The College Board released its 2012 SAT college-entrance test results this week, and here are some highlights:
1. Continuing an uninterrupted trend that dates back to at least 1972, high school boys outperformed girls on the 2012 SAT math test with an average score of 532 points compared to the average score of 499 for females, see top chart above. The statistically significant 33-point male advantage this year on the SAT math test is two points higher than the 31-point difference last year, and just slightly below the 35-point average SAT math score gender gap favoring boys over the last 25 years.
2. For scores in the highest 100-point range of 700-800 on the 2012 math SAT test, boys outnumbered girls by 73,844 to 43,838, which would mean that there were 165 boys for every 100 girls scoring at 700 or above. To account for the large difference in the number of girls taking the SAT test (886,000) compared to boys (778,000), we can calculate that 9.5% of boys taking the math SAT exam scored between 700-800 compared to 5.1% of the girls taking the test, for an adjusted ratio of 186 boys per 100 girls scoring at 700 or above.
The College Board also reports 2012 SAT math test results by gender for all scores between 200 to 800 in 10-point increments (data here), and the male-female ratios for each of those 10-point increments are displayed in the bottom chart above. Here are some observations:
3. Male students outnumbered females for all 2012 math SAT scores of 570 (67th percentile) and above.
4. As SAT math scores increased by 10-point intervals from 570 to 800, the male-female ratio increased in almost all cases, reaching a peak male-female ratio of 2.01-to-1 for test scores of both 780 and 800.
5. Like above, we can adjust for the fact that more young women than men took the SAT test in 2012, and compare the percentage of males who earned perfect scores (1.0%) to the percentage of females with perfect scores (0.43%), which produces an adjusted male-female ratio of 2.29-to-1 (vs. the 2.01 unadjusted ratio) for students who scored a perfect score of 800.
One possible explanation for the fact that high school boys consistently score higher on average than girls on the math SAT test, and outnumber girls 2-to-1 for perfect scores, would be that boys are better students on average than girls and are better prepared in mathematics than their female classmates. But that explanation would be false, when you consider the following data provided by the College Board for students taking the 2012 SAT test:
6. For 2012 SAT test-takers, high school girls had superior overall academic high school records compared to boys: 55% of the students in the top 10 percent of their graduating classes were female, 59% of the students graduating with an A+ grade point average were female, and high schools girls graduated with a higher overall average GPA of 3.42 compared to a 3.28 average GPA for their male counterparts.
7. High school girls were over-represented in advanced AP/Honors math classes (54%) compared to boys (46%), and also in advanced AP/Honors science classes by 55% to 45%.
8. For those high schools taking four years of mathematics, high school girls were over-represented (52%) compared to boys (48%), and more of the high school students studying high school science for four years were female students (53%) than male (47%).
Bottom Line: Even though female high school students are better prepared academically on many different measures than their male classmates, both overall and for mathematics specifically, female high school students score significantly lower on the SAT math test, and the +30-point differences in test scores favoring males has persisted for generations. At the high end of math performance, high school males significantly outperformed their female counterparts on the 2012 SAT math test by a ratio of 2-1 for perfect and near-perfect scores.
And yet, despite the persistent, statistically significant differences in math performance by gender on the math SAT test that continue over time, we frequently hear statements like this: “There just aren’t gender differences anymore in math performance,” says University of Wisconsin-Madison psychology professor Janet Hyde, “So parents and teachers need to revise their thoughts about this. Stereotypes are very, very resistant to change, but as a scientist I have to challenge them with data.”
Given the significant and persistent gender differences in SAT math test scores that have persisted over many generations, the scientific data about gender differences in math performance would seem to present a serious challenge to Professor Hyde’s claims that there are no gender differences in math performance.
Further, the fact that women hold a “disproportionately low share of science, technology, engineering and math (STEM) undergraduate degrees,” and “less than 25 percent of STEM jobs” according to a 2011 Department of Commerce report certainly isn’t because female students are being discouraged from studying math and science in high school. In fact, the evidence shows that females are excelling in math and science in high school – they are overrepresented in AP/Honors math and science courses, and are more likely than their male counterparts to take four years of math and science.
Further, compared to boys, high school girls get better grades on average, and are far more likely to graduate in the top 10% of their high school classes. By all objective measures, girls have essentially all of the necessary ingredients that should result in greater representation in STEM fields like engineering except perhaps for one: that huge, statistically significant +30-point gender gap on the SAT math test in favor of boys that persists over time.
My girlfriend at Akron University states that shes one of like 3 or 4 girls in her engineering classes. Women just think differently than men. Not better, not worse. When will people realize that gender based regulation doesn’t allow for people to be who they are as opposed to who the government wants them to be?
Racist.
May the ghost of Larry Summers haunt your nightmares.
Haha funny.
The only reason this is controversial or opposed is that most people don’t understand probability distributions. Some people who read this think that all males are better at math than all females, and if employers use that heuristic to hire, it hurts females who are good at math. Conversely, there are feminists who don’t understand distributions either, which is why they got so mad with Larry Summers when he said there might be more highly math-adept males than females, as they interpret that as “males are always better than females at math,” similar to the aforementioned dumb employers. In the face of all this stupidity about probability, many choose not to address these gender differences whatsoever, as this knowledge has little utility anyway, and go the PC route of not talking about it. Too bad, because the goal should be to find out more regardless.
Yup. That’s exactly right.
I do have to say, though, that I don’t particularly want to work for a dumb employer. When you’re hiring, you’re not hiring “men” or “women”, you’re hiring a person. So, if you are too stupid to focus on a particular person’s individual’s skill level when hiring and promoting, then do I really want to work for you? In part because of these gender differences, we shouldn’t be surprised by male domination in derivatives or the lack of women in our posses of quants. An observed difference is not a call to action.
two questions:
1. is this gender-based difference also present in other countries ?
2. In international comparisons – if we threw out the girl scores for the US scores, would we score better overall against our international competitors?
Hey Larry,
That’s an interesting question. Do you know if there is some sort of international standardized test like the SAT?
There are several international test : pisa , tismm
re: PISA, NAEP, TIMMS … yup
NAEP actually has a fairly precise definition for “basic”, “proficient” and “advanced”.
many center on the ability to understand and articulate ideas, concepts and real world applications of technology and related.
Depends on where you look for analysis. The official line is that “the more ‘gender neutral’ a country, the smaller the gap.”
A couple of issues spring to mind. First, while the gap appears to decrease, it is never eliminated. More importantly, the data typically used is for 5th and 8th graders (versus college-bound seniors), which fails to account for the faster maturation of girls (i.e., it is possible that the gap increases as young people approach college age).
Just as it is possible for researchers to look at SAT scores and claim that no gap exists, so can researchers look at PISA scores and find the same inexistence.
One could look at France’s “Le Bac” (I could not find a score breakdown. Britain’s GCSE has girls outperforming boys across the board. China’s college entrance exams show boys over-represented among top scorers in math by about 2 to 1, and underrepresented in Art & Soc. Sci. by a similar ratio. Composite scores are equal. Interestingly, girls flat out WHOMP boys in Chinese and English. Just sayin’.
I generally agree. I would emphasize culture more than anything. Looking at China, until how long ago were women given the chance to attend schools and pursue a higher level of mathematics? (I really don’t know)
I would assume that males and females have equal capabilities in mathematics, it’s just up to the individual to have that passion/ “Give a damn”
On what evidence would you base the assumption of equal, on average, capabilities in math? Different hormone profiles have a way of doing different things.
In most important issues we are not interested in equal, on average, capabilities. We are interested in the best individuals out there. The sex with the larger variance is going to produce all the top achievers.
You’re simply parroting the conservative line that there are differences between males and females. Or between blacks and whites for that matter. There are NOT differences. Now that that has been settled (and confirmed by the NY Times and MSNBC), it is only logical to attribute differences in performance to the gender discrimination which has hampered women since the very founding of this country. Until this shameful scourge can finally be eradicated we must launch an affirmative action program (say, by adding points to female SAT scores) to make up for centuries of discrimination.
More seriously: And what of the persistent differences for “Critical Reading” as well?
Interestingly, the trend reverses for the relatively newer “Writing” portion of the test. I wonder whether, by combining analysis of the two trends, one could come up with, say, a gendered “test taker” advantage (or “risk taker,” i.e., the tendency for boys to be more likely, when down to two choices, guess and move on rather than dither). Perhaps the Writing section–a more deliberative, “holistic” test, one assumes–can shed light onto the different processes and how one might further modify the test to highlight different groups’ advantageous strategies.
Interestingly, one could also look at post-admissions data for anti-boy bias (if admitted boys have higher scores, then perhaps they faced a higher admissions hurdle).
Here’s s useful synopsis from UC Irvine:
http://www.oir.uci.edu/adm/IA07-fall-fr-mean-sat-by-sch-gender.pdf
Of course, biases are likely embedded in the data (UCI is admittedly admitting not a group per se but a group of individuals), but I find the differences… interesting.
Other tidbits (from the original SAT document cited by Dr. Perry): You can see some self-selection bias in the relatively higher scores of sophomores and juniors. One could infer that only relatively gifted underclassmen would bother with the test, whereas seniors run a larger range (although not a full range…). Interestingly, freshman scored significantly lower, so perhaps education *is* doing… something. The noticeably higher scores of students from high-income families also reinforces certain themes (I leave it to the reader to side with nature or nurture…).
Fascinating stuff!
Speaking from experience: Your SAT scores will probably take a jump from Freshman to Junior/Senior year because you spend a few years learning helpful tricks. At least I did.
This is particularly true when it comes to things like geometry questions on the math section. For example, you’re given a parallelogram with one angle, what are the other three angles? It’s entirely possible to reason to the answers from limited given information and basic principles, but if you know specific tricks you will get through these problems much faster (and speed is highly important due to time constraints).
These results could be just the tip of the iceberg of a more general phenomenon which could, once understood, explain much more than just the difference of math ability between the sexes. There seems to be some weird goings-on at the very top of many distribution curves, which don’t follow the trend the curve itself was pointing to. The Jewish population, for example, would be another example of this: does the normal curve explain, at the high end, their incredible intellectual performance? I don’t think so — there are many more of them there than should be, according to the distribution curve.
Understanding this is crucial, it would seem, and some have posited the “smartest fraction” theory: it is those at uppermost reaches of the intellectual scale that are mostly responsible for advances in civilization. Zeroing in on these outliers will be the key to understanding the phenomenon. What does the distribution curve actually look like at those altitudes?
It might just be that these extreme individuals will be responsible for the next “leap of being” — an expansion — in human consciousness, just like a few ancient Greek geniuses might have been responsible for the last one.
These people are so precious — to humankind — that it would be better to scratch most of the educational programs (to think of the trillions already wasted!) and concentrate mostly on them, on the super-gifted, the 1% of the 1%. It is they who might one day give the rest of us an inkling as to what the future holds for the human race.
Very Eugenics-esque. You’re implying a scary level of social engineering. The 1% of the 1% will figure it out without the help of the rest of us relative dummies. Just get the hell out of their way. IF the ancient Greek geniuses were responsible for a leap of consciousness, they did it without the help you’re advocating.
But, I doubt that many of the inventions that have bettered our daily life have come from this group (which, if you’ve ever met them, has a really hard time figuring out other human beings and how to perform some mundane tasks necessary for life, among other things). I doubt the weed-whacker or the dishwasher or the vacuum cleaner or the little tab that makes packages easier to open was dreamed up by a super genius. I’m wary.
“(which, if you’ve ever met them, has a really hard time figuring out other human beings and how to perform some mundane tasks necessary for life, among other things)”
haha like what? Mating? Maybe that’s why they’re still at such a low percent of that 1%.
sorry, couldn’t help but laugh at that.
and don’t forget matching their socks. Or shoes.
Actually, I was thinking more along the lines that some of them have a hard time channeling themselves into productive activity. They’re just “out there”. Generalizing about people is always limiting, of course, but ….
The “super gifted” will tend to work it out themselves, they don’t really need “help”. What we really need to do is to have an agile educational system that can maximize the performance of those with lower IQs, which is something our stagnant public education system is unlikely to accomplish with current management structure.
On the other hand, it might be useful to understand the genetic basis of IQ, to provide future parents with the ability to tweak their kids IQ up a few notches if possible.
Yep. It might be, but it seems that just measuring IQ is not as easy as you’d think. Nor is IQ the whole story if the goal is success in life (which is itself hard to define). I have a friend who is doing research in that area and after I broached the subject with her, I ended up more confused (which, I remember one of my professors telling me long ago, is a good thing because it means you’re learning something new).
methinks”: “Nor is IQ the whole story if the goal is success in life (which is itself hard to define).”
That’s very true. Thomas Stanley, in his book “The Millionaire Mind”, shows that neither high IQ nor graduation from top schools are predictors of wealth accumulation. Instead, he attributes economic success to excellence in “Discipline 101 and Tenacity 102″. I couldn’t agree more with Dr. Stanley.
}}}} as opposed to who the government wants them to be?
As opposed to who some victim-interest group wants you to be, you mean.
I just had to tell you that your nom de plume is my absolute favourite of all time.
MARK, if there’s a way to tell this system to organize oldest flow down to newest, that would Be A Real Good Thing.
I hate this “newest first” crap that ALL the noob-written commentaria are using these days, since it doesn’t allow responses to FOLLOW their precedents.
I’ve put in a request to see if the comments can be ordered chronologically.
I second the motion, I thought about suggesting it myself.
I guess I am missing the problem, IGotBupkis. You just have to make sure you are using the Reply button to the comment you are replying to and not the topmost Reply button to put your reply under the original post. I prefer the newest post on top myself.
Ah, but the newest does not go to the top on replies. Is is kind of screwy.
*It* And, we have no preview or spell-check options to correct stupid spelling and typo mistakes!
A contrarian spin on the number of boys/girls taking the exam. Over 100K more girls took the exam than boys. What if you added 100K boys to the exam? That would likely make their numbers go down significantly (as those 100K more boys would be less academic than the 700+K already taking it).
Just to clarify: obviously it wouldn’t change the fact that boys are more likely to score high on the exam, but it would affect the “adjusted” ratios. Also, it would likely bring the lower scores closer to parity.
I have not looked at this years statistics, but have in the past. The past standard deviation difference has been about 1/3 of a standard deviation minus, as the sds are set to be around 110 points. The differences actually hold well across races. In the past, the white girls were about .3 sd above the mean. the white boys about .6 sds above the mean. The Asian girls were about .6 sds above the mean and the Asian boys were .9 sds above the mean. Moreover, the variances for both genders is about the same. If an additional factor of discrimination were present, it should manifest itself as a wider variation.
However, normalized statistical scores are designed to show differences, but do not show what the magnitude of the differences are. If two groups differ in 10 questions out of 1000, there would still be variance that could expressed in the terms like the SATs, but really be meaningless in terms of performance. Furthermore, it’s difficult to extrapolate these scores to actual performance in a mathematically related field as persistence and interest play a strong role.
But the differences do suggest that there would be more boys than girls at the top level ability in these fields. However, give the large numbers, there are more than enough girls and boys to fill the positions requiring these skills. So, it would be incorrect to say that of the people in these fields one gender is better than another.
The take home lesson is that there are three fundamental errors in interpreting statistical differences. The first is that the mean describes the entire group. Just because the mean is different does not mean that all the members of one group are better than the other. This neglects the standard deviation. There are plenty of girls that have higher SAT math scores than most boys. The second error is to use the tails to describe the population and neglect the mean. As in saying based on the 2x greater top scores for males means that they are all higher than females. No there plenty of males that have low scores. The last error is that small differences in averages have no meaning. They do have an effect and it’s seen in the tales. As evidenced by the previously 2x higher top male scores than female.
Isn’t this a bit like the observation made in The Bell Curve? That small differences in the mean IQ level translate to fairly large differences at both ends of the distribution curve?
Hopefully you’ll give the same treatment to the verbal parts of the SATs.
And as an aside: THE FACTS ARE YOUR FRIENDS.
Why? No one is claiming that there is no disparity in the verbal parts of the SATs. At least, the usual suspects don’t care that the disparity is largely in favor of women.
We know math requires a lot of repetition and we know women don’t repeat themselves often, although they tend to talk more.
So, that may explain why more men are better than average in math and more women are better than average in language
Hogwash.
What I did notice, going through the data, is that Asians are overrepresented by a factor of two among the testtakers.
I also wonder what the median amount of times Asians took the test(since most who took it, took it several times) compared to others.
Another factor is socioeconomics.
The average Asian household earns a lot more than everyone else, in large part because Asian immigration is very heavily skewed towards high-skill individuals(which is good for us, but it doesn’t really give a clear picture).
For example, in the 2009 PISA test, if you only looked at American white students, they were ranked third in the world, just after S.Korea and Finland and ahead of Japan.
(No, Hong Kong/Shanghai doesn’t count since it’s the elite who live there. Until China tests a nationally representative sample we don’t know how well they’ll do).
But the thing with PISA, however, is that you can only take it once and it has to be representative of the population.
The SAT, on the other hand, can be taken many times and is easy to prepare for.
Not so with PISA.
Commissioner’s Remarks – NCES Statement on PISA 2009
“In reading literacy, female students scored higher, on average, than male students in all 65 participating countries and other education systems.
In mathematics literacy, male students scored higher on average than female students in 35 countries and education systems; female students scored higher on average than male students in 5 countries.
In science literacy, male students scored higher on average than female students in 11 countries and female students scored higher on average than male students in 21 countries.”
Well, it’s amazing.
When a study gives as a result that males are better than females in something, everybody feels that there must be some explanation and that possibly there is a bias in the study. When a study concludes that females are better than males in something, nobody analyzes the causes.
Isn’t it sexism?
Male and females are just different, so they will excel in different areas in the aggregate. I kind of like that that we are different because it makes the pieces fit together better.
Give a male and a female the same problem to solve and the male will go over in the corner and work on it by himself and the female will call a friend and discuss it with her. SAT testing methodology is not working to female collaboration strengths.
Even when a female knows the correct answer they will often leave it blank because they are not sure. I have to caution my students on national certification tests to not leave empty answers on multiple-choice tests, and I see this problem much more with females than males.
I assume that females test better in some areas than males on the SAT is more a matter of male weakness in those areas than female strengths (give a male a writing assignment and his first reaction is why do I have to write about that garbage, but females almost never question my reasons for written assignments).
Statistics hide individuals. Rather than adding to 40 replies here, I will be blogging about this on Necessary Facts. For here and now, Asians do better on the SAT — but need to score 140 points higher than Caucasians to have an equal chance at college admissions. Family income also correlates to SAT score. The best predictor of success in college seems to be your high school grade point NOT weighted for AP classes rather than your SAT scores. Moreover, the definition of “success” is itself a variable. Steve Jobs and Bill Gates only followed in the tradition of Edison.
I am wondering how this display of Male superiority adds up when accounting for the gap in Male and Female participation. If there were say, many more Females taking those tests, would not this bring the average score for Females down? If say the Male participation was made up of more cream of crop achievers, would this not exagerate the score for boys?
Perhaps you should look an INTERNATIONAL statistics before commenting on gender differences. While the data strongly supports female inferiority in mathematical achievement in the U.S., the data does NOT show a gender gap in other countries. The U.S. has some cultural issues when it comes to mathematics, and once females fall behind in mathematics, they are precluded from hard engineering and science disciplines. We have a cultural problem.
Who cares what your gender scores, I think one should only care about personal merits. If you score avg., it means your stupid. If you are smart your goal should be to score above everyone. There are idiots of both genders, so rather than saying one gender is not as smart as another, just make everyone of both genders feel inferior to you.