In 1981, Gene Glass and colleagues then at the University of Colorado at Boulder conducted an extensive meta-analysis (at the time, a brand new way of synthesizing the outcomes of many different research studies) of the research on the relationship between class size and achievement. They concluded that small classes, especially those at 15 and below produced better outcomes. This conclusion has been the source of major debates in the past 18 years.

Few reforms for educational improvement have gotten more attention recently than the results from class size reduction. This is unfortunate in a way because the rush to obtain small classes might produce negative outcomes for other reasons. For example, the state legislature in California has mandated class size reduction. If a class exceeds a certain class size by even one child, an entirely new class has to be constituted.

As a consequence, there have been many horror stories of schools scrambling to obtain warm bodies as teachers in order to meet the mandate. Other sad tales tell of classes being held in closets, restrooms, etc. Beyond the anecdotes this much is known: the applications for "emergency" credentials have exploded.

In the May 26, 1999 edition of Education Week, Randy Ross of the Annenberg Institute claims that reducing class size in California actually harms poor kids. Ross presents no data that the kids are faring worse, but there is a certain logic to the argument: California already had a shortage of teachers. Given the need for more, teachers in poor schools are able to leave them for more attractive, more affluent schools, leaving the poor schools to scramble for "long-term substitutes" and others who have no credentials at all. At this point, all that Ross can point to is that whatever benefits might accrue from smaller class sizes might be countered by the presence of less experience, less able teachers.

Beyond this specific problem, the debate over class size has become so politicized it is hard to find the facts. The first and worst offender in the debate is Eric Hanushek, an economist at the University of Rochester. More recently, the rogue's gallery has also come to include another economist, Caroline Hoxby of Harvard. From their writings, one must wonder if they have ever actually been in a school.

Hanushek's principal crimes are 1) presenting an analysis of trends in pupil-teacher ratio and then using this data as if it pertained to class size (he interleaves the two phrases as if they were identical) 2) refusing to acknowledge results that contradict his own analyses.

Thus, Hanushek begins one paper "There have been consistent and dramatic falls in pupil-teacher ratios over most of the 20^th century. This decline is the result of steady drops in the pupil-teacher ratio at both the elementary and the secondary school level. The obvious conclusion from this is that, if there is a problem of class size today, there must have been larger problems in the past."

Hanushek rolls down a slippery slope from "pupil-teacher ratio" to "class size." Although Hanushek often uses these phrases as synonyms the two are not interchangeable by any means. Pupil-teacher ratio includes not only regular classroom teachers, but also special education teachers, kindergarten teachers, and basically, anyone in the building who has a teacher's certificate, whether or not they have any classroom duties.

It is not surprising, then, that while the pupil-teacher ratio is about 14/1, the average class size, as given by the Digest of Education Statistics, is 22 for elementary classes and 25 for high school classes. While pupil-teacher ratio has been declining, so has the proportion of school staff who are teachers: In 1950 over 70 percent of all staff were teachers, in 1995, only 52% (Digest of Education Statistics 1997 Table 82, page 89).

Most of Hanushek's claims come from analyses of pupil-teacher ratio, and those that do not come from analyses of intact classrooms. The former are invalid for such use and the latter are problematical to say the least. Even Hanushek admits he has no experimental evidence. Experimental evidence would involve changing class size.

Hanushek attempts to dismiss the single, large experimental manipulation of class size. This, of course is Tennessee's Project STAR and we need to take a minute to understand it and Hanushek's pathetic attempts to make it go away.

Hanushek objects that the schools were not randomly chosen. This is true, but it is a ridiculous objection since the program could not force schools to participate. However, for those that opted to participate, teachers and children were randomly assigned to one of three conditions within each building. This means that any selection favoring small classes from the selection of schools would be neutralized. The three conditions were 1) regular class size 2) regular class with full-time teacher's aide and 3) small class (13 to 17 students).

Hanushek also objects to the fact that the kids were not tested before the program began. This objection can only be raised by an ivory tower researcher who knows nothing about children and schools. Hanushek is lamenting that the kids were not given "achievement tests" before they started kindergarten!?!?! Ignoring the logistic nightmare that such a program would cause, children that young cannot be tested with the kinds of tests used in schools (he actually, wrote that folks: "no achievement tests were given before kindergarten" (p. 31).

Hanushek objects that "sizable attrition occurred over the course of the experiment because of mobility and other factors, and this attrition was likely not random. Note the use of the word "likely." He has no evidence one way or another. Logically, since the students were randomly assigned to the three treatments within each building, there is no reason to imagine that the attrition was not identical for all three conditions.

Hanushek claims that the effect shown by small classes occurs only at kindergarten and might "have a is a one-time effect on student performance that is not linked to the acquisition of cognitive skills per se. Hanushek is the only person in the universe who knows what "cognitive skills per se" means, but he is wrong in any event.

Other analyses have found effects and other grades and have found cumulative effects. Alan Krueger, an economist at Princeton who is held in much higher regard by both educators and economists alike than is Hanushek, not only replicated the original findings, but found that for the control group, those who were in smaller classes scored higher on tests. Hanushek has thus far refused to acknowledge the other studies (he also is quite selective in what test data he incorporates to "prove" that class size has no effect.

But does the effect last? It sure seems to: by eighth grade, the small-classes students were a full year ahead of their peers in the control group. Now, the kids in small classes are graduating in greater numbers than their peers (remember, students were randomly assigned to conditions). Their dropout rate is lower, and their GPA is higher. They have higher SAT's and ACT's.

Another Hanushek objection is that the teachers knew they were part of an experiment and acted to make the results come out the way it did. If teachers in grades k-3 can act in ways that sustain higher achievement through grade twelve, then we need to find out what they did and bottle it.

Hanushek's work has been criticized on the grounds that it uses a primitive technology "rarely used in empirical research where sophisticated analyses are expected" (Larry Hedges and others, "Does Money Matter" Educational researcher, March, 1994). He has also been criticized on the grounds that even the data from his primitive technology contradicts his conclusions (Keith Baker, "Yes, Throw Money At the Schools, Phi Delta Kappan, April, 1991).

Hanushek has testified that class size makes little difference in court cases so he is likely trapped into saying it no matter what the evidence looks like.

References

Charles M. Achilles and Jeremy Finn. "How the Critics Err". Paper delivered to the American Association of School Administrators annual convention, New Orleans, LA, February 19-22, 1999.