Monday, August 15, 2005

Money spent on education for naught?

From my last post on education spending it appeared that money might buy results, at least for some groups. The expenditure per student and teacher's salary were adjusted for the cost-of-living in each state, and the NAEP reading test scores of eighth graders across the nation were correlated from there. Afterwards, it occured to me that standard-of-living might shed some light on the spending as well as the results. To find the standard-of-living, I simply took the median household income of the state and divided it by the government's cost-of-living index. Using the standard-of-living in place of the cost-of-living as a means to adjust for expenditures on students and teacher's salaries obliterated the statistical significance and correlation between spending and results. In fact, only white students received any benefit at all, and it was marginal: .057 significance factor (putting it outside the 95% confidence interval) and an r-squared of .11 (meaning about 11% of the test scores could be explained by expenditures). The coefficients showed that it takes a whopping $200 increase per student to garner a single point on the test, something that is not realistically implementable. And money spent on teachers did absolutely nothing (it actually led to an innocuous decline in test scores).

The signficance factor for blacks was astronomical: .97 (compared to .25 with CoL). Simply put, the amount of resources given to black students across the country is not in anyway correlated to their performance. Ditto for Hispanics (.77) and Asians (.67).

The conclusion to be drawn from this puts even more ammo into the munitions of the nature crowd, although the nurture side could put a positive spin on it as well. When we look only at CoL (cost-of-living), the differences in expenditures have a palpable effect on student performance. But when we take into account the actual living conditions of an area, such disparities disinegrate.

Think of it in this way: There are three cities; Onett, Twoson, and Threed. Eagleland's government has determined that the cost-of-living in Onett is $200 per year, $50 in Twoson, and $100 in Threed. For simplicity's sake, you would expect a teacher in Onett to have a salary of $200, $50 in Twoson, and $100 in Threed. That is what the first analysis (CoL) adjusted for. So if teacher's were paid $180 in Onett (less than the CoL) and $100 in Twoson (twice the CoL), Twoson was actually the town spending "more" on teacher salaries for the correlation with test scores. Let's say Threed also spent $100. The students in Onett scored 200, in Twoson they scored 300, and in Threed it was 250. Thus, in Onett, where the teachers were paid less than the CoL the students fared the worst, in Twoson (where the teachers' pay exceeded the CoL) they did the best, and in Threed the students had average performance as the teacher's were paid a CoL equivalent. So it seems that trumping the CoL and giving the teachers big bucks does the trick, right?

Not so fast. Adding the standard-of-living (SoL) into the equation may explain the benefits. Let's say the average income in Onett is $160 (with $200 required to meet CoL and teachers being paid $180), in Twoson is $120 (with $50 to be CoL and teachers at $100), and in that bellweather Threed it is $100 (with $100 to be CoL and teachers at $100). So teachers in Onett are making more than most people in the community, in Threed they are on par, and in Twoson they are living worse than the average Twosonian. Yet the test scores run in the opposite direction (Twoson on top, Threed in the middle, and Onett at the bottom). It does not matter how much the money is put towards teachers relative to the standard-of-living. Instead, the students who come from areas where the living standards are the highest perform the most exemplary, with or without well-compensated instructors.

To recapitulate, in areas where people have a higher SoL, test scores are better. The CoL has little do to with how well students perform--how much their parents bring in relative to that CoL does matter, however (because when we adjust for it--that is, neutralize its effects--the disparities in test scores almost completely vanish). In areas where it is hard to make ends meet, test scores fall, irrespective of where teacher salaries or expenditures per student fall into the mix of the environment. There are two ways to go with this: boosting everyone's living standards will hoist up everyone's test scores (practicality of implementation aside!), or intelligent people make more money and have smarter kids which, when extended to the community at large, leads to areas with higher incomes having smarter children and then performing better on the performance evaulations. Given the slew of evidence that environmental factors have limited lasting influence on cognitive ability and the strong correlations between intelligence and educational achievement even after socioeconomic status is taken into consideration, the latter appears to me a better explanation. But it is likely some combination of the two, with the former providing a much smaller piece of the puzzle. In any case, the quality of life the students enjoy outside of the school environment explains almost all of the test score discrepancies--more money spent per student very modestly helps white kids (and no one else), while teacher salaries are absolutely meaningless for everyone (as aformentioned).

No comments: