Experimental test of a new global discrete symmetry I will present biblio-metric data about “fundamental theory and gender”, and interpret what data means and answer the big question: The M ↔ F symmetry (S 2 generalized to gender S n ) C) is explicitly broken (namely, it’s not a symmetry, M 6= F )? M) is spontaneously broken (namely, due to discrimination)?
26
Embed
Experimental test of a new global discrete symmetry · Science and Nature. Furthermore, female authors write 19± 7% fewer papers in seven years following their Þrst paper than their
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Experimental test of a newglobal discrete symmetry
I will present biblio-metric data about “fundamental theory and gender”,
and interpret what data means and answer the big question:
The M ↔ F symmetry (S2 generalized to gender Sn)
C) is explicitly broken (namely, it’s not a symmetry, M 6= F )?
M) is spontaneously broken (namely, due to discrimination)?
M(ainstream?) Theory
[In string gender conferences, check links]
“All women share the same kind of sad and unfair experiences since the be-
ginning of their scientific career”. “Mansplaning. Gaslighting. White Male
Hetero Privilege. Sexual harassment at epidemic levels. Micro-aggressions”.
Men mobilize their masculinity supporting ... men in ways that advance careers.
You don’t see? You have (unconscious) bias and steal credit to women. “Eval-
uators tend to favour men”. “Scientific quality is a gender social construction”.
“Excellence is the current buzzword. Gender equality should achieve the same”.
“I have a dream: that ... excellence in science is no more distorted and sweletred
by gender stereotypes or creeping discrimination”. “Positive discrimination and
gender mainstreaming”, “We have to help women at all levels ... to academic
positions”, “programmes for women ... challenged in court for discrimination”
[vs men]. “People and culture can be obstacles for change”.
No gender preference in citations in any category at any time, down to % level.
M more cited than F , equally by M and F : it’s merit, not sexism. Ed 6= Rocco.
Similar analysis applied to countries finds instead significant asymmetries.
Sexism in citations?
A speaker claims sexism in citations quoting 1610.08984:October 31, 2016Preprint typeset using LATEX style emulateapj v. 04/17/13
QUANTITATIVE EVALUATION OF GENDER BIAS IN ASTRONOMICAL PUBLICATIONSFROM CITATION COUNTS
Neven Caplar1, Sandro Tacchella2 & Simon Birrer3
Institute for Astronomy, Department of Physics, ETH Zurich, CH-8093 Zurich, Switzerland
October 31, 2016
Abstract
We analyze the role of first (leading) author gender on the number of citations that a paper receives,on the publishing frequency and on the self-citing tendency. We consider a complete sample of over200,000 publications from 1950 to 2015 from five major astronomy journals. We determine the genderof the first author for over 70% of all publications. The fraction of papers which have a female firstauthor has increased from less than 5% in the 1960s to about 25% today. We find that the increaseof the fraction of papers authored by females is slowest in the most prestigious journals such asScience and Nature. Furthermore, female authors write 19± 7% fewer papers in seven years followingtheir first paper than their male colleagues. At all times papers with male first authors receive morecitations than papers with female first authors. This di↵erence has been decreasing with time andamounts to ⇠6% measured over the last 30 years. To account for the fact that the properties offemale and male first author papers di↵er intrinsically, we use a random forest algorithm to control forthe non-gender specific properties of these papers which include seniority of the first author, numberof references, total number of authors, year of publication, publication journal, field of study andregion of the first author’s institution. We show that papers authored by females receive 10.4±0.9%fewer citations than what would be expected if the papers with the same non-gender specificproperties were written by the male authors. Finally, we also find that female authors in our sampletend to self-cite more, but that this e↵ect disappears when controlled for non-gender specific variables.
Keywords: sociology of astronomy — publications, bibliography
1. INTRODUCTION
Gender inequality and biases seem to be persistentin the scientific community. Even though the numberof doctorate degrees awarded to women is constantlyincreasing, women still tend to be underrepresented infaculty positions (National Science Foundation 2015).Numerous studies have shown that both male and femalereferees consistently give higher scores to identical workdone by males than females (e.g., Wenneras & Wold1997, Moss-Racusin et al. 2012). As an example of biasin publishing, the study by Budden et al. (2008) showedthat the number of female authors increased significantlyafter a journal in the field of ecology switched to thedouble-blind refereeing system in which the names ofthe authors are kept hidden from the reviewers.
The recent growth of big databases enables more sys-tematic statistical investigation into the role of genderon the publishing and awarding mechanisms in the sci-entific community. Conley & Stadmark (2012) deducedthat female authors tend to be underrepresented in theprestigious publications; for example, female authorshave contributed only 3.8% of earth and environmentalscience articles for Nature News & Views even thoughthey represent approximately 20% of scientists in thefield. The same conclusion was reached by West et al.(2013) who conducted a large multi-field analysis andfound much of the disparity between male and female
authors was due to lack of females who are first authorsof prestigious papers. The same group has also foundthat men tend to self-cite their work more (King et al.2016). For the domain of engineering, Ghiasi et al.(2015) has recently showed around 10% bias in thenumber of citations.
Focusing on astronomy, Davenport et al. (2014) hasstudied gender balance at the 223rd meeting of theAmerican Astronomical Society and found that eventhough the gender ratio of speakers mirrors that ofconference attendees, women asked fewer questions thantheir male peers. A similar conclusion was reached byPritchard et al. (2015) who studied patterns at theNational Astronomy Meeting 2014 of UK astronomers.A study by Reid (2014) on the success of proposalsfor time on the Hubble Space Telescope concluded thatproposals with a female principal investigator are lesslikely to succeed than proposals with a male principalinvestigator. They also found that the success rates bymales and females for more recent graduates (Ph.D.since 2000) are more comparable to each other. Similardisparity between genders was also recently reportedfor time allocation at European Southern Observatorytelescopes (Patat 2016). Although these di↵erence areobserved in the conference settings and in the proposalsuccess rate, no study has investigated possible di↵er-ences in the number of citations between the genders.
Spurred by these findings, we wish to measure therole of gender on the number of citations that papersreceive in astronomy. Throughout the study we assume
arX
iv:1
610.
0898
4v1
[ast
ro-p
h.IM
] 27
Oct
201
6
Reading it
Gender Bias in Astronomy 11
e↵ect could also lead to underestimation of seniorityfor some female authors as we misidentify establishedauthors as newly arriving in the field (see Section 3).We note that fully accounting for these e↵ects wouldprobably increase the observed di↵erence in citationcounts between females and males in astronomy.
Because our name classification mechanism is mostlybased on data sources in Europe and North America,this means we are less likely to recognize the gender ofnames from di↵erent cultures. This becomes especiallyapparent in later years with a more globalized astronomycommunity. We do not expect this to create any stronge↵ect in our analysis as we have checked that the genderbias is largely independent of the region where the hostinstitution is based.
Of course we cannot claim that we have actuallymeasured gender bias. One could imagine numerousother parameters that should be considered and matchedbefore such a conclusion could be drawn. Our resultstherefore should be taken with care. It is our beste↵ort based on all of the available data that we couldacquire. We encourage the community to work onand/or enhance our dataset for further analysis.
8. CONCLUSIONS
The main goal of this paper is to quantify the genderbias in astronomy. We define “gender bias” as the dif-ference in the citation counts between female and malefirst author papers with matched non-gender properties.We assembled information about all papers published inA&A, MNRAS, ApJ, and about all of astronomical pa-pers published in Science and Nature from 1950 to 2015.In total, we have analyzed over 200,000 papers. Usingthe gender recognition algorithm, we assigned gender tothe first author of every paper where this was feasible(about 70% of all papers). For the majority of the re-maining papers we were not able to deduce the genderof the first author because the author only used initialsthroughout their publishing career. Our main conclu-sions are as follows:
• Female participation has been consistently risingover time. Females authored around 25% of thepapers in the last few years, but this rise has beenthe slowest in the most prestigious journals, suchas Nature and Science, where the fraction amountsto only 17%.
• By simply measuring the di↵erence between thenumber of citations received by male and femaleauthors in our sample we find a clear 5.6 ± 1.0%di↵erence in favor of male authors, when measuredfrom the year 1985 onward. This gender di↵erencedoes not change significantly when choosing a lateryear for the measurement as the di↵erence is de-creasing very slowly or not at all.
• We estimate gender bias by using machine-learningtechniques to control for di↵erences between themale and female first author papers. We find thatfemales receive 10.4 ± 0.9% fewer citations thanwhat would be expected if the papers with the same
characteristics were written by the male authors.This is consistent with our finding that if genderbias did not exist, we would expect males in oursample to receive 4.2±0.8% less citations than fe-males.
• Using the probability of an author having self-citedtheir previous paper as a self-citation metric, wefind that females in our sample are 9 ± 2% morelikely to cite their previous work. When usingmachine-learning techniques to control for di↵er-ences between the male and female samples we findno significant intrinsic di↵erences in propensity ofmale and female authors to cite themselves.
Our conclusions are limited by our inability to deter-mine the gender for all of the authors of the papers. Webelieve that this e↵ect would probably act in a mannerto further strengthen our conclusion about the existenceof gender bias in astronomy.
We make our dataset publicly available and invite fur-ther research on this topic.
We would like to thank Joanna Woo for giving de-tailed comments on the manuscript. We acknowledge thestimulating comments given to us by Meg Urry, RenateSchubert, Ra↵aella Marino, Benny Trakhtenbrot, IzabelaMoise and Evangelos Pournaras. We thank AmandaBluck for proof reading the manuscript. We acknowl-edge support from the Swiss National Science Founda-tion. This research made use of NASAs AstrophysicsData System (ADS), the arXiv.org preprint server, andthe Python plotting library matplotlib (Hunter 2007).
REFERENCES
Breiman, L., Friedman J., Stone, C. J., Olshen, R. A., 1984,Classification and Regression Trees, Taylor & Francis Ltd [5.1]
Budden, A., et al., 2008, Trends Ecol Evol., 23, 4 [1]Conley, D., Stadmark, J., 2012, Nature, 488, 590 [1]Davenport, J. R. A., Morgan, F., Grand, E., Hagen, A.,
Poppenhaeger, K., Watkins, L. L., 2014, arXiv:1403.3091 [1]Ghiasi G, Lariviere V, Sugimoto CR (2015) PLoS ONE 10(12) [1]Hunter, J. D., 2007, Computing In Science & Engineering, 9, 90
[8]Ivie, R., White, S., Garrett, A., Anderson, G., 2013, AIP, College
Park, MD [2.5]King, M. M., Bergstrom, T. C., Shelley, C. J., Jacquet, J., West,
J. D., 2016, arXiv:1607.00376 [1, 6]Moss-Racusin, C., Dovidio, J., Brescoll, V., Graham, M.,
Handelsman, J., 2012, Proceedings of the National Academy ofSciences, USA, 109, 16474 [1]
National Science Foundation, Women, Minorities, and Personswith Disabilities in Science and Engineering 2015 [1]
Patat, F., 2016, The Messenger, 165, 2 [1]Pritchard, J., Karen, M., Akken, J., Contenta, F., Huckvale, L.,
Wilkins, S., Zocchi, A., 2014, Astronomy and Geophysics,55(6),8-12 [1]
Reid, N. I., 2014, Publ. Astron. Soc. Pacific 126, 923–934 [1]Wenneras, C., Wold, A., 1997, Nature, 387, 341 [1]West, J. D., Jacquet, J., King, M. M., Correll, S. J., Bergstrom,