International Women's Day 2022 finds the women-in-STEM data landscape at a moment that resists both optimism and pessimism without evidence. Two years of pandemic disruption have complicated the longitudinal picture — data collected in 2020 and 2021 reflected conditions that were not normal, and attributing trend changes to structural factors rather than pandemic-era distortions requires care. At the same time, the pre-pandemic evidence on women's STEM participation had reached a level of detail and international coverage that allowed for more precise diagnosis than had previously been possible.
This analysis draws on the sources available as of the first quarter of 2022: UNESCO's annual Women in Science factsheet (released around International Women's Day), the OECD's Education at a Glance 2021, and NSF's 2021 Survey of Earned Doctorates. Charts referenced below are described in prose — the data is from these sources and readers are encouraged to consult them directly.
Chart 1: The Global Research Gap
UNESCO's 2022 Women in Science factsheet reported that women represented approximately 33 percent of researchers globally — a figure that had improved slowly from approximately 29 percent a decade earlier. Progress was real but slow; at the current rate, achieving research-workforce parity would take several more decades. The regional variation was the more analytically important story: Latin America and the Caribbean at approximately 46 percent, Central Asia at approximately 48 percent, Sub-Saharan Africa at approximately 33 percent, and South and West Asia at approximately 20 percent described a world where women's research participation reflected vastly different structural conditions.
Chart 2: The Degree Funnel
OECD Education at a Glance 2021 showed women earning the majority of bachelor's and master's degrees in OECD countries in aggregate — approximately 56 percent of bachelor's degrees overall. In natural sciences, mathematics, and statistics, women were close to parity at the bachelor's level. In engineering, manufacturing, and construction, women earned approximately 26 percent of bachelor's degrees. The degree-level gender gap was concentrated in the engineering and ICT fields; in the life sciences, women often exceeded men.
The transition from bachelor's to doctoral degree showed a consistent narrowing of women's representation across fields. In STEM fields where women were near parity at the bachelor's level, they were substantially below parity at the doctoral level. The mechanisms for this transition-stage attrition were consistent with those documented in the pipeline literature: the doctoral period's demands intersecting with the early period of relationship formation and the beginning of career-domestic responsibility trade-offs.
Chart 3: Doctorates by Field
NSF's 2021 Survey of Earned Doctorates provided the most detailed field-level breakdown available for the United States. Women earned the majority of US doctorates in biological and biomedical sciences (approximately 53 percent), psychology (approximately 75 percent), and health sciences. In physical sciences, women earned approximately 37 percent of doctoral degrees. In engineering, approximately 24 percent. In computer and information sciences, approximately 21 percent.
The variation within engineering and computer science by subfield was not fully captured in the top-level figures. Biomedical engineering had substantially higher women's doctoral representation than mechanical engineering. Within computer science, certain subfields showed higher women's participation than the aggregate CS figure; others were dramatically lower. The aggregate masked the important variation.
Chart 4: The Faculty Pipeline
The gap between doctoral degree attainment and faculty representation was visible in every field. Women who earned doctoral degrees in STEM fields were converting to tenure-track and tenured faculty positions at lower rates than men at equivalent career stages. This was the career stage — from PhD completion to first faculty position — that the research literature identified as one of the two most consequential attrition points (the other being the mid-career transition to full professor and research leadership).
The mechanisms were not mysterious: the pool of entry-level faculty positions was smaller than the pool of qualified doctoral graduates; competition for positions was intense; evaluation criteria for faculty hiring were subject to the well-documented gender biases in research assessment; and the geographic mobility required to compete in the national academic job market was more costly for women who had partners, families, or care responsibilities constraining their mobility.
Chart 5: The International Variation in CS
One of the most counterintuitive findings in the international data was the variation in women's computer science representation across countries. Countries like Malaysia, India (in some institutions), and several Eastern European nations showed higher women's CS enrollment rates than the United States, the UK, or Germany. The US decline in women's CS degrees since the mid-1980s was a specific phenomenon, not a universal one — it reflected specific features of US CS cultural identity formation and computing's social construction in the US context.
The international variation was a reminder that "women in CS" was not a single global phenomenon with a single mechanism but a set of national and cultural phenomena that intersected with the global pattern in complex ways. Policy that worked in one national context might not transfer directly to another.
Chart 6: The Pay Gap in STEM
OECD data showed a persistent earnings gap between men and women in STEM occupations, with the gap varying considerably by field. In engineering and physical sciences, the gender pay gap was larger than the economy-wide average in most OECD countries. In biological sciences and health, it was smaller — but still present. The gap was not primarily explained by differences in qualifications: controlling for education level, experience, and field, a significant portion of the pay gap in STEM remained unexplained by observable variables.
Chart 7: Grant Funding and the Gender Gap
Research on gender and grant allocation in major funding agencies — NSF, NIH, the UK Research Councils, the European Research Council — had consistently found that women were awarded grants at slightly lower rates than men with equivalent qualifications and proposals, and that women's grants were funded at lower levels on average. The mechanisms included both explicit gender bias in review (reduced by double-blind review where implemented) and structural factors: women were less likely to apply to the highest-prestige funding mechanisms, possibly reflecting the same self-nomination gap documented in prize and award data.
Chart 8: The Senior Leadership Gap
Women's representation at the senior levels of scientific organisations — as directors of research institutes, as national academy members, as heads of funding agencies — lagged behind their pipeline representation by a margin that was not fully explained by generational lag. In most major national academies in OECD countries, women represented approximately 15 to 20 percent of members, well below their share of doctoral recipients in the fields represented.
Chart 9: The Pandemic Effect — Early Evidence
As of early 2022, the first peer-reviewed studies of the pandemic's effect on gendered research output were beginning to appear. Studies examining journal submission rates in 2020 found consistent increases in the male-to-female submission ratio relative to pre-pandemic periods across multiple disciplines. The effect was concentrated in preprint servers and journals covering fields where fieldwork was most disrupted. Whether the submission gap translated into a citation and career-trajectory gap over longer time horizons was not yet established.
Chart 10: What Actually Moves the Needle
The final "chart" is a synthesis of the intervention evidence: which institutional actions showed the strongest evidence of changing gender-representation outcomes. The answer in 2022 was consistent with what it had been a decade earlier: structural interventions (blind review, structured hiring criteria, career-interruption accommodation, gender-disaggregated reporting with public accountability) showed the most consistent evidence. Individual fellowship and mentorship programmes were valuable for the individuals they reached; they showed weaker evidence of systemic change. The pipeline-entry interventions that had attracted the most public attention (girls' coding programmes, awareness campaigns) showed limited evidence of downstream career-trajectory effects.
Frequently Asked Questions
What percentage of the world's researchers are women as of 2022?
Approximately 33 percent, according to UNESCO's 2022 Women in Science factsheet, up from approximately 29 percent a decade earlier. Regional variation is substantial, ranging from approximately 20 percent in South and West Asia to approximately 46-48 percent in Latin America and Central Asia.
In which STEM fields are women most and least represented at the doctoral level?
Most represented: psychology (approximately 75 percent of US doctorates), biological and biomedical sciences (approximately 53 percent), health sciences. Least represented: computer and information sciences (approximately 21 percent) and engineering (approximately 24 percent).
What interventions have the strongest evidence for closing the STEM gender gap?
Structural interventions — blind review, structured hiring criteria, career-interruption accommodation policies, gender-disaggregated outcome reporting with accountability — have the most consistent evidence. Individual fellowship and mentorship programmes help the individuals they reach but show weaker evidence of systemic change.
← The NCWIT Scorecard | AAUW's 'The STEM Gap' Report: 2021 Earnings Data →