Taxpayers are entitled to know if public schools are doing their job, but it's hard for them to evaluate the empirical research that is published ("New Truths That Only One Can See," The New York Times, Jan. 20).  Opinions, no matter how deeply felt, are no substitute for expertise ("The Death of Expertise," The Federalist, Jan. 17).

Let's begin with replication, which is the sine qua non of science.   Especially in education, researchers are wont to see what they want to see.  It's typically an unconscious bias that causes them to interpret data to support their hypothesis. For example, critics contend that past studies showing the benefits of preschool couldn't be duplicated on a large scale ("California should give all kids the pre-K advantage," Los Angeles Times, Jan. 24). That's not surprising because successful pilot programs generally have low external validity. In plain English, it means they often don't generalize well to other persons, settings and times. 

Click here to continue reading.