Participation of women in science in the developed and developing worlds: inverted U of feminization of the scientific workforce, gender equity and retention

Introduction

Advances in the realms of science and technology, as well as progress toward gender equity, are often touted as key indicators of a nation’s development. In this context, it is of great concern that in most countries, women constitute a minority in the scientific workforce. Recent decades have seen increasing attention being focused on this issue: its causes, effects and possible ways to redress the imbalance. However, much of this attention has been concentrated on the developed world, especially the USA and countries in Western Europe (see, e.g., [1], [2], [3], [4], [5], [6], [7]). Comparatively little attention has been paid to the particular challenges faced by women in science in the developing world. These women could possibly be viewed as being ‘doubly disadvantaged’: in addition to frequently facing gender discrimination, they have to deal with a severe lack of scientific resources and infrastructure, and possibly prejudice. Such a lack of resources is also faced by the male scientists in their countries [8, 9] but women often have to deal, in addition, with complex sociocultural and socioeconomic issues, as well as entrenched patriarchies, that make their struggle to achieve excellence in their scientific careers even more difficult. The work presented here was motivated by a desire to investigate some of the factors impacting the participation in science of women in the developing world; in order to address this question, it also became necessary to compare and contrast their situation with their counterparts in developing countries.

We first define the criterion we will use to divide the countries of the world into those that are ‘developing’ and those that are ‘developed’. In this article, we will follow the convention that developing countries are those which are not among the group of 83 countries listed as a ‘High Income Economy’ by the World Bank [10]. We note that being economically developed/developing need not necessarily coincide with being developed/developing in terms of the scientific workforce and institutions in that country. For example, a country like India, which has a large scientific workforce, several renowned educational institutions in the STEM (Science, Technology, Engineering and Mathematics) fields and a robust scientific tradition [11, 12], is labeled as a developing country; in contrast, a country like Panama, which is comparatively lacking in these respects [13], is classified as a developed country.

Similarly, the fact that a country is classified as ‘developed’ need not necessarily signal that there is a high degree of gender equity in that country, or vice versa. One way of quantifying gender equity would be to look at values of the Gender Development Index or GDI [14]. The GDI is a ratio that measures the disparity between women and men in three basic regions of human development, viz., health, knowledge and standard of living. A value of GDI = 1.0 indicates perfect parity between women and men in that country in these respects, while smaller/larger values indicate that men/women perform better on these development indicators; in the latest table [14], the values range from 0.747 for Pakistan to 1.043 for Qatar. Of the 10 countries in the world with the highest GDI, five are classified as developed countries, and five as developing countries.

There are various factors that can make life difficult for women scientists in the developing world. These challenges can be grouped under the following headings: (i) they may be few in number, and therefore have to deal with a lack of peer group and mentors, (ii) they face pressures due to family responsibilities, (iii) they face discrimination and bias (iv) they experience geographic and scientific isolation, (vi) they frequently encounter hostile work environments, including sexual harassment, (vii) they suffer from a lack of recognition, and (viii) they have a lack of access to resources. Some of these challenges are also faced by female scientists in the developed world, or by male scientists working in developing countries. However, the extent to which these factors impact men and women, or those working in the developed world vs. those in developing countries, are usually different. We will briefly discuss each of these factors below.

The fact that women typically constitute a minority of the scientific workforce of a country, will be a major focus of this paper, and we will therefore discuss it at length later. However, we mention that this is an area where several widely held preconceptions, regarding the worldwide situation of women in science, turn out to be erroneous. It need not necessarily be true that more developed countries have a higher percentage of women in science – in fact, one frequently finds that the reverse is true. Moreover, certain areas of the world or cultures, which, in the popular understanding, are often seen as particularly regressive in the arena of women’s rights, often turn out to have a large proportion of female scientists.

The pressures due to family responsibilities are responsible for a lack of retention of women in science, with typically many women dropping out after obtaining a higher education in science. We will see below that this appears to be a more serious problem in many developing countries, possibly because of different cultural expectations about the societal role of women, especially post-marriage. Interestingly, the burden of housework may be almost as onerous for women scientists in the developed world, as for those in the developing world: in a Global Survey of about 15 000 physicists, when asked “who does the majority of housework in your family?”, the answer “myself” was given by 33 % of women and 8 % of men in developed countries, and by 40 % of women and 8 % of men in developing countries [15].

Discrimination and bias, either explicit or (more insidiously) implicit, is a fact of life for many women in science. Some anecdotal evidence, as well as some earlier studies, suggest that gender stereotyping and the idea that women are intellectually incapable of performing at a high level in science and mathematics seem to be more prevalent in parts of North America and Western Europe, than in many developing countries [16, 17]. In this context, it is interesting that so far, the only woman to win the Fields Medal in Mathematics has been Maryam Mirzakhani, who grew up and had her undergraduate education in Iran, which is a developing country. Some women scientists from developing countries claim that their very appearance (especially if they wear traditional clothing such as hijab or saris) leads to their not being taken seriously (e.g., at international conferences) by a scientific establishment that is dominated by people who predominantly look unlike them, in gender, skin color and dress. In many parts of the developing world, women face additional barriers such as local attitudes hostile to women studying or working outside the house. All of this is, of course, apart from the fact that for women in impoverished or rural areas, or from disadvantaged castes or indigenous groups, even subsistence or a basic education can be a problem.

The lack of geographical mobility is a particular problem for women from developing countries who aspire to careers in science, and has been identified as one of the major factors preventing them from progressing up the academic ladder [18]. In many developing countries, foreign experience (ideally in a developed country) is considered essential in order for career progression. This places women who are unable to travel, due to either childcare responsibilities or cultural taboos, at a considerable disadvantage.

Sexual harassment is only just starting to be acknowledged as a widely prevalent problem, even in the developed world [19], [20], [21], [22]. Women in many developing countries have reported that this is a severe problem with inadequate societal awareness of its seriousness, and few recourses available to victims, for complaints and corrective action.

There are a few awards that specifically recognize the scientific achievements of women in the developing world. These include the OWSD-Elsevier Foundation awards for Early Career Women Scientists in the Developing World; to qualify for these, a woman must have lived and worked in a ‘science-and-technology-lagging country’. While the L’Oréal-UNESCO Prizes for Women in Science are not reserved for women in developing countries, one prize is given to a woman scientist from each of five regions of the world, and thus often these prizes too (in particular, those for Latin America and the Caribbean, the Asia Pacific Region, and Africa and the Arab States) often recognize the work done by women scientists from developing countries.

For scientists to advance in their careers, they need advice and support from both peers and mentors, and in the absence of such a local female cohort, women in science can often feel isolated and bereft of advice. Various organizations have been set up that try to provide networking and mentorship opportunities, both at the national and international level. The Organization for Women in Science for the Developing World (OWSD), based in Trieste, Italy, is a program unit of the United Nations Educational Scientific and Cultural Organization (UNESCO) that aims to increase the participation in science and technology of women in developing countries. It also aims to bring greater recognition of their achievements, and to promote the use of science and technology in advancing women’s development. Several developing countries have local chapters of OWSD that conduct workshops, conferences and educational outreach activities. COACh is a grass-roots organization, initially established in the USA, that has in recent years extended its activities to developing countries, by conducting career building workshops for women scientists. Similar Career Development Workshops for Women in Physics (primarily from developing countries) have been conducted, since 2013, at the Abdus Salam International Centre for Theoretical Physics in Trieste, Italy. These workshops give women scientists from the Global South a space in which to share their experiences, opportunities to meet female achievers in science, and advice on topics such as writing scientific papers, giving presentations, conducting negotiations and dealing with hostile work environments. A few such similar initiatives and programs exist in other fields and other countries.

In the rest of this paper, we will examine numbers contained in publicly available databases, so as to extract patterns related to the situation of women in science worldwide, with a particular focus on developing countries. We will show some (possibly surprising) correlations that we have found between the participation of women in the scientific workforce and the economic development of a country. We note that though we will point out certain patterns and trends, and speculate on their origin, no efforts are made in this preliminary study to prove causation. We will also look for correlations between the percentage of women scientists in a country and the prevailing degree of gender equity, as quantified by its GDI. Finally, we will show that patterns of retention of women in the scientific pipeline appear to be different in developing and developed countries, and also exhibit regional variations.

What do the numbers tell us?

The situation of women in science in the developing world is a complex one, that cannot entirely be captured by listing statistics. However, we will see whether any trends or conclusions can be drawn by examining some of the numbers available. Several facts and trends that thereby emerge might seem, initially, quite startling, in that they contradict widely held preconceptions and beliefs.

It is unfortunate that as yet, no database is available that provides comprehensive statistics about women in science in the developing world; one therefore has to make do with somewhat scanty and unevenly distributed information. One source that enables cross-country comparisons of women in science, to some extent, is provided by the UNESCO Institute of Statistics. It lists the percentage of women (PW) among R&D researchers in most countries [23]. It should be noted that these statistics contain only limited information. It has not been specified exactly how the PW is defined, and (more importantly) whether it is defined in the same way when collecting statistics for each country. For some countries, they include not just women in the STEM disciplines, but also those in the social sciences. Moreover, there are reasons to believe that women, to a larger extent than men, are often either over-qualified for the jobs in which they are employed, or are working in part-time jobs rather than full-time jobs. The UNESCO statistics do not contain information about such disparities. Nevertheless, in the absence of more complete data, we will use these data to make comparisons across countries, with the caveats mentioned above.

According to the UNESCO data, in the world as a whole, women constitute 29.3 % of scientists (for conciseness, we will employ the word ‘scientist’ as shorthand for ‘staff member involved part-time or full-time in R&D research in a STEM field’). However, there are large regional variations: the highest percentages are seen in Central Asia (48.2 %), followed by Latin America and the Caribbean (45.1 %), the Arab States (41.5 %) and Central and Eastern Europe (39.5 %). It is worth noting that (with the exception of some of the oil-rich countries in the Persian Gulf) most of the countries in these areas would be labeled as ‘developing’. Western Europe and the USA, in contrast, have a relatively low percentage (32.7 %) of women scientists; this percentage is almost the same as that in sub-Saharan Africa (31.8 %). By far the lowest percentage is found in South and West Asia (18.5 %).

Let us define gender parity as a situation where women constitute 48–52 % of the scientific workforce. According to the UNESCO statistics, there are only 13 countries where this situation of parity exists: Uruguay, Malaysia, Moldova, Paraguay, Mauritius, Cuba, Bulgaria, the Philippines, Trinidad and Tobago, Serbia, Lithuania, Panama and New Zealand. One is immediately struck by the fact that most of these are developing countries. Perhaps even more interesting is the case of the 14 countries where the proportion of women exceeds parity. In 10 of them, the percentage is only slightly (52.1–55.4 %) over parity; these are Armenia, Latvia, North Macedonia, Kazakhstan, Kuwait, Georgia, Argentina, Guatemala, Thailand and Tunisia. The remaining four are Mongolia (57.5 %), Azerbaijan (59.0 %), Venezuela (61.4 %) and Myanmar (75.6 %). Once again, it is glaringly obvious that (with the exception of Latvia and Kuwait) all of these ‘above parity’ cases would be classified as developing countries. However, it is also true that the 9 countries with the lowest percentages of women scientists are developing countries, with Nepal having the lowest percentage at 7.8 %. The 10th lowest position (16.2 %), however, is occupied by Japan, a developed country with a strong and advanced scientific tradition [24, 25].

Inverted-U correlation between wealth of a nation and women’s participation in the scientific workforce

Above, we have seen that two clear messages emerge from the UNESCO numbers: (i) developing countries do not necessarily have lower percentages of women in science than developed countries, (ii) as far as numbers go, developing countries display varying scenarios. Can we gain further insights by searching for correlations between these data and other numbers? In Fig. 1(a), we have plotted, for 138 countries for which data were available, the percentage of women scientists (PW) in the country [23] vs. its per capita Gross Domestic Product, adjusted to reflect Purchasing Power Parity (GDP-PPP); the latter was obtained from data tabulated by the International Monetary Fund (IMF) [26]. For a few countries, for which the UNESCO data on PW were unavailable, information was obtained from other sources [27], [28], [29]. On looking at Fig. 1(a), we see that there is of course some amount of scatter in the distribution of points, reflecting a complex mix of social, cultural and political factors, and also, especially in the case of some of the less wealthy nations, the large variations associated with small numbers (the inherent spread associated with having a small overall scientific population in these countries). Nevertheless, we see that on the whole, the data points for most of the countries fall within the band defined by the light-blue shaded ‘boomerang’ in Fig. 1(a). This boomerang forms an asymmetric inverted U: the poorest nations have a very small percentage of women scientists, but as the per capita wealth of a nation increases, this percentage rises sharply. The countries where parity is achieved, or even slightly exceeded, mostly have a GDP-PPP between $10 000 and $20 000. Then, as the per capita wealth of a nation increases, the percentage of women scientists falls again. This is a startling pattern that does not appear to have been noticed earlier.

Fig. 1:

Scatter plots showing the percentage of women working in science in a country (PW) vs. the per capita GDP of the country in US Dollars, adjusted for purchasing power parity.

(a) Data for 138 countries throughout the world for which data were available, the light blue shaded inverted-U-shaped boomerang is a guide to the eye, which contains most of the points. Outlier countries, which fall outside the blue-shaded boomerang, are labeled: MM (Myanmar), JP (Japan), KR (South Korea), IL (Israel), CA (Canada) and SA (Saudi Arabia). (b) Data for countries in Europe. The blue points correspond to countries which were in the former ‘Eastern bloc’ and the red points correspond to the remaining ‘Western bloc’ countries. The dashed line is a linear regression to the data. The data for PW were obtained from Ref. [23], and those for the per capita GDP-PPP from Ref. [26].

Fig. 1(a) is notable, for three things: (i) the overall trend displayed, (ii) the vertical spread within the boomerang, (iii) the outliers that fall outside the boomerang. All three of these features merit further investigation that could, in future, be leveraged to guide policies to improve the participation of women in science.

We first discuss the overall trend we have found, i.e., the inverted U-shaped boomerang. This is a rather surprising result. Several previous studies that examined the participation of women in the labor workforce as a whole, found that the data fell onto a (not inverted, but right-side-up) U-shaped curve (see, e.g., [30], [31], [32], [33], [34]). In development economics, this is sometimes referred to as the feminization U hypothesis [35]: it states that as economic development increases, female participation in the labor force initially falls (reflecting a transition from a primarily agrarian to more industrial workforce) and then rises again (due to declining fertility rates and rising educational levels).

The initial rise seen in Fig. 1(a) is not necessarily incompatible with the feminization U hypothesis: in the poorest economies, if an increase in wealth enables women to move away from agriculture-based jobs, this could conceivably free them up to move into other professions such as science-based ones. However, it is harder to understand why, when the trend observed for the participation of women in the labor force as a whole is that it rises with economic development (in the richer economies), we see the opposite in Fig. 1(a).

The fact that we observe the reverse trend suggests that there is something in the nature of the scientific enterprise, or more likely in the culture that has grown to be associated with it, that sets it apart from other forms of labor. To examine this further, in Fig. 1(b), we have re-plotted the data shown in Fig. 1(a), but this time only for the countries in Europe. We note that the scatter of points is now less, possibly because of larger numbers of people in the scientific workforce of most European countries, as well as comparative homogeneity of some sociocultural factors across European countries, allowing the economy-determined trends to dominate more strongly. We see that in Europe, all the data fall within the region of the boomerang in Fig. 1(a) where there is an inverse correlation: i.e., in general, the higher the GDP-PPP of a country, the fewer women scientists (as a percentage) it has.

How can one understand the inverse correlation displayed in Fig. 1(b)? One possible explanation could be that this is a legacy of earlier political systems in these countries. All the countries with the lowest GDP-PPP (and, hence, developing countries) within Europe are the Eastern European countries that were formerly part of the Communist/Socialist bloc; as the blue dots in Fig. 1(b) show, these are also the countries that tend to have the largest proportion of women scientists. Communist and Socialist ideologies that emphasized gender equality, backed up by factors such as the widespread availability of state-sponsored childcare facilities, led to large numbers of women in these countries working in many fields (including science) that had earlier been strongly male-dominated [36, 37]. The differing position of women in science in Capitalist and Communist cultures was perhaps best exemplified by the contrasting situations in East and West Germany before Unification: the former had a much larger percentage of women scientists than the latter; after reunification, the percentage of working women in general (including women in science) in the former East has fallen [38], [39], [40], [41], [42]. However, this does not explain why one should observe this inverse correlation also within the countries of the former Eastern bloc or within the countries of Western Europe. It is interesting to note that Eastern Europe continues to have a higher participation of women in science than is seen in Western Europe, despite the fact that women’s labor force participation in Eastern European countries declined by about 20–25 % in the post-Communist era [43]. It is also notable that the Scandinavian countries, which are noted for being extremely progressive in terms of having gender-equitable policies [44], do not have a particularly high proportion of women in science.

Two other possible explanations can be advanced for the inverse correlation seen in Fig. 1(b). One is that in the relatively wealthier countries, there is less need for women to work in order to supplement the family income. This, however, would not explain why a (right-side-up) U-shaped curve is found when examining the female participation in the labor force as a whole, for the EU countries [45]. The other possible explanation is that in the wealthier nations, the pursuit of science is more prestigious and/or more financially rewarding, and is hence reserved preferentially for men. In the conventional feminization U hypothesis, it is assumed that in highly developed economies, there is a greater percentage of women working in white-collar jobs where there is less social stigma associated with women working, resulting in a rise in the female participation in the workforce [35]. The fact that the reverse trend is seen in the participation of women in science in these highly developed economies, suggests that negative and discriminatory attitudes toward women may persist more strongly in the scientific community than in other areas of a white-collar labor economy.

Let us now consider the vertical spread within the boomerang. Comparative studies of countries that have roughly the same GDP-PPP, but fall at the top or bottom rims of the boomerang, can help shed light on what factors, besides economics, determine the participation of women in science. For example, it would be interesting to understand the reasons for the differences between Uruguay (GDP-PPP = $21 338, PW = 48.2 %) and Chile (GDP-PP = $23 455, PW = 33.1 %). It is possible that vertical spreads in PW reflect similar differences in other indicators of gender inequity within a country, such as the GDI. For the particular pair of countries that we picked as an example, viz., Uruguay and Chile, the GDI values are 1.016 and 0.962, respectively; if we were to rank the 138 countries considered here by their GDI, Uruguay has rank 6 while Chile has rank 77. This is consistent with our observation that Uruguay falls on the upper rim of the boomerang, whereas Chile falls on the lower rim. We will return to this point further below, where we will look for correlations between the PW and GDI.

Next, let us consider the outliers, the countries that fall well outside the boomerang. We see from Fig. 1(a) that there are five countries for which the PW falls well below the boomerang region. These are Japan (JP), South Korea (KR), Israel (IL), Canada (CA) and Saudi Arabia (SA). Case studies of these nations would be particularly interesting, to find out what the extra obstacles might be that are faced by women scientists here, compared to their peers in other countries with a similar economic situation. The values of GDI for JP, KR, IL, CA and SA are 0.976, 0.934, 0.972, 0.989 and 0.879, respectively. Out of the 138 countries considered, these five countries have rank 52, 97, 60, 32 and 113, respectively, in terms of GDI. Thus, for most of these outlier countries, the low value of PW seems to reflect a somewhat inequitable (though, with the exception of South Korea and Saudi Arabia, not strongly so) situation for women in various indicators of development. We note that it is well known that despite the relatively high GDI for Japan, women there face strong societal barriers to working in all sectors, not just in science [46]. We also note that the UNESCO tables do not provide data for the participation of women in science in Canada, these data were obtained from other sources [28]. Possibly even more interesting than the case of these five ‘under-performers’ is the situation of Myanmar (MM). At 75.6 %, it has by far the highest PW of all the countries, and lies well above the boomerang, despite having a relatively low GDI of 0.953 (rank 81). While this number is interesting, we note that it has been suggested that men are less likely to work in higher education in Myanmar, because of the low salaries associated with university jobs [47]. This is not particularly encouraging, i.e., one should therefore not look at Myanmar as a role model for how to increase the percentage of women in science. Indeed, it is reported that universities in Myanmar currently operate a policy of positive discrimination in favor of men, in an effort to ‘correct’ the fact that more women than men enroll for higher education courses [47].

In this context, it is also interesting to note that the naive expectation that the PW in a country is a number that always increases as a function of time, is not necessarily borne out. We have already noted that the fall of the Iron Curtain resulted in a drop in the percentage of women in science in the former East Germany (GDR). A particularly drastic drop has been reported in Lesotho, where the PW fell from 76 % in 2002 to 31 % in 2011 [48]; in the same time period the per capita GDP (note, this is the GDP and not the GDP-PPP, for which data were not available) rose from $441 to $1432 [49]. The latest year for which the PW is available is 2015, when the PW rose slightly to 36.4 %, and the per capita GDP fell to $1116. While this is just one set of numbers for just one country, from which no sweeping conclusions can be drawn, one does again seem to see a pattern of an inverse correlation between PW and per capita GDP. It has also been reported that some countries in Southeast Europe initially lost gender parity in PW after the breakup of the former Yugoslavia, but subsequently recovered it [48], and it would be interesting to try and understand the reasons behind this [50, 51].

In concluding this section, it seems reasonable to speculate that the reasons for the small PW in the initial (small GDP-PPP) and late (large GDP-PPP) parts of the inverted U boomerang may be different; we will find further evidence for this in Section 2.3 further below, when we examine statistics for retention patterns.

Correlation between women’s participation in the scientific workforce and gender equity

Next, we look for correlations between the percentage of women scientists in a country and the general economic, social and educational status of women in that country, as quantified by the GDI. In Fig. 2, we present scatter plots of the PW vs. GDI for the world as a whole, as well as individually for Europe, Asia, Africa and Latin America. The dashed red lines are linear regressions to the data. We see that in all cases, there is a positive correlation: the higher the GDI, the higher the PW, though the degree of scatter is large. This trend contradicts the findings of two much-discussed and controversial recent studies, which reported that there was a “paradox”, in that greater gender equality correlated positively with larger differences between female and male performance and/or participation in science [52, 53]. In the first of these studies, the authors reported that “gender differences were found to be strongly positively associated with economic development as well as gender equality” [52]. Their reporting of a positive association between gender differences and gender equality would appear to be contradicted by what we see in Fig. 2. While the positive association between gender differences in preferences and economic development, as reported by them, might appear consistent with our finding of a negative correlation between PW and GDP-PPP beyond a certain value [see Fig. 1(b)], the interpretation of this finding is open to debate. The authors of this earlier study implied that their results indicated the existence of innate differences in preferences of men and women for science-related subjects; they suggested that once a certain level of economic development was achieved, so that subsistence was no longer an issue, innate preferences played a dominant role in determining choices of careers [52]. In the second of these earlier studies [53], in order to quantify gender equity, the authors chose to use not the GDI, but an alternative index, the Global Gender Gap Index (GGGI). They reported that the higher the GGGI, the lower the percentage of women among STEM graduates; they also reported that “the sex difference in intra-individual strength in science was higher and favored boys in more gender-equal countries”. Once again, these conclusions seem contradicted by the trends observed by us in Fig. 2. We note that concerns have been raised [54] about the data and methodology used in this earlier study [53], and the authors issued a lengthy correction, stating (among other things) that their data measured the “propensity” of girls to graduate in STEM [55]. In a later response to the questions raised [54] about their paper, the authors stood by their findings, and hypothesized that “men are more likely than women to enter STEM careers because of endogenous interests” [56].

Fig. 2:

Scatter plots showing the percentage of working women scientists in a country (PW) vs. the gender development index (GDI) for the country, for countries in

(a) the whole world (b) Europe (c) Africa (d) Asia, and (e) Latin America. The dashed red lines are linear regressions to the data. In all cases, there is a positive correlation between PW and GDI, though the amount of scatter is large. The values of PW and GDI are taken from Refs. [14, 23], respectively.

In concluding this Section, we underline that the trends found by us are not consistent with those reported by Falk and Hermle [52], or Stoet and Geary [53], in that we find that, in general, the participation of women in science is higher in countries with greater gender equity.

Differing patterns in retention rates

One of the difficulties encountered by those trying to increase the participation of women in science is the problem of retention: typically, more women than men tend to drop out as one progresses up the academic ladder. This phenomenon is frequently characterized as the ‘leaky pipeline’ [57], with the percentage of women dropping steadily as one proceeds from elementary school to high school to undergraduate studies to postgraduate studies to permanent employment to high administrative posts. For example, for women in STEM fields in the USA, the numbers are 50 % (university graduates), 37 % (PhD students), 24 % (Assistant Professors), 12 % (Full Professors) and 6 % (National Academy of Sciences members) [58]. The general understanding is that this is not merely a reflection of changing temporal percentages, i.e., this phenomenon does not result merely from earlier historical under-representations of women in STEM fields, but primarily arises from a lack of retention of women who are already in the pipeline.

Is this pattern seen worldwide, and is its nature similar in developed and developing countries? Ideally, to obtain a definitive answer, one would perform longitudinal studies on large populations of male and female students who would be tracked over the course of their careers. In the absence of such data, some indication can be gained by comparing two sets of available data: the percentage of female science students at the tertiary level in a country (PS), vs. the PW that we have already discussed above [23]. If the PW is smaller than the PS, this indicates that more women than men are dropping out after studying science at the tertiary level, if the PW is larger than the PS, then the reverse is true. Data on the PS, for several countries, is available from an earlier report [48].

In Fig. 3, we have plotted bar charts of the PS (blue bars) and PW (orange bars), for all countries where data on both numbers were available. In Fig. 3(a), we present the data for developed countries. We see that the six countries for which the PS was highest (Oman, Kuwait, Bahrain, Brunei, Qatar and Saudi Arabia), all exhibit a similar pattern, with a significant drop from the PS (which is well above parity) to the PW. Such a marked drop is not exhibited by the remaining developed countries, with the exception of South Korea. In general, the drop is small, and in some cases, we even see a rise as we go from the PS to the PW (most notable for Lithuania, Latvia and New Zealand). The five developed countries in this table with the smallest PS have almost the same values for PS and PW. This suggests that problems in attracting women to science in these countries are likely to be more prevalent at earlier stages, and these are the stages that should be preferentially targeted when making efforts to increase the participation of women in science.

Fig. 3:

Bar charts showing the percentage of females among students studying science at the tertiary level (blue bars) and among women scientists in R&D in the country (orange bars), for (a) selected developed countries, and (b) selected developing countries. When the blue bar for a country is higher than the orange bar, this indicates a problem in retention of women in science at the post-tertiary stage. Data were obtained from Refs. [23, 48].

The scenario is quite different for the developing countries, as we can see in Fig. 3(b). For the majority of the countries for which data were available, there is a significant drop from PS to PW, i.e., there is a significant problem regarding retention of women in science at the post-tertiary stage. One does see a clustering of patterns by geographical regions, with some intriguing exceptions. For example, India, Bangladesh and Nepal all exhibit steep drops from PS to PW, but Sri Lanka shows almost no drop; the reasons for this merit further exploration. Earlier studies, as well as anecdotal evidence, suggest that the reason for the precipitous drop in the participation of women in science in India at the post-PhD level is related to cultural biases regarding the societal roles of women, rather than a belief that women were not intellectually capable of doing science. When questioned about the reason that they dropped out of science, or had a break in their education/careers, most women cited marriage and/or motherhood; they also mentioned ‘location change’ (usually arising from marriage to a spouse in another city) [59, 60]. It seems likely that similar factors operate in much of the Middle East and Africa (though again, Egypt and South Africa are interesting exceptions in the trend displayed). It is also worth remarking in Fig. 3(b) that some currently or formerly Socialist/Communist countries display either only a very small drop, or even a rise, from PS to PW.

Clearly, differential patterns in retention, and the reasons for them, are issues that are worth investigating further. The trends displayed in Fig. 3 imply that developing and developed countries display differing scenarios regarding the attraction of young women to science subjects, as well as the retention of women in science. Thus, any intervention policies should be tailored so as to be region-specific or country-specific. This becomes important in the context of recent intergovernmental efforts to build multi-country partnerships to solve the ‘problem’ of women in science.

Summary and conclusions

In this paper, we have examined trends displayed by statistics related to the participation of women in science, with a particular emphasis on women scientists in developing countries, and how their participation ratios differ from those in the developed world. We have seen that the participation of women in science displays an inverted U-shaped boomerang curve with respect to the economic development of a country: as per capita wealth increases, there is at first a sharp rise in the participation of women in the scientific workforce, followed by a gradual decline. As a result, the countries with the largest percentage of women scientists tend to be developing countries. This inverted-U is the opposite of the trend seen for women’s participation in the labor workforce as a whole, suggesting that there may be something in the nature of the working culture within scientific institutions that sets them apart from other forms of labor.

Within Europe, we see an inverse correlation between women’s participation in the scientific workforce and economic development, so that the developing countries in Europe exhibit the highest percentages of women scientists. This can be partly attributed to the legacy of Communist/Socialist ideologies in Eastern European countries. However, this does not explain the presence of this inverse correlation within the countries of the former Eastern or Western blocs. We speculate that discriminatory attitudes toward women in science may be more prevalent in richer countries, where the pursuit of science may also be more financially rewarding and/or viewed as being more prestigious.

We have found a positive correlation between the participation of women in the scientific workforce and the degree of gender equity in a country, as measured by the GDI. This trend appears to contradict previous reports [52, 53] that gender differences in science correlated positively with a greater degree of gender equity. These earlier findings had led to speculations that the low participation of women in science resulted from innate differences in the preferences of men and women for science, rather than arising from discrimination or inequity [52, 53].

We have found different countries exhibit different patterns regarding the retention of women in science, by comparing the percentage of female science students at the tertiary level with the percentage of working women scientists. We find that developing countries (with the exception of the oil-rich countries of the Persian Gulf and Brunei) do not, typically, display a significant drop in the participation of women in science on going from the tertiary education level to working women scientists. However, this is not true for the majority of developing countries, where there is a significant drop at this stage. This suggests that efforts to increase the participation of women in science may need different emphases in developing and developed countries, by targeting different stages and age groups. Thus, a wholesale borrowing, across countries and across cultures, of programs to promote women in science, may not lead to effective solutions. As an example of a case where such retention statistics have been influential in formulating policy, we note the example of India, where it has been identified that the greatest attrition of female scientific personnel occurs at the post-PhD stage. In response, the government has introduced programs specifically designed to bring back to science those women who have had a break in career.

In this paper, in discussing women’s participation in science, we have focused on their participation as a percentage of the overall scientific workforce of a country. This ignores several important factors that may still impact the performance of women in science: first, particularly in those developing countries where the size of the scientific workforce is small, the absolute number of women may be very low. This tends to be especially true for certain fields such as engineering, computer science, physical sciences and mathematics. Research has shown that women students in male-dominated cohorts are much less likely to complete their PhD degrees [61]. In some developing countries today, a female student may still encounter a situation where she has no female peers, not just in her class or university, but in the whole country. For example, it was only in 2018 that Marie Chantal Cyulinyana reportedly became the first woman from Rwanda to get a PhD in physics; one can find similar examples in other fields and other countries [62]. Second, even if the percentage of women working in science-related jobs is high, these women may be employed at relatively low-ranking positions, or only part-time. When one combines the previous two factors, there is a great lack of mentors and role models for young women in developing countries who wish to pursue a career in science.

We note that organizations such as the European Institute for Gender Equality (EIGE) have argued that closing the Gender Gap in STEM will have a positive impact on GDP’s within the EU, primarily through rises in employment [63]. It is not clear how to position our findings (of a negative correlation between GDP and participation of women in STEM) within this discourse. One would have to establish if there is indeed a causal relationship between women’s participation in the STEM workforce and GDP, if so in which direction it is, and (most importantly) whether increasing the number of women in the scientific workforce could indeed result in a reversal of the correlation found by us.

In conclusion, we hope that the trends discovered in this paper will help stimulate thoughts and formulate policies, and also, perhaps help destroy stereotypes that hinder the advancement of women scientists from the developing world.