Thursday, March 15, 2018

New evidence of an extreme gender bias in selective high school admissions demands an independent evaluation of the exam

Correction: A reader pointed out that Al Baker of the NY Times wrote an article about the gender imbalance at the NYC specialized high schools in 2013, so the Daily News is not the first media outlet to cover this important issue.

Today, Ben Chapman of the NY Daily News wrote about the disproportionate number of girls rejected from the highly-selective Specialized high schools compared to the boys. As I was quoted in the article, "You would think that the city would be taking every step they could, to ensure that girls are accepted to these high schools at, at least the same rate as boys...Girls should have the same opportunities as boys and the data suggests that this entrance exam has a gender bias that needs to be addressed.”

I have been writing about the gender imbalance at the specialized high schools since 2010.  There has been much written about the low numbers of black and Latino students admitted through the SHSAT exam, with only 10 Black students and 27 Latino students accepted into Stuyvesant high school this year.  Indeed, in 2012, a complaint was filed with the Civil Rights office of the US Department of Education about the exam's discriminatory impact.

Meanwhile, the Mayor has continued to blame the state for the problem, which passed a law years ago requiring that the exam results be the deciding factor in three high schools, Stuyvesant, Bronx Science and Brooklyn Tech; yet in the case of the other five high schools that use the exam, their method of admissions is completely under his control.  It is also important to note that  NYC is the only district in the country with selective high schools in which a single high stakes exam is the sole criterion for admissions.

Yet today's Daily News article is the first time to my knowledge that the mainstream media has reported on the sharp gender differential between the admission rates of boys vs. girls.  Here is a chart showing a large gender gap of ten percent, with only 15.4% of girls who took the exams admitted to the specialized schools compared to 20.4% of boys:

These disparate results strongly suggests the  exam exhibits gender bias, especially as among NYC public school students, girls tend to get better grades AND better test scores than boys.  Here, for example, are their comparative scores on the 2017 state 8th grade exams:

You can see from the above that according to the state exams, girls obtain higher scale scores, achieve higher proficiency rates and more of them score at the highest level (level 4) in both ELA and math. 

I also checked for the gender differential on the 7th grade state 2017 math tests, since many 8th graders take the Regents exams in math instead.  Girls get higher scores on these exams as well:

In past years as well, according to this paper by Sean Corcoran and Christine Baker-Smith, if state test scores, grades and attendance from 2005-2013 were used as criteria instead of the SHSAT, girls would be 9 to 13 more points more likely to be admitted to the specialized high schools: "In fact, the gender gap would shift dramatically in favor of girls with the use of grades and State tests."

That's a far greater disparity than they found for Black or Latino students (who tend to score lower on the State exams).  Yet among similar applicants with the same performance on the State exams, girls, Blacks, Latinos  and low-income students were all significantly less likely to score high enough to be admitted to these schools, and Whites and Asians significantly more likely to be accepted.

Corcoran and Baker-Smith also said, however, that over that period, girls were less likely to apply to the specialized high schools, which is no longer seems to be the case, with more girls  now taking the SHSAT than boys. 

In 2016, after much criticism of the exam and its racially disparate results, Pearson was awarded a six-year, $13.4 million contract to improve the previous SHSAT, which was also written by the company.  This is despite the fact that Pearson is not noted for its high-quality exams, to say the least.

They did eliminate the scrambled paragraph section of the exam, and the logical reasoning section, but appear to have made few other changes, other than making the exam even longer -- to 180 minutes from 150 minutes.  They also included only non-fiction passages in the ELA section (perhaps a nod to the Common Core/David Coleman personal preference for informational texts.)  

One of the most frequent criticisms in the past has been the highly unusual way in which the SHSAT was scored, to give extra weight to students who scored exceptionally high on the math or the ELA sections, rather than those who received an overall high average score.  Apparently this remained the scoring method as late as 2016. Has the methodology changed?  Is this one of the reasons for the extreme gender disparity in the results?

In any case, whether you believe that using one high-stakes exam as the sole criterion for admissions is itself unfair and highly unreliable (as I do), it is long overdue that the SHSAT be independently evaluated for gender AND racial bias.  There have been calls for this independent evaluation as far back as 2008.  Given the latest stark disparity in admissions for girls vs boys, that should be mandatory.  Or perhaps a Title 9 complaint?  Please leave your comments below.

Below are the offer of admissions by gender and  by school; you can see that the more selective the school the more unbalanced the numbers; with Stuyvesant 58% male and 41% female.


Anonymous said...

When scrambled paras went out, GRAMMAR went in. I can't see how that helped with the diversity goal at. all. Anyone know if grammar favors girls?

I favor the single test entry for these schools. The problem is really about improving all of our K-8 schools. That's how you increase your pool of prepared applicants. The prepping that the DOE paid for last year was for 7th graders for a few spring months only...ahead of an October test. Great - even kids who are coming into it well-prepared forget stuff or lose their feel for the 3-hour limit over the summer months.

Anonymous said...

It's important to look at the gender balance at the specialized high schools, but it needs the context of the gender balance at some of the city's top screened high schools, which are unbalanced in exactly the opposite way. A quick look at gender statistics (I'm using Inside Schools) reveals the following: Laguardia: 25% male, Bard Queens: 42% male, Bard Manhattan: 38% male, Beacon: 36% male, Eleanor Roosevelt: 40% male. Townsend Harris: 31% male. So it would seem that the screened schools that also draw from the pool of high-achieving students are biased in their selection process in exactly the opposite direction.

Perhaps this is part of a broader conversation about how children make their choices about which schools to apply to, and what the implicit biases are of those doing the admissions screening at some of these elite, non-SHS schools.

I'm not saying that you're not right to point out the gender inequity, only that it needs the context of the broader picture of gender inequity across the high school admissions process.

syed ali said...

What percentage of total 8th graders take the test? Does that vary between girls and boys? Can you control for race? Could it be that, say, more Black and Latino girls take the test than Black and Latino boys, and possibly that more Asian boys than girls take the test? Maybe that accounts for the gender differential?

Anonymous said...

The article states females do better on state exams and males do better on the SHSAT. The article then concludes that the SHSAT must be gender biased toward males. How do we know the state exams aren't biased toward females?