In the months after the cheating scandal, discussion has turned from punishing the offenders to justifying their motives. Bloggers, media outlets, and columnists have united to attack the perceived high-pressure environment at Stuyvesant.
In an article published in The New York Times in September, “Stuyvesant Students Discuss the How and Why of Cheating,” reporter Vivian Yee mentions the college office’s website that contains the grade-point averages and SAT scores of students accepted or rejected to various colleges (www.students-stuyhs.theschoolsystem.net/college_stats.rb). Yee goes on to quote alumnus Elias Weinraub (’12) about the site: “It becomes kind of a numbers game,” he told her. “It was kind of addictive, in a bad way, in a sick way. People will assume, well, I have a 92, most kids who got into that school have a 94, so there’s no way I can get in.”
A few days later, Times columnist Joe Nocera wrote an article in which he decried college rankings, claiming they “exacerbate the status anxiety that afflicts so many high school students.” This is not an original idea. Many students at Stuyvesant feel caught in the “numbers game” Elias mentioned. They feel their chances at Ivy League Paradise slipping away with each decimal below their dream GPA and so, in some circumstances, they cheat.
The cheating scandal has drawn attention to this mentality, and now, a question that has long simmered under the surface now rises to the forefront: Should the Stuyvesant grading policy be changed? It was brought up at Tuesday’s School Leadership Team meeting and is the subject of this edition’s staff editorial. Namely, would a different system—such as letter grades or a 4.0 scale—end the numbers game and reduce the rate of cheating in the school? Can a change in methodology transform an ideology?
Some teachers already use different grading scales in their classes in order to relieve some level of student pressure. Eric Grossman, the Assistant Principal of English, has been grading his student’s essays on a 1-6 scale for the last decade. “It’s like when you go to casino,” he explained in English-teacher analogy. “The first thing you do is cash in your money for chips. It’s easier to play, it’s more fun to play, you’re going to play better if you’re not playing with dollar bills. Chips are some kind of artificial currency, but for the purposes of the game, it’s better to use them.” By limiting the range of the scale, Grossman, and other teachers like him, places less of an emphasis on the grade and more on the learning—you’re playing with artificial markers, not actual transcript grades.
But this casino model can be implemented in reality; the chips can become real money. The Department of Education does not mandate the grading systems of its schools: each school can decide on its own to use the 100 scale, the 4.0, letter grades, or pass/fail. If Stuyvesant were to adopt letter grades, competition would decrease as the transcript differences between students decreased—a 94.3 and a 95.2 would now both be A’s; the inherent subjectivity of deciding between giving a student a 90 or a 91 would be eliminated by the wider range of letters. Likewise, the subjective differences between various teachers’ grading styles would even out in letter ranges. I will not go further into the benefits—these can be found in the Staff Editorial—but I will say that letter grades, even with potential disadvantages, would be an improvement in eliminating the Stuyvesant numbers game, in restoring grades to what they should be: “a rough shorthand for how [students] have done, how prepared they are,” as Grossman said.
A change in methodology, however, can only go so far. In order to effectively combat the point-obsessed culture at Stuyvesant, the student population must itself change. We are in a unique point at which this is a real possibility: for the first time since 1972, the admissions process at specialized high schools is being challenged on a large scale.
On Thursday, September 27, the NAACP Legal Defense Fund (LDF) filed a federal civil rights complaint against the admissions process, claiming that admission “based solely on rank-order SHSAT scores causes this unjustified, racially disparate impact, [and] the admissions policy violates Title VI of the Civil Rights Act of 1964.”
Damon Hewitt, a lawyer with the LDF, further stated that the SHSAT “has no demonstrative relationship to past academic achievements or future academic potential.” I do not believe the test itself is racist, but Hewitt’s comment holds some weight. In 2009, former Editor-in-Chief of The Spectator Sam Gerstenzang (’09) conducted a study on the relationship between his classmates’ SHSAT scores and their academic performance at Stuyvesant. Gerstenzang “discovered that on the so-called English portion of the test, there’s actually a slight inverse relationship between how well students do [and] how well they do in their English classes once they’re at Stuy,” Grossman said. “Most of the things I believe members of the English department tend to value […] aren’t measured on the test.”
I acquired the data for 782 students from the Class of 2009, and, running it through an Excel spreadsheet, found what Grossman was referring to: The regression coefficient (the “a” variable in the best-fit equation “y = ax +b”) for English GPA relative to a student’s verbal score on the SHSAT was -0.015, which would indicate that students who scored higher on the verbal section actually did worse in English once at Stuyvesant. In fact, the regression coefficient for a student’s overall GPA relative to his or her total SHSAT score is only 0.036. Unfortunately, any statistician would tell you that these numbers are meaningless: for both equations, the coefficients of determination (the r-squared values), which indicate how well best-fit lines explain the data, are less than five percent for both regression coefficients. The best-fit equations don’t really fit.
For further analysis, I brought my father, who has a doctorate in economics, onto the scene. “In and of itself, the test score isn’t telling you anything,” he said. “If there were more data available, it is conceivable that a positive relationship may show up.” Once accepted into Stuyvesant, a student’s SHSAT score has no bearing on his or her performance at the school. Additionally, the single-score nature of admissions places us in a numbers game before we even get in. These two facts necessitate a change in the admissions process.
Unfortunately, that statement is enough to set off a chain of teacher, administrator, and student rage; when the status of specialized high schools was threatened in the early 1970s, there were mass protests that lead to the passage of the Hecht-Calandra Act, which committed the SHSAT to law in 1972. Sure enough, the feeling was similar this time around. “They have to keep the test,” biology teacher Dr. Maria Nedwidek said. “The test is the only way to ensure that the students come in with the basic level of skills needed.”
English teacher Dr. David Mandler responded to the NAACP complaint on his blog. “While sympathizing with the NAACP’s goal to increase diversity at specialized high schools, I find the logic of the complaint seriously flawed,” he wrote. “Were the New York State Legislature to change the law and allow admission to be based on other factors as well, I fear that it would alter the very nature of specialized high schools in rather unpredictable ways.” Many also point to the democratic nature of the test, claiming other methods of selecting students would corrupt the process—indeed, I had long felt that way as well.
But unlike the current admissions process, the need for change is not a single-reason argument. Improving the student mindset, increasing ethnic and intellectual diversity, decreasing the rate of cheating, and selecting students who are more likely to succeed at specialized high schools—all of these could be accomplished if additional criteria were considered for admission. Perhaps because I am caught up in the midst of the college admissions process myself, I keep on encountering the phrase “a holistic approach.” Why should the specialized high schools not use a holistic application? In addition to the SHSAT could include a brief essay question, and students could submit a middle-school transcript and attendance record. Something like this, which takes into account more than a single test score, is a step in the right direction. Townsend Harris High School, which the US News survey has ranked above Stuyvesant the last few years, considers middle-school GPA, attendance, and a cutoff for standardized math and verbal scores in admission. If they can do it, why can’t we?
We need to recognize that we are accepting students, not numbers. If one of high school’s goals is to prepare students for colleges (most of which claim to admit holistically) and, hopefully, for life, then it should start with a holistic approach to admissions.
This column started as a response to one comment by one alumnus, and look how far we’ve come: Cheating! Grades! Admissions! These issues are undeniably connected. To counter the numbers game at Stuyvesant, we must do away with 1-100 grades and an admissions process that fosters this obsession while reflecting nothing about a student’s performance in the school. For now, we are stuck in this ideology, and simply cracking down on cheating is not going to change that.
Your example, Townsend Harris, has similar issues with diversity. This is a perfect example that adding grades will not change things; we’d still be pulling from the same pool of students who did well on the SHSAT.
Attendance is not a good criterion either; what about the kid who was sick, or who has a bad attendance because they’re afraid to go to their dangerous middle school?
Grades can be subjective as well, but even if grades weren’t, we know it’s easier to get a 99 in an easy middle school than it is at Stuy.
Adding an essay would mean that those who are better at math than expressing themselves would not get in. And are there enough people to read 28,000 essays (that’s how many take the SHSAT nowadays)?
The key is to raise the quality of the middle schools and grade schools so that all New York City students are prepared equally well to take an objective test that measures their knowledge of math, mastery of English and the ability to work under time constraints pressure.
I agree with you that raising the quality of middle schools is the long-term solution. But that is easier said than done. Amending the criteria for admission works better in the short term: students with poor attendance records in middle school are usually poorer students; the same goes for those with high GPAs. It shows that they have the ability to work and study.
Townsend Harris, incidentally, has a 6 percent black population and is 13 percent Hispanic. Stuy has less than 5 percent combined. I am also focusing more on changing the mentality of the student body (from one of numbers to one of education) than on the issue of racial diversity.
I do not think the way to change the test is as simple as the example I suggested in the article. It will take some time and considerable research. But the point is that students should be considered on more than a single number.
Well, when the city’s black and latino populations are over 60%, then you see that Townsend, and the Bards, NEST and similar schools are in the same boat. These schools use the criteria you suggest, and their results are not much better.
But note that in a middle school that is not offering algebra and other advanced courses, a student there with high grades will not do well on the SHSAT. It helps to be in a good middle school.
There is no short-term solution. We need to improve education offerings of the current 4th grader at the oldest. That means, we’ll see results four years after that.
Also, word needs to spread. Some people think our school is in Bed-Stuy!
Adam, while I appreciate your use of data analysis, and I know that your father surely has more knowledge of can say with certainty that your choice of analysis is flawed in at least a few ways. For one, a linear regression is an inappropriate tool to use in this case, given that SHSAT scores are derived from the raw scores in an intricate and nonlinear fashion (http://www.nytimes.com/2005/11/12/nyregion/12exam.html?pagewanted=all). In fact, grades are a similarly non-linear story; it’s much easier to move from 85 to 88 than it is to move from 94 to 97. In this sort of situation, a best-fit line isn’t going to work super well. It might be more productive to look order the students (1st to 782nd) in both shsat scores and grades (or organize them into quintiles), and see how predictive one rank (SHSAT score) is of the other rank (GPA).
Moreover, as the NYT article I link to notes, Stuyvesant is not exclusively looking for people who will get the best overall GPA; they are often looking for “lopsided geniuses,” which is why the SHSAT score is skewed in favor of those with strong scores on one of the tests. If this is the case, then many high scorers will have low grades in many subjects, but will be highly successful one subject. For example, a high scorer may be a lopsided math genius, who will have great grades in math and mediocre grades everywhere else. Selecting GPA as the thing to compare to SHSAT scores makes the assumption that what Stuyvesant is aiming for is getting students with the maximum overall GPA, yet Stuyvesant is actually also very interested in lopsided students. Moreover, students taking difficult classes may have lower GPAs than those taking easy classes, yet we would agree Stuyvesant is more interested in a 95 taking 4 AP classes than a 95 taking an easy schedule.
Finally, it is somewhat important to know whether the GPA your data set uses is end of senior year GPA, or GPA from some other point in time. Most people stop doing much work starting second term senior year, or even earlier if they get in early. If, for example, more high achieving students get in to college early and thus stop doing work, or if the harder classes top students take punish lack of work more than easy classes, then this could also weaken the correlation.
I think it is important to put the test to the test when it comes to data analysis, but this is going to be much more complex than a simple linear regression between SHSAT scores and GPA.