1. Gender, beauty and support networks in
academia: evidence from a field
experiment
Magdalena Smyk
Michał Krawczyk
Warsaw Economic Seminar 2016
Group for Research in Applied Economics
2. 2
Motivation
Gender differences in academic productivity – large
unexplained component
Social support networks and gender inequalities in academia
Participation in social network increase probability of
receiving job offer (McDonald, 2011) and scientific
productivity (Reagans and Zuckerman, 2001).
Differences in experience between women and men:
Mentoring (Chandler, 1996)
Collaboration (Gersick et al. 2000)
3. 3
Can we blame „ old-boys network”?
Do scholars prefer to „lend a hand” to male researcher
rather than female?
4. Study I (data request)
• 247 papers (with experiments that meet certain
criteria)
• Ask for raw data from their experiments
• E-mails from two accounts:
– Female student
– Male student
• Randomly chosen samples of subjects:
– equal distribution of male and female subjects
– three geographical regions (Europe, Australia and Asia,
Americas).
• After three weeks - reminder
4
5. 5
Study I – measures of success
Two measures:
Response rate = number of responses we received/
number of e-mails sent (succesfully)
Compliance rate = number of datasets we
received/ number of e-mails sent
6. 6
Study I (data request) - RESULTS
Female Student Male Student
No. of requests 100 105
Response rate 75% 74.3%
MWW test (p-value) 0.91
Marginal effects* -0.01 (insignificant)
Compliance rate 34% 35.2%
MWW test (p-value) 0.85
Marginal effects* -0.02 (insignificant)
Notes: *probit regression; gender, university region, fixed effects of journal, date of sending
the request and number of datasets we asked for.
7. 7
Study II
Extension:
10 fields of study: psychology, sociology, economics,
mathematics, law, computer science, philosophy, medicine,
physics and chemistry
two types of request (much smaller):
Article treatment – we ask for full text of subject’s paper
Meeting treatment – we ask for a meeting during office
hours or Skype/phone call to discuss possible mentoring
for graduate studies
additional dimension: physical attractiveness
10. 10
Subjects in Study II
One hundred top faculties from QS World University Rankings
Four (randomly chosen) scholars from each faculty
Faculties without websites or without list of employees –
excluded
Article Treatment – 1287 scholars (without non-English writers
and scholars without papers)
Meeting Treatment – 1488 scholars
Lack of gender balance in the sample (much more males)
11. 11
Study II – measures of success
Response rate = number of responses we received/
number of e-mails sent (succesfully)
Article Treatment:
Compliance rate = number of full papers we
received/ number of e-mails sent
Meeting Treatment:
Compliance rate = number of meeting
aggrement or proposition/ number of e-mails
sent
12. 12
Study II – results (Article Treatment)
Attractive
Female
Less
Attractive
Female
Attractive
Male
Less
Attractive
Male
No. of requests 343 307 337 300
Response rate 56.6% 67.1% 63.2% 62.4%
MWW test p-value
(vs. attractive female) 0.006 0.08 0.08
(vs. less attractive female) 0.3 0.33
(vs. attractive male) 0.97
Compliance rate 49% 60% 56.7% 54.8%
MWW test p-value
(vs. attractive female) 0.005 0.04 0.2
(vs. unattractive female) 0.4 0.14
(vs. attractive male) 0.5
13. 13
Study II – results (Meeting Treatment)
Attractive
Female
Less
Attractive
Female
Attractive
Male
Less
Attractive
Male
No. of requests 370 378 374 366
Response rate 45.7% 47.6% 43.9% 44.3%
MWW test p-value
(vs. attractive female) 0.59 0.62 0.7
(vs. less attractive female) 0.3 0.36
(vs. attractive male) 0.91
Compliance rate 29.2% 34.4% 27% 27.6%
MWW test p-value
(vs. attractive female) 0.13 0.51 0.63
(vs. unattractive female) 0.03 0.05
(vs. attractive male) 0.86
14. 14
Study II – results (probit)
Article
treatment
(1)
Article
treatment
(2)
Meeting
treatment
(1)
Meeting
treatment
(2)
response compliance response compliance
attractive female -0.08* -0.11*** 0.02 0.02
less attractive female 0.03 0.18 0.05 0.09***
less attractive male -0.02 -0.06 -0.005 0.001
female scholar -0.05 -0.07** -0.09*** -0.09***
Observations 1287 1287 1488 1488
Notes: Marginal effects from probit regressions; reference category is attractive male; regressions include
subjects’ characteristics (gender, university region, university ranking position, field of study), date of sending
the request and year of the paper publication (in Article treatment); *** p<0.01, ** p<0.05, * p<0.1.
15. 15
Robustness check and additional dimensions
Interactions: gender of the subject and treatment –
insignificant = lack of „old-boys network” signs
Stronger results (higher marginal effect) in subsample of
subjects who has G-Talk option available
Lack of specific field effects
Nr of unique vistors on websites = 44% of the nr of subjects
Attractive senders websites more popular by 10 pp on average
Refusals in Meeting Treatment:
55/124 (males) to 34/111 (females) negative e-mail with explanation why
someone cannot meet with reqeustor
16. 16
Conlusions
GOOD NEWS :
No gender bias in responding to or fullfilling requests
Result was strong and robust in both studies (and in many
different fields)
BUT…
Attractivness can play a role – but only in the case of female
students
17. Thank you for your attention!
Author: Magdalena Smyk, Michał Krawczyk
e-mail: msmyk@wne.uw.edu.pl
More about our research on
http://grape.uw.edu.pl
Twitter: @GrapeUW