[N.B.: I originally published this post on 2012.10.06; it was last updated on 2014.11.22. I revised it for clarity and to replace broken links.]
From the Huffington Post, via Reuters:
The largest U.S. civil rights group plans to file a complaint on Thursday over the admissions test at New York City’s specialized high schools, among the nation’s most elite public schools, citing effective discrimination against black and Latino students…
While more than half [of] the population of New York City is black or Latino, black students made up only 1.2 percent of the Stuyvesant student body last year, while Latino students represented 2.4 percent, city data showed.
The National Association for the Advancement of Colored People, which is filing the complaint, said the highly competitive, 2-1/2-hour, multiple-choice Specialized High School Admissions Test [SHSAT] was at fault for the disparity.
Oh boy – another “achievement gap” dilemma. So according to NAACP, the disparities in specialized high school demographics must mean the SHSAT is a flawed test; after all, these schools should represent the City’s overall demographic – right? As I will soon show, this line of reasoning is wrong.
Mayor Bloomberg recently chimed in on the matter:
Life isn’t always fair…these are the schools designed for the best and the brightest.
There’s nothing subjective about this…you pass the test, you get the highest score, you get into the school, no matter what your ethnicity, no matter what your economic background is. That’s been the tradition in these schools since they were founded and it’s going to continue to be.
I’m not entirely partial to Bloomberg, but on this issue, he’s right.
Now, I have no affinity for affirmative action (hereafter AA) since it denudes Black/Hispanic achievement and can stigmatize them (e.g. “they only made it thanks to AA, not because they had the smarts/skill/ability”); methinks that’s exactly what will happen should the specialized high schools ditch the SHSAT or otherwise lower standards (and I hate lowered education standards). Given the obstacles we already face (some of which I will show empirically soon), stigmatization is the last thing we need.
That said, I will boldly state that the SHSAT does cause racial/ethnic disparities in specialized student bodies – but not for the reasons NAACP claims.
You see, exams like the SHSAT are cognitively loaded (i.e. psychometric); though they test various subjects, their psychometric loading is comparable to that of IQ tests. A lengthy research article suggests that racial/ethnic differences in average IQ exist even after taking family/home environment, scholastic environment, and socioeconomic status (SES) into account; these differences, in turn, can explain racial disparities in other measures of ability (like the SHSAT; I underline average here because individuals of any race or ethnicity can have an IQ above or below said average). However, the findings (namely, lower average IQs among Blacks/Hispanics) remain controversial and subject to heavy debate. Let’s be careful, however, not to ignore the truth of the matter due to this controversy; consider the various U.S. scholastic cheating scandals. Pressure to close achievement gaps led administrative staff to fudge numbers and even demographic information to obscure their own shortcomings. What this shows is, absent recognition of partly hereditary cognitive differences between groups, the only way to truly close achievement gaps is to fix the results.
Is this the message we want to send to disadvantaged groups? “Hey, y’all can’t make it without a push or a crutch” or “You’re not smart enough to make it on your own, so we’ll ‘help’ you out.”
Now, I’ll look at some other points in the HuffPo/Reuters article:
The NAACP said in its complaint that the test had never been shown to predict reliably a student’s academic potential, and breached the Civil Rights Act by having an “unjustified, racially disparate impact.”
Unfortunately for the NAACP, psychometric exams (whether IQ tests or other highly g-loaded tests) do predict a student’s academic potential (see also here); this does not mean, however, that other factors such as higher-quality schooling in Black/Hispanic hoods, stable home environs, and equal access to tutoring/test prep/extracurricular programs aren’t significant. They are – and I favor such improvements – but we cannot ignore cognitive ability. (Indeed, once the playing field truly becomes level, differences in cognitive ability become clearer.)
This brings me to NAACP’s claim of the SHSAT “breaching the Civil Rights Act by having an unjustified, racially disparate impact.” Straight from the National Center for Education Statistics is this report of racial/ethnic trends in education from July 2010. Let’s examine some of the data.
[Figs. 1, 2] National Assessment of Educational Progress (NAEP) results by race/ethnicity, 2005/2007 reading results and 2005/2009 mathematics results, respectively (click images to enlarge).
The U.S. uses the NAEP to assess student proficiency in various subjects in the 4th, 8th, and 12th grades; here I highlight reading and math scores. The reading results show that, at each grade, the Black/Hispanic proportion scoring “below Basic” was about 2-3 times greater than that of Whites; meanwhile, the White proportion scoring “at or above Proficient” exceeded that of Blacks/Hispanics by a factor of two to three. Asian and White results are comparable.
The mathematics results show even greater disparities, with the Black/Hispanic proportion scoring “below Basic” being two to four times greater than that of Whites; no more than 22% of Blacks or Hispanics scored “at or above Proficient” at any grade.
Moreover, the NAEP has psychometric elements that mimic those of IQ tests! Given the racial/ethnic disparities in NAEP results, the test’s psychometric properties, and the U.S.’s use of said results as an achievement barometer, is the NAEP “breaching the Civil Rights Act by having an unjustified, racially disparate impact?”
[Fig. 3] Trends in International Mathematics and Science Study (TIMSS) results for U.S. 4th and 8th graders by race/ethnicity, 2007 (click to enlarge).
The U.S. uses TIMSS to assess American student achievement in math and science and compare it to that of students from other countries. The highest attainable score on the exam is 1,000 and scores are normed against a mean of 500 and standard deviation of 100. As was the case with the NAEP results, these results depict racial/ethnic disparities. On average – for both grade levels and both subjects (math and science) – Whites and Asians scored above the U.S. average, multiracial children scored around average, and Blacks and Hispanics scored below average. More telling are the so-called “achievement gaps” between Whites and Blacks/Hispanics:
[4th grade math]
White – Black gap: 68 points (0.68 s.d.)
White – Hispanic gap: 46 points (0.46 s.d.)
[8th grade math]
White – Black gap: 76 points (0.76 s.d.)
White – Hispanic gap: 58 points (0.58 s.d.)
[4th grade science]
White – Black gap: 79 points (0.79 s.d.)
White – Hispanic gap: 65 points (0.65 s.d.)
[8th grade science]
White – Black gap: 96 points (0.96 s.d.)
White – Hispanic gap: 71 points (0.71 s.d.)
While the TIMSS Web site cautions against directly comparing 4th grade results to 8th grade results, one must note the consistently larger gaps in the 8th grade results. The gaps are even greater when comparing Black/Hispanic performance to that of Asians, who outscored Whites in all cases except for 8th grade science. Moreover, research conducted by Heiner Rindermann (and widely discussed by others) suggests strong correlations between student assessments such as the NAEP and TIMSS and between said assessments and IQ tests; this implies that the TIMSS may also be psychometric in nature. Given this evidence, is the TIMSS “breaching the Civil Rights Act by having an unjustified, racially disparate impact?”
[Fig. 4] Advanced Placement (AP) exam results by race/ethnicity, 2008 (click to enlarge).
Many high schools offer AP courses and exams (scored from 1-5, with 1 the lowest, 5 the highest, and 3 (generally) the lowest required for “advanced placement” in college courses (i.e. college credit); NYC’s specialized high schools are no exception. Fig. 4 shows overall AP exam results, along with results for selected subjects. As with the NAEP and TIMSS, we have more racial/ethnic disparities; on average, Whites and Asians scored above the overall average while Blacks/Hispanics scored below. For further perspective, I copied the data into Excel and calculated the standard deviations for each result:
[Fig. 5] AP exam data from 2008, with standard deviations (click to enlarge).
Average Black scores, aside from being lower than White/Asian scores, also have lower standard deviations (i.e. less “spread” about the mean); this means that Black proportions scoring 3 or more are substantially lower than that of Whites/Asians. Standard deviations for Hispanic scores were also generally lower than White/Asian deviations but not by much; overall, the Hispanic deviation is higher than the White deviation.
To add to this, here’s a note on AP exam scores from the College Board:
The total scores from the free-response section and the multiple-choice section are combined to form a composite score. These composite scores are then translated into the 5-point scale using statistical processes designed to ensure that, for example, a 3 this year reflects the same level of achievement as a 3 last year.
Do these facts mean that AP exams are “breaching the Civil Rights Act by having an unjustified, racially disparate impact?”
[Fig. 6] SAT scores by section and by race/ethnicity, 1998-2008 (click to enlarge).
The SAT (formerly the Scholastic Assessment Test and Scholastic Aptitude Test) is the most widely used college admissions test in the U.S.. Scores on each section range from 200-800; the College Board added a new writing section to the overall test in 2005. As with the preceding exams, we see racial/ethnic disparities. Across the board (i.e. all sections and all years shown), average White scores are roughly 100 points higher than average Black scores and roughly 70-80 points higher than average Hispanic scores. Average Asian scores in the critical reading section trail Whites, while math scores exceed the White average; White/Asian writing scores are similar. Is the SAT, therefore, “breaching the Civil Rights Act by having an unjustified, racially disparate impact?”
[Figs. 7, 8] ACT results by race/ethnicity. Above: English and Math scores from 1998-2008; below: percentage of test-takers meeting college-readiness benchmarks by race/ethnicity, 2008 (click images to enlarge).
The ACT (formerly the American College Test) is a college admissions test developed in 1959 as an alternative to the SAT. It tests students in four subjects; for each, scores range from 1-36. The left table shows an average White – Black gap of about 5 points and an average White – Hispanic gap of about 3-4 points. Average Asian and White scores for English are similar, while Asians outscored Whites in math. Fig. 8 is even more damning; the average White – Black gap for college readiness in each subject ranged from 28 to 40 percentage points, while the average White – Hispanic gap ranged from 20 to 28 percentage points. Average Asian and White performance is similar except for math, where Asians scored substantially higher. With these results, is the ACT “breaching the Civil Rights Act by having an unjustified, racially disparate impact?”
The answer to all these questions – no, they aren’t.
And neither is NYC’s Specialized High School Admissions Test.
If the NAACP really wants to attack racial disparities, methinks they should look at this…
[Fig. 9] Percentage of teachers without a college major or certification in their main teaching assignment, by racial/ethnic school demographic, 2007-2008 (click to enlarge).
…and fight for higher-quality education (e.g. better teachers, strengthened/diversified curricula, and extracurricular activities) in highly Black/Hispanic neighborhoods and schools. Going after the SHSAT will neither erase racial/ethnic disparities in achievement nor improve long-term educational outcomes for Blacks and Hispanics.
Lastly, I’ll briefly address other points in the article:
[The NAACP] said that other elite, academically successful schools in New York City that use broader criteria such as a student’s grade-point average, attendance, teacher recommendations, interviews and writing samples had far higher enrollments of black and Latino students.
If these other schools are also “elite and academically successful,” why go after the SHSAT? Is it due to a student’s status, perhaps?
“Quite literally, a kid could have straights ‘A’s from kindergarten to Grade 8, could have won a national spelling bee.” [Damon Hewitt, the education director at the NAACP Legal Defense and Educational Fund] added, “But none of that matters – all that matters is the test.”
This is because not all A’s are truly equal, given that grading standards (indeed, even classwork/homework load) vary from school to school.
Rapper Himanshu Suri, who was vice president of the Stuyvesant student union from 2001 to 2002, said knowledge of and preparation for the test varied widely across neighborhoods…
This goes back to what I said earlier about equal access to tutoring, test prep, etc. NAACP should address this disparity, not disparities in SHSAT results!
In conclusion, when it comes to so-called “affirmative action” at NYC’s specialized high schools, just say no.
UPDATE: In November 2014, there was a campaign to save the SHSAT; I learned about it through an email from the Brooklyn Technical High School Alumni Association (I am a Tech alum, after all). I signed it, for the record. I wouldn’t be surprised if the SHSAT – and NYC’s specialized schools in general – end up in the news again.