Does Class Size Matter?

In June, the DeKalb Schools (DCSD) board passed a Proposed Class Size Flexibility Resolution allowing for roughly 6 children over the state allowed max.  That means, for example, that kindergartens with class sizes over 18 would normally not be funded.  However, with this waiver, DCSD may now have kindergartens with as many as 24 children.
You can see the state allowed max and waivers granted to DCSD since 2011 in this chart.
The Atlanta Public Schools (APS) board recently rejected the administration’s request for class size waivers.  In a called special meeting last night, the APS board reversed that decision.
Jarod Apperson, forensic auditor working on his doctorate in economics, sent this note to APS’ Superintendent, Dr. Carstarphen.
From: Jarod Apperson
To: Dr. Meria Carstarphen, Atlanta Public Schools Superintendent
I am writing in hopes of influencing your priorities with respect to class size as you continue to formulate a vision for our district’s schools. From my understanding of the class-size research and knowledge of the Atlanta schools, I have become persuaded that a substantial reduction in class size would be the easiest action you could take to improve student learning.
Understanding that the district faces a number of challenges and competing priorities, I write not to make demands, but with confidence that if you have a thorough understanding of the issue, the appeal of class-size reductions will be evident.
Below, I present a series of relevant questions and attempt to provide informative answers.
Are smaller class sizes an effective means to raise student achievement?
Yes. As most Georgians are aware, APS lags behind the state in student achievement. What fewer realize is that the size of this gap is not insurmountably large. The average APS student scores about 0.25 standard deviations below the state average. I begin with this information to provide context that will help you evaluate the research on class size in terms of its implications for the district.
Credible research design is essential to developing good causal estimates, and both randomized experiments and quasi-experimental research indicates that class size reductions positively impact student achievement.
Evidence from the Tennessee STAR experiment shows that students assigned to classes with a maximum of 17 students scored 0.15 to 0.20 standard deviations above students assigned to classes with a maximum of 25 students.
Thus, the experiment’s results suggest by reducing its maximum class size by 8 students, APS could close between 60% and 80% of its achievement gap with the state.
Quasi-experimental designs, which are more common because they can be conduced with observational data, have found similar results. The most famous of these is an Angrist and Lavy (1999) study using Israeli data. The authors use a fuzzy regression discontinuity design to evaluate differences in achievement for schools just above and below a maximum class size threshold. They find results consistent with the STAR study.
Importantly, both studies indicate that the positive effects are even larger for disadvantaged students, a significant fact in a district were approximately 80% of students qualify for free and reduced lunch.
Unfortunately, the debate on class size was muddied by a number of ill-designed studies in the 1980’s and 1990’s that purported to show no effect, but in fact did not employ empirical designs that would allow the researchers to isolate the effect of class size on student achievement. Though the academic literature has moved toward more credible designs, these studies continue to influence popular culture and were most recently featured in Malcolm Gladwell’s David and Goliath. Northwestern economist Diane Whitmore describes additional research in her 2014 summary of the class-size literature.
For more local and (admittedly) anecdotal evidence, we can turn to an APS charter school that explicitly prioritizes class size. I serve on the Board of Directors for the Kindezi School, an Atlanta charter that sets a maximum class size of eight students across all grades.
The average Kindezi student scores about 0.31 standard deviations above the state average, and according to the state’s Beating the Odds measure the school ranks in the 99th percentile statewide when benchmarked against schools serving similar students.
So, yes, reducing class size is an effective means to raise student achievement. Credibly designed research supports the importance of class size and anecdotal evidence in our own back yard confirms this body of work. If APS were to substantially reduce class size, it could decrease or potentially even eliminate the gap between its achievement and the state average.
Are smaller class sizes easier to implement than other initiatives?
Yes. For reasons that are not always clear to me, class-size discussions in the district often meander into a territory where class size is pitted against effective teachers. In response to a suggestion that the district prioritize class size, it is not uncommon to hear “the most important thing for student achievement is placing an effective teacher in every classroom.” This is a flawed argument for two reasons.
First, it is a false choice. Reductions in class size need not come at the expense hiring effective teachers. The district’s historical struggles to attract top talent are not the result of financial constraints. APS offers one of the most competitive compensation packages of any district in the nation.
Instead, a perceived culture of incompetence is what has long dissuaded talented people from joining the district. Under your leadership, the district can work to improve this culture while prioritizing class size.
Second, reducing class size is easy while placing an effective teacher in every classroom is easier said than done. A recent Education Week report showed that New York City has been able to turn around its first-year teaching pool, but it took a very long time.
In 1985, 42% of the city’s teachers came from the bottom 1/3 of the SAT distribution. Today, only 24% come from the bottom third, while 40% come from the top third. That transition took 30 years.
By developing a pipeline at higher-caliber universities and continuing its partnership with alternative recruitment programs, APS can and should raise the bar for teacher selection, but results will undoubtedly be incremental. Class size reductions are an effective policy that can be implemented immediately, and there is no credible reason they should come at the expense of prioritizing effective teaching hires.
Does the return on investment for class size reduction make it worthwhile?
Yes. When APS publishes estimates of what it would cost to reduce class size, the district typically uses a cost per teacher of $80,000. While this may be accurate from a cash-flow perspective it is not appropriate for long-term decision making because state funding the following year is a function of the number and experience level of teacher employed by APS in the prior year.
Here’s the reality: for every incremental dollar APS invests in class size reductions, the state reimburses it 32 cents, and it gets to keep another 30 cents of local property tax revenue. So when the finance department presents you a proposal with a $20M price tag, if you are willing to set short-term cash flow issues aside, the real incremental cost is about $8M.
I will attempt to explain the basics of this without getting too wonky on the Quality Basic Education (QBE) formula. QBE is designed to incentivize the prioritization of teacher hires over alternative uses of district money. The way the formula works, districts are responsible for paying the base salary of certified teachers, payroll taxes, and contributions to the Teacher’s Retirement System. The state then reimburses districts $11k for health insurance.
Additionally, the following year, the state pays the district the incremental salary earned by the employee as a result of having years of experience and/or any advanced degrees. Both of these payments (T&E/HI) impact the share of local property taxes the district distributes to charter schools. When all three sources are combined, APS ends up net down about $30,000 per teacher rather than $80,000.
The short story is this: investing in smaller classes makes solid financial sense because a significant portion of the expenditure ultimately comes back to the district in the form of higher revenue the following year.
Do smaller class sizes disproportionately benefit non Title 1 schools?
No. The final topic I want to address is the notion that class sizes are a “Northside issue,” and students in the district’s Title 1 schools have no stake in the class size discussion. It is frustrating that some misappropriate the language of social justice to buttress opinions that reinforce the status quo.
As I explained above, the class size research indicates disadvantaged students actually benefit more from small classes than middle-class students. It is true that a number of Title 1 schools in APS already reduce class sizes by using their Title 1 earnings and/or supplemental resources such as EIP teachers.
However, we must acknowledge the limitations that choice poses on their educational program. If supplemental resources are being dedicated to class size reduction, they aren’t being used for other interventions. They aren’t funding individualized after-school tutoring. They aren’t funding small group pullouts.
If APS allocates additional teachers to all its schools, including Title 1 schools, that frees up supplemental resources. It returns those resources for use in targeted interventions aimed at the students most in need.
I hope that this information proves useful as you evaluate ways to raise student achievement in the district. The financial benefits and proven effectiveness of class size reductions suggest you should find ways to make it a priority in your plans.