Can you explain more what these institutions are, exactly?
Itâs not surprising that admitting purely on an academic test does produce a well balanced racial distribution. The expected result would be Asian applicants are overrepresented among high scoring students, and URMs are underrepresented. The more important questions are should Stuyvesant or others admit purely on test scores? And is the test score admissions producing a class that does well in whatever metric they are trying to maximize during performance at the high school and beyond?
As has been noted, the SHSAT is unique with its high test ceiling and has unique scaling favoring extreme scores, which leads to favoring spiky one-sided kids (especially ones who did a portion of grade school in China/Korea/âŠ) over well balanced kids. For example, the study at https://nepc.colorado.edu/sites/default/files/pb-feinman-nyc-test_final.pdf provides the following examples:
Kid #1 â 99th percentile verbal, 76th percentile math â Accepted at Stuyvesant
Kid #2 â 98th percentile verbal, 94th percentile math â Rejected at Stuyvesant
Kid #1 â 25th percentile verbal, 99th percentile math â Accepted at Brooklyn Tech
Kid #2 â 81th percentile verbal, 84th percentile math â Rejected at Brooklyn Tech
If the goal is to maximize chance of success at the HS, then Iâd expect the balanced kid to have the better chance than the spiky kid who is relatively weaker in one key dimension. This is especially true if you arenât emphasizing things like GPA in middle school or preparedness for a tough high school by middle school course selection. You could be admitting English/verbal superstar kids who get mostly Câs math/science/STEM and get mediocre scores on related math/science/STEM standardized tests. This group would not be expected to do well in the especially challenging math/science/STEM type courses at Stuyvesant and similar. However, if the goal is to make sure that someone who is especially talented in a specific area doesnât fall through the cracks and has a chance to pursue that talent with advanced courses; then favoring the spiky kids might be preferred.
This system that does not emphasize GPA tends to overrestimate men and underrestimate women. For example, the study at https://academicworks.cuny.edu/cgi/viewcontent.cgi?referer=https://www.google.com/&httpsredir=1&article=2168&context=gc_etds found that 88% of the low GPA (calling below 80% as âlowâ) students at Stuyvesant were males; while the higher GPA students were mostly female. Some might think this results in more deserving females being rejected over less deserving males.
One also needs to consider which students this system is helping. If you are admitted to Stuyvesant, but get a low <80 average in your courses, are you necessarily better off or well deserving of being one the few admits? I havenât seen many long term outcome studies. Noble prize winners and similar rare few exceptionals are overrepresented among Stuyvesant alumni, but itâs not clear to me that these types of kids are the ones who would have fallen through the cracks, if the admission considered things like middle school GPA or middle school course preparedness.
How does THAT happen? I assumed they took highest total combined score, period. Or is there some weird factor that makes a combined 99/76 higher than a 98/94?
Test prep in middle school (for SHSAT/etc.) or high school (for SAT/ACT/etc.) shows a much stronger parental influence than test prep in college (for GRE/LSAT/MCAT/etc.), simply because middle and high school students are much more heavily influenced and controlled by parental choices than college students are.
Well, âaptitudeâ was what the A in SAT used to mean, but it is not necessarily obvious that a test (back then) of English vocabulary words (the vast majority of the verbal section) and algebra and geometry (the math section) was necessarily measuring âaptitudeâ in isolation rather than a mix of that, achievement, and other factors.
I skimmed the article @Data10 posted. Iâd like to point out some key points.
The exam schools arenât the only good high schools in NYC and the other good high schools DO admit by grades, attendance, etc. And some of these include consideration of the state tests; they donât use a special exam, but they do use results of regularly administered standardized tests as one component of admissions. Here is a link to an article which lists the admissions criteria at these other good NYC public high schools. https://nypost.com/2016/09/17/the-top-40-public-high-schools-in-nyc/
The most competitive of these schools admit more females than males. Thus, Townsend Harris, a fine school by anyoneâs standards and often considered the best of the non-exam schools, is 69% female.
So, kids who do well on standardized tests have an advantage at one type of school and those who do better in school do better in the other admissions process. Although there are many exceptions, males in the aggregate do better with the first and females in the second.
Thatâs not unique to NYC. Males in the aggregate do better than females on the SAT and females in the aggregate have better grades than males in high school. And based purely on anecdotal evidence, the situation is worse in middle school, i.e., the discrepency between male and female GPAs is even greater.
I suspect that the probable result of making the changes in the admissions process the author wants would be to increase the number of females in the exam schools while the number of females in the other top schools wouldnât be reduced. The end result would be fewer boys would get into good high schools. I personally donât see that as a wonderful result.
Would it result in better representation of black and Hispanic kids in these schools? Unless race is explictly taken into account, IMO, the improvement would only be marginal.
Why do I think this? Of the good high schools that do use other factors in admission
Townsend Harris is 58% Asian, 22% white, 6% Black and 12% Hispanic.
The Baccalaureate School for Global Education is 49% Asian, 31% white, 15% Hispanic and 2% Black.
NEST+m 42% white,33%Asian, 11% Hispanic and 9% Black.
Eleanor Roosevelt (which is on the Upper East Side in a heavily white neighborhood) 62% white; 21%
Asian,8%Hispanic, 4% Black
These stats are taken from US News.There are a few of these high schools which have a greater percentage of Black kids, but the highest percenage I saw was 19%
So, IMO those that say that the primary reason why Black kids donât get into the NYC high schools that rely on the SHAT is test prep and especially test prep by Asians are missing the point that when admission is based on multiple factors and not just one exam the results for Black kids usually arenât much better.
Itâs the SCHOOLS they attend that are the problem, NOT test prep. The middle schools must be fixed. Do that instead of using smoke and mirrors to get more Black and Hispanic kids into the exam schools. Because if all you do is make it easier for them to get in without preparing them any better in middle school, you really arenât solving the problem.
The linked sample also includes a kid who was accepted to Stuvyesant with only a 58th percentile verbal, yet the more balanced kid with 98th percentile verbal and 94th percentile math was rejected. This occurs for two reasons â the tests have a high ceiling, such that the 99.x percentile kid gets a different raw score than the 99.y percentile kid; and the test is weighted such that the difference between 99.x percentile and 99.y percentile corresponds to a much larger SHSAT score delta than does the difference between a 98.x percentile and 98.y percentile. Roughly estimating percentiles from the graph in figure 4 of the previously linked study.
Percentile â SHSAT Section Score
90th percentile â 255
95th percentile â 270
98th percentile â 285
99th percentile â 310
99.9th percentile â 350+
The issue in the study and emphasized in my post was that SHSAT score overpredicted academic performance in males, not differences in average scores between gender. Women at Stuyvescant with lower SHAT scores were frequently outperforming makes with higher SHSAT scores, such that the kids towards the bottom of the class at Stuy with a less 80% average were ~90% male and the kids towards the top of the class were mostly female. Yes, the score admission system creates a selective HS option for kids who test well on a specific subsection, but donât do well in school and get mediocre grades. Itâs good to have different options for kids who have different strengths and weaknesses, but getting decent grades seems like a fundamental requirement for admission to a selective school.
It depends how you define marginal. You emphasized Townsend Harris, so Iâll use it as an example. Demographic percentages among graduates at Townsend Harris and Stuyvesant are below, as listed on https://data.nysed.gov/ . Iâd consider a increase from ~0% to 6% to be more than just a âmarginalâ improvement, even if there is still poor racial and gender balance.
2018 Stuyvesant â 38% Female, 68% Asian, 0% Black (0.5% Black in 2017)
2018 Townsend Harris â 72% Female, 61% Asian, 6% Black
I donât know anything about the SHSAT. Does it try to eliminate the distortions in the SAT, like making the math portion harder so both sections show similiar score distributions? Is that the âweird factor that makes a combined 99/76 higher than a 98/94â?
No. If you arenât familiar with the history of the US, or the social science around these topics, a light discussion on a message board wonât help. Itâs very complex and spans hundreds of years.
I would recommend taking a 100-level US history series at an accredited college or university or joining an adult education group on the topic at a local library if you are interested in learning more about the United States and its social structures.
In the years of the study, the math section was indeed a little harder than the verbal. The mean raw score on the math was ~21/50 (SD = ~11), while the mean raw score on the verbal was ~26/50 (SD=11). So the really stellar math kid who can score in the 99+ percentile region may have an advantage over the comparably stellar verbal kid.
However, what makes a 99/76 higher than a 98/94 was large weighting for small percentile differences at the tails, such that both the SHSAT score delta between 98th and 99th percentile was similar the the SHSAT score delta between 94th and 76th percentile. Both deltas were ~30 points. In contrast, the SAT is not as weighted at the tails. The corresponding difference for SAT would be ~10 point difference in SAT score for increasing from 98th to 99th percentile, and ~100 point score difference in SAT for increasing from 76th to 94th percentile. So a 98/94 kid would have a ~90 point higher combined SAT score than a 99/76 kid, while the 99/76 or 58/99.9 kid would do better on the SHSAT than the 98/94 kid.
The specific scores and percentiles for the kids referenced above are listed in more detail:
Kid #1 â Rejected (98/94)
Verbal: 98th percentile = 286 SHSAT Score
Math: 94th percentile = 271 SHSAT Score
Sum = 286 + 271 = 557
Kid #2 â Accepted (99/76)
Verbal: 99.0th percentile = 317 SHSAT Score
Math: 76th percentile = 241 SHSAT Score
Sum = 317 + 241 = 558
Kid #3 â Accepted (58/99+) â Estimating scores from graph
Verbal: 58th percentile = ~205 SHSAT Score
Math: 99+th percentile = ~370 SHSAT Score
Sum = ~205 + ~370 = 575
I think the numbers posted by @data10 shows that a test-based system is not quite as objective or transparent as one might assume. The high ceiling upper tail end of the scale creates once set of results if admissions decision is based on looking at raw numerical scores rather than percentiles â so, in theory, they could use the same test but weight it differently and have different results. For example, instead of selecting based on the highest numerical score, simply focus on the percentiles â letâs say, hypothetically, requiring a combined 95% percentile score on both math & verbal to be accepted. Then the kid with the 99/76 verbal/math split (average 87.5%) is turned down; and the kid with the 98/94 split (average 96%) gets in.
Or the rules could be set to require a minimum score value on each test â for example â the student needs to have at least a score of 250 on either test to be qualify So the kids with the 205 or 241 scores get eliminated, notwithstanding their 317+ scores on the other subtest.
An alternative approach would to have an academic index that combined consideration of middle school grades with score values â or simply to tack on a minimum GPA requirement to the test â rather than test only.
And yet another alternative would be to set a minimum combined test score for entrance that guaranteed that scores were high enough to qualify, but created a larger admission pool and select from that by random lottery. (That is, for example, a bar is set at 500 â everyone who scores 500 goes into the lottery) â so you still end up with a pool of very smart students â but less motivation to cram for those high-end points, and presumably, you end up with a somewhat different demographic mix.
Iâm not advocating one position over another â just pointing out that the test-only admission system is in itself arbitrary, as is the determination of how the test is scaled and graded and evaluated. And that ties into underlying educational goals and philosophies â both in terms of how the leaders of the public educational system see its role, and how parents see their roles.
All these posts make me realize that I am just sad to know that many kids will never have the opportunity to get an excellent public school education. Whether itâs due to early parental involvement ( of lack thereof), economic status, race or any other factorâthe answer is clear. We need better public education for more kids in the US. Not only in NYC, but elsewhere so that the notion and meaning of meritocracy come back into vogue.
Attacking one race, or another. Or saying that the system should be thrown out so that some other group gets in etc. doesnât change the fact that you have many kids/parents who are trying to create opportunities for their children and they are stymied by politicians and others who refuse to make public education better. Stop trying to infuse politics into education and make the pie bigger!! A bigger pie means more kids have access to better exam schools. A smaller pie means we keep dividing the pie into various slices and no one ever thinks they got the big piece!
The City is looking to overturn this. The ruling to keep the testing in place stems from the fact that even in 1971, people opposed the test because the test was thought to be culturally and racially biased.
I remember when I was admitted in 1972, there was no test prep, you tested at the school you wanted to attend on a school day, you took the test, you later got the results that determined if you were in or out. I went to Brooklyn Tech (because my mother would not allow me to ride the subway) with approximately 35 of my friends from our junior high school in Bedford Stuyvesant. We had friends who went to Bronx Science and friends accepted to Stuyvesant. No, it was not an easy feat being both black and female at any of these schools. When my D was admitted to Stuyvesant (and chose stay at her 6-12 school) it was done without test prep, showed up and took the test.
Is there any data on the number of students of each race that took the test?
I usually recommend this to those who think so. The speaker is the Marvin D. Dunnette Distinguished Professor of Industrial-Organizational Psychology at the University of Minnesota:
https://www.youtube.com/watch?v=Gv_Cr1a6rj4
Three of the more pertinent topics to this discussion are: creativity (6:35), predictive power (9:15), and social class (11:15).
BTW, what happened to the Miller Analogy Test? I have not heard anything about it in recent years.
Since objective reality is unknowable (because we do have a G**'s eye view of life), we have to base everything on our subjective reality. In a world of competing realities, the most powerful wins. So in society, the most powerful groups can force their version of reality onto the rest, in the same way that the most powerful nations can do the same on a world-wide basis.
Actually it is quite objective and transparent. Perhaps some people donât agree with the objectives, and that makes it undesirable, but it is remains objective. Note that the SAT was also once a high ceiling test, and the combined score used to reward extreme talent in one area, especially on the verbal section. Now an 800 measures only strong competence.
As I said at other times, people often assume that beyond a certain high percentile (letâs pick 95%) that students are similarly talented. But as you get further into the right tail, the differences expand, not shrink. This is easily seen in sports. Even though all Olympic sprinters are roughly in the top 0.001% of all runners worldwide, you have people like Usain Bolt completely dominate their event for many years. In the NBA the dominant players are LeBron James and Steph Curry. These top performers are so far above the median performers in the elite group that they are barely comparable.
The SHSAT recognizes and rewards this right tail outperformance. So did the pre-1995 versions of the SAT. It is in effect saying that it values students that could become a future math prodigy or a poet laureate, even if they are weaker (percentile-wise) in the other area. Whether or not this should be the case is a judgment call for society to make.
There are schools/programs that require their own exams and interviews (Bard, IB programs, etc.). There are a wide variety of spectacular options with different admissions processes. The SHSAT schools are amazing, but not by any stretch of the imagination the only spectacular programs in our city.
The SHSAT is a strange and very preppable test.
One of the problems with using power score-type admissions processes is how easy it is to cheat. The private schools absolutely inflate the grades of kids they want to be admitted (there is prestige in the number of kids you get into the SHSAT schools and with that comes more applications and more money). This happened in microcosm in a very spectacular way for a particular and unique school on Staten Island. Then the process had to be changed because the power score process was so corruptible. Google Michael J. Petrides admission scandal and you may get a sense. (If youâre interested, of course)
There isnât a good answer w/r/t the SHSAT schools. But what is good is that there are fantastic programs in our city for every child. It is a problem that many in the Asian immigrant community really do feel that their child has failed in some way if they make a different choice. Which is not the case. It would be great if the other magnificent programs were viewed as equally worthy, which they are. Then maybe some of the kids who attend the SHSAT schools but would be a better fit elsewhere would not feel their only appropriate path involves those schools and would apply/attend elsewhere and free up some of those seats for other kids who would be best served there.
The problem with those who want to rework things like standardized tests is clear. Usually there is some amount of political or experimental psychology involved to replace what is currently used with some utopian version that is going to work better ( it rarely does). The thing I like about standardized tests is they are standard so there are obvious winners and losers. The thing I donât like is that people prep for them and some kids are aware of the content until it is too late to do something about it. Are there kids in particular socio-economic and racial groups who will do better or worse on standardized tests? Yes. Is the solution to throw out the test and replace it with holistic acceptance like college? No.
âThe SHSAT recognizes and rewards this right tail outperformance.â
âThe SHSAT is a strange and very preppable testâ
These two statements seem to be contradictory. Extremely long tail tests are the least âpreppableâ, because preparation is very unlikely to get a kid who is in the 95th percentile by ability into the 99.5th percentile of achievement. Familiarity with the test format is helpful, so a 99.5th percentile kid might only score on the 95th percentile if (for example) they didnât realize that they only had x minutes per question so had to work fast, but that doesnât need hours a day of prep, and doesnât work the other way round. The SAT and ACT are much more preppable precisely because they are not very long tail and the difference in the very top scores is mostly a question of a small number of random mistakes.
The irony for most of these kids is that their subsequent college admissions generally wonât be based on the long tail abilities that got them into these schools.