You are on page 1of 41

Managers Political Ideology and Gender Inequality within Organizations*

Seth Carnahan
Ross School of Business
University of Michigan
scarnaha@umich.edu
Brad Greenwood
Fox School of Business
Temple University
brad.n.greenwood@gmail.com
Abstract
Although prior research provides significant evidence that political ideology (i.e. liberal or
conservative) can influence an individuals views about gender roles in economic life, little is known
about whether these views influence organizations. In this work, we theorize as to why conservative
(liberal) managers may have higher (lower) rates of gender inequality in the training, promotion, and
turnover of their subordinates, relative to moderate managers. We further theorize that conservative
(liberal) managers will be less (more) likely to lead their firms diversity initiatives, relative to
moderate managers. Using novel microdata from the legal services industry, we find support for the
majority of our arguments. We discuss the implications of our theory and findings for individual
careers and firm performance.

*
For detailed comments we would like to thank Elizabeth Campbell, Jerry Davis, Abhinav Gupta, Jim Westphal,
and audience members at the University of Michigan and the Wharton People and Organizations Conference. We
express gratitude to Adam Bonica for sharing data. Financial support from the Strategy Research Foundation,
Temple Universitys Young Scholars Forum, and the University of Michigan is gratefully acknowledged. Errors
remain our own.

Electronic copy available at: http://ssrn.com/abstract=2721979

Introduction
Scholars who study the organizational roots of gender inequality increasingly focus on the
organizations managers (e.g. Briscoe & Kellogg, 2011; Castilla, 2011; Huffman, Cohen, and
Pearlman 2010), because these individuals, rather than an abstract organization, make personnel
decisions (see Huffman, 2013 for a review). This literature suggests that female managers might
decrease gender inequality among their subordinates (e.g. Cohen & Huffman, 2007; Hultin &
Szulkin, 1999; Cohen, Broschak & Haveman, 1998; McGinn & Milkman, 2013), perhaps by taking
direct action on behalf of their female employees (Abraham, 2015) or by changing the attitudes
towards women within the organization (Ely, 1995) 1.
Women, however, are often underrepresented in leadership positions, notably in powerful
organizations like large professional services firms or large corporations (Catalyst, 2014). As a
result, a paradox lies at the center of the literature examining how managers influence gender
inequality. While female managers may reduce gender inequality among subordinates, it is unclear
how women obtain such leadership positions in the first place (Cohen, Broschak & Haveman, 1998:
723), which, in turn, may limit the ability of a manager-focused theory to explain gender inequality.
We argue that the managerial perspective can make progress on this predicament by relaxing the
assumption, implicitly held in most prior work (see Dahl, Dezs & Gaddis Ross, 2014 for an
exception), that male managers do not vary in their attention to gender inequality in their
organizations.
In examining variation in male managers attention to gender inequality, we focus on their
political ideology. Researchers have long argued that a managers political ideology, situated on a
liberal-conservative continuum and defined as a set of beliefs about the proper order of society and
how it can be achieved (Erikson and Tedin 2003: 64), can influence organizational outcomes (e.g.
1

See Duguid 2011; Duguid, Loyd, and Tolbert 2012; Srivistava & Sherman, 2014 for a contrary perspective; scholars also
examine manager race and racial inequality (e.g. Elliott & Smith, 2004).

Electronic copy available at: http://ssrn.com/abstract=2721979

Tetlock, 2000), such as investments in corporate social responsibility initiatives (Chin, Hambrick &
Trevino, 2013), targeting by LGBT activists (Briscoe, Chin & Hambrick, 2014), and allocation of
resources among business units (Gupta, Briscoe & Hambrick, 2014). However, researchers have not
yet examined whether a managers political ideology influences gender-related outcomes within their
organizations. This gap is striking when one considers the deep body of political science research
which suggests that conservative individuals tend to favor a society with more traditional gender
roles, while liberal individuals tend to favor a society with more progressive gender roles
(Bolzendahl & Myers, 2004; Davis & Greenstein, 2009). If these preferences influence managerial
decision-making, a managers political ideology may drive considered choices or unconscious biases
that have an important influence on gender inequality within his organization. Anecdotally, recent
high profile managerial actions suggest that these preferences extend into organizational life. 2
Combining organizational perspectives on gender inequality with work in political science,
we argue that conservative and liberal managers will likely differ in the extent to which they: 1)
believe that ameliorating gender inequality is an organizational imperative (e.g. Hambrick & Mason,
1984); 2) expect that female subordinates will be burdened by traditional family responsibilities (e.g.
Eagly, 2013; Phelps, 1972); and 3) believe that female subordinates possess leadership characteristics
(e.g. Eagly & Karau, 2002; Schein, 1973). Relying on these mechanisms, we hypothesize that
conservative male managers will be less likely to serve on their firms diversity committees, will be
less likely to train and develop female subordinates, and that their subordinates will have larger
gender differences in promotion and turnover. In contrast, we predict that liberal male managers will
be more likely to serve on diversity committees and to have lower rates of gender inequality in the
training, promotion, and turnover of their subordinates.
2

For example, Salesforce.com CEO Marc Benioff, well known for his support for liberal politicians, announced an
organizational policy to review pay of all 16,000 employees, stating that My job is to make sure that women are treated 100
percent equally at Salesforce in pay, opportunity and advancement . . . when Im done, there will be no [gender pay] gap (Peck,
2015). Netflix, which recently announced new parental leave policies, provides another prominent example (Steel, 2015).

We test these ideas in large corporate law firms in the United States, a rich, socially
important context where gender inequality is of significant interest to both practitioners (e.g.
American Bar Association, 2013) and researchers (e.g. Beckman & Phillips, 2005; Ely, 1994, 1995;
Phillips, 2005; Gorman, 2005). Combining information from several sources, and casting partner
attorneys as managers and associate attorneys as subordinates, we create a novel dataset that allows
us to capture four key measures: 1) political ideology using individual-level political donations; 2)
partners participation in their firms diversity committees; 3) associate -partner work relationships
(both at the practice area and client-team level); and 4) the promotion and turnover of associates.
Our findings suggest that conservative (liberal) male partners are less (more) likely to be involved in
diversity committees and that they have significantly higher (lower) rates of gender inequality in
terms of associate training, promotion, and turnover, as compared to politically moderate partners.
Our study provides, to the best of our knowledge, the first comprehensive examination of
political ideology and gender inequality in organizations. We extend current theory, which links
managers to organizational inequality, by explaining why male managers with differing political
ideologies may vary widely in their preferences and actions related to gender inequality. This
contribution has clear implications for research examining individual careers and organizational
performance, and we explore these implications in the discussion section.
Managers, Political Ideology, and Gender Differences within Organizations
A growing body of work suggests that managers ascriptive characteristics (e.g. gender, race)
may significantly influence ascriptive inequality among subordinates (e.g. hiring, promotion,
compensation). Indeed, much of this work focuses on manager and subordinate gender, building
primarily on social identity theory (Tajfel, 1982) to explain why managers might be more likely to
support subordinates who share their ascriptive characteristics. For example, both Castilla (2011)
and Tsui & OReilley (1989) find that managers provide better performance ratings to employees
3

who share their gender, and many studies find smaller gender differences in hiring, promotion, and
compensation within organizations that have more female managers (e.g. Abraham, 2015; Bilimoria,
2006; Cohen, Haveman & Broschak, 1998; Ely, 1995; Kulis, 1997; Cohen & Broschak, 2013; Cohen
and Huffman, 2007; Gorman, 2005; Hultin and Szulkin, 1999; Jacobs, 1992; Kurtulus and
Tomaskovic-Devey, 2012; Matsa & Miller, 2011; Shenhav and Haberfeld, 1992) 3.
This work provides compelling evidence that managers play an important role in determining
gender inequality, but it focuses almost exclusively on the managers gender. Cohen, Broschak and
Haveman (1998: 723) note an important limitation of this approach: [a]lthough having women
present in managerial positions is crucial to bringing more women into management, it remains
unclear how women initially attain managerial positions. We address this gap by focusing on male
managers and exploring why they may vary in their support for female subordinates4. We do so by
examining managers political ideology, motivated by political science research that links political
ideology to attitudes towards gender roles and by organizational research that demonstrates that
political ideology has an important influence on managerial decision making (Tetlock, 2000;
Christensen et al., 2015; Chin et al., 2013; Briscoe et al., 2014; Gupta et al., 2014).
While multiple dimensions of ideology are possible (Jost, 2006), most political researchers
argue that the majority of individuals fall on a liberal-conservative continuum. A large body of
political research shows marked differences among individuals with different ideologies. Liberals,
for example, tend to favor social change, social justice, equality, and diversity, while conservatives
place importance on stability and respect for authority (Jost, Glaser, Kruglanski, & Sulloway, 2003;

The value-threat approach, which describes why minority managers may resist providing assistance to members of their
ascriptive group, has emerged as a counterpoint to this work (Duguid, 2012; Duguid, Lloyd & Tolbert, 2012).
4 Dahl, Dezso & Gaddis Ross (2014) also argue for this approach when they examine how the birth of a daughter influences male
CEOs compensation of female employees.

Tetlock, 2000). 5 Prior work indicates that these differences often lead conservatives to prefer more
traditional gender roles in the home and workplace, e.g. men working outside the home and women
raising children (Eagly et al., 2000), while liberals may prefer more progressive gender roles
(Bolzendahl & Myers, 2004; Davis & Greenstein, 2009; Mason and Lu, 1988; Klein, 1984). Within
an organization, traditional gender roles typically find men occupying leadership positions more
often than women (e.g. Eagley & Dabreu, 2002).
With regard to the household, conservatives preferences for more traditional roles may stem
from concern about the unintended consequences that working mothers may have for children (e.g.
Bolzendahl & Myers, 2004), while liberal tolerance for more progressive roles may stem from a
higher aversion to inequality (Jost, 2006). Evidence of these beliefs can be found in the nationally
representative General Social Survey (GSS; see Huffman & Cohen, 2007; Cotter et al., 2011), which
asks Americans about their social preferences and their political ideology, among other items. In
responses summarized in Table 1, we see, for example, that male conservatives are much more likely
than male moderates and liberals to agree that it is much better for everyone involved if the man is
the achiever outside the home and the woman takes care of the home and family.
With regard to leadership in the workplace, conservatives may prefer more individualistic,
agentic leaders (Winter, 2010), a characteristic that is more stereotypically associated with men
(Eagly & Johnson, 1990), while liberals may prefer leaders who exhibit more communal
characteristics (Winter, 2010), a quality that is more stereotypically associated with women (Eagly &
Johnson, 1990). In Table 1, we see that conservative men are more likely to believe that men are
emotionally better suited for politics than women, which prior scholars have used to infer
respondents attitudes towards gender differences in leadership characteristics (e.g. Rosenwasser &

Jost (2006: 654) summarizes this research by identifying two relatively stable, core dimensions that seem to capture the most
meaningful and enduring differences between liberal and conservative ideologies: (a) attitudes toward inequality and (b) attitudes
toward social change versus tradition.

Dean, 1989).
While researchers have examined how ideologically-driven beliefs about gender roles might
manifest themselves in the voting booth (e.g. Edlund & Pande, 2002), their effect on organizations
has not been examined. However, Table 1 also provides suggestive evidence that political ideology
may influence organizational decision making around gender inequality: liberal men are much more
likely than conservative men to agree that because of past discrimination, employers should make
special efforts to hire and promote qualified women. In what follows, we build theory that
describes why male managers political ideology will relate to observable markers of their support
for gender equality within their organizations.
Managers Political Ideology and Participation in Organizational Diversity Committees
We first consider the relationship between political ideology and participation on diversity
committees. According to Kalev, Dobbin & Kelly (2006: 593), diversity committees typically are
charged with overseeing diversity initiatives, brainstorming to identify remedies, and monitoring
progress. The prevalence of these committees has grown substantially over time, particularly in
professional services fields. While the teeth and influence of these committees likely varies across
organizations (Ely, 2004; Ferguson, 2015), participation on a diversity committee is an observable
signal of an individuals interest in affecting the organizations diversity climate and/or formal
rules around the hiring and promotion of women (Kim, Kalev & Dobbin, 2012).
We argue that conservative male managers may be less likely to serve on a diversity
committee than moderate male managers. Jost et al. (2006: 310) note that [i]deologies [..] endeavor
to describe the world [..] as it should be, specifying acceptable means of attaining social, economic,
and political ideals. A conservative ideology, therefore, may be less sensitive to gender inequality
and less likely to believe that society (e.g. Jost et al., 2003), and organizations (see Table 1), should
take steps to reduce it. As a consequence, if diversity committee participation is voluntary,
6

conservatives may devote their time elsewhere. Alternatively, if membership on these committees is
determined by appointment by supervisors or by voting by peers, conservative managers may be less
likely to be selected than moderate managers, as conservatives may be less likely to have signaled
their interest in gender equality issues to coworkers and superiors.
Hypothesis 1a: Conservative male managers are less likely than moderate male managers to
serve on their firms diversity committees.
By contrast, diversity committee participation may allow liberal managers to push the
organization (and thus society more broadly) towards what they perceive to be a desirable state, i.e.
one with higher levels of gender equality in the workplace.
Hypothesis 1b: Liberal male managers are more likely than moderate male managers to
serve on their firms diversity committees.
Managers Political Ideology and Gender Inequality in Subordinate Training and Promotion
We next consider the relationship between political ideology and gender inequality in the
training and promotion of subordinates. Managers allocate training and promotion opportunities
based, in part, on their evaluation of a subordinates potential to contribute to the organization.
Ascriptive biases influence this evaluation when managers allow gender or other ascriptive
characteristics to inform their expectations about a subordinates capabilities (e.g. Bielby & Baron,
1986). Crucial to this theory of statistical discrimination is the idea that different managers hold
different beliefs about the ability of ascriptive characteristics, such as gender, to signal future
behavior (Phelps, 1972; Perry et al., 1994). We argue that conservative managers may be more
likely to use gender as a negative signal of female subordinates' family responsibilities (e.g. Kelly &
Dobbin, 1999) and leadership potential (e.g. Eagly & Karau, 2002; Koch, DMello & Sackett, 2015;
Ridgeway, 2001; Schein, 1973, 1975), which may discourage them from training and promoting
female subordinates.
To explain such behaviors, researchers have emphasized the injunctive or prescriptive
7

nature of gender role beliefs, such that an individuals expectations about mens and womens future
behavior is reflected in the individuals preferences about the way that men and women should
behave (Cialdini & Trost, 1998; Eagly, 2013; Eagly & Karau, 2002). As a conservative manager is
more likely to believe that society should be characterized by traditional gender roles, he may be
more likely to believe that men and women will behave in a manner which is consistent with these
traditional roles. For example, because conservative managers may be more likely to believe that
women should be the primary caretakers of children (Davis & Greenstein, 2009), they may be more
likely to believe that female subordinates will miss work or quit the organization to fulfill family
responsibilities. Similarly, because conservative managers are more likely to believe that leadership
positions should be occupied by men (who are more likely to be perceived as agentic[ Eagly &
Johnson, 1990], a characteristic favorable to conservatives [Winter, 2010]), conservative managers
may be more likely to believe that female subordinates will not excel in leadership positions.
As a consequence of these beliefs, a conservative male manager may be more likely to view
female subordinates as less promising, more risky investments than male subordinates. Accordingly,
he may provide female subordinates with fewer training opportunities. Fewer training opportunities
may, in turn, cause female subordinates to be less competitive for promotions, as they may develop
fewer skills and social connections than male competitors (e.g. Ibarra, 1992). Even if male and
female subordinates receive similar training, a dimmer view of female subordinates leadership
qualities may cause conservative managers to support male subordinates in promotion decisions.
Hypothesis 2a: Conservative male managers will have larger gender differences in the
participation in the training and development of male and female subordinates than
moderate male managers.
Hypothesis 3a: Subordinates will have larger gender differences in promotion rates when
their male managers are more conservative.
Liberal male managers, by contrast, are less likely to view traditional gender roles as a social
8

ideal. Liberal male managers may thus be less likely to expect female subordinates to miss work or
leave the organization due to family responsibilities, and may have a more positive evaluation of
female subordinates leadership potential (e.g. Winter, 2010). In addition, liberal managers may be
motivated to provide training and promotion opportunities to female subordinates in order to further
a broader social goal of inequality reduction.
Hypothesis 2b: Liberal male managers will have smaller differences in their participation in
the training and development of male and female subordinates than moderate male
managers.
Hypothesis 3b: Subordinates will have smaller gender differences in promotion rates when
their male managers are more liberal.
Managers Political Ideology and Gender Inequality in Subordinate Turnover
Our previous arguments suggest there will be a larger gender gap in skill development and
advancement among subordinates who work for conservative male managers. It is thus natural to
expect that we will also observe a larger gender gap in turnover among subordinates who work for
conservative male managers, for at least two reasons. First, turnover may be voluntary, because
female subordinates may be more likely to feel impeded and seek employment elsewhere when they
work for conservative male managers. Second, turnover may be involuntary. Decreased access to
training opportunities may prevent female subordinates from developing skills at the same rate as
their male peers, or conservative male managers may be more likely to push female subordinates out
of the organization over concerns about family responsibilities or lack of leadership potential.
Hypothesis 4a: Subordinates will have larger gender differences in turnover rates when
their male managers are more conservative.
The gender gap in turnover may, by contrast, be smaller under liberal male managers. If
liberal male managers provide female subordinates with more access to training and development,
and if female subordinates perceive greater opportunities for upward mobility, they may be less
likely to leave the organization. Liberal male managers may also be less likely to terminate female
9

subordinates, because such a decision may contribute to gender inequality.


Hypothesis 4b: Subordinates will have smaller gender differences in turnover rates when
their male managers are more liberal.
Data
We test these hypotheses in the context of large American law firms, an ideal setting which
has been used extensively in prior work regarding the role of gender in organizations (e.g. Beckman
& Phillips, 2005; Gorman, 2005; Phillips, 2005; Ely, 1995). In addition, gender issues are highly
salient to members of the legal services community (e.g. Williams and Richardson, 2010), thereby
elevating the practical relevance of our results.
The primary data source for our analysis is a nationwide electronic version of the Martindale
Hubbell Law Directory (Martindale) from 1999-2012 (our analyses cover 2007-2012, for reasons
discussed below). Utilized in organization theory (Phillips, 2002, 2005) and economics (Parkin &
Baker, 2005), Martindale has been in print since 1868 with the purpose of providing a listing of
lawyers so that clients and other members of the legal services community can locate talent.
Providing an accurate and up to date listing in Martindale is a strong norm in the legal services
industry, particularly for the large firms that are the basis of our analysis.
The basic unit in the data is the attorney-quarter, which we collapse to the year level, because
the early years of the data do not contain data in every quarter. For each entry, Martindale captures
name, firm affiliation, street address location, miscellaneous personal information (e.g. birth year,
law school name and year of graduation), and practice area (e.g. criminal or corporate law).
Martindale and the American Bar Association furnish attorneys with an International Standard
Lawyer Number (ISLN) which allows us to track attorneys over time and across firm 6. In order to
calculate variables from other sources (e.g. profits, client gender), our sample consists of attorneys
working for the largest 200 law firms by revenue (the Am Law 200). We describe the dependent
6

A small number of attorneys are assigned to multiple firms or multiple offices (defined at the city level) within the same year.
We assign attorneys to unique firms and unique offices in each year using the procedure outlined in Parkin and Baker (2005).

10

variables and samples used to test each hypothesis after describing our independent variables.
Independent Variables
Political ideology. Following Chin et al. (2013), Briscoe et al. (2014), Gupta et al. (2015), and
Christensen et al. (2015), we measure political ideology using attorneys political donation behavior.
We calculate political ideology with two primary variables: Donations to Democrats ($) and
Donations to Republicans ($). These variables allow ideology to vary along left-right dimensions
simultaneously. With some differences among individual politicians, Democrats consistently occupy
the liberal side of American politics while Republicans occupy the conservative side in our sample.
We include donations to candidates, party committees, and Political Action Committees [PACs].
For H1 and H2, we are examining the behavior of individual partner attorneys, so we
measure these variables at the individual partner level. For H3 and H4, our unit of analysis is the
associate-year, and we calculate the political ideology of the associates male managers. We identify
managers as partner attorneys who work in the associates office and share at least one of her practice
areas (e.g. criminal or corporate law). We then calculate Average donations to Democrats by male
partners in practice area ($) and Average donations to Republicans by male partners in practice
area ($) at the associate-year level. In robustness checks we also consider alternate measures of
ideology, where we examine whether the donation pattern of a partner or practice area is
predominantly liberal or predominantly conservative.
An important consideration relates to the timing of political donations relative to the timing
of the tests of our hypotheses. We follow Chin et al (2013) by measuring an attorneys political
donations for ten years, 1996-2006. We then test our hypotheses during the period 2007-2012. This
approach ensures that donations occur prior to the outcomes that we study. We describe our linkage
between donation data and Martindale data in Appendix A.
Chin et al. (2013: p 207) describe why donations are a reasonable marker of a managers
11

underlying political ideology:


[P]olitical scientists have concluded that donations from individuals, those we examine, are
overwhelmingly motivated by personal ideology (Francia et al., 2003, 2005; Ensley, 2009).
Based on a comprehensive review of prior studies, one set of authors concluded that
individual contributions should not be viewed as an investment, but rather as a form of
consumptionor, in the language of politics, participation (Ansolabehere, de Figueiredo,
and Snyder, 2003: 117). Thus we believe our use of political donation data, especially when
examined over an extended period, yields a valid, albeit imprecise indicator of ones political
ideology.
Chin et al. (2013) also take the step of surveying managers and linking their responses to their
political donations (finding a strong correlation between donation behavior and underlying beliefs).
Gender. Martindale does not contain explicit information about gender. To measure gender,
we follow prior literature and use attorney first name (e.g. Phillips, 2005; Gorman, 2005). We first
match the attorneys first names to the dominant gender indicated by US Social Security Data, and
second, to the greater than 95,000 first names present in the Gender Checker Directory 7. Consistent
with prior literature, we exclude unclassified or unisex names (e.g. Pat; <~4%) from the analyses.
Controls
See Appendix B for a listing and description of variables.
Individual level controls. We include numerous controls for attorney experience and quality,
including Years since JD, Tenure with firm, Age, Law school ranking, and Attorney peer ratings
(note that this variable is only available for partners). We control for social ties using % of shared
law school among office partners and control for associate political ideology using Donations to
Democrats by associate ($) and Donations to Republicans by associate ($.) Further, we control for
the attorneys political activity using donation count (any), and we control for the type of law
practiced by the focal attorney using 23 separate Practice area dummies (see Appendix C) because
particular types of law may be systematically more likely to value/devalue gender diversity. Finally,
in the associates practice area, we include controls for size (# male partners in attorneys practice
7

www.genderchecker.com

12

area), gender (% partners in practice area who are female, % associates in prac. area who are
female), and age (average age of male partners in the prac. area).
Office location controls. The regressions that we report contain office or office-year fixed
effects, which absorb differences across offices in local politics and local labor markets. We also
perform robustness tests where we drop highly liberal and highly conservative locales. We also note
whether the attorneys Office is acquired.
Firm level controls. We control firm performance (Profit per equity partner), whether or not
the Firm reports a diversity committee, and whether the Firm acquires another firm. To account for
client-driven motivation to donate to liberal politicians and support gender equality, we control for %
female leadership among client personnel.
Dependent Variables, Samples, and Results
Within the sample of AmLaw 200 firms, each hypothesis addresses different questions and
requires a different set of estimations. To more clearly communicate our results, we describe the
dependent variable, sample, estimation, results, and robustness tests for each hypothesis separately.
Testing H1a-b: Political ideology and participation in diversity committees
Our DV for this analysis is a dummy coded as one if the partner attorney serves on the firms
diversity committee. We obtain data on the membership of law firms diversity committees from
Leadership Directories 8. On a quarterly basis, Leadership Directories compiles information on the
leadership of law firms and other organizations for use by interested parties (e.g. marketing
professionals) who pay significant subscription fees. Data comes from law firm websites and direct
communication with law firm personnel. We link Leadership Directory data to the Martindale
Hubbell data using firm and attorney names. Each firm lists an average of two attorneys on its
diversity committee.
This sample includes partners of Am Law 200 firms whose diversity committee is reported to
8

https://www.leadershipdirectories.com/

13

Leadership Directories. We are able to link all firms contained in the Leadership Directory to the
Martindale data. Of the firms in the AmLaw 200, roughly 60% report at least one member of their
diversity committee in the Leadership Directory.
As the unit of analysis is the attorney-year, and our dependent variable is dichotomous, we
use two different estimators to test our hypotheses. First, we use a series of linear probability models
(LPM) to allow for easily interpretable coefficients. Second, we use conditional logit models to
account for the non-linearity of the DV. We place a fixed effect at the office-year level to account
for variation that is common across partners who work in the same office in the same year.
Table 2 reports the summary statistics for the partner attorneys in this sample, broken down
by gender and political ideology. A comparison across columns indicates liberal male attorneys are
three times as likely conservative male attorneys to serve on diversity committees, providing initial
support for H1. Female attorneys are about 80% more likely than liberal male attorneys to serve on
diversity committees. Consistent with the results of Bonica, Chilton & Sen (2015), we also see a
substantially larger number of liberal than conservative male attorneys in the data. Finally, donation
amounts ($) are skewed, so we take the log of donation values in all analyses that follow.
We estimate regression models in Table 3. Model 1 is an LPM that contains the full sample,
with no control variables. Results suggest that the more a partner has donated to Democrats, the
more likely he or she is to serve on the firms diversity committee (supporting H1b). Further, the
opposite is true for donations to Republicans (supporting H1a). Model 2 adds control variables and
office-year fixed effects. Results are consistent. Importantly, the estimated size of the coefficients
does not change, despite the substantial increase in parameters and R-squared, giving us confidence
that results are not sensitive to the inclusion of particular controls. Model 3 drops female partners
from the sample, and we continue to see similar results.
Race or ethnicity is an important alternative explanation for these patterns. As non-white
14

males are more likely to be liberal, it is intuitive that they may be more likely to serve on diversity
committees. To account for this possibility, we use an attorneys full name to estimate his ethnicity
using data from Origins Info (https://www.originsinfo.eu/; see Belenzon, Chatterji & Daley 2014; for
a recent use of these data). Origins Info assigns an individual to one of several dozen ethnic
categories probabilistically, using his or her first and last name. In Models 3-6, we exclude male
partners whose names flag them as likely to be American Black, African, Middle Eastern, Indian,
East Asian, or Hispanic. We continue to see similar results.
In Model 5, we replace donation amounts with dummies indicating whether an attorney
donates primarily to Democrats or Republicans. These dummies allow for the easiest interpretation
of effect size: we see that liberals are about .56% more likely to serve on a diversity committee.
Because only .6% of individuals in the sample serve on these committees, this translates to an
increase of about 90% from the base rate. Conservatives are about .15% less likely to serve on
diversity committees, or a 25% decrease from the base rate. Model 6 uses conditional logit
estimation, with similar results. Thus, we see strong support for H1a and H1b.
Testing H2: Managers political ideology and gender inequality in subordinate training
For these estimations our dependent variable is a dummy which takes the value of one if the
associate is selected for client team. In the legal services context, law firms often form teams around
specific client legal matters (e.g. Chatain & Meyer-Doyle, 2015), and training and development can
be proxied by whether an associate works on the same client teams as a particular partner. For new
associates, working on client teams is a critical avenue for the development skills and social capital
(Briscoe & Kellogg, 2011), and associates who are not selected for teams often have difficulty
advancing in the firm. Although partners may vary in the extent to which they provide hands-on
training to the associates on their client teams, selecting an associate for a team provides the associate
with a valuable opportunity for development.
15

The context that we examine is team formation around clients merger and acquisition
transactions. Clients often engage law firms to perform a variety of tasks around mergers, such as
completion of due diligence, negotiation of price and terms, analysis of tax implications, and
preparation of regulatory documents (Aspatore, 2010). To obtain client and team formation data, we
use Mergermarket, a private firm that tracks worldwide M&A deals and records the attorneys who
serve on many transactions. Chatain & Meyer-Doyle (2015) use similar data to examine team
formation. Mergermarket obtains deal-level attorney information from press releases, government
documents, and self-reported information from attorneys. 9 We obtain a list of all M&A transactions
covered by Mergermarket from 2004-2012 where at least one of the participants was a US-based firm
and link the names of the attorneys to the Martindale data. Our analysis covers 5,702 deals completed
from 2007-2012 with a median value of $250M and contains 16,860 unique partners and 18,215
unique associates. Each client team in the sample has an average of 2.4 partners and 1.4 associates.
Conditional on completing one deal in the current year, partners complete an average of 1.98 deals
per year, while associates complete an average of 1.63. Additional information, omitted in the
interest of space, about this sample and deals are available from the authors.
Using these data, we examine partners selection of associates for their client teams. We do
so by identifying all possible associates that a partner could have selected for a given client team,
defined as associates who share a partners office location and practice area and have served on an
M&A deal (with any partner) at some point in the last year. Each observation is at the partner-dealassociate level. Roughly 13% of eligible associates are selected for a given deal team 10.

Mergermarket maintains the worldwide league tables for M&A advisors, showing an up-to-the-moment listing of the top
M&A attorneys by deal count and deal value. Headhunters and rival firms track these league tables closely, so attorneys have an
incentive to ensure that their information (and that of their competitors) is reported accurately. See
http://www.mergermarket.com/pdf/deal_criteria.pdf for more information.
10 Interviews with M&A advisors indicate that firms differ in the manner with which they allocate associates to teams. Some
firms allow HR to perform this allocation, with input from partners and associates, while others use a more informal system.
Firms also differ in the extent to which associates have input in the allocation. Our modeling strategy is predicated on the

16

Our analyses consists of linear probability and conditional logit models, each with a partnerdeal fixed effect to account for unobserved heterogeneity. Characteristics of the partner, his firm, and
the deal itself are absorbed by the fixed effect, leaving only variation in associate characteristics.
Table 4 presents tests of H2a-b where the indicator Female associate has been interacted with
the measures of the individual partners political ideology. Model 1 excludes controls, containing
only these interaction terms and the partner-deal fixed effects. Note that the main effect of the
partners political ideology is absorbed by the fixed effect. We see a positive interaction between
donations to Democrats and the probability that a partner selects an associate, supporting H2b. By
contrast, we see a negative interaction between donations to Republicans and the selection of female
associates, supporting H2a. Model 2 (and all remaining models) excludes female partners and shows
consistent results. Model 3 adds control variables, including the a count of the deals on which an
associate worked in the last year (a strong indicator of the associates human capital and desirability
as a team member), age, experience, tenure, personal political donations, and whether the associate
and partner attended the same law school. Coefficient estimates remain stable in the presence of
these controls. Model 4 eschews linear measures of ideology for dummies, indicating whether a
male partner is liberal or conservative. Results show large effect sizes: conservative partners are
2.7% less likely to select a female associate, relative to a moderate partner. This is a 20% decrease
from the base probability of 13% that an associate is selected for a deal. Liberal partners are .8%
more likely than moderates to select a female associate, a 6% increase from the base rate.
Model 7 uses a conditional logit and shows similar results, though the interaction between
donations to Democrats and Female associate becomes marginally significant (p=.11). Model 6
excludes minority male partners from the analysis with similar results. For many deals, the
Mergermarket data contain the names and titles the client personnel involved, including CEO, CFO,
assumption that partners preferences have at least some influence in the allocation process, which seems reasonable given
partners rank and authority.

17

and internal legal positions. Liberal male partners may be more likely to have female clients who
may spur the partner to staff more female associates on their deals (e.g. Beckman & Phillips, 2005).
To account for this possibility, Model 7 excludes deals on which a member of client personnel has a
female first name. The interaction between donations to Democrats and Female associate becomes
marginally significant (p=.15), but the difference between the Liberal * Female and Conservative *
Female interactions remains statistically significant (p<.05). Taken in sum, results suggest strong
support for H2a and intermittent support for H2b.
Testing H3: Managers political ideology and gender inequality in subordinate promotion
The dependent variable for these estimations is a dummy that indicates whether an associate
is promoted to partner in the following year. This is calculated by identifying instances where an
attorneys title changes from associate to partner while remaining with the same firm. We verify
these promotion events by linking Martindale data to the New Partner Promotions database
maintained by American Lawyer. We use the New Partner Promotions information to identify
promotions that occur in 2013, because our Martindale sample ends in 2012.
To construct this sample, we first identify all associates working for Am Law 200 firms in
2006. We then follow these associates over time until they either (a) receive a promotion to partner
or (b) exit the organization. Our level of analysis is the associate-year. Consistent with prior
research examining employee mobility (Campbell et al., 2012) we use a combination of LPMs and
conditional logit models with office and office-year fixed effects. Further, our results are consistent
using a logit hazard specification (Allison, 2012; McGinn & Milkman, 2013). We report the fixed
effect models due to their more conservative nature.
Table 5 presents the summary statistics for this sample, broken out by gender and the
ideology of partners in the associates practice area. Women comprise roughly 45% of the sample,
which is consistent with gender diversity statistics for large law firms, and the raw promotion rate of
18

men is nearly twice that of women. Across practice area, we note that liberal practice areas tend to be
larger, have more female partners/associates, and are less likely to be located in the Southern US 11.
Table 6 tests the hypothesis that political ideology will influence gender inequality in
associate promotion rates. Model 1 excludes control variables and tests the hypothesis using an
LPM. We see a negative interaction between Female associate and the average amount of donations
that male partners in the associates practice area made to Republicans, supporting H3a. We see a
positive interaction between Female associate and the average amount of donations that male
partners in the associates practice area made to Democrats, supporting H3b. Model 2 adds control
variables and office fixed effects, and we see similar results that do not change much in magnitude,
lending confidence to the stability of our estimates. Model 3 uses a conditional logit model with
office fixed effects and shows similar results. Models 4-6 split the sample into liberal, conservative,
and moderate practice areas using the same criteria as Table 5. We see that the Female associate
coefficient is the most negative in the conservative subsample; it is about 80% larger in magnitude
than the liberal subsample, which indicates large practical effect. A Wald test confirms that this
difference is statistically significant (p<.05).
Table 7 provides robustness tests to these promotion analyses. An important alternative
explanation for our results is that female associates who work for conservative partners may be lower
quality, because the female associate might select into working for liberal partners if they could.
Law school affiliation is an imperfect but valuable marker of associate quality. Model 1 of Table 7
adds 50 dummies to our original model, one for each of the 50 law schools with the most graduates
in our data. Results remain consistent. While our office fixed effects help control for geographic
differences, we ensure that our results are not driven by offices located in the more conservative
Southern US or the more liberal New York City legal market by dropping these offices in Models 2
11

The Southern US tends to be more conservative than the rest of the country.

19

and 3. Model 4 employs a highly restrictive office-year fixed effect, and results remain similar.
Model 5 implements a different measure of ideology, calculating the percentage of the practice areas
partners who are liberal or conservative (the denominator is the count of all male partners, not just
those who make donations). We no longer see that more liberal offices have lower gender inequality
in promotion (H3b), but we continue to see a strong relationship between conservatism and gender
inequality (H3a). Importantly, the difference between the interaction terms is statistically significant
(p<.05). Results (not reported) are also similar if we exclude minority male partners. We thus see
strong support for H3a (conservatives), and relatively strong support for H3b (liberals).
Testing H4: Managers political ideology and gender inequality in subordinate turnover
For these estimations the dependent variable is a variable that takes a value of one if the
associate exits the firm in the following year. These exits are identified by instances where the
attorney is no longer affiliated with her current firm in the following year in the Martindale data.
Roughly half of these departures involve moves to other firms in the Martindale data, while the other
half involve exit from the data. Although the Martindale data do not allow us to directly infer
voluntary versus involuntary turnover, we perform additional analyses that examine an attorneys
destination to try and infer the extent of voluntary versus involuntary turnover.
Our sample replicates that used to examine promotion, with one difference. As we must rely
solely on the Martindale data to measure turnover, we are only able to measure turnover until 2012.
We again use a combination of LPMs and conditional logit models.
Table 8 presents our tests of H4. Model 1 is an LPM that only includes the critical
interaction and lower order terms. We see a positive interaction between Female associate and
donations by male partners to Republicans, implying that more conservative practice areas have
higher rates of gender inequality in turnover, supporting H4a. The interaction between Female
associate and average donations to Democrats is not statistically significant (not supporting H4b),
20

but the difference in the interactions terms is statistically significant (p<.05). Model 2 adds control
variables and office fixed effects. Results remain consistent, and the interaction between Female
associate and donations to Democrats becomes significant. Model 3 uses a conditional logit, with
consistent results. Models 4 and 5 splits the turnover outcome into turnover that entails exit from the
Martindale data (arguably more likely to consist of involuntary turnover, as associates in this case are
joining in-house legal departments or changing industries, likely taking a pay cut in the process 12)
and turnover that entails the joining of a different law firm in the Martindale data (arguably more
likely to be voluntary turnover). We do not see differences in the interaction between conservative
ideology and Female associate across these outcomes. However, there is some evidence that
practice areas with more liberal male partners have less gender inequality in departures to non-law
firm destinations. One possible interpretation is that liberal partners have more gender equality in
their firing rates. Overall, however, it is difficult to discern whether our results are driven more by
differences in voluntary or involuntary turnover.
Table 9 performs robustness tests. Similar to the analysis for promotion, we add 50
additional law school dummies to the data (Model 1), we exclude Southern (Model 2) and NYC
offices (Model 3), and we use highly restrictive office-year fixed effects. Results remain consistent.
Model 5 implements the alternative measure of ideology; here we see that H4a-b are not supported.
Finally Model 6 excludes observations where the associates office is acquired in the current year, as
indicated by mobility of >50% of the office to the same firm in the following year. Results remain
consistent. Results (not reported) are also similar if we exclude minority male partners. Overall, we
see support for H4a-b, though it is weaker than the other hypotheses.

12

We performed Google searches on a random sample of 200 attorneys who exit the data. About 33% of them move to in house
counsel positions, 25% of them move to small firms not included in the Martindale data, about 20% move to government
agencies or other occupations, and the remaining 20% could not be located. Importantly, comparison of Linkedin profiles to
Martindale data reveal only one minor discrepancy in the timing of these exits, giving us confidence in the quality of the
Martindale data.

21

Contributions and future work


This paper is, to our knowledge, the paper is the first to comprehensively examine political
ideology and gender equality within organizations, thereby making a critical contribution to the
organizations literature. Findings suggest that liberal male law partners are more likely than
moderate partners to serve on diversity committees and to select female associates for training and
development opportunities, while conservative partners are less likely to do so. We also find that
associates have larger gender disparities in promotion and turnover when partners in their practice
area are more conservative. We find mixed support for the idea that associates will have smaller
gender disparities in promotion and turnover when partners in their practice area are more liberal.
Our theory and findings help extend a broader research agenda linking managerial
characteristics to inequality among subordinates. To the extent that existing work draws primarily on
social identity theory (Tajfel, 1986; Tolbert, Graham & Andrews, 1999) to explain why managers
might or might not support subordinates from their ascriptive group (e.g. Castilla, 2011), our theory
significantly extents this literature by describing why a managers political ideology might influence
his support for gender equality. While we provide three possible explanations (concern for
inequality, beliefs about family responsibilities, and beliefs about leadership qualities), disentangling
the precise mechanism is beyond our scope, given that our paper is an initial foray into this domain,
and a valuable opportunity we leave for future work.
We also contribute to literature that examines how gender roles influence men and womens
access to leadership positions (e.g. Eagly, 2013; Eagly & Karau, 2002). In describing how future
work could advance role congruity theory, Eagly & Karau (2002: 577) note that cultural and
subcultural variation in evaluations of womens leadership qualities warrants further exploration.
We expand this work (e.g. Parker & Ogilvie, 1996) by making a case for political ideology as an
important marker of a managers injuctive beliefs about gender roles. This contribution has the
22

potential to stimulate significant additional research, because a major barrier to studying how gender
roles influence organizations is a lack of information about managers gender role beliefs and
preferences. We provide a useful measure (political ideology) that may help overcome this barrier.
Our examination of diversity committee membership further contributes to the literature
examining organizational practices and policies that affect ascriptive inequality. While most work in
this space takes place at the organization or industry level, examining the causes and consequences of
the implementation of diversity taskforces and diversity training (Dobbin, Kim & Kalev, 2012; Ely,
2004; Kalev, Dobbin & Kelly, 2006), we provide a first glimpse of the types of individuals who lead
these programs. Our finding that, among male partners, political liberals are more likely to lead a
firms diversity committee indicates that these programs may allow individuals who are concerned
with equality issues to push for organizational change. A valuable path for future work is to examine
how the leaders of these organizational initiatives influence the initiatives effectiveness.
Our theory and results also contribute to the burgeoning literature examining how managers
personal values affect their organizational decision making (Hambrick & Mason, 1984; Hambrick,
2007; Christensen et al., 2015). While this work notes the tendency of managers to hire and promote
clones, i.e. those who share their demographic characteristics (e.g. Hambrick, Glaswecisz &
Finkelstein, 1993), researchers do not typically examine how hiring and promotion processes create
an organization that reflects managers underlying values. Our work suggests that an organizations
level of gender inequality may reflect managers political ideology. In a similar way, our work also
extends recent research that suggests that managers may use their organizations to advance social
causes that fit with their preferences (e.g. Chin et al., 2013).
Perhaps the most valuable opportunity for future work is to examine whether ideologicallydriven gender role preferences affect firm performance. Theory and evidence suggest that
organizations often pay for their ascriptive preferences (e.g. Becker, 1957/2010; Siegel, Pyun, &
23

Cheon, 2014). If conservative managers under support or liberal managers over support gender
equality within their firms, organizations may need to consider managers ideology when monitoring
their personnel decisions.
We must highlight limitations and alternate explanations for our findings. The first is
political homophily. To the degree that women are more likely to be liberal, a plausible explanation
for our results is that partners are rewarding associates who share their political views 13. To reduce
this concern we control for political ideology of associates in our estimations. However, we must
caution that it is still a possible mechanism because associates, being younger and less wealthy, have
lower rates of political donations. 14 The second is associate quality. If our theory is correct, the
highest quality female associates may advocate to work for liberal male partners. Although strong
controls for associate quality are used, particularly in the client deal sample, we must emphasize that
associates and partners are not matched randomly in our data, and the ability of liberal partners to
attract the highest quality female associates may affect our results. It is worth noting that this
explanation is largely consistent with our theory because it relies on the assumption that political
ideology affects gender differences in partners treatment of associates. 15 The third explanation
involves resource dependence. If an attorney has more female clients, he may have business
incentives to 1) support liberal causes and 2) to value gender diversity (see Beckman & Phillips,
2005). We are able to account for this possibility in a relatively rich way, particularly in the client
deal analyses where we are able to observe the gender of client personnel involved with the specific
deal. Perfect data would allow us to control for female client personnel at the partner level in all of
our models, but these data are not available, and we must alert readers to this limitation of our work.

13

See Huckfeldt & Sprague (1995) for evidence of political homophily in social networks; Lazer et al., 2010 provide contrary
evidence.
14 Roughly 30% of associates in the promotion sample make a donation during the 1996-2006 period
15 In interviews, associates in our sample suggest that it is uncommon for them to know the political ideology of the individual
partners in their offices, particularly when they are first joining the organization and selecting a practice area.

24

References
Abraham, M. L. B. 2015. Pay Formalization Revisited: Considering the Effects of Manager Gender
and Discretion on Closing the Wage Gap. Academy of Management Journal, Forthcoming.
American Bar Association. 2013. Closing the Gap: A Road Map for Achieving Gender Pay Equity in
Law Firm Partner Compensation.
Aspatore. 2010. M&A Negotiations, 2010 ed.: Leading Lawyers on Structuring Transactions,
Negotiating Agreements, and Addressing Management Concerns (New edition). Boston, Mass.:
Thomson West, Aspatore Books.
Beckman, C. M., & Damon J. Phillips. 2005. Interorganizational Determinants of Promotion: Client
Leadership and the Attainment of Women Attorneys. American Sociological Review, 70(4):
678701.
Belenzon, S., Chatterji, A. K., & Daley, B. 2014. Eponymous Entrepreneurs. Working paper.
Bielby, W. T., & Baron, J. N. 1986. Men and women at work: Sex segregation and statistical
discrimination. American Journal of Sociology, 759799.
Bilimoria, D. 2006. The Relationship Between Women Corporate Directors and Women Corporate
Officers. Journal of Managerial Issues, 18(1).
Bolzendahl, C. I., & Myers, D. J. 2004. Feminist attitudes and support for gender equality: Opinion
change in women and men, 19741998. Social Forces, 83(2): 759789.
Bonica, A. 2014. Mapping the Ideological Marketplace. American Journal of Political Science,
58(2): 367386.
Bonica, A., Chilton, A., & Sen, M. 2015. The Political Ideologies of American Lawyers.
http://scholar.harvard.edu/files/msen/files/lawyers-ideologies.pdf.
Briscoe, F., Chin, M. K., & Hambrick, D. C. 2014. CEO ideology as an element of the corporate
opportunity structure for social activists. Academy of Management Journal, 57(6): 17861809.
Briscoe, F., & Kellogg, K. C. 2011. The Initial Assignment Effect Local Employer Practices and
Positive Career Outcomes for Work-Family Program Users. American Sociological Review,
76(2): 291319.
Castilla, E. J. 2011. Bringing Managers Back In Managerial Influences on Workplace Inequality.
American Sociological Review, 76(5): 667694.
Chatain, O., & Meyer-Doyle, P. 2015. Alleviating managerial dilemmas in human-capital-intensive
firms through incentives: Evidence from m&a legal advisors. Strategic
Chin, M. K., Hambrick, D. C., & Trevio, L. K. 2013. Political Ideologies of CEOs The Influence of
Executives Values on Corporate Social Responsibility. Administrative Science Quarterly,
58(2): 197232.
Christensen, D. M., Dhaliwal, D. S., Boivie, S., & Graffin, S. D. 2014. Top management
conservatism and corporate risk strategies: Evidence from managers personal political
orientation and corporate tax avoidance. Strategic Management Journal.
Cialdini, R. B., & Trost, M. R. 1998. Social influence: Social norms, conformity and compliance. In
D. T. Gilbert, S. T. Fiske, & G. Lindzey (Eds.), The handbook of social psychology, Vols. 1 and
2 (4th ed.): 151192. New York, NY, US: McGraw-Hill.
Cohen, L. E., Broschak, J. P., & Haveman, H. A. 1998. And then there were more? The effect of
organizational sex composition on the hiring and promotion of managers. American
Sociological Review, 711727.
Cohen, P. N., & Huffman, M. L. 2007. Working for the Woman? Female Managers and the Gender
Wage Gap. American Sociological Review, 72(5): 681704.
Cotter, D., Hermsen, & Vanneman, R. 2011. The End of the Gender Revolution? Gender Role
Attitudes from 1977 to 2008. American Journal of Sociology, 117(1): 259289.
25

Dahl, M. S., Dezs, C. L., & Ross, D. G. 2012. Fatherhood and Managerial Style How a Male CEOs
Children Affect the Wages of His Employees. Administrative Science Quarterly, 57(4): 669.
Davis, S. N., & Greenstein, T. N. 2009. Gender ideology: Components, predictors, and consequences.
Annual Review of Sociology, 35: 87105.
Deshpande, R., Hoyer, W. D., & Donthu, N. 1986. The intensity of ethnic affiliation: A study of the
sociology of Hispanic consumption. Journal of Consumer Research, 214220.
Dobbin, F., Kim, S., & Kalev, A. 2011. You Cant Always Get What You Need Organizational
Determinants of Diversity Programs. American Sociological Review, 76(3): 386411.
Dolan, K. 2014. Gender Stereotypes, Candidate Evaluations, and Voting for Women Candidates
What Really Matters? Political Research Quarterly, 67(1): 96107.
Duguid, M. 2011. Female tokens in high-prestige work groups: Catalysts or inhibitors of group
diversification? Organizational Behavior and Human Decision Processes, 116(1): 104115.
Duguid, M. M., Loyd, D. L., & Tolbert, P. S. 2012. The impact of categorical status, numeric
representation, and work group prestige on preference for demographically similar others: A
value threat approach. Organization Science, 23(2): 386401.
Eagly, A. H. 2013. Sex differences in social behavior: A social-role interpretation. Psychology Press.
Eagly, A. H., & Karau, S. J. 2002. Role congruity theory of prejudice toward female leaders.
Psychological Review, 109(3): 573.
Eagly, A. H., Wood, W., & Diekman, A. B. 2000. Social role theory of sex differences and
similarities: A current appraisal. The Developmental Social Psychology of Gender, 123174.
Edlund, L., & Pande, R. 2002. Why Have Women Become Left-Wing? The Political Gender Gap and
the Decline in Marriage. The Quarterly Journal of Economics, 117(3): 917961.
Elliott, J. R., & Smith, R. A. 2004. Race, gender, and workplace power. American Sociological
Review, 69(3): 365386.
Ely, R. J. 1994. The effects of organizational demographics and social identity on relationships
among professional women. Administrative Science Quarterly, 203238.
Ely, R. J. 1995. The Power In Demography: WomenS Social Constructions Of Gender Identity At
Work. Academy of Management Journal, 38(3): 589634.
Ely, R. J. 2004. A field study of group diversity, participation in diversity education programs, and
performance. Journal of Organizational Behavior, 25(6): 755780.
Erikson, R. S., & Tedin, K. L. 2000. American Public Opinion: Its Orgin, Contents, and Impact.
Longman Publishing Group.
Ferguson, J.-P. 2015. The Control of Managerial Discretion: Evidence from Unionizations Impact on
Employment Segregation. American Journal of Sociology, 121(3): 675721.
Fernandez-Mateo, I., & King, Z. 2011. Anticipatory sorting and gender segregation in temporary
employment. Management Science, 57(6): 9891008.
Gorman, E. H. 2005. Gender stereotypes, same-gender preferences, and organizational variation in
the hiring of women: Evidence from law firms. American Sociological Review, 70(4): 702728.
Gupta, A., Briscoe, F., & Hambrick, D. 2015. Organizational Ideology and Corporate Strategy: A
New Vantage on Socialistic Resource Allocation. Working Paper, University of Washington.
Hambrick, D. C. 2007. Upper echelons theory: An update. Academy of Management Review, 32(2):
334343.
Hambrick, D. C., & Mason, P. A. 1984. Upper echelons: The organization as a reflection of its top
managers. Academy of Management Review, 9(2): 193206.
Huckfeldt, R. R., & Sprague, J. 1995. Citizens, Politics and Social Communication: Information and
Influence in an Election Campaign. Cambridge University Press.
Huffman, M. L., Cohen, P. N., & Pearlman, J. 2010. Engendering Change: Organizational Dynamics
26

and Workplace Gender Desegregation, 19752005. Administrative Science Quarterly, 55(2):


255277.
Hultin, M., & Szulkin, R. 1999. Wages and Unequal Access to Organizational Power: An Empirical
Test of Gender Discrimination. Administrative Science Quarterly, 44(3): 453472.
Ibarra, H. 1992. Homophily and differential returns: Sex differences in network structure and access
in an advertising firm. Administrative Science Quarterly, 422447.
Jost, J. T. 2006. The end of the end of ideology. American Psychologist, 61(7): 651670.
Jost, J. T., Federico, C. M., & Napier, J. L. 2009. Political Ideology: Its Structure, Functions, and
Elective Affinities. Annual Review of Psychology, 60(1): 307337.
Jost, J. T., Glaser, J., Kruglanski, A. W., & Sulloway, F. J. 2003. Political conservatism as motivated
social cognition. Psychological Bulletin, 129(3): 339375.
Kalev, A., Dobbin, F., & Kelly, E. 2006. Best Practices or Best Guesses? Assessing the Efficacy of
Corporate Affirmative Action and Diversity Policies. American Sociological Review, 71(4):
589617.
Kelly, E., & Dobbin, F. 1999. Civil rights law at work: Sex discrimination and the rise of maternity
leave policies. American Journal of Sociology, 105(2): 45592.
Kim, S., Kalev, A., & Dobbin, F. 2012. Progressive corporations at work: The case of diversity
programs. NYU Rev. L. & Soc. Change, 36: 171.
King, D. C., & Matland, R. E. 2003. Sex and the grand old party an experimental investigation of the
effect of candidate sex on support for a republican candidate. American Politics Research,
31(6): 595612.
Klein, E. 1986. Gender Politics: From Consciousness to Mass Politics. Harvard Univ Pr.
Koch, A. J., DMello, S. D., & Sackett, P. R. 2015. A meta-analysis of gender stereotypes and bias in
experimental simulations of employment decision making. Journal of Applied Psychology,
100(1): 128161.
Kulis, S. 1997. Gender segregation among college and university employees. Sociology of Education,
151173.
Kurtulus, F. A., & Tomaskovic-Devey, D. 2012. Do female top managers help women to advance? A
panel study using EEO-1 records. The ANNALS of the American Academy of Political and
Social Science, 639(1): 173197.
Lazer, D., Rubineau, B., Chetkovich, C., Katz, N., & Neblo, M. 2010. The Coevolution of Networks
and Political Attitudes. Political Communication, 27(3): 248274.
Mason, K. O., & Lu, Y.-H. 1988. Attitudes toward womens familial roles: Changes in the United
States, 1977-1985. Gender & Society, 2(1): 3957.
Matsa, D. A., & Miller, A. R. 2011. Chipping Away at the Glass Ceiling: Gender Spillovers in
Corporate Leadership. American Economic Review, 101(3): 63539.
McGinn, K. L., & Milkman, K. L. 2012. Looking Up and Looking Out: Career Mobility Effects of
Demographic Similarity Among Professionals. Organization Science, 24(4): 10411060.
Morrison, A. M., White, R. P., & Velsor, E. V. 1992. Breaking the Glass Ceiling: Can Women Reach
the Top of Americas Largest Corporations? Basic Books.
Parkin, R., & Baker, G. 2006. The Changing Structure of the Legal Services Industry and the Careers
of Lawyers. North Carolina Law Review, 84: 16351682.
Peck. 2015. Salesforce CEO Takes Radical Step To Pay Men And Women Equally. The Huffington
Post.
Perry, E. L., Davis-Blake, A., & Kulik, C. T. 1994. Explaining gender-based selection decisions: A
synthesis of contextual and cognitive approaches. Academy of Management Review, 19(4): 786
820.
27

Phillips, D. J. 2002. A Genealogical Approach to Organizational Life Chances: The Parent-Progeny


Transfer among Silicon Valley Law Firms, 1946-1996. Administrative Science Quarterly, 47(3):
474506.
Phillips, D. J. 2005. Organizational Genealogies and the Persistence of Gender Inequality: The Case
of Silicon Valley Law Firms. Administrative Science Quarterly, 50(3): 440472.
Philpot, T. S., & Walton, H. 2007. One of our own: Black female candidates and the voters who
support them. American Journal of Political Science, 51(1): 4962.
Rosenwasser, S. M., & Dean, N. G. 1989. Gender role and political office: Effects of perceived
masculinity/femininity of candidate and political office. Psychology of Women Quarterly, 13(1):
7785.
Sanbonmatsu, K. 2002. Gender stereotypes and vote choice. American Journal of Political Science,
2034.
Shenhav, Y., & Haberfeld, Y. 1992. Organizational demography and inequality. Social Forces, 71(1):
123143.
Siegel, J. I., Pyun, L., & Cheon, B. Y. 2014. Multinational firms, labor market discrimination, and the
capture of competitive advantage by exploiting the social divide. Harvard Business School
Strategy Unit Working Paper, (11-011).
Steel. 2015. Netflix Offers Expanded Maternity and Paternity Leave - The New York Times. New
York Times.
Tajfel, H. 1982. Social Psychology of Intergroup Relations. Annual Review of Psychology, 33(1): 1
39.
Tetlock, P. E. 2000. Cognitive biases and organizational correctives: Do both disease and cure
depend on the politics of the beholder? Administrative Science Quarterly, 45(2): 293326.
Tsui, A. S., Egan, T. D., & OReilly III, C. A. 1992. Being different: Relational demography and
organizational attachment. Administrative Science Quarterly, 549579.
Williams, J. C., & Richardson, V. 2010. New Millennium, Same Glass Ceiling-The Impact of Law
Firm Compensation Systems on Women. Hastings LJ, 62: 597.

28

Table 1: Gender Role Attitudes and Political Ideology of Male Respondents to the General Social Survey, 2000-2014
% Male Respondents Supporting / Agreeing with the Statement
Question
name

Question text

Gender and economic roles within the family


A working mother can establish just as warm and secure
a relationship with her children as a mother who does not
1 fechld
work
A preschool child is likely to suffer if his or her mother
2 fepresch
works
It is much better for everyone involved if the man is the
achiever outside the home and the woman takes care of
3 femfam
the home and family .
4 hubbywk1

A man's job is to earn money; a woman's job is to look


after the home and family.

Extremely
Slightly
Slightly
Extremely
Liberal
Liberal Liberal Moderate Conserv. Conserv. Conserv.

67%

69%

67%

67%

60%

51%

42%
4847

38%

39%

39%

43%

50%

58%

67%

29%

27%

28%

35%

38%

53%

65%

4802

4818
29%

35%

30%

43%

48%

61%

81%

78%

74%

52%

62%

55%

47%

48%

43%

38%

31%

31%

24%

19%

23%

16%

19%

18%

21%

27%

32%

45%

552

Organizational responses to gender inequality


5 fehire
6 fejobaff

Because of past discrimination, employers should make


special efforts to hire and promote qualified women.
Do you support the preferential hiring and promotion of
women?

Differences in leadership characteristics by gender


Most men are better suited emotionally for politics than
7 fepol
are most women
Source: General Social Survey http://www3.norc.org/Gss+website/

Bolded cells are different from Moderate respondents at p<.05. Calculations include survey weights (wtssall). Unweighted trends are similar.
Respondents include men only.
Table includes gender role questions with at least 500 male responses during the sample period. For additional questions with smaller samples, see GSS website.

29

2379
2321

4530

Table 2: Summary Statistics: Partner Attorney Sample for Tests of H1 (diversity committee membership)
Female
Male: Liberal
n
Mean
StD
n
Mean
StD
Attorney serves on diversity committee
35281
0.020 0.140 22116
0.012 0.107
Donations to Democrats by attorney ($)
35281
4406 131066 22116 12000 85851
Donations to Republicans by attorney ($)
35281
1414 23495 22116
500
1977
Attorney is liberal (>75% don$ to Dem)
35281
0.153 0.360 22116
1.000 0.000
Attorney is conservative (>75% don$ to Repub)
35281
0.035 0.184 22116
0.000 0.000
Donation count (any)
35281 12.689 33.356 22116 32.214 48.592
Avg donations to Dem by male part. in prac area ($) 35075
4396 12119 22053
6283 11140
Avg don. to Repub. by male part. in prac area ($)
35075
2814 9165
22053
2793 5865
Age
35281 47.573 8.430 22116 54.842 8.475
Years since JD
35281 20.290 7.813 22116 28.459 8.761
Tenure with firm
35281
8.405 4.246 22116
9.243 4.289
Tenure is left-censored
35281
0.399 0.490 22116
0.548 0.498
% partners in office from attorney's law school
35281
0.119 0.158 22116
0.109 0.146
Female
35281
1.000 0.000 22116
0.000 0.000
Top 18 law school
35281
0.286 0.452 22116
0.442 0.497
Peer Rating: A
35281
0.389 0.488 22116
0.695 0.460
Peer Rating: B
35281
0.090 0.287 22116
0.049 0.215
Peer Rating: C
35281
0.009 0.095 22116
0.003 0.059
% associates in prac. area who are female
33360
0.423 0.189 21152
0.430 0.172
# male partners in practice area
35281 25.090 23.943 22116 28.882 25.663
% partners in practice area who are female
35281
0.187 0.131 22116
0.181 0.113
Avg age of male partners in prac area
35249 51.523 3.228 22116 52.025 2.876
Firm reports a diversity committee
35281
1.000 0.000 22116
1.000 0.000
Firm size (# attorneys, 1000s)
35281
0.472 0.313 22116
0.502 0.314
Profit per equity partner ($1M)
26776
0.924 0.499 18218
0.955 0.483
% of female leadership among client personnel
33978
0.078 0.061 21407
0.077 0.061
Firm acquires another firm
35281
0.072 0.259 22116
0.073 0.260
Office is acquired
35281
0.004 0.061 22116
0.004 0.064
NYC office
35281
0.088 0.283 22116
0.107 0.309
San Francisco office
35281
0.032 0.176 22116
0.029 0.167
Chicago office
35281
0.089 0.285 22116
0.078 0.269
DC office
35281
0.113 0.317 22116
0.159 0.366
Office located in Southern US
35281
0.234 0.423 22116
0.179 0.383
Observations
35281
22116

Male: Moderate
n
Mean
StD
113231 0.005 0.073
113231 2865 18293
113231 2187 12482
113231 0.000 0.000
113231 0.000 0.000
113231 14.952 37.070
112612 4054 9725
112612 2679 7585
113231 50.516 9.378
113231 23.880 9.597
113231 8.472 4.381
113231 0.429 0.495
113231 0.111 0.156
113231 0.000 0.000
113231 0.316 0.465
113231 0.499 0.500
113231 0.074 0.263
113231 0.005 0.072
106751 0.416 0.190
113231 25.848 23.860
113231 0.176 0.122
113231 51.453 3.137
113231 1.000 0.000
113231 0.470 0.317
84800
0.942 0.503
108388 0.078 0.062
113231 0.072 0.258
113231 0.004 0.064
113231 0.108 0.310
113231 0.026 0.161
113231 0.094 0.292
113231 0.104 0.305
113231 0.246 0.431
113231

Male: Conservative
n
Mean
StD
12731 0.004 0.061
12731
489
2013
12731 11352 44898
12731 0.000 0.000
12731 1.000 0.000
12731 31.058 42.036
12684 3666 19229
12684 4099 7531
12731 54.354 8.594
12731 28.118 8.806
12731 9.926 4.031
12731 0.628 0.483
12731 0.141 0.182
12731 0.000 0.000
12731 0.306 0.461
12731 0.731 0.444
12731 0.070 0.256
12731 0.003 0.057
12137 0.407 0.193
12731 26.918 24.819
12731 0.168 0.113
12731 51.629 3.016
12731 1.000 0.000
12731 0.457 0.302
9179
0.878 0.480
12244 0.078 0.059
12731 0.062 0.242
12731 0.004 0.064
12731 0.048 0.214
12731 0.004 0.063
12731 0.052 0.223
12731 0.098 0.297
12731 0.284 0.451
12731

Level of analysis is the attorney-year. Liberal partners are those who make at least five donations, with at least 75% ($) to Democrats. Conservatives are those who make at least
five donations, with at least 75% to Republicans. Moderates are the remainder; 29% have zero donations.

30

Table 3: DV: 0/1, Partner attorney is member of firm's diversity committee


(1)
(2)
Full sample
Full sample
Donations to Democrats by attorney (log$)
Donations to Republicans by attorney (log$)

0.0010***
(0.0001)
-0.0007***
(0.0001)

0.0011***
(0.0001)
-0.0005***
(0.0001)

(3)
Male only
0.0009***
(0.0001)
-0.0003***
(0.0001)

(4)
White male
only
0.0008***
(0.0001)
-0.0003***
(0.0001)

(6)
White male
only
0.1532***
(0.0115)
-0.0442***
(0.0105)

0.0056***
(0.0008)
-0.0015*
(0.0006)

Attorney is liberal (>75% don$ to Dem)


Attorney is conservative (>75% don$ to Repub)
Female
Age
Years since JD
Tenure with firm
Tenure is left-censored
% partners in office from attorney's law school
Top 18 law school
Peer Rating: A
Estimation
Fixed effects
Additional controls from Table 1
Additional prac. area dummies
Year dummies
N atty-year obs
R-sq
Log Likelihood
Mean of DV

(5)
White male
only

OLS
None
No
No
No
189008
0.00
177826.1
0.009

0.0126***
(0.0008)
0.0000
(0.0001)
-0.0001
(0.0001)
0.0008***
(0.0001)
-0.0015+
(0.0009)
-0.0051**
(0.0019)
0.0024***
(0.0005)
0.0004
(0.0006)
OLS
Office-Year
Yes
Yes
N/A
189008
0.04
181641.0
0.009

-0.0000
(0.0001)
-0.0001
(0.0001)
0.0005***
(0.0001)
-0.0012
(0.0008)
-0.0029
(0.0019)
0.0010*
(0.0005)
-0.0003
(0.0006)
OLS
Office-Year
Yes
Yes
N/A
148078
0.05
170186.6
0.006

-0.0000
(0.0001)
0.0000
(0.0001)
0.0005***
(0.0001)
-0.0014+
(0.0008)
-0.0012
(0.0019)
0.0015**
(0.0005)
-0.0000
(0.0006)
OLS
Office-Year
Yes
Yes
N/A
144105
0.05
170906.2
0.006

Sample consists of partner attorneys working for firms who list a diversity committee in the Yellow Book Leadership Directory.
Robust standard errors clustered on offices in parentheses. + p<0.10, * p<0.05, ** p<0.01, *** p<0.001, two-tailed tests
Donations are measured between 1996-2006, years of estimation are 2007-2012.

31

-0.0000
(0.0001)
0.0001
(0.0001)
0.0005***
(0.0001)
-0.0013
(0.0008)
-0.0010
(0.0019)
0.0016***
(0.0005)
0.0002
(0.0006)
OLS
Office-Year
Yes
Yes
N/A
144105
0.05
170864.3
0.006

-0.0015
(0.0167)
-0.0010
(0.0163)
0.1086***
(0.0199)
-0.3764**
(0.1358)
-0.0853
(0.3133)
0.2560**
(0.0803)
-0.0394
(0.1094)
CLogit
Office-Year
Yes
Yes
N/A
33595
-2629.4
0.025

Table 4: Testing H2: Gender inequality in associate training and development as a function of a partners political ideology.
(1)
(2)
(3)
(4)
(5)
DV in all models: Associate is member of
Full sample Male partners Male partners Male partners Male partners
partners team for client deal (e.g. M&A, IPO)
only
only
only
only
Female* Donations to Dem by partner (log$)
0.0010*
0.0009+
0.0009+
0.0077
(0.0005)
(0.0005)
(0.0005)
(0.0049)
Female*Donations to Repub by partner (log$)
-0.0022***
-0.0021***
-0.0020***
-0.0202***
(0.0005)
(0.0006)
(0.0006)
(0.0055)
Female*Partner is liberal
0.0077+
(0.0045)
Female*Partner is conservative
-0.0279**
(0.0098)
Female associate
-0.0012
-0.0025
-0.0018
-0.0043*
-0.0140
(0.0023)
(0.0025)
(0.0025)
(0.0021)
(0.0234)
Deals by associate last year
0.0087***
0.0087***
0.0756***
(0.0006)
(0.0006)
(0.0051)
Top 18 law school
-0.0021
-0.0021
-0.0189
(0.0019)
(0.0019)
(0.0183)
Years since JD
0.0000+
0.0000+
0.0001+
(0.0000)
(0.0000)
(0.0001)
Age
0.0001
0.0001
0.0010
(0.0003)
(0.0003)
(0.0019)
Tenure with firm
-0.0023***
-0.0023***
-0.0195***
(0.0005)
(0.0005)
(0.0044)
Tenure is left-censored
-0.0045
-0.0046
-0.0419
(0.0090)
(0.0090)
(0.0836)
Donations to Republicans by associate (log$)
-0.0006
-0.0006
-0.0050
(0.0006)
(0.0006)
(0.0061)
Donations to Democrats by associate (log$)
-0.0007+
-0.0007+
-0.0084+
(0.0004)
(0.0004)
(0.0045)
Associate and partner, same law school
0.0038
0.0038
0.0317
(0.0034)
(0.0034)
(0.0329)
Estimation
OLS
OLS
OLS
OLS
CLogit
Fixed effects
Partner-Deal Partner-Deal Partner-Deal Partner-Deal Partner-Deal
N associate-deal obs
179866
154025
154025
154025
154025
Log Likelihood
-46390.9
-38843.8
-38700.9
-38700.5
-38143.6
Mean of DV
0.132
0.131
0.131
0.131
0.131

(6)
White male
partners only
0.0010+
(0.0005)
-0.0020***
(0.0006)

(7)
Exclude deals
w/fem. clients
0.0006
(0.0005)
-0.0016**
(0.0006)

-0.0029
(0.0027)
0.0083***
(0.0006)
-0.0012
(0.0020)
0.0000**
(0.0000)
0.0002
(0.0003)
-0.0024***
(0.0005)
-0.0058
(0.0094)
-0.0007
(0.0006)
-0.0008+
(0.0005)
0.0038
(0.0036)
OLS
Partner-Deal
136157
-34213.9
0.131

-0.0017
(0.0026)
0.0085***
(0.0006)
-0.0019
(0.0020)
0.0000
(0.0000)
0.0001
(0.0003)
-0.0023***
(0.0005)
-0.0068
(0.0092)
-0.0006
(0.0006)
-0.0007+
(0.0004)
0.0018
(0.0035)
OLS
Partner-Deal
148317
-37278.5
0.131

Choice set for each deal consists of all associates who share partner's office and practice area and have served on a deal in the last year. Donations measured 1996-2006.
Data stem from 5,702 unique client deals completed from 2007-2012. Median deal size is $250M, median deal team contains four attorneys.
Robust standard errors in parentheses, clustered on partner-deals. + p<0.10, * p<0.05, ** p<0.01, *** p<0.001, two-tailed tests

32

Table 5: Summary Statistics: Associate Attorney Sample for Tests of H3 (promotion) and H4 (turnover)
(1)
(2)
Female
Male
n
Mean StDv
n
Mean StDv
Promoted to partner, t+1
35990 0.024 0.152
45052
0.042 0.201
Exits firms (any dest.)
35990 0.217 0.412
45052
0.207 0.405
Joins competing law firm
35990 0.062 0.241
45052
0.074 0.262
Exits legal services industry
35990 0.145 0.352
45052
0.122 0.327
Donations to Democrats by attorney ($)
35990
511
5394
45052
602
5025
Donations to Republicans by attorney ($)
35990
386
7185
45052
479
6776
Attorney is liberal (>75% don$ to Dem)
35990 0.020 0.140
45052
0.028 0.164
Attorney is conserv. (>75% don$ to Repub)
35990 0.002 0.041
45052
0.009 0.094
Donation count (any)
35990 3.086 13.57
45052
4.042 16.08
Avg don. to Dem. by male part. in prac area ($)
35990 7291 106087 45052
5650 53121
Avg don. to Rep. by male part. in prac area ($)
35990 2801 5230
45052
2882 5347
Age
35990 33.448 4.640
45052 33.874 4.595
Years since JD
35990 6.594 3.748
45052
6.547 3.454
Tenure with firm
35990 3.917 2.719
45052
3.690 2.497
Tenure is left-censored
35990 0.025 0.156
45052
0.014 0.117
% partners in office from attorney's law school
35990 0.088 0.116
45052
0.089 0.121
Female
35990 1.000 0.000
45052
0.000 0.000
Top 18 law school
35990 0.321 0.467
45052
0.342 0.474
% associates in prac. area who are female
35990 0.416 0.134
45052
0.405 0.133
# male partners in practice area
35990 24.016 22.773 45052 25.764 23.278
% partners in practice area who are female
35990 0.176 0.117
45052
0.164 0.108
Avg age of male partners in prac area
35990 50.494 3.152
45052 50.271 3.163
Firm has a diversity committee
35990 0.470 0.499
45052
0.453 0.498
Firm size (# attorneys, 1000s)
35990 0.574 0.286
45052
0.567 0.289
Profit per equity partner ($1M)
35990 1.044 0.564
45052
1.078 0.610
% of female leadership among client personnel
35990 0.077 0.069
45052
0.075 0.067
Firm acquires another firm
35990 0.097 0.296
45052
0.094 0.292
Office is acquired
35990 0.007 0.083
45052
0.007 0.082
NYC office
35990 0.191 0.393
45052
0.191 0.393
San Francisco office
35990 0.041 0.199
45052
0.029 0.168
Chicago office
35990 0.075 0.264
45052
0.074 0.262
DC office
35990 0.140 0.346
45052
0.139 0.346
Office located in Southern US
35990 0.198 0.398
45052
0.212 0.409
Observations
35990
45052

(3)
Liberal Prac. Area
n
Mean StDv
40351 0.030 0.170
40351 0.213 0.409
40351 0.068 0.252
40351 0.134 0.341
40351
647
5731
40351
415
5300
40351 0.030 0.170
40351 0.004 0.065
40351 4.008 16.38
40351 10005 114660
40351 1454 1915
40351 33.677 4.626
40351 6.560 3.591
40351 3.699 2.587
40351 0.018 0.135
40351 0.081 0.104
40351 0.459 0.498
40351 0.374 0.484
40351 0.422 0.127
40351 26.290 23.359
40351 0.175 0.107
40351 50.445 3.078
40351 0.462 0.499
40351 0.569 0.302
40351 1.113 0.589
40351 0.077 0.069
40351 0.094 0.291
40351 0.007 0.082
40351 0.248 0.432
40351 0.052 0.222
40351 0.066 0.249
40351 0.157 0.364
40351 0.119 0.323
40351

(5)
Conserv. Prac. Area
n
Mean StDv
14878 0.040 0.196
14878 0.202 0.401
14878 0.071 0.257
14878 0.120 0.325
14878
419
4121
14878
441
7271
14878 0.017 0.129
14878 0.007 0.083
14878 2.916 11.93
14878 1179 1506
14878 5634 10100
14878 33.722 4.659
14878 6.445 3.524
14878 3.799 2.572
14878 0.018 0.133
14878 0.108 0.148
14878 0.428 0.495
14878 0.246 0.431
14878 0.388 0.150
14878 18.075 17.855
14878 0.164 0.126
14878 50.063 3.093
14878 0.444 0.497
14878 0.562 0.268
14878 0.926 0.505
14878 0.072 0.064
14878 0.107 0.309
14878 0.006 0.079
14878 0.071 0.257
14878 0.012 0.109
14878 0.062 0.241
14878 0.082 0.275
14878 0.369 0.482
14878

Level of analysis is the attorney-year. Liberal practice areas are those where the average donation ($) of male partners to Democrats is at least twice that of Republicans. Similarly,
conservative practice areas are those where the average donation ($) of male partners to Republicans is at least twice that of Democrats.

33

Table 6: Testing H3: Gender inequality in associate promotion as a function of partners political ideology
(1)
(2)
(3)
(4)
(5)
DV in all models: Associate is promoted to partner in t+1
Full sample Full sample Full sample Liberal Prac. Moderate
Area
Prac. Area
Female* Avg don. to Dem. by male part. in prac area (log$)
0.0019*
0.0021*
0.0424+
(0.0008)
(0.0008)
(0.0249)
Female*Avg don. to Repub. by male part. in prac area (log$) -0.0026***
-0.0025***
-0.0393+
(0.0007)
(0.0007)
(0.0228)
Female associate
-.0212***
-.0229***
-.7245***
-0.0192***
-0.0261***
(.0063)
(.0065)
(.2059)
(0.0022)
(0.0029)
Avg donations to Dem. by male part. in prac area (log$)
-0.0024**
-0.0011
-0.0153
(0.0009)
(0.0007)
(0.0180)
Avg donations to Repub by male part. in prac area (log$)
0.0034***
0.0010
0.0086
(0.0007)
(0.0007)
(0.0176)
Donations to Democrats by associate (log$)
0.0001
0.0101
0.0006
-0.0002
(0.0004)
(0.0084)
(0.0004)
(0.0007)
Donations to Republicans by associate (log$)
0.0012**
0.0240**
0.0001
0.0018*
(0.0005)
(0.0093)
(0.0006)
(0.0008)
Years since JD
0.0065***
0.1672***
0.0065***
0.0064***
(0.0006)
(0.0119)
(0.0008)
(0.0010)
Tenure with firm
0.0145***
0.3139***
0.0126***
0.0151***
(0.0009)
(0.0143)
(0.0013)
(0.0014)
Tenure is left-censored
-0.0813***
-1.6074***
-0.0894***
-0.0630***
(0.0108)
(0.1727)
(0.0165)
(0.0176)
% partners in office from attorney's law school
0.0115
0.1214
0.0054
0.0200
(0.0082)
(0.1699)
(0.0124)
(0.0153)
Top 18 law school
0.0018
0.0355
0.0010
0.0022
(0.0016)
(0.0458)
(0.0022)
(0.0031)
Estimation
OLS
OLS
CLogit
OLS
OLS
Fixed effects
None
Office
Office
Office
Office
Additional controls
No
Yes
Yes
Yes
Yes
23 Legal specialty dummies
No
Yes
Yes
Yes
Yes
Year dummies
No
Yes
Yes
Yes
Yes
N atty-year obs
81042
81042
72933
40351
25813
R-sq
0.00
0.04
0.03
0.04
Log Likelihood
11722.8
14757.9
-11033.4
9792.5
4000.6
Mean of DV
0.046
0.046
0.051
Robust standard errors clustered on offices in parentheses. + p<0.10, * p<0.05, ** p<0.01, *** p<0.001, two-tailed tests
See Table 1 for list of additional controls (attorney age, firm size, firm profits, female clients, etc.)
Donations are measured between 1996-2006, years of estimation are 2007-2012.

34

(6)
Conserv.
Prac. Area

-0.0353***
(0.0042)

-0.0002
(0.0011)
0.0024*
(0.0010)
0.0081***
(0.0013)
0.0186***
(0.0021)
-0.1001***
(0.0280)
0.0080
(0.0163)
0.0035
(0.0046)
OLS
Office
Yes
Yes
Yes
14878
0.06
2004.0

Table 7: Robustness tests of H3: Gender inequality in associate promotion as a function of partners political ideology
(1)
(2)
(3)
(4)
DV in all models: Associate is promoted to partner in t+1
Full sample
Exclude
Exclude NYC Full sample
Southern
offices
offices
Female* Avg don. to Dem. by male part. in prac area (log$)
0.0022**
0.0017+
0.0021*
0.0014+
(0.0008)
(0.0009)
(0.0010)
(0.0008)
Female*Avg don. to Repub. by male part. in prac area (log$)
-0.0025***
-0.0021**
-0.0025**
-0.0026***
(0.0007)
(0.0008)
(0.0008)
(0.0008)
Female associate
-.0229***
-.0205***
-.0242***
-.0134***
(.0065)
(.0073)
(.0076)
(.0065)
Avg donations to Dem. by male part. in prac area (log$)
-0.0011
-0.0012
-0.0009
-0.0002
(0.0007)
(0.0008)
(0.0009)
(0.0008)
Avg donations to Repub by male part. in prac area (log$)
0.0010
0.0005
0.0012
0.0013+
(0.0007)
(0.0007)
(0.0008)
(0.0007)
Female* % liberal male partners in prac. area
Female*% conservative male partners in prac. area
% liberal male partners in prac. area
% conservative male partners in prac. area
Estimation
Fixed effects
Additional controls
23 Legal specialty dummies
50 additional law school dummies
Year dummies
N atty-year obs
R-sq
Log Likelihood
Mean of DV

OLS
Office
Yes
Yes
Yes
Yes
81042
0.04
14790.8
0.046

OLS
Office
Yes
Yes
No
Yes
64360
0.03
12954.3
0.044

OLS
Office
Yes
Yes
No
Yes
65571
0.04
9241.7
0.051

Robust standard errors clustered on offices in parentheses. + p<0.10, * p<0.05, ** p<0.01, *** p<0.001, two-tailed tests
See Table 1 for list of additional controls (firm size, profits, female clients, etc.)
Donations are measured between 1996-2006, years of estimation are 2007-2012.

35

OLS
Office-Year
Yes
Yes
No
N/A
81042
0.04
18842.1
0.046

(5)
Full sample

-.0201
(.0032)

0.0009
(0.0092)
-0.0347**
(0.0106)
0.0160+
(0.0087)
0.0291**
(0.0103)
OLS
Office
Yes
Yes
No
N/A
81042
0.04
14761.9
0.046

Table 8: Testing H4: Gender inequality in associate turnover as a function of partners political ideology
(1)
(2)
(3)
DV: Exit
DV: Exit
DV: Exit
(any dest.)
(any dest.)
(any dest.)
Female* Avg don. to Dem. by male part. in prac area (log$)
-0.0011
-0.0029+
-0.0247+
(0.0015)
(0.0016)
(0.0137)
Female*Avg don. to Repub. by male part. in prac area (log$)
0.0039**
0.0043**
0.0364**
(0.0015)
(0.0014)
(0.0120)
Avg donations to Democrats by male part. in prac area (log$)
0.0006
0.0016
0.0143
(0.0013)
(0.0015)
(0.0125)
Avg donations to Republicans by male part. in prac area (log$)
-0.0040**
-0.0018
-0.0154
(0.0013)
(0.0014)
(0.0112)
Female associate
-.0161
.0100
-.0353
(.0119)
(.0120)
(.1072)
Donations to Democrats by associate (log$)
0.0004
0.0029
(0.0005)
(0.0045)
Donations to Republicans by associate (log$)
0.0004
0.0037
(0.0007)
(0.0054)
Age
0.0027***
0.0223***
(0.0005)
(0.0039)
Years since JD
0.0003
0.0015
(0.0008)
(0.0059)
Tenure with firm
0.0015
0.0135+
(0.0010)
(0.0076)
Tenure is left-censored
-0.0542***
-0.4814***
(0.0132)
(0.1283)
% partners in office from attorney's law school
-0.0461***
-0.4080***
(0.0130)
(0.1172)
Top 18 law school
0.0187***
0.1532***
(0.0033)
(0.0268)
Estimation
OLS
OLS
CLogit
Fixed effects
None
Office
Office
Additional controls / 23 Legal specialty dummies
No
Yes
Yes
Year dummies
No
Yes
Yes
N atty-year obs
74906
74906
73700
R-sq
0.00
0.02
Log Likelihood
-28929.7
-26018.6
-27390.0
Mean of DV
0.149
0.149
0.150

(4)
(5)
DV: Moves to DV: Exits data
competitor
-0.0011
-0.0023*
(0.0011)
(0.0011)
0.0022*
0.0019+
(0.0010)
(0.0010)
-0.0007
0.0025*
(0.0011)
(0.0010)
-0.0001
-0.0017+
(0.0010)
(0.0009)

-0.0000
(0.0004)
0.0009*
(0.0005)
0.0024***
(0.0004)
-0.0024***
(0.0005)
0.0007
(0.0006)
-0.0229**
(0.0078)
-0.0288**
(0.0092)
0.0003
(0.0022)
OLS
Office
Yes
Yes
74906
0.02
2210.1
0.064

0.0001
(0.0004)
-0.0006
(0.0005)
0.0002
(0.0003)
0.0027***
(0.0005)
0.0007
(0.0007)
-0.0279**
(0.0106)
-0.0153
(0.0099)
0.0177***
(0.0024)
OLS
Office
Yes
Yes
74906
0.01
-4694.9
0.076

Robust standard errors clustered on offices in parentheses. + p<0.10, * p<0.05, ** p<0.01, *** p<0.001, two-tailed tests
See Table 1 for list of additional controls (firm size, profits, female clients, etc.). Donations are measured between 1996-2006, years of estimation are 2007-2012.

36

Table 9: Robustness tests for H4: Gender inequality in associate turnover as a function of partners political ideology
(1)
(2)
(3)
(4)
DV in all models: Associate exits the firm in t+1
Full sample
Exclude
Exclude
Full sample
Southern
NYC offices
offices
Female* Avg don. to Dem. by male part. in prac area (log$)
-0.0031+
-0.0031+
-0.0035+
-0.0027+
(0.0016)
(0.0019)
(0.0018)
(0.0016)
Female*Avg don. to Repub. by male part. in prac area (log$)
0.0044**
0.0043**
0.0041**
0.0037**
(0.0014)
(0.0016)
(0.0016)
(0.0014)
Avg donations to Democrats by male part. in prac area (log$)
0.0017
0.0023
0.0008
0.0010
(0.0015)
(0.0017)
(0.0017)
(0.0014)
Avg donations to Republicans by male part. in prac area (log$)
-0.0019
-0.0015
-0.0028+
-0.0013
(0.0014)
(0.0015)
(0.0015)
(0.0013)
Female associate
-.0042
-.0026
.0034
-.0013
(.0129)
(.0143)
(.0146)
(.0013)
Female* % liberal male partners in prac. area
Female* % conservative male partners in prac. area
% liberal male partners in prac. area
% conservative male partners in prac. area
Estimation
Fixed effects
Additional controls
23 Legal specialty dummies
50 additional law school dummies
Year dummies
N atty-year obs
R-sq
Log Likelihood
Mean of DV

OLS
Office
Yes
Yes
Yes
Yes
74906
0.02
-25988.1
0.149

OLS
Office
Yes
Yes
No
Yes
60246
0.02
-21145.0
0.151

OLS
Office
Yes
Yes
No
Yes
59206
0.02
-19425.2
0.145

Robust standard errors clustered on offices in parentheses. + p<0.10, * p<0.05, ** p<0.01, *** p<0.001, two-tailed tests
See Table 1 for list of additional controls (firm size, profits, female clients, etc.)
Donations are measured between 1996-2006, years of estimation are 2007-2012.

37

OLS
Office-Year
Yes
Yes
No
N/A
74906
0.00
-22581.1
0.149

(5)
Full sample

.0052
(.0061)
-.0204
(.0149)
.0243
(.0196)
.0231
(.0156)
-.0158
(.0199)
OLS
Office
Yes
Yes
No
N/A
74906
0.02
-26020.7
0.149

(6)
Exclude
acquired
offices
-0.0031+
(0.0016)
0.0044**
(0.0014)
0.0018
(0.0015)
-0.0020
(0.0014)
-.0037
(.0129)

OLS
Office
Yes
Yes
Yes
70520
0.01
-25885.8
0.145

APPENDIX A: Linking Political Donation Data to Martindale Attorney Data


We obtain donation data from Bonicas (2013) Database on Ideology, Money in Politics and
Elections (DIME, data.stanford.edu/dime). See Bonica (2014) for a detailed discussion of these data.
Political donations in state and federal elections of $200 and above must be reported to the Federal
Election Commission (FEC), which then makes these donation data public. Bonica (2014) standardizes
the FEC data. Each donation record provides the date and amount of donation, as well as the donors
name, address, and employer. Information about the candidate or committee receiving the donation,
including name, political party, and office sought is also included. We link donations to unique attorneys
in the Martindale data using donor name, donor employer name, and donor location information in the
DIME data to the same information about attorneys in Martindale data.
First, we extract all donors from the DIME data whose employer corresponds to an organization
in the Martindale data, using exact string matching supplemented by fuzzy string matching using the
Jarowinkler algorithm. We then link this subset of donors to specific attorneys in the Martindale data by
matching their personal names, requiring an exact match on last name and using a fuzzy match on first
name. When a donor matches more than one attorney in the same firm on the basis of name, we break
ties using location information, assigning donations to attorneys who work in the same core based
statistical area (CBSA) indicated by the donor in the DIME data. For a small number of donations
(<1%), we encounter a donor who matches multiple attorneys working for the same firm in the same
CBSA. We exclude these duplicates from the analyses, though results are robust to randomly assigning
them to a matched attorney.
Our matching algorithm identifies at least one donation during the 1996-2006 period for about
50% partners and 30% of associates working the Am Law 200 firms. Bonica, Chilton & Sen (2015), who
also match DIME data to attorneys in the Martindale data, find a comparable 43.5% donation rate across
all ranks, giving us confidence in the quality of our matching procedure.

38

APPENDIX B: Variable descriptions


VARIABLE NAME
Dependent variables
Attorney serves on diversity committee (H1)
Associate is selected for client team (H2)
Associate is promoted (H3)
Associate exits the firm (H4)
- to another law firm
- to a non-law firm destination
Independent variables

VARIES BY DESCRIPTION

SOURCE

Individual-year
Individual-deal
Individual-year
Individual-year
Individual-year
Individual-year

Attorney's name is listed as member of firm's diversity committee


Attorney's name is listed included among lawyers who worked a given merger transaction
Associate appears as a partner in the same firm in the following year, or promotion is announced in press release
Associate is no longer affiliated with the current employer in the following year
Associate is affilaited with a different law firm in the following year
Associate no longer appears in the Martdinale directory in the following year

Leadership Directory Yellow Books


Mergermarket
Martindale Hubbell; American Lawyer
Martindale Hubbell
Martindale Hubbell
Martindale Hubbell

Female

Individual

Is the attorney's first name female?

US Social Security Administrationa

Donations to Democrats by attorney ($)


Donations to Republicans by attorney ($)
Attorney is liberal (>75% don$ to Dem)
Attorney is conservative (>75% don$ to Repub)
Avg donations to Dem by male part. in prac area ($)
Avg donations to Repub by male part. in prac area ($)
Control variables
Donation count (any)

Individual
Individual
Individual
Individual
Individual-year
Individual-year

Sum of all donations ($) to Democratic actors in state and federal elections, 1996-2006
Sum of all donations ($) to Republican actors in state and federal elections, 1996-2006
Attorney makes at least 5 donations, with 75% ($) to Democrats, 1996-2006
Attorney makes at least 5 donations, with 75% ($) to Republicans, 1996-2006
Total don. ($) by male part in same office and prac. area / total male part in same office and prac. area
Total don. ($) by male part in same office and prac. area / total male part in same office and prac. area

DIME
DIME
DIME
DIME
DIME
DIME

Individual

Count of all donations by attorney in state and federal elections, 1996-2006

DIME

Top 18 law school


Minority / Nonwhite
Age
Years since JD
Tenure with firm
Tenure is left-censored
% partners in office from attorney's law school
Peer rating
% associates in prac. area who are female
# male partners in practice area
% partners in practice area who are female
Avg age of male partners in prac area
Firm reports a diversity committee
Firm size (# attorneys, 1000s)
Profit per equity partner ($1M)

Individual
Individual
Individual-year
Individual-year
Individual-year
Individual-year
Individual-year
Individual-year
Individual-year
Individual-year
Individual-year
Individual-year
Firm-year
Firm-year
Firm-year

Did the attorney attend a top 18 law school?c


Attorneys name suggests that s/he is American Black, African, Middle Eastern, Asian, or Hispanic
Year - year of birth
Year - year of law school graduation
Year - first year observed with current firm
0/1, is first year with current firm<=1999?
Part. from atty's law school in office and prac. area / total part in office and prac. area
Three dummies indicating whether peers gave attorney a rating of A, B, or C in current year
Female assoc. in office and prac. area / total assoc. in office and prac. area
# male partners in office and practice area
Female partners in office and prac. area / total part. in office and prac. area
Sum of age of male partners in office and prac. area / total male part. in office and prac. area
Does the firm report any members of a diversity committee?
Count of unique US-based attorneys
Total profits / # equity partners

Martindale; US News
Origins Info
Martindale Hubbell
Martindale Hubbell
Martindale Hubbell
Martindale Hubbell
Martindale Hubbell
Martindale Hubbell
Martindale Hubbell
Martindale Hubbell
Martindale Hubbell
Martindale Hubbell
Leadership Directory Yellow Books
Martindale Hubbell
American Lawyer

% of female leadership among client personnel


Firm acquires another firm
Office is acquired

Firm-year
Firm-year
Office-year

# female top 5 executives among client firms / total top 5 executives among client firmsd
Firm makes an acquisition in the current year
>50% of the partners in an office appear in a different, shared firm in the following year

American Lawyer; Execucomp


American Lawyer
Martindale Hubbell

Data are supplemented with information from www.genderchecker.com


Database on Ideology, Money in Politics and Elections (data.stanford.edu/dime; Bonica, 2014)
c
18 schools have occupied the top 15 of the US News Law School Rankings since the rankings inception: Yale, Harvard, Stanford, Columbia, NYU, Cal, Chicago, Penn, Northwestern, Michigan,
Virginia, Cornell, Duke, Georgetown, Vanderbilt, Texas, UCLA, and Southern Cal.In robustness tests, we also provide dummies for each of the 50 largest law schools in our data.
d
This measure is weighted by the number of matters on which for law firm represents the client firm (e.g. antitrust, litigation, etc.).
b

39

APPENDIX C: Practice Areas: Partner Attorneys


The Martindale data contain self-reported information on each attorneys legal specialty. Attorneys can list multiple
specialties in a given year. These 3,000 unique free-text strings were pared down to list of 26 cleaned legal
specialties using a three step process. First, we matched the strings with an exact match to 215 different areas of
practice provided by Martindale. For the last several years, Martindale has forced attorneys to choose specialties
from this list. This exact match covered about 75% of the records in the data. Second, for those strings that did not
match, a collaborator, who holds a JD from a top five law school and who worked for six years as an attorney before
entering academia, matched the strings by hand to the list provided by Martindale, using his/her expert judgment to
choose the most appropriate match. This matching process covered another 15% of the records in the data, for a
total coverage of 90%. Remaining unmatched strings were coded as Other. In the third and final step, s/he
matched the 215 areas of practice to a list of 25 areas of practice provided by MLAGlobal, a prominent legal
consulting firm 16, again using expert judgment. The table below displays the frequency of each specialty area in the
sample.
(1)
Female

Administrative
Bankruptcy
Civil rights
Corporate
Criminal
Education
Energy
Entertainment
Environmental
Family
General practice
Government
Healthcare
Immigration
Insurance
Intellectual property
International
Labor
Litigation
Military
Native populations
Personal injury
Real estate
Tax
Trusts and estates
Observations

n
72419
72419
72419
72419
72419
72419
72419
72419
72419
72419
72419
72419
72419
72419
72419
72419
72419
72419
72419
72419
72419
72419
72419
72419
72419
72419

Mean
0.094
0.051
0.029
0.426
0.025
0.016
0.026
0.029
0.070
0.018
0.006
0.071
0.072
0.011
0.099
0.133
0.014
0.207
0.436
0.001
0.004
0.019
0.137
0.063
0.057

(2)
Male: Liberal

(3)
Male: Moderate

n
40739
40739
40739
40739
40739
40739
40739
40739
40739
40739
40739
40739
40739
40739
40739
40739
40739
40739
40739
40739
40739
40739
40739
40739
40739
40739

n
244112
244112
244112
244112
244112
244112
244112
244112
244112
244112
244112
244112
244112
244112
244112
244112
244112
244112
244112
244112
244112
244112
244112
244112
244112
244112

Mean
0.126
0.065
0.021
0.534
0.059
0.009
0.037
0.048
0.091
0.008
0.007
0.110
0.058
0.004
0.083
0.145
0.028
0.113
0.481
0.001
0.002
0.022
0.172
0.076
0.042

Mean
0.111
0.068
0.020
0.511
0.036
0.010
0.033
0.030
0.070
0.008
0.007
0.070
0.052
0.005
0.115
0.164
0.021
0.150
0.464
0.001
0.002
0.034
0.167
0.076
0.051

(4)
Male:
Conservative
n
Mean
23635 0.149
23635 0.056
23635 0.020
23635 0.540
23635 0.032
23635 0.014
23635 0.053
23635 0.025
23635 0.084
23635 0.004
23635 0.005
23635 0.114
23635 0.071
23635 0.007
23635 0.076
23635 0.126
23635 0.030
23635 0.155
23635 0.437
23635 0.000
23635 0.003
23635 0.024
23635 0.170
23635 0.103
23635 0.066
23635

Level of analysis is the attorney-year.


Donations take place from 1996-2006, sample covers 2007-2012.
Liberal partners are those who make at least five donations, with at least 75% ($) to Democrats.
Conservatives are those who make at least five donations, with at least 75% ($) to Republicans.
Moderates are the remainder.
This table contains more partners than Table 2, because Table 2 consists only of firms who list a diversity committee in the
Leadership Directory, while this table includes partners of all Am Law 200 firms.

16

http://www.mlaglobal.com/community/thought-leadership/practice-area-summary

40

You might also like