Gender in Tech at Harvard

We surveyed 900+ undergrads on their experiences with CS at Harvard. Here are our results.

To better understand the gender in technology gap, we constructed a survey in 2015, which aimed to assess students' experiences within the CS department at Harvard. Over Fall 2015 and Spring 2016, we publicized this survey over various channels, aiming to reach all current Harvard college undergraduates.

The survey targeted three audiences: those who had never taken a CS class, those who had taken at least one class but were not intending on concentrating, and those who considered themselves CS concentrators. By targeting these three audiences, we sought to gain a holistic perspective of how CS is viewed on campus.

The full survey is available here. The questions asked were both quantitative and qualitative in nature, and inquired about students' experiences with CS coursework, internships, community, faculty and staff. Some questions specifically asked participants for their thoughts on the gender gap in tech. When conducting the survey, we were sure to not indicate that we were studying the impact of gender on students' experiences with CS, and any gender-related questions were left for after all other responses had been collected.

We view this survey as a first step in understanding the role that gender plays with technology at Harvard. While these results point to interesting conclusions, they also lead to various questions that would be interesting to investigate. Some insights from the quantitative portion of the survey are presented below, but more analysis is possible, for instance by class year or race. If you would like to discuss additional analysis based on this data, contact us.

Overview of Respondents

Breakdown of Respondents

We surveyed three different groups of students based on their prior level of CS background. The three groups surveyed were: 1. never taken a CS class, 2. taken at least one CS class but did not concentrate in CS, and 3. taken at least one CS class and intend to concentrate in CS. Note that each category had large enough sample sizes to analyze further in this report.

Summary of Respondents by Gender and Prior CS Background

Gender Gap in Numbers

Concentration Data

Below we present the total number of concentrators based on gender over the past few years, by graduating class. This data was not based on our survey, but rather is based on registrar data across these years. We provide this information to provide more longitudinal context. Raw data for the total number of CS concentrators in 2015 and 2016 is currently not presented.

Number of CS concentrators by gender

CS Concentrators that are Female or Minority

Secondary Data

Below we present the total number of secondaries based on gender over the past few years. Again, this data was not based on our survey, but rather is based on registrar data across these years. We provide this information to provide more longitudinal context. We note the large increase in secondaries over the past couple years.

Number of CS secondaries by gender

Proportion of CS Secondaries that are Female

Class Data

Each CS concentrator respondent on the survey was asked for the list of classes they had taken. We used this data to approximate the percent of concentrators in each class who were female. Note that this is distinct from the percent of all students in the class who are female; that data would be better obtained from the registrar. All the categories depicted below have at least 10 respondents, and we also provide a chart depicting the total number of respondents in each category.

This demonstrates some interesting trends. For instance, the percent of women in CS50 is much higher than that in most other CS classes. We also see trends in the types of CS classes that students enroll in. For instance, whereas grad level CS classes (2xx) have only 16% female enrollment, CS 171 and CS 179 together have 47% female enrollment.

Percentage of CS concentrators that are female in each class

Total CS concentrator respondents per class

Perceived Gender Gap

Most respondents viewed the gender gap as a problem, and we interestingly do not see much difference between the genders in their perception of this problem.

Perception of the Gender Gap in Tech

Where does the Gender Gap Appear?

Experience levels

Here we visualize how male and female Computer Science concentrators differ in programming experience. A vast majority of women, 76 percent, have four or fewer years of programming experience compared to 53 percent of men. A majority of women, 67 percent, said they had one or fewer years of programming experience before arriving at Harvard compared to only 41 percent of men. A near majority of men, 45 percent, said they had three or more years of programming experience before arriving at Harvard compared to only 21 percent of women.

Programming Experience by Gender

Years of programming experience

Motivation to study CS

I like CS because I am interested in CS problems

I like CS because I want to make a difference

Roughly equal proportions of men and women are interested in CS because it gives them a tool to make a difference. 33 percent of women are neutral or disagree that their interest in CS is related to the problems that they solve, compared to 10 percent of men. 67 percent of women agree or strongly agree that their interest in CS is related to the problems that they solve, compares to 90 percent of men.

Perception of abilities

How Good are You at Programming?

How Good are You at Theory?

Female CS concentrators are much more likely to report that they are average or worse at programming compared to peers in their classes, while male CS concentrators are much more likely to report that they are average or better at programming compared to peers.

There were very similar responses for programming and theory, which is surprising since it is likely that most respondents entered college with little to no computer science theory experience. This might suggest a lack of confidence beyond true abilities.

Average Confidence in Programming Ability by Programming Years of Experience

Here we present the average confidence in programming ability as it varies with number of years of programming experience. We might hypothesize that women's lack of confidence in programming ability solely stems from a lack of experience. However, as demonstrated in this graph, other factors lead to women's lower confidence in their programming abilities. For instance, women with up to 8 years of experience programming only report being as confident as men with 0-1 years of programming experience.

Office hours

How often do you go to office hours?

How often do you think the average student goes to office hours?

Almost half of female CS concentrators, 46 percent, reported that they attend office hours roughly every week, compared to 15 percent of male CS concentrators. Men and women reported similar predictions on how often the average person goes to OH.

Teaching staff accessibility

How Accessible are CS Professors?

How Accessible are CS TFs?

Women CS concentrators were more likely to rate professor accessibility as being average or below average, while male CS concentrators were more likely to rate professor accessibility as above average. Men and women rated TF accessibility similarly. Both genders found TFs more accessible than professors

Involvement with CS community

Interestingly, female respondents vary highly in the sense of community they feel within the CS concentration, with nearly equal numbers feeling at least somewhat detached as somewhat involved. On the other hand, most male respondents reported that they felt an average level of involvement with the community. We found that sense of community was correlated (0.3 correlation coefficient) with ability to find problem set partners; we believe that a sense of community is crucial for both succeeding in the classroom and socializing within the concentration.

How involved do you feel with the CS community at Harvard?

Finding problem set Partners

We see that most students find it easy to find partners in CS classes, with women slightly less likely to report that it is average, somewhat easy, or easy to find partners. We also note that women are 23% likely to report that it is somewhat difficult or difficult to find partners, and that men are 11.4% likely to fall in this category.

How easy is it to find partners?

Finding a job or internship

We can see that most CS concentrators on campus find the job and internship search stressful or very stressful. Nearly half of the women (47.5%) reported this process as very stressful, whereas ony 27% of men chose this category. Over 80% of women and 60% of men reported the process as at least somewhat stressful.

How stressful is recruiting?