Debating the Achievement Gap

At the Freakonomics blog on the New York Times Blog: From Caroline Hoxby:

The first suggestion may seem pointy-headed but it’s the most important: All major interventions in education should be evaluated using scientific methods — experimental methods, if at all possible.

Most interventions in education (class size reductions, pre-kindergarten programs, classroom technology, paying students for performance, drop-out prevention) are based not on evidence that they work, but rather on the “cardiac test” (e.g., “we just know in our heart that this is right”). Moreover, the interventions are not scientifically evaluated, sometimes because advocates oppose evaluation, but more often because no one bothers to set up pilot, randomization, or baseline data in the first place.

Well, duh you say, but as Hoxby points out, evaluative studies almost never happen in education and the result is that we don't have hard data to point at for finding solutions that actually work. Hoxby notes one of her studies:

In a recent study of New York City charter schools, I compared students who were admitted to the charter schools via random admissions lotteries to students who applied but were “lotteried out.” The beauty of randomization is that the lotteried-in and lotteried-out students were the same — not just in background and prior achievement — but also in motivation. The overall result was that New York City charter school students outperformed the lotteried-out students in math and reading, but not all charter schools had identical success.

One factor that was found to correlate with a charter school’s success was a longer school year (210 days, say, as opposed to 180 days) and day (9 hours, say, as opposed to 5.5). Such policies may hold great promise for schools that serve disadvantaged students, and (unlike the policies I mentioned above) have not been tried much in the past. I would like to see large-scale demonstrations of such policies with scientific evaluations so that we’ll know whether they work or not. Even if the typical public school might hesitate to implement such policies, there are many charter schools in the U.S. that would be glad to try them, with a suitable increase in their budgets.

Be sure to check out the comments.

My comment: We as a society fear the very thing Hoxby supports, actual scientific research on education. To accomplish a randomized sample means teaching two kids in different ways, one of which at least we are sure will work (if not both of them). So we try our new idea on everyone with nary a clue as to what we expect to get or whether it will even work. Then there are teh various stakeholders in teh current status quo, the education industry, the teachers' unions, and to a certain extent the political elites who need a crisis to mandate their involvement. Too much success leads to less of a "crisis" and less need for intervention. In short, we fear that finding something valuable in a real experiment will simply highlight our societal ineptitude when it comes to educating our children.