The new RAND report, based on an analysis of National Assessment of Educational Progress (NAEP) tests given between 1990 and 1996 in 44 states, stands apart from other research efforts in some provocative ways. "Improving Student Achievement: What NAEP Test Scores Tell Us" attempts to lay the theoretical groundwork for making cost-effectiveness a key factor in allocating educational resources.

Comparing Apples to Apples
By using demographic data, the authors of the study were able to compare achievement test scores between groups of students with similar family backgrounds—i.e., race, ethnicity, income—in different states. This method enables the researchers to identify differences in achievement that are NOT dependent on demographics, and therefore MIGHT be dependent on how well a state's educational system is working.

One great value of such an approach is that it allows for more informed discussion about which educational policies lead to an effective use of resources. Since research has consistently found that achievement is strongly linked to family characteristics, it is only by comparing similar groups of students that researchers can find achievement differences with their roots in the schoolhouse rather than the family home.

When comparing achievement by students from similar families, the researchers found that Texas is doing the best job of raising test scores, while California is doing the worst job. While the two states share many demographic characteristics, Texas students, on average, scored 11 percentile points higher on the NAEP math and reading tests than those in California.

Spending Wisely
Not surprisingly, it seems that money does matter: per-pupil expenditures are correlated to achievement, especially when the money is used to do certain things. The three most effective uses of resources identified by the researchers were lower pupil-teacher ratios in the lower grades, a greater percentage of students enrolled in public prekindergarten programs, and a higher percentage of teachers who report that they are satisfied with the resources their school systems provide. Less important, but still measurable, was the effect of teacher turnover.

Higher salaries, and higher percentages of teachers with advanced degrees and extensive teaching experience, did not produce measurable achievement improvements in this study.

The researchers note that Texas, and another high-achieving state, North Carolina, both have state-wide educational policies that involve setting standards, rigorously assessing whether those standards are met, and holding schools accountable for their results. The authors speculate that those policies "change the incentives and productivity within public schools and even introduce competition among public schools."

Changing the Equation
The report offers powerful support for those who argue that the money that flows to public schools should be redirected to meet the greater needs of students from low-income and minority families. A cost-effective way to target resources, the authors argue, would be to provide all K-8 teachers with more resources for teaching; expand public prekindergarten in states with high percentages of low-income and minority families; and to reduce pupil-teacher ratios in the lower grades to well below the national average in those same states. States with only moderate percentages of such families should use their resources to bring pupil-teacher ratios down to the national average, but not necessarily below.

Such a targeted approach, according to the report, could produce achievement test gains of 12 to 15 percentile points in states with the highest percentage of low-income and minority students, at a cost of less than $1,000 per pupil per year. "The current disparity in per-pupil spending across states represents a source of major inefficiency in educational spending," the authors argue.

The report does not, however, take on one of the obvious implications of this argument: since schools are primarily locally funded, and states with many low-income and minority students in their public schools can be expected to have greater difficulty in coming up with additional resources, the targeted funding suggested by the research would have to come from somewhere else.

There is not much in this report to comfort those who argue that a better-paid teaching force is a better teaching force. "Significant productivity gains can be obtained with the current teaching force if their working conditions are improved," the report suggests. "Teachers by and large respond to better conditions, and know what to do to raise achievement."

The authors do note that in a tight labor market, teacher salaries become more important factors in the quality equation. And a drive to lower pupil-teacher ratios would have a clear effect on the competition for new teachers.

The researchers also fire a not-too-gentle broadside at the nation's teacher education programs: "For universities and colleges, providing teachers with master's degrees produces significant income but seems to have little effect on improving teachers' abilities to raise achievement. Teachers themselves are motivated to spend significant time and money pursuing such degrees largely because of the structure of the current compensation system. It is arguably one of the least-efficient expenditures in education," they write.

Spending Social Capital
The authors argue that educational researchers need to learn more about the differences between students of different socio-economic backgrounds. They note, for instance, that in school districts where families have more resources, those resources can "subsitute for and supplement school resources in indirect and unmeasured ways." For instance, parents may spend more time overseeing homework in those districts, substituting for individual teacher time, "allowing the teacher of higher SES students to spend more time lecturing and thus avoiding the opportunity costs of individualized instruction inside the classroom."

In districts where families don't have many resources, "more school resources could be a substitute for lower family resources." For instance, smaller class sizes, allowing for more individualized attention, could be a substitute for parental homework oversight.

This effort to consider what is sometimes called "family capital" and "social capital" in the design of educational policy and practice is interesting, but raises a host of political issues that aren't easy to resolve. It's hard to imagine a state where higher-income families will support a system that places their children in large classes devoted mostly to lectures, while the children of low-income families are placed in small classes with plenty of hands-on learning and one-on-one instruction. Things don't seem to work that way, though perhaps they should. But the discussion is definitely worth having.

The RAND report contains plentiful data on the state test results, an explanation of the methodology used by the authors, and a detailed examination and discussion of the different achievement test results in Texas and California.

RAND is a non-profit organization. The NAEP study was led by David Grissmer, with Ann Flanagan, Jennifer Kawata and Stephanie Williamson.