“The question is not why I don’t work in a maths department; the question is why should I?” Women mathematicians’ experiences of power relations and gender symbols during their PhD

ABSTRACT

This paper aims to investigate the reasons some female mathematicians give to justify their choice to not work in academia after finishing their doctoral studies. Nine female mathematicians who finished a PhD in Sweden answered a written questionnaire. Through collective narrative analysis, two main tracks were identified. One narrative described the struggle with self-identity in a gendered structure which included implicit power, while the other was more positive about exposure to discrimination, and highlighted the desire to work with applied mathematics. Through deductive thematic analysis, the results show that the main obstacle raised was the difficulty of getting a job in academia after their doctoral studies, especially permanent positions, without support. Compared to previous research, the lack of family-oriented political policies was not considered a main problem. Instead, the reasons provided by the respondents are structural problems, such as access to post-doc positions, and the stress of having to get research grants, as well as cultural aspects within the structure, including implicit and explicit use of power.

KEYWORDS:

• Gender
• higher education
• PhD
• women mathematicians

1 Introduction

Reports and research studies have in recent decades noted that in undergraduate studies, in many countries and in most subjects, women are in the majority (e.g. OECD, 2015; Reuben, Sapienza, & Zingales, 2014; UNESCO, 2007). This is a global trend and is valid for students at graduate level as well, for instance, in 2012, the OECD average was 47% female doctoral (or equivalent graduates) with the EU21¹ average at 48% (OECD, 2015). At the same time, there is also evidence of a powerful, gendered structure within higher education with respect to profession – one third of the world’s researchers are women and the numbers of women full professors and grade A academic staff are much lower (European Commission, 2009; UNESCO, 2007). It appears that the problem is not attracting women to higher education but women disappearing, especially between PhD level and further academic careers. An example of this potential ‘leak’ is in post doc positions, a position that affects opportunities for advancing in academia (Nerad & Cerny, 1999; Ooms, Werker, & Hopp, 2019). With fewer women getting post docs, fewer can move on to senior positions. This is true for Sweden too (Isopahkala-Bouret et al., 2018; Lindberg, Riis, & Silander, 2011; Sumpter & Sumpter, 2021), an interesting case since this is a country with a reputation for being ahead of the rest in terms of gender equality (Borchorst & Siim, 2008; Weiner, 2005).

A segregated labour market in academia is not just about the type of profession but also in which area/subject people are working. Here, the focus is on mathematics. In many countries, such as western countries like the US, UK, and Sweden (Brandell, 2008; Burton, 2004; Herzig, 2004; Lindberg et al., 2011; Piatek-Jimenez, 2015; Sumpter & Sumpter, 2021), and eastern countries such as Malaysia and Taiwan (Chang & ChangTzeng, 2020; Goy et al., 2018), mathematics is considered a male domain. However, when the gender gap has starting to decrease in countries such as Malaysia (Goy et al., 2018), the situation in mathematics in Sweden is different: all subjects, science included, show a better track record for the last five decades (Lindberg et al., 2011; Sumpter & Sumpter, 2021). At the same time, the girl-boy ratio in the Natural Science programme, the most mathematically intense programme at upper secondary school level in Sweden, is around 50:50 (e.g. Brandell, 2008; Mellén, 2021). In addition, different areas within mathematics show variations – some areas have more women full professors than others, such as mathematical statistics versus pure mathematics (Wedege, 2011). Another example illustrating mathematics as a male domain is the optional courses at upper secondary school level, where the choices appear to reinforce traditional gender symbolism (Mellén, 2021), and where girls and boys in accelerated mathematics classes describe their experiences of school mathematics, their stories differ (Szabo, 2017). Such gender division – between subjects (e.g. Sumpter & Sumpter, 2021) and within a subject (e.g. Szabo, 2017) – is, according to Connell (2006), a result of gendered institutions. These gender divisions include relations of power and symbols, and it is relevant to understand how individuals experience them (e.g. Foyn, Solomon, & Braathe, 2018; Wedege, 2011). The aim here is to explore, one, how a group of women mathematicians experienced mathematics as a gendered institution during their Phd, and two, what they understood to be the reasoning for changing career direction. The research questions are: (1) How are support/lack of support and obstacles expressed in some women mathematicians’ stories about their PhD? (2) What reasons do female mathematicians give for leaving mathematics as an academic profession? This is a qualitative study describing some female mathematicians’ voices with no aim to generalise nor to capture all their experiences.

2 Background

The background is divided into two sections – first a presentation of gender as a theoretical concept, and second an overview of research on gender and academia.

2.1 Gender

A common starting point for understanding gender is as a social construction more than a consequence of a biological sex (e.g. Acker, 2006). Here, we follow Connell’s (2006) definition, where gender is regarded as

“a pattern of social relations in which the positions of women and men are defined, the cultural meanings of being a man and a woman are negotiated, and their trajectories through life are mapped out”. (Connell, 2006, p. 839)

These patterns are the result of dynamic processes (Damarin & Erchick, 2010), creating gender structures that support inequality through gendered beliefs and identities (Acker, 2006). The concept of gender can be divided into distinct aspects or dimensions, such as gender division of labour, gender relations of power, emotion and human relations, and gender culture and symbolism (Connell, 2006). In this paper, we are interested in what is expressed in relation to and attributed to these relations of power, including emotions and different gender symbols. Gender division of labour captures aspects such as the number of female PhD students in mathematics in Sweden in relation to male students (e.g. Sumpter & Sumpter, 2021) or the percentage who pursue academic careers (e.g. Lindberg et al., 2011). Symbolism, according to Connell (2006), exists in images of femininity and masculinity. One example of this is the idea that women “are clearly irrational, illogical and too close to their emotions to be good at mathematics. Or so the story goes”. (Walkerdine, 1998, p. 15). Symbols and discourses are attributed to a specific gender, thereby creating norms and trajectories that tell us what is normal and what is deviant, for instance, the idea of mathematics as a male domain (Brandell, 2008) or how girls/women and boys/men are regarded as preschool children, as mathematics students (Forgasz, Leder & Tan, 2014; Edström, 2014; Frid, Sumpter, & Nortvedt, 2021; Sumpter, 2012; Walkerdine, 1998), or as researchers (Ahlqvist, Andersson, Söderqvist, & Tumpane, 2015).

Several researchers have focused on how different individuals perceive this culture with its symbols; for instance, Burton (2004) reported on female professional mathematicians’ descriptions of what it is like to work in a mathematics department or Mendick’s (2002) study of girls’ views of themselves as mathematics students. Others have looked at how a person develops gender within a structure (e.g. Walls, 2010). Another example is Leder’s (2010) study of women who had been very talented in mathematics at secondary school but had chosen not to become mathematicians. The following quote illustrates the experience of not fitting in to the created norm:

An advantage of being male would be to have been more encouraged to pursue a career in mathematics/engineering/technology. I would also have fitted in at high school better than I did—my Years 9 and 10 were spent on an all-girls campus where it was supremely uncool to be good at maths and science. (Leder, 2010, p. 453)

Similar experiences are shared in the narratives of three Norwegian girls, especially Anna who continued with advanced mathematics (Foyn et al., 2018). There is a tension between being good at mathematics and what that entails with regard to social norms and it appears that there is a social cost to being a ‘clever girl’. In a study of women undergraduate students in mathematics, they were asked to work with their identities, including their self-concept as ‘a woman in mathematics’ and how they talked about themselves and their situations (Solomon, 2012). The participants reported having to adapt to fit the prevailing norms. It should be noted that there need not necessarily be overlap between the way a person experiences themselves within a power-relation structure and the symbols of that power relation structure. In fact, the relationship may be contradictory. For instance, both girls and boys in upper secondary school consider girls to be more insecure about mathematics but when asked directly, more boys than girls express insecurity (Sumpter, 2012).

Research has generally shown that to survive in a male domain, women develop several coping mechanisms (Husu, 2005; Piatek-Jimenez, 2015; Walls, 2010). This is particularly true in areas where what is considered female or a female identity is associated with several negative symbols. As a woman, one must separate what others attribute to you from one’s personal identity (Volman & Ten Dam, 1998). Being one of the few entails constantly being in the spotlight – one is not just representing oneself also all members of one’s minority, here women – one is the token (Kanter, 1977). The different dimensions interrelate and one example of this is how women face implicit stereotypes when ranked for mathematical performance (Reuben et al., 2014). The symbols attributed to women result in women being ranked lower for employment. Such behaviour could be described as part of a norm-controlled self-selection (Lindberg et al., 2011). Even before an interview, implicit stereotyping can be part of such a decision. A recent Swedish study, which looked at whether there was a difference between whether recruiters called men or women applicants for an interview, found that male recruiters were more likely to call male applicants (Erlandsson, 2019). This included professions with a gender ratio between 40–60%, where one might expect more gender neutrality. The concept of homosociality (Lipman-Blumen, 1976) can be used to illustrate such a filter. In PhD studies in mathematics, male professors pick male PhD students similar to themselves, and male students would study mathematics since it is a ‘good’ environment (e.g. Angervall, Beach, & Gustafsson, 2015). Taken together, the symbols, self-concept, and interactions all shape the structure which then generates the norms.

2.2 Gender and academia

In a review that sought to explain US women’s underrepresentation in STEM, Wang and Degol (2017) concluded that there are six explanations, some more valid than others: (a) absolute ability differences, (b) relative ability strengths, (c) career preferences, (d) lifestyle preferences, (e) field-specific ability beliefs, and (f) gender stereotypes and bias. They suggested that socio-cultural factors are more likely to have greater influence than biology, emphasising the importance of how power is connected to cultural and structural factors. Looking at higher education as a gendered institution, several factors have been identified as the explicit and/or implicit exercise of power (Husu, 2013). One example of an explicit factor is how research grants and other funding are distributed (Ahlqvist et al., 2013, 2015; Lindberg et al., 2011). These can be relatively easily measured. Implicit use of power is, however, more veiled and could be described in terms of “non-events”, such as being ignored at meetings or in the coffee room, or not being invited to conferences or selected as keynote speakers (Husu, 2013). Studies have shown that women mathematicians face this use of power in many different ways (Burton, 2004; Wang & Degol, 2017); the two main factors behind women’s struggles to advance in their academic careers being lack of support and discrimination (Heilbronner, 2013; Henrion, 1997; Husu, 2005; Piatek-Jimenez, 2015; Xu, 2008). An example of the latter is the so-called ‘motherhood penalty’ – women, but not men, are punished for having children by, for instance, being regarded as less present (Correll, Benard, & Paik, 2007). However, research looking at countries such as Finland and Japan also indicates that motherhood might not explain the academic gap with respect to research productivity and that other factors appear to have stronger impact (Aiston & Jung, 2015).

One key factor seems to be support. Female researchers and graduate students in mathematics as well as other subjects, often mention that the presence of professors and colleagues of the same sex is important (Henrion, 1997; Husu, 2005; Leder, 1995; Piatek-Jimenez, 2015). In addition, a recent study of data from Germany and the Netherlands signals that a mentorship system has a greater impact on early career positions in academia (post doc) compared to senior positions (e.g. from associate to full professor [Ooms et al., 2019]). This implies that the support of a mentor is more important early in an academic career. The importance of support is one of the main findings of a study by Angervall et al. (2015). When asking how people got into research and succeeded in their careers, one pattern of reply related to being one of ‘the chosen, privileged, and fortunate few’. Young men were more likely to be part of networks and to get positions, whereas women, especially those slightly older and with a long professional background, did not share this experience. At the same time, improvement of the institutional culture is vital if reaching a balanced gender division within a department and if integration of women into a faculty is desired (Hill, Corbett, & St. Rose, 2010; Xu, 2008). It is plausible to believe that it would be more problematic to pursue a career in research if one was less integrated into the department and in scientific networks, especially when the integration itself is subjected to a dynamic process with gender patterns, and takes place within a gendered structure (Acker, 2006; Damarin & Erchick, 2010). Another concept related to this self-selection is ‘the ideal worker’, first presented by Kanter (1977), who showed how organisations created routines and work practices based on the idea that an employee does not have commitments or responsibilities other than work. Budig (2002) extended this reasoning:

“women are not disadvantaged simply because they lack work experience, seniority, or other human capital. Instead, or in addition, women are disadvantaged because the typical woman does not fit the disembodied category of the ideal worker” (Budig, 2002, p. 261).

Hence, women cannot be seen as mathematicians simply because they are ‘non-male’.

3 Methods

The context of this study is Sweden, which is often seen as country with a long tradition of active gender equality work (Borchorst & Siim, 2008; Lindvert, 2002; Weiner, 2005). This makes it an interesting case. Swedish gender equality policy has the following starting point:

“Sweden’s overarching objective of gender equality policy is for women and men to have the same power to shape society and their own lives.” (www.government.se)

This is the cornerstone of the Discrimination Act, including sub-goals such as equal distribution of power and influence, and equal education. The governmental criterion for equality in Sweden is a minimum of 40/60, meaning that the representation in a workplace of each gender should be in the span 40–60%. Such division is regarded as balanced (Kanter, 1977) and should be valid for higher education too (Ahlqvist et al., 2015; Lindberg et al., 2011; Sumpter & Sumpter, 2021). Several political decisions in the last four decades have had the aim of supporting family life, such as extensive parental leave and substantial nursery care (Hedlin, 2013). They stem from the idea that women should participate equally in work life, and subject to the same conditions as men. This was established officially in the 1970s (Lindvert, 2002) but the idea of equal participation had already appeared in the school curriculum in 1969 (Hedlin, 2013).

3.1 Methods of data collection

A written questionnaire was sent to nine female mathematicians who had graduated from six Swedish universities (of different sizes, spanning the length of the country). This was a case of purposive sampling based on the criterion that they had finished their PhD in mathematics during the years 2002–2012. Mathematics is here interpreted as mathematical sciences such as pure and applied mathematics, mathematical statistics, and computational mathematics and optimisation but the decision was to exclude mathematics history and mathematics education. Because of the relatively small Swedish population, it was difficult to avoid the possibility of connections being drawn between the respondents and the author – of the nine mathematicians participating, the author knows four and the other five were found through a mutual contact or the Swedish network, Women and Mathematics, a sub-organisation of IOWME (the International Organization of Women and Mathematics Education). However, the assumption is that the different connections did not affect the quality of the replies since the answers to the questionnaire were kept anonymous throughout the research process, and the respondents were informed of this. The questionnaire had four main questions: (1) Why did you become a mathematician/Why mathematics? (2) Why did you do a PhD in mathematics? (3) How was your experience as a (female) PhD student in mathematics? (4) You have a career outside the university. How did that come about? Several optional sub-questions were listed to each of these four questions, each with the aim to gather more information. The respondents were informed that they could write as much or as little as they wanted. They were also encouraged to write something about their background. Their responses to the questionnaire were returned within two weeks.

3.2 Method of analysis

A preliminary inductive analysis was conducted to look for patterns yielded by the responses (e.g. Braun & Clarke, 2006). The analysis showed that four of the respondents (from three different universities) had similar answers, describing a more positive experience, whereas the other five (from five different universities) shared a more negative view. These two sets of replies constituted the basis for the second analysis which was the creation of two collective narratives. The main reason for choosing a collective narrative format was the request from several respondents to keep them as anonymous as possible given that they came from a small community to start with (female mathematicians in Sweden) and that the study specifically focused on a specific sub-group of this small community (female mathematicians in Sweden who had decided to leave academia after their PhD studies). In a collective narrative, one story is created by interweaving the (written) answers from several respondents which works as a tool to emphasise the meaning behind a collection of responses (patterns) rather than to specify individual replies (Mendick, 2002). In that sense, collective narrative analysis shares similar goals to content analysis when aiming to identify the traits of the material (Smith, 2000). In this study, the responses of the two sets of respondents are summarised into the two separate fictive voices, Anna and Sarah. These narratives have been published separately (Sumpter, 2014a, 2014b), but have never been compared. Such a comparison can highlight not only similarities and differences between the narratives, but also nuances within each narrative. In narrative analysis, the author’s voice is part of the story (Smith, 2000) but here it has been minimised as much as possible by using the respondents’ own formulations. In some cases, the formulations have been combined into one sentence and when needed the context has been changed (e.g. seminars have become lessons) to ensure that the specific person/situation cannot be identified. In either of these cases, the interpreted meaning of the replies remains the same. The respondents were given the opportunity to read ‘their’ story which functioned as a triangulation (Madill, Jordan, & Shirley, 2000). One respondent accepted this offer and concluded that the story as a whole, and the meaning, described her well even though some details did not match her experience.

The narratives were then broken down into data points that were compared using the themes generated from previous research in a deductive thematic analysis (Braun & Clarke, 2006). The first theme was support/lack of support (Heilbronner, 2013; Henrion, 1997; Husu, 2005; Leder, 1995; Xu, 2008). When responses explicitly mentioned some kind of support or lack of support, the entities were marked and coded accordingly. The second theme was tokenism, as presented by Kanter (1977), here looking for data points where respondents described experiences that represented not only themselves but all women. The third theme was obstacles (Correll et al., 2007; Hill et al., 2010; Xu, 2008) and was divided into several sub-themes such as motherhood, ranking when applying for jobs/grants, etc. The last theme was about reasons for leaving academia, an overarching theme that functioned as a validation of those themes that were identified as explanations for why respondents chose not to remain in academia.

4 Results

Both Anna and Sarah started their PhDs almost directly after their undergraduate studies in mathematics/applied mathematics in combination with engineering/physics/statistics/computer science (the five responses creating Anna’s story) or mathematics/applied mathematics/mathematics statistics in combination with engineering/physics/statistics (the four responses creating Sarah’s story). Anna had had children during her PhD and took parental leave. She finished her PhD around the age of 35. Sarah did not have children and therefore finished her PhD when she was somewhat younger (between 30 and 35). For Sarah, the step to doing a PhD was straightforward. They got a PhD position at the same department where they did their undergraduate studies, therefore remaining in the same working environment. Anna had also worked in her department as an amanuensis.² At the time the study was conducted, Sarah and Anna were both working as mathematicians/researchers in private corporations or in council/governmental institutions.

The first theme comprises the expressed instances of support or experiences of lack of support, including subthemes, see Table 1:

Table 1. Expressed support or lack of support.

	Anna	Sarah
Support
Individual		Encouragement from her supervisor (male) for final essay to apply for a PhD.
Collective		Several female PhD students in her group. Collaboration with other research groups where women were working.
Lack of support
Individual	Discouraged from pursuing a PhD where teacher said she was not clever enough.	Some but not all expressed that they were not clever enough. The support was greater.
Collective	No help from supervisor or other people after PhD – all efforts to try to get positions or grants had to be done alone.

As we can see in Table 1, Anna’s and Sarah’s responses differ regarding the explicit support they experienced both before, during, and after the PhD, both regarding individual support and collective support. Anna said that she had been alone, while Sarah described a community. Sarah was in a group with several women and had many contacts with other research groups where several women were working, so although there were some individuals that made comments that could be construed as lack of support, being part of a group compensated for this. When analysing the data with regard to respondents feeling like one of the few, both shared the following experience albeit from two different angles, see Table 2:

Table 2. Expressed experiences about being the token.

	Anna	Sarah
Tokenism	Often you were the only woman (or one of few). Most people were positive and friendly, but it was like an underlying structure – one should not think that they were as good as or as clever as the male students.	When you study, there are so few students, so you get both good contacts with the other students and with the teachers, as well as a look into the world of academia.

Looking at Table 2, we see that Sarah had a good experience being one of the few during undergraduate studies, leaving with the feeling of having good contacts with the mathematics department, but Anna described a different experience. The women who constituted Anna’s voice all reported experiencing being confronted with an underlying structure when they faced obstacles, which they illustrated with several examples (see Table 3). The sub-categories are grants/positions, symbols, use of power, parenthood, and pregnancy.

Table 3. Expressed obstacles.

Obstacles	Anna	Sarah
Grants/ Positions	After PhD, started to apply for different positions which she didn’t get. Once, they gave a position to a man who was not as qualified as she was, explaining that his work was much more “developed”. Do not miss the working climate and the stress of applying for grants you are most likely to not get. Didn’t like the competition, a focus on everyone should be ‘best’, and the lack of common goals to strive for.	Employment at the university insecure. It was hard to get a permanent position and you had to go abroad to do a post doc, which conflicted with starting a family. In academia, you are forced to apply for grants, which you are most likely to not get.
Symbols	Women professor were regarded as having “sharp elbows”. Explicitly told that there was no point with women PhD students since they should end up next to the stove anyway.	No data points about symbols.
Implicit/explicit use of power	In a course, the professor ignored her, would not answer her questions. In seminars, when raising critique, she was told not to be so troublesome. Male students were instead praised for their ability to scrutinise and for flexible thinking. Several occasions when the department actively tried to block initiatives to promote gender equality.	Felt she was listened to during meetings. She was the PhD representative on the department board and they took her comments seriously.
Parenthood and pregnancy	During the first years of being a parent, seminars were scheduled late in the afternoon, and she could not attend since she had to pick up children from preschool. Leadership/lecturers/professors showed no sensitivity to her situation”? Hard to focus when you are tired, maths is a subject where you need to be focused the whole time. Tough coming back from parental leave, work felt so distant. Lost time and energy, costed a few months to get back on track.	Didn’t have any children while a PhD student. During the first years, parenthood was not on the agenda, and later found it hard to be on parental leave.

Anna describes a struggle and lists several incidents and some unique events, but also experiences that were repeated, such as being ignored in seminars. This is presented both from a personal but also general perspective. In Anna’s story, the department knew about the problems, but all initiatives aimed at addressing these challenges were discouraged and blocked in subtle and non-specific ways so it was not possible to object. The result was that no improvements were ever made. During her PhD, Anna fell pregnant, had children, took parental leave, and then returned to work, trying to combine parenthood with a job. Two issues stand out in Table 3 regarding parenthood and pregnancy. First, the example of the exercising of implicit power – the scheduling of seminars late in the afternoon so it collided with family life. The second, how motherhood impacts the ability to keep up with one’s PhD studies and research work. This should be viewed in light of the fact that in Sweden you are employed by a university department for 48 months to do your PhD. Despite this, Anna concluded that becoming a mother had cost her research time and this understanding was part of Sarah’s decision not to have children as it would negatively affect her studies.

Examining their reasons for leaving academia, it appears that the choice was easy for both Sarah and Anna. See Table 4.

Table 4. Reasons for leaving academia presented with respect to themes.

	Anna	Sarah
Support/ Lack of support
Individual	Alone: can’t get a position without support. Supervisor(s) didn’t help.	Hard to get a permanent position [at the university].
Collective	Applied for lots of positions, didn’t get them. Didn’t have network to work for her, to argue for her case. In current workplace, well taken care of. They appreciate her work. Clear plan of how she will develop. Work in teams with the same goals. The sum is greater than the parts. In maths department, you sit in your office behind closed doors.	In Sweden, there is not much encouragement in researchers changing between the university world and outside academia. Few mathematicians work with other subjects (e.g. Biology and Medicine). Researchers (in other subjects) struggle with models they don’t understand. Maths researchers should be encouraged to work in other subjects, to collaborate with other people.
Tokenism	No data	No data
Discrimination/Lack of discrimination	At current workplace, not pushed to work overtime, respect working hours.	Employment at the university felt insecure, especially the system with Post doc.
The nature of the subject/work	The work is interesting. Lots of feedback, push each other forward. Product is useful for society.	Didn’t want to stay in academia and there was no plan to stay since she wanted to apply her knowledge. Work is joyful – get to solve problems, which is the most fun part of research, but don’t have to pursue my own research project. The work is much more meaningful.

Anna comes back to the issue of support, or more specifically, the lack of support when trying to get a position in a mathematics department. Anna and Sarah agree that their current work and workplace are better: they use words such as “meaningful” and “appreciation”. Both Anna and Sarah, representing all nine respondents, emphasise the stress of applying for and getting research grants as well the insecurity of work contracts. Sarah expands her reasoning by talking about academia being narrow-minded regarding cross-subject research and applications of mathematics. In sum, Anna and Sarah are unanimous in their replies about why they left academia. The following quote summarises the situation: “The question is not why I don’t work in a maths department; the question is why should I”.

5 Discussion

The aim of this paper is to describe how a group of women mathematicians experienced mathematics during their PhD, and what they understood to be the reasons for changing career direction. Women in academia face both explicit and implicit power where the latter includes “non-events” (Husu, 2013). In the responses creating Anna’s story, there are several examples of the subtlety of the resistance women encounter; the term “constant headwind” being used as an illustration. Almost all respondents were at some point told that they were not good enough. Being a clever girl in mathematics was not always easy (e.g. Foyn et al., 2018; Szabo, 2017). The success of various efforts, made by the different mathematics departments, to change the situation were dependent on the attitudes of people. It was not related to regulations or policy documents. Previous research has shown that improving institutional culture is vital for women staff to be successfully integrated (Hill et al., 2010; Piatek-Jimenez, 2015; Xu, 2008), and the results from the present study suggest that it is not just a case of absent policies although these policies represent socially accepted principles (see Borchorst and Siim (2008) for a longer discussion on the public–private split). The situation where caring for children results in being unable to give adequate time to one’s PhD studies because seminars are scheduled for the late afternoon is an example of family-protection policies failing to address variations of the ‘motherhood penalty’ (e.g. Correll et al., 2007). The present results are therefore an addition to previous research.

Being a woman in a male domain can lead to a struggle to preserve ones self-concept in an environment where one is pressured merely for being the ‘wrong’ sex (Budig, 2002; Kanter, 1977; Leder, 2010; Reuben et al., 2014; Solomon, Radovic, & Black, 2016; Szabo, 2017; Walls, 2010). There are several examples in Anna’s story where the respondents talk about themselves in relation to the context, about how to maintain self-concept and prevent the environment from affecting them negatively. This could indicate the presence of a coping mechanism (Husu, 2005; Volman & Ten Dam, 1998). There is also talk about changes, for instance, the view of work after having children. The negotiation of self-identity appears to be a dynamic process, an ongoing work. This finding is consistent with previous research (e.g. Foyn et al., 2018; Solomon, 2012; Solomon et al., 2016).

The main reason all respondents gave for leaving academia was the lack of permanent positions. These included the types of job, short-term contracts, and the uncertainty of getting them. This obstacle has been recognised as a main filter for women scientists (Husu, 2005; Ooms et al., 2019). None of the respondents mentioned the size of the salary; it was knowing that they would have a salary (at all) that mattered. In addition, if one wants to pursue a career and at the same time start and/or provide for a family, such work contracts are not an option. Also, only one respondent explicitly said she would have stayed if she could have which is reflected in the last category in Table 4 – The nature of the subject/work. There is something else outside academia which could offer different conditions (e.g. Wang & Degol, 2017). This topic, the nature of the subject, is a suggestion for further studies: respondents get to speak freely about ones work as contextual, here away from academia and located in a new context. Such study would illuminate other factors than present study given the deductive set up of the analysis.

In this study, post doc positions were raised as an obstacle, especially if one does not want to leave the country. Post-doc positions are often used to boost academic careers (Nerad & Cerny, 1999) and thereby function as a form of norm-controlled self-selection resulting in fewer women on the career ladder (Lindberg et al., 2011). Related to this, is the difficulty of getting a job in contexts where one lacks a conducive network. This challenge has been identified as a major obstacle for women’s advancement in academia both in mathematics and other STEM subjects (Heilbronner, 2013; Henrion, 1997; Leder, 1995; Xu, 2008) as well as more generally (Angervall et al., 2015; Husu, 2005). The stress of grants (applying and not getting them) is another obstacle (Ahlqvist et al., 2013, 2015; Lindberg et al., 2011). This is a structural problem which has been reported as a general problem for female researchers in STEM subjects (Xu, 2008) – if you want to pursue a career in research you need grants, and in order to get grants you need support, especially if the embodied you, what is attributed to you, is valued less highly (Ahlqvist et al., 2013, 2015; Budig, 2002; Volman & Ten Dam, 1998). Several researchers (e.g. Wang & Degol, 2017) highlight the need for active decisions that will lead tochanging this environment and culture, especially if such a structure appear to be static.

The results also indicate the absence of some obstacles in certain research groups. The four women that comprise Sarah’s story did their PhDs in mathematical sciences focusing on applications. The analysis reveals several potential factors that could have contributed to the positive experience. One is collaboration and the presence of various networks where several people provided support (e.g. Leder, 1995; Xu, 2008). There were also already several women in the research group (e.g. Henrion, 1997; Husu, 2005; Ooms et al., 2019). This raises the question about micro-climates, which would not be revealed in studies on gender patterns in bigger structures (Lindberg et al., 2011; Sumpter & Sumpter, 2021), but would be highly relevant when trying to understand what does and does not shape gendered institutions (Connell, 2006). Micro-climates has not been addressed in this study, but further research could study potential links between the view of mathematical sciences (including the ranking of different mathematical areas as so-called ‘pure’ and ‘not-pure’ [e.g. Ooms et al., 2019]) and variations in micro-climates from a gender perspective.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1. The 21 countries that are members of both the European Union and the OECD: Austria, Belgium, the Czech Republic, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Ireland, Italy, Luxembourg, the Netherlands, Poland, Portugal, the Slovak Republic, Slovenia, Spain, Sweden and the UK.

2. Amanuensis carries different meanings depending on the country; here it is teaching assistant.