The Dutch and Swedish medical students differed in gender awareness and showed significant differences in outcome on all three N-GAMS subscales. Dutch students reported higher gender sensitivity but they also expressed more gender-stereotypical attitudes towards patients and doctors. There were differences between men and women in stereotypical thinking, and in both countries female students disagreed more with stereotypical statements about patients. The pattern was that Swedish female students expressed the least stereotypical thinking, followed by Swedish male students, Dutch female students and finally the Dutch male students. The students' age had some impact on outcome in both countries, father's country of birth was related to gender sensitivity among the Dutch students, and mother's education was related to stereotypical thinking among the Swedish students.
Data for this study were collected during the students' first week in medical school, implying that the results mirror the gender awareness the students had when entering their studies. The large sample and high response rate was a strength.
N-GAMS was carefully elaborated and validated in a Dutch context , but our factor analysis showed that N-GAMS could be translated and used also among Swedish medical students. Factor structure was the same and the loadings similar in the Swedish and the Dutch samples. The items that needed to be excluded from the subscales had low loading in both samples. From GRI-P and GRI-D, one item from each subscale needed to be excluded (Table 1). The item added to the GRI-D subscale loaded as expected and seemed to add something to the scale. The GS subscale needed the largest revision, with five items excluded due to low factor loading. The remaining nine items all concern whether differences between men and women are relevant to consider in clinical work. In fact, "sensitivity for gender differences" might be a more appropriate name than "gender sensitivity" for this subscale.
Using a scale as a tool for exploring gender attitudes and stereotypes obviously has both strengths and limitations. The N-GAMS scale was suggested to be used for research purposes and to evaluate gender awareness raising courses and one obvious strength with using such an instrument is that a large number of medical students could be included in the study. We also found that N-GAMS could be translated and used outside the Netherlands, meaning that the instrument makes it possible to compare medical students in different countries and cultural settings. However, the methods needs certain reflection; when understanding the differences we found between Dutch and Swedish students the background variables used in the questionnaire do not in themselves explain the results. Still, the significant differences in students' characteristics (Table 2) provide clues to the context of the medical students' attitudes and awareness. While the N-GAMS scale could be used for example as a baseline assessment to evaluate gender awareness raising courses, or to make a comparison of a large amount students, qualitative research might give further depth and provide more explanations through highlighting the social discourses and reasoning that lies behind the students results and the difference between the Netherlands and Sweden.
The fact that the students in Nijmegen compared to the students in Umeå were more sensitive to gender differences and also more gender stereotyping of patients and doctors might have different explanations. Since the students came from all over Sweden and the Netherlands and answered the questionaire during their first days of medical school it is reasonable to believe that the differences reflect national differences rather than University differences. As outlined in the introduction, Sweden and the Netherlands are both welfare states, but the Netherlands has less far-reaching gender equality legislation and a more obviously gendered distribution of chores and duties than Sweden [19, 20, 23]. Childcare and household work are the responsibilities of Dutch women who to a large extent work part-time, while most Dutch men are the main breadwinners and work full-time. In Sweden a majority of both women and men work full-time while the children are taken care of at day-care centres . Conditions in society and people's behaviour have an impact on people's attitudes, expectations and values , and inasmuch as Dutch students see more differences between men and women in everyday activities it seems reasonable that they perceive gender differences as more relevant and also score higher for GS than the Swedish students do.
While the statements in the GS subscale consider the importance of biological and communicative gender differences for the clinical work, the statements in the GRIP and GRID subscales contain rather outspoken and evaluative statements about how male and female patients and doctors are, for example that patients or doctors of one gender are "better", "too much" or "less", compared to the opposite gender (Table 1). Agreeing with the statements implies believing in gender differences, but also accepting a hierarchy in the differences - the features of one gender are described as being superior to and more desirable than the features of the other. In line with common gender stereotypes in society [24, 25], the statements describe men as more competent, effective, instrumental and trustworthy than women, who are seen as more emotional, worried and in need of more attention and time to talk. That the Swedish students disagreed more than the Dutch with gender-stereotypical statements was not so surprising. The dominant political discourse about gender in Sweden proclaims that women and men are equal and should be treated equally . These social norms reasonably influence not only students' values but also what they consider to be suitable and political correct answers in an enquiry.
Although communal conceptions about men and women underpin attitudes towards both patients and doctors, the students' gender stereotyping of patients was more pronounced than their stereotyping towards doctors. Previous research shows that people's self-reports of their instrumental and communal traits are usually less gender-stereotypical than are their estimates of the "typical person" [24, 27]. Perhaps students in the first term already identify with the group of "doctors", and knowing the diversity of their own features they avoid judging doctors according to preconceptions about typical male and female behaviours?
The finding that male students held stronger gender stereotypes towards patients than their female peers is consistent with the previous Dutch study (using N-GAMS) of medical students , and with other research comparing gender-stereotypical attitudes in men and women [24, 25]. In the present study both male and female students scored in the direction of disagreeing with stereotypes, but the female students stated more clearly that they disagree. The reason that men admit more to gender stereotypes could be that such stereotypes are in general more positive towards men, which is also true for the statements on the GRI-P subscale .
Proportions of women in full-time work and educational level in both men and women are often used as indicators of gender equality in a country [19, 20, 23]. We therefore expected that these background variables would contribute to explaining the differences in gender awareness between the medical students in the Netherlands and Sweden. However, no such clear pattern emerged. How can that be understood?
It is well described in research that prevailing norms and beliefs about gender in a society are created and maintained by multiple, complementary processes acting simultaneously and at different levels of analysis, e.g., individual, societal, political, normative and cultural levels [22, 24, 25]. Changes on a single level will not be sufficient to eliminate or change the gendered norm patterns or attitudes. For instance, even if political decisions aiming at gender equality have had an impact on women's career choices as well as on men's participation in traditional female tasks [19, 20], studies on gender stereotypes show that the core structure of beliefs about typical men and typical women have been largely unchanged for decades [24, 25, 27]. Gender stereotypes have a high momentum of inertia and change slowly. Transferring this reasoning to our study, Swedish and Dutch students with parents acting in contrast to traditional gender roles are still living in societies where gendered segregation of duties is a prevailing pattern and where cultural beliefs about gender are exposed in everyday talk as well as in newspapers, books, films, and other media. Thus it becomes hard to relate the differences in attitudes between the Dutch and Swedish students to one or the other of the characteristics used as background factors. Moreover, within each country of study, the variation among the students in socio-cultural background is small and a majority of the students have similar background characteristics (Table 2). When the countries are separated and the background variables analysed in linear regression, most students are grouped in the same categories, which would explain that even the model including all background variables does not explain more of the variation in outcome on N-GAMS (Table 4). Still, our study indicates that mothers' educational level is important for gender attitudes.
Implications for education
It is important to know the students' attitudes as a take-off point in the implementation of gender in the medical curricula [12, 17]. Different attitudes between the Dutch and Swedish students informs us that medical education has to take cultural and societal differences between countries into account, and even when comparing two welfare western states there are differences in values and preconceptions about gender. Inasmuch as the goal in education about gender is to make students interested and aware of the significance of gender in medical work, the examples used in discussions need to be trustworthy, relevant and challenging in the context of the specific country.
In our study the male students agreed more with gender-stereotypical statements about patients than their female peers, which is in line with previous research showing that men to a larger extent than women hold on to gender-stereotypical ideas [11, 24]. Earlier studies have shown that male students are less knowledgeable about gender issues and also more sceptical about the implementation of gender in education than female students [15, 16]. We suggest that these phenomena are related. Lack of insight and knowledge implies difficulties in understanding the relevance of gender in medical work, and not understanding the relevance makes you less interested in, and more sceptical about, education on gender. Furthermore, lack of interest means that you are less motivated to reflect on your own attitudes and values concerning gender and you are thus more subject to stereotyped preconceptions about men's and women's needs, wishes, and behaviour. Is it possible to break through this circular process? Yes, to some extent at least. Previous research claims that students become more positive and interested as they learn more and get used to discussing gender [15, 16]. However, to the best of our knowledge, there is a lack of research about whether tendencies to make stereotyped assessments of patients decrease as a result of the implementation of gender aspects in education.
Since gender-stereotypical assessments might lead to bias and misjudgements of patients, as well as misunderstandings in collegial cooperation, it is important to find out how to help students to reduce judgments that are grounded in stereotypes [9, 12]. Reflecting about one's own reactions to patients in clinical situations might be a tool that helps students to see their own attitudes and the consequences these might have for medical treatment. Such reflections can proceed from the students' own experiences from e.g. clinical training in hospitals and primary care, in prefabricated paper-cases, or in symptom presentations from simulated patients.
Our experiences from gender education indicate that, parallel to the importance of good examples to discuss and reflect on, it is crucial to create a climate for dialogue where students feel permitted to disclose ideas and attitudes, including such ideas as are not "politically correct" in a specific context. If problematic ideas and reactions are not disclosed, the students' chances to catch sight of their own attitudes and values, and to discuss them, will be small and as a result the education will have less impact. On the other hand, when a student realizes that her/his own, or other students' or staff members', values and preconceptions might lead to problems for patients, then s/he usually becomes curious and interested.
Further research about the effects of gender perspectives in education on students' attitudes and behaviour need to focus the gender topics included as well as the education methods used.