Column: Smart kids come in all class sizes

By Joshua Benton
Staff Writer

Page 1B

In 1968, Jack Singley scheduled middle-school classes for a living. “In all your core classes, you were lucky if you only had 33 kids in a classroom,” he said. “Thirty, 35 – that was the standard.”

In 1984, a new state law required that every Texas classroom from kindergarten to fourth grade have no more than 22 students. Educators were thrilled.

In 2004, Mr. Singley is the superintendent of Irving schools. Much of his job nowadays is finding ways to cut the district’s budget. And he wants to see the hard-and-fast 22-student rule come to an end.

“It’s very expensive,” Mr. Singley said . “Is there any difference between a class with 22 kids and one with 24 kids? Absolutely not.”

With Texas scrounging for school money, the class-size cap is back in the news. For those who didn’t read the fine report on the subject by my colleague Terrence Stutz a week ago Sunday, some people are saying it costs too much to hire the extra teachers the 22-student cap requires. They want more flexibility.

The most common idea being bounced around Austin: Make the 22-to-1 ratio a district average instead of a classroom requirement. So a school could get away with a 24-student classroom if it had a 20-student classroom to balance it out.

As you might expect, there’s some disagreement about the wisdom of such a move. But what does the research say? Do smaller classes lead to smarter kids?

Surprisingly, not much. Some would argue not at all.

Take California: In 1996, it had the largest elementary school classes in the nation – about 29 kids to a classroom. The state Legislature decided to change that and has since spent somewhere north of $10 billion to cap class sizes at 20. Some people called it the biggest state education reform effort in American history.

But did it work? The state board of education asked four esteemed research organizations to study the impact of class-size reduction on academic achievement. Their findings: Smaller classes were very popular among teachers and parents. But there was “little connection” between smaller classes and better test scores.

One reason: Having more jobs to fill meant schools were forced to hire more uncertified, underqualified teachers. Teachers hired after the cap were seven times more likely to be less than fully certified than those hired before. And those least-qualified teachers were most likely to be assigned to kids who are struggling.

It’s harder to learn from a bad teacher, no matter how many kids are in the class.

Studies in other states have found that smaller classes have little or no impact on test scores for most kids. (Non-English speaking kids – which Texas has a lot of – do seem to be the exception.)

To be fair, both sides of the debate can point to research that seems to support their arguments. Both sides can also point out the flaws in the other side’s studies, point out the flaws in those alleged flaws, and so on. (Education researchers have yet to find a subject they can’t disagree about.)

But what about Texas? Our student-teacher ratios shrank noticeably in the 1990s. In an average Texas district, classes contracted by slightly more than one child between 1994 (when Texas introduced its school ratings system) and 2002 (the last year of the TAAS test). Some shrank by a little; some shrank by a lot.

If smaller class sizes lead to better student performance, classes that shrank the most should see the biggest gains, right?

To test that theory, I pulled test scores for every Texas school district from 1994 and 2002. I also gathered data on average class sizes at the elementary and secondary levels.

At first glance, there was no apparent pattern in the numbers. Some districts with small classes scored very well; some didn’t. The same was true with large-class districts: some did very well; some didn’t. No clear connection.

So my colleague Jennifer LaFleur, a data whiz, ran what statisticians call a regression analysis to search for a connection. I won’t bore you with the details of the analysis, but I will use one term that might be new to you: r-squared.

(See, these education columns can be … educational!)

An r-squared value is always between zero and 1. A zero indicates absolutely no statistical relationship between two items (like class size and test scores). An r-squared of 1 means there is a perfect connection – if you knew how small a school’s classes are, you could predict with precision how strong their test scores would be.

In Texas elementary schools, the r-square is 0.002. In high schools, it’s 0.003.

Those are teeny, tiny numbers. They say that in Texas, there’s been almost no connection between shrinking class sizes and growing test scores.

That matches up with what much of the education research says. Yes, 35 in a classroom probably is too many. And yes, there are some kids with specific needs – such as language problems or learning disabilities – who might need classes of 15 or fewer.

But for the great majority of kids, there’s precious little difference between a class of 22 and a class of 24. Except that, under the law, one is legal and one isn’t.

“The quality of the instruction you give is just as important as how many kids there are in the room,” Mr. Singley said.

Joshua Benton covers education for The Dallas Morning News. He can be reached at