Hacker News new | comments | ask | show | jobs | submit login
[flagged] Where Boys OutperformGirls in Math: Rich, Whiteand Suburban Districts (nytimes.com)
27 points by jbredeche 8 months ago | hide | past | web | favorite | 20 comments

I have a hard time with these studies because they obviously had an agenda before they started.

Rural areas likely do not have as much communication with friends and the income to support a tech lifestyle.

While you grow up in the suburbs, you see your friends get the latest video games, you spend more time online. One day you decide you want to 'hack' and 10 years later you are making apps and websites.

There seemed to be a culture among my friends that math/tech was extremely important. I know my sister had very little interest, they'd play outside with my both male and female neighbors who none ended up being good at math/tech.

I'm very difficult to please, but I think this piece was actually pretty well done --> using large standardized samples, pointing out trends without drawing conclusions.

As a rural (non-US) person who did ok at math and whose children are doing great at it, I can only guess that you have an agenda before you read the article ;)

To be fair, this is an area where it is impossible not to have an agenda. Both being pro-status-quo or pro-something-here-should-change are relatively radical in the eyes of the other, and there isn't a lot of middle ground.

Ideally, more money would be sunk into raising the statistical literacy of everybody. Then we can all talk sensibly about the statistics involved here and take it for granted that Simpson's Paradox is being considered in all studies involving gender :P.

The way the article's written, it's putting much more emphasis on the slight and sporadic outperformance of boys in math, rather than on the much stronger and more consistend outperformance of girls in language. That's the problem with the agenda, IMO.

Edit: as mlthoughts2018 points out, the title could as well be:

"Boys, increasingly abandoned in English class, turn to math as a domain in which they can be rewarded for success-- especially in rich, white suburbs."

I grew up in a rural area and had the same experience with math and tech as you. I don't think whether or not I was in a rural area was a significant factor in encouraging or discouraging an interest in tech.

What's the agenda?

I can't find it now, but I seem to remember a study that showed that boys dominated both ends of the "bell curve" when it came to academic achievement: the top achievers were boys, but the lowest performers were, too. The author here doesn't seem to be particularly interested in averages, which would make for a fairer comparison.

Yeah, I was also missing some measures and discussion of absolute performance (both girls vs boys, and rich vs poor).

Alternative title: "Where Girls Outperform Boys in Math: Low Income, African-American districts"

Why does it matter?

Not the top level commenter, but my take on that would be that you can tell a lot of different stories with the same data

Maybe if it was framed this way, we might be looking into solutions to help struggling boys in lower income areas

Not that one is more or less valid than another, just that different framings of data can lead researchers to different conclusions/solutions

Wow, pretty interesting to see my school district as one of the closest 15 to the bottom right corner. Looking back though, I can definitely see it. Most of the kids in my AP computer science class were boys and I now see way more guys from my high school in STEM jobs then girls.

In my opinion, the biggest cause was due to gender stereotypes. We weren't the actual school based on for the movie mean girls (although lots of people joked that we were), but it had a very similar vibe in that math was not seen as being cool for girls to do and popularity was definitely a major focus for a lot of people.

Edit: reading further into the actual paper, the first possible reason they site are gender stereotypes. It's a pretty interesting read. I would at least skim if it if you are at all curios and it's a shame that this article seems to have been shadow hidden from the front page despite having 25 points in an hour: https://cepa.stanford.edu/sites/default/files/wp18-13-v20180...

"Instilling children early with motivation and confidence to do well in school is crucial, researchers say. When students reach high school and have more choice in the classes they take, the gender gaps in achievement grow even larger."

Interestingly, it doesn't appear to be the gender gap in math that grows larger, but instead the gap in writing and language.

The correct title should be: Boys that live in rich, white and suburban districts outperform girls in standardized test scores.

It borders on dangerous lack of journalistic responsibility to write a story like this which does not even mention the effect size observed.

The closest statement I can find to something like mentioning an actual effect size is this:

> "In the Montgomery Township district in New Jersey, for example, the median household income is $180,000 and the students are about 60 percent white and 30 percent Asian-American. Boys and girls both perform well, but boys score almost half a grade level ahead of girls in math. Compare that with Detroit, where the median household earns $27,000 and students are about 85 percent black. It’s one of the districts in which girls outperform boys in math."

'Boys score almost half a grade level ahead' -- ahh so much imprecise terminology that we can't know what it even means, let alone if this example was representative or cherry-picked.

I'm not talking about holding a popular newspaper to some unrealistic standard, like giving us informative plots of the spread (uncertainty) of results or minor description of the methodology or anything. I wish, but there's no way.

I'm saying they aren't even throwing any numbers into the clickbait. The takeaway from this article is purely subjective, because we don't know what ' grade level' or 'almost' or 'better' mean in any of the description. What does "5 months of grade level" mean, and how does it vary by state or district, etc?

Normally there's at least something like, "Boys did X% better on SAT math sections between 2010-2016" or something, and you can look at X% and decide whether it looks huge and meaningful or looks small enough that it might be spurious or have no meaningful effects in students' lives down the road or something.

These just seem like made up numbers in units of months or "grade level". How do I know what it means? Given that the blue point cloud is roughly around the 0 line, and I don't know how much "1 unit" "means" on the vertical axis, I seems like there's no conclusion to draw.

In fact the most salient effect size mention is about girls and about language arts:

> "In no district do boys, on average, do as well or better than girls in English and language arts. In the average district, girls perform about three-quarters of a grade level ahead of boys."

Maybe the title should be something like,

"Boys, increasingly abandoned in English class, turn to math as a domain in which they can be rewarded for success-- especially in rich, white suburbs."

(The point is, it's just as consistent with the utter lack of detail of the article.)

> 'Boys score almost half a grade level ahead' -- ahh so much imprecise terminology that we can't know what it even means, let alone if this example was representative or cherry-picked.

Well the paper is linked in the third sentence if your curios and want to get all the details [0]. Newspapers generally try to keep it simple for the average person. From the paper: "We estimate the mean math and ELA test scores for male and female students for each of roughly ten thousand U.S. school districts in grades three through eight from the 2008-09 to 2014-15 school years. These data enable us to estimate male-female testscores gaps, as well as changes in the gaps over grades and cohorts within districts, providing adescription of gender differences in academic performance at an unprecedented level of detail.

[0]: https://cepa.stanford.edu/sites/default/files/wp18-13-v20180...

Why does the paper being linked have a bearing on this?

The point is that even in a popular article or cursory summary, you at least need the context of an effect size with interpretable units.

If you have to follow a link to the paper even for that then the article itself would have to be worded in an entirely neutral way (as in, not indicating that the data supports any particular conclusion).

Any non-neutral presentation (like saying the data supports any type of conclusion about boys’ math performance in rich, white suburbs) will create subjective impressions about what the result means (which can be fine, so long as an effect size and interpretable units are attached).

People seem to disagree with my comment because the study (among others) was hyperlinked in the Times article. This strikes me as entirely missing the point of my comment.

Because this is a newspaper, not a scientific journal or peer review and newspapers are generally written for the average reader who probably does not want nor even understand all of those details.

But a basic notion of effect size and units are often included in popular news coverage of statistics. That's a basic expectation, involving zero nuance or statistical rigor.

"This is a newspaper" is a reason why they might leave off details about methodology, metrics of uncertainty, etc. Not a reason why they would make an entirely unqualified claim attached to no notion of the effect size.

Essentially you're saying we should hold the New York Times to the same reporting standards as some clickbait site that fuels confirmation bias with sensational headlines, and not expect large world-spanning newspapers to have even the slightest of better practices than that.

Applications are open for YC Summer 2019

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact