@QuantMech I can certainly see how kids from different social environments may differ in understanding and familiarity with these type of questions, but I fail to see why boys and girls would differ.
Not sure that’s a great example as EPGY, as good as it is, had a lot of bad grammar and questions (back when my kids were using it). They just don’t have the unlimited budget to test stuff like College Board does. 
With the “equation” vs. “number sentence” issue, you would still have the question whether “box” means anything, or it is just a place to put the answer to a normal question. I could see boys and girls reacting to this a little differently. Also, even these days, sports-related questions probably slightly favor boys–perhaps by a question or two. Depending on the test, that could generate a three-months difference in scores.
Despite its validity studies, it took CB years to eliminate “regatta” from its vocabulary questions. There used to be a writing prompt given as a sample for the CB writing test that asked for an essay on country clubs. (Oh yeah, that has no socioeconomic correlation.)
I thought of another example, which is not gender-related, but which does depend on environment to a certain extent: A question asked about the units that would be appropriate to measure a parking lot. If the students had all been exposed to “acres” in their direct instruction, that is reasonable. Otherwise, I would think that rural students would probably know the answer from their day-to-day experience. Children of real estate professionals and students whose family had recently purchased a house with a large lot would probably have been hearing a lot about acreage. Students whose families watch HGTV might have encountered the term. I don’t whether urban students without those circumstances hear about acreage much at all. To the best of my recollection, “acre” was just one of those unused terms in our house up to the time my daughter was eight. We had no particular call to use it.
In Classical Electrodynamics (for many years a standard text for advance undergraduate and graduate-level classical E&M), J. D. Jackson comments about the “natural” unit system used in relativistic quantum field theory. The speed of light in a vacuum and Planck’s constant are not only equal to one, they have no dimensionality. This system of units “has only one basic unit”, usually “chosen to be length.” Then all other quantities, “whether length or time or force or energy, etc. are expressed in terms of this one unit and have dimensions which are powers of its dimension. There is nothing contrived or less fundamental about such a system than one involving the meter, the kilogram, and the second as basic units (mks system). It is merely a matter of convenience.” As Yoda says, “You must unlearn what you have learned.”
With regard to bluebayou’s statement, I think that beyond a certain age, people know what “two out of three” is intended to mean–that it does not refer to some specific set of three that is a subset of the whole group, but rather expresses a proportionality that applies to the entire set. But it takes time and circumstances to be exposed to and learn that type of expression. I think it is definitely more common in some households than others. But I would not expect it to be flagged in testing the validity of questions.
My daughter recalled all of the questions on the 3rd grade standardized test that she had doubts about, and asked me about them. I recalled several of the questions as being identical to questions I had seen in the 1960’s, when I took what was apparently the same standardized test or a closely related one. Once they had been “validated” for students 60 years ago, I am not sure whether they were re-validated later.
The underlying issue is: The testers can’t share the questions, because it is expensive to validate them, and they can’t keep rewriting them. But if they don’t share the questions, it is hard to tell whether the differences actually matter or mean anything. At the same time, “girls are no good at math,” is being drawn as a conclusion by many of the children, if not the adults.
I think the SAT lost a lot of its mystique when people learned that missing a single question in math could drop the score from 800 to 780, 770, or even 760 on an unlucky day.
If the testers cannot share questions (as they probably can’t), at least they might provide raw score data, so that people could judge how large a difference “three months” or “six months” boils down to.
I think the current study is being used to sell newspapers. It would not be a problem if it did not have the side effect of making girls uncertain about their general math ability, based on their gender.
@QuantMech – SNL had a funny take on the testing bias issue back in the 1970s (I think). Aykroyd, in his bow tie and suit, versus Morris arguing about bias in a test question that asked something along the lines of, you’re on your yacht in Martha’s Vineyard and are having fish for dinner, which wine do you serve: red or white?". I, as a hick from TX, sure as hell didn’t know the answer!
BTW, as an amateur crackpot physics guy, think that the units issue is only a problem for people who have “over thought” CEM. I think that the Doppler Effect is the basis for all the confusion out there (e.g. the Compton Effect is most definitely a Doppler effect plus more unknown stuff).
I need to amend post #39. Of the moms not working outside the house in that group, it turns out many of them hold STEM degrees, often from prestigious universities, including PhD’s in science, etc. They worked in their field for a few years, then got married, had their first child, and stopped working.
One thing to note is that this study only talks about average math scores. But if you look compare girls that do well in math contests with boys that do well, girls are more likely to be found at schools where a lot of students do well in math contests (see this study - http://news.mit.edu/2009/math-gender). So at the top levels, you get a slightly different effect than what is talked about in the OP.
^ also girls are (even) more likely to be Asian.
Re #46, It’s heavily a social effect. If a lot of students do well in math contests, there is less social stigma associated with doing well in math contests, especially for girls.
The American Mathematical Society published a study of participation in the Math Olympiad teams by girls, broken down country by country. The US has had very few female members of the teams that compete internationally. There were some fascinating studies of the difference of participation by girls in West Germany vs. East Germany (prior to reunification), in the Czech Republic vs. Slovakia (after the split) and in North vs. South Korea. In terms of ethnic composition, you would expect both nations in each of the pairs to be fairly similar. But the participation in IMO teams by girls was radically different.
In my opinion, if it is widely publicized that girls do not do as well in math as boys, this is likely to have a depressing effect on the number of young women who pursue mathematics. There may be a few “rebels” who rise to the challenge. But I certainly have had periods of “gender threat” where I felt that the pursuit of quantum mechanics (and the math that accompanies it) might mean that I was less feminine. Some people can withstand this better than others.
The most important point is that there may be a very small number of questions that account for 3 to 6 months difference in the reported scores. If one could see the questions, the reasons might be discernible. Or it may just correspond to small, almost random differences in the responses to multiple questions.
One of the tests for validity of a standardized test involves seeing whether questions tend to be answered better by the higher scorers overall. Obviously, this type of test for validity will not detect a gender differential in itself, if the girls as a group do not score as well. I think that practically all validity tests now include a breakdown by gender, to test for validity. But if girls really were worse at mathematics than boys, it would be difficult to maintain the differential while making all of the question outcomes gender neutral. (IQ tests give the same average scores for boys and girls because they are specifically rigged that way.)
Also, the range from high to low scorers is likely to be multiple years, and that would correspond to quite a few more questions.
IMO, the results of the study in the OP is primarily social effect. When my kids started GATE in elementary school, the classes had slightly more girls than guys, but only a few. But when middle school rolled around, I was struck by all the kids who voluntarily dropped down to the slower math track. In particular, was noticeable in that there were a lot more girls that dropped down than guys (as a %), essentially precluding the possibility of Calc BC as a senior. (District does not offer AB.)
It was not that many years ago where AP Cal BC had 65% guys taking that test, with a similar number of girls taking the (less rigorous) AB. Fortunately, we are getting closer to 50:50 in BC.
fwiw: middle-upper class district consistently testing as one of the top public districts in the state.
I agree with you, bluebayou. We have seen the same effect around here, but mitigating somewhat now.
I always wonder if it even matters. Pay equity is now very narrow and predominantly female held professions have made large strides in pay (eg teachers and nurses). So what if slight differences are hardwired in the dna? Or so what if slight differences in career aspirations are gender preferred? Is that necessarily something that needs to be manipulated through “special programs?” And let’s face it, many women do drop out of the workforce to stay at home and/or drop down to part-time work chose majors that lead to more flexible types of industries and jobs. That concept is not foreign in the least.
Well it does for the purposes of the study referenced in the OP.
Few students are a year ahead in English, few double up on English courses, and high schools and colleges are less likely to consider English a leveled subject like math. So “four years of English” and “English every year” tend to be as close to a hard requirement as there is for college prep.
With respect to math, most college bound students should continue taking math through high school.
Navigation apps give turn direction. Where signs refer to N/S/E/W, they also give that. For example, “turn right onto GA 400 south”.
Note that sometimes it is not obvious beforehand whether an entry ramp to a freeway is on the right or left, so the one giving directions should give notice beforehand to get into the correct lane, rather than expect the driver to cross three lanes in the last 100 feet.
No sure what the N/S/E/W direction thing has to do with math. I don’t/can’t follow directions that way but I was/am very strong in math. Was in honors math in HS and had no problems in calc… Interestingly my favorite math class was Geometry which made perfect sense to me.
Another example of rich white male privilege: they get good grades handed to them. Very racist.
I searched our district … girls outperform boys in both math and English.
I kind of wonder if some of this is because rich, white girls are allowed to consider more liberal arts interests while poorer girls are encouraged to take up stem careers because that looks more secure.
@turtletime I noted that possibility in comment #5 and that it is similar to the explanation put forward for why more girls do STEM courses in developing countries
http://talk.collegeconfidential.com/discussion/comment/21590798/#Comment_21590798