There is no doubt anymore that there is a steady and significant underrepresentation of women in philosophy. This underrepresentation manifests itself in a variety of ways, such as in the numbers of published articles in highly ranked journals and of philosophy faculty positions held by women, in the list of invited speakers at academic events, in the number of women contributors to edited volumes, and in the citations of publications by women authors (e.g., Beebee & Saul 2011; Benetreau-Dupin & Beaulac 2015; Haslanger 2008; Healy 2015; Paxton, Figdor, & Tiberius 2012).¹ While the underrepresentation of women is not unique to philosophy (Beebee & Saul 2011), it is particularly pronounced in philosophy. The situation is comparable to that in STEM fields, that is, science, technology, engineering and mathematics, and—surprisingly—worse than in other humanities (Antony 2012; Leslie, Cimpian, Meyer, & Freeland 2015; Paxton et al. 2012).²

While the underrepresentation of women in philosophy is apparent on all professional levels, the underrepresentation already begins at an early career stage. For example, Paxton et al. (2012) analyze an existing gender gap in philosophy across different professional levels. Their results show that the statistically significant dropout rate in the United States can be found at the bachelor’s level and is particularly pronounced between introductory courses and the decision to major in philosophy (Paxton et al. 2012; Baron et al. 2015, Dougherty, & Miller 2015; Hutchison & Jenkins 2013). Along the same lines, Thompson, Adleberg, Sims, & Nahmias (2016: 2) find that women leave philosophy disproportionately often after taking only a few courses.

That the underrepresentation of women in philosophy is much larger than in other humanities is particularly striking. This discrepancy suggests that there might be features specific to philosophy to which men and women respond differently and in turn contribute to this situation. Examples of such discipline-specific features could be the methodologies used in philosophy (e.g., argumentation, logic, thought experiments, etc.), its subject matter, its strong emphasis on argument and discourse, and the canon of philosophical ideas, among others.³ Discussions around whether such features are ultimately responsible for women’s underrepresentation often rest on the implicit assumption that women and men have inherently different interests, attitudes, perceptions, and professional goals that philosophy can or cannot satisfy. This difference may lead to gender differences in attraction towards philosophy as, for example, a discipline providing a comfortable and constructive learning environment and promising an interesting and achievable career.

In this paper, we empirically explore some of the mechanisms that rest upon such “essentialist assumptions” (Thompson et al. 2016: 5) about philosophy and gender and how they play out in German academia. By undertaking an exploratory study at one of the biggest and best-known German universities, namely, the Ludwig Maximilian University of Munich (LMU), we ask whether the underrepresentation of women in the profession can be explained by factors inherent to philosophy as an academic discipline and/or specific to women in order to better understand why women eventually drop out of philosophy.⁴ Gaining a better picture of whether there are gender differences in interests, abilities, beliefs, attitudes, perceptions and goals among philosophy students could prompt further research to explore why those gender differences exist in the first place.

We conducted our study in 2014. We compared the interests, abilities, beliefs, attitudes, perceptions, and goals of bachelor’s students in a semester-long philosophy of science course at LMU. LMU houses one of the largest philosophy departments in Germany. As of today, only 3 out of 11 chairs in philosophy are held by women.⁵ This roughly reflects the situation in Germany as a whole, where between 2005 and 2016 only an average of 15% of all philosophy chairs were held by women.⁶

In Germany, the dropout rate among philosophy students is highest between the bachelor’s level—where an average of 45% of women graduated between 2005 and 2016—and the master’s level, where on average only 37% of women graduated in the same period.⁷ In contrast, an average of 77% of all bachelor’s students in other humanities and an average of 72% of all master students graduating in the same period were women. At LMU, dropout rates are also highest between the bachelor’s and the master’s level. On the bachelor’s level, the dropout rate among philosophy students at LMU generally is rather high. Between 2013 and 2016, the dropout rate was 41% among women and 42% among men. While those dropout rates are similar, the reasons behind this dropout rate might differ across gender. The underrepresentation of women in Germany becomes more evident at the master’s level. While on average, 42% of students in the philosophy master’s program are women, on average only 32% of those women finished the master’s program between 2013 and 2017.⁸

Our analysis of students’ interests, beliefs, abilities, attitudes, perceptions, and goals is unique in that it provides, first, a set of preliminary insights into possible causal factors behind women’s underrepresentation in philosophy in Germany that could be studied further. Second, by introducing an empirical design that allows for comparing mixed-gender and women-only study environments, our analysis provides a basis for re-applying this study design elsewhere and thereby allows for cross-country comparisons. Our analysis points out a set of avenues that should be explored in Germany and elsewhere as well as on a large scale. It also can initiate a substantial discussion of concrete institutional changes that could potentially counteract underrepresentation, such as whether women-only environments can encourage women to continue studying philosophy.

By studying the underrepresentation of women in philosophy in German academia, our analysis complements existing studies conducted in other countries, such as the United States, the United Kingdom, and Australia (Adleberg, Thompson, & Nahmias 2014; Baron et al. 2015; Beebee & Saul 2011; Paxton et al. 2012; Thompson et al. 2016). First, while more empirical data is available, empirical studies that aim at better understanding the relevant factors and their significance across countries are missing. Causally responsible factors, the way those factors interact, and their level of influence might vary across geographical and cultural borders. Therefore, results from Germany might help us to assess such potential variations. Second, while our study is similar in focus to existing studies, we consider additional factors such as those embodied in the essentialist assumptions mentioned above.

Gaining a better understanding of the mechanisms underlying women’s underrepresentation in philosophy is important for multiple reasons. Primarily, if universities can identify the mechanisms, they can take concrete steps towards mitigating underrepresentation, which is central to promoting equal career opportunities for women in philosophy. Furthermore, diverse representation is essential for the quality of philosophy as a discipline. Diversification guarantees that the best candidates—men or women—get the chance to produce high quality work in the profession, allowing us to draw from a wider pool of original ideas.⁹ This exploratory study takes a first step towards a better understanding of the mechanisms that cause women to abandon philosophy at an early stage.

Likely, not a single factor, but multiple mechanisms operate simultaneously and together contribute to the underrepresentation of women in philosophy (see also Antony 2012; Thompson et al. 2016: 4).¹⁰ While there is thus no single causal model to explain women’s underrepresentation among philosophy students, a number of mechanisms have been identified—sometimes only on the basis of anecdotal evidence—as being potentially responsible for it (e.g., Dougherty et al. 2015; Thompson 2017).¹¹ In this section, we present the most frequently discussed mechanisms underlying underrepresentation and select those that we explore in our analysis.

A first mechanism (M1) relates to gender differences in students’ scientific and/or mathematical abilities (Thompson 2017: 3). Students’ abilities may play a role when they select study programs or decide whether to continue with their selected program. Regarding philosophy as a field of study, the underlying idea is that men and women differ in their abilities of abstract thinking and rational reasoning, and their skills in using formal and mathematical tools, such as logic. Because philosophy is a field that uses those skills, women drop out once they realize that those skills are required.

A second mechanism (M2) relates to gender differences in students’ interests. Women may have different interests than men. For instance, women might care more about problems that are practically or socially relevant (Baron et al. 2015; Dougherty et al. 2015: 2). Because from their point of view large parts of philosophy do not address such problems, they switch disciplines and choose a field in which they see their interests better realized.

A third mechanism (M3) refers to gender differences in beliefs about one’s own abilities and about general abilities that are required for success in a field (Thompson 2017: 6). The concern is not what abilities students actually have but what they believe their own abilities are (Dougherty et al. 2015; Leslie et al. 2015); those beliefs can clash with their actual abilities.¹² M3 as well as M1 and M2 can interact, as students’ actual abilities and their ability beliefs can both influence their interest in a specific field and vice versa. Receiving low grades in a certain subject—say, in philosophy—may decrease interest in the subject and the motivation to work hard. It can lead to a shift in interest towards an area in which one performs better. At the same time, an originally large interest in a subject as well as higher marks may come with a greater motivation to work hard and the desire to continue studying the subject (Arcidiacono 2004).

A fourth mechanism (M4) refers to gender-specific schemes that we unconsciously use to collect and save information about the world (see, e.g., Valian 1997). They allow us to generalize and form stereotypes that are grounded upon characteristics that we attribute to a particular gender. These schemes can clash with reality when they do not align with our observations. For instance, when historically thinking of a philosopher as male, women may find it difficult to imagine themselves as philosophy majors, as the schema for ‘philosopher’ and the schema for ‘woman’ clash (Calhoun 2015; Hutchison & Jenkins 2013). Because we draw on schemes that rely on connotations of ‘philosopher’, ‘rational reasoning’, ‘logical thinking,’ ‘abstractness,’ etc. as male, such a clash comes with the risk that women are (unintentionally) excluded from the profession or leave philosophy voluntarily (Haslanger 2008).

A fifth mechanism (M5) refers to gender differences regarding the sense of belonging and the perceived atmosphere in philosophy as a field, which in turn influences confidence and the willingness and ability to perform well. The idea is that there are gender differences regarding the perceived atmosphere in seminars, colloquia, and other educational contexts and that, because women perceive the atmosphere negatively in those contexts, they feel less comfortable in them than men. M4 and M5 can be further enforced by the lack of female role models in philosophy.

Two additional mechanisms are implicit bias (M6) and gender stereotype threat (M7). Implicit bias captures the idea that women are generally and unconsciously viewed more negatively than men in philosophy (Saul 2013), which also manifests itself in class discussions, student assessments, etc. (see, e.g., Boring, Ottoboni, & Stark 2016). M6 can potentially interact with other mechanisms, such as M1, M2, and M3, as for example women students’ abilities might be judged lower than they objectively are, if instructors hold implicit biases against women (Dougherty et al. 2015: 4). Gender stereotype threat describes the fear that because of consciously or unconsciously formed negative stereotypes about members of an underrepresented group (here women), members of that group are judged in light of those stereotypes or will conform to those stereotypes. This in turn can lead to their reinforcement (Hill, Corbett, & St. Rose 2010; Spencer, Steele, & Quinn 1999). Other effects can be a lower self-identification with philosophy and worse performance on skill tests. If implicit bias and/or gender stereotype threat operates in philosophy, it can make philosophy a hostile environment for women (Saul 2013).

A last mechanism (M8) potentially operating in academic philosophy is gender-specific discrimination, that is, the systematic discrimination against a person because of their gender. Such behavior can result from implicit bias; as such, this mechanism can be closely connected to M6. Gender-specific discrimination can also include sexual harassment. M8 is particularly relevant when pursuing a career in philosophy, such as, for example, at the level of selection procedures in job searches, or at the level of professional positions, such as editorships, as well as the selection and refereeing procedures of journals (Friedman 2013).¹³

Those mechanisms cannot be cleanly separated, as they frequently operate simultaneously or reinforce each other. However, to further systematize M1 to M8 and make them fruitful for empirical analysis, Dougherty et al. (2015) have offered a set of hypotheses about how they manifest themselves in the classroom. For our purpose, we selected the following five hypotheses that rest upon some essentialist assumptions about gender and philosophy as a discipline.

The Adversarial Argumentation Hypothesis suggests that the discussion style in philosophical debates is often aggressive and arguments are expressed in a particularly adversarial manner (Beebee 2013). Because women tend to shy away from an aggressive discussion style, they are less inclined to engage in such discussions (Dougherty et al. 2015: 4). That results in male-dominated seminar discussions. As a consequence, men not only appear more active and interested but also shape the discussion content and culture in seminars. The hypothesis mainly rests on M4 and M5 in that an aggressive discussion style is stereotypically associated with male behavior. Because of that and other factors such as stereotype threat, women do not feel a sense of belonging in such seminar settings.

We also explore our data with respect to several hypotheses that relate to the content of philosophy courses. First, we study whether there is any empirical support for the Subject Matter Hypothesis, which is grounded in M2. It holds that female and male students have different interests and that the problems discussed in philosophy courses tend to be more tailored to the interests of men (see also Baron et al. 2015). A variant of this hypothesis is what we call the Formal Methods/Abstractness Hypothesis. This hypothesis can be interpreted in two ways. On the externalist interpretation of the hypothesis, women perform worse when using abstract methods than men, which implies that their instructors and/or peers treat them as less able to succeed in formal or more abstract subjects than men, processes which are described by M1. A possible underlying mechanism to explain women’s lower performance could be discriminatory behavior (M8), which could but does not have to result from implicit bias (M6). On the internalist interpretation, women (correctly or incorrectly) hold beliefs about their own abilities (as captured by M3) and draw inferences about how their perceived abilities relate to the skills required for being successful in philosophy.

Applied to the classroom situation, both interpretations imply that women find the abstract problems, the narrow methodological focus on logic, and the methodology of rigorous argument especially in analytic philosophy less interesting, more difficult, and/or not as easily acquirable for them compared to their male classmates. Both interpretations individually or together indicate that women’s success in subjects relying on formal methods (such as logic or mathematical philosophy) is compromised (see also Thompson 2017). The hypothesis is grounded in M3; men are more confident than women in their own abilities, including the ability to do mathematics. Because women more often deprecate their own abilities and skills, they are more prone to think of themselves as having lower chances of pursuing philosophy successfully.

As Dougherty et al. (2015) have pointed out, the Subject Matter Hypothesis closely connects with the Role Model Hypothesis. According to the latter, female students lack a sense of belonging in academic philosophy (captured by M5) mainly because of a lack of female role models, reinforcing a gender-specific scheme of philosophy as being mainly done by, and offered for, men (captured by M4) (see also Thompson 2017: 4). This lack of role models becomes visible, for example, in the absence of women teachers, the men-dominated canon of the core curriculum, the presentation of images of male philosophers on presentation slides, posters, or philosophy websites, and the lack of women authors on course syllabi (Dougherty et al. 2015: 3; Paxton et al. 2012). The lack of role models may signal to women that philosophy is not a place for them, and even less so will offer them a successful career, reinforcing women’s lack of a sense of belonging and self-confidence (M5).

The absence of role models further comes with a risk of generating a study environment in which stereotype threat (M7) becomes more likely than would be the case in an environment in which women were well represented (Saul 2013). A similar effect applies to implicit bias (M6), as an environment lacking role models facilitates ignoring the excellent work done by women and instead associates good work with role models who are men (Saul 2013).

The Subject Matter Hypothesis is also related to the Impractical Subject Hypothesis. The latter states that women are disproportionately less inclined to major in philosophy because besides striving for career goals, such as a secure job and a high income, and/or for personal life goals, such as personal growth, they consider it important that those goals help them have a social impact. The idea is that women want to engage with problems they consider relevant and which are more practical or concrete in nature (see also Dougherty et al. 2015: 7 f.). Because the problems and skillset of philosophy are perceived as impractical for these matters, it does not help women achieve their life goals and they eventually lose interest in philosophy (also captured by M2). The hypothesis further captures the idea that there are gender differences in long-term career and personal life goals and in the beliefs about the usefulness of philosophy for achieving them. Women are more concerned about getting a secure job than men. They perceive an education and an academic career in philosophy more generally as riskier than an education and career in disciplines that address concrete problems and teach applicable skills. In Germany, this could be reinforced by the fact that because, for instance, the tenure-track system has not yet been broadly implemented, the conditions for successfully pursuing an academic career and having a family are often not in place. Studying a more practical subject is perceived as advantageous by women because of their immediate concern to balance family life and professional career (see Antony 2012).¹⁴

In this section, we present our research setup to explore those five hypotheses. We take students’ reports of their interests, beliefs, abilities, and goals, and we regard their perceptions of the academic environment they are part of as equally important as their expressed attitudes towards that environment. We understand attitudes as individual beliefs about the attributes of a specific object (say, academic philosophy), which can be influenced by other attributes (Aijzen 2001; Crano & Prislin 2006; Aijzen & Fishbein 1975). But first, to better understand the demographics of the philosophy discipline beyond the UK, the United States, and Australia and to enable a cross-country comparison of the possible causes of women’s underrepresentation, some notes about the German educational and academic systems are in order.

First, given the federal organization of the German educational system, only some students arrive at the university with prior knowledge of philosophy. In all federal states, philosophy replaces religious education in high school and is thus open only to pupils who do not officially belong to any denomination. The content of philosophy classes is typically limited to ethics but does not have to be (see Report of the KMK 2020). This also holds for the federal state of Bavaria, which hosts the LMU. In Bavaria and other federal states, high school students can further pursue their interest in philosophy by taking voluntary philosophy courses, if offered at their particular school (Report of the KMK 2020: 19f.).¹⁵ Consequently, students arrive at university with different backgrounds, and most of those who have some background in philosophy have it limited to practical philosophy.

Second, German high school students choose at least four areas they wish to focus on in their final high school exam. Depending on the federal state and the other courses they choose, philosophy can but does not have to be one of them. Specialization typically begins only with entering the university system. Once students enter university, they specialize in the field they major in. If students enroll in a philosophy major program, they specialize in philosophy from the first semester onwards. Given that there are typically no prerequisites for enrolment in a philosophy program at German universities, all students can do so, given that they have successfully finished high school. This also applies to LMU.¹⁶

Third, changing academic specialties during their bachelor’s program is generally not as difficult as in other countries. Because there are usually no high tuition fees, changing one’s degree program is rather easy, especially if a student changes to a degree program that does not have any entry requirements. Nevertheless, some factors could make changing one’s major difficult. For instance, not all course assessments might be recognized as valid in the new study program. Furthermore, changing one’s major can mean changes to a student’s funding.¹⁷ Finally, switching from philosophy to a different study program can be challenging if the new study program has specific entry requirements. All three aspects can influence a student’s decision to change his or her study program in Germany.

Procedure. We conducted this study with bachelor’s students enrolled in a philosophy of science class. The course was organized in terms of a lecture format. It was obligatory for all bachelor’s students at LMU who majored in philosophy, that is, who studied philosophy as a so-called ‘Hauptfach’. There were no entry requirements for participating in the course, but students had to pass the course to successfully finish their degree program in philosophy.¹⁸ The lectures were given by a male professor and accompanied by weekly tutorials taught by male and female PhD and postdoctoral students. All but one tutorial were organized as mixed-gender study environments, one of them taught in English. One tutorial was open only to women and was taught by a woman. Students could choose whether to attend a tutorial, and if so which one. While students had to answer weekly questions and write two practice essays, the formal requirement for successfully finishing the course was passing a two-hour exam at the end. Final grades were based solely on the exam.¹⁹

Materials. In total, we used three pen and pencil questionnaires that we distributed at the beginning and at the end of the term. To explore the aforementioned five hypotheses, our questionnaires asked about the following:

The first questionnaire contained 17 items, consisting of closed-ended questions and Likert scales. It included some control questions and (1) questions for collecting demographic and biographical data about the participants and questions about (2) their motivation to study philosophy, (3) their high school grades, (4) their skills in mathematics and computer science, and (5) their perception of the role and usefulness of mathematics in philosophy. The questionnaire was accompanied by a math test to check students’ actual mathematics skills. The test contained 14 problems of basic high-school mathematics that students were asked to solve. Questionnaires and tests were distributed after the first lecture.

The second questionnaire was distributed at the end of the term. It contained 33 items, including close-ended questions, Likert scales, and open-ended questions. Students could fill out this questionnaire during the last session of each tutorial or after the exam. Items were subdivided into the following sections: (1) demographics, (2) participants’ major and minor, (3) their performance in high school, (4) their interest and active participation in the course, (5) their assessment of course topics and philosophical problems more generally, (6) their assessment of philosophy as an academic discipline, (7) their feelings and attitudes in seminar discussions and towards the discussion culture more generally in philosophy, (8) their active participation in seminar discussions, (9) their career expectations in philosophy, and (10) their interest in philosophy as an academic discipline.

The third questionnaire was attached to each student’s exam. It contained four items consisting of close-ended questions, asking about (1) the discipline they studied as major and minor, (2) the frequency of their course participation, (3) the tutorial they had chosen, and (4) how frequently they had attended the tutorial. An additional datum was (5) the grade that students obtained on the exam. Students’ participation in all three questionnaires was voluntary.

Participants. The number of participants varied across questionnaires. This can be partly explained by the fact that throughout the semester, some students decided to postpone the exam or not attend the (nonobligatory) lecture and/or tutorials. Accordingly, there were three groups of participants in this study. The first group consisted of those students who attended the first lecture and filled out the first questionnaire. The second group consisted of those students who attended the last lecture and/or the last tutorial and filled out the second questionnaire. The third group consisted of all students who took the final exam and filled out the third questionnaire. Of course, the three groups overlap. We can assume that students taking the final exam and attending the final lecture also attended the first lecture. It is less likely that the opposite was true. Overall, the participation rate in all three questionnaires was high, given that between 2013 and 2017 on average 156 students were annually enrolled in the class.²¹

First questionnaire. For our first questionnaire, we had a sample of N =153 participants:²² 64 women (42%) aged between 18 and 60 years (Mage=23.87 years, SD=6.85), and 88 men (57.5%) aged between 18 and 45 years (Mage=22.69 years, SD=3.91). Nearly 65% of all students majored in philosophy (68.6% of all women and 61.4% of all men). Nearly 11% (10.9%) of all women and 19.3% of all men majored in a social or natural science. All others majored in some other program.

Second questionnaire. The turnout rate for the second questionnaire was N =98, 54.1% of them women and 45.9% of them men. Sixty-one percent of all students in this sample majored in philosophy. Students were aged between 19 and 61 years (Mage women=24.04,SD=6.34; Mage men=22.25,SD=2.32).

Third questionnaire. The turnout rate for the third questionnaire was N =129. More than 46% (46.5%) of those students majored in philosophy, a total of 37.2% being women and 44.2% being men. More than 18% (18.6%) of the students did not reveal their gender in this questionnaire.

Using cross-tabulation analysis with gender as the independent variable, we explored potential differences among women and men students regarding the aforementioned indicators. For every statistical analysis minimum, alpha levels of .05 were established for contrasts; values equal to or lower than .05 were taken as significant.

To summarize our results: We found significant gender differences with respect to the Adversarial Argumentation Hypothesis. Women were significantly more afraid, less relaxed and generally less communicative during seminar discussions than men. While we do not know the causal factors for why women felt more afraid in seminar discussions than their male colleagues, we found that women felt less afraid and more relaxed in women-only environments. Our results also indicate on average a slightly better exam performance among those women participating in the women-only tutorial. We did not find significant gender differences in students’ perception of the discussion style as particularly competitive or aggressive. This result is consistent with Thompson et al.’s (2016) research, which also found no significant gender difference in the perception of seminar discussions as particularly aggressive or confrontational. Notably, however, in the women-only tutorial, women perceived the discussion-style as significantly less competitive and friendlier than women in mixed tutorials. Men also reported higher levels of discussion participation compared to women. At the same time, men reported having larger circles of acquaintances among their classmates in the tutorials.

While we know that women were more afraid and less communicative during seminar classes, we do not know why. Numerous mechanisms could underlie this result. Philosophy is often perceived as a male discipline (see also Baron et al. 2015; Calhoun 2015; Haslanger 2008),³¹ which is why women might experience a lower sense of belonging (M5) or see themselves confronted with a gender schema clash (M4). The deviant feelings of women as well as M4 and M5 could also be (partly) explained by implicit bias (M6) or stereotype threat (M7). However, another possibility could be that, for example, the presence of stereotype threat could explain why women do not feel that they belong in philosophy, which could manifest itself in feelings of fear and anxiety. Because all of these mechanisms could individually or jointly explain why women are more afraid and less likely to speak up in seminars, their possible presence should be studied further in studies of larger scale. Another mechanism underlying our finding could simply lie in a lower interest of women in the subject matter (M2). However, as we discuss in the next section, our results do not point in such a direction.

We did not find evidence for the Subject Matter Hypothesis or for the Impractical Subject Hypothesis, as our results do not show any gender differences regarding students’ interest (M7) in and their perceptions of the usefulness and relevance of the subject matter. Women reported even more often than men that they chose philosophy out of interest; overall, they were highly intrinsically motivated.

We also did not find evidence for the Formal Methods/Abstractness Hypothesis: both women and men equally perceived the course to be difficult. We also did not find gender differences regarding the degree of perceived difficulty of specific topics. All students reported that they found formal topics more difficult than non-formal topics. As such, our results do not show gender differences in perceptions of one’s abilities with respect to different philosophical topics (M3). Furthermore, we did not find gender differences regarding the perceived usefulness of formal methods in philosophy. On average, however, more men than women reported that they thought of mathematics as fascinating and fun and that they deeply enjoyed doing it. Also, fewer men than women reported that they avoided math-related issues. This indicates that there was a gender difference in attitudes in our cohort, not towards philosophy (M2), but towards mathematical methods.

Relatedly, we found significant gender differences in students’ mathematical skills. While women did on average not only perform worse than men on the math test, we also found significant gender differences in students’ willingness to try to solve a math problem. We can only speculate about the underlying mechanisms. Experienced stereotype threat (M7) could be a possible cause for lower performance and fewer attempts. M7 would also explain women perceiving math as less fun and their reported attitude of avoiding math-related issues. It could also be that women are simply less interested in mathematical problems (M2) and therefore dedicate less time and effort to each one of them. Our results thus point to the need to investigate in more extensive studies whether M7 or M2 operates and, if the latter, why women have a lower interest in the first place.

Regarding the Role Model Hypothesis, our analysis shows that it was considerably easier for all students to name men as opposed to women philosophers. Given that almost no female role model on the professor level was present before the start of our study, we can confirm that the established fact of a lack of role models in philosophy also applies to students at LMU. One promising route for exploration would be to study possible correlations between a large exposure to role models and the willingness of women to continue their studies in philosophy and their aspirations for an academic career in philosophy. Our findings also suggest this route in that they point towards a difference in perceived chances for a successful academic career between those women who were able to name three or more women philosophers and women who were not able to name three women philosophers.

Generally, our results suggest that mechanisms other than those that we primarily explored are responsible for gender differences in seminar experience. We therefore suggest that mechanisms pointing to the social environment in which academic philosophy takes place should be studied to explain them. Several hypotheses formulated in the literature are possible candidates to explain our results about reported students’ feelings in seminars. First, the Sexist Mistreatment Hypothesis suggests that women become victims of sexist, sexually harassing or otherwise discriminatory behavior in their study environment (Baron et al. 2015). Second, it would be worthwhile exploring gender discrimination. Implicit bias is viewed as a specific variant of this mechanism, namely, that teachers and students hold negative implicit biases towards women as being less capable philosophers than men (Saul 2013). Being subject to such biases can be reflected in lower performance and discriminatory treatment by teachers and other students (Dougherty et al. 2015). The presence of stereotype threat is another mechanism potentially underlying some of our results, especially the differences in math performance and women’s lower active engagement in class. Those hypotheses suggest that female students’ feelings, which are triggered by experiences of discrimination or indirect discrimination effects during their first philosophy classes, decrease their willingness to major in philosophy.

Our results about gender differences in grades also point towards further exploring the Gender Differences in Grades Hypothesis. This hypothesis suggests that students tend to major in subjects in which they receive the best grades (Arcidiacono 2004). Several mechanisms can be thought of to explain grade differences. One mechanism is gender differences in interest (M2) (Thompson 2017). However, our results do not support such a difference. Mechanisms that relate performance to discriminatory behavior have also been said to explain grade differences. Again, women could possibly achieve lower grades because they are exposed to stereotype threat (Haslanger 2008; Thompson et al. 2016).³² Because gender discrimination is a highly complex phenomenon, it has to be studied in relation to different mechanisms.

While we did not directly explore any of those hypotheses and mechanisms, our results provide indirect evidence for one or more of them. Our study is exploratory and results are not generalizable. However, the aforementioned results open up possible avenues for future research that is dearly needed in Germany and elsewhere.

Our results indicate that attending women-only tutorials provided an experience for women that was overall more pleasurable and less stressful than attending mixed tutorials. Moreover, women in the women-only tutorial performed slightly better on the final exam. This suggests that the effect of women-only environments on the motivation and performance of women could be a route for mitigating such gender differences, which in turn would most likely motivate more women to remain in academic philosophy.

As a step in this direction, we suggest the creation of women-only learning spaces and events such as tutorials, colloquia, workshops, roundtables, or informal study gatherings. These suggestions are supported by empirical research in other fields. For example, Booth, Cardona-Sosa, and Nolen (2014) found that women in a first-year introductory economics course at the University of Essex were 7% more likely to pass exams when randomly assigned to women-only classes. Furthermore, a UCLA study of 6,000 female students found that those students who entered the university after attending women-only schools showed a higher academic self-confidence at the beginning of their university studies (Riggers-Piehl, Lim, & King 2018). Those results suggest that such measures help limit potential stereotype threat and increase the presence of role models on different levels.

Given that some women in our study reported that they did not engage in seminar discussions because they felt that they did not have enough time to think or were afraid to say something wrong, concrete steps can be taken in a seminar to foster women’s engagement in discussions. This could mean, for example, designing exercises in which students prepare something for themselves first, and, in general, establishing learning methods that enable students to think first with sufficient time before they have to speak up in class.

Lastly, we recommend the targeted support for and promotion of women philosophers at universities and particularly in the classroom in Germany as elsewhere. The inclusion of women can be achieved in a number of ways, such as, for example, by including more women on course syllabi, by inviting more women to talks and conferences, and by offering seminars in which the work of women is at the center. We suspect these measures would signal a learning environment to women in which their presence is natural and their contributions are appreciated and welcome.

Our study is the first systematic empirical investigation in the German-speaking context that explores some of the hypotheses regarding the origins of underrepresentation of women in philosophy at the student level. Besides providing results about existing gender differences and their underlying mechanisms among philosophy students in the classroom, the contributions of our study are also methodological. It is the first study that implements a design allowing for comparisons of, for example, feelings and performance of women in women-only study environments and mixed study environments to better understand how the social context potentially contributes to drop-out rates and, more generally, women’s underrepresentation. Furthermore, using a panel design with multiple questionnaires, including a mathematics test to check for students’ background knowledge, for the same group is equally unique. Finally, the specific combination of interests, abilities, beliefs, attitudes, perceptions, and goals has also not been focused on before. This innovative design is part of the exploratory aspect of the study and can function as an exemplary design for future studies in other countries and/or on a larger scale in Germany to frame and make comparable future research in this area.

Despite its contributions, our study has limitations. First, it was carried out on a small scale. More research at other German institutions would be necessary to capture the actual situation of philosophy in Germany and enable local generalizations. Research in other cultural and geographical contexts is equally needed to enable cross-country comparisons and further support generalizations. Moreover, we did not separate causes for gender differences that originate in the pre-university period from those that originate in the period when students begin their studies.³³

Second, our focus was on exploring factors internal to philosophy and/or women that may explain gender differences among students. We particularly looked at a student’s interests, abilities, beliefs, attitudes, perceptions, and goals with respect to philosophy as an academic discipline, while ignoring factors such as gender discrimination and/or sexism. Research suggests that such reasons might not always be apparent to the women themselves and are thus not (or only partly) reported (Saul 2013: 42; Brennan 2013). They tend to become harmful only when they accumulate and are therefore not easily detectable but instead might lead to an intangible feeling of not belonging or wanting to do something else. As such, they require an extensive and sophisticated design, which might also partly be based on qualitative methods.

Third, as the different indicators that we included in the questionnaire were part of different hypotheses, it might be worthwhile designing specific surveys to gather more in-depth information about individual hypotheses.

Fourth, we lack evidence for how much our results are unique to students of philosophy compared to students in other disciplines, such as in STEM fields. Study results about the factors that lead women to drop out of STEM fields in Germany would be particularly helpful with regard to whether women’s underrepresentation has similar causes in those fields. Furthermore, it would allow for a discussion about whether remedial measures should be applied across disciplines or even at the university level, or be tailored specifically to the context of academic philosophy.

Finally, we want to address the worry that a self-selection bias might have partly driven the results when comparing the two different study environments (i.e., the mixed and the women-only tutorial), in that a particular set of women may have chosen to attend the women-only tutorial in the first place. This possible bias could go as follows: If there were pre-existing systematic differences in the groups regarding their feeling of fear, the women-only environment attracted women who were generally more afraid on average in a seminar environment than their colleagues (men and women) and they chose the tutorial to escape such a situation. If such self-selection occurred, we would consequently expect the women in the women-only tutorial to report feeling more afraid than women in the mixed tutorials. As our results state the opposite, namely, that women in the women-only tutorial felt less afraid than those in mixed tutorials, we assume that self-selection bias did not compromise this result.

In this paper, we studied the underrepresentation of women in philosophy at the undergraduate level Germany with a focus on five hypotheses currently discussed in the literature. We explored these hypotheses by analyzing factors related to the interests, abilities, beliefs, attitudes, perceptions, and goals of students in the context of their studies in a semester-long philosophy of science course at LMU, a major German university. Furthermore, we compared women-only learning environments with mixed learning environments and analyzed them with respect to students’ feelings and performance differences. While our results could not support most of the proposed hypotheses explaining dropout rates, we found some gender differences in attitudes, perceptions and feelings in tutorials, which varied when we compared mixed-gender and women-only study environments. Our results suggest that significant factors leading to an underrepresentation in philosophy are to be sought in the social and institutional environment within which philosophy is taught. Finally, our results point towards the potential usefulness of women-only learning environments that should be implemented parallel to mixed learning environments. Given that we present an exploratory study, we see our results as opening avenues for future hypothesis development and for a systematic study of such hypotheses on a large scale to empirically support and justify concrete measures that encourage women to stay in philosophy.

With our study, we want to encourage researchers to conduct studies in different geographical and educational contexts. While research has shown that dropout rates of women are particularly high (Paxton et al. 2012; Thompson et al. 2016), exploring reasons why women leave philosophy at all levels is important. High dropout rates of more advanced female philosophers also mean a loss of role models for younger generations. A survey of former students who have already left the field could further reveal why women turn away from philosophy. However, while this step might seem like the most obvious, one must remain cautious. As causal factors resulting in dropout rates are complex and interrelated, it cannot be expected that students who quit their studies are necessarily able to identify the exact reasons. We therefore think that it is particularly promising to study students’ perceptions, attitudes, preferences, interests, feelings and experiences in and with regard to philosophy as extensively as possible and thereby create a realistic picture of female students’ reasons for dropping out of philosophy