In June, Stuart Reges, principal lecturer in Computer Science and Engineering at the University of Washington, published a blog post Why Women Don’t Code that led to several articles and blog posts in response (e.g., Seattle Times and GeekWire). Reges argues that women are simply never going to enter computing at significant numbers, and 20% is about all that we’re ever going to get.
Our community must face the difficult truth that we aren’t likely to make further progress in attracting women to computer science. Women can code, but often they don’t want to. We will never reach gender parity. You can shame and fire all of the Damores you find, but that won’t change the underlying reality.
It’s time for everyone to be honest, and my honest view is that having 20 percent women in tech is probably the best we are likely to achieve. Accepting that idea doesn’t mean that women should feel unwelcome. Recognizing that women will be in the minority makes me even more appreciative of the women who choose to join us.
Hank Levy, Director of the U-W CSE School, wrote a great statement in response (see here). Levy disagrees with Reges’s conclusions, but supports Reges’s right to make his argument. Levy puts the current gender ratio in computer science in context by comparing to other disciplines.
I was most struck by the 20% claim. That’s easily proven wrong. There are many CS educational programs in the US with more than 20% female (like Computational Media at Georgia Tech). There are countries where CS is more than 50% female. How can Reges claim that 20% is the best that we can possibly do?
Here’s something important about Stuart Reges that people outside of CS education might not know: he’s a rockstar. He packs the house when he speaks at education conferences. He publishes regularly in the field. He has written a popular book on how to teach Java in introductory computer science (see Building Java Programs). Students love him, and teachers want to be like him. When Stuart Reges speaks, CS educators listen.
In this post, I want to step back and consider how Reges is making his argument, because it says something about how we make decisions in computing education. I am going to characterize the argument style in computing education as argument from authority which Wikipedia describes as "a claimed authority's support is used as evidence for an argument's conclusion." We need to recognize the form before we can move beyond it.
Appeal to Authority: Teaching Java like it’s Pascal
Stuart Reges is most famous in CS education for his radical approach to teaching introductory computer science. He’s decided that it doesn’t work to teach object-oriented programming in the introductory course. Instead, we should teach Java with all programming in the main function — as if it were Pascal. Quoting from his 2006 SIGCSE Paper Back to Basics in CS1 and CS2
Our new version of CS1 looks a lot like a 1980’s course taught in Pascal. We have gone back to procedural style programming. I was motivated to do this after attempting and failing to teach a broad range of introductory students at the University of Arizona using an "objects early" approach. I found that my best students did just fine in the new approach, but the broad range of mid-level students struggled with the object concept.
Note the reason for his approach — he personally attempted and failed to teach objects. Most of Reges’s argument in support his approach is based on quotes from Donald Knuth. He has data in his paper about enrollment rising at U-W in his class, but this was a time when everyone’s enrollment was rising in CS. It’s not really evidence about his approach — it’s correlation, not causation. There’s plenty of evidence that we can teach Java and object-oriented programming effectively in CS1.
In 2005, the SIGCSE Conference staged a debate, Resolved: Objects Early has Failed. In support of the proposition was Reges and Elliot Koffman. Koffman was selected because he wrote the original course definitions for CS1 and CS2 in 1984. Against the proposition was Kim Bruce and Michael Kölling, both of whom regularly teach object-oriented programming in their first course, have written textbooks about their approach, and have published evaluations about their approach. The debate had authority in support of the proposition, and evidence in opposition. Authority has sway in CS education.
A recent publication, a panel at SIGCSE 2016, considers why some students succeed and some fail at CS. Reges writes:
My 27 years of teaching programming to novices has left me with some deep intuitions about how people learn to program. I share Don Knuth’s belief that there is a mode of thinking that is particular to computer science (CS) and that some students have a greater aptitude than others.
Moving Beyond an Appeal to Authority
There’s a pattern here. It’s not unique to Reges. He makes his arguments the way that most arguments are made in computer science education, as an appeal to authority. Reges bases his position on his own many years of experience and references to one of the founders of our field, Don Knuth.
Arguments from authority led to a belief that the number of chromosomes in humans was 24 (see story here), that life emerged from spontaneous generation, and that Earth was the center of the universe. Science is how we learned to respond to appeals to authority. We use evidence rather than our gut or the experience of people who have been around for awhile. NCWIT just sent out a post to its Academic Alliance about the empirical data that contradicts Reges’s claims. The concluding paragraph captures the idea well:
In the end, these recent public discussions reinforce the old adage: just because you always hear it, doesn’t mean it’s true. For a field that claims to be "data-driven," it’s time to stop the so-called "debate." But, as long as these "debates" continue, our commitment to using evidence-based research to educate and to correct these misunderstandings and fallacies must also continue.
Computer science faculty usually make educational decisions without using evidence and research. Faculty in other fields tend to use research more. The earliest paper that I know that actually measured this was Clayton Lewis’s 2007 work that showed that CS faculty have beliefs about students that are in contradiction to measurable facts about them. The best work that I know on this question is Lecia Barker’s interview study with 66 CS faculty across the US, showing that research had little impact on their teaching practices. In contrast, a 2008 survey of 722 US Physics faculty found that 87% were familiar with one or more research-based instructional method, and 48% use at least one regularly.
Stuart Reges is making the argument that we can’t have more than 20% women in CS based on his authority. That’s an argument that will likely sway a lot of CS educators. We have to raise the standards of our arguments in CS education. We can use research and evidence to do much better than just an argument based on authority.