The impact of gender diversity on scientific research teams: a need to broaden and accelerate future research

The paper investigates how gender diversity influences the processes and outcomes of scientific teams, moving beyond simple counts of women to examine integration and interaction patterns. Prior work shows diversity enhances creativity and collective intelligence, with gender-balanced teams often performing best, yet the mechanisms remain unclear. Much past research relies on bibliometric indicators that risk tokenism by focusing on presence rather than participation and integration. This study responds to calls for mixed methods approaches that analyze team processes and relational dynamics to understand the role of women on teams. The authors pose two research questions: (1) What is the role of women on scientific teams? and (2) How do women impact team interactions? They test five hypotheses grounded in literature on gendered roles, leadership perceptions, mentoring/service burdens, and collaboration patterns.

The literature documents that team diversity can drive innovation and collective intelligence, with studies noting benefits of gender diversity and balanced teams (Bear & Woolley, 2011; Woolley et al., 2010; Riedl et al., 2021). Earlier perspectives often viewed diversity as detrimental due to conflict (Guimerà et al., 2005), but more recent SciTS work emphasizes the importance of integration over mere representation (Bear & Woolley, 2011; Smith-Doerr et al., 2017). Women are frequently socialized into caregiving and service roles and may face higher service loads (mentoring, committees) than men (Guarino & Borden, 2017; Misra et al., 2011). Perceptions of leadership are often gendered, with men more likely to be seen as leaders or influential even when in the minority (Baugh & Graen, 1997; Bunderson, 2003; Craig & Sherif, 1986; Joshi, 2014). Collaboration patterns also differ by gender, with women engaging more in interdisciplinary research and often maintaining broader collaboration networks (Rhoten & Pfirman, 2007; Bozeman & Gaughan, 2011). Scholars have called for multi-pronged, qualitative and mixed-methods approaches to unpack the meanings of collaboration, mentorship, and the informal processes that shape equity and integration (Bozeman et al., 2013; Madlock-Brown & Eichmann, 2016; Keyton et al., 2008). Theories of gender as social performance (West & Zimmerman, 1987; Butler, 1988) suggest organizational practices can reproduce or disrupt gendered patterns in roles and recognition.

Design and setting: The study was conducted at a land-grant R1 university in the western US, focusing on 12 self-formed interdisciplinary scientific teams funded through a competitive institutional team science program (2015–2020). Teams spanned STEM-related topics and were supported by periodic progress meetings, customized trainings (e.g., SciTS principles, SNA interpretation, DEI, leadership), and administrative mentorship. One additional team not in the formal program volunteered; Team 2 was excluded due to author involvement. Data sources: Multiple evaluation modalities were used: rosters (names, emails, self-identified gender, title, college, department, team role), participant observation (2–6 meetings per team; exceptions: Team 1 had no face-to-face meetings; Team 5 did not consent), field notes, and social network surveys administered at or near the program conclusion. Gender in surveys was self-entered to avoid forced categories; for two nonresponses, roster data filled in gender. Participation followed IRB protocol #19-8622H. Social network survey: Two sections—demographics and relational questions—captured collaboration and social/professional relations among named team members. Collaboration items included: discussed possible joint research/ideas; worked on research/collaborations/tech/consulting; coauthored publications/presentations/proceedings; worked on or submitted grants; sat on a student's committee together. Social/professional items included: I learn from; I seek advice from; I hang out with for fun; is a leader on the team; is a mentor to me; is a friend; energizes me. Constructed measures: Three composite measures were created by combining items: (1) Collaboration (research/collab/tech, joint publications/presentations/proceedings, grants), (2) Social (friend, hang out for fun), (3) Professional support (advice, mentor, student committee). Network diagrams were constructed with directed edges from A to B if A reported the relation with B. Observation and field notes: Field notes (constant comparative method) provided qualitative insights into team development, collaboration patterns, and gender interactions and informed the classification of teams with a senior woman leader. Classifications: Senior woman denoted a woman PI or a woman on the team leadership group (based on rosters and field notes). Faculty included assistant/associate/full professors; non-faculty included undergraduates, graduate students, postdocs, research associates, community partners, and project managers. Analytic approach: Using RStudio, the study computed indegree (number of others who reported a relation with an individual), outdegree, and average degree. To enable cross-team comparison (varying team sizes/response rates), indegree/outdegree were scaled by the number of respondents (yielding proportions). Confidence intervals for scaled indegrees used t-distributions due to limited sample sizes. Pearson correlations between networks (leadership, advice, mentoring, student committees) were computed across teams. For overlap among collaboration, social, and professional support networks, the authors calculated, for each individual, proportions such as collaboration-given-social, and compared averages by gender using t-tests with Cohen's d effect sizes. Sample and response: Of 204 team members, 160 (78.2%) responded; 84% of women and 73% of men completed the survey. Team sizes ranged from 6 to 30 (average ~15), with representation from 3–7 colleges per team.

Overall, only one of five hypotheses was supported by the data; most hypothesized gendered patterns were not observed. - Hypothesis 1 (women higher indegree in mentoring/student committees; men higher in advice/leadership): Not supported. Mentoring appeared team-norm specific rather than gender-specific; on teams with high mentoring indegree for women, men also had high mentoring indegree. - Hypothesis 2 (men more likely to be considered leaders at all career stages): Not supported at 95% CI. Faculty were more likely than non-faculty to be viewed as leaders, but no significant gender differences. Scaled leader indegree estimates (95% CI): Women faculty 0.29 (0.15, 0.42); Men faculty 0.37 (0.16, 0.58); Women non-faculty 0.11 (0.04, 0.18); Men non-faculty 0.17 (0.01, 0.32). - Hypothesis 3 (network correlations reflecting gendered perceptions): Partially supported. Leadership and advice networks were strongly correlated (r = 0.83), consistent with the hypothesis. Contrary to expectations, mentoring was also highly correlated with both leadership (r = 0.82) and advice (r = 0.84). Mentoring and student committees were only weakly correlated (r = 0.32). - Hypothesis 4 (social and collaboration relations more positively correlated for women): Not supported. No significant gender differences in the overlap between collaboration, social, and professional support networks. Both men and women exhibited intertwined relationships. Average overlaps (men vs women; p-values; Cohen's d): collaboration|social 0.77 vs 0.71 (p=0.34; d=0.19); collaboration|professional support 0.78 vs 0.76 (p=0.76; d=0.06); social|collaboration 0.60 vs 0.58 (p=0.77; d=0.06); social|professional support 0.66 vs 0.67 (p=0.96; d=0.01); professional support|collaboration 0.59 vs 0.55 (p=0.43; d=0.15); professional support|social 0.66 vs 0.58 (p=0.22; d=0.24). - Hypothesis 5 (non-faculty have more social connections on teams with a senior woman): Not supported. Average scaled social indegree for entire teams: 0.28 with a senior woman vs 0.20 without (t-test p=0.44; Cohen's d=0.51). For non-faculty: 0.27 with a senior woman vs 0.16 without (p=0.42; d=0.55). No evidence that the presence of a senior woman increased social connectivity. Additional descriptive results: 78.2% overall response rate (women 84%, men 73%).

The authors expected to observe reproduction of gendered inequalities within team interactions, but their analyses largely did not support this. Strong correlations among leadership, advice, and mentoring suggest that perceived leaders are also central sources of guidance and mentoring, irrespective of gender. The lack of significant gender differences across leadership perceptions, network overlaps, and social connectivity raises the possibility that these scientific teams—supported by intentional team science training, facilitation, and institutional structures—may disrupt rather than reproduce traditional gendered hierarchies. The program environment emphasized DEI and team development, potentially fostering ingroup inclusion of women and enhancing dialogic, supportive interactions. Nevertheless, the authors caution against overgeneralization: field notes documented instances of gender inequality, and individual experiences likely vary. The findings align with SciTS evidence that effective processes and authentic integration, not mere compositional diversity, underpin successful team functioning and may enable broader participation and recognition for women.

This study advances understanding of gender diversity on scientific teams by analyzing relational and process data rather than relying solely on bibliometrics. Most hypothesized gendered patterns in mentoring, leadership, and collaboration were not observed, suggesting that, in intentionally supported team science contexts, team norms and inclusive processes may supersede gendered expectations. The key takeaway is that the integration and empowerment of women on teams matter more than achieving a specific proportion of women. Recommendations: (1) Publish statistically insignificant and null findings to broaden the evidence base; (2) Expand research to examine how team science structures and facilitation change scientific culture and access for marginalized groups, focusing on team development, facilitation, and tacit expertise; (3) Move beyond binary gender frameworks to include non-binary identities, develop richer measures of inclusivity, and study how power imbalances constrain or enable expertise. Ultimately, high-functioning scientific teams that authentically integrate diverse members may accelerate knowledge creation while promoting a more equitable profession.

The authors note six limitations: (1) Participation concerns and attrition (one team exited early), limiting data for some teams; (2) Potential Hawthorne effects due to researcher presence in meetings; (3) Senior woman classification combined formal titles with observational assessments of leadership, which may conflate positional and perceived leadership; (4) Variability in respondents’ interpretations of terms such as mentor, advice, and leader, partly mitigated by composite measures; (5) Team success outcomes were not measured, preventing linkage of interaction patterns to performance; (6) Cross-sectional network data captured a single time point, limiting causal or temporal inferences about the sequencing of social and collaborative relationships.