AP Exams

At our last meeting we talked a lot about AP exams - specifically the AP Computer Science Exam. I know Anthony already discussed the extraordinarily low numbers of students who take the AP CS Exam, but I wanted to discuss that further and also the gender distribution of the exams.

After a lot of poking around on collegeboard.com and downloading lots of files, I came up with some graphs and a spreadsheet of some AP CS gender related data from 1996 through 2009.

I found that the AP CS Exams - both the A and AB exams - have never accounted for more than 1.7% of all AP exams in a year (and that, unsurprisingly, was in 2001). It has been steadily declining since then - leveling off in recent years. The AP CS exams now make up .72% of all AP exams taken.

A small distinction is the number of students who take one of the AP CS Exams (referred to from now on as AP CS students) out of all students who take AP exams (referred to from now on as AP students). AP CS students have never accounted for more than 2.85% of all AP students and in 2009, that number was down to 1.24%.

These are alarmingly low numbers. In 2009, only 16,061 students took the AP CS A Exam and only 4,900 took the AP CS AB Exam for a total of 20,961 taking either AP CS Exam. Compare this to something like Calculus (which isn't even the most popular AP Exam) which has 108,249 students taking the AP Calculus AB Exam and 28,974 students taking the AP Calculus BC Exam for a total of 137,223 students. That's more than 6.5 times the number of students!

Well, okay, you say, Calculus is far more general subject than CS. It is far more applicable to more fields. Okay, then. Let's compare CS data against CS data. In 2009 we were less than 2500 exams off from the peak number of AP CS exams in 2002. In 2002 23,459 students took an AP CS exam. In 2009, 20,961 students took an AP CS exam. That's about 89% of what it was. "Not bad," you might say, "I still don't see the problem - 90% of the peak." However, it should be noted that the number of AP students in 2009 was 1.8 times the number of AP students in 2002. And that the number of AP exams taken in 2009 was 1.85 times the number of AP exams taken in 2002. Taking that into account, the number of AP CS exams in 2009 is 61% of what it was in 2002.

Then there's the question of women taking AP CS exams. In 2009, women accounted for 55% of all AP students. However, they did not account for 55% of AP CS students. In fact, they accounted for less than 18% of AP CS students. Women accounted for 19% of AP CS A students but only 13% of AP CS AB students. And this is the best it's ever been.

While the number of students taking the AP CS Exams has been going up over the past few years, the percentage of all AP Exams that are AP CS Exams and the number of AP Students that are AP CS Students has been declining and is only now leveling off. While the percentage of women taking AP CS Exams has been slowly increasing over the past few years, still fewer than 4,000 women are taking AP CS Exams - and the percentage (18%) is nowhere close to the number of women taking any AP Exams (55%). Fewer than 700 women are taking the AP CS AB Exam.

While I recognize that not everyone who takes CS in high school takes one of the AP CS exams, the numbers are alarmingly low. Something needs to be done to introduce students to CS in high school! I don't know what, but something...

AP Tests

So I've been conducting focus groups with International students recently. I've been listening to lots of different inputs and insights into computer science perception and how women fit into the picture of computer science both in their home country and in the United States. A recurring theme has been the exposure to computing and computer science prior to arriving in college and being forced to make an educational decision that will impact the next 4 years of their life at least and how women aren't enrolling at numbers comparable to men in computer science. At one point one person told me that making computer science compulsory in high school would close the gap significantly. This got me to wondering exactly how many people in high school are getting exposed to computer science in some way, shape, or form.

Now, I know the AP test is by no means an exact or even remotely close to exact figure of how many people are taking computer science in high school. I also know that the AP test is more a measure of your programming knowledge than anything else. However, for someone to take the Computer Science AP test, it suggests that they believe to have had some exposure to something akin to collegiate level in computer science. After digging around on the College Board website for a while, I made a list of the number of people who took all of the AP tests in 2009. The chart is below:

From this information we've learned some information that at least to me is a little bit shocking. I mean, I know that not very many people get exposed to computer science prior to high school. Going in I knew computer science would be dominated by tests like Calculus and English. However, there were quite a few that surprised me considerably. For one Art History has significantly more takers than Computer Science, almost 33% more. More people in high school are exposed to a collegiate level of art history than they are to computer science. I have never met anyone who's met anyone who's taken the Art History exam but it's still taken more often than Computer Science. In fact, about the same number of people are exposed to Computer Science as are exposed to Spanish literature. I don't know this for a fact but I get the feeling there's a lot less people majoring in Spanish literature than in computer science. Why then are the numbers taking the AP tests comparable? While collecting this data I cringed with every new page I opened up (and not just because half of College Board's data is corrupted, which is why several figures are missing including Spanish Language and Music Theory). I could make arguments for issues of class difference that keeps computer science being as popular as microeconomics since it's probably more expensive to teach, but it's hard to make that argument for all of the tests it's outstripped by.

Comparing Global Statistics

This week I've been trying to compile some data on global levels on computer science enrollment. I think it would be really interesting to be able to access an all-in-one resource that would give data on computer science students all around the world so that I had a better picture of what exactly was happening internationally. However like most things the reality of the data is a lot more complex than it is in my head. I've run into more than a few little caveats that make my question just how valuable any numbers I find will be.

Finding data in the US generally isn't a problem. The NSF keeps statistics regularly for most disciplines and for computer science we have the luxury of the Talbee reports which have demographic information for undergraduate and graduate students that give a very nice picture of how Ph.D. granting America universities are doing enrollment wise.

However, looking internationally the picture gets a lot more confusing. It is difficult to find the equivalent of the NSF or other large reporting organization body that will handle statistics like this if they have one at all. Reports are littered with all kinds of figures that are often outdated or not sourced at all. It's hard to find where exactly these numbers come from if they arise at all.

Vashti Galpin has been sort of my inspiration for a lot of this as her 2002 paper Women in Computing Around the World gave one of the clearest pictures I've seen of the international enrollment levels. However, 2002 was 8 years ago and I imagine some of these numbers have changed so I was hoping to get updated data. It's only once I've begun digging that I've realized how important the many footnotes her paper included on the statistics are.

Most specifically, I've been wondering about the large section of her paper that details the many different computing disciplines she found statistics for. I remember blogging early about our lack of a common language but I'm wondering how can we measure a degree in Informatics against a degree in Computer Science. I've spoken to many people who say they're the same but I wonder again how a name can affect enrollment.

I find this point especially intriguing after our first focus group. A couple of the participants who are now graduate students at Carnegie Mellon in computer science did not do their undergrad work in computer science. Rather, they found their passion through computer engineering. The work they were doing in computer engineering was still computer science by Carnegie Mellon's definition but in their home countries they still had computer science programs. While I've been so focused on collecting statistics for only computer science it's becoming more obvious that I may need to be doing more investigating into computer engineering, informatics, and other disciplines.

I think if I stop and take my bigger question of "Who is doing Computer Science?" I can look at numbers all I want but I'll never get the answer. Only a closer reading of Galpin and the actual work being done around the world can I better understand where my data is coming from.

Stereotypes

A friend of mine just e-mailed me a link to two articles about women in computer science (thanks Jack!) which I thought were interesting and I thought that I would share.

The first article is about a psychology researcher, Professor Sapna Cheryan who did her PhD at Standford and is now a professor of psychology at University of Washington. She is looking into computer science stereotypes and how they affects women's and men's interest in computer science. Professor Cheryan sets up two computer science classrooms - one which is filled with stereotypically geeky things like Star Wars posters, computer parts lying on the floor, cans of soda all over the table, and Star Trek figurines; and another which is filled with items that are not associated with the "geek" stereotype such as water bottles, neutral abstract art posters, and potted plants. She brings subjects in and asks them to rate their interest in computer science.

Cheryan [...] has placed students in situations like this for nearly five years. She has found that women rate themselves as less interested in computer science than men in the “geek room” described above. But in a room decorated more neutrally with art posters, nature photos, and water bottles, their interest levels were about the same.

The article goes on to describe Cheryan's explanation of why there is such a difference in interest levels between men and women when placed in these situations:

Women don't identify with the archetypal image of computer scientists. Cheryan's subjects describe this image as “nerdy, techie, stay up late coding, energy drinks, no social life.… They don't frequently take showers.” The geek room conjures this picture in our minds, Cheryan says, based only on the stuff we find lying around.

Now Cheryan has moved to virtual classrooms which allows more control over the environment and has allowed her to place avatars in seats in the classroom or a teacher. Going back to this theme of "seeing people like you" - Cheryan can place all women in the seats or put a female teacher in the front of the virtual classroom to see if these things change interest levels.

The virtual classrooms have been put into all sorts of configurations with different avatars or no avatars. Students were again asked about interest in computer science and also about which room had a more masculine vibe. Cheryan found the following:

Every time they changed the study, the results were the same: Most women avoided the geek space. When prompted, many said it gave them a masculine vibe. The more masculine they found the room, the less they liked it.

Very interesting article!

The second article was much shorter and focused on "pair programming." The studies conducted seem to based on a gender differences model - that pair programming would make programming a more social activity and thus attract more women. However, the findings were that pair programming helps retention regardless of gender.

“I thought it would give a boost to women, but the results were gender-neutral,” McDowell says. “It helped everybody.”

Students were paired in some introductory programming classes and not paired in others. Students who were paired were supposed to complete homework assignments as a pair, but do exams solo and students in the other class did everything alone. The findings were pretty drastic:

A year later, 59 percent of the paired female students were still computer science majors, compared with 22 percent of the women who programmed solo. The results were striking for men as well: the department retained 74 percent of male pair programmers, versus 47 percent of lone coders.

Again, these studies show that a gender differences model is not acceptable for explaining women's low enrollment in computer science - the first article shows that it is clearly a cultural thing. The second article shows that the social aspect of computer science appeals to both men and women despite researchers belief that it would only boost women's retention.

Webcomics

If there's one thing that almost every one of my friends have in common it's that they read some form of webcomic on the Internet. At some point the funnies were translated into online format and have since exploded. I still remember back when Penny-Arcade was the only real webcomic online but they have now grown quite numerous and cross many genres. It used to be another way to gleam humor out of video games but now you can find web comics on everything from tech news to history.

It was by chance that a friend linked me to a paper that explored webcomics from a different perspective than any I had previously been acquainted. Specifically, Tatalovic in his article "Science comics as tools for science education and communication: a brief, exploratory study" examines the multitude of ways that this genre of webcomic that has exploded in the past 10 years has increased the way we discuss and perceive science. It wasn't until this was made salient that I became aware of how often I learn some new scientific concept through the reading of a comic. Moreover, I get a sense of accomplishment from "getting" some nerdy joke in some comic that I can then explain to friends who may not be as familiar with some of the concepts.

So where does computer science fit in? Tatalovic spends the majority of the paper addressing specifically educational comics and how they spur communication but the humor comics that evoke conversation through science references still receive mention. Unfortunately a search for computer science comics comes up relatively dry. I'm not aware of any real paper computer science comics and the webcomic world is significantly dry minus XKCD. While XKCD hits computer science topics fairly regularly and is notably popular around the CMU computer science department, someone unfamiliar with computer science may not immediately recognize XKCD as having anything to do with computer science, especially considering the author describes his background as physics, not CS.

This leaves me with some questions I'm going to take into further exploration of the undergrad culture. I'm continually interested in the role models that people take and who they identify as computer scientists. Given the recent popularity of webcomics, I wonder if XKCD has transformed or introduced any budding computer scientists to the field and what role these comics have as a medium. I remember a long time ago discussing with some people getting a Numb3rs like tv show on the air to introduce computer science and do for CS what CSI has done for law enforcement. Maybe we don't need a tv show though, maybe we just need a really sweet computer science comic?

In exploring different countries and how they perceive computer science, I've read a lot of different papers about the state of computer science. However, it didn't strike me until today while reading a paper about Australia that I realized how similar people around the world see this problem of participation. Eileen Trauth, Susan Nielsen, and Liisa von Hellens's paper "Explaining the IT Gender Gap: Australian Stories for The New Millennium," many of their insights into the world of IT professionals come to near identical to the attitudes we adopt.

In the beginning they identify the issue of low participation in computer science compounded with a substantial gender imbalance. However, they run into similar issues as we do with problems of gender difference. Trauth et. al. describe the collection of gender differences as flawed, in part as their case study identifies several women who upset the principles assumed by gender difference. Many of the approaches centering around gender differences ignore the countless women already participating in computer science and instead of focusing on improving this value they spend more time alienating the women already participating in computer science.

In interviewing women who are currently participating in the IT field they found a number of different factors that don't fall into the gender differences spectrum. For example, they found that imbalanced environments create hostile cultures for minorities. Crude jokes demeaning women and hostility are more likely to be present in male-dominated populations. Additionally, not seeing women regularly in the workplace creates false perceptions that women are somehow not qualified to participate in IT careers. Further, many of the women identified prominent female role models in their academic careers that helped to create or continue their interest in realizing that IT was an acceptable path for them.

So while we may be looking globally to operate on the US, these studies have implications globally outward as well. Many countries around the world suffer the same gender imbalances and are buried under the same gender differences suggestions that plague the United States.

Retention and Attrition Rates in CS

This past week I have been looking into retention and attrition rates of both men and women in CS programs. One of the first papers that I looked at was the paper about the University of Malaysia - Women in Computer Science: NO SHORTAGE HERE! by Mazliza Othman and Rodziah Latih. Here are the tables which show intake and withdrawal / failure rates based on gender.


The university has fairly equal intake rates and in almost all years (with the exception of 1999/2000) examined in this study, more men failed / withdrew from the program each year.

While attrition rates may be fairly low overall, and in particular for women at the University of Malaysia, that is not the case at many U.S. universities.

In the paper Scavenger Hunt: Computer Science Retention Through Orientation by Talton et al. from the University of Illinois:

"Like many large research universities, the University of Illinois has struggled with the high attrition rate of first-year students in computing disciplines. In the five year period prior to 2003, roughly 25% of the total number of entering freshmen have dropped out of the program by the end of their first year (see Table 1). In particular, the attrition rate of women and minority students is quite high, averaging about 35% for the same period."

Similar situations exist even at CMU.

In a paper on W@SCS, (Women in Computer Science: The Carnegie Mellon Experience by Lenore Blum), Blum looks at retention rates of men and women in CS.

In a paper (not the book), Unlocking the Clubhouse: The Carnegie Mellon Experience, Margolis and Fisher examine the same data at CMU and show promising trends of retention of women in CS.


These graphs so something that should not be surprising, that higher percentages of women in the major seems to lead to higher retention. In Blum's paper Peter Lee, Associate Dean for Computer Science Undergraduate Education at the time is quoted as saying, "[Retention] seems unnecessarily negative to me, and at any rate seems to aim too low. The goal, it seems to me, is to take advantage of the great recruiting success and produce a crop of graduating women who will be the future leaders, world-class scientists, visionaries, and captains of industry. . . . "

Other factors which seem to raise retention rates of women in CS are discussed in IMPROVING THE PERSISTENCE OF FIRST-YEAR UNDERGRADUATE WOMEN IN COMPUTER SCIENCE by Rita Manco Powell. Powell did a study at UPenn about retention and found that there was a 50% persistence rate of women before changes were implemented compared with an 85% persistence rate after changes were implemented which is equal to the male persistence rate. Some changes include more faculty-student interaction, social gatherings for students, mentoring groups, and the organization of WICS (Women in Computer Science). These are very similar to the findings at CMU with the organization of W@SCS.

Powell also examined why women were leaving CS - why the attrition rates were so high:

"This research study found that many of the study participants began the computer science major with an inadequate background from high school in the subject, causing them to struggle to perform as well as their peers with more computer science experience. Because of this fact, which was further heightened by the women’s perception that the male students knew more than they, several women lost confidence in their ability to be successful in the major and subsequently lost interest in the major. Social isolation accompanied their gender minority status within their peer group further weakening their resolve to persist."

I am interested in examining whether women at CMU have similar reasons for leaving CS, and also examining what men's reasons are. I am currently working on coming up with interview questions for men and women who have switched out of SCS at CMU. I am also interested in looking at attrition and retention rates, because I know of 4 women in my year who have left CS (and a few in other years), but I don't believe that I know of any men who started in CS and are now in a different major.