Underrepresentation of women in the senior levels of Brazilian science

Introduction

Although some progress has been made, there is evidence of a persistent gender gap in science (Shen, 2013) (see a special 2013 issue of the journal Nature: http://www.nature.com/women; Martin, 2014; McAllister, Juillerat & Hunter, 2016). All over the world, many women are deterred from pursuing a career in science to the most senior levels. Women scientists are still promoted less, win fewer grants, and are more likely to leave research than similarly qualified men. Whereas around half of those who gain doctoral degrees in science and engineering in the United States and Europe are women, only one-fifth of full professors are women (Handelsman et al., 2005).

Women represent less than 12% of the membership of the world’s national science academies, according to a study conducted by the Academy of Science of South Africa (ASSAf; Academy of Science of South Africa-ASSAf, 2016). An analysis of 69 National Science Academies revealed that the average proportion of members who were women stood at just 12% in 2013–14. Women were relatively more represented in the Academies of North America, Latin America and the Caribbean, with the Cuban Academy of Sciences having the highest proportion of female members (27%). Examining the membership of the science academies of Brazil, China, France, and the USA according to gender, it was found that the percentage of women varies between 6 and 14% (China 6%, France 11%, USA 13%, Brazil 14%; Academy of Science of South Africa-ASSAf, 2016).

Some progressive countries have recognized the need to address barriers to gender equality in order to retain their best scientists and innovators, and to ensure research excellence. Policies that focus on gender equality in science would also assist with making sure that social and economic returns on the investments made by taxpayers each year on training women scientists are not wasted. The Athena SWAN Charter is an accreditation and improvement program for higher education and research organizations focusing on gender and other forms of inequality, which was established in the UK in 2005 (Equality Challenge Unit (ECU), 2016; Munir, 2014). This program is proving successful in improving gender equality in terms of promotion and retention of women in science (Munir, 2014). Similarly, in Australia, the Science in Australia Gender Equity (SAGE) was created in 2015 in order to address problems related to gender equality (Science in Australia Gender Equity (SAGE), 2016). SAGE is an initiative of the Australian Academy of Science in partnership with the Australian Academy of Technological Sciences and Engineering, and was adapted from the Athena SWAN Charter (Munir, 2014).

Innovative national policies that affect research funding include gender balance targets (e.g., Slovenia, Switzerland) and legislation on gender quotas of up to 40% of the minority gender on committees (e.g., Finland, Norway). Some countries also have policies to increase university funding based on their performance in terms of gender equality (e.g., Ireland, Germany and Netherlands) (Boyle et al., 2015).

Aims of the Current Study

Here, we studied the gender distribution of: (a) the productivity scholarship holders (PSH) of the Brazilian National Council for Scientific and Technological Development (CNPq; Conselho Nacional de Desenvolvimento Cientifico e Tecnológico (CNPq), 2016a), (b) the members of the Brazilian Academy of Science (ABC; Academia Brasileira de Ciěncias (ABC), 2016), and (c) the amount of funding awarded for research (“Universal” Call of CNPq; Conselho Nacional de Desenvolvimento Cientifico e Tecnológico (CNPq), 2016b). We analysed the data according to gender, area of knowledge, and level in the research ranking system. More specifically, the goal of the present research was to offer a more comprehensive picture of gender imbalances in Brazilian science, by comparing the areas of Engineering, Exact Sciences, Earth Sciences, Life Sciences, and the Humanities and Applied Social Sciences. According to previous research, we expected a higher frequency of male researchers among scholarship holders and among the Academy of Science members. We also hypothesized that the gender gap would be less pronounced in the fields of Humanities and Social Sciences compared to Exact Sciences. The overarching goal of our research is to raise awareness of gender imbalance in science and stimulate to future policy changes that address will this problem.

Materials and Methods

The study sample included a total 13,625 productivity scholarship holders (PSH) of the Brazilian National Council for Scientific and Technological Development (CNPq); 4,859 from the area of Engineering, Exact Sciences, and Earth Sciences (ETEC), 5,687 from the area of Life Sciences (LS), and 3,079 from the area of Humanities and Applied Social Sciences (HASS), according to lists publicly available at the site of the agency at January 2016 (Conselho Nacional de Desenvolvimento Cientifico e Tecnológico (CNPq), 2016a). The study sample also included a total of 899 active members of the Brazilian Academy of Sciences according to lists organized by gender publicly available at the site of ABC at January 2016 (Academia Brasileira de Ciěncias (ABC), 2016). Their area of knowledge was identified according to Lattes curricula (unified system of academic information for Brazilian researchers and students), photos were checked, and CNPq research categories (1A, 1B, 1C, 1D, and 2) were determined for those who were productivity scholarship holders.

We analysed an additional sample of 3,836 researchers awarded funding for research in the UNIVERSAL MCTI/CNPQ CALL (No 14/2014), comparing the results in the three funding ranges by gender: less than 30 thousand Brazil reals (BRL; 3.12 BRL = 1 US dollar, USD), 30–60 thousand BRL and 60–120 thousand BRL. The call was open to all scientific disciplines and types of research, from basic research to applied research, including scientists who are not PS holders. Researchers with PhD > 7 years ago could apply for the lower funding range (<BRL 30,000 = USD 9,619). The intermediate funding range (<BRL 60,000 = USD 19,240) was open to Level 2 researchers. Level 1 PS holders could only submit proposals to the higher funding range (<BRL 120,000 = USD 38,480).

Results

For the CNPq’s productivity scholarship holders, we found that women are involved in all areas of research, but equally represented only in Humanities and Applied Social Sciences (50%, χ2 = 0.094, N = 3, 079, df = 1, p = 0.75), compared with Life Sciences (41%, Fχ2 = 171.992, N = 5,687, df = 1, p < 0.001) and Engineering, Exact and Earth Sciences (20%, χ2 = 1739.17, N = 4,859, df = 1, p < 0.001).

Gender distribution of the Brazilian academy of science

There were 126 female members (14%) of the Brazilian Academy of Science (ABC; χ2 = 465.64, N = 899, df = 1, p < 0.001). Examining the gender distribution as a function of areas of knowledge, using the classification of the National Council for Scientific and Technological Development (CNPq), the gender gap is more pronounced in Engineering, Exact and Earth Sciences (8.9%), and less pronounced in Life Sciences (20.4%), and Humanities and Applied Social Sciences (18.2%), (χ2 = 23.87, N = 899, df = 1, p < 0.001). Male scientists predominate in all areas of knowledge, and the representation of Humanities and Applied Social Sciences in the academy is low (Table 5).

Table 5:

Distribution of the members of the Brazilian Academy of Science by gender and area of knowledge (W, Women; M, Men; n, frequencies; AR, Adjusted residuals).

Area	W (n)	M (n)	Total n	W (%)	M (%)	AR W
Engineering, exact sciences and earth	44	450	494 (54.9%)	8.9	91.1	−4.9
Life sciences	76	296	372 (41.4%)	20.4	79.6	4.7
Humanities and applied social sciences	6	27	33 (3.7%)	18.2	81.8	.7
Total	126	773	899

DOI: 10.7717/peerj.4000/table-5

The percentage of women in the Brazilian Academy of Sciences does not correspond to the percentage of women among Productivity Scholarship Holders; it is two times lower. This occurs in all areas of knowledge (Fig. 1). For Brazilian Academy of Science members who were Productivity Scholarship Holders, there was a significant association between gender and scholarship level in Life Sciences (Fisher’s Exact Test = 10,08, p < 0.05), but not in Engineering, Exact Sciences and Earth Sciences (χ2 = 4,765, df = 4, NS). The small sample size in Humanities and Applied Social Sciences precludes detailed analyses. However, in LS, the number of women was greater than expected in the lower level of the research ranking system (1D), whereas the number of men was greater than expected in the higher level (1A, Table 6).

Table 6:

Distribution of the members of the Brazilian Academy of Science who are Productivity Scholarship Holders by gender and scholarship level (W, Women; M, Men; n, frequencies; AR, Adjusted residuals).

Level	W (n)	M (n)	Total n	W (%)	M (%)	AR W
Exact sciences and earth sciences
PS-1A	17	137	154	11.0%	89.0%	0.7
PS-1B	3	27	30	10.0%	90.0%	.0
PS-1C	0	13	13	0.0%	100%	−1.2
PS-1D	4	18	22	18.2%	81.8%	1.4
PS-2	2	42	44	4.5%	95.5%	−1.3
Total	26	237	263
Life sciences
PS-1A	20	99	119	16.8%	83.2%	−2.8
PS-1B	4	13	17	23.5%	76.5%	.0
PS-1C	7	11	18	38.9%	61.1%	1.6
PS-1D	9	13	22	40.9%	59.1%	2.0
PS-2	8	17	25	32.0%	68.0%	1.0
Total	48	153	201
Humanities and applied social sciences
PS-1A	4	4	8	50.0%	50.0%	.8
PS-1B	0	0	0	–	–
PS-1C	0	1	1		100%	−.9
PS-1D	0	1	1		100%	−.9
PS-2	1	1	2	50.0%	50.0%	.3
Total	5	7	12

DOI: 10.7717/peerj.4000/table-6

When divided into two age categories (30–50 and >50 years), there was a significant gender difference in both age categories in Engineering, Exact and Earth Sciences, and Life Sciences, and a significant difference between men and women in Humanities and Applied Social Sciences (Table 7). When we compared the proportions of women between the younger and the older category, we found a significant difference in the total number of women (χ2 = 22.35, df = 1, p < 0.001), and also specifically in Engineering, Exact and Earth Sciences (χ2 = 8.22, df = 1, p < 0.001), Life Sciences (χ2 = 11.87, df = 1, p < 0.001), while there was no difference in Humanities and Applied Social Sciences (χ2 = 0.45, df = 1, p = 0.50).

Table 7:

Percentages of men and women among the members of the Brazilian Academy of Sciences (ABC) divided into a younger group (30–50 years) and an older group (more than 50 years).

Age categories	Total N (%)		Engineering, exact and earth sciences		Life sciences		Humanities and applied social sciences
	Men	Women	Men	Women	Men	Women	Men	Women
30–50	140 (78.7)	38 (21.3)	78 (86.7)	12 (13.3)	60 (70.6)	25 (29.4)	2 (66.7)	1 (33.3)
χ2	58.45***		48.40***		14.41***		.33
>50	708 (91.1)	69 (8.9)	411 (94.9)	22 (5.1)	262 (86.5)	41 (13.5)	24 (82.8)	5(17.2)
	527.45***		350.46***		162.12***		12.45***

DOI: 10.7717/peerj.4000/table-7

Notes:

***p < .001.

Table 8:

Distribution of gender by amount of funding in the UNIVERSAL MCTI/CNPQ CALL-No 14/2014 in the major areas of knowledge: Humanities and Applied Social Sciences (HASS), Life Sciences (LS) and Engineering, Exact Sciences and Earth Sciences (ETEC).

W, Women; M, Men; n, frequencies; AR, Adjusted residuals.

Amount	W (n)	M (n)	Total n	W (%)	M (%)	AR W
Exact sciences and earth sciences
60–120 thousand reais	38	152	190	20.0%	80.0%	−1,6
30–60 thousand reais	76	231	307	24.8%	75.2%	,1
Less than 30 thousand reais	172	492	664	25.9%	74.1%	1,2
Total	286	875	1,161
Life sciences
60–120 thousand reais	112	218	330	33.9%	66.1%	−3,3
30–60 thousand reais	236	362	598	39.5%	60.5%	−1,5
Less than 30 thousand reais	503	593	1,096	45.9%	54.1%	3,8
Total	851	1,173	2,024
Humanities and applied social sciences
60–120 thousand reais	25	37	62	40.3%	59.7%	−1,8
30–60 thousand reais	86	78	164	52.4%	47.6%	,4
Less than 30 thousand reais	222	203	425	52.2%	47.8%	,8
Total	333	318	651

DOI: 10.7717/peerj.4000/table-8

Discussion

Our findings show gender imbalances in senior levels of Brazilian science, considering scientific productivity scholarships, grants and membership of Brazilian Academy of Sciences. Our analysis of a sample of 13,625 productivity scholarship holders (PSH) of the Brazilian National Council for Scientific and Technological Development (CNPq) showed that female scientists were more represented among PSH at the lower level of the research ranking system, whereas male scientists were more represented at higher senior levels (1A or 1B). This was evident in Engineering, Exact and Earth Sciences, and in Life Sciences. This result is in line with the gender gap in science found at senior levels worldwide (Shen, 2013; Handelsman et al., 2005). In Brazil, the gender gap in science was reduced in Humanities and Social Science compared to Engineering, Exact and Earth Sciences, which has also been found in other countries such as the UK (Boyle et al., 2015).

Our analysis of a sample of 3,836 researchers awarded funding by the Universal CNPq Call shows that female scientists were more frequently awarded lower amounts to conduct their research than male scientists. We did not have access to the amounts of funding requested by gender, and therefore we do not know if women applied for similar amounts. However, it is notable that there were no significant differences in the amount of funding awarded according to gender both in Humanities and Applied Social Sciences, and in Engineering, Exact and Earth Sciences, and Life Sciences. The smaller amount of funding awarded to female scientists was only found in Life Sciences (<30,000 BRL versus <120,000 BRL). In Biomedical sciences, women also get smaller grants than men in the US (Pohlhaus et al., 2011) and the UK (Bedi, Van Dam & Munafo, 2012).

There are relatively more women members of the Brazilian Academy of Science (ABC) compared to other academies of science (Academia Brasileira de Ciěncias (ABC), 2016). Nevertheless our study revealed several asymmetries in ABC. For example, the proportion of men is greater in all areas of knowledge (Engineering, Exact and Earth Sciences: 8.9% women (W) versus 91.1% men (M), Life Sciences: 20.4% W versus 79.6% M; Humanities and Applied Social Sciences: 18.2% W versus 81.8% M). The proportion of researchers working in Humanities and Applied Social Sciences is very low in the ABC (3.7%) in comparison to Engineering, Exact and Earth Sciences, and Life Sciences (54.9% and 41.4%, respectively). The proportion of women in the ABC is approximately half that of women among Productivity Scholarship Holders in all areas of knowledge (Exact Sciences and Earth Sciences: 8.9% versus 19.8%; Life Sciences: 20.4% versus 41.3%; Humanities and Applied Social Sciences: 18.2 versus 49.7%). Finally, among members of the ABC who were Productivity Scholarship Holders of the area of Life Sciences, the number of women was greater than expected in the lower level of the research ranking system (1D), whereas the number of men was greater than expected in the higher level (1A). As can be expected, most ABC members are holders of higher levels of PS, and thus the differences between men and women points to the general finding of gender imbalance in the senior academic positions. Our additional analyses including age of the ABC members support this notion showing that the gender gap is smaller among a younger category of individuals.

Our results are in agreement with previous research showing that in several scientific areas, in particular Exact Sciences, women are heavily underrepresented which can be result of several factors (e.g., Ceci & Williams, 2011), such as women’s self-exclusion from competition, and time out for maternity leave. For example, the Brazilian system of classification of scientists by CNPq does not take into account maternity leave (four or six months) or any other career breaks, which could be one of the reasons for the expected reduction in women in the higher classification levels, in particular at the beginning of their careers. In addition, the very fact of being in a minority may make women more likely to drop out. Moreover, we can speculate that scientific areas can be avoided by women because of great workloads and stress caused by such professions that is incompatible with family life. However, this would be true also for other areas, such as social sciences, veterinary medicine, medicine, or law where in some countries women dominate (e.g., Adamo, 2013). One of the plausible factors seems to be competitiveness and job insecurity in science (compared to medicine, for example, Adamo, 2013) that can be biased in many ways, such as recommendation letters (Dutt et al., 2016). In addition, and as supported by our data, it is important to remember that there may be generational effects; if the more senior grades are people who started careers a long time ago then there may have been different pressures and opportunities for them.

Another major reason for gender imbalances in science is that women more commonly tend to opt for areas such psychology and veterinary, while more men choose more traditional areas, such as engineering (for a meta-analysis, see Su, Rounds & Armstrong, 2009). This is due to many factors, such as conditioning from early ages, or cognitive differences between men and women (that can be caused by both biological and cultural factors; Su, Rounds & Armstrong, 2009). However, cognitive differences between men and women are quite small or even non-existent (for review see Halpern, 2012; Lippa, 2005), which means that most men and women overlap greatly in their abilities. The effect size of sex difference in career preferences is, by contrast, large (Su, Rounds & Armstrong, 2009), which points to the fact that differences in cognition only marginally contribute to the gender imbalance in academic representatives of many scientific fields. Thus, public opinion and popular literature exaggerating “natural” sex differences in cognition, and arbitrarily creating two distinct categories of men versus women can largely affect expectations of representation of women in science. For example, women scientists represent a minority in school science textbooks, and they are also invited less for conferences as plenary speakers (Martin, 2014). Changing this might start to change the public opinion that women are less competent for scientific jobs. Overall, a lack of role models and the public message of lack of competency decreases self-esteem and motivation of women to compete with men in numerous dimensions of scientific work, such as publications quantity (Cameron, Gray & White, 2013). As suggested by Ceci & Williams (2011), it is highly important to offer to women “realistic information about career opportunities and expose them to role models” to “ensure they do not opt out of inorganic fields because of misinformation and stereotypes”. Thus, early education should focus more on not biased information about scientific fields, leaving misinformation based on social stereotypes behind.

We show that there are great differences for women between Life Sciences and Engineering and Exact Sciences (ETEC), versus the Humanities and Social Science. It seems that during the career there is a significant drop of women in ETEC and most Life Sciences, while the gender gap is not much pronounced on the senior level of Humanities and Social Sciences. Again, this can be caused by many factors, similar to those discussed above. One study (Glass et al., 2013) compared adherence of women in STEM in general and other professional positions, showing that in many dimensions, such as family factors, these professional areas would be comparable. However, factors such as job satisfaction and in general motivation to remain active in the field, seem to be lower in STEM (Glass et al., 2013). Thus, the policy might focus on motivating women, and creating more women-friendly environments in STEM professions to reduce the exit rate.

Our data from the Brazilian Academy of Science contains the age distribution of their members. We were thus able to test whether the gender difference remains stable or changes through the age cohorts. We showed significantly higher proportion of women among the younger ABC members than among the older members. Thus, it seems that with lower age groups the gender difference even in the top scientific levels tends to diminish. This is in agreement with a new study showing that only recently Brazilian women publish half of the scientific papers, in comparison to the past (Elsevier, 2017). Although the situation seems to be improving, still the gender difference in all the scientific areas was significant among the ABC members.

The Brazilian government has programs, such as Programa Pró-Equidade de Gênero e Raça (Secretaria de políticas para as mulheres, 2015) (The Pro-Equity Gender and Race Program), and Observatório de Igualdade de Gênero (Brazil Observatory of Gender Equality) (Ministério das Mulheres, da Igualdade Racial e dos Direitos Humanos , 2015). They focus on people management and organizational culture to achieve equality between women and men in the labour market, political participation, on reduction of violence against women, but not specifically on gender inequality in science. The Brazilian Academy of Science has a leadership role in the country, contributing to the policy of Science, Technology and Innovation. The achievements of Brazilian women in science are discussed by the ABC, recognizing the importance of retaining them and making the best use of the investment made in training women scientists. Nevertheless, our research clearly shows that even with more female presence than in other countries, the imbalance is still evident.

Our research has several limitations. For example, we do not possess data on gender balance in application for Productivity Scholarships and other research grants. Thus, the present data do not show if lower numbers of women with grants are because fewer women apply, or because those who do apply are less likely to be funded. Because this would be a key factor for developing effective solutions to gender imbalance in science, future studies should focus on the difference between applicants and funded researchers.

Future research might focus not only on mapping gender differences in scientific or any other areas of human behavior, but also on specific mechanisms that lead to such gender gaps, which might help to better understand and in particularly to reduce the gender gaps. For example, Nosek et al. (2009) analysed Implicit Association Tests completed by citizens of 34 countries and found that nation-level implicit stereotypes associating science with males more than with females predicted nation-level sex differences in 8th-grade science and mathematics achievement. They suggested that implicit stereotypes and sex differences in science participation and performance are mutually reinforcing, and contributing to the persistent gender gap in science engagement. Thus, the differences in top scientific levels are just the tip of an iceberg that starts as early as during childhood with different expectations that the society has from men and women.

Overt sexism has decreased over the last few decades, but subtle gender biases maybe still be exhibited by both men and women, and are held even by individuals who consider themselves as egalitarian (Nosek, Banaji & Greenwald, 2002; Kite, Deaux & Haines, 2008; Moss-Racusina et al., 2012). Our results show clear evidence of imbalance of gender in Brazilian science, and this probably has deep institutional and cultural roots. We agree with Mühlenbruch & Jochimsen (2013) that research policies are needed and only wholesale reforms will bring equality to Brazil and other countries. The Athena Swan program in the UK and similar initiatives in other countries provide important examples of how long-term progress can be achieved (Munir, 2014; Science in Australia Gender Equity (SAGE), 2016).

Supplemental Information