Description
Learning from Bursztyn, Fujiwara, and Pallais (2017), analyze the data obtained from the primary experiment. Questions are listed on [PS1.pdf]. Please use STATA to answer each question.Specification of the report: Page limit (excluding Stata codes, but including tables and figures) is 4. Font size is 12, and line spacing is 1.5.Table I and II in q1-3 are in the [Table_Template.xlsx]. You also need to submit the Stata code.
Unformatted Attachment Preview
American Economic Review 2017, 107(11): 3288–3319
https://doi.org/10.1257/aer.20170029
‘Acting Wife’: Marriage Market Incentives
and Labor Market Investments†
By Leonardo Bursztyn, Thomas Fujiwara, and Amanda Pallais*
Do single women avoid career-enhancing actions because these
actions signal undesirable traits, like ambition, to the marriage market? While married and unmarried female MBA students perform
similarly when their performance is unobserved by classmates (on
exams and problem sets), unmarried women have lower participation grades. In a field experiment, single female students reported
lower desired salaries and willingness to travel and work long hours
on a real-stakes placement questionnaire when they expected their
classmates to see their preferences. Other groups’ responses were
unaffected by peer observability. A second experiment indicates the
effects are driven by observability by single male peers. (JEL C93,
D82, J12, J16, J31)
Even in the twenty-first century, men prefer female partners who are less professionally ambitious than they are (Fisman et al. 2006). Men tend to avoid female
partners with characteristics usually associated with professional ambition, such
as high levels of education (Brown and Lewis 2004; Greitemeyer 2007; Hitsch,
Hortaçsu, and Ariely 2010). It is relatively unlikely that a woman will earn more
than her husband, and when she does, marital satisfaction is lower and divorce is
more likely (Bertrand, Kamenica, and Pan 2015). Promotions increase the chance of
divorce for women, but not for men (Folke and Rickne 2016).
Single women may thus face a trade-off: actions that lead to professional success might be sanctioned in the marriage market because they signal ambition and
* Bursztyn: Department of Economics, University of Chicago, 1126 E. 59th Street, Chicago, IL 60637, and
NBER (email: [email protected]); Fujiwara: Department of Economics, Princeton University, 131 Julis
Romo Rabinowitz Building, Princeton, NJ 08544, CIFAR, and NBER (email: [email protected]); Pallais:
Department of Economics, Harvard University, Littauer Center, Cambridge, MA 02138, and NBER (email:
[email protected]). This paper was accepted to the AER under the guidance of Stefano DellaVigna, Coeditor.
We would like to thank Daron Acemoglu, Nava Ashraf, David Autor, Marianne Bertrand, Stéphane Bonhomme,
David Deming, Esther Duflo, Patrick Francois, John Friedman, Georgy Egorov, Paola Giuliano, Lawrence Katz,
Mario Macis, Alexandre Mas, Emily Oster, Ricardo Perez-Truglia, Gautam Rao, Jesse Shapiro, Francesco Trebbi,
four anonymous referees, and seminar participants at Berkeley, CIFAR-IOG, LSE, MIT, NBER Culture and
Institutions Meeting, Princeton, UCLA, and Zurich for helpful comments and suggestions. Maxim Ananyev, George
Cheimonitis, Stephanie Cheng, Mikhail Galashin, Vasily Korovkin, Juan Matamala, Imil Nurutdinov, Sebastian
Ottinger, Benjamin Smith, Maria Lucia Yanguas, and especially Jenna Anders provided excellent research assistance. Financial support from NSF CAREER grant no. 1454476 (Pallais) is gratefully acknowledged. This project
received IRB approval from Harvard, Princeton, and UCLA. The experiments and survey reported in this study
can be found in the AEA RCT Registry (0001456, 0001686, and 0001774). The authors declare that they have no
relevant or material financial interests that relate to the research described in this paper.
†
Go to https://doi.org/10.1257/aer.20170029 to visit the article page for additional materials and author
disclosure statement(s).
3288
VOL. 107 NO. 11
Bursztyn et al.: Acting Wife
3289
assertiveness.1 For example, while volunteering for leadership roles or asking for a
promotion might help women’s careers, they may also send negative signals to the
marriage market. This trade-off can be pervasive and is not limited to large, discrete decisions. Daily activities such as speaking up in meetings, taking charge of a
project, working late, or even certain outfits, haircuts, and makeup can be desirable
in one market and not in the other. Hiding career-enhancing actions from potential
partners may be challenging for single women: it is likely difficult to hide working
late or traveling for work, for example. Moreover, the workplace is the most common
place to meet a partner (Rosenfeld, Thomas, and Falcon 2015). Similar to minority
students who shy away from educational investments to avoid “acting white” and
improve their standing with peers (Austen-Smith and Fryer 2005; Fryer and Torelli
2010), single women might try to improve their marriage options by “acting wife.”
On the other hand, for men, the consequences of actions in the labor and marriage
markets are more closely aligned: women value their partner’s intelligence and education, even when these exceed their own (Fisman et al. 2006; Lee 2016).
In this paper, we test for the existence and the implications of this trade-off by
studying students in an elite US MBA (Master of Business Administration) program. Graduate school is a natural place to study this trade-off. Many students are
both investing in their professional career and looking for a long-term partner.2 For
example, a 2015 survey of Harvard Business School (HBS) MBA alumni indicates
that among the youngest surveyed cohort (“Generation Y,” aged 25–30), 31 percent
of married women and 16 percent of married men are married to an HBS alum.3
Many actions in graduate school are observable to peers—and thus may influence
marriage outcomes—and are potentially important for the labor market. These
include joining professional clubs and organizations, on-campus interviews, trips
to career fairs, and participation in case competitions. There are substantial gender
differences in career outcomes for elite MBA program graduates (Bertrand, Goldin,
and Katz 2010). Male preferences for less ambitious women have been documented
in similar contexts: Fisman et al. (2006) studies Columbia University graduate students (including MBA students), for example.
We start by providing observational evidence that single women avoid activities
that could help their careers to avoid signaling traits that may be penalized in the
marriage market. We conducted a survey asking first-year MBA students whether in
the two years before business school they had avoided certain actions they thought
would help their careers because they were concerned it would make them “look too
ambitious, assertive, or pushy.” Sixty-four percent of single females said they had
avoided asking for a raise or a promotion for that reason, relative to only 39 percent
1
While in the paper we mostly refer to the “marriage market,” this trade-off may also apply to a more general
“dating market” where women also seek romantic relationships not expected to lead to marriage.
2
For example, a New York Times article describes how a female Harvard MBA student dealt with such tradeoff: “Judging from comments from male friends about other women (‘She’s kind of hot, but she’s so assertive’),
Ms. Navab feared that seeming too ambitious could hurt what she half-jokingly called her social cap, referring to
capitalization,” and wondered about her goals: “Were her priorities purely professional, were they academic, were
they to start dating someone?” It also describes how after she “started dating … [she] felt freer to focus on her career
once she was paired off.” (Jodi Kantor, “Harvard Business School Case Study: Gender Equity,” New York Times,
September 7, 2013)
3
The rates of marriage are also high for other cohorts. In the “Generation X” cohort (aged 31–47), 23 percent of
married women and 9 percent of married men are married to an HBS alum. The respective numbers for the “Baby
Boomer” generation (aged 48–66), are 23 percent and 5 percent (Ely 2015).
3290
THE AMERICAN ECONOMIC REVIEW
November 2017
of women who were married or in a serious relationship and 27 percent of men.
Over one-half (52 percent) of single women reported avoiding speaking up in meetings, relative to 33 percent of non-single women, and 28 percent of men. Overall,
almost three-quarters (73 percent) of single women said they had avoided actions
they believed would help their career because they were worried about looking too
ambitious.
Next, we document that unmarried female MBA students have lower class participation grades than married ones. Class participation is observable to peers and may
signal students’ ambition or assertiveness.4 It is not the case that unmarried women,
in general, perform worse in class than married women: both groups perform similarly on their midterm exam, final exam, and problem sets. The difference—approximately one-third of a standard deviation—is only present for the observable part of
the grade. It is not driven by observable differences (e.g., age or citizenship status)
between the two groups. For men, who do not face the same trade-off, the differences in participation grades by marital status are negligible. Note that lower participation grades are consequential: they are a component of final grades, which this
school discloses to potential employers.
Our main results come from two field experiments that directly test whether single
women respond to the studied trade-off by explicitly changing their behavior, making themselves look less professionally appealing. In the first (primary) experiment,
we randomize whether actions with positive labor market consequences (potential
signals of ambition) are expected to be observed by one’s classmates. Classmate
observability allows the actions to have larger marriage market consequences.
On the first day of the MBA program, during a career advising session for the
newly admitted class, a career counselor asked students to complete a questionnaire
about their job preferences. The questionnaire asked about students’ preferences over
fields and geography and included a number of questions for which we expect that
the answers that would make respondents more appealing job candidates would also
make women less attractive spouses. It asked about students’ desired compensation,
hours of work, and days per month of travel. It also asked students to rate their leadership abilities and professional ambition. These questions are commonly used in
recruiting MBA students and shape what positions are considered a good fit for a
candidate. From the students’ standpoint, this questionnaire had substantial stakes.
This was the first information the career center collected on students’ preferences
and they were told the information would be used to place them into summer internships.5 The career office informed us that students’ reported preferences would have
impacted their placement: for example, if a student was not willing to travel at least
four days per week, she would be steered away from consulting, while if she reported
not wanting to work long hours, she should not be placed in investment banking.6
4
A New York Times article about HBS reports that “after years of observation, administrators and professors
agreed that one particular factor was torpedoing female class participation grades: women, especially single women,
often felt they had to choose between academic and social success” (Jodi Kantor, “Harvard Business School Case
Study: Gender Equity,” New York Times, September 7, 2013).
5
The summer internship is a key step for job placement: 38 percent of the last graduating cohort accepted a
post-graduation job at the firm of their summer internship.
6
In practice, after learning about the large effects of our treatment for single women, the career center decided
not to use the questionnaires in placement. However, when completing the questionnaires, students did not know
this was an experiment or have any reason to believe the questionnaires would not be used.
VOL. 107 NO. 11
Bursztyn et al.: Acting Wife
3291
There were two slightly different versions of the instructions; which version a
student received was randomized. Both stated that a career counselor would see
their answers. In the public version, students were also told that “your” answers will
be discussed in the career class, while in the private version, students were told that
“anonymized” answers would be discussed. The two versions of the questionnaire
looked essentially identical—they differed only by one word ( “your” versus “anonymized” )—so students did not know there were two versions of the questionnaire.
When students thought their answers would only be viewed by a career counselor,
single and non-single women answered similarly.7 However, when single women
expected their classmates to see their answers, they portrayed themselves much
less favorably to the labor market. They reported desired compensation $18,000 per
year lower. They said they would be willing to travel seven fewer days per month
and work four fewer hours per week.8 They also reported significantly lower levels
of professional ambition and tendency for leadership. Neither non-single women
nor men, regardless of their relationship status, changed their answers when they
expected their peers to observe their choices.
To assess an alternative interpretation that single women are simply more humble
in public, we included a placebo question on self-reported writing ability. Writing
skills are valued in the labor market, but not sanctioned in the marriage market. Thus,
while we would predict that single women would report similar writing abilities in
both treatments, a humility explanation would not. Consistent with our hypothesis,
single women (and all other groups) rated their writing skills equally in the public
and private treatments.
Perhaps surprisingly, there are very small gender differences in the private treatment. Women report lower desired compensation than men, but similar willingness to travel and work long hours, professional ambition, and leadership abilities.
Gender gaps emerge once students expect their preferences to be shared with their
peers. In other words, the bulk of the gender gap in responses is driven by (single women’s response to) expected observability and not by differences in private,
which are perhaps more likely to reflect “true” preferences.
The primary experiment results indicate that single women, but not women in a
relationship, avoid actions that could help their careers when these actions have negative marriage market consequences. A supplementary experiment shows that single
women present themselves less favorably to the labor market—and more favorably
to the marriage market—when they believe their choices will be seen by men as
opposed to women. During a career class, students were asked to make choices
over three pairs of hypothetical jobs. Students made these choices individually, but
were told that if there was time at the end of class, the instructor would discuss their
answers with the small groups in which they were completing other class activities. These groups change from day-to-day and, on this day, they were randomized
such that some single women were in all-female groups while the remainder had
all male group-mates. This was a natural activity during a session discussing job fit
7
The questionnaire asked a more detailed question about relationship status than the one available in the grades
data. We classify as non-single those who were “in a serious relationship,” “cohabiting,” “engaged,” or “married.”
8
In the private treatment, single female students reported desired compensation of $131,000, and were willing
to travel 14 days a month and work 52 hours a week, on average.
3292
THE AMERICAN ECONOMIC REVIEW
November 2017
and students did not know this was an experiment. They were told that these forms
would be collected at the end of class, so they knew the career center would see
them.
When placed in all-female groups, 68 percent of single women reported that they
would prefer a job with a higher salary that required 55–60 hours of work per week
over a job with a lower salary requiring only 45–50 hours per week. But, when
placed with male peers, women were 26 percentage points less likely to make this
choice. Similarly, in all-female groups, 79 percent of single women reported preferring a job with quicker promotion to partner but substantial travel over a job
with slower and less certain promotion but no travel. When placed with male peers,
women were 42 percentage points less likely to choose this option. Single women’s
answers to a placebo choice between a job with a positive social impact and a job
with collegial coworkers were not affected by their group-mates’ gender. We also
exploit the random variation in the share of married men within the groups. Single
women were less likely to choose the career-focused option when there were more
single (as opposed to married) men in the group, consistent with marriage market
signaling.9
We assess whether another difference between single and non-single women can
explain why single women, but not women in relationships, downplay their ambition in public. While we cannot rule out that possibility, our results are not easily
explained by such a difference. In the primary experiment, differences in single and
non-single women’s responsiveness to the public treatment are robust to allowing
students’ responsiveness to vary with covariates. Both groups report similar preferences and self-assessments in the private treatment. They also answer similarly (in
public and private) when answers are unlikely to be sanctioned in the marriage market. Similarly, married and unmarried women have similar performance on exams
and problem sets: differences only arise for the participation grades, and remain
unchanged after controlling for covariates. Finally, the supplementary experiment
shows that single women’s decision to portray themselves as less ambitious in public is driven by the presence of male peers, and especially single male peers.
We discuss a simple model of dating market signaling (presented in online
Appendix A) that helps interpret the experimental results. The model predicts that
when their decisions are publicly observed, single women downplay their ambition
and incur a labor market cost to increase their probability of attracting a desirable
partner. When their decisions are not observed, they make the choice that maximizes their labor market outcomes. All other groups (non-single women and all
men) make the decision that maximizes their labor market outcomes regardless of
whether it is publicly observed.
Our results suggest that single women avoid actions that would help their careers
because of marriage market concerns. Many schooling and initial career decisions—
9
To keep the discussion of the results concise, we implicitly abstract from the possibility that some respondents
might be interested in same-sex partners. No data on students’ sexual orientation are collected. Ninety-seven percent of the US population labels itself as heterosexual (2014 National Health Interview Survey). The interpretation
of our observational and primary experiment results is unchanged if one assumes that homosexual and bisexual
women have similar partner preferences as heterosexual men, although whether this is true is an open question.
The interpretation of the supplementary experiment is based on the assumption that single women in our sample
are interested in male partners. The presence of a substantial share of women interested in same-sex partners would
attenuate the results.
VOL. 107 NO. 11
Bursztyn et al.: Acting Wife
3293
such as whether to take advanced math in high school, major in engineering, or
become an entrepreneur—occur early in life when most women are single. These
decisions can have labor market consequences that last long after these women get
married. While extrapolating to other settings is beyond the scope of this paper, elite
female MBA students comprise a selected group that presumably places a higher
value on career success than the general female population. This suggests the effects
of marriage market signaling are perhaps even larger in other contexts.
These findings point to marriage market signaling as an additional explanation
for gender differences in the labor market. Marriage market signaling is related to
explanations surrounding norms over gender identity and the propensity to negotiate.
However, these existing explanations have difficulty explaining our results. Gender
differences stemming from these explanations would have likely appeared when
comparing answers by male and female students in private and would have likely
not been restricted to single women, particularly given that single and non-single
women behave similarly in private. Our results also add to the literature on how
individuals’ economic decisions are affected by social image concerns.10
The remainder of this paper proceeds as follows. In the next section, we present
descriptive statistics on our sample and present results from the observational data:
the survey about pre-MBA career choices and classroom grades. Section II lays out
the design of the primary and supplementary experiments, presents the experimental
results, and discusses alternative interpretations. Section III concludes.
I. Observational Evidence
A. Descriptive Statistics
We have four datasets on students in an elite MBA program. The first is administrative data on students’ grades in their (required) introductory economics class for
the 2010–2016 entering cohorts. The other three datasets—the survey, the primary
experiment, and the supplementary experiment—were collected on the 2016 entering cohort. Except for the survey, which was anonymous, we link all of the datasets
to admissions records, which has information on student characteristics.
The first column of Table 1 provides descriptive statistics on the 2,235 students
who entered the program between 2010 and 2016. Almost 70 percent of students
are male. They average 28 years old, with just over 5 years of work experience.
Two-thirds are US citizens, with most of the remainder coming from Asia. The
average GMAT score is above the ninetieth percentile of the national distribution,
consistent with admission to an elite business school. The fourth column provides
statistics on the primary experiment sample. Because the experiment was conducted
10
See, for example, Akerlof and Kranton (2000), Fernández, Fogli, and Olivetti (2004), Niederle and Vesterlund
(2007), Eckel and Grossman (2008), Fernández and Fogli (2009), Dohmen et al. (2011), Alesina, Giuliano, and
Nunn (2013), Baldiga (2014), Coffman (2014), Bertrand, Kamenica, and Pan (2015), and Bordalo et al. (2016) on
gender norms and identity; and Flory, Leibbrandt, and List (2014), Leibbrandt and List (2015), and Exley, Niederle,
and Vesterlund (2016) on the propensity to negotiate. A large literature, surveyed by Jayachandran (2015), studies
the role of social norms in explaining gender inequality in developing countries. Austen-Smith and Fryer (2005),
Fryer and Torelli (2010), and Bursztyn and Jensen (2015) study social image and educational choices; DellaVigna
et al. (2017) study voting; Bénabou and Tirole (2006) study prosocial behavior in general; and Charles, Hurst, and
Roussanov (2009) study consumption. Bursztyn and Jensen (2017) review the topic.
3294
November 2017
THE AMERICAN ECONOMIC REVIEW
Table 1—Descriptive Statistics
Male (percent)
Age
Has children (percent)
Years of work experience
GMAT score
2010–2016 cohorts
2016 cohort
Grades data
Primary experiment
Survey
Overall
Married
women
Unmarried
women
Overall
Non-single
women
Single
women
Overall
68.3
28.0
4.7
5.1
709
28.9
15.9
5.8
707
27.1
0.2
4.6
703
67.9
27.9
5.7
5.2
715
27.5
0.0
5.0
701
27.3
1.7
4.9
707
46.2
22.0
3.7
5.9
20.0
2.3
0.0
51.9
9.6
15.4
23.1
0.0
100.0
0.0
0.0
0.0
0.0
0.0
Marital/relationship status (self-reported, percent)
Single
In a serious relationship
Cohabiting
Engaged
Married
No response
67.3
28.1
53.3
18.0
2.9
4.8
19.9
1.1
Marital/relationship status (admissions data, percent)
Single
77.0
0.0
100.0
Married or in a domestic
18.1
100.0
0.0
partnership
No response
4.9
0.0
0.0
Citizenship (percent)
United States
North America (without US)
Asia
Europe
South America
Africa
Oceania
66.0
4.1
21.7
3.8
3.7
0.5
0.2
51.2
7.3
34.1
2.4
4.9
0.0
0.0
68.2
2.4
23.7
2.9
1.9
0.5
0.3
62.5
3.9
24.5
2.0
6.2
0.8
0.0
71.2
5.8
19.2
1.9
1.9
0.0
0.0
58.3
3.3
35.0
1.7
1.7
0.0
0.0
Observations
2,235
82
582
355
52
60
272
Notes: The table shows descriptive statistics from the grades, primary experiment, and survey samples. The grades
and primary experiment data are linked to admissions records. The survey was anonymous and cannot be linked
to admissions records. Non-single refers to individuals who report being in a serious relationship, cohabiting,
engaged, or married.
on the first day of the program at the career center introductory session, 98 percent
of the class participated. The 2016 cohort looks similar to the earlier cohorts.
Admissions records for the 2010 to 2016 cohorts contain information on whether
students were married or in a domestic partnership at the time they applied to the
program: only 18 percent of students were. For the 2016 cohort, we collected more
detailed information on students’ relationship status during the primary experiment.
While only 20 percent of students are married, less than one-half (46 percent) call
themselves single. Twenty-two percent of students are in a serious relationship at the
start of the MBA program, while just under 10 percent are cohabiting or engaged.
These fractions are similar in the survey data (column 7). The survey was voluntary
and conducted during the economics class during the middle of the semester and
76 percent of the class participated.11 In the grades analysis, we can only compare
11
We do not know whether nonparticipation is due to class absence or active nonresponse. Based on usual
absence rates, we estimate a 90 percent response rate among present students. Attendance is not required or measured in this course.
VOL. 107 NO. 11
Bursztyn et al.: Acting Wife
3295
married and unmarried students, while in the experimental and survey data, we compare single students and students in serious relationships, including engagements,
cohabitations, and marriages. To distinguish this from the comparison of married
and unmarried students, we refer to these students as “single” and “non-single.”
The second and third columns of Table 1 show descriptive statistics separately
for married and unmarried women. Married women are about two years older on
average than unmarried women and have more work experience. They are less likely
to be US citizens and are more likely to be from Asia or South America. However,
their GMAT scores are similar to those of unmarried women. Married men are also
older (2.5 years on average), less likely to be US citizens, and more likely to be from
Asia or South America than unmarried men (online Appendix Table 1). Married and
unmarried men also have similar GMAT scores.
Table 1 shows that single and non-single women look much more similar than
married and unmarried women. Non-single women are neither significantly older nor
do they have significantly more work experience. Single women are less likely to be
US citizens, although the difference is not statistically significant (online Appendix
Table 1 shows the significance of these differences.) Again, there is little difference in
the GMAT scores between single and non-single women. In the supplementary experiment, we only analyze data from women who reported being single in the primary
experiment. We discuss the descriptive statistics of this sample in Section IID.
B. Survey on Past Behaviors
In October 2016, the first-year MBA cohort was asked to answer a short anonymous survey on its prior work experience.12 The survey, presented in online
Appendix Figure 1, was conducted during a required class (economics). It was
intended as motivational evidence to assess (i) how often single women avoid
actions beneficial to their careers to avoid appearing too ambitious or assertive and
(ii) whether single women avoid these actions more than other groups, in particular
non-single women. Specifically, it asked:
In the last two years, are there behaviors or activities at your work that
could have helped you professionally that you didn’t undertake because
you might have looked too ambitious, assertive, or pushy?
We asked students who responded affirmatively to mark any of four behaviors
they did not undertake for that reason: (i) speaking up at meetings, (ii) offering to
make a presentation or sales pitch, (iii) asking for a leadership role in a team or task
force, and (iv) taking initiative in negotiating a raise or asking for a promotion. We
also left space for students to write in other activities that they avoided, but no one
did. Almost all (98 percent) of the students reporting avoiding some activity marked
one of the four listed. We also asked students’ age, gender, and relationship status.
While this survey was not intended to provide causal evidence that single women
adjust their behavior because of marriage market concerns, the results displayed in
Table 2 are striking. Relative to the other groups and across all options p rovided,
12
Admissions data show that 96 percent of students in the 2016 cohort had at least two years of prior work
experience.
3296
November 2017
THE AMERICAN ECONOMIC REVIEW
Table 2—Avoidance of Workplace Behaviors by Gender and Relationship Status (Survey Data)
Taking
Asking for
initiative in a leadership
Offering
negotiating
role in a
to make a Speaking
a wage raise team or task presentation
up at
Any
or promotion
force
or sales pitch meetings behavior Observations
Single women
Non-single women
Single men
Non-single men
63.5
39.4
25.3
30.3
40.4
24.2
23.0
23.6
25.0
15.2
18.4
6.7
51.9
33.3
27.6
29.2
73.1
60.6
43.7
50.6
52
33
87
89
p-values of differences
Single versus non-single women
Single women versus others
0.030
0.000
0.129
0.014
0.284
0.031
0.095
0.002
0.234
0.002
85
261
Notes: Data are from a survey administered to first-year MBA students in the fall of 2016. Each number in the first
four rows of data is the percentage of the group indicated by the row that avoided the action indicated by the column in their previous two years of work, despite the fact that they believed it could help their careers because they
were concerned about appearing too ambitious, assertive, or pushy. Non-single refers to respondents in a serious
relationship, cohabiting, engaged, or married.
single women were more likely to report having avoided these workplace behaviors, by amounts that are both economically and statistically significant. For example, 64 percent of single women did not take initiative in asking for a raise or a
promotion because they were worried about looking too ambitious, relative to
39 percent of non-single women (the p-value of the difference is 0.030), 30 percent of non-single men, and 25 percent of single men (the p-value of the difference between single women and all other groups is below 0.001). Fifty-two percent
of single women avoided speaking up at meetings for the same reason, relative to
33 percent of non-single women (the p-value of the difference is 0.095), 29 percent of non-single men, and 28 percent of single men (the p-value of the difference
between single women and all other groups is 0.002). Moreover, 40 percent of single women avoided asking for a leadership role, and one-quarter refrained from
offering to make a presentation or a sales pitch, despite the fact that they thought
these activities could help them in their careers. Almost three-quarters of single
women (73 percent) reported avoiding activities that they thought would help them
professionally because they were concerned about how the activities would make
them look. Adding up the number of these actions that each group avoided in the
past generates a similar picture: the average number of avoided actions was 1.81 for
single women, 1.12 for non-single women ( p-value of the difference is 0.013), 0.94
for single men, and 0.90 for non-single men (the p-value of the difference between
single women and all other groups is below 0.001). Overall, women’s relationship
status is predictive of avoiding these behaviors. Across all four actions, non-single
women look more similar to men than to single women; the largest differences are
between single and non-single women.13
13
As described in the registration of the survey (AEARCTR-0001686), the differences we observe between single and non-single women may actually underestimate the behavior differences between these two groups. Many of
the cu