Powerful visualization of gender-skew in AP CS from USA Today
January 30, 2014 at 1:01 am 10 comments
The chart below (above, here in the blog) shows the ratio of boy to girl test-takers across AP exam subjects. In subjects whose bars do not reach the orange line, girls outnumber boys. In subjects where the bar extends past the orange line, boys outnumber girls.
via AP Test Shows Wide Gender Gap in Computer Science, Physics – Data Mine usnews.com.
Entry filed under: Uncategorized. Tags: APCS, BPC, NCWIT, women in computing.
Computing Education Research Blog 
1.
Max Hailperin | January 30, 2014 at 1:08 am
Powerful, yes, but also powerfully misleading. The linear scale of ratios makes a ratio of 2 look twice as far from a ratio of 1 as a ratio of 0.5 is, even though they are in fact symmetrical. This visually accentuates the gender imbalances that are male-heavy, such as CS and Physics, versus those that are female-heavy, such as Studio Art.
2.
Mark Guzdial | January 30, 2014 at 8:57 am
Interesting — thanks for the insight. I’m sure you’re right, Max. I do know that the Studio Art numbers and the CS numbers are not symmetric. There are more males taking AP Studio Art than there are females taking AP CS.
3.
Max Hailperin | January 30, 2014 at 9:00 am
In a properly drawn graph, we could see that the Studio Art bars don’t extends as far to the left as the CS bar extends to the right. (The bars would be based at the ratio 1 and extend in one direction or the other on the log scale.) In the current graph, you can’t see how much worse the imbalance in CS is, because you can’t see anything, really.
4.
gasstationwithoutpumps | January 30, 2014 at 2:28 am
What Max said. When you are touting ratios, you should DEFINITELY be using a log scale. Come on, this is simple presentation honesty. Even someone without statistics background should know when to use a log scale.
5.
Dennis (the other one) | January 30, 2014 at 5:44 pm
USA Today isn’t known for ‘presentation honesty’. Unfortunately, sensationalism helps their bottom line more than honesty (at least, they think it does). I always turn to USA Today when looking for examples of misleading visualizations. I hope that any sensationalism of the real gender-imbalance does not detract from the honest efforts of people attempting to address the issue.
6.
Alfred Thompson | January 30, 2014 at 9:22 am
We worry rightly about the imbalance when female students are under represented but is anyone worried about the areas were male students are underrepresented? If not,why not?
7.
Mark Guzdial | January 30, 2014 at 9:26 am
I think we do in nursing, when there is a shortage. We don’t currently have a shortage of Studio Art majors. While there’s some question of whether we have a shortage of STEM majors overall, it does seem pretty clear that we have a shortage of computing majors — and NSF believes that we can’t resolve it without bringing in more women and under-represented minorities.
8. The Problem with Computer Science | Geeky Mom | January 30, 2014 at 8:55 pm
[…] struggling to make CS count or to make CS more compelling, especially to young women. Mark Guzdial recently cited this US News article that breaks down all AP test takers by gender. By far the […]
9. Creating a better data graphic for Advanced Placement exam data | Brian A. Danielak | January 31, 2014 at 1:57 am
[…] News then went ahead and made this graph, which Mark Guzdial linked to on […]
10.
Brian Danielak | January 31, 2014 at 2:09 am
Mark and Max,
I went and pulled the original AP Data and created a new graphic that I think addresses some of the concerns commenters are raising:
Here’s my blog post on creating the graphic, where I detail the same kind of reasoning Max is talking about:
http://briandk.com/2014/01/creating-a-better-data-graphic-for-advanced-placement-exam-data/
In short, there are many visual issues with the graph that have a bearing on how we think about gender disparities.