Introduction

In the past few decades, girls and women have successfully increased their educational participation: girls constitute a higher share of the student body in upper secondary schools than boys (DiPrete & Buchmann, 2013), they obtain better grades in high school than their male counterparts (Voyer & Voyer, 2014), and in higher education young women account for a higher share of entrants (Clancy & O’Sullivan, 2020) as well as graduates (Lörz & Mühleck, 2019) than young men. Up until high school, girls outperform boys in terms of grades (Burusic et al., 2012), yet some research shows that this gender difference is reversed at university, at least in some fields (Francesconi & Parey, 2018). Understanding women’s lost grade advantage from high school to university is of substantial importance. As university grades have a high signalling value for subsequent income and employment (Røberg & Helland, 2017; Zou et al., 2022), women’s worse chances of translating their advantage in school grades into an advantage in university grades might contribute to the gender wage gap, and thus gender inequalities on the labour market. Therefore, it is crucial to identify the factors that might hinder young women’s ability to achieve their full potential.

Given that these factors were not examined in the previous studies, our study (i) scrutinises the existence of a gendered gap between high school grades and university grades for a broad set of fields of study, and (ii) investigates perceived discrimination against women and perceived competition among students, as two potential factors associated with university grades, independent of high school grades in Germany. Our analysis is conducted separately for six popular fields of study that differ in their social context, namely law, medicine, economics, engineering, natural sciences, and social sciences. First, law and medicine, as fields that have been described as leading to elite professions, are characterised by a high degree of social closure within higher socio-economic classes and a high degree of competitiveness: for medicine, this is especially the case in regard to entrance to study programmes; for law, this applies throughout the programme (and subsequent career). The large inflow of women into these fields (German Federal Statistical Office, 2022; Mann & DiPrete, 2013) has not altered the elite character of these professions (Strømme & Hansen, 2017). Second, the field of economics has likewise seen an increase in the share of female students: women now account for up to 54% of students in the economics and management subfield, which can today be considered a gender-integrated field of study (Alon & DiPrete, 2015). Third, engineering subjects are characterised by a high share of male students and the climate in these fields has often been described as “chilly” for women (Hall & Sandler, 1992; Morris & Daniel, 2008). Fourth, in the natural sciences, the share of women varies tremendously between subject areas, with higher shares in biology and agricultural sciences and lower shares in physical sciences (Mann & DiPrete, 2013). Fifth, social sciences (which are often categorised as part of the humanities) are characterised as female-dominated (Hägglund & Lörz, 2020).

Theoretically, we base our argument on status characteristics theory, which claims that in most societies women (alongside those things that are considered “female”) are perceived as less competent and valuable than men (Ridgeway, 2011). We argue that this perception of women’s lower competence is more pronounced in social contexts that are considered to be “male”, either because of a strong dominance of men, such as in engineering, or because of a sex-typing of the field of study that has prevailed despite an increasing share of women studying in that area, such as in medicine and law (Torre, 2018). In male-dominated fields of study (but not fields of study that are sex-typed as male but that have become gender-integrated in terms of student numbers), this expectation is also supported by the theory of tokenism (Kanter, 1977). This theory argues that individuals who are in a minority position, such as women in male-dominated fields of study (e.g., engineering), are perceived by the majority—in this case by male students—not as individuals but as representatives of their group. This leads to several challenges: first, it makes “tokens” more visible; second, those who are dominant become more aware of what sets the tokens apart; and third, the tokens are not judged on their real behaviour but on the expected behaviour of the minority group. Thus, in the case of women studying male-dominated fields of study, women are stereotypically perceived to be less competent than men.

The German Context

We draw on a nation-wide German survey of higher education students that contains data collected from 1982 to 2013. Germany is characterized by a stratified secondary education, in which children (at the age of 10) are placed into three tracks: Hauptschule, Realschule, or Gymnasium. Only the last leads to direct higher education entrance qualifications (Abitur), while the other two offer mainly access to vocational training. After graduation at Gymnasium (Abitur, around age 18), students have the choice to pursue higher education at either traditional universities or universities of applied sciences (Fachhochschulen, that have a teaching focus on the practical application of knowledge in fields such as engineering, business studies, social work), pointing towards a horizontal differentiation of higher education rather than a vertical one (Leuze, 2011). What is important to note in Germany, moreover, is the international comparison low rate of students in higher education. This has been attributed to the attractive alternative of vocational educational training that is chosen not only by those who finish secondary school on the lower or medium track but also those who attain the entrance certificate for higher education (Abitur) (Powell & Solga, 2011). In terms of gender differences, it is important to emphasize that men—despite lower grades in secondary school—are still more likely to enter higher education as compared to their female counterparts (Lörz & Mühleck, 2019; Lörz et al., 2011). At the same time, women, compared to men, are more likely to choose the track that leads to the traditional university than the university of applied sciences (Uunk & Pratter, 2020). As in other countries, women are less likely to choose technical degrees compared to men (Lörz et al., 2011), even though the share of women in gender-atypical fields of study has increased over time in Germany (Galos & Strauss, 2023).

Gender Gap in Grades: What Do We Know?

Previous research has shown a difference between the gender gap in grades in primary and secondary school, as compared to those in tertiary education: in primary school, despite equal performance in standardised tests, girls outperform boys in terms of grades (Burusic et al., 2012). Girls in high school have been shown to have better reading grades than their male counterparts, a finding that has been attributed to girls’ better classroom behaviour (Downey & Vogt Yuan, 2005) and their stronger involvement in cultural activities, which aid their educational trajectory (Dumais, 2002). Some studies suggest that young girls have better maths grades than boys (Downey & Vogt Yuan, 2005), while others indicate that there is no significant difference in maths grades until later stages of high school, when boys have a slight advantage (Leahey & Guo, 2001). Therefore, while, generally, girls tend to have better grades in high school, this gender difference is slightly reversed at the university (Francesconi & Parey, 2018). However, there is heterogeneity in regard to this effect: girls who enrol in medicine and economics especially outperform boys in university grades, while this is not the case for girls enrolling in science, technology, engineering and mathematics (STEM) and humanities subjects. Likewise, men at graduation from higher education have better grades in STEM and humanities than their female counterparts (constant over time for STEM subjects, declining for humanities), yet, while the gap is not significant in economics, it is substantial and increasing in medicine (Francesconi & Parey, 2018).

Considering this change in the gender gap in grades from high school to tertiary education, the next natural step would be to consider to what extent high school grades predict university grades. Although better school grades have been shown to be a strong predictor of better university grades (Cyrenne & Chan, 2012; Giese, 2020; Vulperhorst et al., 2018), it is unclear if this holds true for young men and women alike. Moreover, there might be differences as to the predictive power of high school grades for university grades in different subject areas. Evidence from the area of law shows a lower correlation between school and university grades for men than for women (Hinz & Röhl, 2016). Since Author’s study is based on administrative data using results from written final exams, where examiners did not have access to information on the students’ gender, they exclude the possibility of direct discrimination. Instead, several possible mechanisms are discussed, such as positive discrimination against young women in high school that could be linked to better school grades. School grades might contain a norm compliance component and because young men are less norm compliant, their grades less reliably measure their academic potential compared to young women. In addition, gendered study motivation, gender-specific attrition rates and labour market perspectives adjusted to gender-specific career expectations might be associated with worse grades for women at university. However, these mechanisms could not be assessed empirically in Hinz & Röhl (2016) study. Based on these arguments, we expect that high school grades are more strongly associated with university grades for women than for men (Hypothesis 1a).

Since not all final examinations at the university are standardised to the same degree as those for law (Hinz & Röhl, 2016), it is still possible that processes of perceived discrimination against women play a role. Aditionally, women also have a lower preference for competition and tend to perform less well in competitive situations (Gneezy et al., 2003; Morin, 2015; Ors et al., 2013; Schram et al., 2019). Both factors might be related to a lower correlation between high school and university grades for young women as compared to young men. We thus alternatively expect that high school grades are more strongly associated with university grades for men than for women (Hypothesis 1b).

Moreover, from previous research it can be assumed that the relationship between school and university grades is not universal but differs between study contexts. Thus, the experience of studying varies between different fields of study: fields of study differ in their academic demands and standards (such as admission requirements, levels of scholastic intensity, grading policies and achievement norms) and also in the structure of the curriculum, class size, culture, the level of student integration, the availability of study groups, the accessibility of faculty, and the social climate (Alon & Gelbgiser, 2011; Marczuk, 2023). Additionally, the experience of studying differs by the gender composition of the field of study. Thus, male-dominated fields, like STEM fields, tend to have a larger number of compulsory and “barrier courses” in their curriculum, while female-dominated fields, like the humanities and social sciences, allow for greater freedom in course selection (Xie & Shauman, 2009). The gender segregation between fields of study thus contributes to women’s grade advantage in higher education (Alon & Gelbgiser, 2011). This finding does not, however, speak to the question of how far the predictive power of high school grades for university grades differs by gender. Women in gender-atypical fields of study who are perceived as “tokens” (minority), could be judged not on their individual performance but rather on the performance expectations regarding the social group (of women, in this case) to which they belong. This might lead to a biased perception of the performance of these minority groups.

While the argument on the situation of “tokens” is gender-neutral (Kanter, 1977), authors in the tradition of status characteristics theory argue that women and female-associated fields of study or occupations are culturally devalued (Ridgeway, 2011). Thus, the minority status of women in some fields of study might encourage male students (and teachers) in particular—as members of the majority group—to advantage the negative stereotypes that female (fellow) students are less competent than men (for a similar argument, see Meyer & Strauß, 2019). We therefore expect that women in male-dominated fields of study (such as engineering)Footnote 1 might find it more difficult to translate their good high school grades into good university grades. From this argument, we expect that in engineering, as a male-dominated field of study, relative to other fields, high school grades are less strongly associated with university grades for women than for men (Hypothesis 2).

When reflecting on the factors that might be associated with the better university grades achieved by men, compared to women, several arguments have been raised. It has been argued that the reversal in the gender grade gap from high school to university might be due to selection: since more men than women drop out of higher education, male graduates might be more positively selective, e.g., they are academically stronger (Francesconi & Parey, 2018). Further, without providing empirical evidence, it has been speculated that the academic curricula might fit men’s abilities better than women’s, or build men’s skills more efficiently than women’s (Francesconi & Parey, 2018).

In our own empirical analysis, we focus on two other potential factors that might be associated with worse university grades for women: perceived discrimination against women and perceived competition among students. While perceived discrimination might make women feel unfairly treated compared to their male peers, perceived competition emphasises the impression of studying in an environment in which students are eager to outperform each other.

Regarding discrimination, we expect that the negative stereotypes associated with women’s competence (Bolton & Muzio, 2008; Walton & Spencer, 2009) contribute to women’s higher likelihood of perceiving there to be discrimination against women, as compared to their male counterparts. If young women feel that they are subject to unfair treatment at universities because of their gender, this is likely to have negative consequences for their university grades. This phenomenon has been described as “stereotype threat”, which occurs when individuals’ performance is being judged and there is a possibility that the negative stereotype applied to their group may be proven true, leading to a negative impact on their performance (Spencer et al., 1999, 2016; Steele, 1997). Empirical evidence, including two meta-analyses of stereotype-threat studies, illustrates that individuals belonging to social groups associated with negative stereotypes related to performance often face the pressure to disprove these negative stereotypes, ultimately hindering their performance (Spencer et al., 2016; Walton & Spencer, 2009). Yet some studies have cautioned that stereotype threat might be more evident when tested in experimental scenarios (such as lab settings), as compared to real-world settings, and that the extent of the effect sizes should be subject to scrutiny (Inglis & O’Hagan, 2022; Shewach et al., 2019). If young women perceive themselves to be treated unfairly at university, in comparison to their male counterparts, particularly when negative stereotypes about their competence (Ridgeway, 2001) become salient (such as during exams or assessments), they may experience additional pressure to achieve high grades. This pressure can ultimately have a negative impact on their performance (Nosek et al., 2002; Walton & Cohen, 2003; Walton & Spencer, 2009). For example, a study carried out in the United States illustrated that stereotype threat accounts for 50 to 82% of the gender gap in SAT maths performance (Walton et al., 2013). Additionally, perceived discrimination might translate into worse academic performance for women, but not for men, as discrimination is presumed to be more harmful for individuals from the relatively disadvantaged group than for those from the relatively advantaged group (Schmitt & Branscombe, 2002). This might be the case as the stereotypes associated with the disadvantaged group are likely to be internalised by members of that group, leading them to devalue their own abilities and adopt the perceptions of other groups about themselves. Considering the potential detrimental outcomes of perceived discrimination for academic performance (measured in grades), we expect that perceived discrimination against women is correlated with lower university grades for women but not for men (Hypothesis 3).

When considering the relationship between competition and university grades, we know that boys/men perform better in competitive circumstances than girls/women (Gneezy et al., 2003; Morin, 2015; Ors et al., 2013), and that they are not as negatively affected by situations of stress (Cahlíková et al., 2020). Indeed, men have higher academic self-confidence, especially in male-dominated fields of study, such as STEM subjects (Litzler et al., 2014 who only find a gender gap for those of White ethnicity). Since women have been shown to perform less well in competitive situations (Cheryan, 2012; Smyth & Nosek, 2015), it can be expected that perceived competition negatively impacts their university grades. In other words, we expect that perceived competition is correlated with lower university grades (especially for women) (Hypothesis 4).

Table 1 provides an overview of the theoretical as well as the empirical implications of our study.

Table 1 Overview of hypotheses
Full size table

Data, Variables, and Method

Data

Our study uses 12 different waves of the cross-sectional German Student Survey (Studierendensurvey), a representative survey of students of universities and universities of applied sciences that has information about their educational performance, educational choices and perspective on their future careers. The survey was conducted every 2 to 3 years from 1982/1983 to 2015 (Multrus et al., 2017) providing 13 cross-sectional waves (because the last wave of 2015 did not ask for perceived discrimination and competitive climate we use twelve waves from 1982/1983 to 2013). The response rate was constantly over 40% until the late 1990s when it dropped markedly. In wave 12, the last wave in our study, the response rate was down to 18.6%. The sampling strategy of the German Student Survey followed a two-step procedure. In the first step, the universities were included according to their distribution across Federal States and by size, institution type (university or university of applied sciences) and the range of academic disciplines offered. The number of universities (of applied sciences) participating varied from 11 to 25. In a second step, a random sample of students was drawn from within these institutions. The number of students participating in the survey was approximately 8000 per wave (for more information on the sampling procedure, see German Student Survey, 2024). The survey offers the unique opportunity to study a representative sample of German higher education students over time. Moreover, the possibility of pooling data across a long period of about 30 years allows for sub-groups that are large enough to permit us to investigate our research questions by looking at different fields of study.

The initial sample (pooled across all waves) contained 108,969 students (both men and women) studying at universities or universities of applied sciences, but we restrict the analytical sample in several ways. First, we include only students who report both pass grades at high school and preliminary average grades at the university, and who are not in the first two semesters of university (as many students are not yet graded at this stage). The selection of students with passing grades at school level restricts our sample to students who are allowed to enter higher education directly after finishing high school and excludes the very small group of students who gain their higher education entrance admission through indirect pathways, such as occupational training and work experience. In our discussion, we address the implications of these selections for our findings. Second, we restrict our sample to the six fields of study of interest for our analysis (law, medicine, economics, engineering, natural sciences and social sciences). Law, medicine, natural sciences and economicsFootnote 2 are gender-balanced fields of study, while engineering is a male-dominated field, and social sciences are female-dominated. Although these fields of study are available at both universities and universities of applied sciences, law and medicine are largely exclusive to universities. Third, we exclude the last wave of the survey (2015) as it does not have data on perceived discrimination against women; thus our data cover 1982–2013. After considering these restrictions, using listwise deletion, we keep only the observations with complete information on the variables of interest (N = 49,576). In the Supplementary Information (SI), Fig. A1 details the restrictions on our analytical sample.

Variables

The dependent variable, university grade, is a continuous variable from 10 to 40, where in our analysis, 40 is the highest (“best”) grade and 10 is the lowest (“worst”) pass grade. The second digit indicates the first decimal place, i.e., 10 indicates a grade of 1.0. Note that for the purpose of the analysis, we inverted the German grading scheme, so a higher grade indicates a better grade. It represents the self-reported average grade measured between the third and fifteenth semester (depending on the semester in which the students are enrolled when the survey takes place), and in our analysis, we also control for the semester in which the university grade was measured. The main independent variable, high school grade, is a continuous variable from 10 to 40, where 40 is the highest grade and 10 is the lowest pass grade (again, the German grading scheme was inverted).

Both context variables, perceived discrimination and perceived competition, are measured on a seven point rating scale ranging from zero (level of perceived discrimination against women/competition not high) to six (level of perceived discrimination against women/ competition very high). Fields of study at the university has six categories: law, medicine, economics, engineering, natural sciences, and social sciences. Wave represents the survey year in which the students completed the survey, differentiating between 12 waves: 1982/83, 1984/85, 1986/87, 1989/90, 1992/93, 1994/95, 1997/98, 2000/01, 2003/04, 2006/07, 2009/10, 2012/13. University represents the university that students attended, and they are included in the analysis as dummy variables. Further, to account for pre-university variables that might be relevant for major choice, we included in our analysis the type of higher education entrance qualification (as a proxy for high school type) and parental education (at least one parent has a higher education degree versus neither parent has a higher education degree). The type of higher education entrance qualification reflects the nature of preparation that the students receive, which predisposes them to certain fields of study. We differentiate between general higher education entrance qualification, subject-specific higher education entrance qualification, entrance qualification for universities of applied sciences, and other university entrance qualification. Similarly, parental education, as a proxy for parental background, is associated with students’ aspirations and access to different fields of study (for example, girls with highly-educated parents are more likely to follow gender-atypical fields of study, see Berggren (2008). In our survey, gender is measured as a dichotomous variable that elicits the respondent’s sex assigned at birth, rather than their gender identity. It is measured with 1 indicating female sex and 0 indicating male sex.

Method

In the first part of the analysis we provide a descriptive analysis of the claim that (i) women have an advantage in high-school grades, but not in university grades; and (ii) women perceive, on average, more discrimination against women and competition among students in their degree programme, compared to men. First, we evaluate the association between high school grades (\({Y}_{i})\) and gender (including wave, high school diploma and parental education as control variables):

$${Y}_{i}={\beta }_{0}+{\beta }_{1}gender+ {\beta }_{2}wave+{{\beta }_{3}high school diploma+{\beta }_{4}parental education+\varepsilon }_{i}$$
(1)

and, second, to evaluate the association between university grades (\({Y}_{i})\) and gender (including the previous controls, field of study, university and the semester when university grade is assessed).

$${Y}_{i}={\beta }_{0}+{\beta }_{1}gender+{\beta }_{2 }field of study+{\beta }_{3 }university +{\beta }_{4 }wave+{\beta }_{5} high school diploma+{\beta }_{6}parental education+ {\beta }_{7}semester+ {\varepsilon }_{i}$$
(2)

To statistically test the second claim, using the same controls and robust standard error as in Eq. 2, we evaluate the relationship between perceived discrimination against women/perceived competition (\({Y}_{i})\) and gender.

In the second part of the analysis, linear regression models with robust standard errors are used to explore the association between university grades (\({Y}_{i})\), and factors that might have disparate impacts on men and women, such as high school grades, perceived discrimination against women, perceived competition (controls included).

$${Y}_{i}={\beta }_{0}+{\beta }_{1}high school grades+{\beta }_{2}gender+{\beta }_{3 }discrimination+{\beta }_{4 }competition+ {\beta }_{5 }high school grades*gender{+\beta }_{6 }discrimination*gender+{\beta }_{7}competition*gender+{\beta }_{8}field of study+{\beta }_{9}university+{\beta }_{10}wave+ {\beta }_{11} high school diploma+{\beta }_{12}parental education+ {\beta }_{13}semester{+ \varepsilon }_{i}$$
(3)

In the third part of the analysis, we scrutinize whether in engineering (as the male-dominated field of study) versus other fields (binary variable), high school grades are less strongly associated with university grades for women, compared to men. Therefore, as a first step, we conduct the same analysis as in Eq. 3 (without field of study) but on two different samples: (i) engineering and (ii) the other fields aggregated. We also further investigate this in the full sample with a three-way interaction between high school grades, gender, and fields of study (measured both as a binary variable and categorical variable) in the SI. We estimate all models by OLS regression with robust standard errors to account for heteroscedasticity.Footnote 3

Results

Descriptive Analysis

Before we report the main findings, Figs. 1 and 2 illustrate the distribution of grades at high school and university level across fields of study by gender. We note that female students generally achieve better high school grades than males in all fields of study (besides natural sciences, see also Table A1). Despite the lower female representation in engineering (18%) and economics (41%), these women, selected on their strong high school performance maintain their good grades also in university. Additionally, in predominantly female-dominated fields, such as social sciences (69%), female students seem to continue to outperform their male counterparts.

Fig. 1
figure 1

Source: German Student Survey (1982–2013). Note: high school grades are measured on a continuous scale ranging from 10 (“worst”) to 40 (“best”) as the original German grade scale was inverted

Distribution high school grades across fields of study, by gender.

Full size image
Fig. 2
figure 2

Source: German Student Survey (1982–2013). Note: university grades are measured on a continuous scale ranging from 10 (“worst”) to 40 (“best”) as the original German grade scale was inverted

Distribution university grades across fields of study, by gender.

Full size image

Besides reporting descriptive statistics on grades, Table A1 in SI presents the descriptive statistics of the main variables used in the analysis by field of study and gender. It is worth noting that engineering is the most male-dominated field in our analysis, while social sciences is the most female-dominated one. Moreover, students enter different fields of study with different average high school grades as the grading conventions might be different. This is particularly true for the field of medicine since admissions are widely directly based on very good high school diploma grades (see the truncated distribution of grades for medicine in Fig. 2). In other fields of study, local thresholds of admission might have been in place at some universities and at some points in time. However, as Fig. 2 clearly illustrates, this does not translate into a restricted distribution of grades since in all fields of study under focus, we see a broad variation of high school diploma grades.

Furthermore, while perceived discrimination is higher among women than men across all fields of study, except social sciences, the difference in perceived competition is less pronounced (see Table A1 in SI). Over the period from 1982 to 2013, women consistently reported higher levels of perceived discrimination and competition than men, although these levels decreased over time, particularly for discrimination (Fig. A2 in SI).

Next, we report the results of the main empirical analysis in three steps. First, we provide descriptive evidence of the potential gender disparities in grades (both at high school, and at university) as well as in perceived discrimination against women and perceived competition. Second, we consider the relationship between university grades and factors that might have a gendered dimension in their association (high school grades, perceived discrimination, and perceived competition). Third, to gain a deeper understanding of the results, we scrutinise whether the association between university grades and high school grades varies by gender in engineering, compared to other fields of study.

Table 2 shows the associations between on the one hand, (i) high school grades, (ii) university grades, (iii) perceived discrimination against women, and (iv) perceived competition among students, and, on the other hand, gender (including controls). When examining the model that estimates high school grades, it is apparent that there is a significant correlation between gender and grades (Table 2, first column). Namely, women, compared to men, have better grades in high school (a higher coefficient indicates a better grade). On average, being a young woman, compared to being a young man, is associated with receiving a 1.02 better grade in high school, which translates into a gap of 0.1 better grade on the original scale.

Table 2 Gender differences in grades, perceived discrimination and competition
Full size table

At the university level, men are likely to have better grades than women (Table 2, second column). This difference is statistically significant at 0.05 levels. This indicates that although women receive better grades in high school, this advantage does not carry over to university. Further, this gender difference remains when controlling for fields of study which means that it cannot be explained by different grading conventions in male- or female-dominated fields of study. Yet, it is worth noting that compared to social sciences students, law and economics students have the lowest grades at the university level. Social sciences graduates have the highest grades at the university level (for descriptive statistics on grades, see Figs. 1 and 2 in the manuscript and Table A1 in the SI).

As we argued in the theoretical considerations, the fact that women fail to maintain their advantage in grades gained in high school at the university level might be related—amongst other things—to two factors: discrimination against women and the perceived level of competition among students. At first sight, there is a significant correlation between gender and both perceived discrimination against women and competition (Table 2, columns three and four). Specifically, women, compared to men, tend to perceive there to be higher levels of discrimination against women and competition among students. Additionally, the coefficient size of discrimination and competition variables indicates that women (compared to men) perceive higher levels of perceived discrimination against women than competition among students. That means that men are less perceptive about discrimination against women (note that the survey question on “discrimination against women” was posed to women in the same way as to men). What we also see is that the level of perceived discrimination against women and the level of perceived competition tend to be higher in almost all fields of study compared to social sciences (the most female-dominated field).

What Factors are Associated with University Grades?

Turning to the factors associated with university grades, Table 3 shows the correlation between university grades and the factors that might have a gender dimension, including, high school grades, perceived discrimination and perceived competition (controls are included).

Table 3 Factors associated with university grades
Full size table

At first sight, we can see that the association between high school and university grades varies by gender: namely, high school grades are less strongly associated with university grades among female students than male students (H1b supported). Specifically, a point increase in high school grades amounts to a 0.371-point increase-equivalent to a 0.055 standard deviation in university grade (based on the analytical sample)-in university grades for men and a 0.27- point increase (main effect 0.371 minus interaction effect 0.091), equivalent of 0.042 standard deviations for women. Further, in analysing the claim that perceived discrimination and competition could hinder women’s university grades, Table 3 shows that only the interaction between perceived discrimination and gender is statistically significant (H3 supported; H4 not supported). This indicates that higher levels of discrimination are indeed associated with lower university grades for women but not for men (the marginal effect is − 0.185, p = 0.000 and 0.051, p = 0.034 for women and men, respectively). More precisely, this means that, for women, a one-unit increase in perceived discrimination (on a scale from 0 to 6) is associated with a decrease in university grades by 0.185 points. In practical terms, if women’s perceived discrimination were to increase from the minimum to the maximum level, it would amount to a decrease of 1.11 points in university grades. Although the sizes of the coefficients are small, the negative effects on women’s performance associated with high school grade translation and perceived discrimination could have significant implications, especially in a setting where university grades are relevant for labor market entry.

Further, to facilitate the interpretation of the interaction, Fig. 3 illustrates the predicted relationship between university grades and a given predictor for men and women, respectively (based on the model presented in Table 3). Panel A shows a stronger relationship between high school grades and university grades for men than for women. In a similar vein, Panel B illustrates that whereas the level of discrimination against women is negatively related to university grades for women, it is practically unassociated with university grades for men. Panel C shows that across both genders, higher levels of competition have a roughly similar (and weak) negative association with predicted university grades.

Fig. 3
figure 3

Source: German Student Survey (1982–2013); both high school grades and university grades are measured on a continuous scale ranging from 10 (“worst”) to 40 (“best”) as the original German grade scale was inverted; both high perceived discrimination against women and perceived competition among students are continuous variables, measured on a seven-point Likert scale where higher values indicate a greater average level of perceived discrimination/competition

Differing effects of high school grades, perceived discrimination, and perceived competition on university grades for men and women.

Full size image

Thus far, we have shown the factors that are associated with university grades. To test H2, we need to take a closer look at the relationship between university grades and high school grades for men and women in engineering (as the male-dominated field of study) versus the other fields (law, medicine, social sciences, economics, natural sciences).

As a first step, Table 4 illustrates the association between university grades and high school grades for young men, compared to young women for engineering and the other fields of study (natural sciences, economics, law, medicine, social sciences). It is evident that the lower association for women is about equal in both fields (− 0.092 and − 0.118, respectively) and a three-way interaction between high school grades, gender, and engineering (versus other fields) confirms that the difference is not statistically significant (see Table A2 and Fig. A3 in the SI). In other words, men and women’s high school grades are equally predictive of their university grades in both engineering and other fields of study.

Table 4 Gender differences in university grades, by male-dominated versus other fields
Full size table

However, if we disaggregate the other fields of study in the analysis (using a three-way interaction between high school grades, gender, and each specific field of study), we find a statistically significantly (p = 0.005) stronger association between university grades and high school grades for men than for women in the field of engineering compared to the field of social sciences (see Table A3 and Fig. A4 in the SI).Footnote 4 Interestingly, we also find a statistically stronger association between high school grades and university grades for men relative to women in the natural sciences—the second most male-dominated field—compared to the field of social sciences. In short, we find some support for H2 (in engineering, compared to other fields, high school grades are less strongly associated with university grades for women than for men), although this depends on the specific field used as a benchmark. More generally, this points to paying further attention to qualitative nuances between fields.

Lastly, we also observe interesting gender-differential relationships between perceived discrimination and perceived competition. In engineering, perceived discrimination is uncorrelated with university grades for men (marginal effect is 0.006, p-value = 0.845), but strongly negatively associated (marginal effect is − 0.336, p-value = 0.000) for women as expected. In the other fields of study, perceived discrimination is again negatively associated with university grades for women (marginal effect is − 0.149, p-value = 0.000), but positively associated with university grades for men (marginal effect is 0.266, p-value = 0.000). As for perceived competition, we observe a statistically significant association for women (marginal effect is 0.115, p-value = 0.035) and, conversely, a statistically significant negative relationship for men (marginal effect is − 0.068, p-value = 0.013) in engineering. By contrast, there are very strong negative associations between perceived competition and university grades for both men (marginal effect is − 0.622, p-value = 0.000) and women (marginal effect is − 0.700, p-value = 0.000), with a slightly stronger association women. We elaborate on these results in the discussion.

Limitations

It is worth noting that our study considers only individuals with passing grades at both high school and university. While the limitation to those students with passing grades in high school is not considered problematic since it only excludes a very small group of students who gained their higher education entrance via other ways, such as occupational training and work experience, it does not allow us to capture low-performing students who have to leave university because they did not pass an exam. If women’s grade-related dropout is associated with discrimination or competition, this might lead to an underestimation of this relationship. However, if male students are more likely to drop out since they have, on average, lower grades in high school, the effects might be overestimated. Unfortunately, our survey does not provide data for the students who drop out of their courses and therefore, we cannot account for dropout.

Further, our study relates to imprecise measurement in survey data—in particular, in regard to self-reported grades. Although high school grades have been received some years before survey participation they are likely to be remembered with a relatively high degree of precision because of the high salience of this number in German high schools. Again, we assume a positively biased reporting. In addition, the salience of average grades received at university increased in the years after the Bologna reform, since students are now provided with preliminary transcripts of records that were not available in earlier years. The limitation regarding the self-reporting also applies to our measure for perceived discrimination and perceived competition—a criticism that applies to most studies on social perceptions: students might have different views on what constitutes discrimination or competition within different fields of study. Last, but not least, this study is correlational in nature and the results can only be interpreted carefully regarding the complex selection processes described. Nevertheless, main descriptive results on the grading gap and its correlation with field of study and perceived discrimination and perceived competition seem to be robust.

Discussion

Our study started with the question of why the positive gap between female and male students in terms of their high school grades does not translate into a positive gap between their grades at the university level (Francesconi & Parey, 2018). University grades play an important role during the initial stages of individuals’ careers (Røberg & Helland, 2017; Zou et al., 2022); thus, it is crucial to comprehend the nature of the obstacles women might encounter at university that potentially impair their performance. Empirically, we provide a descriptive analysis of the claim that women have a grade advantage in high school but not at university, and we investigate the factors associated with university grades by taking advantage of the German Student Survey, which spans a period of approximately 30 years.

Our findings have marked implications for understanding the performance of young women in university relative to high school. This is arguably especially important in Germany, as women, compared to men, are less likely to enroll in higher education (Lörz & Mühleck, 2019; Lörz et al., 2011) and therefore, in that sense, a more selected group than their male counterparts. We show that high school grades are more strongly correlated with university grades for men, compared to women. Further, we find that there is a stronger association between university grades and high school grades for men than for women within the field of engineering (male-dominated) compared to the field of social sciences (the most female-dominated field in our analysis). In fields of study in which the representation of women is low, young women’s good grades in high school seem to fail to be translated into better grades at university. This is especially noticeable since young women who enter STEM fields have, on average, better high school grades than their male counterparts who seem to choose these male-dominated fields as “a default” option. Thus, women studying engineering might encounter a “chilly climate” (Hall & Sandler, 1992; Morris & Daniel, 2008), as demonstrated by our finding that in engineering, for women, compared to men, perceived discrimination tends to have pronounced negative consequences on grades. Thus, an environment characterized by a lack of inclusivity and support for minority members, can negatively contribute to women’s lower performance. This result is in line with the theoretical considerations that this form of tokenism might hinder women’s ability to capitalise on their talents.

What have we learnt about the potential factors that hinder young women’s achievement of their academic potential? The descriptive analyses show that female university students report higher levels of perceived competition among students and discrimination against women than their male counterparts—supporting the idea that studying at a university is impacted by some gender-specific barriers. However, only perceived discrimination turned out to be a pronounced factor associated with young women’s university grades. While overt discrimination against female students might be rare, the perception of unequal treatment among female students translates into negative consequences. Substantively, this finding is in line with status characteristics theory (Ridgeway, 2011) which argues that women are negatively stereotyped with regard to their studying/working capacities, which might result in perceived discrimination. While these analyses identified perceived discrimination as a factor that is correlated with poorer university performance of women compared to men, this study cannot determine the circumstances in which perceived discrimination is likely to occur. We can speculate that “stereotype threat”—a phenomenon that occurs if performance is being evaluated and there is a potential for confirmation of a negative stereotype associated with a given group (Spencer et al., 1999, 2016)—might contribute to perceived discrimination by exacerbating performance disparities between young women and men.

Further, while perceived competition is associated with worse university grades, there is no statistically significant difference between men and women in the overall sample of graduates. Therefore, young men and women are equally likely to receive lower grades where there are high perceived levels of competition. While research shows that men seem generally to perform better under competition than women (Gneezy et al., 2003; Ors et al., 2013), this does not seem to apply in the case of university grades in the overall sample of fields studied. However, our analysis of subsamples shows that perceived competition for women, compared to men, in male-dominated fields of study (relative to other fields) might be a positive motivator rather than a hindrance. In fields of study in which women are underrepresented (such as engineering), they might suffer from lower performance expectations, but it might also be expected that, on average, they are “different” compared to those who followed gender-normative academic paths. This resonates with previous literature from the U.S. that suggests that women who pursue engineering are strongly selected on academic confidence and resilience, as they encounter significant barriers in this male-dominated field compared to women in female-dominated fields (Cech et al., 2011). Therefore, they might perceive the competition in these fields as an opportunity to prove themselves and demonstrate their skills and abilities, leading to an underestimation of the association between discrimination/competition and grades.

The study highlights the gender-specific barriers that hinder women’s academic advantage over men in university settings, making this environment challenging for their academic performance. Although blatant discrimination might be less common, women’s pervasive perception of “non-fit” in certain fields of study, compared to their male peers, contributes to their lower academic performance. These results highlight the need to address gender sexism in higher education. Cultivating inclusive learning environments—where all students, regardless of their gender, receive equal treatment—is crucial for ensuring fair opportunities for academic success (Li & Singh, 2022). Additionally, the importance of such environments extends beyond academic achievement. As gender-inclusive environments foster, among others, creativity and innovation (Vedres & Vásárhelyi, 2023; Yang et al., 2022), it becomes evident that removing gender barriers is essential to enable women to achieve their full potential.

Can our findings be generalized to other countries? When it comes to grades, Germany is fairly similar to other European countries, as high school grades are a strong predictor of success in higher education (Danilowicz-Gösele et al., 2017; Trapmann et al., 2007). Likewise, university grades in Germany (Piopiunik et al., 2020), have a similarly high signalling value for employment as in other countries (Humburg & van der Velden, 2015 for Czech Republic, France, Germany, Italy, Poland, Spain, Sweden, the Netherlands and the United Kingdom). Yet, further research is needed to confirm to what extent our findings are applicable to other country contexts and to integrate more recent data.