Male And Female Anatomy Drawing

Female programmers are less confident than male programmers

A new study of computer science students suggests that a confidence gap exists between men and women in software development

Phil Johnson By

Writer/Editor at ITworld, ITworld |

Last year, journalists Katty Kay and Claire Shipman wrote a book titled The Confidence Code: The Science and Art of Self-Assurance—What Women Should Know, in which they posit (and back up with lots of research) that, despite having the same abilities as men, many women fail to achieve as highly in the corporate world due partly to a lack of confidence in their own abilities. Is it possible that a lack of confidence is also one of the reasons for the well-known gender gap in STEM fields , which are just 24% female, according the White House, including software development, which is made up of just 20% women, according to the U.S. Bureau of Labor Statistics? A new study looking at how male and female college students view their abilities suggests that such a gender confidence gap in STEM is real and is particularly pronounced among those studying computer science.

The study was done by Piazza, a social learning platform used by college teachers and students. Instructors set up areas on the site for their classes, where students can ask questions (anonymously, if they choose) that their fellow classmates can collectively answer (again, anonymously, if they prefer). Piazza recently looked at over 2 million questions asked and answered during four recent semesters by almost 1 million undergraduate and graduate students at schools in the U.S. and Canada in a wide range of fields.

They defined the gender confidence gap as the percentage difference between the average number of questions answered by male and female students. Overall, they found a noticeably greater confidence gap between the genders in computer science than in other STEM fields. Among CS students, women answered 37% fewer questions than men, as compared to 18% fewer in non-CS STEM fields. Outside of STEM, the confidence gap is much lower and sometimes reversed, where women answered just 7% fewer questions than men in humanities while actually answering more questions than the men in business and social sciences, 11% more and 5% more, respectively.

The confidence gap among CS students was persistent throughout all levels of study, starting at 43% for students in lower level classes, then dropping to 27% of students in upper level classes before rising back up to 42% in graduate level classes. Piazza also found that female CS students who did answer questions were more likely to do so anonymously, 35% of females, vs. 23% of male CS students.

I reached out to Piazza for additional data on the gap at prominent computer science departments, which they kindly shared. The gender confidence gap was even more pronounced at some of the county's top CS programs, such as -62% at Carnegie Mellon University, -60% at Cornell, -51% at the University of California, Berkeley, -49% at Stanford, and -39% at Harvard.

Some CS departments, though had a smaller gender confidence gap than average, or even saw it reversed, with women answering more questions than their male counterparts. For example, the gap was just -18% at the University of California, San Diego, -12% at Harvey Mudd, and -1% at Duke, while flipping the other way to +3% at Boston University and a whopping +22% at Caltech.

Is the confidence gap between male and female students in computer science a real thing? It's hard to say for sure, but these numbers certainly suggest there's something to it, which might help to partially explain why fewer women choose careers in tech. The infographic below from Piazza has more details, so take a look and draw your own conclusions.

Piazza

Editor's note: An earlier version of this article said that male computer science students answered questions anonymously 25% of the time; the correct amount is 23%.

This story, "Female programmers are less confident than male programmers" was originally published by ITworld .