Gender-biased perceptions of important ecology articles

Gender bias is still unfortunately rife in the sciences, and men co-author most articles (> 70%) in ecology. Whether ecologists subconsciously rate the quality of their peers’ work more favourably if men are the dominant co-authors is still unclear. To test this hypothesis, we examined how expert ecologists ranked important ecology articles based on a previously compiled list. Women proposed articles with a higher average proportion of women co-authors (0.18) than did men proposers (0.07). For the 100 top-ranked articles, women voters placed more emphasis on articles co-authored by women (0.06) than did men (0.02). However, women voters were still biased because they ranked men-dominated articles more highly, albeit not by as much as men did. This effect disappeared after testing read-only articles. This indicates a persistent, subconscious bias that men-dominated articles are considered to be of higher quality before actual assessment. We add that ecologists need to examine their own subconscious biases when appointing students, hiring staff, and choosing colleagues with whom to publish.


30
Despite a general reduction in gender disparities within academia over time [1][2][3] , there 31 remains ample gender-bias across scientific disciplines. Experimental evidence shows 32 that scientists tend to rate writings authored by men higher than those authored by 33 women 4 , and that academic scientists tend to favour men applicants over women for 34 student positions 5 . In the United Kingdom, there is also evidence that women academics 35 in science, engineering, and mathematics have more administrative duties on average 36 than men, and hence, less time to do research 6 . Women scientists there also have fewer 37 opportunities for career development and training, and tend to earn lower salaries, hold 38 fewer senior roles, and are less likely to be granted permanent positions than men 6,7 . 39 40 Gender bias -in its myriad forms of expression and consequences -is also likely to 41 vary among scientific disciplines. In ecology, despite undergraduates and young 42 researchers having gender ratios closer to parity (as is now the case in most science 43 disciplines 8,9 ), senior academic positions in ecology and evolution are still dominated by 44 men 10 . This means that most ecology papers are written by men; for example, in a study 45 examining the proportion of women authorships in papers published from 1990-2011 46 across 21 science and humanities disciplines, ecology and evolution had the seventh lowest proportion of women authors (22.76% of 279012 total authorships) 3 . Women 48 scientists are also consistently under-represented in ecology textbooks compared to 49 baseline assumptions of no bias 9 . 50 51 Scientific journals also tend to appoint more men than women on their editorial boards, 52 and editors tend to select reviewers of the same gender as themselves (known as 53 homophily) 11 . Ecologists are also guilty of homophily; for example, men editors selected 54 < 25% women reviewers, but women editors consistently selected between 30 to 35% 55 women reviewers for all papers submitted to the journal Functional Ecology from 56 January 2004 to June 2014 12 . Yet, this is not due to the actual performance of women 57 reviewers, because reviewer scores for that journal did not differ between men and 58 women reviewers, and the proportion of papers rejected did not differ between women 59 and men editors 12 . However, there are gender differences in how papers are reviewed. For 60 example, from a much broader sample of journals in ecology and evolution, a survey of 61 1334 ecologists and evolutionary biologists identified that women took longer to review 62 papers than men, and women reviewed fewer manuscripts on average (a logical outcome 63 of being asked less frequently than men to review). In seeming contradiction to the lack 64 of a gender difference in reviewer scores for Functional Ecology 12 , men from the broader 65 sample recommended rejection more frequently than did women 13 . 66 67 Ecologists can take some heart in the observation that there is little evidence for gender 68 bias in acceptance or citation rates of their papers. In one regional ecology journal (New 69 Zealand universities), men produced almost 40% more papers than did women, and this difference 93 appeared as early as two years from initial publication 19 . Likewise, a sample of 182 94 academic biologists (69 women and 113 men) with at least ten years of experience in 95 academia indicated that women produced between 19 and 29% fewer papers after ten 96 years of employment than did men 20 . 97 98 Gender differences in publication frequency can occur for many reasons, including 99 possibly having less time to do research 6 , higher demands of motherhood 21-23 , a lower 100 relative tendency compared to men to seek self-promotion 24-26 , fewer academic grants 101 and accolades 27-29 , among other reasons 9,30,31 . Despite no strong evidence yet for gender 102 biases in citation rates in ecology, it is still unclear whether established ecologists -both 103 women and men -subconsciously rate the quality of their peers' work more favourably 104 than if men are the dominant co-authors, as has been shown for postgraduate students 105 enrolled in communication programs 4 . To test this hypothesis, we have recently compiled 106 a unique dataset to determine which ecology articles are most recommended by ecology 107 experts 32 . From this list of most-recommended articles, we compiled the gender of both 108 the proposers and voters of the articles, as well as the gender of each co-author of the 109 articles themselves (including the gender of the lead author). Specifically, we asked 110 whether ecologists of both genders were swayed by their learned perceptions of article 111 'quality' outside of the review process in terms of: (i) whether men and women proposed 112 or voted for articles more often if they had a higher proportion of men co-authors, and (ii) 113 if there was a correlation between the proportion of women co-authors on an article and 114 its mean rank (as measured and reported previously -see Material and methods 32 ). 115 116

117
Editor pool 118 The overall proportion of women among the 665 editors we originally contacted to 119 propose articles was 22.1% (i.e., 141 women, 524 men); of these, 14 women (10.0%) and 120 137 men (26.1%) responded, and 12 women (8.5%) and 101 men (19.3%) proposed 121 articles. These show that men were more likely to respond and propose articles than were 122 women. 123 124 Proposed articles and voting differences 125 The proportion of women co-authors on the articles proposed by men were on average 126 lower (0.06 to 0.09; mean = 0.07) than those proposed by women (0.13 to 0.27; mean = 127 0.20), although the data were highly skewed and most proposed articles (77%) had no 128 women co-authors at all (Fig. 1a). When we examined the 100 top-ranked articles voted 129 by women or men only, the bias remained: women voters ranked articles in the top 100 130 that had more women co-authors (0.029 to 0.093 proportion women) than did those voted 131 by men (0.001 to 0.029) (Fig. 1b). 132 133 However, even for women voters, there was a tendency to rank men-dominated articles 134 more highly. For women voters only, there was a weak (β = 0.03), but non-random (p ran = 135 0.011) correlation between the proportion of women co-authors and the article's score 136 (from the voting), such that the lower the proportion of women co-authors, the higher 137 they were ranked by women (Fig. 2a). For men voters only, the relationship was stronger 138 (β = 0.11) and also non-random (p ran < 0.0001) (Fig. 2b). The inverse-score-weighted 139 mean proportion of women co-authors (Σw i /s i = for i voters, where s = score from 1 to 4, 140 and w = proportion of women co-authors) was 0.0277 for women voters, and 0.0251 for 141 men voters (Fig. 2a,b). 142 143 Read-only articles 144 These non-random relationships could be partially driven by the observation that older 145 articles were more highly ranked than younger articles 32 , and that gender biases in 146 authorship are generally stronger in older articles. So, we also used the 'read-only' article 147 scores (in the original survey, voters were asked to indicate whether or not they had in 148 fact read the paper they were scoring; for this new list, article score was unrelated to 149 article age) 32 for women-only and men-only voters separately. Indeed, the relationships 150 between article rank and proportion of women co-authors disappeared for both women 151 voters (Fig. 2c) and men voters (Fig. 2d), although the inverse-score-weighted mean 152 proportion of women authors was again higher for women voters (0.0458) than men 153 voters (0.0303). 154 155 Lead author 156 Examining just the gender of the lead author, 510 of the 544 papers (93.8%) proposed 157 had a man as a first author. For the 100 top-ranked papers (read or not), 98 were led by a 158 man; when men alone voted, 99 of the 100 top-ranked papers were led by a man, and 159 when women voted, 96 were. As above, the difference between women and men voters 160 largely disappeared when we examined the read-only list of the 100 top-ranked papers -161 when women voted, 93 of these was led by a man, and 92 were when men voted. 162 163 Temporal trends 164 The articles proposed by the entire sample of ecologists indicated a general trend of 165 increasing proportion of women co-authors, from < 5% women co-authors before the 166 1990s, to the most recent articles published in the last decade exceeding one quarter 167 women co-authorship (Fig. 3). 168 169

170
Our results show that at least for well-established, expert ecologists, both men and 171 women tend to propose and rank articles more highly when they are co-authored by more 172 men, perhaps indicating a degree of homophily when assessing article importance. These 173 results endure despite little evidence that men biologists view themselves as having 174 relatively higher self-perceived expertise than women biologists (according to a sample 175 of 61 men and 190 men tropical biologists) 30 . That article score and the proportion of 176 women co-authors were correlated for both genders can be explained largely by the fact 177 that older papers with which ecologists are at least familiar are generally ranked higher 32 . 178 But because older articles had fewer women co-authors, women ecologists appear to have 179 had little choice but to score the 'classics' more highly. Indeed, when we restricted the 180 ranked articles to those that voting ecologists had actually read, the relationship 181 disappeared. 182 183 We contend, however, that because assessing the read-only papers demonstrated less of a 184 bias toward men-co-authored papers, this is in fact evidence of a lingering, subconscious 185 gender bias among ecologists. Both men and women ecologists rated articles that they 186 had not actually read higher when they were more men-dominated, yet once they did 187 personally evaluate (read) them, this bias disappeared. This appears to indicate that they 188 had the a priori assumption that men-dominated papers would somehow be better. This 189 assumption was stronger in men than women, but it seems that women ecologists are still 190 subject to a persistent form of auto-sexism, perhaps kept flourishing by a remaining 191 academic culture of valuing women's contributions less than men's. 192 193 This read-only group of younger articles (by 14 years, on average) 32 , combined with the 194 observation that there is an increasing proportion of women co-authors on highly ranked 195 ecology articles, are nonetheless encouraging signs. Indeed, that these highly ranked 196 papers are now (over the last decade) exceeding 25% women co-authors agrees with the 197 approximate overall pool of women co-authors in the general discipline (22.76% women 198 co-authors for articles published from 1990-2011 in ecology and evolution, based on 199 279012 total authorships) 3 (Fig. 3). 200 201 Despite this increasing trend, our results show that women ecologists are still very much 202 in the minority, both in terms of high-ranking article authorships (i.e., less than one third) 203 and editorships (i.e., less than one fifth). Further, women experts were much less likely to 204 respond to requests to contribute their suggestions of potentially important articles. The 205 underlying reasons for this are unclear, although we hypothesize that it could be 206 explained in part by the observation that expert women ecologists are increasingly and 207 disproportionately requested to take part in surveys, consortia, juries, and committees in 208 an attempt to seek gender parity (e.g., reference 20 ). Excessive requests to participate 209 might be exasperating and time-consuming, thus discouraging participation rates relative 210 to men. 211 212 Our results highlight two important remaining biases persisting among today's expert 213 ecologists: (i) we all subconsciously bias our opinions of article importance toward those 214 that have at least traditionally been dominated by men co-authors, and (ii) men ecologists 215 are still more gender-biased than women ecologists in this regard. While homophily 216 might partially explain these results, it seems apparent that some gender biases against 217 women remain when ecologists assess article quality, and even more so when they judge 218 apparent quality without actually reading the article (i.e., via reputation only). The 219 potential solutions to these problems are varied, including increasing the discussion of the 220 contribution of women ecologists more explicitly in university teaching material 9 , 221 improving flexibility and opportunity in the workplace 2,33 and at conferences 2,34 for 222 women, embracing positive discrimination in academic appointments 33 , increasing the 223 prevalence of double-blind reviews 16 , and advocating alternative metrics of citation 224 performance that do not disadvantage women 19 . We further add that all ecologists -225 especially men -would benefit from serious, personal introspection about their own 226 biases 35 , no matter how uncomfortable an admission of gender bias might be. Denial of 227 one's own contribution to the problem only serves to perpetuate it 33 . Consciously 228 increasing the number of women ecologists among our students, in our labs, on our 229 editorial boards, requested to review papers, and as co-authors on our manuscripts 230 (something we admittedly failed to do here), will also help to reduce these subconscious 231 biases. 232 233

234
The full details of how we generated the list of most recommended ecology articles and 235 how they were ranked are given in Courchamp & Bradshaw 32 ; however, we briefly 236 describe the approach and main characteristics of the list here. We contacted the editorial 237 members (ipso facto, ecology 'experts') of some of the most renowned journals in 238 general ecology: Ecology Letters, Trends in Ecology and Evolution, Ecology, Oikos, The 239 American Naturalist, Ecology and Evolution and Ecography, as well as all the members 240 of the Faculty of 1000 Ecology Section (f1000.com/prime/thefaculty/ecol). Of these, we 241 contacted 665 by e-mail to ask them to send us three to five peer-reviewed papers (or 242 more if they wished) that they deemed each postgraduate student in ecology -regardless 243 of their particular topic -should read by the time they finish their dissertation, and that 244 any ecologist should also probably read. 245 246 We successfully elicited 147 respondents of the 665 we contacted, who in total 247 nominated 544 different articles to include in the primary list. We then asked these same 248 665 experts to vote on each of the papers to obtain a ranking, assigning each article to one 249 of four categories: Top 10, Between 11-25, Between 26-100 or Not in the top "100". We 250 gave one (1) point for each selection of the Top 10 category, two points for the Between 251 11-25, three points for the Between 26-100, and four points for the Not in the top "100". 252 As described in Courchamp & Bradshaw 32 , we averaged all article scores across all 253 randomly sampled sets of votes for each article, and then applied a simple rank to these 254 (ties averaged), thus avoiding any contrived magnitude of the differences between 255 arbitrary score values (i.e., 1 to 4 base scores  and 95% confidence interval (error bars) of the proportion of women co-authors for the proposed articles relative to the gender of the proposer (articles proposed by 68 women only, and 418 proposed by men only). The values (proportion women co-authors) are 'scattered' to show their distribution within each proposer gender; note that 55.9% and 80.1% of the articles proposed by women only and men only, respectively, had no women co-authors (i.e., zero values). b) Mean (dashed horizontal lines) and 95% confidence interval (error bars) of the proportion of women co-authors of the 100 top-ranked articles relative to the gender of the voter (62 women and 292 men voted in total). The values (proportion women co-authors) are 'scattered' to show their distribution within each voter gender; note that 83% and 94% of the articles proposed by women and men, respectively had no women co-authors (i.e., zero values).

Figure 2. a)
Proportion of women co-authors on articles relative to their mean rank (score; where lower scores indicate a higher ranking) when voters were restricted to women. There was a weak (β = 0.03), but non-random (p ran = 0.011) correlation between article gender ratio and score, such that the lower the proportion of women co-authors, the higher they were ranked by women. b) Proportion of women co-authors on articles relative to their mean rank when voters were restricted to men. There was a stronger (β = 0.11) and non-random (p ran < 0.0001) correlation between article gender ratio and score, such that the lower the proportion of women co-authors, the higher they were ranked by men. Also shown in both panels is the inverse-score-weighted mean proportion of women co-authors (Σw i /s i = 0.0277 for i women voters, or 0.0251 for i men voters, where s = score from 1 to 4, and w = proportion of women co-authors). c) As in a, but when the scored articles were only those actually read by the voters 32 . d) As in b, but when the scored articles were only those actually read by the voters. The inverse-score-weighted mean proportion of women co-authors for these read-only articles was higher for womenonly (0.0458) versus men-only voters (0.0303).

Figure 3.
Time series of mean (± 2 standard errors of the mean; grey dashed lines) decadal gender ratio (proportion women co-authors) for all 544 proposed articles. Numbers above the graph indicate sample size (number of articles) used to calculate decadal means ('←1' indicates one article from 1858) 37 . For comparison, the proportion of women authorships in articles published from 1990-2011 in ecology and evolution (22.76% of 279012 total authorships; lower black horizontal dashed line) 3 , are shown. Also shown are the 95% confidence limits of the proportion of women co-authors of the 100 top-ranked articles assessed by women (green shaded area: 0.029 to 0.093; Fig. 1b) and men voters (orange-shaded area: 0.001 to 0.029; Fig. 1b).