My unshakable faith in the value of a Boston Latin School education
The school that I love has come under attack. As has been the case periodically over the past forty years, the primary issue is admissions testing, specifically concerns about unequal representation and inequities resulting from admissions testing. Added to the mix this time is the argument that perhaps it’s time to shut down exam schools because there is no added value to attending them – a claim supported by data and science. I am more than willing to engage in discussions about the fairness and appropriateness admissions testing. The value of a Boston Latin School education, however, is not open to debate.
Let’s first address admissions testing. We know the issues surrounding admissions testing all too well. Concerns about the use of tests for entrance to select public secondary exams schools like those in Boston and New York are largely the same as those with college admissions testing (complete with countersuits by Asian students). Some argue that the consequences are greater with exam schools, however, because the students are younger and access to a quality K-12 education is at stake.
A first step in solving the admissions testing problem is to determine whether the admissions test itself is the cause. Emerging perspectives on racism and testing may uncover new issues, but the technical quality of the tests has not been a primary concern recently. Susceptibility of the tests to test preparation programs and inequitable access to those programs has been a concern, and one unlikely to go away with a new test. At some point it becomes necessary to acknowledge admissions testing reveals an underlying problem. That problem, however, is not a measurement problem and does not have a measurement solution.
Ultimately, as Boston Latin Head of School Rachel Skerritt (BLS ’95) wrote in response to the call to consider closing Boston’s exam schools, the solution lies in the city and Boston Public Schools (BPS) figuring out “how we can ensure that students from all racial and socio-economic backgrounds have the preparation and access to pursue the opportunity of Boston’s exam schools.”
Ideally, the powers that be in Boston will arrive at an interim policy solution on their way to achieving that ultimate goal; and the Boston Latin faculty and students will make the best of it as they always do. A less than ideal solution is one in which BPS adopts a more extreme position as they may or may not have earlier this year, suspending enrollments in Advanced Work Classes for elementary students, a position akin to throwing the baby out with the bathwater. In a 2018 post, I wrote about my own experiences as a fifth- and sixth-grade student in a BPS Advanced Work Class and the positive learning opportunities it offered prior to my years at Boston Latin.
Questioning The Benefits of Attending an Exam School
The call to close Boston’s exam schools is based on a study that finds “… most of these students would likely have done well without the benefit of an exam school education.” (A quote that is much more highly charged when removed from the context of the study.)
The study used an aptly named fuzzy regression discontinuity design and was published in Econometrica with the title The Elite Illusion. Ah, econometrics – the younger, flashier field that stole the hearts of education policymakers. Policymakers still party with psychometricians, but it’s an econometrician they go home to at the end of the day.
We all know an econometrician, or at least someone with econometric tendencies. They present at our conferences, serve on the boards of our organizations, and sit on our TACs. They blend right in for the most part, but there are telltale signs. They are just as socially awkward as psychometricians (i.e., painfully so), but for some inexplicable reason are much more outgoing. If you know someone who slips words like counterfactual and well, discontinuity, into casual (or causal?) conversation, they just might be an econometrician.
I will not claim to know enough about regression discontinuity designs, fuzzy or otherwise, to comment on the quality of The Elite Illusion. I trust that if it appears in Econometrica, it was a well-designed and well-executed study. That is why we have peer-reviewed journals and do not rely solely on blogs and commentaries.
Regression discontinuity design (RDD) has gained in popularity recently, but it is not new, going back to the work of Thistlethwaite and Campbell in the 1960s. We are familiar with Campbell – those of us of a certain age recall the reverence with which we treated our copies of Campbell & Stanley. It follows, therefore, that RDD is a quasi-experimental design. It focuses on students on either side of a cut score (e.g., the test score necessary for admission into an exam school), and the generalizability of findings from RDD is often limited, applicable to students performing near that cut score. RDD also relies on the accuracy of assumptions made about the students performing on either side of that critical cut score.
All of the above is made clear in the article, but we are not surprised when much is lost in translation from research article to newspaper column to social media posts. Having read all three, I have a few takeaways:
- I was taught to be cautious when the words regression, quasi-experimental, and causal are used together. That lesson still applies with regard to RDD.
- The authors applied several filters in their selection of the sample of students. One must be aware of them to fully understand whose performance is being compared in this study.
- One figure that jumped out at me from the newspaper article is that the students in Boston’s three exam schools account for one-third (33%) of all seventh through twelfth grade BPS students. The corresponding figure was approximately 25% in the 1990s and closer to 15% when I entered Boston Latin in the early 1970s (BPS enrollment being about half of what it was in 1970 while enrollment at the exams schools remains in the same ballpark).
- Last time I looked at the relationship between percentile ranks and test scores, percentile ranks between 60 and 70 are getting into that really fat part of the normal curve. I am somewhat skeptical of the generalizability of inferences based on students near the 65th percentile to the broader Boston Latin population.
- Related to the point above, the article does not report raw data that would enable me to get a better sense of the students in the study and their performance on the outcome variables of MCAS performance and post-secondary experiences.
- Finally, the authors acknowledge that those outcomes variables may not tell the whole story of the Boston Latin School experience.
Understanding The Latin School Experience
With regard to the Latin School experience, the authors note that “students who attend Boston Latin School almost certainly learn more Latin than they would have otherwise.” I really want to believe they inserted that line for its comedic value; something my parents would have described as a fine example of “that Latin School sense of humor” while shaking their heads. (For the record, “Gallia est omnis divisa in partes tres.” That’s it. That’s all I’ve got.)
Personally, I choose to believe that the Latin School students holding up signs saying that the MAP assessment should be used for admissions, implying that it was a better choice than using an actual map (i.e., admissions based on ZIPCODE), is all the evidence I need that the Latin School experience I remember is alive and well. (Or maybe Chris Minnich paid them to hold up the signs. That works for me, too.)
Watching Goodbye, Mr. Chips on TCM last week, I couldn’t help but reflect on my Latin School experience as Mr. Chipping reflected on his long life at Brookfield.
It began walking into the auditorium as a sixie and seeing those familiar names on the upper frieze: FRANKLIN, ADAMS, HANCOCK, EMERSON, BULLFINCH, … and toward the end of the list one name that looked really different from the rest: SANTAYANA (I just had to find out who he was). It continued through the years as I learned of others, leaders in my areas of interest, who preceded me at BLS: Leonard Bernstein, Robert Coles, Theodore White, William Angoff (yup, Angoff, too). If you can see it, you can be it.
And then there were the teachers and administrators, several of whom also attended Boston Latin. No, I didn’t learn a lot of Latin, but I will never forget Mr. Jameson and the many times he stood in front of our AP Latin class, putter in hand, simulating a smooth stroke, while telling us that Form Follows Function, a lesson that I would apply so many times during my career in large-scale assessment.
But perhaps the most important part of the Latin School experience were my fellow classmates, the friends who pushed, pulled, and helped each other in so many ways through those six years and the forty-plus that followed. People tell me it’s kind of unusual that we were able to get 20% of our graduating Class of 1977 together on an impromptu Zoom call last December between Christmas and New Year’s. I don’t know, it didn’t feel any different to me than when I last saw some of them at the Thanksgiving football game in 1978. That’s my Latin School experience and that’s the peer effect that matters to me.