What can the Uber Gender Pay Gap Study tell us about improving diversity in computing?
March 5, 2018 at 7:00 am 6 comments
The gig economy offers the ultimate flexibility to set your own hours. That’s why economists thought it would help eliminate the gender pay gap. A new study, using data from over a million Uber drivers, finds the story isn’t so simple.
Source: What Can Uber Teach Us About the Gender Pay Gap? – Freakonomics
A fascinating Freakonomics podcast tells us about why women are paid less than men (by about 7%) on Uber. They ruled out discrimination, after looking at a variety of sources. They found that they could explain all of that 7% from three factors.
They found that even in a labor market where discrimination can be ruled out, women still earn 7 percent less than men — in this case, roughly 20 dollars an hour versus 21. The difference is due to three factors: time and location of driving; driver experience; and average speed.
The first factor is that women choose to be Uber drivers in different places and at different times than men. Men are far more often to be drivers at 3 am on Saturday morning. The second factor is particularly interesting to me. Men tend to stick around on Uber longer than women, so they learn how to work the system. The third factor is that men drive faster, so they get more rides per hour.
When someone from Uber was asked about how they might respond to these results, he focused on the second factor.
But for example, you could imagine that if we make our software easier to use and we can steepen up the learning curve, then if people learn more quickly on the system, then that portion of the gap could be resolved via some kind of intervention. But that’s just an example. And we’re not there yet with our depth of understanding, to just simply write off the gender gap as a preference.
Improving learning might help shrink the gender pay gap. Obviously, I’m connecting this to computing education here. What role could computing education play in reducing gaps between males and females in computing? We have reason to believe that our inability to teach programming well led to the gender gap in computing. Could we make things better if we could teach computing well?
Here are two thoughts exploring that question.
- We know (e.g., from Unlocking the Clubhouse) that men tend to sink more time into programming, which can give them a lead in undergraduate education (what Jane Margolis has called ‘preparatory privilege‘). What if we could teach programming more efficiently? Could we close that gap? If we had a science of teaching programming, we could improve efficiency so that a few hours of focused effort in the classroom might lead to more effective learning of tens of hours of figuring out how to compile under Debian Linux.
- When I first started thinking about the “phonics of computing education” and our ebooks, I was inspired by work from Caroline Simard that suggested that helping female mid-level managers keep up their technical skills could help them to progress in the tech industry. Female mid-level managers have less time to invest in technical learning, and at the mid-level, technical education still matters. If you have a project that needs a new toolset, you’ll more likely give it to the manager who knows that toolset. If we could teach female mid-level technical managers more effectively and efficiently, could they make it into the C-suite of tech companies?
Maybe better computing education could be an important part of improving diversity, along multiple paths.
Entry filed under: Uncategorized. Tags: BPC, computing for all, computing for everyone, NCWIT.
Computing Ed Research - Guzdial's Take 
1.
Bonnie | March 5, 2018 at 7:42 am
I would prefer that we focus on the reasons women spend less time on their programming assignments. I believe computer science is an inherently project based discipline, and like other project based disciplines (writing, studio art) require a lot of practice. The idea that somehow we are going to be more “efficient” is an illusion. And even if it were possible, without addressing the reasons why women spend less time, men will still be advantaged because they will still be spending more time on task.
I have never heard reports that women spend less time practicing their writing in composition courses – and in fact, the stereotype is that men spend less time on those courses. Perhaps we should be investigating a)if that is really true, and b) if so, why? What is it about programming that leads women to spend less time compared to writing?
Lastly, my experience in industry leads me to believe that the kind of “keeping up the technical toolset” needed to move mid level managers to the C-suite is different from the “keeping up the technical toolset” needed for hands on developers. Mid level managers need to know why a technical skill or product is important (or not) and have just enough hands on experience to understand some key examples. Managers at that level need to understand more at the architectural level, and also to understand process and the company’s core business. So yes, mid level managers probably can “keep up” more efficiently. I think having that skill is what distinguishes employees destined for upper management from employees who stay on the technical track.
2.
Mark Guzdial | March 5, 2018 at 10:14 am
Bonnie, I’m pretty sure that women do spend more time on their programming assignments in Media Computation. I believe that that’s why they end up with higher grades on tests of CS knowledge than men. I suspect that the context (the framing, in Kahneman and Tversky terms) matters more to female students.
I believe that we can be more efficient. I understand the skepticism. It’s a similar kind of skepticism that leads reading instructors to prefer whole language learning over phonics, despite the studies showing the benefits of phonics. I think our work with subgoal labeling and with Parsons problems is demonstrating that we can teach programming more efficiently than just having people program all the time.
3.
gasstationwithoutpumps | March 5, 2018 at 11:57 am
While we can probably make programming education more efficient (it is currently bad enough that improvement is easy to believe in), I don’t see how that would help close the gender gap if men continue to put more time in than women. Your example of media computation argues not for more efficient education, but for more inspiring education—particularly, education more inspiring to women, so that gap in effort is removed.
On the phonics vs. whole-language question: different people learn differently, and relying entirely on either approach is a bad idea (particularly in English, where phonics is useful, but orthography irregular enough that many words just need to be learned whole). Other languages change the balance: in Spanish, with highly regular orthography, phonics clearly wins, but in Chinese and Japanese, with non-phonetic writing systems, whole-language clearly wins.
4.
Mark Guzdial | March 5, 2018 at 12:08 pm
As I said to Greg Wilson in this thread — reduce not close. Inefficiency contributes. Yes, Media Comp is about something other than efficiency, but was an example of how to get more time on task.
People have different learning preferences, for sure. I was only speaking of English and CS when I offer that phonics plays a role in learning. Certainly, using a variety of approaches and methods is the best recommendation from the literature on learning to read and learning to program.
5.
Greg Wilson | March 5, 2018 at 7:54 am
When I read, “Men are far more often to be drivers at 3 am on Saturday morning,” I interpret it as, “Men are less likely to have family care responsibilities and feel safer being out in the middle of the night.” I don’t think education will close that gap – I think it will require larger structural changes.
6.
Mark Guzdial | March 5, 2018 at 10:08 am
We can’t close the gap with education. We can make it smaller.