One thing has struck me lately: There is almost no reference to evidence when education is discussed. When I'm speaking about evidence, I'm talking about trials with the following characteristics:
1. The whole curriculum is tested, not just elements. Too often, we hear about curricula based on "research". Unfortunately, individual program elements, even if they are based on good science, don't translate into effective programs. For a curriculum to work, *all* the elements must work together to provide a positive result.
2. Many classrooms. I once spoke with an administrator about the need for evidence. His response was "well, there are just too many elements in a classroom to figure out what is working and what isn't." This is true if you're talking about a single classroom .. . we don't know if its the classroom teacher, those particular students, or some other factor. However, if we involve multiple classrooms at multiple schools, we start to get somewhere, as random factors start to cancel each other out. While this is expensive, curriculum developers, like drug manufacturers, should bear the burden of testing their curriculum *before* they introduce it to classrooms.
3. Beginning and end points. Often, medical researchers will talk about "end points" in a study. This refers to end outcomes that they are going to measure. In academia, this means 3rd party tests that make sure students know the material. These tests should be administered at a minimum before and after the "treatment" (i.e., the curriculum in question). To make sure the curriculum developer is dealing honestly, multiple 3rd party tests should be used, to make sure curriculum developers aren't gaming the trial by using tests that favor their curricula, and so outside parties can figure out which measure they trust if results are vastly different.
4. Controls. In medical trials, one group will get a placebo while the other gets the medicine. This doesn't apply directly, but curriculum developers can simply let schools either implement whatever curriculum they like, or they can stipulate a specific curriculum they'd like to test theirs against. Since the curriculum developers are claiming that their curriculum is better, then it should handily beat the other curricula, and this should show up in greater net achievement in beginning vs. end testing no matter what other curricula is chosen. If the chosen curricula can't achieve this result on the average, then the curriculum just isn't better.
This approach could change the whole game. One could imagine, for example, an FDA-like body for curricula. Publishers would have to submit their testing results to prove efficacy before the curriculum would be used in schools. So rather than testing the "treatment" on the next generation of children (e.g., whole language in CA in 1987), curricula would be tested *before* kids are subjected to them.