What makes good magnet programs attractive?

Gather advanced students, then teach them where they're at

Justin Rosentover is a Junior Fellow at the Center for Educational Progress and recent graduate of the Science, Mathematics, and Computer Science Magnet Program at Montgomery Blair High School in Silver Spring, Maryland. This fall, he will be attending Case Western Reserve University in Cleveland, Ohio.

Charting the Course is a series from Education Progress featuring pro-excellence education commentary, news, and policy analysis.

---

Last month, in response to Katie Arnold-Ratliff’s viral criticism¹ of gifted and talented programs in New York City public schools, Matthew Yglesias argued that the real problem with these programs is that parents barely care about their “actual educational efficacy.” Contra Arnold-Ratliff, who doubts there are “gifted” children to identify at all, Yglesias writes that,

“Precisely because you really can identify which kids are the most promising ones in a pretty reliable way, the mere fact that the graduates of a gifted program do well in life does not convey any information at all about whether the program is actually any good. [...] Parents are just massively, massively under-rating the power of selection effects and wasting a lot of time, money, and political capital.”

One study he points to finds that attending NYC’s specialized high schools “has little impact on [SAT] scores, college enrollment, or college graduation.” Some magnet schools and gifted programs do earn this finding: they crow about improving test scores and college prospects while contributing little more than the ability to gather smart kids together in one place.

But an out and out dismissal of the value of selective programs would be misguided; SAT scores and college admissions are not where to look for their value. Yglesias himself doesn’t dismiss them; instead he implores selective schools and the parents choosing between them to embrace a basic tenet for improving “the actual quality of education,” one that everyone outside of education already knows: “teach the material to the kids who are ready to learn it when they’re ready, regardless of their age.”

This principle — teach students where they’re at — needs more defenders and better execution. When gifted education programs deliver on this mission, they don’t gather advanced students to coast on selection effects, but to make advanced instruction practical. One of the most concrete benefits they offer is giving advanced students the chance to take advanced classes that are only available when enough of them are pooled in one place.

---

A model program

The magnet program at Montgomery Blair High School in Maryland, located just outside of D.C., shows what delivering this looks like in practice.² Though the Science, Mathematics, and Computer Science Magnet Program only launched in 1985, already in 1993 the New York Times would write in amazement that Blair produced three Westinghouse Science Talent Search Finalists, more than Stuyvesant or Bronx Science. The local superintendent, Paul Vance, remarked at the time that “I have never seen a high school’s image turn around so quickly.”

Since then, Blair’s magnet program has maintained its position as a premier science and technology program. Students and parents who fixate on test scores still have plenty to celebrate. In 2024, the average SAT score of Blair magnet students was 1529, higher than the better known Bronx Science, Thomas Jefferson, and Boston Latin.

Yet you won’t find Blair alongside the high-profile magnets in the best public high school rankings of U.S. News and World Report or Niche. One reason is that these rankings grade whole student bodies, whereas the magnet program is only a fraction of Blair, enrolling a little over 400 students in a school of 3,200. The more important reason is that at Blair, chasing prestige is besides the point.³

It is made clear to Blair students from the minute they walk in the door that better scores and better college admissions are not the magnet’s goal. This is expressly opposed to what many parents want from the school because the school realizes where it is actually capable of adding value.

The number of American high schools that offer a course in complex analysis can be counted on your fingers and yet between a quarter and a third of Blair students take it. It is only because Blair has brought together a group of students who have managed to complete multivariable calculus before their final year of high school can this be done, and therein lies Blair’s great advantage. The desired material benefit of a specialized magnet program should not be that it boosts SAT scores; it should be to offer classes that would otherwise be impossible for high school students to take. Despite the discourse one hears around individualized instruction, the practical reality is that providing truly one-on-one individualized math instruction at a level beyond high school is incredibly difficult. The cost-effective way to offer classes like Complex Analysis is to bring together enough students who can take it all at once.⁴

To teach advanced material to advanced students who are ready for it requires reaching a critical mass of advanced students. When only a couple of students in a school would be ready, it’s unlikely to devote the resources for offering it. But when a school system has teachers who can teach advanced classes like Complex Analysis and enough students who can take those classes, then it is just wasting resources by not putting them together.

The people who administer the Blair program realize this and optimize for it. Rather than maximizing the number of standardized tests its students can do well on, Blair offers students an extra period a day to take more accelerated classes, meaning Blair students take eight classes a semester. It explicitly encourages students not to take AP classes (not that the students listen) since the program’s own content is more advanced than the AP material.

It is also vastly more enjoyable to attend a school with similarly able and interested peers. To Blair magnet students, the robotics team was a much bigger deal than any school sport was. By attending the Blair program, a student was automatically sorted into a group of people who liked the same things you liked and that was simply more personally enjoyable.

The fact that students have been selected for ability is an irrelevant criticism if your school is, like Blair, actually offering more content than would otherwise be available. This is how you actually go about “teach[ing] the material to the kids who are ready to learn it when they’re ready.” And so it is exactly what successful programs like Blair are trying to do. Teaching this material may not improve SAT scores, college admissions, or graduation rates.⁵ But that doesn’t matter when the school is providing additional learning that is of even more value to its students. I’m sure there are other GT programs that are set up badly and don’t recognize where their value lies, but the existence of such programs should not be used to discredit the entire gifted and talented enterprise.

The standard model for advanced high schools such as Thomas Jefferson or Stuyvesant is to select the top applicants from across a city, county, or region, concentrate them in one location, and offer advanced content, often with a focus on STEM or other subjects.⁶ The Blair model, on which the program makes up only a fraction of the school, offers considerable benefits for both the students in the program and the rest attending the school. Students in the program benefit from the existence of extracurriculars which would be hard to maintain in an entirely STEM-focused school, in addition to being able to interact with peers with more diverse interests.

Students outside the program, meanwhile, benefit from the fact that upper-level magnet classes are open to them so long as they can satisfy the prerequisites.⁷ Students who couldn’t enter the program in ninth grade, either because they did not want to or they were incapable of doing so, are thus able to benefit from the program. The whole school can draw on those classes once the critical mass that they need exists.⁸

Making attractive magnets

Montgomery County Public Schools currently intends to create four new magnet programs on the Blair model. In theory, I cheer this development. If there is more demand than the supply can accommodate, then more supply should be created. Between Blair and its sister program at Poolesville High School, there were 1,548 applicants and only 230 invites. However, there are no prerequisites to apply. It could be that all of the 1548 applicants would be able to healthily complete the program; it could be that only 230 can. It is probably somewhere in the middle, but the school district has failed to justify the number of new seats being created with any sort of assessment of the number of eligible students. If the district dilutes the number of students in each program such that there are no longer enough to take Complex Analysis, then it will be actively taking away opportunities.

This is in addition to other important-but-unanswered questions, like whether the new programs will have an eighth class, or how equivalent quality will be enforced across them. There is already a great deal of inconsistency between the two existing programs, and how this will be addressed when four new ones are being added simultaneously is unclear. But I would guess that the reason these problems are unaddressed is that the people who conceived this plan didn’t think about how the programs actually benefited their students.⁹ “Program makes scores go up and is overenrolled, therefore we should add more programs” is an insufficient viewpoint.

But it is entirely possible that this expansion will destroy the greatest benefit of the Blair program. If there are no longer enough students gathered in one place who want to take Complex Analysis or Biochemistry or Quantum Physics because they’ve been dispersed among half a dozen magnets, then the new programs really will just be selection effects masquerading as accelerated programs. If all you do is gather all the students taking AP Calculus BC to the same place without adding additional content, then you are not actually adding value. Some gifted and talented programs are probably like this, maybe even some of the famous ones. But that is not the same as adding value because you have selected advanced students and as a result are teaching them more.

As a method for school districts to provide more instruction to their top-performing students, gifted and talented programs are cost-effective. But if your goal is to maximize district performance on standardized tests, then sure, creating magnet programs doesn’t work. The students you will attract are not high-leverage data points and it will be much more cost-effective to focus on low-performing students. But if your goal is to provide the maximum possible number of opportunities for the maximum possible number of students, then concentrating your advanced students in one place is the way to do it. Blair’s particular model is then one step better because it expands its opportunities to those outside the program.

When districts are designing new accelerated programs and parents are choosing where to send their promising kids, the question they need to ask is not how to get the school with the best SAT scores ranking, but how to provide students with the advanced learning opportunities that otherwise would not exist. Those opportunities are perhaps the best way for teaching advanced students what they’re ready to learn and when, and for exactly that reason they can actually increase school quality.

1. Ironically, magnet/GT program discourse is animated both by critics who think these programs are simply ineffective and by critics who think they are incredibly effective but roped off from students from low-income and minority backgrounds.

2. I use the terms “magnet program” and “gifted and talented program” mostly interchangeably, but it is worth noting that there are magnet programs around non-academic topics, like music or theater, that likely experience similar benefits from agglomeration. Similarly there are GT programs which do not draw from multiple schools which can derive the same benefits.

3. We may also question whether these rankings capture meaningful information beyond just the wealth of the students.

4. The other key requirement being that you can find a teacher to teach the class. Attracting teachers who know the material is quite difficult, especially when unionized school districts can not offer financial incentives. Because of this, GT schools are generally no more expensive than regular schools on a per pupil basis, excepting that non-urban districts, like Montgomery County, need to pay to transport students to their magnet schools.

5. It’s worth noting that the graduation rate from programs like Blair is almost 100% anyway, the SAT has no advanced mathematics content on it, and college admissions are incredibly opaque, making all three of these metrics rather poor ones.

6. Even when Thomas Jefferson has a focus on STEM, of course, the rate of AP class enrollment in non-STEM subjects is still much higher than average, since performance in different areas is heavily correlated.

7. While it is highly unlikely for a student outside the magnet program to take a specialized class like Complex Analysis, classes like Multivariable Calculus, Quantum Physics, and Neuroscience almost always have non-magnet students.

8. Examining the data, there’s a suggestion that these spillover benefits exist. If we attempt to account for the high influence of socio-economic status on SAT score, we see that Blair lies 2.47 residual standard errors above the trendline:

On its own, that number is just what we expect selection to produce. But selection stops explaining the result when we remove the selected students. If we add a datapoint for a hypothetical Blair, where we subtract 422 magnet students with SATs at the Blair average, and we conservatively assume none of those students are economically disadvantaged, the hypothetical Blair that remains (which no one tested into!) still sits 1.87 residual standard errors above the trendline. Correlation is not causation; it is entirely possible that Blair is doing something else entirely that explains its enhanced performance. But something is working.

9. There is hope on this front. The longtime coordinator of the Blair program was recently promoted to oversee the countywide rollout which restores some confidence in the district.