In a new Brookings Institution study (pdf), Matthew Chingo, Grover Whitehurst, and Michael Gallaher use data from Florida and North Carolina to determine whether there is a relationship between school district reform and student achievement.

Researchers found that school districts “account for only a small share (1 to 2%) of the total variation in student achievement.”  Nevertheless, they note that the differences between low and high performing districts are large enough to be significant for policymakers to consider.  Here are a few notable findings:

  • Teachers matter.  “As in Florida, [North Carolina] teachers account for more of the variation in student achievement than schools and districts and the three institutional components account for more variance in math than in reading. But districts explain more than twice the share of variance in North Carolina as they do in Florida: 1.9 and 1.3 percent in math and reading, respectively, compared to 0.8 and 0.4 percent in Florida (using our preferred model that controls only for student demographics). This may be due to the fact that North Carolina districts are smaller than Florida districts, on average.” (p. 13)
  • Teachers matter a lot.  “In North Carolina, our preferred estimates apportion 20-24 percent of the variance to the district level, 22-27 percent to the school level, and 53-54 percent to the teacher level. In other words, schools appear to be roughly equal in importance to districts in North Carolina (but not in Florida).” (p. 14)
  • District-wide performance changes are uncommon. “Among the 115 North Carolina districts, five experienced gains and six experienced declines in math, whereas four gained and seven declined in reading. These results indicate that significant changes were uncommon, occurring in fewer than 10 percent of districts, but not unheard of.” (p. 17)
  • Districts still matter. “But our results suggest that there are important differences in student achievement across school districts after taking into account differences in student characteristics. Moving from approximately the 30th to the 70th percentile corresponds to 0.07-0.14 standard deviations in fourth- and fifth-grade student test scores, or 20-33 percent of a year of learning.” (p. 17)

So, do the Brookings researchers identify any North Carolina school districts in their study?  Oh, yeah. (These passages come from a companion Brookings Institution report.  Emphasis added.)

Beaufort County is a consistently above average performer, with a mean at the 76th percentile over the decade, i.e., it was better than ¾ of the districts in the state on a covariate-adjusted basis. Pasquotank County is the mirror image of Beaufort, scoring at a consistently low level, with an average of the 14th percentile over the decade.  Halifax County is a dramatic decline, having moved from one of the top to one of the lowest performing districts in the state over the decade.  In contrast, Edenton/Chowan is a notable gainer, having moved from the 18th to the 65th percentile in 10 years. These four districts are similar in most dimensions other than academic performance. For example, they are each districts in largely rural counties from the northeastern section of the state with relatively small student populations. They have high proportions of students eligible for free and reduced price lunch, high concentrations of minority students, and similar levels of per-pupil expenditure.” (p. 16)