Pregnant women should take thalidomide because research shows that it’s effective.What’s wrong with that assertion? It’s true that thalidomide has been proven effective, as measured by its ability to reduce morning sickness. Why shouldn’t pregnant women take it, then? We all know the answer: because it has other effects, too – some of which, as we all now know, are terrible. Before taking a drug, you’d want to know what all of its effects were, and whether the good effects outweighed the bad. No one would argue that we should reduce morning sickness at all costs.
But that basic principle is completely absent from most discussions of empirical research on education. Educational “reformers” are eager to proclaim that their proposals “work” based on very incomplete information about their effects on children. If an intervention appears to cause short-term standardized test scores to rise, it is enthusiastically embraced, regardless of what its other effects may be. But does anyone think that we should raise test scores at all costs? If beating the kids made their test scores go up, is that all you would want to know about the practice?
Ultimately, test scores are just a proxy for something much harder to measure: whether we have achieved our educational goals by the time our kids reach adulthood. But as soon as you put it that way, you realize how pathetically inadequate those proxies are. Education has lots of goals: we want our kids to be proficient in reading and math, but we also want them to grow up to be independent, to be intellectually curious, to be appropriately skeptical, to be self-supporting, to be mentally and physically healthy, to participate willingly and capably in democratic self-government, to be ethical, and assertive, and honest, and brave, and kind, and fair, and to lead lives they find fulfilling. Many of these goals are impossible to assess empirically, especially over the long-term. Even if we could measure them, a rise in one value would often come at the expense of another. Trying to raise test scores too single-mindedly, for example, can produce stressed-out unhappy people who dislike learning. Similarly, an exclusive focus on reducing disciplinary referrals can turn schools into obedience training academies that teach authoritarian values. Where is the assessment of those effects?
“Evidence-based education” has an army of enthusiasts, but I have never heard a response from any of them on this basic point. There are gaping holes in their evidence, and they don’t seem to care at all. Yes, analysis is easier if you don’t hold more than one value in your head at a time, and if you don’t ask any questions that you don’t want to know the answer to. But is that the sign of a scientific mindset, of a respect for evidence and empiricism?
What is the response?