The president of the United Federation of Teachers, Randi Weingarten, is still peddling the myth that smaller classes are the cure for what ails New York City’s public schools. Under the banner of New Yorkers for Smaller Classes — and the slogan “class size counts!” — Ms. Weingarten’s union has gathered more than 100,000 signatures on a petition to ask voters whether the City Charter should limit the number of students that can be in a kindergarten, grade school, or high school class. Never mind that hundreds of studies have failed to show any convincing link between reducing class sizes and improving students’ test scores. Since the city has more pressing funding priorities than adding members to the UFT’s rolls, this initiative will test Mayor Bloomberg’s mettle.

In an opinion piece set to run in papers in Brooklyn, the Bronx, and Queens, Ms. Weingarten dredges up a 1985 experiment by Tennessee to testify to the virtues of smaller classes. Ms. Weingarten writes that a study of Tennessee’s Project STAR shows that “students in small classes consistently outperformed other students in math and reading every year at all grade levels.” Actually, as this column pointed out back in May, the Tennessee study is bunk. The students in small classes pushed ahead slightly in kindergarten, but after that did not gain compared to their large-class counterparts; if there truly were a benefit in the later grades, the gap ought to have widened every year. Even more problematic, the experiment couldn’t isolate class size from other factors that might have affected student achievement. Teachers and administrators involved in the program knew that future money for class-size reduction rested on its outcome, likely skewing the results.

A far more scientific study was done by noted educational economist Caroline Hoxby, an associate professor of economics at Harvard. Ms. Hoxby conducted a “natural experiment” involving 1,035 schools in 165 districts over 20 years in Connecticut. Using the natural fluctuation in the number of students entering a class each year, she created a sample with little statistical noise. A family might have one daughter in a first-grade class of 25 students, and her younger brother, two years later, might have only 18 students in his first-grade class. The students in this study were coming from the same economic background and sometimes had the same teachers. The result: no difference between students in small and large classes. This mirrors the national trend. America has been reducing class sizes steadily for the last half-century, as education economist Eric Hanushek, of the Hoover Institute at Stanford University, notes. Despite a 35% reduction in student-teacher ratios, scores on the National Assessment of Educational Progress have remained stagnant from 1970 to today.

The UFT’s referendum, which it wants to put before voters in November, would establish a commission to propose City Charter language that would possibly limit class sizes to 18 students a class in grades K–3, 22 in grades 4–8, and 25 in high school. However, the mayor’s office has hinted that it will not allow the question on the ballot this year. In years that a mayor proposes charter revisions of his own, the state’s Municipal Home Rule Law mandates that the mayor’s proposals “bump” others — so that a plethora of questions do not distract voters. In 1998, Mayor Giuliani used this provision to thwart an attempt by the City Council to field a referendum that would have prevented him from funding the construction of a stadium for the Yankees at Manhattan. Mr. Bloomberg, as luck would have it, already has a Charter Revision Commission set to make a proposal. Any failure to challenge the UFT on this matter would be an abdication not only of a duty to uphold state law, but also of guarding the city against having its education policy dictated by its teachers union through the chartering process.