Improving the representation of women in academia is critical for providing role models and mentors for women students and early-career researchers to improve retention1. A diverse workforce is required for population of recruitment and other important functional committees to ensure equal representation. Diverse role models are also required for education and outreach programmes in schools to avoid stereotype threat2. Only with gender parity can these three goals be achieved simultaneously within an organization without overloading women in senior positions with more mentoring, education, outreach, service and committee work than their male colleagues3,4,5.

In Australian astronomy, women comprise 30–35% of PhD students, and fewer than 20% at the highest professorial level6,7,8. Theoretical workforce models predict that if this status quo is maintained the fraction of women at all levels in Australian astronomy will remain below 30% for at least 60 years9. This lack of workforce diversity is self-propagating. Studies have shown that universities hire and promote more women and people from minority groups into faculty positions when these positions have been more frequently held by women or people from minority groups in the recent past10. Gender biases, exclusionary behaviour and stereotypes are more pronounced in organizations with fewer women, producing a ‘chilly’ culture that can impact the recruitment and retention of women11. These experiences lead to women having lower expectations than men about their ability to reach the highest levels within their organization12,13.

To break this cycle and to improve the pipeline for promotion of diverse candidates to permanent positions in Australia, the Australian Research Council Centre of Excellence for All Sky Astrophysics in 3 Dimensions (ASTRO 3D) designed an evidence-based approach to recruitment and retention that was implemented across its nine Australian universities from 2017 to 2022. ASTRO 3D contains 300 members across nine Australian partner universities and eight international partner organizations. The centre is organized into 11 science teams, covering studies of the epoch of reionization, identifying the first stars and galaxies, studies in galaxy evolution, local optical and radio surveys, and a theoretical galaxy formation and evolution programme. Within the ASTRO 3D membership, the term ‘women’ refers to all people who self-identify as women and the term ‘men’ refers to all people who self-identify as men. Estimates of the total numbers of people in the centre include members who self-identify as non-binary and those who prefer not to self-identify. As of January 2023, the centre contained 147 investigators (researchers), 32 postdoctoral researchers employed by the centre, 17 staff employed by the centre and 109 students at honours, masters or PhD level.

Figure 1a shows the percentages of women postdoctoral researchers, students and investigators in the centre over a 5 yr period. Before the introduction of our recruitment and retention initiatives in June 2018, the percentage of women remained approximately constant at 29-42%, depending on level. After the new initiatives were introduced, the percentage of women increased steadily in all categories from June 2018 to January 2023, with 56% (18/32) women postdoctoral researchers, 52% (57/109) women students and 50% (152/305) women overall as of 1 January 2023. This increase in women was not made at the expense of men in the centre: Fig. 1b shows the growth in centre members across the January 2018 to January 2023 time-frame. The growth in the membership fraction of women is due to a faster increase in the numbers of women investigators, postdoctoral researchers and students.

Fig. 1: Percentage of women and growth in personnel in ASTRO 3D over time.
figure 1

a, Percentage of women investigators, postdocs and students in ASTRO 3D over time. b, Number of women and men staff, students, postdocs and investigators in ASTRO 3D over time. Students include honours, masters and PhD students. Postdoctoral researchers include both postdoctoral fellows who work on independent research programmes and postdoctoral researchers employed for specific projects. Investigators are junior, mid-career and senior researchers affiliated with the centre. Staff refers to management, administration and education/outreach staff employed by the centre.


Centre of Excellence funding for ASTRO 3D is provided to the Australian-based universities within the centre, which hire postdoctoral researchers and recruit PhD students into the centre teams. A total of 55 postdoctoral researchers were recruited over the June 2017–June 2022 time-frame across nine Australian universities. Unless specified otherwise, for this Analysis, postdoctoral researchers include both postdoctoral fellows who work on independent research programmes and postdoctoral researchers employed for specific projects.

A total of nine individuals were recruited into advertised independent 4 yr fellowships located at five universities, and 39 into advertised 3 or 4 yr fixed-term project-based postdoctoral positions (that is, 48 advertised positions in total). The project positions required expertise specific to the projects, and these requirements were provided in the advertisement and the position description. All advertisements included wording stating a commitment to a diverse work environment, and/or a commitment to the principles of equal opportunity. We received a total of 1,348 applications, with an average of 34 applications per position. Despite the advertisement wording, we found no increase in the number of women applicants above the fraction of women in the field, with 32% of applications from people who identified as women on the job application. For the 48 postdoctoral researchers recruited from advertisements, 52% were women. Among these, 51% (20) of the 39 advertised project postdoctoral positions and 56% (5) of the 9 fellowships were filled by women. These outcomes were achieved without quotas through the implementation of a core set of recruitment practices at all nine universities. These core recruitment practices included having widely broadcast gender targets, mandatory implicit bias training, 50% women committee chairs, 50% women on selection committees and 50% women on short-lists (see Methods for further details).

Throughout the five years from 2017 to 2022, supervisors were also free to hire experienced researchers already located at their universities into positions without advertisement. A total of seven people were recruited into postdoctoral positions without advertisement. Of these, one (14%) was a woman. Figure 2 shows the gender balance of advertised (postdoctoral positions and fellowships) versus non-advertised appointments. It is clear that advertised appointments achieved a notable improvement in gender balance over non-advertised positions, and that our recruitment practices produced gender balance across the nine universities for the 48 advertised positions.

Fig. 2: ASTRO 3D postdoctoral applicants and appointments.
figure 2

a, Percentages of men and women applicants for all 48 postdoctoral researchers recruited through advertisements (39 project-defined postdoctoral researchers and 9 fellowships). b, Percentages of men and women recruited for the 48 postdoctoral researcher positions through advertised appointments. c, Percentages of men and women recruited for the seven non-advertised appointments. Over 50% women personnel was achieved through the advertised appointments through implicit bias training, 50% women on search committees, 50% women on short-lists and 43% women committee chairs.

PhD student recruitment contributed strongly to the achievement of 50% women personnel within ASTRO 3D over the 5 yr time-frame. The fraction of women students in ASTRO 3D is now significantly larger than the fraction of women students in Australian astronomy overall, which has remained constant since 2003 at 30–33% women9. In Australia, PhD students receive federal government or university-funded scholarships, supplemented by a limited number of international or corporate-funded scholarships. PhD selection committees are at the university, college or faculty level, with candidates ranked across the sciences or across all fields on the basis of undergraduate grades and previous research experience. Students who receive scholarships may choose their department, team, project and supervisor from among the many offered at each university. Potential supervisors in the same department effectively compete for students to work on their projects.

The majority of students enter their PhD programmes at the beginning of the southern hemisphere academic year in January, and PhD programmes last 3.5 yr. As of June 2022, ASTRO 3D had 152 students, with 151 students who self-represented as men or women. Figure 1a shows that the percentage of women PhD students in ASTRO 3D increased steadily from 33% in January 2018 to 53% in January 2023. The main increase in the number of women students came after the fraction of women postdoctoral researchers reached 50% and the fraction of women overall in the centre reached 40% in 2021. This ‘tipping point’ of 40% women is consistent with sociology theory of groups, which proposes that at ~35% representation minority members are potential allies, can form coalitions and can influence the culture14.

The increase in women PhD students choosing ASTRO 3D projects is associated with the rise in the number of potential women supervisors (Australia-based women postdoctoral researchers and investigators) from 33% in 2017 to 43% in 2022. The percentage of women students that chose women supervisors was 38% between 2017 and 2020 and 58% between 2020 and 2022. At the same time, the percentage of men students choosing women supervisors remained roughly constant (32% between 2017 and 2019 and 34% between 2020 and 2022). Women supervisors had a larger overall supervisory load, with women supervising 3.2 total students on average, and men supervising 2.4 total students on average between 2017 and 2022. This difference in supervisory load is primarily due to a larger fraction of women students choosing women supervisors than men supervisors; women students supervised 1.9 women and 1.3 men students on average, and men supervised an average of 0.95 women and 1.4 men students.


Recruitment initiatives to achieve diverse teams need to be paired with successful retention initiatives. Longitudinal studies tracking the progress of physics PhD students indicate that, in the past, women departed physics more frequently than men due to a lack of women role models who are seen to have a good balance between their family life and academic career, a dislike of the culture or atmosphere, and doubts that they would attain a senior position1. These concerns are likely to be relevant to Australian astronomy, which has fewer than 20% women professors9. Women in astronomy also experience sexism and microaggression more frequently than men15. Both sexism and microaggression must be addressed at the departmental level to ensure an inclusive and welcoming culture for all astronomers.

We introduced a range of retention policies focused on leadership development, promotion of work–life balance, partner recruitment, monitoring of workplace climate with actions, and reporting pathways for misconduct and action against the code of conduct as described in Methods. To assess the efficacy of these initiatives in the retention of women, we investigate the retention rates of men and women affiliated with the centre over 5 yr from Dec 2017 to June 2022. This time-frame is longer than the standard Australian PhD (3.5 yr) and standard ASTRO 3D postdoctoral researcher positions (3 yr). During the first two years of the centre (2017–2019), students and postdoctoral researchers were recruited. If students and postdoctoral researchers who were in the centre in 2018 or in 2019 remained in the centre in June 2022, this indicates a choice to remain in astronomy and to remain affiliated with ASTRO 3D. Figure 3 shows the percentage of men and women students and postdoctoral researchers in the centre in December 2017 and December 2018 who remained in the centre as of June 2022. In all categories, larger percentages of women were retained than men. The high retention of women students (55–58% women retained versus 37–48% men) was primarily due to women students obtaining postdoctoral positions at other universities within the centre after their PhD. Similarly, a larger percentage of women postdoctoral researchers were retained in the centre (67–70% women versus 55–69% men) due to women postdoctoral researchers obtaining postdoctoral positions or faculty positions within predominantly Australian universities. The retention rates of women postdoctoral researchers in ASTRO 3D are in sharp contrast to the broader Australian astronomy retention rates, where 62% of women and 17% of men in Australian astronomy at the junior postdoctoral level A are no longer in the Australian astronomy pipeline at the more senior postdoctoral level B9. These results strongly suggest that the ASTRO 3D recruitment, retention and culture initiatives have contributed to the retention of women in the centre, and have counteracted the retention problems that effect the broader Australian astronomy community.

Fig. 3: The fraction of women and men students and postdoctoral researchers retained in ASTRO 3D over time.
figure 3

The fraction of students and postdoctoral researchers in the centre in December 2017 and December 2018 who were still in the centre in June 2022. The proportion of women retained was larger than the proportion of men retained over these time-frames, suggesting that the retention and culture initiatives introduced have had an impact.

The importance of diverse leadership

Previous work on the importance of women leaders has been mixed. In some studies women managers appear to be correlated with the hiring of women employees16, while other work suggests that women managers have no effect17. We investigate whether the gender of the team leader has influenced the proportion of men and women within each of the 11 scientific teams in ASTRO 3D. ASTRO 3D contains six teams led by men and five teams led or co-led by women. A total of 10 out of 11 teams are observational astronomy teams focusing on extragalactic or galactic optical and radio surveys, with the remaining team focused on theoretical studies of galaxy formation and evolution. Figure 4a shows that the percentage of Australia-based women in women-led teams increased from 38% to 54% from 2017 to 2022, while the percentage of Australia-based women in men-led teams remained roughly constant (29% to 33%) over the same time-frame. Figure 4b shows that the women-led teams consistently hired a majority (50% or more) of women postdoctoral researchers between 2017 and 2022, while the teams led by men consistently hired substantially fewer (20–30%) women over the same time period. Women-led teams also contained more women investigators overall, providing a larger proportion of women collaborators, role models and potential supervisors for junior team members. A cultural audit of ASTRO 3D was conducted in 2023 by Dattner Group, with interviews of 22 members of the centre. During this audit, many interviewees noted the importance of seeing women in positions of leadership: “this demonstrates to young women that they do have a place and can be ambitious in driving to achieve their career aspirations”.

Fig. 4: Percentage of women in women-led and men-led teams in ASTRO 3D over time.
figure 4

a, The percentages of all women in women-led and men-led teams as a function of time. The percentage of women in women-led teams increased from 38% to 54% from 2017 to 2022, while the percentage of women in men-led teams remained roughly constant (29 to 33%). b, The percentage of women students, postdocs and investigators in each year of the centre for women-led or women co-led teams (left) and the percentage of women students, postdocs and investigators in each year of the centre for men-led teams (right). Students include honours, masters and PhD students. Postdoctoral researchers include both postdoctoral fellows who work on independent research programmes and postdoctoral researchers employed for specific projects. Investigators are junior, mid-career and senior researchers affiliated with the centre. The increase in women in women-led teams was primarily driven by the increase in women students and postdoctoral researchers. Women-led teams consistently hired 50% or more women postdoctoral researchers between 2017 and 2022, while the teams led by men consistently hired 20–30% women over the same time period.

Workplace climate and support

Building a supportive workplace climate requires a broad range of support for members with different backgrounds and different social and family needs. Intersectionality plays an important role in the experience of women in astronomy. Women of colour experience more sexism and microaggression than white women18. A 2019 report on the workplace for LGBT+ physical scientists prepared by the UK Institute of Physics, Royal Astronomical Society and Royal Society of Chemistry concluded that nearly one-third of UK physical scientists from sexual and gender minorities have considered leaving their jobs because of their workplace climate. In a recent study of 324 LGBT+ physicists in the UK, exclusionary behaviour was experienced by 24% of all respondents and 49% of transgender respondents19. To improve awareness and empathy around these issues, our training over 2017–2022 included implicit bias training, LGBTQIA+ awareness training, cultural awareness training and Indigenous awareness training. This training was supplemented by initiatives to support and promote opportunities for all centre members, regardless of background or gender. We implemented family-friendly core meeting and colloquium times (10:00–14:00), financial support for carers to attend conferences, financial support for people experiencing financial difficulty such as during Covid-19, mentoring circles including minority circles, equal access to professional development and leadership opportunities, equal access to nominations for awards and monthly centre-wide seminars on diversity and inclusion. We emphasize that diversity training alone is insufficient to change recruitment or retention20 and is more effective when responsibility for diversity is assigned to managers throughout an organization21.

The workplace culture within ASTRO 3D is focused on equity and diversity and the centre undertook a comprehensive, independently administered and analysed climate survey soon after the beginning of the centre in 2018, and again in 2020. The goals were to to (1) quantify ASTRO 3D culture, (2) identify areas for action and (3) measure the improvements resulting from these actions. Details of the climate survey methodology are provided in Methods. In aggregated feedback on a five-point scale, the percentage of members who agreed or strongly agreed with positive statements regarding ASTRO 3D leadership and collaboration was 81% in the 2020 survey, while for culture, values, equity, diversity and inclusion it was 83%. These values showed improvements from 72% and 78% respectively in the 2018 survey, indicating successful outcomes from the ASTRO 3D initiatives in diversity and inclusion described above. In particular, in the 2020 survey 90% of individuals agreed or strongly agreed that “I believe that ASTRO 3D is doing a good job to promote equity and diversity".

Sustaining a supportive workplace climate requires support mechanisms and communication channels that are tailored to each cohort, so that problems can be identified and mitigated early. ASTRO 3D has multi-institutional student and postdoctoral researcher committees, each with representation from all universities. The student and postdoctoral researcher committees have ex officio seats on the centre’s executive management committee. Students and postdoctoral researchers are encouraged to conduct surveys of their needs and issues, which are brought to the executive management committee, helping to identify problems and to find solutions. Students and postdoctoral researchers are brought together twice a year at an annual all-centre science meeting and an annual strategic planning retreat, and also meet for their own retreats. Students and postdoctoral researchers are empowered to apply for funding for workshops and training programmes to meet their needs. Additional training and professional development is offered by the centre management, and includes training on how to build collaborations, how to give effective presentations, scientific writing, peer review, advanced scientific programming, media training, how to have difficult conversations and leadership training (including women’s leadership and mindful leadership).


The ASTRO 3D Centre of Excellence achieved 50% women personnel over a 5 yr time-frame by implementing a suite of evidence-based recruitment and retention initiatives. These initiatives changed how postdoctoral researchers were recruited and helped to develop a supportive and positive culture. The increase in potential women supervisors, mentors and role models for students appeared to pass a tipping point, after which the number of women students enrolling in the centre increased significantly. The recruitment and retention initiatives were applied consistently across nine universities, indicating that the initiatives themselves, rather than the individual university environment or other local factors, were responsible for the increase in diversity with time.

We conclude that improvements in binary gender diversity in teams can be readily be achieved in postdoctoral and student cohorts through a top-down approach of (1) setting a diversity target with regular monitoring of progress, (2) selecting a diverse set of team leaders, (3) in-person diversity training for all organization members, (4) ensuring 50% women on postdoctoral selection committees and (5) ensuring 50% women on postdoctoral short-lists.

The retention policies and culture initiatives implemented in ASTRO 3D have retained women postdoctoral researchers and students at higher rates than men. Large national and international teams would benefit from the application of similar programmes to help achieve equal retention of people within teams, regardless of gender. While retention within a Centre of Excellence is different from the retention of untenured or tenured staff at a standard academic organization, the initiatives implemented can be applied to any organization to improve the academic culture and support of staff.

We find that diverse leadership of teams is crucial for improving the diversity within teams: women-led teams recruited and retained more women postdoctoral researchers, attracted more women students and worked with more women collaborators, while the converse was true for men-led teams. This suggests that the presence of women supervisors and role models is critical for attracting and retaining women.

While this work focused on binary gender diversity, true diversity requires the recruitment and retention of people from all cultural, ethnic and religious backgrounds, people of all socioeconomic status, people with disability and people from the LGBTQIA+ community. Reporting these types of demographic information is voluntary and requires people to feel comfortable reporting their membership of the LGBTQIA+ community and disability status. This voluntary reporting is now underway within ASTRO 3D and we aim to report our progress on improving representation across a broader range of diversity in a forthcoming publication.


The centre chose a top-down approach to gender diversity, with the advisory board, executive management committee and science management committees selected to contain at least 50% women. The centre had a female director with a male deputy director from June 2017 to June 2022. A total of 5 out of 11 of the science teams were initially led by women. ASTRO 3D set an ambitious gender target of 50% women by 2022. Accountability is important for achieving tangible outcomes in diversity. Our gender diversity goal (50:50 at all levels by end of 2021) was widely broadcast across the centre. To achieve this goal, recruitment practices were drawn from evidence-based studies in academia, social psychology and linguistics. Where possible, we used evidence from randomized double-blind trials to inform our practices.


Postdoctoral researchers were recruited into nine Australian universities: the Australian National University, Curtin University, Macquarie University, Monash University, Swinburne University, University of Melbourne, University of New South Wales, University of Sydney and University of Western Australia. The majority (six out of nine) of the fellowships were filled through a single advertisement in 2019, with the selection committee chaired by a woman. The remaining three fellowships were filled through individual advertisements with one woman and two men committee chairs. Of the 48 advertised positions, 23 candidates (48%) were selected by committees chaired by a woman. During the first stages of recruitment we drafted advertisements stating positive organization attributes such as gender targets, diverse leadership and diversity practices within the advertisement to increase the number of applications from women candidates22. Supervisors and team leaders were named in the job advertisements and all jobs were available part time. Advertisements included a link to the ASTRO 3D website, which contained public information about the centre diverse leadership and a detailed diversity and inclusion action plan.

Down-selection processes need to stringently avoid bias against women. Implicit bias can enter the down-selection process itself as well as the metrics used to assess productivity, mentorship and research impact. Randomized double-blind trials show that identical applications are judged by men and women to be less qualified if the candidate name is female23,24,25,26. Biases against women have also been proven in peer review of publications27, citations to publications28, the number of speaking invitations received29,30, the number of invited journal articles31, the number and size of grants awarded32,33, the allocation of awards34 and student teaching assessments35. Such biases may accumulate over a career, producing a type of ‘legacy’ bias on career metrics. Unequal opportunities also arise from career breaks such as maternity leave, which are often invisible in applications. These biases can be minimized if search committees are trained to recognize their own implicit bias and stereotypes36.

Additional biases can enter the hiring process when evaluating and discussing candidates. Men are more frequently discussed in agentic terms (for example, intelligent, exceptional, outstanding, leader) than are women in reference letters37,38. During a study of recruitment committee meetings, men used concrete and abstract language differently to describe male and female applicants, which led to more negative outcomes for women39. In interviews, both interviewer and applicant stereotypes can influence interview performance and evaluations36. Women applicants are frequently expected to meet more criteria than men, including sociability40. Achievement-oriented behaviours, including highlighting one’s accomplishments, may be interpreted favourably for men but negatively for women, unless women are also seen as ‘nice’41,42.

To overcome gender-based biases and stereotypes during the search process, all Australian members of the centre, including members serving on recruitment selection committees, were required to participate in a two-day in-person diversity workshop in 2017–2018 run by a management consultancy that advises businesses and universities on gender stereotypes and unconscious bias issues in the workplace. The same training programme was provided to all members at all universities using the same training company. To further avoid gender-based biases and stereotypes during the search process, we required all search committees to be 50:50 men:women and reinforced the importance of combating implicit bias through the hiring process. To ensure that our small numbers of senior women were not overburdened on hiring committees during the first years of the centre, we requested help from women from outside the centre, or from other science areas (such as geophysics, physics and chemistry), where needed to fulfil the 50:50 requirement. Search committees typically consisted of six members, with the project supervisor serving as committee chair. The team lead was the project supervisor 63% of the time. The search committee members were different for each position and were drawn from the 80 ASTRO 3D chief, partner and associate investigators (mid-career to senior researchers). A woman supervisor chaired the project postdoctoral committees 44% (17/39) of the time.

Academic search committees typically fill short-lists and positions according to the gender ratio in the applicant pool. This strategy is based on the assumption that men and women applicants follow the same distribution of track records and qualifications for the position. However, people who persist in astronomy through their postdoctoral years may not be randomly selected from the original distribution of PhD students. To avoid assumptions about persistence and to overcome implicit bias in assessment of job applications, we required short-lists to contain 50% women. No restrictions were made on the sizes of short-lists, rankings on the short-lists or on the gender of the candidates offered positions. Short-listed candidates were interviewed by the committees remotely, with the same set of questions used for each candidate short-listed for a position. Each position required different expertise, requiring different sets of questions for different positions. Committees used predetermined sets of criteria to rate candidate suitability for each position. Out of the short-listed candidates, the person whom the committee considered most suitable for the role was offered the position. This judgement was based on expertise and fit to the project, not metrics or gender.

The team leaders were responsible for recruitment at their universities and reported on these recruitments to the executive management committee after the positions were filled. Annual diversity reporting was made to the Australian Research Council, the centre advisory board and the centre through annual reports. The gender diversity goal was incorporated into a centre-wide inclusive hiring policy designed to avoid implicit bias.

Climate and retention

We introduced a range of retention policies aimed at supporting the development and workplace environment of women. These included (1) increasing the number of women in senior positions through leadership programmes and recruitment, (2) multiple pathways for complaints, including an anonymous feedback form and ombudspersons outside the centre, (3) clear action against sexism, insults, microaggression, exclusionary behaviour and other factors that produce a poor work culture and atmosphere for women and minorities by consistently following and acting on codes of conduct, (4) help for partners of astronomers including postdoctoral researchers to obtain positions in the same geographical location through university spousal hire programmes and (5) climate surveys with actions to address any issues identified. To provide visible role models with good work–life balance, we highlighted centre members’ hobbies in our bi-weekly newsletter.

ASTRO 3D ran membership climate surveys in 2018 and 2020 to provide the centre with an understanding of the key factors required for its members to contribute at their best and be part of a positive, professional working environment. The feedback provides insights about how the centre is successfully engaging with its members as well as how it could be further improved to support them more. The 2020 survey was similar in structure to the 2018 survey to allow a comparison of the results between the two years, and to identify areas of improvement and those needing more attention.

Leaderskill Group, an external survey provider, was contracted to implement the membership survey to maintain independence throughout the survey process and confidentiality of all responses. Only data from large subgroups were reported to eliminate the possibility of individuals or small groups being identified. All survey emails were sent from the centre’s director, including communication explaining the purpose and scope of the survey and a clear statement about the confidentiality of the feedback, as well as subsequent reminders. The final participation rate was approximately 59%, representing the feedback from a significant portion of the centre’s members with even coverage across ASTRO 3D nodes and member cohorts. The survey asked questions grouped into the following areas of feedback: (1) leadership and collaboration, (2) information sharing, (3) communication channels, (4) professional development and (5) culture, values, equity, diversity and inclusion. All survey questions were assessed by respondents on an industry-standard climate survey Likert ‘agree/disagree’ scale: 1, strongly disagree; 2, disagree; 3, neutral; 4, agree; 5, strongly agree.