Working groups, gender, and publication impact of Canada’s ecology and evolution faculty
Abstract
Working groups are recognized as a highly effective method for synthesizing science. It is less clear if participating in working groups benefits individual researchers, or if benefits differ between men and women. This is a critical question, for the working group method is not sustainable if the benefit to science comes at a cost to academic careers or gender equity. Here we analyze the publications of Canadian university faculty specialized in ecology and evolution (N = 1244), a field that has embraced the working group method. Researchers were more likely to have participated in a working group as their academic age and prior H-index increased, but controlling for these factors there was no effect of gender. Using a longitudinal analysis, we find that researcher H-indices accrue 14% faster following their first working group publication, regardless of gender. Part of this acceleration may be the 3- to 5-fold higher citation rate of working group synthesis publications. In a survey (N = 169), researchers also report indirect benefits of working groups, at similar rates for men and women. Working groups are therefore good not just for science but also for scientists. Formalized mechanisms for collaborations such as working groups may also offset gender inequities in science.
Introduction
Progress in science depends on our ability to draw out general principles from large amounts of heterogeneous data. While the grist of science is empirical observations and experimental manipulations—that is, primary research—the full strength of individual studies is only realized when they are synthesized, either statistically, mathematically, or conceptually. Rapid development of computational tools to analyze large datasets, combined with the increasing availability of data through public repositories, has allowed scientists to harness the power of large-scale syntheses of previously published results (Hampton et al. 2013). Such “synthesis science” allows researchers to determine which processes are truly general (by comparing multiple studies) and to develop new paradigms (by exploring the interface between disciplines).
As synthesis science is often conducted through collaborative methods, the rise in synthesis science may also be driven by the increase in research conducted in collaborative networks. The ascent of collaborative research networks can be seen, for example, in the impact of the Human Genome Project or the Intergovernmental Panel on Climate Change, or more generally in the rise in the number of authors per publication (Wuchty et al. 2007; Huang 2015). Collaboration promotes research productivity and publications produced by collaborative networks have a higher impact than those produced by individual researchers (Bordons et al. 1996; Lee and Bozeman 2005; Wuchty et al. 2007; Abramo et al. 2009; Gazni and Didegah 2011).
In the fields of ecology and evolution, synthesis science is often conducted through a specific type of collaboration research network known as working groups (WGs). These consist of a small network of researchers—typically 5 to 15 people—that meet to work intensively on a critical problem that requires the synthesis of large amounts of information and ideas and often involves insights from multiple (Rajan and Subramanian 2008) disciplines. Although WGs can be funded through a variety of mechanisms, most WGs in ecology and evolution are formally organized and funded through synthesis centers (Baron et al. 2017). The WG method for synthesis science has been highly influential in ecology and evolution. Specifically, the publications from WGs: (1) are cited more than other papers on similar topics; (2) bridge different knowledge clusters (subdisciplines) more often than other studies; and (3) play a pioneering role in the development of a new fields by synthesizing overall findings and resolving methodological issues (Halpern et al. 2020). Although science may benefit from the WG method, it is not clear whether individual scientists that participate in WGs also benefit in terms of their academic careers. This matters, as the WG method is only viable in the long term if the interests of both science and scientists are aligned.
There are several reasons to expect that WGs may catalyze the careers of individual scientists by accelerating their publication rates and publication impacts. First, because WG publications are often highly influential (Halpern et al. 2020), participating in WGs may raise the publication impact of all participants. Here, the high impact of WG publications could either occur because synthetic research, the type of publication typically created by WGs, is cited more than primary research (Miranda and Garcia-Carpintero 2018), or because publications from collaborations are generally cited more than those produced individually (Lariviere et al. 2015; Leimu and Koricheva 2005). Second, there may be longer lasting impacts on careers as researchers develop new paradigms, construct large databases, and build collaborative networks that define and accelerate future research activities. Alternatively, participation in WGs may take time away from the primary research activities of researchers, depressing their overall publication rate.
Therefore, the first aim of this paper is to examine the impact of participation in WG on researchers’ publication impact. As our analyses are based on observational data, it is difficult, if not impossible, to establish causality. If researchers who participate in WGs have stronger publication records than those who do not, it is not clear if this association reflects an effect of WGs on the careers of researchers or if stronger researchers are simply selected to participate in WGs. We can start to decouple this association by first testing for selection effects using the entire population of researchers who might feasibly participate in WGs, and then conducting a longitudinal analysis (publication record before vs. after WG participation) of just those researchers who have participated in WGs.
If there are career benefits emerging from WGs, another crucial question is whether access to these benefits is equitable. Patterns of collaborative research can disadvantage women, with some analyses suggesting they are less likely to be integrated into international research networks and to assume leadership roles in interdisciplinary research centers (Bozeman and Corley 2004; Misra et al. 2017). For example, when international collaboration requires travel, gendered family constraints may disproportionately prevent women’s participation (Uhly et al. 2017). Furthermore, women tend to receive less recognition than their male collaborators despite similar contributions—a phenomenon known as the “Matilda Effect” (Rossiter 1993; Sarsons 2017). Collaboration can benefit science but not female scientists, and whether this is also true for WGs is far from clear.
Alternatively, women may receive more benefits from WGs organized via synthesis centers than less formalized types of collaboration. A number of synthesis centers have formal policies about gender balance for WGs, some going back to the 1990s. Some centers ask for diverse teams and use gender balance as an assessment criterion (e.g., National Center for Ecological Analysis and Synthesis, John Wesley Powell Center for Ecological Analysis and Synthesis) whereas other centers mandate minimum proportions of women and non-binary researchers amongst the participant’s team (e.g., Canadian Institute of Ecology and Evolution, German Centre for Integrative Biodiversity Research). The presence of formal policies should help address gender inequities in participation which may in turn help erode biases linking masculinity with perceived competence (Stainback and Tomaskovic-Devey 2009; Tomaskovic-Devey and Avent-Holt 2019). However, formal mandates do not always effectively counter interactional dynamics that can disadvantage marginalized groups. Subtle bias can persist, and even be magnified, where presumptions of meritocracy prevail, and mandated equity and diversity initiatives can prompt backlash (Castilla and Benard 2010; Dobbin and Kalev 2016; Kang et al. 2016). Women may participate in WGs at equal rates, but not in ways that ensure equitable credit via authorship in immediate research outputs. They may also fail to reap the same indirect career benefits of WG participation such as the development of useful network connections that facilitate future research and publication activities.
In what follows, we investigate the relationship between WG participation and career publication impact with a gendered lens. We ask: (1) Is the probability of participating in WGs affected by previous publication records or by gender? Do men and women accept/decline opportunities to participate at the same rate and for the same reasons? (2) Does participation in WGs lead to an increase in the publication impact of ecology and evolution researchers? If so, does this benefit differ for men and women? (3) Finally, do synthesis publications from WGs have a higher impact than other publications because of the type of science (synthesis vs. primary) or because of the method (WG vs. traditional), or both?
Methods
Data
Our study population is all Canadian university faculty members who received at least one NSERC Discovery grant through the Evolution and Ecology evaluation committee from 1991 to 2019 (N = 1244). NSERC Discovery Grants are the main source of research funding for ecology and evolution researchers in Canadian universities, providing a useful means of identifying active researchers in this area. We realize that some researchers who consider themselves ecologists or evolutionary biologists hold Discovery Grants through other committees. It is also possible to have a successful research career while never holding a Discovery Grant. Nonetheless, this is a simple and unambiguous criterion for defining our study population. We focus on this group of scientists for three reasons: (1) ecology and evolution researchers are strong users of the WG method. The first synthesis science center in ecology and evolution, NCEAS (National Center for Ecological Analysis and Synthesis) was established in the U.S. in 1995 (Hampton and Parker 2011) and the number of similar synthesis centers worldwide has now grown to 16; (2) Canadian researchers have been active in synthesis science WGs from the beginning, with WG publications appearing soon after the first synthesis center was established. Canadian researchers have participated in WGs in Canada and internationally. Canada’s own synthesis center, CIEE (Canadian Institute of Ecology and Evolution), started in 2008 and since then more than 500 researchers have taken part in WGs organized by CIEE; (3) women comprise 23% of this study population; i.e., there is a gender imbalance to examine, but still enough women for meaningful statistical analysis.
We collected information on researchers’ assumed gender, year of PhD and all affiliated institutions (to aid in the matching of researcher names and outputs) through their careers from publicly available sources (Supplemental Information: Appendix 1). We coded gender (man, woman) from first names (where strongly associated with gender) and from pronouns and photographs on current websites. We acknowledge that such assumptions necessarily have a small degree of error: we may have mis-assigned gender in some cases or failed to capture non-binary or dynamic gender identities. The online sources for biographic information, from the most preferred to the least preferred, included curriculum vitae, websites of the researcher’s current institution, personally-maintained website of the researcher, websites of the researcher’s former institution(s), LinkedIn, Research Gate, Google Scholar, and other sources such as obituaries. Missing data are unavoidable since some information is not available from open sources. We initially identified 1405 researchers from the NSERC Discovery Grants database. However, 161 entries were excluded due to missing data, resulting in a final sample of 1244 individuals.
For each researcher, we quantified the trajectory of their publication impact using the H-index. A researcher has an H-index of H when they have published at least H publications, each of which is cited at least H times. The rate at which a researcher’s H-index increases through time describes the growth of their career publication impact. Please note that our focus on H-index as a response variable in our analyses should not be taken as advocacy for this index as a reliable indicator of scientific impact. Instead, we use this index simply because it is so widely used—or misused—as an indicator of impact (Koltun and Hafner 2021). Indeed, our analyses add to previous work that shows systemic gender bias in this index.
We constructed each researcher’s longitudinal H-index using (1) a retrospective publication record of their peer-reviewed articles and (2) the yearly distribution of citation counts for each publication. To our knowledge, there is no free, publicly available, and readily formatted data source that captures the H-index over time at the individual-researcher level.
To reconstruct a retrospective publication record of each researcher, we started with the Web of Science Core Collection (hereafter, WOS) as 36% of researchers, especially older researchers (modal PhD year = 1972), had no other online publication record. There are several challenges in developing this publication record. First, because WOS uses initials of all names but the surname, there is a possibility of erroneously including publications by other scholars with similar names (e.g., Jane Doe vs. John Doe), a type of false positive. Second, researchers with middle names can vary in which initials they include in their name across publications (e.g., a WOS search on JM Doe would not include publications under J Doe), a type of false negative. Finally, researchers may change their names during their publication career (e.g., with marriage). We developed a Python-based workflow (described in full in Supplemental Information: Appendix 2), to reduce both the false positives and false negatives within the constraints of WOS. In brief, this workflow uses the researcher’s full last name and first initial of their first name (based on all name variations in our database) to download from WOS all publications starting from 5 years prior to their PhD until 2019. This creates a large pool of potential publications (mean of 504 publications per researcher), which is subsequently filtered to a clean subset by cross-referencing with known variants in authorship names for the researcher (from online curriculum vitae or Google Scholar profile) as well as their institutional affiliations, fuzzy matching of publication titles from curriculum vitae or Google Scholar profile where possible, and recursive identification of previously unidentified affiliations to fine-tune the cross-referencing procedure. Once we had cleaned the publication record, we then calculated cumulative citations over years for each publication from WOS yearly citation counts as a precursor to calculating the H-index.
Our analysis also required us to identify which publications are products of WGs. To achieve this, we matched WOS titles with known WG publications funded by the 15 synthesis centers that comprise the International Synthesis Consortium (publication lists from synthesis center websites or obtained directly from centers), or by searching the funding and acknowledgement sections of publications for synthesis center names or acronyms (Supplemental Information: Appendix 3, Table A1). We captured WGs funded through other organizations or mechanisms by searching for keywords commonly used to describe WGs (“WG”, “synthesis group”, “synthesis WG”, “synthesis committee”, “synthesis workshop”, “catalysis group”). All publications were then manually validated by two researchers experienced in biology and synthesis research, and coded as primary research versus synthesis research, and as WG method versus non-WG method. We further categorized synthesis research publications into the following types: statistical synthesis (statistical analysis of previously published or archived data collected by multiple different researchers and/or studies), conceptual synthesis (qualitative review of the literature or proposal of new frameworks for scientific concepts or investigation), or mathematical synthesis (theoretical mathematical models or specific application of general models for the purpose of prediction).
We scored non-WG publications using similar criteria. However, given the large number of publications involved, we changed methods to allow for programmatic approaches to identify publications based on keywords indicative of the three types of synthesis science (Supplemental Information: Table A2). This process yields 2541 new WOS titles that were then manually validated.
To provide additional contextual information on gender and WGs we conducted an online survey (Supplemental Information: Appendix 4) of ecology and evolution faculty in Canada using our researcher database as the sample frame. Of the 1244 ecology and evolution faculty members in our researcher database, we were able to find 1151 researchers’ email addresses. An email invitation was sent to these researchers containing a link to an online questionnaire with a consent cover letter. The survey was carried out from July to September 2019. Two rounds of reminders were sent to improve the response rate as well as in-person recruitment at the Canadian Society of Ecology and Evolution annual conference (Fredericton NB Canada, 18–21 August 2019). After clearing invalid questionnaires with too much missing data or no identification information, we had 169 valid responses, for an effective questionnaire response rate of 14.7%. The questionnaire (Supplemental Information: Appendix 4) asked for information designed to confirm or complete the researcher database (e.g., academic history, gender) as well as information about why researchers participated or not in WGs, and the perceived costs and benefits of participation.
Statistical models
The survey results were evaluated with simple chi-square tests of association.
The effects of research type (synthesis vs. primary) and method (WG vs. traditional) on publication citation rates were evaluated with zero-inflated generalized linear model based on a negative binomial error distribution with a log link (R package glmmTMB). The zero-inflated term was modelled as the logarithm of years prior to 2019 (year of data collection), as we reasoned that some recently published papers would not yet have had time to be cited even if they eventually will be. We excluded publications from 2019 or which drew equally from primary research and synthesis science approaches.
Our statistical analysis of the relationship between H-indices, gender, and WG participation was more complex. This analysis proceeded in two stages, a first stage that examines which researcher characteristics (prior H-index, gender) predict WG participation, and a second stage that tests for changes in researchers’ H-index trajectory following participation in their first WG.
Survival analysis
Our first research question is whether gender or prior H-index predict which researchers participate in WGs. Of the 1244 researchers in our database, 183 (15%) had participated in at least one WG and obtained at least one identifiable publication from that WG. We used survival analysis to test if gender or pace of career progression (H-index adjusted for time since PhD) predicts the hazard rate of participation in WGs. Survival analysis is a statistical technique used to examine the expected time until an event of interest occurs, such as death, failure, recovery, or other specified events. In our study, the event is the scholar’s first WG participation experience, with the rate representing how fast this participation occurs at a specific time. We included an interaction between gender and H-index to assess whether potential selection effects tied to research records captured by the H-index are the same for women and men. In our sample, the first WG publication was in 1993, so researchers enter the risk set either in the year 1993, or in the year of their PhD graduation for those who finished their PhD after 1993.
We estimated hazard ratios for attending WGs using Cox proportional hazards models. This model takes the form: log [h(t)] = a(t) + B'X, where a(t) is a function of time; hazard rate (h(t)) is the rate of participating in a WG, given that the researcher has not yet participated; X is a vector of covariates and B the vector of related coefficients. Since the left-hand side of the equation takes the logarithm of the hazard rate, we use hazard ratios (exp(B)) as opposed to coefficients (B) to make our interpretation more intuitive. Specifically, hazard ratios represent the relative change in the rate associated with a one-unit change in a covariate, holding all other variables constant: a hazard ratio above one represents an increased rate, while a hazard ratio below one suggests a decreased rate.
Fixed effects regression with linear spline
Our second analysis focuses on the subset of researchers who participated in WGs. We employed a fixed effects model with a linear spline to investigate the effects of WG participation and gender on researchers’ trajectory of H-indices over time. We chose a fixed-effects model since this technique nets out all unobserved time-invariant heterogeneity of individuals. In other words, each individual is treated as their own control in a longitudinal analysis (Allison 2009). In our case, this meant comparing the trajectory of researchers’ H-indices in years before and after participating in WGs, and then averaging those differences across researchers. This approach ensures that we do not confound the effect of WG participation on researcher H-index with the effect of H-index on researcher selection for WG participation. To account for autocorrelation within individuals and heteroscedasticity across individuals, we clustered on individuals as suggested by Wooldridge (2016) to obtain robust standard errors. Although for most researchers, the increase in their H-indices over time appears almost linear, overall citation rates have increased dramatically in recent decades. If we factor out the effects of calendar year, H-index trajectories actually grow more slowly as time since PhD increases. We therefore implemented a 0.67 power transformation on the “time” variable to linearize the relationship (the 0.67 exponent was estimated from a log H-index versus log time model incorporating calendar year effects).
To model the temporal structure of H-index growth, we use a linear spline with three knots to allow for differing H-index growth trajectories in the years prior to and following researchers’ first WG experience. Unlike a simple linear regression that assumes a constant slope over the entire period, a linear spline is a piecewise linear function that allows the slope to change at predefined time points, i.e., knots. In this context, the knots divide the time trajectory into three segments, enabling us to model differing growth rates of H-index before and after the first WG experience. Specifically, the first segment of time is 0–5 years before the first WG; the second segment is 1–5 years after the first WG; and the third interval is 6 and more years after the first WG. We code the spline specification in marginal form, which makes interpretation simple: coefficients of the second and third interval capture changes in H-index growth rates from their prior intervals.
Results
The researcher population
We represent the career publication impact of researchers by their H-index, which summarizes the number of papers cited a minimum number of times. The H-index of researchers in our database varied substantially (Fig. 1a). We partitioned our data into quartiles based on the annual H-index from Year 0 to Year 66 after the PhD. While our sample includes a maximum of 69 years post-PhD, the years 67, 68, and 69 post-PhD with only two, one, and one observation, respectively, were excluded from our quartile calculations. The top 25% of researchers with the highest H-index scores were classified as “High H-index”, whereas the bottom 25% with the lowest scores were labeled “Low H-index”. The remaining 50% in the interquartile range were designated as “Medium H-index”. The mean H-index in the upper quartile is 25 (i.e., at least 25 papers cited at least 25 times) whereas the lower quartile records a mean H-index of 5.
Fig. 1.
In our researcher population, gender and H-index progression are not independent: women have significantly slower H-index progression than men (Fig. 1b). Women are also underrepresented among the ecology and evolution faculty, as revealed both in our survey (33.1%) and researcher database (23.15%). The database likely has a lower proportion of women than the survey because it includes older cohorts of researchers (higher proportions of men) that are no longer active or deceased, whereas we recruited survey respondents by email, and younger cohorts of researchers are more likely to have functional and public emails.
Influence of H-index and gender on working group participation
We start our analysis by assessing whether previous H-index or researcher gender affects the chances that a researcher participates in a WG. In our researcher database, 2.9% of women and 11.8% of men have WG experience. However, we cannot directly compare these percentages, because of potentially confounding effects of gendered patterns in researcher age (quantified here as time since PhD) and, as we have just shown, H-index. Instead, we conduct a survival analysis, the results of which indicate a strong association between higher H-indices and increased participation rates in WGs. We first applied the nonparametric Nelson–Aalen estimator and found a significant difference in the cumulative hazard rates of the initial WG experience across High, Medium, and Low H-index categories (Fig. 1c). The results of Cox model, listed in Table 1, demonstrate that the hazard ratio of the H-index is 1.07, which means that a one unit increase in H-index is associated with a 7% increase in the hazard rate of participating in the first WG. Meanwhile, neither gender nor its interaction with H-index are significant predictors of WG participation, suggesting there is no significant difference between men and women in their probability of participating in WGs once we account for academic age.
Table 1.
| Hazard ratio | |
|---|---|
| H-index | 1.07*** |
| Woman | 1.24 |
| Woman*H-index | 0.98 |
| Time since PhD | 0.93*** |
Note: Data are from the period 1993–2019 as no working group publications were found in our database in earlier years. Over this period, 183 out of 1244 researchers had at least one working group publication. *p < 0.05, **p < 0.01, ***p < 0.001.
Findings drawn from our survey data echo this result: the men and women who responded to the survey have similar rates of participating in WGs. The majority of the 169 faculty who took part in our survey had participated in at least one WG (Fig. 2a): 54% of women researchers, and 64% of men researchers (a non-significant difference: = 1.2, p = 0.27). A similar proportion of women (36%) and men (38%) declined at least one invitation to take part in a WG ( = 0.016, p = 0.90). Both women and men listed “work-related duties” as the main reason for declining an invitation, but the second-most common reason for women researchers was “family-related duties” whereas men were at least as likely to give other reasons, such as the research topic being outside their interests or expertise. Despite these trends, the reasons for declining WG participation do not significantly differ between men and women ( = 3.06, p = 0.22)
Fig. 2.
Working groups and H-index progression
Since our survival analysis indicates that the subset of researchers with WG experience is biased with respect to prior H-index, analyzing H-index trajectories in the full population would confound effects of H-index on WG participation with effects of WG participation on H-index progression. Instead, we focus our analysis on longitudinal change in individual H-index trajectories within the subset of 183 researchers who participated in WGs. We use a time spline with knots at the time of WG participation, and 5 years later to evaluate the immediate and longer-term impacts of WG participation on the rate of H-index growth (Table 2). Model 1 reveals that WG experience is associated with a steepening of researchers’ H-index trajectory (Fig. 3a). Specifically, WG participation is associated with a 14% increase in the rate of H-index progression (p = 0.034); the H-index of a researcher with WG experience grows by 1.64 per year, while in the 5 years before participating, their H-index is predicted to increase by only 1.44 per year.
Fig. 3.
Table 2.
| Spline regression with fixed effects of H-index on working group experience and gender | ||
|---|---|---|
| Model 1 | Model 2 | |
| Years since first WG publication | ||
| Pre-working group (−5, 0) | 1.44*** | 1.52*** |
| Post-working group (1, 5) | 0.20* | 0.22* |
| Long post-working group (6, 6+) | −0.13 | −0.23 |
| Woman* pre-working group (−5, 0) | −0.44*** | |
| Woman* post-working group (1, 5) | 0.028 | |
| Woman* long post-working group (6, 6+) | 0.33 | |
| Constant | 11.56*** | 11.56*** |
Note: Spline regression model 1 (r2 = 0.86) and model 2 (r2 = 0.87) was used to test for changes in the H-index accrual rate, relative to that 5 years prior to the first working group publication (“pre-working group”), at specific points in time (years relative to first working group publication given in parentheses): immediately after the first working group (WG) publication (“post-working group”), and 5 years later (“long post-working group”). The effect of researcher gender being woman as opposed to man (“woman”) on H-index accrual rate, pre- and post- working group was also included in the model. Significance (alpha = 0.05) is demarcated as: * (p < 0.05), ** (p < 0.01), *** (p < 0.001).
This acceleration in researchers’ H-indices was realized within 5 years of the first WG experience but does not change further after the 5 years (i.e., neither long term costs nor further acceleration, Table 2).
Gender and the benefits of working group participation
Given that WGs indeed catalyze researchers’ publication impacts, the next important question is whether this benefit is similar for men and women. Here we ask: Does WG participation magnify or ameliorate women’s H-index disadvantage? To assess this, we interact gender with the time intervals before and after the first WG experience (Table 1, Model 2).
Consistent with our earlier analysis (Fig. 1b), before the WG experience, women have significantly lower H-index progression than men, with women’s H-index predicted to increase by 1.08 per year and men’s H-index by 1.52 per year (Fig. 3b). After their first WG, the H-index progression of both women and men accelerates. However, there is no significant interaction between gender and WG participation, meaning that this acceleration is similar for men and women (Table 2,Fig. 3b).
Furthermore, our survey suggests perceptions of indirect benefits from WGs are also similar for both women and men, with 78% and 80%, respectively, reporting benefits in addition to publications from their most recent WG (Fig. 2c). The majority of respondents reported that they developed new collaborations in their WG which carried forward into new projects. Roughly a quarter of respondents reported that their participation in a WG enabled them to reuse a database developed in a WG for a new purpose, and a quarter reported that their WG participation led to funding opportunities. Importantly, there was no gender difference in any of these proportions, suggesting that men and women perceive similar future benefits, at least for their most recent WGs (all tests: < 0.18, p > 0.67).
Citation rates of working group publications
The immediate effects of WGs on H-indices may be related to the high citation rates of WG publications. Our analysis found that WG publications have on average 4.4× more citations than those produced outside of WGs (Fig. 4). We found that this high citation rate appears to originate both from the type of research (synthesis vs. primary) commonly produced by WGs (>97% of WG publications are synthetic) as well as the actual WG method. Specifically, synthesis science—whether mathematical, conceptual, or statistical—is more highly cited than primary research (Figure 4; χ2 = 35.3, p = 10−9). Independent of the type of research, publications from WGs are cited more than publications based on other, more traditional, methods (Fig. 4; WGs: χ2 = 4.9, p = 0.03). Although, we had only a few examples of primary research produced by WGs in our database (largely from globally distributed experiments or original data collection on the scientific process), these also tended to be more cited than primary research from more traditional methods (WG × research type interaction, χ2 = 0.62, p = 0.43).
Fig. 4.
Discussion
This study explores the impact of WGs—an effective approach to doing synthesis science—and gender on the H-index advancement of Canada’s ecology and evolution faculty. By combining analyses of researcher citations with a survey of the target population, several interesting findings emerged. First, participation in WGs leads to a 14% increase in the rate of H-index growth within 5 years of the first WG publication, and this higher rate is maintained subsequently. Second, despite marked gender disparities in representation and publication impact among Canada’s ecology and evolution faculty, we found no evidence of gender disparities in either participation or benefits from WGs.
There are several possible reasons for why WGs accelerate the H-indices of researchers. The most direct reason is that publications from WGs are typically more cited than other publications in ecology and evolution. In our analysis, synthesis research publications from WGs were cited 4.4× more than primary research publications produced through other means. This estimate is slightly higher than a previous analysis of synthesis science publications from WGs which showed that, depending on the subject, such publications were cited 1–3× more than the average publication (Halpern et al. 2020). This slight discrepancy may be because synthesis science publications were not removed from the pool of non-WG publications in the analysis by Halpern et al. (2020). By contrast, our analysis separates research type (synthesis vs. primary) from method (WG collaboration vs. traditional) and so is able to show that the high impact of WG publications is due to both aspects of WGs.
Previous analyses of ecology publications have also shown that longer papers with more authors tend to be cited more, with potential mechanisms including a greater diversity of ideas, greater findability, and higher self-citations (Fox et al. 2016). As WG publications are often substantial and multi-authored, similar mechanisms may underlie our finding that they are cited more than traditionally produced publications. Although we did not attempt to attribute mechanisms, we note that Halpern et al. (2020) demonstrated that the greater citation rate of WG publications was independent of author number. If the high impact of WG publications drives the acceleration of researchers’ H-indices, then the benefit in co-authoring such publications must exceed the cost in terms of foregoing other research opportunities.
It is possible that this benefit may not only be the result of higher impact of WG publications, but also reflects greater efficiency of WGs in creating publications. For example, WGs often result in multiple publications that differ in the lead author, creating the potential for efficient sharing of the scientific workload. The composition of WGs may also lead to greater productivity. Many WGs are funded through synthesis center competitions which favour groups assembled from a diversity of different institutions, disciplines, and genders. As both institutional and gender diversity are positively associated with the number or impact of publications produced by WGs, there is a positive selection for productive teams (Hampton and Parker 2011; Campbell et al. 2013).
We also evaluated the potential for WGs to indirectly accelerate H-index progression by enabling research networks, databases, or funding opportunities that facilitated new projects. In our survey, most WG participants reported such indirect benefits, particularly in terms of research networks. This corroborates an earlier analysis of WG participants which reported an increase in the number of co-authors following their first WG experience (Hampton and Parker 2011). It is possible that such indirect benefits, depending on their timing, contributed to the acceleration of H-indices subsequent to WG participation.
These effects of WGs occur in the context of profound gender disparities in the representation and H-index growth rates of Canadian faculty in ecology and evolution. Such gendered disparities are not unique to Canada, or to ecology and evolution. Many disciplines report that women are underrepresented among university faculty and have lower research productivity and impact (Astegiano et al. 2019; Huang et al. 2020). In Canada, women comprised only 42.1% of all university faculty in 2022 compared to 47.4% of the Canadian labour force (Statistics Canada 2023; World Bank 2023). Such gender inequities have many contributing factors, including systemic disadvantages faced by women in funding, institutional expectations, opportunities for collaboration, experiences of harassment and mistreatment, and domestic workloads (Bornmann et al. 2007; McLaughlin et al. 2012; Weisshaar 2017; Astegiano et al. 2019). Our study was not designed to parse such mechanisms. Instead, we ask if the disadvantages faced by Canadian women faculty in their research careers as a whole are also reflected in WGs. Here, the answer is no: the gender of researchers does not predict whether they will be selected for a WG (after taking into account their prior H-index and academic age), nor how much their H-index accrual increases after their first WG publication. Even when we examine WG participation and benefits in more detail, in our survey, we found no gender differences in rates of participation, of declining invitations or of receiving indirect benefits.
The lack of gender-bias in WG participation may reflect the policies of many, although not all, synthesis centers in either mandating or encouraging a minimum participation rate of women in funded WGs. We do caution that our analysis can only show that WG participation rates are equal (similar between men and women conditional on their H-index), not that they are equitable (similar between men and women conditional on their abilities as researchers), as the H-index may be a gender-biased metric of researcher quality.
Even if participation rates in WGs are gender-neutral, this does not guarantee that WG career benefits will also be gender-neutral; for example, not all participants may be included as authors in all publications. However, our longitudinal analysis and survey both suggest that such benefits are similar for women and men. It is remarkable that women have been able to achieve such parity in WGs while facing disadvantages in other research activities, such as promotion and funding (Astegiano et al. 2019). We do note that our analysis, by its narrow focus on H-indices, does not consider whether gender matters for WG's relationship with other metrics of academic achievement, such as the prestigious first, last and corresponding authorship positions on publications. Previous analyses have shown that women are systematically disadvantaged in terms of authorship position (Broderick and Casadevall 2019), including in ecology and evolution (Fox et al. 2018).
One of the strongest predictors of WG participation is the prior H-index of researchers. The funding of WGs is competitive and includes the curriculum vitae of participants (especially the WG leads). Such funding is needed to cover the substantial travel, accommodation and meal expenses of WGs. There is therefore a strong selective pressure on the H-indices of WG participants. Given that participation in WGs results in an acceleration of researcher H-index, there is a potential for positive feedback such that WG participation results in higher H-indices which enables greater WG participation. Researchers that participate in WGs also learn modes of collaborating particular to WGs and this, combined with their personal benefits realized from WG participation, may also result in continued use of this research mode. In our survey, 85% of respondents with WG experience had participated in multiple WGs. Informally, many synthesis centers report that the same researchers are often in multiple WGs, supporting such a mechanism (pers. comm. International Synthesis Consortium). One limitation of our study is that we considered only the effects of researchers’ first WG experience, not the cumulative effect of subsequent WGs, but this would be a rich area for further study.
Our findings have several implications for science policy. First, we provide evidence that the synthesis science carried out by WGs not only benefits science (Halpern et al. 2020) but also scientists. This alignment of interests is a prerequisite for the long-term sustainability of the method. However, sustainability is also contingent on funding mechanisms for WGs, such as financial support of synthesis centers (Baron et al. 2017), and the benefits outlined in this study provide a rationale for such continued funding. Second, the parity demonstrated between men and women in WG participation and benefits stands out against a backdrop of relentless gender disparity in science. One policy mechanism to offset this disparity is therefore to fund structured collaborations like WGs that require diverse teams. Just as gender-diverse corporate boards are more profitable (Đặng et al. 2020), there is evidence that gender diversity in WGs leads to higher impact publications (Campbell et al. 2013). Third, given the benefits of WG participation to academic careers, it is essential that the skills and opportunities for WG participation be equitably and widely available. Specifically, we need widespread training of graduate students and postdoctoral researchers in the collaboration and data science skills required for WGs, such as by the CIEE’s Living Data Project (Bledsoe et al. 2022). Finally, although we were restricted by data availability to compare WG benefits between men and women, there is an urgent need to examine equity in terms of other demographics such as race, Indigenous status, (dis)ability, sexual orientation, and other gender identities. Collection and equity analysis of such data should be a priority for synthesis centers.
Acknowledgments
This project was funded by a NSERC (Natural Sciences and Engineering Research Council of Canada) “Studies in NSE Research in Canada” grant to Diane S. Srivastava, SylviaFuller, and Catherine Corrigall-Brown. The survey was conducted under Human Ethics certificate H19-00891 approved by the University of British Columbia’s Behavioral Research Ethics Board. We thank the undergraduate assistants—Mirkka Puente, Rodrigo Vallejo, and Liam DelGesso—for their help in building the databases, CIEE Program Coordinator Kelly Haller for her assistance with administration, survey logistics, and reporting, and Catherine Corrigall-Brown for her input in the survey.
References
Abramo G., D'Angelo C.A., Di Costa F. 2009. Research collaboration and productivity: is there correlation? [journal article]. Higher Education, 57(2): 155–171.
Allison P.D. 2009. Fixed effects regression models. SAGE publications.
Astegiano J., Sebastián-González E., Castanho C.d.T. 2019. Unravelling the gender productivity gap in science: a meta-analytical review. Royal Society Open Science, 6(6): 1–12.
Baron J.S., Specht A., Garnier E., Bishop P., Campbell C.A., Davis F.W., et al. 2017. Synthesis Centers as critical research infrastructure. Bioscience, 67(8): 750–759.
Bledsoe E.K., Burant J.B., Higino G.T., Roche D.G., Binning S.A., Finlay K., et al. 2022. Data rescue: saving environmental data from extinction. Proceedings of the Royal Society B Biological Sciences, 289(1979): 20220938.
Bordons M., Gómez I., Fernández M.T., Zulueta M.A., Méndez A. 1996. Local, domestic and international scientific collaboration in Biomedical research [journal article]. Scientometrics, 37(2): 279–295.
Bornmann L., Mutz R., Daniel H.-D. 2007. Gender differences in grant peer review: a meta-analysis. Journal of Informetrics, 1(3): 226–238.
Bozeman B., Corley E. 2004. Scientists’ collaboration strategies: implications for scientific and technical human capital. Research Policy, 33(4): 599–616.
Broderick N.A., Casadevall A. 2019. Gender inequalities among authors who contributed equally. Elife, 8, e36399.
Campbell L.G., Mehtani S., Dozier M.E., Rinehart J. 2013. Gender-heterogeneous working groups produce higher quality science. PLoS ONE, 8(10): e79147.
Castilla E.J., Benard S. 2010. The paradox of meritocracy in organizations. Administrative Science Quarterly, 55(4): 543–676.
Đặng R., Houanti L.H., Reddy K., Simioni M. 2020. Does board gender diversity influence firm profitability? A control function approach. Economic Modelling, 90, 168–181.
Dobbin F., Kalev A. 2016. Why diversity programs fail [Article]. Harvard Business Review, 94(7/8): 52–60. https://stanford.idm.oclc.org/login?url=https://search.ebscohost.com/login.aspx?direct=true&db=bth&AN=116330233&site=ehost-live&scope=site.
Fox C.W., Paine C.E.T., Sauterey B. 2016. Citations increase with manuscript length, author number, and references cited in ecology journals. Ecology and Evolution, 6(21): 7717–7726.
Fox C.W., Ritchey J.P., Paine C.E.T. 2018. Patterns of authorship in ecology and evolution: first, last, and corresponding authorship vary with gender and geography. Ecology and Evolution, 8(23): 11492–11507.
Gazni A., Didegah F. 2011. Investigating different types of research collaboration and citation impact: a case study of Harvard University's publications [journal article]. Scientometrics, 87(2): 251–265.
Halpern B.S., Berlow E., Williams R., Borer E.T., Davis F.W., Dobson A., et al. 2020. Ecological synthesis and its role in advancing knowledge. Bioscience, 70(11): 1005–1014.
Hampton S.E., Parker J.N. 2011. Collaboration and productivity in scientific synthesis. Bioscience, 61(11): 900–910.
Hampton S.E., Strasser C.A., Tewksbury J.J., Gram W.K., Budden A.E., Batcheller A.L., et al. 2013. Big data and the future of ecology. Frontiers in Ecology and the Environment, 11(3): 156–162.
Huang D.-w. 2015. Temporal evolution of multi-author papers in basic sciences from 1960 to 2010. Scientometrics, 105(3): 2137–2147.
Huang J., Gates A.J., Sinatra R., Barabási A.-L. 2020. Historical comparison of gender inequality in scientific careers across countries and disciplines. Proceedings of the National Academy of Sciences, 117(9): 4609–4616.
Kang S.K., DeCelles K.A., Tilcsik A., Jun S. 2016. Whitened résumés:race and self-presentation in the labor market. Administrative Science Quarterly, 61(3): 469–502.
Koltun V., Hafner D. 2021. The h-index is no longer an effective correlate of scientific reputation. PLoS ONE, 16(6): e0253397.
Larivière V., Gingras Y., Sugimoto C.R., Tsou A. 2015. Team size matters: Collaboration and scientific impact since 1900. Journal of the Association for Information Science and Technology, 66: 1323–1332.
Lee S., Bozeman B. 2005. The impact of research collaboration on scientific productivity. Social Studies of Science, 35(5): 673–702.
Leimu R., Koricheva J. 2005. Does scientific collaboration increase the impact of ecological articles?. BioScience, 55: 438.
McLaughlin H., Uggen C., Blackstone A. 2012. Sexual harassment, workplace authority, and the paradox of power. American Sociological Review, 77(4): 625–647.
Miranda R., Garcia-Carpintero E. 2018. Overcitation and overrepresentation of review papers in the most cited papers. Journal of Informetrics, 12(4): 1015–1030.
Misra J., Smith-Doerr L., Dasgupta N., Weaver G., Normanly J. 2017. Collaboration and Gender Equity among Academic Scientists, 6(1): 25. https://www.mdpi.com/2076-0760/6/1/25.
Rajan R.G., Subramanian A. 2008. Aid and growth: what does the cross-country evidence really show? Review of Economics and Statistics, 90(4): 643–665.
Rossiter M.W. 1993. The Matthew Matilda effect in science. Social Studies of Science, 23(2): 325–341.
Sarsons H. 2017. Recognition for group work: gender differences in academia. American Economic Review, 107(5): 141–145.
Stainback K., Tomaskovic-Devey D. 2009. Intersections of power and privilege:long-term trends in managerial representation. American Sociological Review, 74(5): 800–820.
Statistics Canada. 2023. Statistics on full-time academic teaching staff at Canadian universities. Available from https://www150.statcan.gc.ca/n1/pub/71-607-x/71-607-x2019027-eng.htm.
Tomaskovic-Devey D., Avent-Holt D. 2019. Relational inequalities: an organizational approach. Oxford University Press,USA.
Uhly K.M., Visser L.M., Zippel K.S. 2017. Gendered patterns in international research collaborations in academia. Studies in Higher Education, 42(4): 1–23.
Weisshaar K. 2017. Publish and perish? An assessment of gender gaps in promotion to tenure in academia. Social Forces, 96(2): 529–560.
Wooldridge J.M. 2016. Introductory econometrics: a modern approach. 6th ed. Cengage Learning.
World Bank. 2023. World development indicators. The World Bank.
Wuchty S., Jones B.F., Uzzi B. 2007. The increasing dominance of teams in production of knowledge. Science, 316(5827): 1036–1039.
Supplementary material
Supplementary Material 1 (DOCX / 40 KB).
- Download
- 39.23 KB
Information & Authors
Information
Published In
FACETS
Volume 10 • 2025
Pages: 1 - 11
Editors: Paul Dufour and Marie-Claire Shanahan
History
Received: 18 July 2024
Accepted: 21 February 2025
Version of record online: 14 April 2025
Copyright
© 2025 The Author(s). This work is licensed under a Creative Commons Attribution 4.0 International License (CC BY 4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original author(s) and source are credited.
Data Availability Statement
Data and codes are available from Dryad (https://doi.org/10.5061/dryad.t4b8gtj86).
Key Words
Sections
Subjects
Plain Language Summary
Collaborative Boost: How Working Groups Enhance Scientific Careers and Promote Gender Equity in Academia
Authors
Author Contributions
Conceptualization: QW, SF, DSS
Data curation: QW, FL, DSS
Formal analysis: QW, SF, DSS
Funding acquisition: SF, DSS
Investigation: QW, DSS
Methodology: QW, FL, SF, DSS
Project administration: QW, DSS
Resources: DSS
Supervision: QW, DSS
Validation: QW, DSS
Visualization: QW, DSS
Writing – original draft: QW, DSS
Writing – review & editing: QW, SF, DSS
Competing Interests
The authors declare there are no competing interests.
Funding Information
Natural Sciences and Engineering Research Council of Canada: UIF-537833-18
Metrics & Citations
Metrics
Other Metrics
Citations
Cite As
Qian Wei, Francois Lachapelle, Sylvia Fuller, and Diane S. Srivastava. 2025. Working groups, gender, and publication impact of Canada’s ecology and evolution faculty. FACETS.
10: 1-11.
https://doi.org/10.1139/facets-2024-0143
Export Citations
If you have the appropriate software installed, you can download article citation data to the citation manager of your choice. Simply select your manager software from the list below and click Download.
There are no citations for this item