Bad Report Card

Strange study habits at GAO

Does private management of failing public schools make a difference? A recent General Accounting Office (GAO) report looked for a consistent trend, but with little success. Comparing privately managed public schools with similar public schools in Cleveland, Denver, Detroit, Phoenix, St. Paul, and San Francisco, the GAO noted significantly higher reading and math scores in some cities, lower scores in others, and no difference in the rest.

The GAO's results are, predictably, being seized upon by opponents of private school management. "GAO finds no gains in privately run public schools," clucked a report by the National Education Association, the national teachers' union. But can the study really justify such a broad conclusion?

Schools that get handed over to private management are often those with the poorest performance records to begin with. Thus, while it's possible to compare these with similar schools in their regions, the privately managed schools are starting out at a disadvantage. Comparing performance over time at privately managed schools would be more revealing than comparisons between schools, and here there is more reason to believe that private management's track record is good. The GAO report's "lack of test score data for successive years," not to mention the odd choice to focus on only one grade, rather than the performance of the school as a whole, makes it a thin reed upon which to hang any generalizations about the effectiveness of private management.

Of course, private management without genuine competition -- in other words, the sort of private management GAO examined -- omits one of the biggest benefits of genuine private education: When parents can compare performance for themselves, the companies that improved achievement find more business flowing their way, while those that failed to do so would soon be displaced.

So under monopoly conditions, does private management make an individual school better, make it worse, or not make much difference either way? The data, and the jury, are still out.