Why Do Women Leave Philosophy? Surveying Students at the ...

Philosophers’

Imprint

volume 16, no. 6 march 2016

WHY DO WOMEN LEAVE PHILOSOPHY? SURVEYING STUDENTS AT THE INTRODUCTORY LEVEL Morgan Thompson∗ , Toni Adleberg† , Sam Sims‡ , Eddy Nahmias§ University of Pittsburgh∗ , University of California, San Diego† , Florida State University‡ , Georgia State University§ © 2016, Thompson, Adleberg, Sims, and Nahmias This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 3.0 License

1.

When do women leave philosophy?

In 2012 in the United States, for every 100 men graduating with a college degree, 141 women graduated.1 For decades now, more women have been enrolled in American universities than men. Yet, during these same decades, the proportion of women who major in philosophy has remained stagnant, hovering below one-third. So, while almost 60% of college graduates are now women, only 30% of philosophy majors are women (Department of Education, 2013; Paxton et al., 2012). In the humanities, religion and theology (35.6%) is the closest major to philosophy when it comes to the underrepresentation of female majors. Among all majors, the only ones with similarly low ratios are economics (31%), physics (19.7%), computer science (22%), and engineering (20%).2 With women getting just 30% of philosophy bachelor’s degrees, it’s no surprise that the ratio of women to men is so low among philosophy graduate students (30%) and professors (20.7%) (Paxton et al., 2012; Norlock, 2012).3 The underrepresentation and treatment of female graduate students and professors in philosophy has, for good reason, received increasing attention in recent years. But there has been limited discussion, and very few empirical investigations, of why so many women say goodbye to philosophy just after being introduced to it. In this article, we offer our initial attempts to gather data to test various hypotheses aimed at answering this question and to suggest 1. Project was conceived by TA and MT; survey was written and data collected by MT, TA, and EN; statistical analyses were conducted and described by SS; paper was written by MT and EN. 2. Data compiled by Integrated Postsecondary Education Data System and accessed via WebCASPAR (https://webcaspar.nsf.gov/) using the 2013 NCEJ data set. 3. There is no systematic data on the gender proportion of graduate school applicants or of first-year graduate students in philosophy. The number of faculty includes both full-time and part-time instructors. Women make up only 17% of full-time faculty in philosophy departments (Norlock, 2012). The proportion of female graduate students in philosophy is lower than all fields except for music theory and composition, physics, and engineering and roughly equal to mathematics and religious studies (Healy, 2011).

thompson, adleberg, sims, and nahmias

Why Do Women Leave Philosophy?

some ways to increase the number of women who continue in philosophy after their introductory courses, ideally while also improving the philosophy classroom for all students.4 Paxton, Figdor, and Tiberius (2012) investigated the underrepresentation of women across various levels in the profession. In their sample of over 50 universities, they examined potential drop-off points in women’s representation in philosophy: between introductory courses and majoring, between majoring and going to graduate school, and between attending graduate school and becoming a faculty member. They found that the only statistically significant drop-off in women’s representation was between the introductory courses (43%) and majoring in philosophy (35%). This finding suggests that women disproportionally drop out of philosophy after taking one or two courses. At Georgia State University (GSU), where we did our research, we found a similar trend. Although women take Introduction to Philosophy (55%) at a rate proportional to women’s undergraduate enrollment at GSU (60%), women make up a significantly smaller proportion of philosophy majors (38–41% from 2006–2011).5

Men and women likely enter college and their introductory philosophy classes with different opinions about, and levels of interest in, the philosophy major.6 However, because philosophy is rarely taught in high school in the U.S., most students’ first exposure to the subject is in their introductory college course, and most students likely choose to major in philosophy — or not to — only after taking such a course. Hence, introductory philosophy courses are likely to be the most significant point in turning students toward, or away from, philosophy. For these reasons, we primarily examine hypotheses focused on college-level courses, though some of them may be useful in influencing perceptions of philosophy at the pre-college level. We focus on what happens in the Introduction to Philosophy classroom because we believe that this is one of the most useful places to intervene on various factors that may increase the number of women who want to continue in philosophy. We think intro-level instructors are in a unique position to set the tone for students regarding what philosophy is, how it is done, and what it is good for. We also believe that finding ways to make introductory courses more interesting and appealing to women

4. The initial focus of our research was on gender disparities in philosophy, but we also used our study to examine hypotheses for why Blacks are less likely than Whites to major in philosophy, presumably one of the contributing factors to the dismally low proportion of Black graduate students and faculty in philosophy (see Botts et al. 2014). Our initial data suggest that there are some overlapping problems and solutions to the gender and race disparities, but we also think there are important differences. Finally, there may be some problems and solutions uniquely impacting Black women. Issues regarding race are discussed in a separate paper, see Thompson, Bright, and Kummerfeld (ms). For a discussion of one empirical test for intersectional hypotheses, see Bright, Malinksy, and Thompson (2015). 5. Interestingly, the proportion of women philosophy majors at GSU has trended upward since 2011 and is currently at 52%, explained by a roughly proportional increase in the number of women majoring and decrease in the number of men majoring. This period corresponds to the period when we developed and carried out the research discussed below, including more discussion among the instructors of Introduction to Philosophy about attempts to increase the diversity of the major and the proportion of women on the syllabus and showing a presentation on the value of philosophy in every Intro class (see Section 4.4 below).

6. See Baron, Dougherty, and Miller (2015) for evidence that women in Australia enter introductory philosophy classrooms already less interested in majoring compared to men. In work examining why women are less interested in philosophy than men before college, Dobbs (2015) reports data from the Higher Education Research Institute (HERI) showing that from 2004–2009, out of over 2 million students (55.7% women) enrolling in U.S. colleges and universities, 4,838 men and 2,463 women declared philosophy as their intended major (totaling just 0.33% of the students surveyed). Hence, of the (very few) students entering college with the intention to major in philosophy, only 33.7% are women. Very few high schools teach philosophy, with the exception of some private schools and International Baccalaureate programs that offer a course called Theory of Knowledge. A few high schools offer Philosophy clubs, and there exist some excellent outreach programs (e.g., the Outreach Program at the University of North Carolina, Philosophy Learning and Teaching Organization (PLATO), High-Phi, Ethics Bowl). Aymelek (2015) and Chan et al. (ms.) provide qualitative evidence that women who continue in philosophy were often exposed to philosophy before beginning introductory coursework. We discuss in Section 5 possibilities for altering the image and gender schema of philosophy such that it might influence students’ impressions of the discipline even before they get to college.

philosophers’ imprint

-2-

vol. 16, no. 6 (march 2016)



will likely improve the courses for everyone and that it is the most effective way to increase the number of women in philosophy at all career stages.

sulting in a looping effect (Saul, 2013) — the speculative mechanism by which fewer women continue in philosophy due to the already low number of women in the field, which in turn perpetuates the low numbers of women in the field. When talented women leave philosophy (opting instead to study psychology, sociology, literature, education, etc.), the philosophy profession loses out on the high quality of work that would have been performed by those individuals had they continued in philosophy. Some feminist philosophers have argued that diversifying the demographics of philosophy will bring with it more diversity in philosophical content and methods. It is an empirical question whether women philosophers disproportionately prefer certain areas of philosophy, methodologies, or topics. But if these philosophers are correct, then philosophy may be missing out on a variety of fruitful methodologies, arguments, and research questions. Given the importance of retaining women in philosophy for pedagogical and professional reasons, we need a better understanding of why women leave philosophy when they do. To achieve this, we begin with the assumptions that philosophy should be enjoyable and useful for most students — though part of our project is to consider ways to make it more appealing — and that taking more philosophy courses, getting a minor, or obtaining a major (or second major) in philosophy would be better for some students than their alternatives. Obviously, the philosophy major or minor is important for the very few students, including women, who might go on to graduate school or careers in philosophy. But for many other students, philosophy may serve them better than other courses or majors for a wide variety of careers.7 Philosophy courses may also improve students’ quality of life

2.

Why it’s problematic that women leave philosophy

Before we consider what causes, and what might be done about, the underrepresentation of women in philosophy, we want to say a little about why it is problematic. First, women may be underperforming in philosophy or simply avoiding philosophy classes altogether because of stereotype threat, which occurs when implicit or explicit stereotypes about one’s selfidentified group (e.g., gender or race) influence one’s performance. Psychological research shows that women experience stereotype threat (see Section 5) in a number of contexts, especially when they are underrepresented, which in turn makes women perform worse than they otherwise would. To the extent that stereotype threat applies to women in philosophy (for arguments, see Schouten, 2015), women may be producing work of lesser quality than they would otherwise be able to produce. When women are preoccupied (either consciously or unconsciously) with worries about confirming stereotypes about women in a given area, they feel more stressed in those settings, are less motivated, and are less able to devote their full attention and effort to the tasks of responding in class discussions, writing papers, and preparing presentations. This consequence is unfair to women students. It is also evidence of a pedagogical problem, since instructors should aim to connect with and help all of their students. For students in introductory philosophy courses, stereotype threat may result from the low proportion of women authors and instructors in philosophy, because of a schema of philosophy as male (see below), or because of the association of philosophy with logic and mathematics, a field in which stereotype threat has been shown to exist (Steele, 1997; Maloney et al., 2013). Furthermore, stereotyped individuals tend to avoid situations in which stereotype threat could arise (McKinnon, 2014), potentially re-


7. Many recent articles and sources of data have highlighted the practical benefits of an undergraduate degree in philosophy, including improved skills for a variety of careers, higher test scores for graduate programs, and higher salaries and salary increases relative to other majors. See, for instance, this article showing the lifetime earnings gain for philosophy majors relative to other majors: http: //www.forbes.com/sites/jeffreydorfman/2014/11/20/surprise-

-3-

vol. 16, no. 6 (march 2016)



by providing methods for approaching questions of great importance to them and their society. Philosophy encourages investigation of questions about how we should live our lives, which may provide meaning and enrich the individual’s life. It also provides people with the tools to analyze important debates in the real world, such as economic inequality, abortion, climate change, conflicts between science and religion, or whether race is a biologically real category. Though some may disagree with us about the value of philosophy courses for students, we find it entirely plausible that, to the extent that philosophy is valuable, it should be just as valuable for women as for men. Finally, we find it plausible that it would be valuable to both students and teachers of philosophy to have a more diverse set of philosophy students in their courses at all levels (regardless of whether that diversity results in different philosophical methodologies or projects), and that it would benefit philosophy, and perhaps society in general, to have more women major in philosophy.

though we suspect that some of the factors are not related directly to sex discrimination, especially as they contribute to women’s leaving philosophy after their initial exposure to it. On this model, there will not be a single factor that explains why philosophy has such a badly disproportionate gender balance in comparison to fields like neuroscience, psychology, political science, English, and biology. Rather, a number of factors in common with other fields, along with some unique to philosophy, interact to constitute philosophy’s substantial underrepresentation problem. Indeed, the Perfect Storm Model can help make sense of the puzzling fact that roughly 50 or 60 years ago, every academic field had similarly disproportionate gender ratios, but philosophy is now decades behind most of these fields in overcoming the problem. In some cases relevant differences can be found. For example, English departments responded to the underrepresentation of women and minorities in their field in part by broadening the canon to include more women and non-European authors in a way that has not occurred in philosophy and has been resisted by some philosophers (see Dotson, 2012 on philosophy’s culture of justification vs. a culture of praxis). But in other cases, the differences are harder to discover. Why, for instance, have history and biology both managed to alter their demographics so that roughly half their majors are women, while philosophy has not? Very few of the factors that have already been addressed in other fields or of those potentially unique to philosophy have been empirically explored in the field of philosophy. In the following sections, we discuss our empirical studies that explore some of these factors. However, we begin here with consideration of some factors that we did not or could not explore with our studies, some of which may have a direct or indirect influence on women’s willingness to continue in philosophy after their initial course(s). We mention one possibility only to set it aside. When describing our concerns about the relatively low number of women pursuing philosophy, a few philosophers and blog discussants have suggested that the primary explanations may have something to do with the “inher-

3.

The perfect storm

Given the harmful effects of the underrepresentation of women in philosophy and the lack of focus on the early exit of women from philosophy, we decided to investigate some potential factors that may contribute to this drop-off after initial philosophy courses. Of course, the disproportional loss of women in philosophy at various stages surely results from many causes, interacting in complex ways. Here, we agree with Louise Antony’s Perfect Storm Model (2012, p. 231), which “seeks to explain women’s low representation within philosophy as a kind of interaction effect among familiar kinds of sex discrimination that are operative throughout society, but that take on particular forms and force as they converge within the academic institution of philosophy,” humanities-degrees-provide-great-return-on-investment/. (Alas, one explanation for philosophy majors’ higher earnings, relative to those majors in other humanities, is the higher proportion of men with philosophy degrees combined with the fact that, on average, men have higher salaries than women.)


-4-

vol. 16, no. 6 (march 2016)



ent” nature of philosophy, or of women, or some combination of the two. We are highly dubious about such claims for a number of reasons — for instance, the implausible essentialist assumptions about philosophy and gender. But more importantly, even if one could somehow make sense of the claim that a “natural” proportion of women philosophy majors would be about 30%, it would be impossible to know that without first exploring and ruling out the many other factors that may contribute to such a disproportionate ratio. So, our project of motivating such an exploration and taking some initial steps is required, even if one starts with the rather bizarre views described in this paragraph. Discrimination and harassment against women in a given field is likely to drive women away from that field. The frequent discussion of sexual harassment and gender discrimination in philosophy suggests that these factors have surely led some women to leave the field. Obviously, these incidents are unjust, in some cases illegal, and they must end. However, we will focus on other factors for women’s underrepresentation. Many departments likely do not have a culture of sexual harassment or overt discrimination and yet still have few women majors. These departments may have other forms of harm (e.g., Rowe, 2008; Brennan, 2013 on the effects of micro-aggressions). We will not pursue the issue of explicit gender discrimination further in this paper, in part because we did not attempt to collect data about it from our participants, though these issues may be important in combating climate problems in philosophy (see Saul, 2014 on informal responses to and responsibilities concerning instances of sexual harassment). One prominent hypotheses about women’s underrepresentation in philosophy is that implicit bias (among other factors) causes women to leave the field. This hypothesis suggests that instructors may have implicit biases against women in their courses, which cause them to devalue work done by these students, call on these students less often in class, or fail to mentor these students to the same extent as men. Implicit biases are unconscious attitudes about members of a particular group, which affect the way that an individual with bias interacts with or makes decisions concerning members of that group. Saul (2013) has

suggested that implicit bias may be one factor for women’s initial dropoff in philosophy. We find this claim plausible, but our study focused on other potential factors.


3.1 Grade Discrepancy between Philosophy and Other Introductory Courses Another factor that might lead women to leave philosophy is if they receive grades in philosophy courses lower than the grades they receive in other courses. (Indeed, if women receive lower grades than men relative to overall grades, this may be in part due to implicit biases of instructors.) In general, students who receive worse than usual grades are more likely to switch majors, whereas students who receive better than usual grades are more likely to continue in that major (Arcidiacono, 2004). Other research suggests that women rely more heavily than men on grades to assess their performance in some field-specific tasks (Correll, 2001). In STEM fields and economics, women who receive B’s are less likely to continue taking courses compared to women who receive A’s, whereas there is no such difference for men (Rampell, 2014). There is also evidence that women have better overall GPAs than men (Chee et al., 2005). So, if women and men receive on average the same grades in introductory philosophy courses, then it is evidence that, on average, women are performing worse than men compared to their performance in other fields, and as a result, they may be more likely than men to stop taking philosophy courses. In Appendix C we report results of a study examining grades in all Introduction to Philosophy courses at GSU offered over 13 semesters (a total of 8,918 students, 4,655 women and 4,263 men). We found that, while men and women received on average the same grades in Intro, women’s grades were lower relative to their overall GPA (since women’s overall GPAs were higher than men’s). This gender difference in grade discrepancy had an influence on the fact that fewer women, relative to men, took philosophy courses after their Intro course according to university enrollment records (see Appendix C).8 We do 8. We also found that, in our climate surveys, women and men reported no

-5-

vol. 16, no. 6 (march 2016)



not, however, have data to support various hypotheses that might explain why women, compared to men, received lower grades in Intro relative to their other courses. While implicit biases of instructors and perhaps even other students may play a role, we suspect more significant contributors include that, because women, in general, identified less than men did with the topics or methods in the Intro class (see Section 4.1 and Section 4.3), women may have accordingly put less effort into the class relative to their other courses than did men.9 Furthermore, women reported feeling less confident in their philosophical abilities compared to men (see Section 4.2), which may have caused women to perform worse in their philosophy courses. Women may also have experienced stereotype threat in philosophy in a way that caused some to perform worse in their philosophy course relative to their other courses. Some of the factors described above are likely to be a part of the Perfect Storm of contributing factors that influence women to drop out of philosophy at disproportionate rates. To test the impact of some other potential factors, we conducted climate surveys involving over a thousand students in dozens of sections of Introduction to Philosophy during two semesters.

ness to continue in philosophy. We examined the following factors (in the order discussed below): confidence in philosophical skills, comfort and sense of belonging in the classroom, dislike of the method of thought experiments, philosophical issues seen as irrelevant, perceptions of philosophy major as useless, aggressiveness of class discussion, lack of strong opinions or different opinions from peers, the presence of women role models, proportion of women authors on syllabi, and holding brilliance-based field-specific ability beliefs. Our goal was to get feedback from a large and diverse sample of students in Introduction to Philosophy courses, but we begin with the caveat that future research will be required to see if similar results are found with philosophy students at other universities. Furthermore, we did not carry out our survey on students taking introductory courses in other fields, which will be necessary to test for similarities and differences in results between philosophy and other fields. The climate survey was filled out by 1,540 undergraduates (798 women, 686 men, and 56 who did not report gender) using Scantrons during one of the last few class sessions of their Introduction to Philosophy course at Georgia State University in Fall of 2012 or Fall of 2013. Both surveys had nearly a 70% response rate, with non-response due primarily to absence on the day the survey was administered (1,050 total undergraduates received grades in Introduction to Philosophy in Fall 2012, and 1,187 students in Fall 2013; IRB protocol also allowed students to opt out). One survey was administered in 34 sections of Intro (with 16 different instructors, 3 women) in Fall 2012 and included 35 questions about students’ experience in and perceptions of the course and philosophy, as well as some demographic questions about demographic categories such as race and gender, and five open-ended questions. Another survey was administered in 27 sections (with 14 different instructors, 3 women) in Fall 2013, and it included the exact same 35 questions as the 2012 survey plus an additional 20 questions designed to test further hypotheses and the demographic questions but no open-ended questions (see Appendix A for survey questions). 724 students (385 women and 323 men) took the 2012 version of the climate

4. A climate survey of students in Introduction to Philosophy courses We designed a climate survey to test a number of hypotheses that have been suggested in the literature, and a few hypotheses of our own, regarding factors in Introduction to Philosophy courses that might differentially influence women and men, and their interest in and willingdifference in their expected grade (see Q55 in Appendix A), which suggests that women are aware that they are getting grades that are, on average, lower than they expect based on their overall GPA. Furthermore, there was a marginally significant trend towards women reporting less satisfaction with their expected grade than men (Q40). 9. Most instructors in Intro at GSU use anonymous grading methods for most major assignments, so any effects of implicit bias would presumably be minimized in those classes.


-6-

vol. 16, no. 6 (march 2016)



survey, and 816 students (413 women and 363 men) took the 2013 version. Since some of our survey questions are related to each other, some of them likely measure the same aspects of students’ experience in Intro. In other words, different questions might cluster together into groups that are highly correlated with each other and measure the same latent variable that explains people’s responses to that group of questions. Factor analysis is a statistical method of looking for such latent variables or “factors.” We conducted a factor analysis on the questions from our longer 2013 survey. Note that factor analysis can separate factors that may seem conceptually related given that the questions load onto different factors. We found seven factors that a subset of the survey questions appear to measure, which we then used for some of the analyses described below (see Appendix B for details).10 Based on the questions correlated with each factor, we interpreted these factors as:

in addition to factors. For these analyses, as an indicator of likelihood of taking more philosophy after Intro, we calculated a new variable, Willingness to Continue, by averaging two survey items: ’I plan to take another philosophy course after this one’ (Q11) and ’I would consider majoring in philosophy’ (Q12).11 The rest of this section is a brief primer on the statistical analyses used below that might be useful to those without background in statistics (see also note 13, note 18, Appendix A, and Appendix B).12 In this article, we test a number of hypotheses about the relationships between gender, IdentificationWithPhilosophy, Willingness to Continue, other factors, various individual survey items, and so on. Each hypothesis is about the relationship between some variable, such as gender, and some other variable, such as IndentificationWithPhilosophy. We test each hypothesis by conducting an analysis involving the variables the hypothesis considers. The result of each analysis has a p-value, which indicates the probability of getting that result given that the null hypothesis is true. For our analyses, the null hypothesis is always that some variable has no real relationship with some other variable. For example, if we conduct an analysis to test the relationship between gender and IdentificationWithPhilosophy, then the null hypothesis is that gender has no real relationship with IdentificationWithPhilosophy. The p-value of the result of this analysis is the probability of getting this result if gender had no real relationship with IdentificationWithPhilosophy. If this p-value is low enough, we can reject this null hypothesis and conclude that gender does have some real relationship with IdentificationWithPhilosophy. We decide whether

1. 2. 3. 4. 5. 6. 7.

IdentificationWithPhilosophy PerceivedInstructorFairness PerceivedStudentRespect AlternativeClassroomPreference ComfortSpeaking Field-SpecificAbilityBelief and GenderAndRaceGapBelief.

Since we could not collect data in this study on whether students signed up for more philosophy courses after this Intro course, we used IdentificationWithPhilosophy as one indicator of how likely students would be to continue in philosophy. We looked for gender differences in IdentificationWithPhilosophy, and then conducted other analyses to see which of the other factors explain those gender differences. We also wanted to test some hypotheses using individual questions

11. Cronbach’s α = .792 for Q11 and Q12, indicating reliable internal consistency. When testing hypotheses about factors, we used IdentificationWithPhilosophy as an indicator of likelihood of continuing in philosophy. When testing hypotheses about individual questions, we used Willingness to Continue as an indicator of likelihood of continuing in philosophy. The reason for having two indicators of likelihood of continuing was that factor analysis normally makes all questions load onto all factors to some degree. So testing correlations between individual questions and factors (like IdentificationWithPhilosophy) would be improper. 12. We thank a referee at this journal for the suggestion to include this material.

10. We follow the convention in the subfield of causal graph theory of presenting variable names (in this case, factors) as italicized without spaces.


-7-

vol. 16, no. 6 (march 2016)



a p-value is low enough by seeing if the p-value is below some threshold, and if it is, then this result is statistically significant. For testing one hypothesis alone, the conventional significance threshold is .05, so a p-value must be less than or equal to .05 for a result to be statistically significant. Testing multiple hypotheses increases the likelihood of getting a statistically significant result even if one or more of the null hypotheses are true. To keep this false positive rate from increasing, we use the Sidak formula to calculate a lower significance threshold (see Appendix A for the Sidak-corrected p-values we used when comparing gender differences in responses to individual survey items and note 13 for p-values we used when making other comparisons).

tinue measure, and while the mean responses to these questions suggest that most students do not plan to continue in philosophy, women expressed less willingness to continue than men.14 Since students’ scores in IdentificationWithPhilosophy and Willingness to Continue reflect gender differences in the likelihood of continuing in philosophy after Intro, we investigated what might contribute to the gender differences on these measures. In the following sections, we describe how we tested a number of different hypotheses about the explanation of these gender differences and the results of these analyses. 4.2 Sense of belonging, comfort, and confidence One hypothesis is that, compared to men, women feel less comfortable and confident in the philosophy classroom. This hypothesis has been explored in other disciplines. A mere sense of social belonging can influence people’s goals and motivations, but only for people with at least some identification with the field (e.g., who felt the field was important to them) (Walton et al., 2012). Good, Rattan, and Dweck (2012) found that merely priming the stereotype that women are worse than men at mathematics causes students who have a fixed mindset about mathematical skill to report a lower sense of belonging to the field than those students who either were men or had a malleable mindset (as described in Section 4.7). While social belonging is very important for students’ persistence and confidence in a field, Cheryan, Plaut, Davies, and Steele (2009) have shown that even physical cues in the environment can lower individuals’ sense of belonging. Participants were asked to wait in a room filled with either stereotypical

4.1

Gender differences in variables measuring identification with philosophy and willingness to continue One of the purposes of our climate surveys was simply to get information about what students think about philosophy and the way it is taught in Intro courses (at one university) and whether and how women and men perceive the course differently. For this purpose the responses to many of the questions, and the gender differences (or lack of differences) to these questions, as reported in Appendix A, are useful to examine. For instance, in response to Q2–Q4, on average students agreed that they enjoyed the course and that they find philosophy interesting and relevant, but on average women agreed to a significantly lower degree than men. Indeed, significant gender differences were found for all of the other questions that make up factor 1, IdentificationWithPhilosophy, including those that ask about students’ perception of their abilities in philosophy (Q5, Q30, Q31), about their sense of how well they fit in philosophy (Q7, Q9, Q10), and about whether they are interested in taking more philosophy or majoring (Q11, Q12) (see Appendix A).13 The last two questions make up the Willingness to Con-

A, we corrected our significance threshold for multiple comparisons. We used a Sidak correction with n = 16 for hypotheses tested outside Appendix A, which would make our significance threshold .0032. We considered p-values at or below this threshold to be significant. 14. We conducted a simple linear regression with gender as a predictor of Willingness to Continue, and found that women score significantly lower than men (β = −.165, t(1479) = −6.427, p < .001).

13. We conducted a simple linear regression with gender as a predictor of IdentificationWithPhilosophy, and found that women score significantly lower on this factor than men (β = −.212, t(739) = −5.902, p < .001). Since we conducted 16 analyses in addition to the t-tests on gender differences described in Appendix


-8-

vol. 16, no. 6 (march 2016)



computer science items (e.g., Star Wars posters, video games) or nonstereotypical items (e.g., art, nature posters), but asked not to pay attention to the contents of the room (as it was being used by another group on campus). They were then asked to fill out surveys about their feelings toward computer science under the guise of a Career Development Centre survey. Women felt less interested than men in computer science in the stereotypical room, but there was no difference in interest in the non-stereotypical room. In another study, Cheryan et al. asked participants to choose a team to work with in their hypothetical employer’s company, and women consistently chose the team in the non-stereotypical room despite both teams’ containing all women. We asked several questions to gauge students’ level of comfort in the classroom, as well as their sense of belonging and their sense of confidence in their abilities. Women agreed less than men that they have a lot in common with the typical philosophy major (Q7) and with their philosophy instructor (Q9), that people like them can be successful in philosophy (Q10), that they are confident in their ability to talk about philosophy (Q31), and that they are confident in their ability to do well on papers and exams (Q32).15 However, students also generally felt comfortable with their instructors as both discussion leaders and evaluators. Both men and women agreed that they felt comfortable asking the instructor questions after class or in office hours (Q15), that the instructor made an effort to involve all students regardless of race or gender (Q16), and that the instructor treated all students with respect regardless of race or gender (Q17, Q18). Students generally reported feeling that the instructor graded them fairly too (Q24). Finally, there was no gender difference on the factor PerceivedInstructorFairness.16

Women scored significantly lower than men on the factor ComfortSpeaking.17 To see if gender differences in comfort speaking in the classroom partially “mediate,” or account for, why women score lower than men in IdentificationWithPhilosophy, we conducted a mediation analysis with ComfortSpeaking as a mediator of the effect of gender on IdentificationWithPhilosophy.18 We found that ComfortSpeaking is a partial mediator of the effect of gender on factor 1, indicating that gender differences in comfort speaking in the philosophy classroom partly explain why women score lower than men in IdentificationWithPhilosophy.19 It will be important to carry out further research to determine if these gender differences in sense of belonging, comfort, and confidence are particularly significant in philosophy or which, if any, other subjects have introductory courses that show similar gender differences in these factors. We now turn to a few hypotheses that we have reason to believe are more specific to philosophy. 17. We found a significant correlation between gender and ComfortSpeaking (r = −.114, p = .002), which is equivalent to step 2 of the Baron and Kenny method described in note 18. 18. A mediator is a variable that explains or accounts, at least in part, for the effect of one variable on another. For example, we found that gender is a significant predictor of IdentificationWithPhilosophy. When we call some other variable a “mediator” of the effect of gender on IdentificationWithPhilosophy, we mean that this other variable accounts for the effect of gender on IdentificationWithPhilosophy. So ComfortSpeaking is a mediator of the effect of gender on IdentificationWithPhilosophy just in case ComfortSpeaking accounts for the effect of gender on IdentificationWithPhilosophy. If ComfortSpeaking accounts for all of the effect of gender on IdentificationWithPhilosophy, then ComfortSpeaking is a “full” mediator of this effect. If ComfortSpeaking accounts for some, but not all, of the effect of gender on IdentificationWithPhilosophy, then ComfortSpeaking is a “partial” mediator of this effect. We looked for full and partial mediators by following the three steps that Baron and Kenny (1986) recommend, and by using the Sobel test. 19. In step 3 of the Baron and Kenny method, we found that both gender (β = −.167, t(738) = −5.042, p < .001) and ComfortSpeaking (β = .394, t(738) = 11.889, p < .001) were significant predictors of IdentificationwithPhilosophy. Since gender was still a significant predictor of IdentificationwithPhilosophy even when controlling for ComfortSpeaking, it was not a full mediator of the effect of gender on IdentificationwithPhilosophy. Using the Sobel test, however, we found that ComfortSpeaking was a partial mediator of the effect of gender on IdentificationwithPhilosophy (z = 3.013, p = .003, effect size rβ = −0.045).

15. See Appendix A, Table 1, which indicates significant gender differences for these questions for both the 2012 and 2013 surveys (though, for Q7 and Q10, the 2012 results indicate trends that are not significant when corrected for multiple comparisons). 16. See Appendix A and Appendix B for means and p-values.


-9-

vol. 16, no. 6 (march 2016)



4.3 Thought experiments and other methods in Intro classes One set of hypotheses suggests that some of the unique methods used in the philosophy classroom may be disproportionately unappealing to women. Perhaps women are less likely to hold similar philosophical intuitions as men, such that when thought experiments are used in philosophy classrooms, women may feel out of place. Alternatively, it may be the very method of thought experiments, rather than a difference in prompted intuitions, that helps to explain the gender gap. Perhaps women are more likely than men to find the method of thought experiments used in many philosophy courses to be ineffective or unsatisfactory for answering philosophical questions. These hypotheses have the benefit of applying specifically to philosophy among the humanities and other subjects. Buckwalter and Stich (2013) conducted one of the earliest empirical investigations of the initial drop-off of women in philosophy. They presented evidence for gender differences in philosophical intuitions in response to 14 scenarios, such as the Trolley Problem and a Gettier case, and suggested that women may feel, or be made to feel by their instructors, that they have the wrong intuitions, leading them to lack confidence in their own philosophical skills or feel alienated in the classroom. If so, this might be a plausible explanation for why some women stop taking philosophy after Introductory courses. Although Buckwalter and Stich presented an interesting hypothesis and provided data to support it, further studies have failed to replicate gender differences on the thought experiments presented in their papers (Adleberg et al., 2015; Seyedsayamdost, 2015; Chernykhovskaya, ms). Beyond the empirical problems with their results, others have noted problems with Buckwalter and Stich’s argument (Antony, 2012; Adleberg et al., 2015). We conclude that if gender differences in philosophical intuitions exist, then they likely play a very minimal role in women’s exodus from philosophy, and it would be more fruitful to consider alternative hypotheses.

Buckwalter and Turri (2016) suggest a different explanation for how thought experiments in Intro classes might have a disproportionate effect on whether women continue in philosophy. Women, more than men, may simply find the methodology of using thought experiments and intuitive responses to them less interesting or valuable, or they may prefer the methodologies used in other fields. Buckwalter and Turri examined participants’ responses to scenarios about investigations run by teams or an individual about knowledge or about human selfishness using either the distinctive methodology of psychology (i.e., observation) or one methodology particular to philosophy (i.e., thought experiments). They found that, although all participants preferred observational methods to the use of thought experiments, women preferred observational methods more strongly than men. Their results also showed that women valued research questions pursued by a group of researchers as opposed to a lone researcher, whereas men had no preference based on the number of investigators. Buckwalter and Turri’s results suggest interesting potential factors in women’s early exit from philosophy. If philosophical thought experiments are often used in introductory philosophy classes and women prefer observational methods over methods that rely on intuitions, then perhaps this causes some women to migrate away from philosophy and towards fields that use observational methods, such as psychology. On this hypothesis women may even enjoy investigating philosophical questions just as much as men, but it is the method of using thought experiments that they find uninteresting or unfruitful. Buckwalter and Turri suggest that all students, perhaps especially women, would likely be more interested in philosophy if it were introduced using fewer thought experiments and, conversely, using more scientific work, including experimental philosophy.20


20. This hypothesis is complicated by the fact that women are also underrepresented in some STEM fields that use experimentation and observation (e.g., physics and economics).

- 10 -

vol. 16, no. 6 (march 2016)



To explore whether the use of thought experiments might disproportionately discourage women from continuing to enroll in philosophy classes, the 2013 climate survey asked students whether they liked “the method philosophers use of testing our intuitions in response to ’thought experiments’ (imaginary situations)” (Q36). The mean responses suggest agreement to this question, though women agreed less than men that they liked the method of testing intuitions with thought experiments (see Appendix A). To see if this difference helps to explain why women scored lower than men in Willingness to Continue in philosophy, we used the Baron and Kenny (1986) method to conduct a mediation analysis with Q36 as a mediator of the effect of gender on Willingness to Continue. We found that Q36 partially mediates the effect of gender on Willingness to Continue, suggesting this difference in responses does partially explain why women are less willing to continue in philosophy than men.21 This result provides some support for Buckwalter and Turri’s hypothesis. However, we should also note that men agreed more than women on our Q37, which asked if the students would prefer philosophy more if it used scientific experiments (see Appendix A). It would also be interesting to know the extent to which Intro courses use thought experiments and the degree to which they present the philosophers’ or the (majority of) students’ intuitions as a type of evidence. Finally, we did not find evidence of differences between women and

men in whether they report that they would have enjoyed the course more if there had been more discussion (Q50, Q51), or more discussion of real-world problems (Q53), and only marginal trends towards women wanting more non-philosophy texts (Q54). Together, most of these questions (Q50–53) make up factor 4, AlternativeClassroomPreference, and we found no gender differences on this factor (p=0.84).22 Overall, these results suggest that some women may be leaving philosophy after their first classes partly because they do not like some of the specific methods used in philosophy courses, or like them less than men do. 4.4 Perceptions of philosophy as irrelevant or useless Another way that the methods or content of philosophy classes may impact women’s interest in the subject involves the stereotypical perception of philosophy as irrelevant or useless. Philosophy is often seen as an impractical major that will not help students get a job.23 To the extent that people recognize certain questions as philosophical, those questions are often viewed as irrelevant or esoteric. It may be that women hold these stereotypes about philosophy more strongly than men, or that women’s perceptions of these issues impact their desire to study philosophy more than for men. Our surveys included several questions aimed to explore these hypotheses. We asked whether students believed that the topics that philosophers discuss are relevant to their lives (Q4), and in both 2013 and 2012 surveys we found that women agreed that the topics are relevant at a significantly lower level than men (see Appendix A). As we discussed above, both men and women may have preferred to read more newspaper articles, literature, etc. in their philosophy course (Q54), and such readings may offer one way to increase the relevance of philosophy

21. In step 1 of the Baron and Kenny method, gender significantly predicted Willingness to Continue (β = −.165, t = −6.427, p < .001). In step 2, gender significantly predicted scores on Q36 (β = −.105, t = −2.920, p = .004). In step 3, both gender (β = −.145, t = −4.347, p < .001) and Q36 (β = .351, t = 10.504, p =< .001) predicted Willingness to Continue. The effect size of the indirect path was (−.105)(.351) = −0.037. Since the effect of gender on Willingness to Continue was still significant even when controlling for Q36 in step 3, Q36 was not a full mediator of the effect of gender on Willingness to Continue. Using the Sobel test, however, we found that Q36 was a partial mediator of gender on Willingness to Continue (z = 2.813, p = .005).


22. Note, however, that while the means for most of these questions were near or below the midpoint, both women and men tended to agree that they would have enjoyed more discussion of how philosophical issues relate to real-world problems (Q53). See below. 23. See, e.g., “Does Philosophy Matter?” in The New York Times, August 1, 2011.

- 11 -

vol. 16, no. 6 (march 2016)



for all students, perhaps especially women. Or perhaps when students find their course syllabi populated mostly by authors who are different from them (e.g., in gender, race, religion, etc.), this impacts their sense of relevance (see Section 4.6 and Appendix D). Students primarily choose majors in fields that match their interests, that allow students to maintain a high GPA, and that offer lucrative, autonomous careers after graduation (Beggs et al., 2008; Malgwi et al., 2005). We used Q34 to measure the latter — the perceived usefulness of a philosophy major for getting a job. Women perceived philosophy to be less useful then men, but the differences were not significant when corrected for multiple comparisons (see Appendix A). There were no significant gender differences in questions that asked about the importance of majoring in a subject useful for getting a job (Q41), with agreement very high. To see whether the stereotype that philosophy is useless could be counteracted, we created a 10-minute presentation on the value of philosophy, which included information about careers for philosophy majors, the high mid-career income and income growth of philosophy majors (vs. other majors), high test scores, etc. In roughly half the sections of Intro in Fall 2013, instructors showed this presentation and discussed the value of the philosophy major. We found a significant difference in student responses to Q34 on the usefulness of the major between those sections that included the presentation and those that didn’t. We found that students who viewed the presentation perceived philosophy to be a more useful major than those who didn’t, but we did not find that the presentation affected the gender gap in perceptions of usefulness.24 We did, however, find that students who found

philosophy to be a useful major were more likely to report wanting to continue taking philosophy classes.25 4.5 Aggressive, Strong, or Different Opinions Another set of hypotheses concerns the manner in which students engage in philosophical discussions in the classroom. It may be that the sometimes aggressive nature of discourse in philosophical discussions discourages women more than men from participation or interest in philosophy. Alternatively, the strength with which students hold their views on philosophical topics may affect students’ desire to engage in philosophical discussion. Finally, the extent to which students feel their views on philosophical topics differ from others’ in the classroom may correspond to an increase in those students’ willingness to engage in philosophical debate — the student may feel that their ideas must be defended against the status quo and, since they are one of the few students in the room who hold that view, they must be the ones to defend it. One pedagogical factor that might influence women to leave philosophy more than men involves the argumentative atmosphere in some philosophy classes. The hypothesis is not that women cannot handle or dislike argumentation (in the technical sense), but rather that women are socialized to avoid public confrontation and arguing (in the colloquial sense). Further, this hypothesis predicts that women who act aggressively during philosophical discussions will not reap the benefits of this behavior to the same extent as men who act within the aggression norm and may even experience backlash for acting within the norm (Rudman and Glick, 2001; Rudman and Fairchild, 2004). The norm is that men are aggressive, so it is more surprising and less accepted when woman act in the same aggressive manner (Moulton, 1983). If the idea of philosophical competence is associated with argumentative behaviors, then philosophically competent women may be

24. For students who viewed the presentation, the means and standard errors for Q34 were M = 3.22, SE = .060, while for students who did not view the presentation, they were M = 2.66, SE = .036. This difference was significant (t(1521) = −7.852, p < .001). We conducted a 2x2 ANOVA to see if viewing the presentation affected women’s responses to Q34 differently from those of men, but we did not find a significant interaction effect (F (1, 1470) = 0.301, p = .583).


25. For Q34, students who viewed philosophy as a useful major scored higher on Willingness to Continue (β = .462, t(1526) = 20.37, p < .001).

- 12 -

vol. 16, no. 6 (march 2016)



more likely to be seen as aggressive compared to philosophically competent men. So, under this hypothesis, even women who “play by the rules of the game” may be viewed or treated negatively. We asked students whether they thought classroom discussions were “too combative” (Q19) and whether they felt “uncomfortable with the confrontational nature of some philosophical discussions” (Q39). There were no differences in responses between women and men, and students tended not to think that classroom discussions were too combative or confrontational (see Appendix A). Hence, our evidence does not support the hypothesis that women find Intro classes to involve overly aggressive or confrontational discussions. Further, in general, students regardless of gender felt that the other students in their philosophy course treated each other with respect. There was no significant gender difference in PerceivedStudentRespect. However, we did not have any questions directly testing a related possibility: that men in Intro classes dominate class discussions and may be called on more by instructors, which women notice and dislike (see Chan et al. (ms.) for qualitative evidence of this phenomenon). Even if instructors sometimes try to "shut down" domineering students or find ways to balance class discussions, they may also implicitly or explicitly reward students who contribute to discussions (there is, after all, no inconsistency between talking too much in class and often raising interesting points or questions). This phenomenon of overly discursive (and perhaps over-indulged) male students may be particularly problematic in philosophy courses. The topics are often controversial, students often feel passionate about their views and have had limited opportunities to discuss them in academic settings, and it is not always obvious what counts as a correct, or cogent, answer to each question. Another possibility is that women have less strongly held opinions about these philosophical topics than men. Women did report less agreement than men with Q35, which asks whether they have strong opinions about the philosophical issues discussed in the class (see Appendix A). We conducted another mediation analysis and found that Q35 was a significant partial mediator of the effect of gender on Will-

ingness to Continue, so the extent to which men and women take themselves to have strong philosophical opinions also partly explains why women score lower than men in Willingness to Continue.26 Finally, students may not enjoy philosophy classes if they feel their opinions differ from other students’ or from their instructors’. Q25 asked whether students often had different opinions from the instructor, and there was no difference between responses from women and men (see Appendix A). Q26 asked whether students believed their opinions differed from the opinions of other students in the class. Here, women agreed less than men (significant only in the 2012 survey), so men more than women believed their philosophical opinions differed from other students’. To see if this gender difference in responses to Q26 helps to explain differences in Willingness to Continue, we conducted a mediation analysis with Q26 as a mediator of the effect of gender on Willingness to Continue. Q26 was a partial mediator of the effect of gender on Willingness to Continue.27 That is, women reported less of a difference between their own opinions and those of others in the class, and this gender difference in responses to Q26 partly accounts for the difference between men and women in Willingness to


26. In step 2 of the Baron and Kenny method, gender was a significant predictor of Q35 scores (β = −.116, t(1476) = −3.248, p = .001). In step 3, gender (β = −.139, t(1474) = −4.189, p < .001) and Q35 (β = .369, t(1474) = 11.125, p < .001) were both significant predictors of Willingness to Continue. The effect size of the indirect path is (−.116)(.369) = −0.043. Since gender was still a significant predictor of Willingness to Continue even when controlling for Q35, it was not a full mediator of the effect of gender on Willingness to Continue. Using the Sobel test, however, we found that Q35 was a partial mediator of the effect of gender on Willingness to Continue (z = 3.118, p = 0.002). 27. In step 2 of the Baron and Kenny method, gender was a significant predictor of Q26 scores (β = −.096, t(1477) = −3.725, p