Ask two people whether class size matters and you will get two confident, opposite answers. What makes this frustrating is that both of them are usually citing real evidence.
There is a large randomised experiment showing meaningful benefits from smaller classes. There is also a multi-billion dollar statewide programme that produced no detectable improvement in test scores. Neither side is inventing anything. They are describing different interventions, in different grades, for different children, and calling both of them “smaller classes.”
Here is what the evidence actually supports, and what it means for a parent rather than a legislator.
Before any of the research is useful, it helps to know that schools quote two different figures and most parents are not told which one they are looking at.
A pupil-teacher ratio divides all teaching staff into total students across a school. Actual class size is how many children are in the room with your child. The ratio is nearly always the lower number, and a school quoting fifteen to one may still be running classes in the twenties.
So the question worth asking on a tour is simply what the actual class sizes are, grade by grade. Schools built around small groups will give you a figure without hesitating, and it is often far below any district average. A family comparing a local district school against a private school in Coral Springs or a similar independent option may be looking at the difference between a class in the mid-twenties and a class in single digits. A ratio on a brochure will not show you that gap.
Hold that distinction in mind, because it explains most of the apparent contradiction in the research.
The most credible evidence comes from Tennessee’s Project STAR, run in the late 1980s. Students and teachers were randomly assigned to small classes averaging fifteen students or regular classes averaging twenty-two. Random assignment matters enormously here, because it rules out the usual problem that wealthier schools tend to have both smaller classes and better outcomes for reasons that have nothing to do with class size.
Analysing that experiment, Alan Krueger found students in the small classes outperformed their peers by roughly 0.22 standard deviations after four years, which works out at something like three additional months of schooling. Worth knowing: that effect was concentrated in the first year a student took part, rather than accumulating steadily across all four. A later follow-up using tax records found those students were around two percentage points more likely to be enrolled in college at twenty, though the same study found no clear effect on income at twenty-seven.
Brookings researchers Grover Whitehurst and Matthew Chingos surveyed this literature in their 2011 review of what class size research says, and it is worth reading if you want the full picture rather than the version each side quotes. They also flag the ratio problem above: the pupil-teacher figure is almost always lower than real class size, because it counts teachers in specialist roles alongside classroom teachers. It is still a useful signal, since within a given state the two numbers track each other reasonably closely. It is simply not the number you actually want.
STAR remains the anchor study, though it has drawn methodological criticism over the years, including recent work questioning how well its results would hold up if the approach were scaled.
Now the other side.
In 2002 Florida voters amended the state constitution to cap class sizes in core subjects. Implementation cost roughly twenty billion dollars over the first eight years. Chingos examined the results and found no evidence that the policy improved test scores in grades three through eight.
That is not a small footnote. Connecticut research found something similar, with Caroline Hoxby finding no relationship between class size and achievement in fourth and sixth grade. Studies in California found positive effects roughly half the size of Tennessee’s, much of which was offset by the wave of inexperienced and uncertified teachers hired to staff all the new classrooms.
The reconciliation is fairly simple once the numbers are in front of you.
Tennessee cut classes by seven students, a reduction of about a third, in kindergarten through third grade. Florida’s caps were set at eighteen through grade three, twenty-two in grades four to eight and twenty-five in high school, reached by trimming district averages by around two students a year. Those are different interventions wearing the same name.
The pattern across the credible studies is that large reductions, in the order of seven to ten fewer children, produce meaningful effects. Small trims mostly do not. Effects are also strongest in the earliest grades and fade as children get older, and Florida’s evaluation could not see kindergarten through second grade at all, because state assessments do not begin until third grade.
There is also a resourcing trap. Shrinking every class in a state requires hiring thousands of teachers quickly, and in the STAR data itself, differences in teacher quality produced considerably larger effects than differences in class size. A smaller class taught by someone hired in a hurry is not obviously an improvement.
This is the part that matters more to a parent than to a policymaker, and it is where the evidence is most consistent.
Krueger’s analysis found the benefits were largest for Black students, for economically disadvantaged students, and for boys. Brookings summarise the wider pattern the same way: effects appear largest in the earliest grades, for children from less advantaged backgrounds, and possibly in classrooms where the teacher is less experienced.
There is a second thread worth knowing about. Dee and West looked at eighth graders and found no overall effect on test scores, but did find modest positive effects on what researchers call non-cognitive outcomes, including attentiveness and attitudes toward learning, plus a small test score effect in urban schools. Small classes may not reliably raise scores for a typical middle schooler. They may still change how engaged that child is.
That distinction lines up with what teachers tend to describe. In a room of twenty-eight, a child who does not raise a hand can go a long time unnoticed. In a room of eight, going unnoticed is not really available. Whether that shows up on a standardised test is a separate question from whether it changes a child’s experience of school.
Policy research answers the question “should a state cap class sizes for everyone.” That is not the question a parent is asking.
The useful translation is roughly this. If your child is young, class size is more likely to matter. If your child is already doing fine, is confident about speaking up, and is not struggling with attention or anxiety, the evidence that a smaller class will lift their test scores is weaker than the marketing suggests. If your child is one of the ones who disappears in a crowd, the case is stronger, and the outcomes likely to shift are engagement and confidence rather than a jump in scores.
One honest limitation is worth stating plainly. None of these studies tested very small classes. STAR’s small classes averaged fifteen students, and the credible research covers reductions of seven to ten children from a baseline in the low twenties. Schools running classes of four, six or eight are operating well outside anything that has been rigorously measured. That does not mean those settings do not work, and there are good reasons to think the mechanisms would hold. It does mean nobody should cite Project STAR as proof that a class of six is better than a class of twelve, because that comparison has never been run.
What the evidence does support is that the size of the difference has to be real. Moving a child from a class of twenty-four to a class of twenty is unlikely to change much. Moving them somewhere dramatically smaller is a different proposition, which is why the actual number is worth asking for.
Is there an ideal class size?
No single figure has research consensus behind it. What the evidence supports is that the size of the reduction matters more than hitting a particular target, and that reductions of seven to ten students are where meaningful effects start appearing.
Why do studies disagree so much?
Mostly because they are studying different things. Grade levels, the size of the reduction, which students were involved, and how the reduction was staffed all change the result. A study of a two-student trim in middle school and a study of a seven-student cut in kindergarten are not measuring the same intervention.
Is a low student-teacher ratio the same as a small class?
No. The ratio divides all teaching staff into total students, including specialists who may never teach your child, so actual class size is almost always higher. The two do correlate within a state, but ask for the class size figure directly.
Does class size matter more for children with additional needs?
The evidence points that way, though the large studies were not designed to answer that question specifically. Effects were consistently strongest for children who were already at a disadvantage, and smaller settings make it harder for a struggling student to go unnoticed.
If the research is mixed, is class size a marketing gimmick?
Not quite. It is a genuine variable with genuinely uneven effects. The honest position is that it matters a great deal for some children and very little for others, and the useful work is figuring out which group yours is in rather than treating the number as a proxy for quality.
What should I ask a school about class size?
Ask for actual class sizes by grade rather than a ratio. Ask what the largest class currently is, not the average. And ask what happens when a class fills up, since a school with a firm cap and a school that quietly exceeds its stated average will hand you the same number for very different realities.
