Content uploaded by Arnfinn H. Midtbøen
Author content
All content in this area was uploaded by Arnfinn H. Midtbøen on Apr 01, 2021
Content may be subject to copyright.
Gender Bias in Academic Recruitment? Evidence
from a Survey Experiment in the Nordic Region
Magnus Carlsson
1
, Henning Finseraas
2
, Arnfinn H. Midtbøen
3,
*
and Guðbjo¨ rg Linda Rafnsdo´ ttir
4
1
Department of Economics and Statistics, Linnaeus University, Va¨xjo¨ 35195, Sweden,
2
Department of
Sociology and Political Science, Norwegian University of Science and Technology (NTNU), Trondheim
7491, Norway,
3
Institute for Social Research, Oslo 0208, Norway and
4
Faculty of Sociology, Anthropology
and Folkloristics, Faculty of Social and Human Sciences, University of Iceland, Reykjavı´k 102, Iceland
*Corresponding author. Email: ahm@socialresearch.no
Submitted June 2020; revised September 2020; accepted September 2020
Abstract
Gender disparities in top-level academic positions are persistent. However, whether bias in recruit-
ment plays a role in producing these disparities remains unclear. This study examines the role of bias
in academic recruitment by conducting a large-scale survey experiment among faculty in Economics,
Law, Physics, Political Science, Psychology, and Sociology from universities in Iceland, Norway, and
Sweden. The faculty respondents rated CVs of hypothetical candidates—who were randomly
assigned either a male or a female name—for a permanent position as an Associate Professor in their
discipline. The results show that, despite the underrepresentation of women in all fields, the female
candidates were viewed as both more competent and more hireable compared to their male
counterparts. Having children or a stronger CV do not change the overall result. Consequently, biased
evaluations of equally qualified candidates to Associate Professor positions do not seem to be the key
explanation of the persistent gender gap in academia in the Nordic region.
Introduction
Gender disparities in top-level academic positions are
slow to change. Despite a remarkable progress in wom-
en’s educational attainment since the 1960s (van Hek,
Kraaykamp and Wolbers, 2016;OECD, 2018), as well
as a slow gender convergence in attainment of faculty
positions (European Commission, 2019), women still
tend to be underrepresented in professor positions (Ceci,
2018;Cech and Blair-Loy, 2019). This holds true even
in the Nordic countries, well known for their institution-
alized gender equality norms (Teigen and Skjeie, 2017).
Although the Nordic countries rank at the top of the
Global Gender Gap Index, benchmarking 153 countries
on their progress towards gender parity (World
Economic Forum, 2020), women are strongly underre-
presented in professor positions in the Nordic region
(e.g., Bergman and Rustad, 2013;Nielsen, 2017).
To some, the persistent gender gap in top-level aca-
demic positions reflects historical differences in educa-
tional attainment and academic careers that will
naturally evaporate with time, as women eventually will
catch up with men (e.g., De Groot, 1997;Allen and
V
CThe Author(s) 2020. Published by Oxford University Press.
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommon s.org/licenses/by/4.0/),
which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited.
European Sociological Review, 2020, 1–12
doi: 10.1093/esr/jcaa050
Original Article
Downloaded from https://academic.oup.com/esr/advance-article/doi/10.1093/esr/jcaa050/6000745 by University of Oslo Library. Library of Medicine and Health Sciences user on 25 November 2020
Castleman, 2001). Others point to various sorting mech-
anisms as the key explanation for persistent gender in-
equality in academia: Because fewer women embark on
an academic career, especially in the fields of science,
technology, engineering, and mathematics (STEM), the
underrepresentation of female professors in these disci-
plines might be a matter of choices—constrained or
not—made long before they enter the academic system
(Ceci and Williams, 2011;Ceci, 2018). The higher rep-
resentation of women in the social sciences and human-
ities compared to STEM indeed indicates that sorting is
an important mechanism producing gender inequality.
However, women tend to be underrepresented at the
professor level in social sciences and humanities as well,
despite a strong female dominance among students
(Santos and Dang Van Phu, 2019), suggesting that the
phenomenon of ‘leaky pipelines’ are present even in
these fields.
A range of studies point to various forms of gender
bias as a major explanation for the persistent underre-
presentation of women at the top of the academic hier-
archy. Research has documented that male and female
students receive differential access to mentoring (Blau
et al., 2010) and entry-level positions as lab assistants
(Moss-Racusin et al., 2012). Some studies show that fe-
male researchers face bias-driven barriers in peer-review
processes (Wennera˚ s and Wold, 1997) and experience
more difficulties in achieving promotion and tenure
(Barbezat and Hughes, 2005). Research also shows that
family responsibilities affect the academic performance
of men and women differently (e.g., Lutter and
Schro¨ der, 2019), yet whether this is due to bias is a
debated question. Anders (2004) suggests that women
self-select away from academia in response to barriers
related to parenthood, while Xu (2008) claims that
women’s stronger turnover is due to dissatisfaction with
research support, advancement opportunities, and free
expression of ideas, rather than gender-based differences
in family responsibilities. Supporting the latter view, re-
cent research finds that hiring committees penalize fe-
male scholars in relationships because they are
considered ‘unmovable’—and hence less attractive—
compared to single women and men regardless of rela-
tionship status (Rivera, 2017).
In terms of gender bias in academic recruitment,
however, the existing evidence is mixed. To be sure, gen-
der discrimination in academic recruitment has been a
topic of study for decades, often suggesting that there is
a female advantage in hiring (Baldi, 1995;Merritt and
Reskin, 1997;Wolfinger, Mason and Goulden, 2008;
National Research Council, 2009;Lutter and Schro¨ der
2016). However, this body of work mainly consists of
observational studies. More recently, experimental stud-
ies have examined the occurrence of gender bias directly,
but, as in the literature on gender discrimination in the
labour market in general (Rich, 2014;Neumark, 2018),
the findings vary. Experiments involving hiring to non-
faculty positions within universities tend to find a bias in
favour of male job applicants (Foschi, Lai and Sigerson,
1994;Steinpreis, Anders and Ritzke, 1999;Moss-
Racusin et al., 2012), but for higher-level positions in
academia, experimental studies of gender bias in evalua-
tions find that female applicants have an advantage over
male applicants with similar qualifications (Williams
and Ceci, 2015;Ceci, 2018).
This article presents the results of a survey experi-
ment that investigates gender bias in recruitment among
faculty (N¼775) at universities in Iceland, Norway,
and Sweden in six disciplines (Economics, Law, Physics,
Political Science, Psychology, and Sociology) where
women are underrepresented at the level of Full
Professor. In the experiment, the faculty respondents
rated CVs of hypothetical candidates—who were ran-
domly assigned either a male or a female name—for a
permanent position as an Associate Professor in their
discipline. We asked faculty members to rate the candi-
dates according to their competence and hireability. The
design allows us to detect the occurrence of possible gen-
der bias in evaluations of male and female candidates to
Associate Professor positions. As each participant eval-
uated two CVs of the same candidate (one of them hav-
ing a more extensive publication record) and were
randomly assigned candidates with two children or no
children, we are moreover able to examine the effects of
children and publication records on faculty’s assess-
ments of competence and hireability, and whether these
effects differ for men and women.
Our research design combines the merits of previous
studies of gender bias in academic recruitment. We build
on Moss-Racusin et al.’s (2012) experimental design,
asking faculty to rate fictitious applicants’ competence
and hireability based on actual applicant material (CVs),
while similar to Williams and Ceci (2015), we study fac-
ulty’s evaluations of applicants for permanent associate
professor positions rather than early-career or even stu-
dent assistant positions. We used actual CVs because
evaluating a CV is an important part of the hiring pro-
cess in which stereotypes may play a role, while at the
same time being more neutral than a highly positive nar-
rative, which may leave limited room for bias in the
evaluation. Also, highly positive narratives, as in
Williams and Ceci (2015), may signal very qualified can-
didates, which could limit the external validity of the
results (see Heckman, 1998, for an elaboration of this
2European Sociological Review, 2020, Vol. 00, No. 0
Downloaded from https://academic.oup.com/esr/advance-article/doi/10.1093/esr/jcaa050/6000745 by University of Oslo Library. Library of Medicine and Health Sciences user on 25 November 2020
point). Thus, we carefully calibrated the CVs against an
existing pool of real CVs and pilot tested them to make
them representative for the respective discipline. Each
participant evaluated only female or male applicants to
avoid that they realized that the purpose of the experi-
ment was to study gender disparities (and perhaps pro-
vided socially desirable answers, e.g., rated the female
applicant higher).
Existing experimental studies on gender bias in aca-
demic recruitment are conducted exclusively in the
United States and focus only on the STEM fields, which
prevents the results from being generalized to the
European context and to other fields of study. Hence, the
current study is the first of its kind to examine whether
bias in recruitment is a likely explanation for the prevail-
ing gender gap in top-level academic positions in Europe.
We continue by presenting the theoretical framework
and the preregistered hypotheses of the study. Next, we
detail the research design before presenting the main
results. In conclusion, we interpret our findings and dis-
cuss their implications for the broader understanding of
persistent gender inequality in academia.
Theoretical Framework and Hypotheses
Most theories in the social sciences would suggest that
there is a bias against women in academic recruitment.
The stereotype content model, for example, is a model
of cognition that has received growing support for
explaining bias between groups of people in society,
including in the labour market (Fiske et al., 2002;Abele
et al., 2008). In this model, communion and agency are
central dimensions of social judgements that are
assumed to stem from evolution. Research in social cog-
nition has established that people judge other people on
the basis of these dimensions (Judd et al., 2005;Fiske,
Cuddy and Glick, 2007). The communion dimension
concerns people’s intentions and involves judgements of
other’s intent to either harm or help. The agency dimen-
sion involves judgements of other people’s capacity to
carry out the intent. People high on communion tend to
be characterized as trustworthy, empathic, and friendly,
while people high on agency tend to be characterized as
intelligent, skilled, creative, and efficient.
When people rate individuals on these dimensions,
women tend to receive higher scores on communion than
men, while men tend to receive higher scores on agency
(e.g., Fiske and Dupree, 2014). Apparently, such gender
stereotypes also exist for more specific groups of people,
such as professors. In line with the stereotype content
model, previous research has shown that female professors
are perceived as less qualified and have their work less val-
ued compared to males, but are instead perceived to pos-
sess feminine traits that include warmth and accessibility
(Miller and Chamberlin, 2000;Monroe et al., 2008).
An implication of the stereotype content model is
that a gender bias in hiring situations can emerge if there
are gender stereotypes about the traits that men and
women possess and these traits are incompatible with
the traits perceived to be needed for the job (Fiske et al.,
1991;Burgess and Borgida, 1999). The question is then
what traits are perceived to be needed to work as an
Associate Professor? Fiske and Dupree (2014) find that
the professor occupation is perceived to require more of
typical male traits (the agency dimension) and less of fe-
male traits (the communion dimension). To the extent
that these findings apply also for Associate Professor
positions, they suggest a potential mismatch between the
perceived traits that female applicants possess and the
perceived traits needed in the occupation. Together with
the empirical fact that women are underrepresented in
top-level academic positions, this motivated the follow-
ing main hypothesis in the survey experiment:
H1: Female applicants have lower ratings on compe-
tence and hireability than males.
Of course, the main hypothesis could also be moti-
vated from theories of discrimination in other disciplines,
e.g., economics and sociology, although the mechanisms
might look slightly different. In economics, theories of
statistical discrimination suggest that a gender bias in hir-
ing situations could emerge because of incomplete infor-
mation about the productivity of the applicants (Phelps,
1972;Arrow, 1973). In our case, men could, on average,
be perceived as having greater learned abilities and being
more productive in the professor occupation, which could
be a result of historical gender imbalances and biases. It is
also possible that women are expected to be less product-
ive in the professor occupation because of a higher likeli-
hood of having career disruptions due to greater family
responsibilities and a perception that long-term parental
leave could risk the progress of ongoing research projects.
These cases belong to the type of statistical discrimination
in which a prospective employer classify applicants on
the basis of group belonging and make judgements based
on the groups average productivity rather than on the
individual’s productivity which is partly unobserved.
Taking a sociological perspective, cultural beliefs
about the gendered nature of jobs could also lead to bias
in evaluations (e.g., Reskin and Roos, 1990;Ridgeway
and England, 2007). Indeed, much research has shown
that gender segregation in the labour market is not only
European Sociological Review, 2020, Vol. 00, No. 0 3
Downloaded from https://academic.oup.com/esr/advance-article/doi/10.1093/esr/jcaa050/6000745 by University of Oslo Library. Library of Medicine and Health Sciences user on 25 November 2020
a product of gender differences in human capital, but
also a consequence of how women and men are viewed
as suited for different jobs and work tasks (e.g., Correll,
2004;Correll, Benard and Paik, 2007;Ridgeway, 2011;
Orupabo, 2018). In order to understand the sources of
such social inequality, expectation states theory (e.g.,
Correll and Ridgeway, 2003;Ridgeway and Bourg,
2004) suggests that hierarchies of evaluation and influ-
ence are based on gendered beliefs about social status,
which produce different expectations as to what men
and women can accomplish. Following this logic, we
would expect that members of academic hiring commit-
tees could—consciously or not—translate ideas about
the gender of the applicant into discriminatory behav-
iour against women based on gender categorization and
stereotyped ideas of skills and suitability.
We further formulated an inferior hypothesis, which
concerns gender differences in the perceptions of appli-
cants with and without children. The idea is that having
children could be an important moderator for a potential
genderbiasinhiring.Cuddy, Fiske and Glick (2004) show
that working women who become mothers receive higher
ratings on communion, while their ratings on agency de-
crease. Thus, women seem to trade perceived agency for
perceived communion. However, working men who be-
come fathers gain in perceived communion and maintain
perceived agency. If this finding applies also for applicants
for Associate Professor positions, the mismatch between
the traits that women possess and the traits perceived to
be needed for the job should increase. Thus, we expect a
greater gender bias in favour of men among applicants
with children. This motivates the following hypothesis:
H2: Females have a lower return to children than males.
Finally, we formulated an additional inferior hypoth-
esis concerning the effect of having a strong CV. As we ex-
plain below, to help conceal that the main intent of the
experiment was to evaluate the prevalence of gender bias,
and thereby avoid socially desirable answers, we varied
the quality of the CVs. Thus, mainly as a byproduct, we
were able to investigate the quality of the CV as a possible
moderator for the gender bias. A certain type of statistical
discrimination models would predict that females have a
lower return to a strong CV. In particular, this type of
model may apply in environments where female applicants
are uncommon, such as in male-dominated disciplines in
academia and/or in environments in which there are few
female professors. The idea is that in this context a CV
could be a noisier productivity signal for female applicants
than for male applicants. As a result, there could be a gen-
der bias in favour of men, even if men and women have
the same perceived average productivity. We formulated
the following hypothesis:
H3: Females have a lower return to a strong CV than
males.
Research Design
In the experiment, the faculty respondents rated CVs of
hypothetical candidates for a permanent position as an
Associate Professor in their discipline. The experimental
design closely resembles academic appointments in the
Nordic region, where hiring procedures for academic
staff in public higher education in these countries are
strictly regulated. Vacant posts are normally advertised
openly and internationally. Applications to vacant posi-
tions typically include an application letter, a CV, and
the applicants’ choice of their most relevant publica-
tions. Applicants are ranked by an external committee,
and the top candidates are invited to an interview and
trial lecture, which are supervised by an internal com-
mittee. The internal committee conducts the final rank-
ing (Fro¨ lich et al., 2018).
Each CV in the experiment included information on
a number of attributes, most importantly the applicant’s
name (gender), whether the applicant has children (two
or none), and the applicant’s publication list. The
respondents were randomized into four groups, A–D,
and evaluated two CVs each. Group A evaluated a
hypothetical male candidate without children, Group B
a female candidate without children, Group C a male
applicant with two children, and, finally, Group D a fe-
male applicant with two children. The two CVs were
identical along the dimensions gender and number of
children, but, as mentioned above, we varied the quality
of the CV. The first CV had four publications, while the
second had six publications. All CVs had at least two
articles published in well-known journals in the respect-
ive fields. The randomization of gender and number of
children means that we can make causal interpretations
of the effect of these variables on the evaluation of the
CVs.
The effects of gender and children were evaluated
using a between-subjects design, while the quality of CV
effects was evaluated using a within-subjects design. The
between-subjects part of the design has an important ad-
vantage, as it helps conceal that the main intent of the
project is to evaluate the between-subject treatments
(gender and children). If the respondents would under-
stand the purpose of the experiment, they may give so-
cially desirable answers and create a bias in our
4European Sociological Review, 2020, Vol. 00, No. 0
Downloaded from https://academic.oup.com/esr/advance-article/doi/10.1093/esr/jcaa050/6000745 by University of Oslo Library. Library of Medicine and Health Sciences user on 25 November 2020
estimates. Moreover, the between-subjects part of the
design makes the data we collect less sensitive because
we will never measure a gender bias of a particular sub-
ject. For a particular subject, the data only reveal the
subject’s ranking of the publication list. Effects of the
quality of the CV were evaluated using a within-subjects
design.
Our research design has several advantages over
other designs. With observational studies, it has proven
to be very difficult to measure unequal treatment in a
credible manner due to confounding factors. With
laboratory experiments, it is difficult to create realistic
situations to study, where the results also apply outside
of the laboratory. With interviews with job applicants, it
is difficult to know if the answers reflect real or per-
ceived discrimination and with interviews with faculty
members, one cannot be sure that the answers are con-
sistent with their behaviour. A survey experiment of gen-
der bias addresses many of the problems associated with
these other methods. Although we will not study real
hiring processes, our research design will be useful to
substantiate whether discrimination is likely to occur in
real hiring processes at the sampled universities.
Before collecting the data, we described all aspects of
the research design, variable operationalizations, model
specifications, robustness checks, and handling of miss-
ing data in a preregistered research plan that was sub-
mitted to the Evidence in Governance and Politics’
(EGAP) research register.
Choice of Subjects and Institutions
We started from three criteria as guidelines when choos-
ing disciplines. The first was to include disciplines with
a low share of female professors. The second criterion
was to include disciplines with a high share of female
PhD students relative to the share of female professors.
These first two criteria indicate problems in recruiting
female professors. The third criterion was that the num-
ber of researchers in the discipline, i.e., potential partici-
pants, had to be sufficiently large to substantially
contribute to the data collection. In practice, it turned
out that the third criterion was most important. The rea-
son is that many broader disciplines consist of sub-fields
in which very specific CVs have to be used, meaning
that, if included, we would have had to treat them as
separate disciplines and create a vast number of specific
CVs for these sub-fields. Therefore, in the end, the
chosen disciplines—Economics, Law, Physics, Political
Science, Psychology, and Sociology—were those with
many employed researchers, and they should give a fair-
ly representative picture of the prevalence of gender bias
in Iceland, Norway, and Sweden. Figure 1 shows the
share of female professors in the included disciplines.
We included all large institutions in the three countries
in the survey experiment. Altogether, 17 institutions
were included. We pool the data from the three coun-
tries, since the sample size is too small to analyse the
data by country.
Construction of Hypothetical CVs
To create realistic CVs, we studied a large number of
real CVs in each discipline available online, typically at
researchers’ personal homepages. We also consulted col-
leagues in each discipline to give feedback on the CVs.
Finally, in March 2019, we conducted a pilot study
(N¼22) at a Swedish university not part of the main
study. The first aim of the pilot was to check that the
distribution of the outcome variables was reasonable
centred on the answering scale, 1–7. We found that the
mean of the competence and hireability indices were 4.5
and 4.0, respectively. The second aim was to gather
comments and feedback through an open question at the
end of the survey. After the pilot study, we made several
minor changes of the survey questions and the CVs.
Each constructed CV fits on a single page to facilitate
that the participants easily can overview the content. At
the very top of the CVs, there is personal information such
as the candidate’s name, birth date, nationality, and civil
status (marital status and the number of children). Next
follows information about the candidate’s research inter-
ests, employment history, education, teaching experience,
experience of professional services and memberships in re-
search networks and organizations, and, finally, publica-
tions in international peer-reviewed journals.
The experimental manipulations are the name of the
applicant, which signals gender, civil status, which states
that the candidates are married and have either no chil-
dren or two children, and the publication list, which
contains either four or six international peer-reviewed
publications. Supplementary Figures S1 and S2 in
Appendix E show two example CVs.
Outcomes and Other Variables
We had two main outcomes in the experiment. The first
was an evaluation of the competence of the applicant.
Our measure of competence is an additive index con-
structed from the answers to the following three ques-
tions (see the preregistered research plan): (i) To what
extent do you consider the applicant as competent for
the position?, (ii) To what extent do you consider the ap-
plicant to possess the necessary qualifications for the
position?, and (iii) How qualified do you consider the
European Sociological Review, 2020, Vol. 00, No. 0 5
Downloaded from https://academic.oup.com/esr/advance-article/doi/10.1093/esr/jcaa050/6000745 by University of Oslo Library. Library of Medicine and Health Sciences user on 25 November 2020
applicant to be? The respondent answered the questions
on a scale 1–7 where 1 is the lowest score and 7 is the
highest score. The second outcome was an evaluation of
the hireability of the applicant. Our measure of hirabil-
ity is an additive index constructed of the answers to fol-
lowing three questions: (i) How likely is it that you
would hire the applicant?, (ii) How likely is it that you
would invite the applicant to a job interview?, and (iii)
What is your assessment of the probability that the ap-
plicant will get the job? The respondent answered these
questions too on a scale 1–7. The distributions of the
outcome variables competence and hireability are shown
in Supplementary Figure S5 in Appendix F.
We also asked the respondents questions about back-
ground characteristics. These were: gender (a binary indi-
cator for being female or male), age (measured as a
continuous variable), number of years since obtaining the
PhD degree (a continuous variable), number of years
employed at the current institution (a continuous
variable), whether having participated in a hiring
committee for permanent positions during the last 5 years
(a binary variable; yes or no), faculty position (a binary in-
dicator for Full Professor or not) and citizenship (a binary
indicator for being a citizen in the survey country, e.g.,
being an Icelandic citizen and participating in the survey
in Iceland). These variables are used for balance and ro-
bustness checks (see Supplementary Appendix B).
Sampling
We employed two approaches to collect data to ensure
that enough respondents participated in the study. In the
first approach, research assistants collected emails for all
relevant faculties at the largest universities in Iceland,
Norway, and Sweden. The participants were then
recruited by email to fill out an electronic survey in the
survey software Qualtrics (www.qualtrics.com). In the
second approach, the participants were recruited at in-
ternal faculty meetings. We contacted departments in
the universities and asked for access to internal meetings
where we could distribute the survey. Respondents were
asked to fill out the survey, using pen and paper, on the
spot without communicating with others. Out of the
775 participants, 706 were recruited through the first
method and 69 through the second method.
Figure 1. Share of women in Full Professor positions in Iceland, Norway, and Sweden, by discipline, 2017
Notes: In Iceland and Sweden, Economics include both Economics and Business Administration. In Norway, the numbers include
both Full Professors and Docents, the latter being an academic appointment between Associate Professor and Full Professor. Data
for Iceland are based on our own compilations of staff lists available at the webpages of all Icelandic universities. Data for Sweden
are obtained from the Swedish Higher Education Authority. Data for Norway are obtained from the Nordic Institute for Studies in
Innovation, Research and Education (NIFU).
6European Sociological Review, 2020, Vol. 00, No. 0
Downloaded from https://academic.oup.com/esr/advance-article/doi/10.1093/esr/jcaa050/6000745 by University of Oslo Library. Library of Medicine and Health Sciences user on 25 November 2020
Altogether, the targeted population consists of ap-
proximately 2,000 individuals, of whom 775 (39 per
cent) participated in the study and voluntarily ranked
applications. The number of responses varied from 93 in
economics to 155 in psychology. Among those who
answered the background questions in the survey, 66
per cent are male, the average age is 49, 59 per cent have
experience from hiring committees, and 46 per cent are
full professors. Eight per cent, 43 per cent, and 51 per
cent work at an Icelandic, Norwegian, and Swedish uni-
versity, respectively. Respondents self-select to partici-
pate in the survey, but the rather high response rate and
the descriptive statistics suggest that the sample is fairly
representative of the population. Also, since we con-
ducted an experiment, self-selection to the study will not
harm the internal validity, but might limit the external
validity, of the results.
Data Analysis
We estimated the treatment effects in a regression frame-
work. To test H1, we ran the following regression separ-
ately both on evaluations of the baseline CV with four
publications (N¼775) and on evaluations of all CVs
(N¼1,550), i.e., the pooled sample of baseline CVs and
CVs with a strong publication list:
yijk ¼bFFEMALEjk þijk (1)
In this model, iindexes respondent, jdiscipline, and
kinstitution. The FEMALE dummy is equal to 1 for
those receiving the gender treatment (CV of type B and
D). Hypothesis H1 (that females have lower ratings than
males) implies that bF<0.
To test H2, we ran the following regression using evalu-
ations of the baseline CV and the pooled sample of all CVs:
yijk ¼pFFEMALEjk þpCCHILDRENjk
þpFCFEMALEjk CHILDRENjk þijk (2)
In this model, CHILDREN is equal to 1 for groups
receiving the children treatment (CVs of type C and D).
Hypothesis H2 (that females have a lower return to chil-
dren than males) implies that pFC <0.
To test H3, we run the following regression:
yijk ¼pFFEMALEjk þpCCHILDRENjk þpVCVijk
þpFCFEMALEjk CHILDRENjk
þpFV FEMALEjkCVijk þijk (3)
In this model, CV is equal to 1 for the evaluation of
the strong CV and otherwise 0. Hypothesis H3 (that
females have a lower return to a strong CV than males)
implies that pFV <0.
Results
We analysed treatment effects using OLS regressions.
Figure 2 displays the main results graphically (while for-
mal tests of whether differences are statistically signifi-
cant are presented in Table 1). The dots in the figure
show the mean scores for male and female CVs, while
the lines show the confidence intervals. Panel A in
Figure 2 presents results for the competence ratings,
using only the baseline CV with four publications (to the
left) and using all CVs, i.e., with four and six publica-
tions (to the right). Panel B in Figure 2 shows the corre-
sponding results for the hireability ratings. Contrary to
our expectations, we find that, for both competence and
hireability, female CVs get higher ratings than male
CVs. However, as expected, the average ratings for both
males and females are higher when we include the CVs
with six publications (to the right in the figure).
Table 1 shows the formal tests of whether female CVs
are evaluated differently from male CVs. Again, panel A
Table 1 uses only the baseline CVs, while panel B uses the
pooled sample of CVs. In these regressions, the constants
are the average evaluations of the male candidate, while
the coefficient for a female CV shows the difference in rat-
ing for the female candidate compared to the male candi-
date. In panel A, we see that, on average, a male
candidate’s CV is rated 4.26 on competence and 3.77 on
hireability and a female candidate’s CV is rated 0.29
higher on competence and 0.28 higher on hireability. Both
differences are statistically significant (at the 1 and 5 per
cent level, respectively). In panel B, using all observations,
we obtain very similar results, but the precision is slightly
better.
In Supplementary Appendix B, we present the results
of the preregistered balance and robustness tests. They
show balance across treatment groups for all background
variables except gender, which most likely is the result of
an unlucky draw from the population. In practice, this
imbalance for gender seems unimportant, since the results
remain unchanged when we control for gender and other
background characteristics of the participants (for
details, see Balance and Robustness Checks section in
Supplementary Appendix B). Moreover, we repeat the
regressions in Table 1 including a dummy for whether the
survey was answered by paper and pencil or online, dis-
cipline fixed effects, institution fixed effects, and discip-
line times institution fixed effects. These variables could
increase the precision of the estimates and change the esti-
mated treatment effects if there is an imbalance in the
share of treated in different groups of respondents (e.g.,
age or gender groups). The treatment effect barely moves
across the robustness checks. We therefore reject H1.
European Sociological Review, 2020, Vol. 00, No. 0 7
Downloaded from https://academic.oup.com/esr/advance-article/doi/10.1093/esr/jcaa050/6000745 by University of Oslo Library. Library of Medicine and Health Sciences user on 25 November 2020
If we estimate the treatment effect for H1 separately
by academic field, we find that the treatment effect esti-
mates are smaller for Law and higher for Physics and
Psychology. However, a formal test does not rule out
that the treatment effect is the same across fields (see
Supplementary Table S3 in Appendix A).
Finally, we find no support for H2 and H3. The treat-
ment effect estimates are always small (below 0.1) and
statistically insignificant (see Supplementary Tables S1 and
S2 in Appendix A). Thus, somewhat surprisingly, we find
no evidence that the pay-off for children and for a strong
CV is lower for women than for men. This conclusion
remains across the robustness and specification checks.
Discussion and Conclusion
For decades, scholars have debated whether bias against
women explains the persistent gender gap in top-level aca-
demic positions. This study has examined the role of bias
in the critical moment when candidates are evaluated for
positions as Associate Professors. In the Nordic countries,
landing a position as an Associate Professor translates into
a permanent career in academia and is a necessary
stepping-stone for later promotion to the position as Full
Professor. To the extent that female researchers experience
bias in this crucial stage in their academic careers, it would
help explaining why women are underrepresented at the
top of the academic hierarchy even in the ‘women-friendly’
Nordic region (Borchorst and Siim, 2002;Nielsen, 2017).
Despite mixed results in the existing experimental lit-
erature on gender bias in academic recruitment (Moss-
Figure 2. Ratings of CVs for male and female candidates (Panel A: Competence. Panel B: Hirability).
Note: The gender difference in each graph is statistically significant (see Table 1).
Table 1. Ratings of CVs for male and female candidates
Competence Hireability
(1) (2)
Panel A (N¼775):
Female CV 0.2949
***
0.2842
**
(0.0992) (0.1153)
Constant (male CV) 4.2596
***
3.7710
***
(0.0707) (0.0821)
Panel B (N¼1,550):
Female CV 0.3075
***
0.2999
***
(0.0937) (0.1070)
Constant (male CV) 4.4095
***
3.9689
***
(0.0688) (0.0769)
Notes: The regressions include no other covariates than the female CV
dummy. In Panel B, standard errors are clustered by respondent.
***Significant at the 1 per cent level,
**significant at the 5 per cent level;
8European Sociological Review, 2020, Vol. 00, No. 0
Downloaded from https://academic.oup.com/esr/advance-article/doi/10.1093/esr/jcaa050/6000745 by University of Oslo Library. Library of Medicine and Health Sciences user on 25 November 2020
Racusin et al., 2012;Williams and Ceci, 2015;Ceci,
2018), we expected our survey experiment to reveal a
male advantage in the evaluation of candidates to
Associate Professor positions. We also expected female
candidates to have a lower return to children and strong
CVs than males. The main rationale behind these
hypotheses was the persistent gender gap at the top level
in all countries and almost all fields examined, com-
bined with theories of discrimination and stereotype
content, which suggest that women—and especially
women with children—would be viewed as less compe-
tent and hireable than men.
Contrary to our main hypothesis, however, we did
not find evidence of a bias against female applicants to
Associate Professor positions in the Nordic region.
Rather, female candidates are perceived as both more
competent and hireable compared to equally qualified
male candidates. Furthermore, we find no evidence of a
child penalty for neither male nor female applicants and
no gender difference in the pay-off from a strong CV.
A potential limitation of the study is that we cannot
rule out self-selection to the survey experiment. This might
lower the external validity of the results as we cannot be
certain that the participants’ ratings fully reflect the atti-
tude of the full population in the chosen disciplines.
However, the rather high response rate (39 per cent)
should make this less of a concern. Another limitation is
that although we aimed for obtaining a fairly representa-
tive picture of the prevalence of gender bias in Iceland,
Norway, and Sweden by including six disciplines with
many employed researchers (Economics, Law, Physics,
Political Science, Psychology, and Sociology), we cannot
rule out the possibility that the inclusion of other disci-
plines would affect the results.
Our results stand in contrast to many previous studies
of various gender disparities in academia, which suggest
that bias against women is widespread (e.g., Wennera˚s
and Wold, 1997;Barbezat and Hughes, 2005;Moss-
Racusin et al.,2012). However, our findings are nonethe-
less in line with a number of existing studies on academic
hiring that suggest a female advantage (Baldi, 1995;
Merritt and Reskin, 1997;Wolfinger, Mason and
Goulden, 2008;National Research Council, 2009;Lutter
and Schro¨der, 2016). In particular, our findings are con-
sistent with Williams and Ceci (2015), the only existing
experimental study that assesses the prevalence of gender
bias in recruitment to tenure track positions. Yet, the fe-
male advantage we find is far less pronounced than the 2-
to-1 female advantage that Williams and Ceci (2015) re-
port. The smaller effects in our study could reflect differ-
ences in research design. Specifically, in Williams and
Ceci (2015), the same subject evaluated both male and fe-
male applicants (while we used a between-subject design)
and their subjects were asked to rate detailed narratives
of exceptionally qualified applicants (instead of CVs).
These differences could have led to a particularly strong
assessment of the female candidates. Another important
difference is that Williams and Ceci (2015) examined
only STEM fields. Finally, differences in results could re-
flect that the studies were conducted at different times
and in different institutional contexts.
What could explain our finding that, if anything, the
bias is in favour of female candidates? This is a difficult
question to answer fully. However, one possibility is
related to the fact that the present study has been con-
ducted in the Nordic region, well known for its institu-
tionalized gender equality norms (Teigen and Skjeie,
2017). These norms could impact, e.g., hiring commit-
tees, making them aware of the demands in the Nordic
region to hire more female academics. To the extent that
this demand is widely accepted in academic commun-
ities, it is possible that the respondents have internalized
this and consequently given more favourable evaluations
of the female candidates.
Regardless of how we explain the results of this study,
one important question remains: How can we account for
the persistent gender gap in top-level academic positions if
bias in recruitment is not the reason? One potential explan-
ation is that the underrepresentation of women in profes-
sor positions is the result of sorting mechanisms occurring
at earlier stages in the academic career, which—con-
strained or not—could lead fewer women to ascend to pro-
fessor positions. Another potential explanation is that bias
against female academics occurs earlier in career trajectory:
The lack of bias against female applicants in our study
does not rule out the possibility that men experience
advantages in other phases of academic life, such as in
monitoring, review boards, or peer-review assessments.
Indeed, the virtue of experiments such as this is the ability
to examine directly whether gender bias exists in evalua-
tions of candidates of different gender, all else being equal.
Bias in processes prior to the event of applying for a pos-
ition as an Associate Professor is not studied in experi-
ments of our kind, nor are sorting mechanisms that may
lead more men than women to embark on an academic
career. If such biases and sorting mechanisms are wide-
spread, they might be an important explanation of the
existing gender gap in academic top positions. Yet when
female applicants have succeeded in getting on par with
their male peers, our study suggests that women are not
less likely than men to being awarded permanent positions
in academia.
European Sociological Review, 2020, Vol. 00, No. 0 9
Downloaded from https://academic.oup.com/esr/advance-article/doi/10.1093/esr/jcaa050/6000745 by University of Oslo Library. Library of Medicine and Health Sciences user on 25 November 2020
Supplementary Data
Supplementary data are available at ESR online.
Funding
This work was supported by NordForsk (grant number 80713)
and is part of the Nordic Centre for Research on Gender
Equality in Research and Innovation (NORDICORE).
Acknowledgements
We thank the two anonymous reviewers and the journal’s edi-
tors for constructive comments to a previous version of the
manuscript, and participants at the NORDICORE meetings at
the Institute for Social Research in Oslo, Norway, for their feed-
back to the research design and analyses. Many thanks also to
A
´stro´ s Anna, Olof Axman, Live Kjos Fjell, Øyvind Skorge,
Maya Staub, and Sahra Ali Abdullahi Torjussen for invaluable
contributions in different stages of the research process.
References
Abele, A. E. et al. (2008). Fundamental dimensions of social judg-
ment. European Journal of Social Psychology,387, 1063–1065.
Allen, M. and Castleman, T. (2001). Fighting the pipeline fal-
lacy. In Brooks, A. and Mackinnon, A. (Eds.), Gender and the
Restructured University Changing Management and Culture
in Higher Education. Buckingham: The Society for Research
into Higher Education & Open Press University. pp. 151–165.
Anders, S. (2004). Why the academic pipeline leaks: fewer men
than women perceive barriers to becoming professors. Sex
Roles,51, 511–521.
Arrow, K. J. (1973). The theory of discrimination. In
Ashenfelter, O. and Rees, A.(Eds.), Discrimination in Labor
Markets. Princeton, NJ: Princeton University Press, pp. 3–42.
Baldi, S. (1995). Prestige determinants of first academic job for
new sociology Ph.D.s. The Sociological Quarterly,36,
777–789.
Barbezat, D. A. and Hughes, J. W. (2005). Salary structure
effects and the gender pay gap in academia. Research in
Higher Education,46, 621–640.
Bergman,S., & Rustad, L. M. (2013). The Nordic region – a step
closer to gender balance in research? Joint Nordic strategies
and measures to promote gender balance among researchers
in academia. TemaNord Report 2013:544. Copenhagen:
Nordic Council of Ministers.
Blau, F. D. et al. (2010). Can mentoring help female assistant
professors? Interim results from a randomized trial. American
Economic Review,100, 348–352.
Borchorst, A. and Siim, B. (2002). The women-friendly welfare
states revisited. NORA - Nordic Journal of Feminist and
Gender Research,10, 90–98.
Burgess, D. and Borgida, E. (1999). Who women are, who women
should be: descriptive and prescriptive gender stereotyping in sex
discrimination. Psychology, Public Policy, and Law,53,
665–692.
Cech, E. A. and Blair-Loy, M. (2019). The changing career tra-
jectories of new parents in STEM. Proceedings of the National
Academy of Sciences of the United States of America,116,
4182–4187.
Ceci, S. J. (2018). Women in academic science: experimental find-
ings from hiring studies. Educational Psychologist,53, 22–41.
Ceci, S. J. and Williams, W. M. (2011). Understanding current
causes of women’s underrepresentation in science.
Proceedings of the National Academy of Sciences of the
United States of America,108, 3157–3162.
Correll, S. J. (2004). Constraints into preferences: gender, status,
and emerging career aspirations. American Sociological
Review,69, 93–113.
Correll, S. J., Benard, S. and Paik, I. (2007). Getting a job: is
there a motherhood penalty? American Journal of Sociology,
112, 1297–1339.
Correll, S. J. and Ridgeway, C. L. (2003). Expectation states theory.
In Delamater, J. (Ed.), Handbook of Social Psychology.New
York, NY: Kluwer Academic/Plenum Publishers, pp. 29–51.
Cuddy, A. J. C., Fiske, S. T. and Glick, P. (2004). When
Professionals Become Mothers, Warmth Doesn’t Cut the Ice.
Journal of Social issues, 60, 701-718.
De Groot, J. (1997). After the ivory tower: gender, commodifica-
tion and the ‘academic’. Feminist Review,55, 130–142.
European Commission. (2019). She Figures 2018. Luxembourg:
Publications Office of the European Union.
Fiske, S. T., Cuddy, A. J. and Glick, P. (2007). Universal dimen-
sions of social cognition: warmth and competence. Trends in
Cognitive Sciences,112, 77–83.
Fiske, S. T. and Dupree, C. (2014). Gaining trust as well as re-
spect in communicating to motivated audiences about science
topics. Proceedings of the National Academy of Sciences of
the United States of America,111, 13593–13597.
Fiske, S. T. et al. (1991). Social science research on trial: use of
sex stereotyping research in Price Waterhouse v. Hopkins.
American Psychologist,46, 1049.
Fiske, S. T. et al. (2002). A model of (often mixed) stereotype
content: competence and warmth respectively follow from
perceived status and competition. Journal of Personality and
Social Psychology,82, 878.
Foschi, M., Lai, L. and Sigerson, K. (1994). Gender and double
standards in the assessments of job candidates. Social
Psychology Quarterly,57, 326–339.
Fro¨ lich, N. et al. (2018). Academic Career Structures in Europe:
Perspectives from Norway, Denmark, Sweden, Finland, the
Netherlands, Austria and the UK. Oslo: Nordic Institute for
Studies in Innovation, Research and Education (NIFU).
Heckman, J. J. (1998). Detecting discrimination. Journal of
Economic Perspectives,12, 101–116.
Judd, C. M. et al. (2005). Fundamental dimensions of social
judgment: understanding the relations between judgments of
competence and warmth. Journal of Personality and Social
Psychology,896, 899.
10 European Sociological Review, 2020, Vol. 00, No. 0
Downloaded from https://academic.oup.com/esr/advance-article/doi/10.1093/esr/jcaa050/6000745 by University of Oslo Library. Library of Medicine and Health Sciences user on 25 November 2020
Lutter, M. and Schro¨ der, M. (2016). Who becomes a tenured
professor, and why? Panel data evidence from German soci-
ology, 1980–2013. Research Policy,45, 999–1013.
Lutter, M. and Schro¨ der, M. (2019). Is there a motherhood pen-
alty in academia? The gendered effect of children on academic
publications in German sociology. European Sociological
Review,36, 442–459.
Merritt, D. J. and Reskin, B. F. (1997). Sex, race, and creden-
tials: the truth about affirmative action in law faculty hiring.
Columbia Law Review,97, 199–311.
Miller, J. and Chamberlin, M. (2000). Women are teachers, men
are professors: a study of student perceptions. Teaching
Sociology, 28, 283–298.
Monroe, K. et al. (2008). Gender equality in academia: bad
news from the trenches, and some possible solutions.
Perspectives on Politics,62, 215–233.
Moss-Racusin, C. et al. (2012). Science faculty’s subtle gender
biases favor male students. Proceedings of the National Academy
of Sciences of the United States of America,109, 16474–16479.
National Research Council. (2009). Gender Differences at
Critical Transitions in the Careers of Science, Engineering and
Mathematics Faculty. Washington, DC: National Academy
Press.
Neumark, D. (2018). Experimental research on labor market
discrimination. Journal of Economic Literature,56, 799–866.
Nielsen, M. W. (2017). Scandinavian Approaches to Gender
Equality in Academia: A Comparative Study. Scandinavian
Journal Ofeducational Research, 61, 295-318.
OECD. (2018). Education at a Glance 2018. Paris: OECD
Publishing.
Orupabo, J. (2018). Cultural stereotypes and professional
self-socialisation in the transition from education to work. Journal
of Education and Work,31, 234-246.
Phelps, E. S. (1972). The statistical theory of racism and sexism.
The American Economic Review,624, 659–661.
Reskin, B. F. and Roos, P. A. (1990). Job Queues, Gender
Queues: Explaining Women’s Inroads into Male Occupations.
Philadelphia, PA: Temple University Press.
Rich, J. 2014. What Do Field Experiments of Discrimination in
Markets Tell Us? A Meta Analysis of Studies Conducted since
2000. IZA DP No. 8584.
Ridgeway, C. and England, P. 2007. Sociological approaches to
sex discrimination in employment. In Sex Discrimination in
the Workplace. Oxford, Blackwell.
Ridgeway, C. L. (2011). Framed by Gender: How Gender
Inequality Persists in the Modern World. Oxford: Oxford
University Press.
Ridgeway, C. L. and Bourg, C. (2004). Gender as status: an ex-
pectation states theory approach. In Eagly, A. H., Beall, A. E.
and Sternberg, R. J. (Eds.), The Psychology of Gender. New
York, NY: The Guilford Press, pp. 217–241.
Rivera, L. A. (2017). When two bodies are (not) a problem: gender
and relationship status discrimination in academic hiring.
American Sociological Review,82, 1111–1138.
Santos, G. and Dang Van Phu, S. (2019). Gender and academic
rank in the UK. Sustainability,11, 3171.
Steinpreis, R., Anders, R. K. and Ritzke, K. D. (1999). The impact
of gender on the review of the CVs of job applicants and tenure
candidates: a national empirical study. Sex Roles,41, 509–528.
Teigen, M., and Skjeie, H. (2017). The Nordic gender equality
model. In Knutsen, O. (Ed.), The Nordic Models in Political
Science. Challenged but Still Viable? Oslo: Fagbokforlaget.
pp. 125–147.
van Hek, M., Kraaykamp, G. and Wolbers, M. H. J. (2016).
Comparing the gender gap in educational attainment: the im-
pact of emancipatory contexts in 33 cohorts across 33 coun-
tries. Educational Research and Evaluation,22, 260–282.
Wennera˚s, C. and Wold, A. (1997). Nepotism and sexism in
peer-review. Nature,387, 341–343.
Williams, W. M. and Ceci, S. J. (2015). National hiring experi-
ments reveal 2:1 faculty preference for women on STEM ten-
ure track. Proceedings of the National Academy of Sciences of
the United States of America,112, 5360–5365.
Wolfinger, N. H., Mason, M. A. and Goulden, M. (2008).
Problems in the pipeline: gender, marriage, and fertility in the
ivory tower. Journal of Higher Education,79, 388–405.
World Economic Forum. (2020). The Global Gender Gap
Report 2019. Geneva: World Economic Forum, available
from: http://repor ts.weforum.org/global-gender-gap-report-
2020/ [accessed 6 April 2020].
Xu, Y. J. (2008). Gender disparity in STEM disciplines: a study
of faculty attrition and turnover intentions. Research in
Higher Education,49, 607–624.
Magnus Carlsson is an associate professor in eco-
nomics at Linnaeus University, Sweden. Current re-
search interests include labour economics and
applied microeconomics, and a large part of his re-
search aims at studying discrimination and which
factors are important in explaining observed group
differences in labour market outcomes. Carlsson’s
work has been published in journals such as The
Review of Economics and Statistics,Labour
Economics, and ILR Review.
Henning Finseraas is an associate professor in polit-
ical science at the Norwegian University of Science
and Technology (NTNU) and a research professor at
the Institute for Social Research in Oslo, Norway.
Current research interests include electoral behaviour
and public opinion, and quantitative evaluations of
public policy, often related to immigration, political
economics, and the welfare state. Finseraas’ work has
been published in the journals American Journal of
Political Science,British Journal of Political Science,
and Journal of Public Economics.
European Sociological Review, 2020, Vol. 00, No. 0 11
Downloaded from https://academic.oup.com/esr/advance-article/doi/10.1093/esr/jcaa050/6000745 by University of Oslo Library. Library of Medicine and Health Sciences user on 25 November 2020
Arnfinn H. Midtbøen is an associate professor in
sociology at the University of Oslo and a research
professor at the Institute for Social Research in Oslo,
Norway. Current research interests include economic
and political sociology, and frequent topics are labour
market discrimination, citizenship policies, and in-
corporation patterns among descendants of migrants.
Midtbøen’s work has been published in journals such
as Proceedings of the National Academy of the
Sciences (PNAS),British Journal of Sociology,and
International Migration Review.
Guðbjo¨ rg Linda Rafnsdo´ ttir is a pro-rector of science
and a professor in sociology at the University of
Iceland. Current research interests include gender seg-
regation, gender quotas, and work/family balance.
Rafnsdo´ ttir’s work has been published in journals
such as Gender Work and Organization,Politics and
Gender,andWork, Employment and Society.
12 European Sociological Review, 2020, Vol. 00, No. 0
Downloaded from https://academic.oup.com/esr/advance-article/doi/10.1093/esr/jcaa050/6000745 by University of Oslo Library. Library of Medicine and Health Sciences user on 25 November 2020
