Introduction

Despite many advances in human rights regarding gender and sexual (GenSex) diversity from innovative HIV/AIDS public health, and evolving conceptions of marriage law, to the creation of public policy that aims to ensure equity between women and men, and the world’s largest pride parade—systems that discriminate against GenSex minorities still persist in Brazil (Mello, Brito, and Maroja 2012). For instance, educational policies to combat GenSex discrimination and guarantee LGBT rights are still underdeveloped and inconsistent. Notoriously, in 2011, Brazil’s President personally announced the cancellation of the ‘Schools Without Homophobia’ project. Congress members that are conservative religious leaders are strongly opposed to the progression of women’s and LGBT rights, shifting Brazilian political agenda to reinforce prejudice (Vital da Cunha and Lopes 2012).

Heteronormativity constitutes the systems of gender and sexual norms, which work together to reinforce ideological and behavioural standards in a society. These systems manifest as discrimination in the form of sexism (Glick and Fiske 2001), heterosexism (Herek 2004; Massey 2009) and transphobia (Hill and Willoughby 2005). As such, there are certain privileges conferred in the mere state of being designated male, solely attracted to the other sex (heterosexual) and identifying with the sex you were given at birth (cis-gendered). Emotional abuse, physical violence, and exclusion are the consequences of the deviation of these standards (Meyer 2003); furthermore, the effect of occupying multiple subordinate identities is cumulative (Hendricks and Testa 2012). It is this reinforcement of conformity and the punishment of dissent that maintains GenSex hierarchies. Transphobia—defined as prejudice against gender non-conformity—presents an informative intersection of gender, sex, and sexuality, yet it is the least studied of this triangulation of prejudices (Hill and Willoughby 2005; Sennott 2010).

Clausell and Fiske (2005) have used the Stereotype Content Model to demonstrate that stereotype content for gay men reflects the degree of gender conformity in the expression of their sexuality. Gay men who ‘pass as straight’ were perceived as high in warmth and competence and therefore often escaped overt forms of prejudice against sexual orientation, or as Cuddy et al. (2007) have termed it, active harm. This is especially true in Brazil where homosexuality tends to be perceived through gender expression and sexual roles. For example, gay men have historically been characterized as either a bofe—masculine, dominant, and penetrative—or a bicha—feminine, submissive, receptive (Fry 1986). With the organization of the Brazilian civil rights movement in the 1990s and influences from North American gay culture, the bofe–bicha model has shifted, and men who engage in same-sex behaviour started to be described as ‘gay’, regardless of sexual roles (Green 1999). Nevertheless, aspects of these traditional notions are still visible in effeminate stereotypes of homosexual men. Conversely, masculine men who have sex with men tend to receive less explicit discrimination and are often perceived as heterosexual (Parker 1999). Recently, Costa et al. (2013) have gathered empirical evidence demonstrating the central role of gender norms in the manifestation of sexual prejudice in Brazil. The study showed that the explicit forms of prejudice are not diminishing and that although there is a clear theoretical distinction between sexual and gender diversity, from the standpoint of manifestation of prejudice, that distinction seems to be more tenuous.

The Trans Murder Monitoring Project has revealed that Brazil has the highest murder rate of transgender individuals in the world (Balzer et al. 2012). Despite these shocking rates, it is common knowledge that most cases go unreported. Individuals who identify with the Brazilian transgender identity travesti are most at risk of these attacks. Travestis are trans people who were assigned male at birth but affirm female gender performance and bodily form, although typically not undergoing neovaginoplasty. Their gender identity varies across individuals and contexts: most identify as male, some as women, and others simply as travestis. This counter-normative shift away from traditional gender binaries is a source of much controversy (Barbosa 2013). One tenuous intersection between gender and sexualityin Brazil can be seen in the instance of travesti sex-workers who are sought by men for penetrative anal sex. Because travestis are conceived of as phallic women, these acts do not seem to challenge the heterosexual identity of the clients (Pelúcio 2011).

The presence of GenSex discrimination in educational institutions has been a phenomenon of academic interest since the mid-1980s (for a comprehensive account of the North American literature, see Rankin (2003)). The negative impact on students’ quality of life and educational outcomes are significant and well documented (Kosciw et al. 2013). Whilst the development of Brazilian empirical research on GenSex prejudice is recent and scarce, the systematic review of Costa et al. (2013) on the available data included various studies in educational contexts. Dunbar et al. (1973) demonstrated that Brazilian undergraduates presented more homophobic attitudes and attributed more feminine characteristics to homosexual men compared to Canadians. Additionally, Araujo et al. (2007) looked at the way students viewed adoption by homosexual couples, stratified by their disciplines. Notably, undergraduate Psychology students tended to cite ‘psychological disturbance’ as a possible outcome faced by children, whilst law students considered ‘questions of morality’ to be a potential issue. In a survey of Brazilian undergraduate theology students by Pereira et al. (2011), explicit prejudice was associated with strong ethical and moral beliefs about the nature of homosexuality. Those who held biological and, especially, psychosocial beliefs about the nature of homosexuality were also biased, but more subtly. Finally, in a study among graduate Psychology and Administration students, Fleury and Torres (2007) demonstrated that they were more likely to attribute positive characteristics to heterosexuals than homosexuals.

The previous Brazilian studies utilized convenience samples and culturally insensitive measures. Understanding how discrimination functions in Brazil in a broader undergraduate context is the first step to ensuring a more positive tertiary educational environment. Consequently, we proposed an evaluation of attitudes towards GenSex at the Brazilian public university using a recently developed, culturally sensitive measure.

The objectives of this study can be summed up by three research questions: How is GenSex prejudice expressed at the university? How does prejudice vary in relation to socio-demographic variables, in particular to religion affiliation? And what are the effects of discipline and exposure to GenSex education on levels of prejudice? Extant literature suggests that the following variables may provide answers to these questions: the length of time at the university; gender; religiosity, religious affiliation, and attendance; birthplace population; sexual orientation; previous GenSex-related education; and the discipline studied at the university. An analysis of the effects of these variables should allow for an informative exploration of the dynamics of GenSex discrimination in Brazilian public undergraduate education systems.

Method

Participants and Procedures

UFRGS is a century-old public educational institution maintained by the federal government and situated in Porto Alegre, the capital city of the State of Rio Grande do Sul, southern Brazil. It offers free undergraduate and graduate academic programs from an extensive array of disciplines following a competitive entrance examination (~7.5 % overall admission rate). The institution accommodates over 30,000 people. In 2014, for the third consecutive year, UFRGS was rated as the best university in Brazil by the Brazilian Ministry of Education.

The Human Ethics Committee of the Institute of Psychology at UFRGS approved this research project (project number 04642712.9.0000.5334). On December 3, 2013, the rectory of the university sent e-mails to all undergraduate students enrolled in the university inviting them to participate in the research (N = 28,410). The e-mail outlined the purpose of the study and asked students to respond to the self-report questionnaire online. Participation was voluntary; however, students who did not respond received two further invitations in the week following the initial request.

A sample of 8,184 undergraduate students responded on the questionnaire (29 % response rate), which is a satisfactory rate for this method of sampling (Nulty 2008). Participants’ disciplines were divided according to the governmental standard of Brazil (CAPES 2012; Table 1). Notably, UFRGS has a Biotechnology course that is the only one classified as ‘Multidisciplinary’. For the purpose of this research, the participants of this course (n = 33) were classified in the Biological Sciences category. Table 1 provides more information regarding the sample and university distributions as well residuals. The percentage of students in each discipline ranged from 24.37 to 35.72 %. These proportions significantly deviate from a proportional representation of the university, as demonstrated by a chi-Square goodness-of-fit test (α = 0.05). The implications of this distribution will be discussed further.

Table 1 Sample per academic discipline
Full size table

The average age of the participants was 25.89 (95 % CI [24.73, 26.04]; Mdn = 24; SD = 7.25), ranging from 16 to 81 years, with a majority (57.3 %) of our sample comprised of adults who were in emerging adulthood (18–25). The majority of participants self-identified as heterosexual (83.9 %), whilst 13.7 % identified as non-heterosexual (gay, lesbian, bisexual, or others), and 2.4 % reported that they did not know their sexual orientation. A total of 45.5 % identified as male, 53.7 % as female, and 0.8 % identified with ‘other [gender]’ or that they did not know it. Characteristics of the sample can be found in Table 2.

Table 2 Sample characteristics
Full size table

Measures

Socio-demographics

“Participants answered socio-demographic questions regarding their gender (male, female, I don’t know/other), age, undergraduate discipline, year of entry in the university, sexual orientation (heterosexual, non-heterosexual [gay, lesbian, bisexual, other/I don’t know]), and birthplace population density (cities with more or less than 100,000 habitants). Participants were also asked if they were religious. Those who answer affirmatively were then asked for their religious affiliation and rate of religious attendance (no, low, and high attendance). The affiliation followed the main categories of the Brazilian population census (IBGE 2012) in addition to an open-ended option. UFRGS does not offer systematic GenSex courses for all disciplines. Some disciplines have seasonal extra-curricular courses on this topic or specific lectures within a course indirectly related to the GenSex. For this reason, participants were asked whether they had already been through any form of class, course, or activity related to gender, sexuality, or sexual diversity at the university. Finally, inquiry was made into whether participants had seen or heard of any humiliation, physical assault, or maltreatment of homosexual (gay or lesbian), travesti, or transsexual (regardless of sexual orientation) students at UFRGS.

Prejudice Against Sexual and Gender Diversity Scale

An 18-item questionnaire assessed GenSex prejudice, asking participants about their attitudes (beliefs, affects, and behaviours) toward gays, lesbians, travestis and transsexual people, and gender non-conformity. This scale was created to evaluate extreme explicit GenSex prejudice specific to the Brazilian context (Costa et al. 2015; Costa et al., manuscript submitted for publication). The scale is based on items from two prior instruments: one evaluating prejudice against non-heterosexual orientation (Attitudes Toward Lesbians and Gays Scale; Herek and McLemore 2011) and the other investigating prejudice against gender non-conformity and transsexuality (Genderism and Transphobia Scale; Hill and Willoughby 2005). The former items were adapted to Brazil, and new items were created, prioritizing the assessment of prejudice in gendered terms, including the Brazilian transgender identity travesti.

This scale is comprised of items such as ‘male homosexuality is a perversion’, ‘masculine girls should receive treatment’, ‘men and women should be prohibited from changing their sex’, ‘travestis make me feel sick’. Participants answered on a five-point Likert scale, ranging from 1 (completely disagree) to 5 (completely agree). The scale was validated using an Item Response Theory (IRT) Rasch model. Cronbach’s α indicated high internal consistency (α = .93).

Data Analyses

Prejudice Against Sexual and Gender Diversity Scale mean levels were calculated using Rasch Family IRT analysis (Andrich 1978) and Winsteps v.3.72.2 software (Linacre 2011). The rating scale model was used to transform the raw prejudice scores into standardized logarithm odds units that describe the latent trait (theta, θ), arranged around a zero mean. This transformation gives the data additive properties.

Additionally, the relationship between prejudice and demographic variables—sex, religiosity, religious affiliation, birthplace population density, previous GenSex education, and sexual orientation—were assessed using independent sample t tests, with Cohen's d for effect sizes. One-way ANOVAs were used to evaluate group differences in religious practice and disciplines. Effect sizes were calculated using η 2 (eta-square). Pearson correlations were used to evaluate the relationship between mean prejudice level and years of attendance in the university. Only those who attended the university for at least 5 years (the average length of undergraduate courses at UFRGS) were considered in this analysis (84.21 % of the sample). Finally, chi-square tests, with Cramer’s V for effect size, were used to calculate the difference in discrimination awareness by sexual orientation. All tests were two-sided, with a significance level of .05.

Results

The data collected enabled us to answer each of our research questions in turn. Focusing on how prejudice was expressed at UFRGS, we found that the mean level of prejudice for the entire sample was −1.78 θ, with a standard deviation of 1.34 (95 % CI [−1.81, −1.75]), a median of −1.63, and a range from −4.29 to 4.87, wherein higher levels of theta (θ) denote a greater degree of prejudice.

Item responses were mapped to determine latent trait clusters, which were used to categorize response levels by item difficulty. Clusters were labelled according to level of prejudice: extreme (>1), high (0–1), moderate (−1 to 0), low (−2 to −1), or very low (−3 to −2). To compensate for ceiling effects, cases with the lowest levels of theta (−4.29) were removed from the ‘very low prejudice’ group. Those cases formed a new group that was labelled ‘minimal prejudice’ and constituted 12.17 % of the total sample (Table 3). However, when considering only heterosexual students, this percentage dropped to 8.90 %. It should be noted that these levels reflect degrees of extreme prejudice; therefore, any grade of prejudice above the lowest category is concerning.

Table 3 Characteristics of groups per prejudice level
Full size table

As homogeneity of variance was not found between prejudice levels, Welch’s adjusted F ratio was used in the one-way ANOVA, which demonstrated statistically significant differences between groups at each level of prejudice: Welch’s F (5, 855.688) = 152260.13, p < .001, η 2 = .94. Additionally, the Games–Howell post hoc procedure was implemented. Results indicated that mean scores for prejudice levels significantly differed from each other (p < .001).

Rounding off results for this research question, we addressed students’ awareness of discrimination at UFRGS. Whilst the majority of non-heterosexual participants (63.7 %) had seen or heard of discrimination on campus, almost half as many heterosexual students (33.4 %) were aware of any discrimination (χ 2 [1, N = 8182] = 431.17, p < .001, V = .23). Looking at the overall sample, a one-way ANOVA demonstrated statistically significant differences between those aware of some level of discrimination and those completely unaware in relation to their level of prejudice (F [2, 8180] = 626.541, p < .001, η 2 = .13). A Tukey post hoc test confirmed that those who had observed discriminatory acts were significantly less prejudiced (−2.73, 95 % CI [−2.83, −2.63], p < 001) than those who had merely been informed of these (−2.33, 95 % CI [−2.37, −2.28] p < .001) and even less so than those who were unaware of such acts (−1.40, 95 % CI [−1.44, −1.37], p < .001).

Regarding our second research question, the effects of socio-demographic group membership are shown in Table 4. Whilst significant differences were found for all variables of less theoretical pertinence, large effect sizes were also found for gender (.57) and sexual orientation (.91), which is consistent with much of the extant literature. Finally, a dismissible effect size was found for the population density of a participant’s birthplace. We also wish to draw attention to the large effect size for religiosity (.60) and the high prevalence of non-religious participants (56.2–50.3 % higher than census levels—IBGE 2012). Prejudice scores were significantly different between the levels of religious attendance (Welch's F (2, 1615.89) = 18.22, p < .001). Games–Howell post hoc analysis revealed that the prejudice increase from ‘no’ to ‘low attendance’ was not significant (p = 2.29); however, a significant difference was found between ‘low’ to ‘high attendance’ (−.29, 95 % CI [−.43, −.16], p < .001). A breakdown of statistics by religious affiliation is presented in Table 5. Although Brazil is a secular state with a Catholic majority, the last Brazilian population census found that the State of Rio Grande do Sul has the highest presence and range of religious diversity (IBGE 2012). This may account for the prevalence of different faiths in our sample. Prejudice by affiliation presented a clear pattern raging from lower levels amongst modernized/individualized beliefs (Afro-Brazilians, Paganisms and Neo-Paganisms, Eastern Asians, Spiritism, Agnosticism, and others), intermediate levels among classic Abrahamic creeds (Judaism, Catholicism, and Islamism), and higher levels among orthodox factions (Protestantism, Neo-Pentecostals, Mormon, and other Christians).

Table 4 Prejudice by socio-demographic groups
Full size table
Table 5 Prejudice by religious affiliation
Full size table

Fundamental to the penultimate research question, we found that the distribution of prejudice levels varies according to student disciplines (Table 6 for mean levels; Fig. 1 for a breakdown per prejudice group levels). To reiterate, when interpreting Fig. 1, one should be reminded that, despite the positive skew towards the less prejudiced categories, all items represent egregious forms of prejudice. Whilst all disciplines demonstrated some degree of the two highest levels of prejudice, the data reflect a concentration of ‘extreme’ and ‘high’ prejudice in students enrolled in Engineering, Agricultural Sciences, and Exact and Geological Sciences. ‘Moderate’ and ‘low’ prejudice was over-represented in these disciplines as well as in Health, Applied Social and Biological Science student samples. Conversely, those who have ‘very low’ or ‘minimal’ prejudice tended to cluster in the Humanities and Linguistics and Arts.

Table 6 Prejudice according to academic discipline
Full size table
Fig. 1
figure 1

Groups per prejudice level according to academic discipline

Full size image

This brings us to the final research question, concerning the effect of GenSex education. The effect of time spent at the university was found to be negligible (r (6892) = .007, p = .57). Nevertheless, we believe that these results are relevant in the context of the rates of GenSex training per discipline. Students who had attended prior training on these issues were significantly less prejudiced. A considerable effect size was also found for previous GenSex training (Δθ = .66, 95 % CI (.60, .72), t (8181) = 21.38, p < .001, d = .51). However, this varied across disciplines. GenSex education was very effective at reducing prejudice in the Agricultural and Applied Social Sciences. It was moderately effective in the Health Sciences, Human Sciences, and Engineering and had no effect in the Exact and Geological Sciences, Biological Sciences, and Linguistics and Arts (Table 7). An interpretation of these results follows.

Table 7 Prejudice by attendance in training
Full size table

Discussion

Despite the intuition that education in general reduces prejudice, anti-LGBT attitudes are still present to various extents in even the most erudite areas (D'Augelli 1989; Finlay and Walther 2003; Herek 1993; Kjaran and Jóhannesson 2013). The majority of our sample demonstrated some degree of explicit GenSex prejudice. Yet despite the ubiquity of these attitudes at UFRGS, most students were unaware of any ongoing discrimination, reporting to have neither seen nor heard of humiliation, physical assault, or maltreatment towards homosexual (gay or lesbian), travesti, or transsexual students at the university. Indeed there is a 10:1 ratio between those who are completely ignorant compared to those who have witnessed any degree of victimization. Furthermore, given that the most prejudiced people were the least likely to detect acts of discrimination, these acts are likely to be even more prevalent than our results portray. The incongruence between these beliefs and reality reflect the heterosexist system that normalizes prejudice against LGBT individuals, making discrimination invisible to the majority of students.

Similarly, Nardi et al. (2013) have described the rituals of UFRGS second-year students, ridiculing ‘freshmen’ with homophobic and sexist insults, in order to affirm hierarchies. Meanwhile, examples of individual victims include the gay student who was verbally threatened because of his sexual orientation or the transsexual student who was repeatedly refused to have her name and gender identity recognized by the university personnel. Instances like these indicate that universities are not immune from this form of oppression, reproducing social inequalities and providing a breeding ground for prejudice. Academic institutions have the potential to play an important role in reforming these social ills; however, if we are to combat these problems, students must first be aware.

There was significant variation in prejudice levels across groups. In the case of religion, our results supported VanderStoep and Green’s (1988) hypothesis that religiosity predicts ethical conservatism, which in turn predicts prejudice against sexual diversity. This suggests that GenSex prejudice does not necessarily stem from religion per se but that it is derived from a generally conservative position on issues of personal morality. This is reflected in the lower prejudice levels among non-religious students, those who identify as religious but with low or absent religious attendance, and among believers of more liberal tenets (Tables 4 and 5). We shall address each group of religions in ascending order of prejudice.

In line with existing literature, Pagan, Neo-Pagan (de Sá Rocha and de Oliveira 2014), and Afro-Brazilian (Birman 2006) followers demonstrated the least prejudice of all denominations. In our sample, their scores were equivalent of those who identified as non-religious. This is consistent with ethnographic research, which indicates that discursive categories used in Afro-Brazilian cults often encompass alternate genders and sexualities without discrimination and prejudice (Fry 1986). It is also worth noting that it is not uncommon to find gender-nonconforming Afro-Brazilian religious leaders (Fernandes 2013).

Eastern Asian religions are over-represented by our sample as the Asian population in the state is proportionally smaller. Eastern Asian religions are the main destination of the Brazilian religious conversion, especially among Catholics (Oliveira 2011). These religions gained popularity in Brazil in the 1980s, during the expansion of the New Age movement, after re-democratization. The positively skewed prejudice levels could be attributed to the fact that the New Age movement overlaps with counterculture movements (feminist, ecological, LGBT).

Contrary to extant literature, which suggests a negative bias inherent in the tenets of Spiritism (Marmolejo 2007), Spiritists represented some of the least prejudiced participants in our sample. Spiritist groups in Brazil are often comprised of constituents who have multiple affiliations, such as Protestantisms, Catholic, and/or Afro-Brazilianisms (Camurça 2009). One explanation is that the general acceptance of diverse affiliations facilitates an acceptance of sexual and gender diversity.

Despite research in Rio de Janeiro suggesting a conservative ideology amongst Jewish Brazilians (Machado et al. 2010), our sample demonstrated relatively low levels of prejudice. It is important to recall that anti-Semitism research was the matrix of the modern study of prejudice (Young-Bruehl 1998). Therefore, it is possible that the Rio Grande do Sul Jewish community is more akin to their North American counterparts, who played an active role in empowering the LGBT community. Further research is required to explore the dynamics in these groups.

Described as the leading religious denomination in Brazil (IBGE 2012), Catholicism has become a relatively moderate community. Whilst many identify with the religion, the majority are not active practitioners. In line with this, Catholics in our sample were significantly less biased than other Christians, reflecting findings in international literature (Finlay and Walther 2003; Hinrichs and Rosenberg 2002). Despite their differences, the similarity of prejudice levels in Catholics and Muslims is striking. However, the sample size of the Islamic group is small, so this comparison should be taken carefully. Islam is a small and heterogeneous religious denomination in Brazil (Montenegro 2002), but this is starting to change with the recent immigration of West Africans to southern Brazil (Tedesco and Grzybovski 2013).

At the other end of the spectrum, Mormon, Neo-Pentecostal, Protestant, and other Christian levels of prejudice are particularly disturbing. Extant literature differentiates between Progressive and Orthodox Protestants, where Progressives demonstrate significantly lower levels of prejudice, equivalent to more moderate religions (Barreto and de Oliveira Filho 2012). Whilst our methodology did not ask participants which faction they fell into, unimodal distributions suggest that there was no categorical division of prejudice levels in any of the religious groups. Further research is needed to clarify any finer divisions. Whilst Mormons were the most prejudiced of the religious affiliations in our sample, Neo-Pentecostal, Protestants, and other Christian groups also demonstrated high levels of prejudice. These churches are gaining media exposure in Brazil in response to their violations of Brazilian LGBT rights (Natividade and de Oliveira 2009). Neo-Pentecostal attacks on Afro-Brazilian religious communities for their gender and racial inclusiveness are not uncommon (da Silva 2007). However, it would be unfair to ignore the emergence of progressive Neo-Pentecostal communities that seek to be inclusive of sexual and gender diversity (Natividade 2010).

Central to the question of confronting GenSex prejudice, these data denote the need for stronger anti-discriminatory messages from the university, with particular emphasis on messages directed at more extensively prejudiced groups. It also drives home the conclusion that for LGBT rights to continue to develop in academic institutions, ecumenism must be reinforced.

One of the most striking findings of this research is the absence of correlation between the time spent in the university and the reduction of prejudice. However, an informative dynamic of our study lies in the interactions between prejudice, discipline, and the effects of GenSex education. As GenSex education at UFRGS is neither compulsory nor standardized, the level of critical thinking around GenSex issues is at the discretion of professors and departments. Given that the non-uniformity of GenSex education is an unreliable method of dealing with prejudice (Tucker and Potocky-Tripodi 2006), it is plausible that some disciplines teach GenSex courses that reinforce prejudiced perspectives rather than challenging them.

Haslam et al. (2002) investigated the relationship between anti-gay prejudice and beliefs about the nature of homosexuality in North America. The authors showed that heterosexist people tend to believe that homosexuality is a stigma that is categorically separate from sexual normality and grounded in chosen conduct rather than biology. That is, anti-gay attitudes are associated with anti-essentialist beliefs. This is especially evident taking into account the literature on attribution of choice to sexual orientation as predictor of anti-homosexual attitudes in North America (for a summary, see Whitehead (2010)). The opposite may be true in Brazil, according to a study by Lacerda et al. (2002), who investigated the relationship between ontological explanations of homosexuality across different academic disciplines. ‘Non-prejudiced’ and ‘subtly prejudiced’ participants were predominantly Psychology students who agreed with a non-essentialist psychosocial explanation of homosexuality and disagreed with moralistic or religious accounts. Conversely, medical and engineering students, who tended to have higher prejudice, agreed with essentialist biological, religious, and/or moralistic explanations and disagreed with the psychosocial ones (also see Pereira et al. (2013) and Araujo et al. (2007)).

This theme is apparent in our results as Health Sciences students lie at the more extreme end of the prejudice spectrum, whilst students in the Human Sciences show lower levels of prejudice in general. GenSex education in the Health Sciences is predominantly focused on sexual and reproductive health, unrelated to GenSex diversity (Rufino et al. 2013; Medeiros, et al 2014). Even though the Psychology Department (classified as a human science, according to Brazilian standards; CAPES 2012) offers courses for Health Sciences students, psychology in Brazil is still strongly influenced by psychoanalytical theories with a history of harmful GenSex curricula (Costa, dos Santos, Rodrigues, and Nardi 2009). This can also explain the moderately effective GenSex education among Health Sciences students.

Moderate effects for GenSex education were also found in Human Sciences, and no effects were found for Linguistics and Arts. Nevertheless, those two disciplines combine the majority of participants who underwent GenSex training and the lowest degree of prejudice. Conversely, although only 30%of the Applied Social Science students had been enrolled in some form of GenSex training and with an intermediate average level of prejudice, this group demonstrated the highest rate of improvement. Gender and sexuality studies in Brazil are particularly strong within Human and Social Sciences disciplines (Matos 2008). Therefore, the low basal prejudice and the effective training in those groups can be related to the combination of GenSex curricular activities based on a non-essentialist and inclusive approach.

Engineering, Agricultural, and the Exact and Geological Sciences are predominantly masculine careers in Brazil (Agrello and Garg 2009). Bilimoria and Stewart (2009) have highlighted overt hostility toward LGBT students from faculty members, LGBT invisibility, interpersonal discomfort, and pressure from peers to mask non-heteronormativity in a North American tertiary engineering context. Similarities have been cited in Brazil (Pinto and Maciel 2013). Potentially a contributing factor, Engineering and Agricultural sciences do not have compulsory GenSex courses at UFRGS. As such, the low prevalence of GenSex education in those disciplines (Table 7) may indicate that students who pursue those courses tend to have pre-existing prejudices that are reinforced by this environment. However, the pro-activeness required to pursue GenSex courses from within that environment suggests a willingness to be open to GenSex diversity and explain the effectiveness of GenSex education within that group.

An intriguing finding was the negligible effect of GenSex education for the Exact and Geological Sciences students. One explanation for this may be that Mathematics, Physics, and Chemistry students have to attend pedagogy courses in order to obtain a teaching degree in addition to the traditional, applied technical qualifications. Teaching is a role predominantly fulfilled by women in Brazil (75.4 %; Gatti 2010), and UFRGS Pedagogy Department possesses a highly active research group on GenSex in education. Consequently, lower prejudice would be expected among the ‘teaching’ subgroup, whilst higher prejudice levels might understandably be found in the ‘applied technical’ field. Thus, in this group, the benefits of specific education would be masked, which would explain our findings. This is supported by the bimodal distribution of the Exact and Geological Sciences students’ prejudice scores. The Biological Science group also presented no effect for GenSex training and a bimodal distribution of prejudice levels. Just as it occurs in the Exact and Geological Sciences, this group is made up of applied and teaching degree courses. However, as stated before, essentialist biological views about the origin of homosexuality appears to be associated with increased anti-gay prejudice in Brazil, which can reinforce general GenSex prejudice in this group and decrease the effect of training.

Beyond speculation, what these results allow us to conclude is that unsystematic GenSex education is an unreliable means of reducing prejudice in students at the institutional level.

Some limitations must be accounted for. The sample has some bias in the distribution across disciplines (Table 1). However, it is notable that there was a lower response rate in disciplines that showed high prejudice levels and a higher response rate in those with lower levels. This fact, along with the overrepresentation of LGB students (13.7 %), may indicate an overall underestimation of actual prejudice levels at the University. Moreover, the study was designed as an exploratory case study, which is to say that these findings have limited generalizability. Future efforts should aim to look beyond UFRGS and Rio Grande do Sul.

Conclusion and Policy Recommendations

The former Brazilian National Education Plan (PNE-2001) failed to emphasize GenSex education (Vianna and Unbehaum 2004). Regarding tertiary education, the plan requested the inclusion of GenSex curriculum guidelines solely for teaching degree courses in an aim to focus future anti-discrimination action at primary and secondary schools, preventing prejudice at an earlier stage. The majority of the action at a tertiary level consisted of courses for teachers in partnership with university research groups. Some of those courses were held in Porto Alegre, where UFRGS matriculated with local LGBT NGOs (Nardi and Quartiero 2012). However, action aimed at tertiary-level students remained overlooked. Furthermore, even action directed at primary and secondary levels is now in decline (Mello, Brito, and Maroja 2012). In the last 10 years, the Brazilian government has invested a great deal in so-called diversity policies. These policies aimed to ensure access in higher education and affirm the identity of groups such as afro-descendent, indigenous peoples, and people with disabilities. Unfortunately, the LGBT community has not received much attention under the umbrella of ‘diversity’ (Moehlecke 2009). The new National Education Plan (PNE-2014) was approved last year with all mentions of GenSex education removed due to pressure from religious members of congress, leaving the Brazilian educational policy without any federal objectives for GenSex anti-discrimination policy in the next decade.

We believe that this exploratory study provides grounds for a number of responses to GenSex prejudice at UFRGS and in the Brazilian educational policy in general. Given the lack of federal policy support, the university must make practical efforts to ensure the well-being of LGBT students. UFRGS already promises to guarantee their students’ rights and promote ‘respect for difference’ on campus (Institutional Development Plan 2010, p.11). The UFRGS Internal Regulations (2011) reaffirm these principles in Article 2, which indicates how vital ‘respect for differences’ is, and in Article 3, which assures respect for human dignity. Article 4 explicitly refers to discrimination (as an ‘action based on prejudice of any kind’). Finally, in Article 10 of the Student Disciplinary Code (2004), discrimination is classified as a severe offense. However, this needs to be supported by real conviction for action.

After this research was conducted, several Brazilian universities guaranteed transgender students the possibility to change their name and gender designation on all campus records and documents. This also occurred at UFRGS in response to a state law that conferred transgender identification rights in public institutions and pressure of the student's union and faculty. However, academic policy must guarantee other LGBT basic rights, such as the inclusion of LGBT identities in all university forms, the adoption of gender-neutral restrooms, locker rooms, and housing, and a clear anti-discriminatory agenda.

Students must also be made aware of prejudice on their campuses, whether this is through advertising campaigns, events, or student initiatives. Heterosexist values need to be confronted and on the terms of those who face this stigma. One example can be found at the University of Ontario, where the student union has generated a submission to the Canadian government, petitioning support to attain universal healthcare and mental health services for students, with a focus on minority access (Pin and Martin 2012). Furthermore, these efforts need to be cognizant of groups with high prejudice, as we have outlined in this study, which should be targeted when resources are being allotted and messages directed.

Supporting the existing literature (Paluck and Green 2009), our results suggest that a step towards better GenSex education involves the production of compulsory, standardized courses for all students, regardless of discipline. These may be assessed and adapted over time to ensure their effectiveness. In light of Egry’s (1985) findings, we believe that these courses will be most effective when supervised by a professor, but run by students. This should foment a non-violent, diverse, and safe academic environment for LGBT students, where difference will be valued instead of ostracized.

Undergraduate education produces long-lasting effects on how students understand and operate in society (Gurin et al. 2002). The ethical and political duty of faculty to facilitate a society free from prejudice and discrimination is part and parcel in the charge of academic production and education.