics. Although dollars spent per student has nothing to do with the success of Hispanics, teacher's salaries seem to modestly correlate with Hispanic success.

For comparison, using a 95% confidence interval for whites and a 90% confidence interval for Hispanics, blacks, and Asians, the "cost" for one extra test point breaks down like this:

Whites: $7 per student

Hispanics: $22 per teacher

Asians: $3.50 per student as well as $5 per teacher

Blacks: N/A

Fascinating differences. Extrapolating, we see that more money means more results as average IQ increases. Based on Weschler IQ tests it is estimated that the average IQ for (all US) Asians is 106, whites is 101, Hispanics is 91, and blacks is 87. In the investment world, the better the business the more return the investor gets for his buck. Education probably works in a similar fashion. This does not bode well for the narrowing of educational attainment by individuals and by various groups displaying differences in cognitive ability. Of course, that is likely to be the great irony of globalization--as the playing field becomes increasinly leveled, the inequities in human accomplishment will become perpetually more pronounced, because environmental effects will become more and more controlled for, leaving nature as the chief variable.

Academic paths should be determined by early testing. A one-size-fits-all educational curricula is foolish--less intelligent children should spend less time in high school and be taught vocational skills while brighter kids should have access to intellectually stimulating material as early as they can handle it (high school-level courses for sharp 12 year-olds, college-level material made available in high-school, etc). More resources should be spent on those with the most intellectual potential. They should not be handicapped by receiving less funding than special-needs children as an artificial means of leveling (by bringing everyone to the lowest common denominator). Privatization of education would likely make this a reality (think private Ivy League schools like Harvard that stringently test for IQ versus state universities that do with much less restriction versus community colleges that basically let anyone in who has the drive to learn).

The data needs to be refined by including state population so that Montana and New York scores are given weights based on the number of students they contain rather than treating them as individual entities of equal weight. Yet average scores (by race) vary little between states, so this is unlikely to shake things up much. In addition, areas that spend more per student tend to be economically better off. And given the correlation of income and IQ, it may be simply that the areas able to spend more on education have brigther citizens (and by extension brighter children). However, given that the data is by state much of that should even itself out. That being said, it certainly appears that money being spent can have an impact on academic achievement, but only if population segmentation occurs. Without breaking groups up (in this case by race), there is no correlation between funding and results. It is yet more evidence of the veritability of human biodiversity and a demonstration of the ignorance that is brought on by intellectual fascism.

In the future I will be looking at math scores, controlling for state population, and controlling for standard of living (in addition to cost of living). As always, if you email me I will be happy to send you the excel data.

## No comments:

Post a Comment