STEM Confidence Gap

Recent studies have shown that engaging students in active learning can drive improved student outcomes, particularly for women and minorities. These findings reinforce the message we’ve been hearing anecdotally from numerous professors since we launched Piazza five years ago.

Piazza recently undertook an extensive study of all U.S and Canadian actions occurring on the platform in 2013 and 2014, which was recently published in the Chronicle of Higher Education. Our hope is that shedding a light on classroom behavior patterns will help create a fruitful conversation among educators about how best to drive improved learning outcomes — especially among those students who need it most. We were fascinated by the results and wanted to share them with you:

Female students are highly engaged and actually ask more questions on average than their male peers enrolled in both computer science and other STEM coursework.

  • Women ask 26 percent more questions than men in computer science classes and 23 percent more questions in STEM classes. More than half of questions asked by female computer science (52 percent) and STEM (60 percent) students are asked anonymously, versus 41 percent and 49 percent for computer science and STEM, respectively, for males.

janimage_1However, female students answer fewer questions compared to their male peers.

  • Female computer science students answer 37 percent fewer questions on average compared to their male peers.
  • Female STEM students answer 18 percent fewer questions on average than their male peers.

Female students are more likely to answer questions anonymously than their male peers.

  • Female computer science students answer questions anonymously 35 percent of the time, whereas their male colleagues answer anonymously 22 percent of the time (a 13 percentage point difference).
  • Female STEM students answer anonymously 39 percent of the time, whereas male STEM students answer anonymously 28 percent of the time (an 11 percentage point difference).


As students progress from lower-level to higher-level computer science coursework, both men and women become slightly less confident, and the confidence gap among women persists.

  • From lower-level to upper-level coursework in computer science, men go from answering 22 percent to 24 percent of questions anonymously, while women go from answering 36 percent to 38 percent of questions anonymously.
  • While answers per student decline for both men and women in upper-level computer science classes, women answer 43 percent fewer questions than men in lower-level classes, and 27 percent fewer questions in upper-level classes.
  • The gap increases again in graduate-level classes, where women answer 42 percent fewer questions than men.
  • The confidence gap also holds true in STEM classes, where women answer 14 percent fewer questions in lower-level classes, 25 percent fewer questions in upper-level classes, and 23 percent fewer questions in graduate classes.


This confidence gap exists, and in some case is more prevalent, at schools typically considered “top” STEM programs.

  • At Berkeley, Stanford and Harvard, the gap between men and women answering questions was greater than the average for all schools in CS and STEM.
  • At MIT, women answered 28% more questions than men in STEM, but answered 35% fewer questions in CS.
  • In computer science, Harvard had the smallest percentage point difference between women and men answering questions anonymously (1 percent), while Stanford had the largest percentage point difference (29 percent).
  • In STEM classes, Harvard and Berkeley had the smallest percentage point difference between women and men answering questions anonymously (7 percent), while Stanford had the largest percentage point difference (32 percent).


Harvey Mudd is one of a few schools that has shown significant progress in narrowing the confidence gap.

This small private school has made significant efforts to not only increase its percentage of women entering computer science classes, but also to adjust curriculum and address systemic challenges throughout the college experience.

  • There was only a 12 percent differential between answers per female and answers per male (compared to 37 percent average overall).
  • Women answered 13 percent of questions anonymously, as opposed to the average of 35 percent (and men answered 9 percent of questions anonymously, versus average of 22 percent).
  • There was a 4 percent gap between male and female rates of anonymous answers, compared to the average gap of 13 percent.


The confidence gap is much greater in STEM than in non-STEM disciplines.

  • While men answer 37 percent more questions than women in STEM classes, this difference is only 7 percent in humanities.
  • In business and social sciences, women actually answer more questions than men (11 percent and 5 percent, respectively).
  • There is a 13 percentage point difference in male and female anonymous answers in STEM classes; this compares to a zero, 6 and 1 percentage point difference, respectively, in humanities, business and social sciences.


Due to heightened awareness of the need for more women to pursue STEM careers, a number of worthy organizations have emerged to encourage girls to engage in coding and technical activities during the K-12 years. However, as this research shows, the college years are also critical years for supporting women and helping them persist in computer science and STEM coursework. Harvey Mudd may provide a blueprint for what works. Piazza’s research demonstrates not only that the confidence gap among female STEM students is real, but also that concrete efforts to address it can make a substantial and measurable impact in bridging the gender divide, particularly in computer science.

If you’d like to learn more about results for your school or subject, or have questions or comments, please email me at


Piazza’s primary research focused on 976,000 college and graduate STEM student enrollments at traditional college and graduate programs within the US and Canada (1.126 million total student enrollments, including humanities and social sciences). Research tracked students over nearly four terms: Spring 2012, Fall 2013, Spring 2013, and Fall 2014 through November 10th. Data collected involved more than 2.3 million questions asked and answers given, 2.1 million of which involved STEM and computer science coursework, and which collectively were viewed 220 million times. All data was generated on a per capita basis; for example, comparing male answers per male student to female answers per female student, and so on. This is to account for differences in overall proportions of men and women within a given subject.

About Piazza

Piazza is a collaborative social learning and college recruiting platform empowering students to work together in real-time with classmates and instructors to find the answers they need at the time they need them. More than 700,000 higher education students use Piazza per year with an average engagement of 3 hours per day and nearly 200 post views per student per semester. Students and alumni can also use the Piazza Careers platform to connect with companies, apply for jobs and internships, and find out more about the skills needed to excel in their chosen careers. Piazza is FERPA-compliant and used at more than 1,000 colleges and universities.

Click to See Full Infographic