Try and Make Me!:
Why Corporate Diversity Training Fails
Alexandra Kalev, Sociology, Tel Aviv University
akalev@post.tau.ac.il
Frank Dobbin, Sociology, Harvard University
frank_dobbin@harvard.edu
Abstract
Diversity training consumes the lion’s share of the corporate diversity budget yet studies suggest
that it may do little to change attitudes or behaviors. We develop locus-of-motivation theory to
suggest that organizational practices can be structured to signal individual, organizational, or
extra-organizational motivation, and that extra-organizational motivation will elicit resistance
and thereby backfire. Three features of diversity training signal extra-organizational (regulatory)
motivation – making it mandatory (implying a legal requirement), special manager training
(implying an effort to stop illegal treatment), and legal/procedural curriculum. By contrast,
voluntary training signals individual choice, while all-worker training and cultural curriculum
signal organizational motivations. Individual and organizational motivation should bring about
more positive responses and lead to increases in managerial diversity. Yet, we argue, the effects
of regulation are complex. Accountability theory suggests that training effects will be moderated
by evaluation apprehension; training should have positive effects where decision-makers believe
that federal regulators, and diversity managers, may scrutinize their decisions. Predictions are
tested using data on 805 workplaces over more than thirty years. Findings have implications for
theory and practice, shedding light on why inequality persists and providing insights for
policymakers and employers seeking to counter discrimination.
Prejudice reduction efforts date back to the race-relations workshops required by federal
agencies in the late 1960s. Today corporate diversity training is pervasive in the Fortune 500
and common even among small to middling firms. Diversity training is often required in
discrimination suit settlements; proposed by plaintiff attorneys and approved by judges
(Schlanger and Kim 2014). Yet there is little evidence that prejudice-reduction training reduces
bias or discrimination. A survey of 985 studies found weak evidence of short- or long-term
effects of training on prejudice or behavior (Paluck and Green 2009). Field studies show that
while diversity mentoring and taskforces have strong, consistent effects on the representation of
white women and minorities in management, diversity training has weak, inconsistent, effects
(Dobbin, Kalev, and Kelly 2007; Kalev, Dobbin, and Kelly 2006).
We build on a growing body of research that brings together insights from laboratory and
field studies to understand workplace inequality (e.g. Legault et al. 2011; Kaiser et al. 2013;
Castilla and Benard 2010). We begin with self-determination theory, which suggests that people
rebel against efforts to control them (Deci and Ryan 1985, 2002). Thus when subjects asked to
read anti-prejudice materials are encouraged to think they are motivated to reduce prejudice by
external norms, rather than autonomous desire, they show increased prejudice (Legault et al.
2011). We argue that the locus of motivation shapes behavior in important ways, and that
organizational practices can signal that the motive for training is extra-organizational (legal
compliance), organizational (workgroup efficacy and inclusion), or individual (autonomous).
When they signal extra-organizational motivation, we predict, training will backfire and lead to
reductions in managerial diversity. When they signal organizational motivation or individual
voluntarism, trainees will internalize training lessons and managerial diversity will rise.
Employers typically structure training in ways that, we predict, will backfire.
Our theory could be read to suggest that innovations adopted to signal compliance with
federal anti-discrimination regulations will backfire generally, but previous studies show positive
effects of those regulations (Leonard 1989, 1990; Kalev et al. 2006). We argue that regulation
may have a positive effect by activating accountability (Kalev et al. 2006; Kalev 2014; Tetlock
1992; Castilla forthcoming; Rissing and Castilla 2014). When subjects in laboratory studies
believe that they may be asked to account for their decisions, evaluation apprehension (Cottrell
1972) leads them to scrutinize their own actions and censor bias (Barley, Gordon, and Gash
1988; Kruglanski and Freund 1983; Bodenhausen, Kramer, and Susser 1994). We posit that
negative responses to training will be moderated where managers are subject to monitoring by
Department of Labor officials who oversee federal contractors or by internal diversity managers.
We suggest, then, that the effects of regulation are complex; when managers think that
external legal controls drive equity efforts they may rebel and exercise bias, yet when they think
that federal officials or diversity managers may call them to account they may censor their own
biases.
We use data from the Equal Employment Opportunity Commission’s annual
demographic census of private-sector work establishments, combined with a retrospective survey
of characteristics of corporate diversity training programs. We test predictions using a national
sample of 805 work establishments for the period 1971-2003. What effects do new training
programs have on white, black, Hispanic, and Asian men and women in management? It may be
that employers who face legal problems will tie training to compliance, thus we control for
discrimination charges, lawsuits, compliance reviews of federal contractors, and a wide range of
other organizational factors. We pool the data in panel models with fixed effects to assess the
consequences of training programs and federal contractor status. We conduct a series of
robustness tests.
THE ORIGINS OF DIVERSITY TRAINING
Corporate anti-discrimination training was stimulated by civil rights regulations of the
1960s, beginning with John F. Kennedy’s 1961 executive order requiring federal contractors to
take “affirmative action” to prevent discrimination (Lawrence 1965, p. 139). By the end of
1971, the Social Security Administration had put fifty thousand staffers through training
(Robertson 1971). By 1976, 60% of big companies offered equal-opportunity training (Bureau
of National Affairs 1976, p. 9). A 2005 study reported that 65% of large firms offered what had
come to be called diversity training (Esen 2005). In our 2002 survey of medium and large
private-sector employers we find that 40% of employers offer diversity training, making it the
most popular diversity program. The popularity of training has been fueled by the rise of white
women in management, who advocate for training (Dobbin, Kim, and Kalev 2011).
In the 1980s, diversity training acquired a new business rationale (Kelly and Dobbin
1998). Consultants and federal agencies began to argue that the workforce was becoming
majority-minority with the growth of women and racial and ethnic minorities, and that training
could help employees to work together (Johnston and Packer 1987). Some consultant/trainers
still focused on legal compliance, while others came to focus on the business case for inclusion.
Our respondents reported little variation in the method of training (97% involved live
workshops), but substantial variation in whether training was mandatory or voluntary, whether it
was for managers or the whole workforce, and whether the curriculum covered legal compliance
or cultural inclusion.
WHY DOESN’T DIVERSITY TRAINING WORK BETTER?
Next we review evidence that training is generally ineffective, and discuss three
explanations that have been put forward; that training reinforces stereotypes and heightens bias,
that it fosters complacency, and that it makes whites feel excluded.
In their exhaustive review of 985 published and unpublished studies of prejudice
reduction interventions, Paluck and Green (2009) conclude that there is little evidence that
training reduces bias (e.g. Kirkpatrick and Kirkpatrick 2006; Kraiger et al. 1993). In their review
of 31 organizational diversity-training studies that used either pre-test/post-test assessments or a
post-test control group, Kulik and Roberson (2008) identify twenty-seven that document
improved knowledge of, or attitudes toward, diversity. Yet improvements were modest, and
most studies that assessed multiple behavioral and attitudinal indicators found change in only
one or two. None looked at changes beyond one year out. In their review of 39 studies,
Bezrukova, Joshi, and Jehn (2008) identify only five that examine long-term effects, two
showing positive effects, two negative, and one no effect.
Studies of the effects of diversity training on workplace outcomes, beyond attitudes or
self-reported behaviors, are rare. Rynes and Rosen (1995) find that only a third of 765 human
resources experts think that their diversity training has positive effects. A study of federal
agencies found no effect of diversity training on the careers of women or minorities (Kellough
and Naff 2004; Naff and Kellough 2003). Kalev and colleagues (2006) found that corporate
diversity training had mixed, and modest, effects on management diversity.
Why does research suggest that training is ineffective? One possibility is that training
activates bias, rather than giving participants the tools to suppress bias. Field and laboratory
studies suggest that training may have such adverse effects, reinforcing stereotypes and eliciting
backlash from white men (Egan and Bendick 2008; Naff and Kellough 2003; Sidanius et al.
2
2001; MacDonald 1993; Rynes and Rosen 1995; Galinsky and Moskowitz 2000; Kulik, Perry,
and Bourhis 2000). Anand and Winters (2008:361) conclude that trainees often leave “confused,
angry, or with more animosity toward” other groups than they began with. Experienced trainers
report that they often encounter anger and resistance (Kulik et al. 2007).
A second possibility is that training makes managers complacent by inspiring unrealistic
confidence in their employers’ anti-discrimination efforts. Several studies suggest this may be
the case. In the lab, Kaiser and colleagues (2013) find that when whites and men are told that
their employers have pro-diversity measures such as training, they presume that the workplace is
free of bias and react harshly to claims of discrimination. In the same vein, Castilla and Benard
(2010) find that when an employer signals that it is meritocratic, subjects do not censor their own
gender biases. These studies suggest that diversity training may lead decision-makers to lower
scrutiny of the workplace and self-control of bias.
If training fails solely because it activates bias, or breeds complacency, we might expect
training of different sorts to have similar adverse effects. However, if training also fails because
it signals that the locus of motivation is extra-organizational, eliciting resistance, formats that
signal extra-organization motivation should be most likely to backfire.
A third possibility is that training makes white decision-makers feel excluded. In a series
of laboratory studies, Plaut and colleagues (2011) find that as compared to the message of colorblindness, the message of diversity and multiculturalism makes whites feel excluded and makes
them less supportive of diversity. If training fails principally because it activates sentiments of
exclusion among whites, we might expect cultural awareness training to be less effective than
legal training, which emphasizes color- and gender-blindness. However, if training also fails
because it signals extra-organizational motivation, legal training should be more likely to fail.
TRAINING AND THE LOCUS OF MOTIVATION
We draw insights from self-determination theory, which suggests that external control of
behavior can backfire, causing people to think and behave in opposition to the designs of the
controller. This finding has broad implications for the organizational and management
literatures, where the command-and-control view still plays a large role. We develop an
organizational-sociology version of locus-of-motivation theory, which suggests that external
motivation for pursuing a goal is less effective than autonomous motivation (deCharms 1968;
Ryan and Connell 1989; Malhotra, Galletta, and Kirsch 2008). In the workplace, there are
multiple perceived sources of motivation. When the source is organizational, we suggest,
managers may react positively -- organizational motivation may foster internalization and
increase the efficacy of practices.
Self-determination theory suggests that autonomous choice contributes to internalization
(Deci and Ryan 1985, 2002). Whether subjects control their own prejudice is, in part, a function
of whether the desire to control prejudice is self-determined -- voluntary (Devine et al. 2002;
Legault et al. 2007; Plant and Devine 1998). People who are autonomously motivated to control
their own racial prejudice show lower levels of prejudice than people who are motivated by an
external incentive or norm (Amodio, Harmon-Jones, and Devine 2003; Legault, Green-Demers,
and Eadie 2009; Plant, Devine, and Peruche 2010). Plant and Devine (2001) find that whites
resent pressure to control prejudice against blacks (see also Kaiser et al. 2013). Moreover,
instructing subjects to suppress stereotypes backfires, making stereotypes more cognitively
accessible (Galinsky and Moskowitz 2000a), and causing subjects to discriminate (Kulik, Perry,
and Bourhis 2000a). Thus, external efforts to control prejudice may prevent both internalization
3
and realization of the desired effect. These findings are consistent with reactance theory (Brehm
and Brehm 1981), which links defiance to threats to autonomy.
Laboratory studies have focused on external versus internal (autonomous) motivation.
We argue that in the workplace, people make a distinction between goals imposed by the
external environment and those chosen by the organization. In the case of training, it is one
thing to signal that workforce diversity is a goal of distant regulators. It is another thing to signal
that workforce inclusion and diversity are important to the success of the firm. We argue that the
framing of practices such as training can influence whether people perceive motivation to be
autonomous, organizational, or extra-organizational, and that extra-organizational motivation
will be most likely to backfire. This intuition is supported by a lab study finding that when legal
regulations and sanctions are framed as the motive for a diversity program, white MBA students
are more resistant than when the organization’s need to manage diversity is framed as the motive
(Kidder et al. 2004; see also Ely and Thomas 2001).
In the organizational context, the motivation for a new management practice can be
signaled by features of implementation. While early self-determination studies depended on
naturally occurring anti-prejudice motivation, contrasting it norms activated in the lab, Legault,
Gutsell, and Inzlicht (2011) show that autonomous and external motivation can be manipulated
with contextual cues. One implication is that the design of diversity training may influence the
perceived locus of motivation. Thus, for instance, we expect that by making attendance at
diversity training voluntary, employers encourage participants to think motivation is autonomous
(individual) – and autonomous choice is a strong predictor of internalization of norms.
Mandatory Training and Extra-Organizational Motivation
We examine three characteristics that signal the motivation for training. First,
mandatory/voluntary status, we argue, distinguishes extra-organizational from autonomous
motivation. There are three arguments for making training mandatory. First, research suggests
that in mandatory courses, top executives model desired behavior for underlings because
everyone participates (Wiggenhorn 1990; Goldstein 1991; Noe and Ford 1991). Second, HR
executives whose companies mandate diversity training assess it as more effective (Rynes and
Rosen 1995). Third, voluntarism brings in the converted, and people knowledgeable about
diversity, but not the skeptics; as uber-consultant R. Roosevelt Thomas argues, diversity training
is “required, and if you can’t deal with that then we have to ask you to leave” (quoted in Johnson
2008, p. 412; see also Kulik et al. 2007).
But, we suggest, making training mandatory signals that the motive for offering training
is extra-organizational, based in the belief that regulators and judges favor mandatory training.
That belief is well founded; high-profile settlements in discrimination lawsuits, such as those
federal judges approved in cases against both Texaco and Coca-Cola, require mandatory
diversity training (Herman et al. 2003; Williamson et al. 2002). According to our locus-ofmotivation theory, by signaling that the motive is extra-organizational, firms may elicit rebellion
on the part of managers.
Building on self-determination and locus-of-motivation theories, we suggest that by
making participation voluntary employers encourage trainees to perceive their participation as
autonomously determined. This should reduce bias, for Legault and colleagues (2011) show that
perceived motivation for reading anti-prejudice materials can be influenced and that perceived
voluntarism decreases bias. Thus employers who make training voluntary should encourage
internalization and see positive effects on workforce diversity.
4
Overall, 29% of firms in our sample offered manager training – 80% of those courses
were mandatory. Nineteen percent of firms offered all-worker training – 50% of those courses
were mandatory. Another survey showed similar results; 79% of managers-only courses, and
53% of all-worker courses, were mandatory (Esen 2005). Thirteen percent offered both special
manager training and all-worker training.
Hypothesis 1: Mandatory diversity training sessions will lead to reductions in managerial
diversity; voluntary training sessions will lead to increases in managerial diversity.
Manager Training and Extra-Organizational Motivation
The second and third training characteristics distinguish organizational from extraorganizational motivation. The second concerns the target group for training. Employers with
manager training signal external, legal, motivation, by indicating that training is offered to
prevent management decisions that could give rise to litigation. Court-approved discrimination
settlements, indeed, typically specify manager training to prevent discrimination (Herman et al.
2003; Williamson et al. 2002). There is a long history of manager training to prevent illegal
discrimination, beginning in the 1960s and early 1970s at employers such as Western Electric
and the Department of Health, Education, and Welfare (Robertson 1971; Dobbin 2009).
Employers who offer training for the entire workforce, by contrast, send the signal that training
is motivated by management goals of improving inclusion and workgroup efficacy (Ely and
Thomas 2001). We expect manager training to backfire, but all-worker training to encourage
internalization and thereby promote diversity.
Hypothesis 2: Manager training will lead to reductions in managerial diversity; all-worker
training will lead to increases in managerial diversity.
Legal Curriculum and Extra-Organizational Motivation
The third training characteristic concerns curriculum. Legal and procedural curriculum,
focusing on the risk of litigation and practices that can prevent lawsuit, signals extraorganizational motivation. Cultural awareness curriculum signals an organizational motive;
improved communication and collaboration among workers. Kochan and colleagues (2003)
argue that interest in promoting cultural awareness to improve workplace relations rose after the
1987 publication of Workforce 2000 (Johnston and Packer 1987), a report sponsored by
Reagan’s Secretary of Labor that predicted a dramatic rise in the racial, ethnic, and gender
diversity of the workforce. Yet when surveyed in the late 1990s, HR managers overwhelmingly
listed legal protection as the first reason for using diversity training (Jordan 1998). While the
motive for training in most organizations may in fact be legal compliance, trainer/consultants can
signal this with legal curriculum, or downplay it with cultural curriculum.
According to our locus-of-motivation theory, by signaling extra-organizational
motivation, legal curriculum will inhibit internalization and elicit rebellion. This prediction is
consistent with case studies by Robin Ely and David Thomas (2001), who find that diversity
programs operating under the discrimination and fairness paradigm, emphasizing the external
legal case, are less effective than programs operating under the learning and effectiveness
paradigm, emphasizing the business case, or hybrid programs (see also Thomas and Ely 1996;
Anand and Winters 2008:262).
In our survey, we asked employers about the curriculum of training programs for the
most common form of training – mandatory manager training. Among 207 workplaces, 25% had
cultural awareness curriculum, 24% had legal and procedural curriculum, and 52% had mixed
curriculum. We expect that any degree of legal content will signal that the motive for prejudice
aversion is external, and will prove counterproductive.
5
Hypothesis 3: Legal training curriculum will lead to reductions in managerial diversity;
cultural curriculum will lead to increases in diversity.
Accountability and Motivation
Thus far our theory suggests that legal regulation can be counterproductive when
managers perceive extra-organizational motivation for diversity interventions. But
accountability theory suggests that external oversight can have a different effect, eliciting
evaluation apprehension that leads people to scrutinize their own behavior for signs of bias
(Cottrell 1972). We consider the effects of both regulatory and managerial accountability. The
prospect of having to account for one’s decisions leads subjects to pay attention to the
appearance of their own behavior, censoring actions that might be perceived as biased (Tetlock
1992). People censor demographic biases in making judgments (Kruglanski and Freund 1983;
Gordon, Rozelle, and Baxter 1988, 1989; Bodenhausen et al. 1994). Accountability has been
shown to reduce bias in both laboratory (Lerner and Tetlock 1999; Dencker 2008) and field
studies (Hirsh 2009; Castilla 2008; Castilla forthcoming; Rissing and Castilla 2014; Kalev 2014).
Monitoring of managers by diversity experts has been shown to catalyze particular diversity
reforms and increase workforce integration (Kalev et al. 2006; Sturm 2001). Thus while locusof-motivation theory suggests that extra-organizational motivation will activate resistance, and
exacerbate discrimination, we expect that organizational and regulatory monitoring will dampen
discrimination.
Organizational and Regulatory Accountability
We consider two factors that may elicit evaluation apprehension; the presence of a local
diversity manager and the presence of a federal contract, subjecting the employer to Department
of Labor oversight.
From the early 1960s, large firms hired full-time managers to handle compliance with
state equal-opportunity laws and with John F. Kennedy’s 1961 presidential order requiring
federal contractors to practice “affirmative action” in the pursuit of equality. While some argue
that these positions are merely symbolic (Edelman 1992), studies suggest that diversity managers
improve the efficacy of diversity programs by activating accountability (Hirsh and Kmec 2009;
Dencker 2008; Kalev et al. 2006) and protect underrepresented groups in downsizings (Kalev
2014). We expect that the presence of a diversity manager will help to inoculate firms against
negative effects of training programs by eliciting evaluation apprehension.
The federal government has held contractors accountable for preventing employment
discrimination since Kennedy’s 1961 edict. Today the Office of Federal Contract Compliance
Programs (OFCCP) in the Department of Labor is charged with conducting on-site compliance
reviews at federal contractor workplaces, at its discretion (Anderson 1996; Leonard 1984b;
Baron, Mittman, and Newman 1991:1386; Skaggs 2001, 2008). We suggest that federal
regulatory monitoring will positively moderate the effects of training programs. Some have
found that actual compliance reviews have positive effects on black employment (Goldstein and
Smith 1976; Kalev and Dobbin 2006; Leonard 1984c, 1984b), despite evidence that federal
agencies do not review the firms that most need oversight (Leonard 1984a). But accountability
theory suggests that it is not actual scrutiny of behavior that causes people to change; it is
apprehension. We predict that even controlling for actual compliance reviews, and the
interaction of compliance reviews with diversity training, the presence of a federal contract will
elicit evaluation apprehension and positively moderate the effects of training.
Hypothesis 4: Corporate diversity officers and federal contracts will positively moderate the
effects of training programs on managerial diversity.
6
DATA AND METHODS
We examine 805 establishments between 1971 and 2002, conducting a fixed effects
analysis of managerial composition to evaluate the effects of different sorts of diversity training
programs. The dependent variables are the log odds of white, black, Hispanic, and Asian men
and women in management. Our pooled time-series, cross-sectional models estimate the effects
of new training programs on the log odds of each group in management. Dependent variables
are lagged by one year. Fixed effects for establishments are achieved with binary variables for
each workplace, using Stata’s AREG procedure. Fixed effects for year are achieved with binary
variables for each but the first year. We use time-varying controls for organizational, regulatory,
and labor market factors thought to affect managerial diversity, which are discussed below.
Coefficients for controls are reported in appendices. We include post-program-change spells in
the models, so that the coefficients represent the average effects of program changes across
subsequent years. The modal training program is observed for 10 years.
Data
Our data come from three sources. The managerial composition data and select
organizational variables come from the annual EEO-1 reports employers submit to the
government. The data on training and most other organizational practices come from our own
retrospective employer survey. Data on external labor market characteristics come from Bureau
of Labor Statistics surveys.
EEO1 data on managerial composition. Private sector employers with more than 100
workers, and federal contractors with more than 50, are required to file annual EEO-1 reports
detailing the race, ethnicity, and gender of their workers, broken down into 9 occupational
categories. State and local agencies, schools, and colleges file different reports (EEOC n.d.).
There are no better data on workforce composition at the organizational level (see Robinson et
al. 2005). Ronald Edwards and Bliss Cartwright at the EEOC graciously gave us access to these
confidential data through an Intergovernmental Personnel Act (IPA) agreement.
We examine the effects of training on different status groups in management, as the best
measure of whether groups are moving into positions of authority and prestige. While the EEOC
data do not distinguish different levels within management, previous studies show white women
and minorities to be clustered in the lower ranks. In figures 1 and 2 we present trends in the
sample between 1971 and 2002. White men held 81% of management jobs in the average
establishment in 1971 and 61% in 2002. White women rose from 16% to 26%; black women
from 0.4% to 1.8%, black men from 1% to 3.1%; Hispanic men from 0.6% to 2.5%; Hispanic
women from 0.1% to 0.9%; Asian men from 0.5% to 1.8%; and Asian women from 0.1% to
0.7%. These figures understate national gains for underrepresented groups because our sample
excludes groups of employers where gains were substantial; small firms, newly established
firms, and the public sector.
Figures 1 and 2 About Here
To ensure that our estimates are robust to extreme values, we take the log of the odds of each
group being in management. We log the odds rather than the proportion because the conditional
distribution of log odds is closer to normal (Fox 1997:78).1 We interpolate workforce data for
three missing years — 1974, 1976, and 1977. The results are not sensitive to data interpolation.
1
1−
!
!!
!
Where the proportion of managers for a group was 0 or 1, we substituted !! for 0 and
for 1, where N is the total number of managers in the firm (Reskin and McBrier 2000).
7
Survey data on organizational factors. To learn about employer diversity training
programs, we drew a stratified random sample from the 1999 EEO-1 files and administered a
survey. We selected half of that sample from establishments that were in the dataset, which
begins in 1971, since at least 1980, and half from those that were in the files since at least 1992.
Thus the youngest establishments in the sample existed in 1992. We selected 35% of the sample
from establishments with less than 500 employees in 1999, to ensure some variation in size. We
sampled from a representative group of industries: food, chemicals, electronics equipment,
transportation equipment, wholesale trade, retail trade, insurance, business services, and health
services, sampling no more than one establishment per parent firm. We follow establishments
through changes in ownership.
The Princeton Survey Research Center surveyed human resource managers or, in firms
without HR managers, general managers. We built on experience from previous surveys of
human resources practices, asking about the history of practices in the target establishment
(Kalleberg et al. 1996; Kelly 2000; Osterman 2000). In the first contact, we asked for the name
of the person most knowledgeable about the history of human resources programs. We
conducted the interview with that person. The modal respondent had 11 years of tenure.
We asked respondents when their establishments had adopted, and terminated, each of
several dozen personnel practices. At the end of each interview we asked the respondent to
consult records or colleagues to fill in any blanks and scheduled a follow-up call. We completed
833 interviews for a response rate of 67%, which compares favorably with other employer
surveys (Kalleberg et al. 1996; Kelly 2000; Osterman 1994). By matching the survey data and
the EEO-1 data, we created a dataset with establishment-year spells. We exclude 28 cases with
large numbers of missing key variables, or with missing data on training program characteristics,
for a final dataset of 805 establishments and 16,570 establishment-years. We set a minimum of
five years of data per establishment. The maximum is thirty-two years, and the median is
twenty-five years.
In Table 1 we report means and standard deviations for key variables. For missing values
on the dependent variables in scattered years, we interpolate. We impute missing values for
years of adoption of independent variables using OLS regression based on industry,
establishment age, and headquarters status. We do not impute missing values for training
programs or their features.
Table 1 About Here
Data on state unemployment, industry size, and the composition of both industry and
state labor markets come from the Bureau of Labor Statistics. Data on federal contractor status
and proportion of industry establishments with contracts come from EEO-1 reports.
In the survey, we first asked “Has (organization name) ever had a training program about
diversity issues for all workers?” and then asked “Has (organization name) ever had a diversitytraining program for managers?” Figure 3 presents the proportion of firms in the sample with
diversity training for managers, and for all workers, in 2002. Nineteen percent of firms had
manager training, 6% had all-worker training, and 13% had both. We followed up the allworker-training question by asking “Is diversity training mandatory?” and the manager-training
question with “Is this training mandatory for managers?” Based on the pilot, which showed
manager training to be more common than all-worker training, we estimated that cell sizes would
Results are robust to other substitutions, and we include a dummy variable for firms with no
group members in management.
8
be adequate for analysis of curriculum variation for manager training. Thus we followed up the
manager-training questions with “Which of the following topics best describes” what is covered,
allowing respondents to choose one, two, or three answers from “the legal context of
discrimination,” “procedures for fair recruitment and selection”, and “cultural information on
minorities, immigrants, and other groups.” Most respondents who chose “legal context” also
chose “procedures” and so we combined those answers into one, binary, variable.
In Figure 4 we report the characteristics of manager training programs in 2002. Twenty
percent of firms made manager training voluntary. Cell sizes are too small to analyze training
curriculum within those that make training voluntary. But 80% of firms with manager training
make it mandatory; half of those offer mixed curriculum, a quarter offer cultural curriculum, and
a quarter offer legal curriculum.
We have time-varying data on training target groups (managers versus all workers) but
asked about mandatory/voluntary and curriculum only for the survey year because in the pilot,
we found that respondents rarely reported change over time in these characteristics. This may
introduce some error, because employers may have changed the mandate or curriculum. But the
effect of measurement error should be to increase standard errors, and make significant effects
more difficult to discover, and below we report a clear pattern of significant coefficients.
Figure 3 About Here
Figure 4 About Here
Controls for Other Factors
In addition to using fixed effects for establishments to control factors that do not vary
over time, and for year to control environmental shifts that affect all employers, we account for
factors that vary across firms or sectors with controls representing the organizational, labor
market, and regulatory forces shown to matter in previous studies.
Organizational structures. We control for the percent of the workforce in management
jobs, and firm size, because growth in management jobs and firm growth signal that a company
offers promotion opportunities, and increase competition for jobs – research suggests that white
men are favored in situations of expanded competition (Reskin and Roos 1990). Growth also
creates opportunities for increasing management diversity through hiring (Baron et al. 1991).
White women have been shown to benefit from the growth in management positions (Konrad
and Linnehan 1995; Leonard 1990, p. 52). Unionization has been found to slow integration
through seniority provisions that favor incumbent employees for management posts (but see Blau
and Beller 1992; Kelly 2003; Milkman 1985; Leonard 1985). Unionization is coded 1 when the
establishment has a contract. Personnel formalization is thought to advantage women and
minorities (Reskin and McBrier 2000). Formal human resources policies is a count of hiring,
promotion, and discharge guidelines; job descriptions; promotion ladders; performance
evaluations; pay grade systems; and internal job posting. Firms with human resources
departments and legal counsel have been more attentive to issues of equal opportunity (Edelman
and Petterson 1999; Holzer and Neumark 2000). We control for non-union grievance procedures
which provide an avenue for civil-rights complaints (Edelman 1990), and for skill tests for
managers, which are thought to prevent discrimination. Attorneys are thought to be attentive to
civil rights law, and so we control for both in-house attorneys and attorneys on retainer.
Diversity practices. We control for the diversity-management initiatives examined in
previous work (Kalev et al. 2006); special recruitment for women and minority managers, workfamily accommodations, affirmative action plan, diversity manager, diversity
committee/taskforce, diversity ratings for managers, affinity networking programs, diversity
9
mentoring. Work-family policies counts paid maternity leave, paid paternity leave, flextime
policies and top management support for work-family programs as assessed by survey
respondents (Williams 2000).
Legal environment. Department of Labor compliance reviews of federal contractors are
controlled with a binary variable representing past review activity. Discrimination lawsuits and
discrimination charges to the government are also controlled in a binary variable. Reviews,
lawsuits, and charges have been shown to raise diversity at the firm and industry level (Baron et
al. 1991; Kalev and Dobbin 2006; (Skaggs 2001, 2008) Goldstein and Smith 1976; Leonard
1984a; Leonard 1984b; Leonard 1984c). More than a third of establishment-spells had
previously faced at least one lawsuit, more than a third had faced an EEOC charge, and nearly
15% had faced a compliance review (only contractors are subject to compliance reviews).
Top management composition. Firms with diverse top management teams appear to hire
more diverse managers (Elliot and Smith 2004; Kanter 1977; Tomaskovic-Devey 1993). Percent
of top managers who are minorities and who are women are calculated based on the top ten
managers. We asked respondents about the percent at 10 year intervals and interpolated values
for intervening years.
Labor market and economy. Corporate diversity is affected by internal and external
labor pools (Cohen, Broschak, and Haveman 1998; Shenhav and Haberfeld 1992). The diversity
of the establishment’s internal labor pool is measured with two variables based on the EEO-1
reports: the proportion of the status group in non-managerial jobs and in the core job. The core
job is the single biggest job category in the establishment. We include a variable coded 1 when
there are no members of the focal group in management. For the external labor pool, we control
for the representation of each of six groups in the industry labor market (2-digit Standard
Industrial Code) and in the state labor market, from the Current Population Survey. The six
groups are white, black, and Hispanic men and women. We omit Asia men and women because
the CPS data do not cover Asians for the entire period under observation. However, we tested
for the sensitivity of models to the exclusion of labor market data, by excluding data for all
groups, and results were robust.
We use the industry’s proportion of government contractors (based on EEO-1 data) to
measure demand for underrepresented workers in industries subject to affirmative-action
regulations. Industry employment captures changing demand for workers. Growing industries
can offer new job opportunities, yet women and minorities have historically been relegated to
less dynamic sectors (Hodson and Kaufman 1982). We control for state unemployment, because
in recessions, minorities and women are less likely to be hired.
Method
We pool the cross-sectional observations, using time series models (Hicks 1994; Hsiao
1986). To account for stable establishment characteristics we introduce fixed effects (Budig and
England 2001; Western 2002), using the Stata AREG procedure. Fixed effects for year are
achieved with binary variables for each year but the first, 1971. The establishment fixed effects
help us to isolate the effects of changes in causal variables, and to deal with heteroskedasticity
arising from the cross-sectional and temporal aspects of the pooled data.
Because the eight dependent variables are related, their error terms are likely correlated.
Our results are robust to taking into account covariance between the errors
FINDINGS
We present three sets of models, examining the most common combinations of
manager/all-worker training, mandatory/voluntary training, and cultural/hybrid/legal training
10
curriculum. For combinations we omit, estimated cell sizes based on the pilot were too small to
permit analysis (hence we omitted subquestions), or actual cell sizes were too small. The
sequence of models permits us to test all hypotheses.
In the first set of models we compare manager training with all-worker training. We
expect that manager training signals that the impetus is extra-organizational (regulatory
compliance) and that this elicits resistance and reduces managerial diversity. We expect that allworker training signals that the impetus is organizational, and that this will have positive effects
on managerial diversity. In the second set of models, we compare voluntary versus mandatory
versions of manager training. We expect that voluntarism will encourage people to think that
their motivation to participate is autonomous and that mandatory training will encourage them to
think the motivation is extra-organizational, or regulatory. In the third set of models, we
compare legal, cultural, and mixed curriculum for the most common form of training, mandatory
manager training. We expect that any mention of the law will signal extra-organizational
motivation, and will lead training to backfire.
We interact training characteristics in each set of models with federal contractor status
and the presence of a diversity manager to test whether regulatory, and organizational,
accountability positively moderate the effects of diversity training. About fifty percent of
establishment-year spells report federal contracts, and about ten percent report diversity
managers. In each set of models, a significant positive coefficient indicates that a group’s share
of management jobs increases in the years following the new training program. A significant
negative coefficient indicates that the group loses. Significant coefficients consistent with
predictions are bold, and those inconsistent are bold italic.
Table 2 reports changes in management diversity following the adoption of manager
training and all-worker training. Without accountability, manager training shows negative
effects on white and black women and on Asian men and women. The interaction with
government contract produces positive coefficients that eliminate the negative effects on Asian
women and turns the effect on Asian men to positive and significant (.063 p<.05). We use the
LINCOM procedure in STATA to estimate the joint effects of factors included in interactions.
The interaction produces an unexpected negative coefficient for Hispanic women, but the net
effect of manager training plus its interaction is non-significant. A contract turns the effect of
diversity training positive only for Asian Men.
The interaction with diversity manager produces positive effects on Hispanic men and
women and Asian women. For Hispanic men, the net effect of manager training plus its
interaction with diversity manager is non-significant. For Hispanic women, the net effect is a
coefficient of .240 (p<.001). For Asian women, net effect of manager training plus its
interaction with diversity is non-significant. The presence of a diversity manager eliminates
some negative effects, but turns the effect positive only for Hispanic women.
All-worker training shows no effects in the absence of accountability, but the contractor
interaction produces positive coefficients for black and Asian men. For both groups, the
combination of all-worker training plus its interaction with contract is also significant.
Moreover, the combination of these variables for White Men is negative (0.085 p<.05). The
diversity-manager interaction produces no significant effects, but the combination of all-worker
training and its interaction with diversity manager is significant (0.163 p<.05). In support of
Hypothesis 2, manager training appears to elicit backlash. The finding for all-worker training
does not support the hypothesis – all-worker training does not promote diversity. In support of
Hypothesis 4, both regulatory and organizational accountability appear to positively moderate
11
the effects of training, eliminating some negative effects of manager training, and catalyzing
positive effects for both types of training for certain groups.
In Table 3, we examine the effects of mandatory versus voluntary diversity training for
managers. Eighty percent of the 258 manager-training programs were mandatory. In the first
row, we see that mandatory manager training had negative effects on all underrepresented groups
but Hispanic women. Again, accountability moderates those effects. The government contract
interaction produces significant positive effects for black and Hispanic men, as well as for Asian
men and women. LINCOM results show that a government contract eliminates the negative
effect on black men, reduces the effect on black women to -.079 (p<.01), eliminates the negative
effect on Hispanic men, turns the negative effect on Asian men positive (.057, p<.05), and
eliminates the negative effect on Asian women. In firms with government contracts, in other
words, mandatory manager training has one negative effect (black women) and one positive
effect (Asian men). In interaction with diversity manager, mandatory manager training shows
significant positive effects on Hispanic men and women and on Asian women. The net
(combined) effect of mandatory manager training and diversity manager is to increase Hispanic
women in management (.292, p<.001) and reduce Asian men (-.162, p<.01).
Voluntary manager training shows positive effects on black men and Hispanic women.
The interaction with government contract produces an unanticipated negative effect on black
men (and a non-significant combined effect), as well as significant positive effects for Asian men
and women, as well as significant linear combinations of effects that similar in magnitude and
significant to the interactions. The interaction with diversity manager produced no significant
effects. All-worker diversity training shows positive effects for black and Asian men which, we
saw in Table 2, were brought about by the joint presence of all-worker training and a federal
contract.
When we broke all-worker diversity training into mandatory and voluntary, with and
without federal contract and diversity manager, cell sizes became very small. Only 25 firms had
voluntary all-worker training, and only 4 of those had diversity managers. Few coefficients were
significant, and there was no clear pattern. We do not report those results.
In Table 4 we examine variation in curriculum for the most common form of diversity
training; mandatory manager training. The effects of curriculum are striking. In the absence of
accountability, cultural training shows one negative effect, on Asian women. Legal and cultural
curriculum combined show negative effects on all minority groups but Hispanic women. Legal
curriculum by itself shows a positive effect on White men, and negative effects on all other
groups but black men and Hispanic women. Without accountability, it appears that any legal
curriculum in mandatory manager training produces broad backlash.
Regulatory and organizational accountability eliminate a number of negative effects and
produce some positive net effects. First, the interaction between cultural training and federal
contract produces positive coefficients for black and Asian men – the linear combination of
coefficients black men is .213 (p<.001) and the combination for Asian men is .112 (p<.05).
Moreover, the linear combination of cultural curriculum and its interaction with government
contract produces a significant negative effect for white men (-.174, p<.05) and a significant
ositive effect for white women (.280, p<.01). Second, the interaction cultural
curriculum*diversity manager produces two unanticipated effects, positive for white men and
negative for white women, but the linear combinations of these with “cultural curriculum” are
non-significant. The interaction with diversity manager produces positive coefficients for
Hispanic men and women and for Asian women, and significant positive combined effects, at
12
.585 (p<.001), .488 (p<.001), and .546 (p<.001) respectively.
Third, for legal and cultural curriculum, interactions with government contract produce
one unanticipated negative coefficient (Hispanic women) and one positive effect (Asian men).
But combined, legal/cultural curriculum and its interaction with contract show only 2 significant
effects in the LINCOM results, for black men (-.149, p<.001) and black women (-.107, p<.01).
Thus, a government contract eliminates negative effects for Hispanic men and Asian men and
women. Fourth, the interaction legal/cultural curriculum*diversity manager produces positive
effects for Hispanic and Asian women. Combined, legal/cultural curriculum and its interaction
with diversity manager produces negative effects on black women (-.165, p<.05) and Asian men
(-.216, p<.01) and a positive effect on Hispanic Women (.278, p<.001). In short, diversity
managers eliminate negative effects of legal/cultural curriculum for black men, Hispanic men,
and Asian women and create a positive effect for Hispanic women.
Fifth, government contract produces significant positive coefficients in interaction with
legal curriculum for white women, black men, Hispanic men, and Asian men and women. When
we combine the coefficients in LINCOM, only the coefficient for white women remains
significant (-.106, p<.05). Government contract thus eliminates five of the six adverse effects of
legal curriculum, and reduces the sixth. Finally, the interaction legal-curriculum*diversitymanager produces a significant positive coefficient for black men, and the combined coefficients
(LINCOM) suggest that under a diversity manager, legal curriculum has four adverse effects
(increasing white men, and decreasing white women and Asian men and women) rather than the
six we saw for legal curriculum on its own.
It appears that when diversity training emphasizes the law, managers resist the goal of
diversity and actually hire more white men and fewer women and minorities for management
than they would have without the training. By contrast, when training curriculum emphasizes
only cultural awareness, managers are less likely to resist. Both regulatory and organizational
accountability appear to dampen the negative effects of legal curriculum.
In this model, the non-interacted voluntary manager training shows positive effects on
black and Hispanic as well as Asian and women. All-worker diversity training continues to
show positive effects on black and Asian men.
Across the models there are 61 significant coefficients that support our hypotheses. Six
scattered coefficients run counter to hypotheses. None of these six is significant at p<.01, and
the significant interactions do not produce significant effects in their linear combinations with
the main training program effects.
Controls are reported in the appendices. Other diversity measures show considerably
stronger, and more consistent, effects than even the best-designed diversity training program.
The other reforms are also measured with binary variables. It appears that special recruitment
programs, diversity task forces, and mentoring programs do more to promote diversity than
training does. Diversity managers themselves have strong positive effects, net of the effects of
the programs included in the models, as we see in tables 2, 3, and 4.
Robustness Tests
Fixed effects account for unmeasured differences across years and establishments. To
further examine whether firms that adopt specific training programs differ from their peers in
unmeasured ways that affect diversity, we re-ran models separately for each training program
using only firms that at some point adopted the program. The results reported in tables 2, 3, and
4 held up, suggesting that unobserved differences between adopters and non-adopters of
particular training programs are not driving the results.
13
To examine the robustness of the results to within-unit serial correlation, we tested
whether each error is partially dependent on the error of the previous year (AR(1)). We found no
evidence that serial correlation affects the findings.
We considered the possibility that some of the non-effects resulted from including large
and small firms or different regulatory periods in the same analysis. Interactions did not reveal
size-specific or presidential-administration-specific effects.
We also looked into the possibility that an unmeasured change at the firm level, such as a
new CEO, brought about new training programs changes in diversity. For each reform, we
omitted post-adoption years and ran identical models, adding a placebo binary variable T equal
to 1 in the three years before policy adoption (Heckman and Hotz 1989). This approach offers a
stringent test for selection bias. If T shows a significant effect in the same direction as the
program effect, unobserved differences between program adopters and non-adopters may be
responsible for observed policy effects. T is significant for several reform/group combinations
but the overall pattern of our findings is robust.
CONCLUSION
We argue that managerial pursuit of organizational goals is shaped by the perceived
locus of motivation for those goals, which may be extra-organizational (e.g. regulatory),
organizational (maangement initiative), or individual (autonomous choice). Psychologists have
shown that when the perceived locus of motivation for pursuing a goal is individual, subjects are
more likely to internalize the goal than when it is external. People frequently rebel against
perceived external control efforts, as self-determination theorists have found in the lab and jobautonomy theorists have found in the field (Deci and Ryan 1985, 2002; Hodson 1996). We
argue that characteristics of organizational practices, such as diversity training, often signal loci
of motivation. Thus special manager training indicates that the goal is to prevent litigation,
mandatory training signals that the goal is to appease judges, and legal curriculum suggests that
the goal is compliance. These formats, we suggest, will make trainees think the motivation for
training is extra-organizational and resist the message. By contrast, voluntary training will elicit
sentiments of autonomous, individual choice, encouraging internalization. And both all-worker
training and cultural curriculum will signal organizational, not regulatory, motivation, also
encouraging internalization. We thus expand locus-of-motivation theory for the organizational
context, suggesting that employees will internalize goals they perceive as important to their
organization as well as those they perceive as autonomously chosen.
We draw implications from this theory for the pattern of effects of diversity training,
suggesting that mandatory training, manager training, and legal curriculum will lead to
reductions in managerial diversity. Voluntary and all-worker training, and cultural curriculum,
are more likely to promote diversity. The findings support this pattern. Others have suggested
different reasons for the failure of training. First, that training strengthens stereotypes, rather
than giving people tools to fight them (Egan and Bendick 2008; Sidanius, Devereux, and Pratto
2001; Galinsky and Moskowitz 2000b; Kulik, Perry, and Bourhis 2000b). Second, that training
breeds complacency, which causes organizational decision-makers to let down their guard and
practice or permit discrimination (Kaiser et al. 2013; Castilla and Benard 2010). Third, that
multicultural training makes white decision-makers feel excluded, and reduces their commitment
to diversity (Plaut et al. 2011). If stereotype-activation or complacency were the sole
mechanisms causing diversity training to fail, we would expect to find different training formats
and curricula to be equally ineffective. If white sentiments of exclusion based on the message of
multiculturalism were the sole mechanism, we would expect cultural curriculum to be less
14
effective than legal training. Instead, there is a clear pattern of effects across the different
formats and curricula that supports our organizational locus-of-motivation theory.
We hypothesized that accountability would positively moderate the effects of training
programs among employers with federal contracts or full-time diversity managers by eliciting
“evaluation apprehension,” which makes managers sensitive to how their decisions appear to
monitors. Results supported this hypothesis as well. In fact, if it were not for accountability, the
typical training program would have significant negative effects. Mandatory manager training
with legal curriculum accounts for over half of training programs, and without accountability,
such programs have broad adverse effects on managerial diversity. Accountability to federal
regulators and local diversity managers moderates these negative effects in many firms, and in
those firms training is a wash. In our sample, 57% (89) of employers with mandatory manager
training and legal curriculum held federal contracts, and 17% (26) had full-time diversity
officers.
That diversity training appears to do little to promote diversity in the management ranks,
and frequently backfires, is troubling given the role it plays in equal opportunity programs
backed by both the private sector and government. Training is regularly at the top of private
best-diversity-practices lists (Catalyst 1998; Society for Human Resources Management 2004).
Mandatory legal training is also at the top of the Equal Employment Opportunity Commission’s
best-practices list for racial equality, which begins: “Train Human Resources managers and all
employees on EEO laws” (EEOC 2015).
Attorneys and judges have also championed mandatory training in the law and in
procedures to prevent discrimination. High-profile settlements in discrimination cases, designed
by attorneys and approved by judges, frequently allocate the lion’s share of the budget for
remedial action to training. The court-approved 2000 settlement in the $192 million Coca-Cola
race case called for mandatory “Equal Opportunity Training” for managers and, separately,
diversity awareness training for all employees (Herman et al. 2003). Our results suggest that
Coca-Cola would have been well advised to either skip training or, if intent on proceeding with
it, to not offer special training to managers, not focus on equal opportunity laws, and not make
training mandatory.
The court-approved $176 million settlement in the 1997 Texaco race case called for
mandatory diversity training for all employees, and then ongoing voluntary diversity training for
supervisors. In its final report, the court-appointed external task force argued that the ongoing
supervisor training should be mandatory, “ChevronTexaco actively encourages diversity
training; however, those who need it most may not be the most likely to respond to such
encouragement. We therefore reiterate our strong belief that ongoing diversity training for
existing supervisors is critical, and we encourage the Company to provide such training on a
mandatory basis” (Williamson et al. 2002:9). Our results suggest that Texaco should not heed
that advice.
The Coca-Cola and Texaco settlements are far from unique. A study of 502 class-action
consent decrees between 2000 and 2008 found that 89% required diversity training – typically
mandatory training. The other popular remedies were posting the anti-discrimination policy
(89%), writing an anti-discrimination policy (62%), and creating a grievance procedure (38%)
(Hegewisch, Deitch, and Murphy 2011:26; see also Schlanger and Kim 2013). Courts appear to
put a lot of confidence in diversity training. Our results suggest that they should not.
These findings should serve as a wake-up call to proponents of workplace equity because
they show that many training programs have done more harm than good. One estimate put the
15
annual cost of diversity programs at $8 billion (Fay 2003), and our analyses suggest that much of
that money is wasted on training that elicits resistance. Companies can, however, tweak their
training programs and turn the tide. Some studies suggest that employers have noticed that
mandatory training can elicit backlash (Esen 2005; Kulik and Roberson 2008). Still, in our
sample 80% of employers with manager training make it mandatory, and other recent studies
show that 60% to 70% make it mandatory (Esen 2005; The New York Times Company 2007).
We see little evidence of a groundswell of support for making training voluntary, and leading
diversity consultants such as R. Roosevelt Thomas and DiversityInc still insist that it must be
obligatory to be effective (Frankel 2007; Johnson 2008:412)
Some companies have also recognized resistance to training about legal compliance. Bell
and Kravitz (2008, p. 303) find that many employers feel that they “must provide training about
antidiscrimination law” and question whether diversity training is “more likely to have positive
effects if it is separated from the antidiscrimination training temporally and through the use of
different trainers.” Many companies have separated legal compliance and diversity management
functions, and some, like the food service giant Sodexo, have separated diversity from legal
compliance training (Anand and Winters 2008, p. 264). Whether that will quell backlash is an
open question, but our findings would seem to suggest that any legal training is too much, for
firms that combine cultural with legal training see only negative effects.
There is still much debate among practitioners about the most effective training
strategies, and new strategies appear regularly. Some experts have recently advocated educating
managers about implicit bias research (Banaji et al. 1993; Greenwald and Krieger 2006;
Greenwald et al. 2003) as a way to help them understand how cognition shapes our response to
people (Egan and Bendick 2008; Stewart et al. 2008). Google has taken up the challenge by
developing a diversity training program for its 49,000 employees based on implicit association
research (Manjoo 2014). This strategy may make managers aware of their own unconscious
biases. Future research might look into whether this, or other alternatives now being put into
place, leads to actual changes in workforce composition.
Our tort and regulatory systems are rooted in the assumption that legal threat will cause
individuals to pursue societal goals. If legal threat actually impedes internalization of those
goals, as our research suggests it may, then this regulatory strategy may be misguided. The
psychological literature suggests that we may overestimate the importance of homo economicus,
and underestimate the role of social and psychological factors in shaping the behavior of
managers and workers (Heath 1999). Some have discounted this area of research as tantamount
to social engineering, and yet it will be key to understanding the failure of regulatory
interventions to achieve the desired results. Regulators need to pay greater heed to social science
research on the effects of public policies and the organizational reforms they stimulate
(Apfelbaum and Sommers 2013).
16
Table 1: Means and Standard Deviations of Key Variables Used in the Analysis. N=16,570.2
Mean
S.D.
Min
Max
Type
Data
Outcome Variables
Proportion of Managers who are:
White Men
White Women
Black Men
Black Women
Hispanic Men
Hispanic Women
Asian Men
Asian Women
0.702
0.219
0.024
0.013
0.017
0.005
0.012
0.004
0.236
0.213
0.057
0.040
0.050
0.021
0.042
0.018
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
1
1
1
0.667
0.714
0.500
0.851
0.500
Continuous
Continuous
Continuous
Continuous
Continuous
Continuous
Continuous
Continuous
EEO-1
EEO-1
EEO-1
EEO-1
EEO-1
EEO-1
EEO-1
EEO-1
Diversity Training Variables
Manager Training
Mandatory Manager Training
Voluntary Manager Training
Mandatory Manager Training, Cultural
Mandatory Manager Training, Legal/cultural
Mandatory Manager Training, Legal
All-worker Training
0.094
0.075
0.019
0.017
0.042
0.016
0.055
0.292
0.264
0.135
0.130
0.200
0.127
0.228
0
0
0
0
0
0
0
1
1
1
1
1
1
1
Binary
Binary
Binary
Binary
Binary
Binary
Binary
Survey
Survey
Survey
Survey
Survey
Survey
Survey
Diversity Manager
Federal Contract
0.061
0.491
0.239
0.500
0
0
1
1
Binary
Binary
Survey
EEO-1
2
Descriptive statistics for control variables are listed in Appendix Table A.
17
Table 2: Fixed-Effects Estimates of the Log Odds of Groups in Management following
Diversity Training Programs, 1971-2002. Controls Presented in Appendix Table B.
White
Men
Diversity Training
Manager Training
* Federal Contract
* Diversity Manager
All-Worker Training
* Federal Contract
* Diversity Manager
Monitoring
Federal Contract
Diversity Manager
Constant
White
Women
Black
Men
Black Hispanic Hispanic
Women
Men
Women
Asian
Men
Asian
Women
-0.012
-0.071*
0.018
-0.116*** -0.031
0.050
-0.125*** -0.101***
(0.033)
(0.034)
(0.031)
(0.029)
(0.030)
(0.027)
(0.030)
0.015
0.050
-0.015
0.049
0.061
-0.073*
0.188*** 0.126***
(0.043)
(0.044)
(0.040)
(0.037)
(0.038)
(0.034)
(0.038)
(0.034)
0.042
0.033
0.028
2.64E-04 0.125*
0.190*** 0.039
0.118*
(0.067)
(0.068)
(0.063)
(0.058)
(0.060)
(0.054)
(0.060)
(0.053)
-0.003
0.029
-0.036
0.039
-0.017
-0.064
-0.031
-0.010
(0.043)
(0.044)
(0.040)
(0.037)
(0.038)
(0.034)
(0.038)
(0.034)
-0.081
0.013
0.134** -0.038
0.049
0.020
0.131** 0.025
(0.054)
(0.055)
(0.051)
(0.047)
(0.048)
(0.043)
(0.048)
(0.043)
0.091
-0.072
0.040
0.124
-0.018
0.089
-0.027
0.116
(0.077)
(0.078)
(0.072)
(0.066)
(0.068)
(0.062)
(0.068)
(0.061)
0.033
0.010
-0.024
-0.048** -0.076*** -0.041** -0.069*** -0.034*
(0.019)
(0.019)
(0.017)
(0.016)
(0.017)
(0.015)
(0.017)
(0.026)
(0.015)
-0.084** 0.128*** 0.180*** 0.096*** 0.012
0.128*** 0.090** 0.069**
(0.032)
(0.032)
(0.030)
(0.026)
0.141
-0.686
2.130*** 2.930*** 2.714*** 2.925*** 2.923*** 3.216***
(0.346)
(0.357)
(0.325)
(0.028)
(0.303)
(0.028)
(0.309)
(0.279)
(0.028)
(0.310)
(0.025)
(0.275)
R-sq
0.867
0.873
0.819
0.859
0.859
0.881
0.852
0.886
Note: Data shown are coefficients from OLS regression with standard errors in parentheses. The analyses
include establishment and year fixed effects (parameters for 30 binary variables for the years 1972-2001 are not
shown, 1971 is the omitted year and 2002 is included only for calculating the outcome variable). All independent
variables are lagged by one year, excluding proportion of managerial jobs. Number of parameters is 76. N
(organization-year; organizations)= 16,570; 805.
***p<0.001; ** p<0.01; * p<0.05; (two tailed test)
18
Table 3: Fixed-Effects Estimates of the Log Odds of Groups in Management following
Diversity Training Programs, 1971-2002. Controls Presented in Appendix Table C.
White
Men
Diversity Training
Mandatory Manager Training -0.004
* Federal Contract
* Diversity Manager
Voluntary Manager Training
* Federal Contract
* Diversity Manager
All-Worker Training
Constant
R-sq
White
Women
Black
Men
Black Hispanic Hispanic
Women
Men
Women
Asian
Men
Asian
Women
-0.080*
-0.083*
-0.144***-0.068*
0.016
-0.195***-0.154***
(0.035)
(0.036)
(0.033)
(0.030)
(0.031)
(0.028)
(0.031)
-0.007
0.058
0.088*
0.065
0.080*
-0.050
0.252*** 0.121***
(0.041)
(0.042)
(0.039)
(0.036)
(0.037)
(0.033)
(0.037)
0.094
0.007
0.067
0.058
0.143**
0.276*** 0.032
0.236***
(0.055)
(0.056)
(0.052)
(0.048)
(0.049)
(0.044)
(0.049)
(0.043)
0.012
-0.035
0.207*** -0.020
0.037
0.109*
-0.020
-0.013
(0.061)
(0.062)
(0.057)
(0.053)
(0.054)
(0.049)
(0.054)
(0.048)
-0.034
0.040
-0.171*
-0.080
0.106
-0.088
0.189*
0.288***
(0.088)
(0.090)
(0.083)
(0.076)
(0.078)
(0.071)
(0.078)
(0.070)
0.016
-0.085
0.111
0.134
-0.005
0.035
0.120
-0.121
(0.132)
(0.135)
(0.124)
(0.115)
(0.118)
(0.106)
(0.118)
(0.105)
-0.037
0.028
0.064*
0.043
0.015
-0.037
0.053*
0.031
(0.029)
(0.030)
(0.027)
(0.025)
(0.026)
(0.023)
(0.026)
(0.023)
0.138
-0.670
2.171*** 2.932*** 2.728*** 2.940*** 2.944*** 3.207***
(0.347)
(0.357)
(0.325)
(0.303)
(0.309)
(0.279)
(0.310)
(0.275)
0.867
0.873
0.819
0.859
0.859
0.882
0.852
0.886
(0.028)
(0.032)
Note: Data shown are coefficients from OLS regression with standard errors in parentheses. The analyses include
establishment and year fixed effects (parameters for 30 binary variables for the years 1972-2001 are not shown,
1971 is the omitted year and 2002 is included only for calculating the outcome variable). All independent
variables are lagged by one year, excluding proportion of managerial jobs. Number of parameters is 77. N
(organization-year; organizations)= 16,570; 805.
***p<0.001; ** p<0.01; * p<0.05; (two tailed test)
19
Table 4: Fixed-Effects Estimates of the Log Odds of Groups in Management following
Diversity Training Programs, 1971-2002. Controls Presented in Appendix Table D.
White
Men
White
Women
Black
Men
Black Hispanic Hispanic
Women
Men
Women
Asian
Men
Asian
Women
Diversity Training
Voluntary Manager Training -0.007
-0.019
0.127** -0.047
0.092*
0.066
0.091*
0.126***
(0.043)
(0.043)
(0.040)
(0.037)
(0.038)
(0.034)
(0.038)
(0.034)
Mandatory Manager Training
Cultural Curriculum
-0.070
0.073
0.014
-0.038
0.103
-0.024
-0.056
-0.106*
(0.066)
(0.067)
(0.062)
(0.057)
(0.059)
(0.053)
(0.059)
(0.052)
-0.105
0.108
0.200*
0.020
-0.101
-0.066
0.168*
0.024
(0.085)
(0.086)
(0.080)
(0.074)
(0.075)
(0.068)
(0.076)
(0.067)
0.269*
-0.259*
-0.008
0.166
0.481*** 0.512*** 0.178
0.546***
(0.112)
(0.115)
(0.106)
(0.098)
(0.100)
(0.089)
Legal & Cultural Curriculum -0.051
-0.044
(0.044)
(0.045)
0.055
(0.055)
* Federal Contract
* Diversity Manager
* Federal Contract
* Diversity Manager
Legal Curriculum
* Federal Contract
* Diversity Manager
All-Worker Training
Monitoring
Federal Contract
Diversity Manager
Constant
R-sq
(0.090)
(0.100)
-0.142***-0.184***-0.101*
0.033
-0.198***-0.096**
(0.041)
(0.038)
(0.039)
(0.036)
(0.039)
0.004
-0.007
0.077
0.075
-0.097*
0.214*** 0.079
(0.056)
(0.052)
(0.048)
(0.049)
(0.044)
(0.049)
0.040
0.109
0.010
0.018
0.102
0.245*** -0.018
0.172**
(0.069)
(0.071)
(0.065)
(0.060)
(0.062)
(0.056)
(0.062)
(0.055)
0.225** -0.377***-0.092
-0.151*
-0.154*
0.014
-0.379***-0.365***
(0.080)
(0.081)
(0.069)
(0.071)
(0.064)
(0.071)
-0.162
0.271** 0.168*
0.064
0.207*
0.022
0.440*** 0.327***
(0.091)
(0.093)
(0.085)
(0.079)
(0.081)
(0.073)
(0.081)
(0.072)
0.079
-0.008
0.263*
0.099
-0.081
0.127
0.106
0.054
(0.142)
(0.144)
(0.133)
(0.123)
(0.126)
(0.114)
(0.126)
(0.112)
-0.020
0.003
0.076** 0.044
0.013
-0.035
0.052*
0.021
(0.030)
(0.030)
(0.028)
(0.026)
(0.026)
(0.024)
(0.026)
(0.023)
0.030
0.013
-0.021
-0.048** -0.070*** -0.041** -0.061***-0.027
(0.019)
(0.019)
(0.017)
(0.016)
-0.078*
0.122*** 0.179*** 0.107*** 0.015
0.133*** 0.094*** 0.075**
(0.031)
(0.032)
(0.029)
(0.025)
0.165
-0.712*
2.195*** 2.908*** 2.688*** 2.925*** 2.932*** 3.172***
(0.347)
(0.357)
(0.325)
(0.303)
(0.309)
(0.279)
(0.310)
(0.275)
0.868
0.873
0.820
0.859
0.860
0.882
0.852
0.886
(0.075)
(0.027)
(0.017)
(0.028)
(0.015)
(0.017)
(0.028)
(0.035)
(0.043)
(0.063)
(0.015)
(0.025)
Note: Data shown are coefficients from OLS regression with standard errors in parentheses. The analyses include
establishment and year fixed effects (parameters for 30 binary variables for the years 1972-2001 are not shown,
1971 is the omitted year and 2002 is included only for calculating the outcome variable). All independent
variables are lagged by one year, excluding proportion of managerial jobs. Number of parameters is 81. N
(organization-year; organizations)= 16,570; 805.
***p<0.001; ** p<0.01; * p<0.05; (two tailed test)
20
Figure 1: White Men and Women in Management
Figure 2: Minority Men and Women in Management
21
Figure 3
Prevalence of Manager and All-Worker Training
19%
Manager Training
Manager Plus All-Worker
13%
All-Worker Training
62%
6%
22
No Training
Figure 4
Characteristics of Training Programs for Managers
19%
20%
Voluntary
Mandatory, Cultural
20%
Mandatory, Cultural & Legal
Mandatory, Legal
41%
23
Appendix Table A. Means and Standard Deviations of Control Variables: N=16,570.
Organizational Structures
Percent Managers
Establishment size
Union agreement
Formal HR policies a
HR Department
Grievance Procedure
Job Tests for Managers
Legal Department
Attorney on Retainer
0.123
746
0.258
4.151
0.830
0.362
0.090
0.288
0.343
0.088
923
0.437
2.362
0.376
0.481
0.287
0.453
0.475
0.001
12
0
0
0
0
0
0
0
0.789
14195
1
8
1
1
1
1
1
Continuous
Continuous
Binary
Count
Continuous
Binary
Count
Count
Count
EEO-1
EEO-1
Survey
Survey
EEO-1
Survey
Survey
Survey
Survey
Diversity Programs
Special Recruitment, Women/Minorities
Affirmative Action Plan
Diversity Taskforce
Diversity Evaluations for Managers
Networking Program
Mentoring Program
Work-family Supportsb
0.167
0.050
0.069
0.067
0.037
0.940
0.464
0.373
0.218
0.253
0.250
0.188
0.991
0.499
0
0
0
0
0
0
0
1
1
1
1
1
4
1
Binary
Binary
Binary
Binary
Binary
Binary
Count
Survey
Survey
Survey
Survey
Survey
Survey
Survey
Legal Environment
EEOC Charge/Title VII Lawsuit
Compliance Review
0.157
0.447
0.364
0.497
0
0
1
1
Binary
Binary
Survey
Survey
Top Management Compositionc
Percent of Top Managers, Minority
Percent of Top Managers, Women
3.244
16.249
9.549
23.374
0
0
100
100
Labor Market and Economy
Proportion of Non-Managers:
White Men
White Women
Black Men
Black Women
Hispanic Men
Hispanic Women
Asian Men
Asian Women
0.411
0.380
0.052
0.058
0.052
0.034
0.015
0.015
0.253
0.251
0.088
0.098
0.137
0.096
0.043
0.037
0
0
0
0
0
0
0
0
1
1
0.940
0.893
1.724
1.340
1.074
0.953
Continued
24
Continuous Survey
Continuous Survey
Continuous
Continuous
Continuous
Continuous
Continuous
Continuous
Continuous
Continuous
EEO-1
EEO-1
EEO-1
EEO-1
EEO-1
EEO-1
EEO-1
EEO-1
Appendix Table A Continued
Proportion in Core Job:
White Men
White Women
Black Men
Black Women
Hispanic Men
Hispanic Women
Asian Men
Asian Women
0.389
0.384
0.055
0.061
0.045
0.031
0.014
0.016
0.317
0.319
0.106
0.114
0.115
0.081
0.041
0.046
0
0
0
0
0
0
0
0
1
1
0.963
1
1
0.757
0.819
0.560
Continuous
Continuous
Continuous
Continuous
Continuous
Continuous
Continuous
Continuous
EEO-1
EEO-1
EEO-1
EEO-1
EEO-1
EEO-1
EEO-1
EEO-1
No Managers are:
White Men
White Women
Black Men
Black Women
Hispanic Men
Hispanic Women
Asian Men
Asian Women
0.007
0.120
0.542
0.706
0.657
0.813
0.701
0.839
0.081
0.325
0.498
0.456
0.475
0.390
0.458
0.367
0
0
0
0
0
0
0
0
1
1
1
1
1
1
1
1
Binary
Binary
Binary
Binary
Binary
Binary
Binary
Binary
EEO-1
EEO-1
EEO-1
EEO-1
EEO-1
EEO-1
EEO-1
EEO-1
Proportion Industry Labor Force:
White Men
White Women
Black Men
Black Women
Hispanic Men
Hispanic Women
0.445
0.324
0.041
0.042
0.064
0.044
0.153
0.145
0.019
0.025
0.037
0.023
0.145
0.103
0.009
0.004
0.006
0.000
0.742
0.624
0.106
0.119
0.231
0.141
Continuous
Continuous
Continuous
Continuous
Continuous
Continuous
CPS
CPS
CPS
CPS
CPS
CPS
Proportion State Labor Force:
White Men
White Women
Black Men
Black Women
Hispanic Men
Hispanic Women
0.387
0.353
0.043
0.048
0.052
0.037
0.061
0.063
0.030
0.034
0.064
0.046
0.116
0.093
0
-0.004
0.001
0.001
0.595
0.496
0.186
0.201
0.286
0.249
Continuous
Continuous
Continuous
Continuous
Continuous
Continuous
CPS
CPS
CPS
CPS
CPS
CPS
Proportion Contractors in Industry
0.492
0.226
0.061
0.821
Continuous
EEO-1
Industry Employment (millions)
State Unemployment Rate
3.738
6.101
2.821
2.023
0.996
2.000
11.458
18.000
Continuous
Continuous
CPS
BLS
a
Includes adoption of a formal HR department, written hiring, promotion and discharge
guidelines, written job description, written promotion ladder, written performance evaluations, pay
b
Includes paid maternity leave, paid paternity leave, policy allowing flexible work hours and top
management support for work-family balance (support was asked at 10-year intervals and
interpolated).
c
Percents were obtained in 10 years intervals (2002, 1992 and 1982). Values for the years in
between were interpolated using a linear function.
25
Appendix Table B: Coefficients for Control Variables for Table 2
White
Men
White
Women
Black
Men
Black Hispanic Hispanic
Women
Men
Women
Asian
Men
Asian
Women
Organizational Structures
Percent Managers
-0.939*** 0.451*** -3.804*** -4.279*** -3.914*** -4.710*** -4.365*** -4.603***
Establishment Size
Union Agreement
Formal HR Policies
HR Department
Grievance Procedure
Job Tests for Managers
Legal Department
Attorney on Retainer
(0.109)
(0.112)
(0.102)
(0.094)
-0.022
-0.064*** -0.499*** -0.662*** -0.616*** -0.722*** -0.656*** -0.731***
(0.012)
(0.012)
(0.011)
0.004
-0.053
(0.034)
(0.035)
0.007
(0.004)
-0.029
-0.064** -0.056** -0.040*
-0.052** -0.069*** -0.074*** -0.069***
(0.020)
(0.021)
(0.087)
(0.097)
(0.086)
(0.018)
0.015
-0.053** -0.072*** -0.043** -0.051** -0.039** -0.087*** -0.028*
(0.017)
(0.018)
(0.016)
-0.013
-0.063*
-0.123*** -0.076*** -0.080*** -0.124*** -0.052*
-0.109***
(0.026)
(0.027)
(0.025)
(0.011)
(0.010)
(0.011)
(0.009)
-0.107*** -0.063*
0.050
0.028
0.002
-0.024
(0.032)
(0.029)
(0.030)
(0.027)
(0.030)
(0.027)
-0.003
-0.001
-0.003
-0.002
0.001
-0.004
-0.006
(0.004)
(0.004)
(0.004)
(0.004)
(0.003)
(0.004)
(0.003)
(0.019)
(0.010)
(0.097)
(0.018)
(0.015)
(0.016)
(0.016)
(0.014)
(0.018)
(0.016)
(0.016)
(0.014)
(0.023)
(0.023)
(0.021)
(0.023)
(0.021)
-0.117*** 0.177*** 0.045*
-0.027
0.019
0.020
0.032
0.017
(0.024)
(0.024)
(0.022)
(0.021)
(0.021)
(0.019)
(0.021)
(0.019)
0.027
-0.057** 0.007
-0.031
-0.019
-0.053*** -0.039*
-0.042**
(0.018)
(0.018)
(0.016)
(0.016)
(0.015)
(0.016)
(0.014)
(0.017)
Diversity Programs
Special Recruitment for
Women or Minorities
-0.065** 0.090*** 0.097*** 0.122*** 0.035*
(0.020)
Affirmative Action Plan
Diversity Taskforce
Diversity Evaluations for
Managers
Networking Program
Mentoring Program
Work-family Supports
0.042**
0.091*** 0.064***
(0.019)
(0.017)
(0.018)
(0.016)
(0.018)
(0.016)
-0.051** 0.021
0.036*
0.013
0.037*
0.018
0.038**
-0.006
(0.016)
(0.015)
(0.014)
(0.015)
(0.013)
(0.015)
(0.013)
(0.020)
(0.017)
-0.111*** 0.165*** 0.108*** 0.215*** 0.116*** 0.120*** 0.259*** 0.188***
(0.027)
(0.027)
(0.025)
0.051
0.040
(0.026)
(0.027)
(0.023)
(0.024)
(0.021)
(0.024)
(0.021)
-0.079** -0.010
0.051*
0.016
-0.031
-0.005
(0.025)
(0.023)
(0.021)
(0.024)
(0.021)
(0.023)
-0.068** 0.081**
-0.078** 0.002
0.067**
0.018
0.022
0.018
(0.026)
(0.027)
(0.025)
(0.023)
(0.023)
(0.021)
(0.023)
(0.021)
-0.026
0.016
0.055
0.173*** 0.080**
0.184*** 0.165*** 0.178***
(0.033)
(0.034)
(0.031)
(0.029)
(0.027)
(0.030)
(0.030)
(0.026)
-0.051*** 0.039*** 0.004
0.026*** 0.029*** 0.030*** 0.044*** 0.038***
(0.008)
(0.008)
(0.007)
(0.008)
(0.007)
(0.008)
(0.007)
EEOC Charge/Title VII Suit -0.065*** 0.072*** 0.016
0.009
(0.009)
Legal Environment
Compliance Review
Top Mgt. Composition
Proportion Minorities in
Top Management
Proportion Women in Top
Management
0.031*
0.026*
0.014
0.012
(0.015)
(0.015)
(0.014)
(0.013)
(0.013)
(0.012)
(0.013)
(0.011)
-0.047*
0.038
0.073*** 0.014
-0.002
0.023
0.063*** 0.051***
(0.019)
(0.020)
(0.018)
(0.017)
(0.016)
(0.017)
0.097
-0.085
0.911*** 0.483*** -0.177
-0.361** -0.356** -0.432***
-0.145
-0.147
-0.137
-0.130
-0.116
-0.130
-0.115
-0.025
0.120
-0.182** 0.221*** 0.071
0.003
-0.069
0.039
-0.074
-0.075
-0.070
-0.060
-0.066
-0.059
(0.017)
-0.126
-0.064
Continued
26
-0.066
(0.015)
Appendix Table B Continued
Labor Market & Economy
Proportion of Group in
Non-managerial Jobs
Proportion of Group in
Core Job
None from Group in Mgt.
Industry Labor Force (log)
White Men
White Women
Black Men
Black Women
Hispanic Men
Hispanic Women
State Labor Force
White Men
White Women
Black Men
Black Women
Hispanic Men
Hispanic Women
Contractors in Industry
Industry Employment
State Unemployment Rate
Constant
1.309*** 1.116*** 1.349*** 2.095*** 0.717*** 0.389**
(0.086)
(0.090)
0.391
2.122***
(0.119)
(0.131)
(0.384)
-0.066
(0.297)
-0.174*** 0.291
-0.977*** 0.286
0.151
2.876*** 0.808***
(0.045)
(0.047)
(0.122)
(0.156)
(0.341)
(0.232)
(0.156)
(0.192)
(0.152)
(0.235)
-0.536*** -0.565*** -0.630*** -0.680*** -0.550*** -0.551*** -0.608*** -0.591***
(0.072)
(0.018)
(0.013)
0.085
-0.057
(0.091)
(0.093)
-0.079
(0.069)
(0.012)
(0.013)
(0.012)
0.380*** 0.088
0.242**
0.078
0.334*** 0.100
(0.086)
(0.079)
(0.081)
(0.073)
(0.082)
(0.072)
0.117
0.197**
0.129*
0.064
0.054
0.127*
0.149**
(0.071)
(0.065)
(0.060)
(0.061)
(0.055)
(0.061)
(0.054)
-0.104*** 0.092*** 0.018
0.026
0.012
-0.015
0.036
0.012
(0.026)
(0.022)
(0.023)
(0.021)
(0.023)
(0.020)
(0.026)
(0.024)
(0.013)
(0.013)
-0.089*** 0.042
0.093*** 0.011
0.058**
0.024
-0.008
0.044**
(0.021)
(0.020)
(0.019)
(0.017)
(0.019)
(0.017)
(0.022)
(0.019)
-0.057** 0.049*
0.092*** 0.115*** 0.018
0.045**
0.003
0.082***
(0.021)
(0.021)
(0.019)
(0.018)
(0.018)
(0.017)
(0.019)
(0.016)
-0.027
0.006
0.005
0.007
0.003
-0.010
0.011
0.007
(0.017)
(0.018)
(0.016)
(0.015)
(0.015)
(0.014)
(0.015)
(0.014)
0.233
-0.453
-0.242
-0.510
-1.184*** -0.632*
-0.801*
-0.402
(0.392)
(0.400)
(0.369)
(0.341)
(0.349)
(0.351)
(0.311)
-0.320
-0.030
-1.576*** -1.789*** -2.428*** -2.610*** -2.340*** -2.006***
(0.434)
(0.443)
(0.407)
1.658*
-0.879
(0.777)
(0.792)
0.161
(0.674)
(0.377)
(0.386)
(0.315)
(0.349)
(0.388)
(0.345)
-4.516*** -2.962*** -2.504*** -1.360*
-0.827
-0.546
(0.730)
(0.675)
(0.692)
(0.694)
(0.616)
0.002
1.550*
1.860**
-2.065*** -2.374*** -2.056*** -1.545**
(0.688)
(0.634)
(0.587)
(0.600)
(0.542)
-0.338
-0.888
-1.038
-1.005
-0.509
-2.023*** -2.647*** -1.288*
(0.691)
(0.704)
(0.649)
(0.600)
(0.615)
(0.555)
1.448
-1.549
-1.383
-0.046
0.614
2.842*** 2.457**
1.280
(0.837)
(0.854)
(0.786)
(0.727)
(0.745)
(0.674)
(0.748)
(0.663)
(0.625)
(0.603)
(0.617)
(0.535)
(0.548)
0.681*** -0.817*** 0.132
-0.371*** 0.112
-0.209*
-0.190
-0.323**
(0.129)
(0.111)
(0.115)
(0.103)
(0.115)
(0.102)
0.040*** -0.045*** -0.010*
-0.001
-0.015*** -0.003
-0.015*** 0.017***
-(0.004)
-(0.005)
-(0.004)
-(0.005)
-(0.005)
0.010*
-0.018*** 0.004
-0.011** -0.013*** -0.008*
-0.013*** -0.010**
(0.004)
(0.004)
(0.004)
(0.003)
(0.004)
0.141
-0.686
2.130*** 2.930*** 2.714*** 2.925*** 2.923*** 3.216***
(0.346)
(0.357)
(0.325)
(0.131)
(0.121)
-(0.004)
(0.303)
(0.004)
(0.309)
-(0.004)
(0.003)
(0.279)
(0.310)
-(0.004)
(0.003)
(0.275)
R-sq
0.867
0.873
0.819
0.859
0.859
0.881
0.852
0.886
Note: Data shown are coefficients from OLS regression with standard errors in parentheses. The analyses
include establishment and year fixed effects (parameters for 30 binary variables for the years 1972-2001 are not
shown, 1971 is the omitted year and 2002 is included only for calculating the outcome variable). All independent
variables are lagged by one year, excluding proportion of managerial jobs. Number of parameters is 76. N
(organization-year; organizations)= 16,570; 805.
27
Appendix Table C: Coefficients for Control Variables for Table 3
Organizational Structures
Percent Managers
Establishment Size
Union Agreement
Formal HR Policies
HR Department
Grievance Procedure
Job Tests for Managers
Legal Department
Attorney on Retainer
-0.939***0.451*** -3.797***-4.277***-3.913*** -4.704*** -4.363***-4.597***
(0.109)
(0.112)
(0.102)
(0.094)
-0.022
-0.064***-0.499***-0.662***-0.616*** -0.722*** -0.656***-0.731***
(0.012)
(0.012)
(0.011)
0.005
-0.054
(0.034)
(0.035)
0.007
(0.004)
-0.029
-0.064** -0.057** -0.041*
-0.052** -0.069*** -0.075***-0.070***
(0.020)
(0.021)
(0.087)
(0.097)
(0.086)
(0.018)
0.015
-0.053** -0.072***-0.042** -0.052*** -0.039** -0.087***-0.030*
(0.017)
(0.018)
(0.016)
-0.012
-0.063*
-0.125***-0.076***-0.082*** -0.124*** -0.056*
-0.114***
(0.026)
(0.027)
(0.025)
(0.011)
(0.010)
(0.011)
(0.009)
-0.111*** -0.064*
0.046
0.026
-0.003
-0.032
(0.032)
(0.029)
(0.030)
(0.027)
(0.030)
(0.027)
-0.003
####### -0.003
-0.001
0.001
-0.004
-0.005
(0.004)
(0.004)
(0.004)
(0.003)
(0.004)
(0.003)
(0.019)
(0.010)
(0.097)
(0.004)
(0.018)
(0.015)
(0.016)
(0.016)
(0.014)
(0.018)
(0.016)
(0.016)
(0.014)
(0.023)
(0.023)
(0.021)
(0.023)
(0.021)
-0.116*** 0.177*** 0.045*
-0.027
0.017
0.021
0.030
0.014
(0.024)
(0.024)
(0.022)
(0.021)
(0.021)
(0.019)
(0.021)
(0.019)
0.028
-0.057** 0.006
-0.030
-0.019
-0.053*** -0.040*
-0.042**
(0.018)
(0.018)
(0.016)
(0.016)
(0.015)
(0.016)
(0.014)
(0.017)
Diversity Programs
Special Recruitment for
Women or Minorities
-0.066** 0.091*** 0.100*** 0.122*** 0.037*
(0.020)
Affirmative Action Plan
Diversity Taskforce
0.042**
0.095*** 0.067***
(0.019)
(0.017)
(0.018)
(0.016)
(0.018)
-0.051** 0.021
0.036*
0.014
0.036*
0.017
0.038** -0.008
(0.016)
(0.015)
(0.014)
(0.015)
(0.013)
(0.015)
(0.020)
(0.017)
(0.027)
(0.025)
Diversity Evaluations for
Managers
0.052
0.041
(0.026)
(0.027)
Networking Program
-0.066*
(0.026)
Work-family Supports
Legal Environment
EEOC Charge/Title VII Suit
Compliance Review
(0.013)
-0.111*** 0.166*** 0.113*** 0.216*** 0.121*** 0.124*** 0.264*** 0.197***
(0.027)
Mentoring Program
(0.016)
(0.023)
(0.024)
(0.021)
(0.024)
(0.021)
-0.078** -0.008
0.053*
0.017
-0.030
-0.001
(0.025)
(0.023)
(0.021)
(0.024)
(0.021)
0.080** -0.075** 0.006
0.067**
0.019
0.024
0.021
(0.027)
(0.025)
(0.023)
(0.023)
(0.021)
(0.023)
(0.021)
-0.024
0.016
0.057
0.174*** 0.081**
0.187*** 0.164*** 0.182***
(0.033)
(0.034)
(0.031)
(0.029)
(0.027)
(0.023)
(0.030)
(0.030)
(0.026)
-0.052***0.039*** 0.004
0.026*** 0.029*** 0.030*** 0.044*** 0.038***
(0.008)
(0.008)
(0.007)
(0.008)
(0.007)
(0.008)
(0.007)
-0.064***0.073*** 0.016
0.009
0.031*
0.026*
0.014
0.012
(0.015)
(0.015)
(0.014)
(0.013)
(0.013)
(0.012)
(0.013)
(0.011)
-0.048*
0.038
0.074*** 0.013
-0.003
0.023
0.064*** 0.050**
(0.019)
(0.020)
(0.018)
(0.017)
(0.016)
(0.017)
0.101
-0.075
0.937*** 0.492*** -0.161
-0.345** -0.340** -0.406***
-0.145
-0.148
-0.137
-0.130
-0.117
-0.130
-0.115
-0.026
0.118
-0.195** 0.213*** 0.064
0.003
-0.082
0.026
-0.074
-0.076
-0.070
-0.060
-0.066
-0.059
(0.009)
(0.017)
(0.015)
Top Mgt. Composition
Proportion Minorities in
Top Management
Proportion Women in Top
Management
-0.126
-0.064
Continued
28
-0.066
Appendix Table C Continued
Labor Market & Economy
Proportion of Group in
Non-managerial Jobs
Proportion of Group in
Core Job
None from Group in Mgt.
1.307*** 1.116*** 1.349*** 2.100*** 0.724*** 0.391**
(0.086)
(0.090)
0.365
2.027***
(0.119)
(0.131)
(0.384)
(0.297)
-0.065
(0.045)
-0.174***0.295
-0.977***0.282
0.160
2.850*** 0.894***
(0.047)
(0.122)
(0.156)
(0.341)
(0.232)
(0.156)
(0.192)
(0.152)
(0.235)
-0.539***-0.565***-0.632***-0.680***-0.551*** -0.550*** -0.610***-0.589***
(0.072)
(0.018)
(0.013)
0.084
-0.057
(0.091)
(0.093)
-0.080
0.118
(0.069)
(0.071)
(0.013)
(0.012)
(0.013)
(0.012)
(0.013)
0.387*** 0.089
0.245**
0.082
0.340*** 0.105
(0.086)
(0.081)
(0.073)
(0.082)
(0.072)
0.195** 0.126*
0.063
0.055
0.123*
0.146**
(0.065)
(0.060)
(0.061)
(0.055)
(0.061)
(0.054)
-0.105***0.092*** 0.021
0.027
0.013
-0.014
0.038
0.013
(0.026)
(0.022)
(0.023)
(0.021)
(0.023)
(0.020)
Industry Labor Force (log)
White Men
White Women
Black Men
Black Women
Hispanic Men
Hispanic Women
State Labor Force
White Men
White Women
Black Men
Black Women
Hispanic Men
Hispanic Women
Contractors in Industry
Industry Employment
State Unemployment Rate
Constant
R-sq
(0.026)
(0.024)
(0.079)
-0.089***0.042
0.095*** 0.011
0.059**
0.024
-0.007
0.045**
(0.021)
(0.020)
(0.019)
(0.017)
(0.019)
(0.017)
(0.022)
(0.019)
-0.057** 0.049*
0.091*** 0.114*** 0.017
0.046**
0.002
0.080***
(0.021)
(0.021)
(0.019)
(0.018)
(0.018)
(0.017)
(0.019)
(0.016)
-0.027
0.006
0.005
0.007
0.003
-0.010
0.011
0.008
(0.017)
(0.018)
(0.016)
(0.015)
(0.015)
(0.014)
(0.015)
(0.014)
0.227
-0.468
-0.276
-0.522
-1.194*** -0.648*
-0.814*
-0.402
(0.393)
(0.400)
(0.369)
(0.341)
(0.349)
(0.351)
(0.311)
-0.309
-0.053
-1.604***-1.778***-2.447*** -2.616*** -2.365***-2.016***
(0.434)
(0.443)
(0.407)
1.668*
-0.873
(0.777)
(0.793)
0.185
(0.674)
(0.376)
(0.386)
(0.316)
(0.387)
(0.344)
-4.481***-2.941***-2.508*** -1.347*
-0.828
-0.577
(0.730)
(0.676)
(0.694)
(0.615)
-0.036
1.463*
1.867** -2.110*** -2.399*** -2.126***-1.576**
(0.688)
(0.633)
(0.587)
(0.600)
(0.542)
-0.316
-0.888
-1.084
-0.992
-0.523
-2.046*** -2.679***-1.294*
(0.691)
(0.704)
(0.648)
(0.600)
(0.615)
(0.555)
1.428
-1.602
-1.358
-0.039
0.584
2.848*** 2.481*** 1.256
(0.836)
(0.853)
(0.784)
(0.726)
(0.744)
(0.673)
(0.747)
(0.662)
(0.692)
(0.349)
(0.625)
(0.603)
(0.617)
(0.534)
(0.547)
0.675*** -0.812***0.148
-0.373***0.124
-0.203*
-0.172
-0.303**
(0.129)
(0.111)
(0.115)
(0.103)
(0.114)
(0.102)
0.040*** -0.045***-0.011*
-0.001
-0.015*** -0.003
-0.015***0.017***
-(0.005)
-(0.005)
-(0.004)
-(0.005)
-(0.005)
0.010*
-0.018***0.004
-0.011** -0.013*** -0.008*
-0.013***-0.010**
(0.004)
(0.004)
(0.004)
(0.003)
(0.004)
0.138
-0.670
2.171*** 2.932*** 2.728*** 2.940*** 2.944*** 3.207***
(0.347)
(0.357)
(0.325)
(0.303)
(0.309)
(0.279)
(0.310)
(0.275)
0.867
0.873
0.819
0.859
0.859
0.882
0.852
0.886
(0.131)
(0.121)
-(0.005)
(0.004)
-(0.004)
(0.003)
-(0.004)
(0.003)
Note: Data shown are coefficients from OLS regression with standard errors in parentheses. The analyses include
establishment and year fixed effects (parameters for 30 binary variables for the years 1972-2001 are not shown,
1971 is the omitted year and 2002 is included only for calculating the outcome variable). All independent
variables are lagged by one year, excluding proportion of managerial jobs. Number of parameters is 77. N
(organization-year; organizations)= 16,570; 805.
** p<0.01; * p<0.05; (two tailed test)
29
Appendix Table D: Coefficients for Control Variables for Table 4
White
Men
White
Women
Black
Men
Black Hispanic Hispanic
Women
Men
Women
Asian
Men
Asian
Women
Organizational Structures
Percent Managers
Establishment Size
Union Agreement
Formal HR Policies
HR Department
Grievance Procedure
Job Tests for Managers
Legal Department
Attorney on Retainer
-0.942***0.451*** -3.791***-4.278***-3.924*** -4.706*** -4.362*** -4.605***
(0.109)
(0.112)
(0.102)
(0.094)
-0.023
-0.063***-0.501***-0.663***-0.616*** -0.722*** -0.655*** -0.729***
(0.012)
(0.012)
(0.011)
0.007
-0.058
(0.034)
(0.035)
0.008
(0.004)
-0.031
(0.020)
(0.087)
(0.097)
(0.086)
0.014
-0.051** -0.076***-0.045** -0.054*** -0.040**
-0.090*** -0.031*
(0.017)
(0.018)
(0.016)
(0.016)
-0.014
-0.060*
-0.125***-0.074** -0.080*** -0.125*** -0.053*
-0.111***
(0.026)
(0.027)
(0.025)
(0.011)
(0.010)
(0.011)
-0.111*** -0.063*
0.050
0.028
-5.40E-05 -0.026
(0.032)
(0.029)
(0.030)
(0.027)
(0.030)
(0.027)
-0.004
-0.001
-0.003
-0.002
0.001
-0.004
-0.005
(0.004)
(0.004)
(0.004)
(0.004)
(0.003)
(0.004)
(0.003)
-0.061** -0.053** -0.038*
-0.051**
-0.070*** -0.073*** -0.071***
(0.021)
(0.018)
(0.016)
(0.019)
(0.010)
(0.097)
(0.018)
(0.015)
(0.016)
(0.014)
(0.018)
(0.009)
(0.016)
(0.014)
(0.023)
(0.023)
(0.021)
(0.023)
(0.021)
-0.116*** 0.178*** 0.046*
-0.027
0.021
0.021
0.033
0.017
(0.024)
(0.024)
(0.022)
(0.021)
(0.021)
(0.019)
(0.021)
(0.019)
0.026
-0.055** 0.010
-0.029
-0.019
-0.054*** -0.039*
-0.043**
(0.018)
(0.018)
(0.016)
(0.016)
(0.015)
(0.016)
(0.014)
(0.017)
Diversity Programs
Special Recruitment for
Women or Minorities
-0.068***0.093*** 0.101*** 0.120*** 0.033
(0.020)
Affirmative Action Plan
Diversity Taskforce
0.041*
0.090*** 0.063***
(0.019)
(0.017)
(0.018)
(0.016)
(0.018)
(0.016)
-0.054***0.025
0.035*
0.013
0.038**
0.016
0.040**
-0.004
(0.016)
(0.015)
(0.014)
(0.015)
(0.013)
(0.015)
(0.013)
(0.021)
(0.017)
-0.113*** 0.165*** 0.105*** 0.210*** 0.109*** 0.123*** 0.263*** 0.196***
(0.027)
(0.027)
(0.025)
Diversity Evaluations for
Managers
0.054*
0.035
(0.027)
(0.027)
Networking Program
-0.064*
(0.026)
Mentoring Program
Work-family Supports
(0.023)
(0.024)
(0.022)
(0.024)
(0.021)
-0.074** -0.012
0.050*
0.022
-0.033
-0.003
(0.025)
(0.024)
(0.021)
(0.024)
(0.021)
0.077** -0.073** 0.006
0.070**
0.020
0.027
0.023
(0.027)
(0.025)
(0.023)
(0.023)
(0.021)
(0.023)
(0.021)
-0.007
-0.008
0.048
0.177*** 0.088**
0.194*** 0.168*** 0.187***
(0.034)
(0.034)
(0.031)
(0.029)
(0.027)
(0.023)
(0.030)
(0.030)
(0.027)
-0.052***0.039*** 0.007
0.027*** 0.029*** 0.030*** 0.044*** 0.038***
(0.008)
(0.008)
(0.007)
(0.008)
(0.007)
(0.008)
(0.007)
EEOC Charge/Title VII Suit -0.065***0.074*** 0.015
0.009
(0.009)
Legal Environment
Compliance Review
0.031*
0.026*
0.015
0.014
(0.014)
(0.015)
(0.014)
(0.013)
(0.013)
(0.012)
(0.013)
(0.011)
-0.048*
0.038
0.069*** 0.012
-0.001
0.022
0.068*** 0.056***
(0.019)
(0.020)
(0.018)
(0.017)
(0.016)
(0.017)
(0.015)
0.099
-0.079
0.915*** 0.499*** -0.139
-0.341**
-0.312*
-0.373**
-0.145
-0.148
-0.137
-0.117
-0.130
-0.114
-0.057
0.158*
-0.185** 0.221*** 0.076
-0.001
-0.058
0.046
-0.075
-0.076
-0.070
-0.060
-0.067
-0.059
(0.017)
Top Mgt. Composition
Proportion Minorities in
Top Management
Proportion Women in Top
Management
-0.126
-0.065
Continued
30
-0.130
-0.067
Appendix Table D Continued
Labor Market & Economy
Proportion of Group in
Non-managerial Jobs
Proportion of Group in
Core Job
None from Group in Mgt.
1.300*** 1.114*** 1.316*** 2.104*** 0.712*** 0.375**
(0.086)
(0.090)
0.340
1.987***
(0.119)
(0.131)
(0.384)
-0.061
(0.297)
-0.169***0.310*
-0.974***0.281
0.179
2.874*** 0.875***
(0.045)
(0.046)
(0.122)
(0.156)
(0.341)
(0.232)
(0.156)
(0.192)
(0.152)
(0.235)
-0.538***-0.565***-0.632***-0.681***-0.552*** -0.549*** -0.608*** -0.589***
(0.072)
(0.018)
(0.013)
0.078
-0.051
(0.091)
(0.093)
-0.079
0.115
(0.069)
(0.071)
(0.013)
(0.012)
(0.013)
(0.012)
(0.013)
0.397*** 0.083
0.237**
0.087
0.339*** 0.104
(0.086)
(0.081)
(0.073)
(0.082)
(0.072)
0.205** 0.129*
0.064
0.058
0.125*
0.143**
(0.065)
(0.060)
(0.061)
(0.055)
(0.061)
(0.054)
-0.106***0.092*** 0.022
0.026
0.011
-0.014
0.037
0.012
(0.026)
(0.022)
(0.023)
(0.021)
(0.023)
(0.020)
Industry Labor Force (log)
White Men
White Women
Black Men
Black Women
Hispanic Men
Hispanic Women
State Labor Force
White Men
White Women
Black Men
Black Women
Hispanic Men
Hispanic Women
Contractors in Industry
Industry Employment
State Unemployment Rate
Constant
R-sq
(0.026)
(0.024)
(0.080)
-0.087***0.039
0.093*** 0.010
0.056**
0.023
-0.009
0.042*
(0.021)
(0.020)
(0.019)
(0.017)
(0.019)
(0.017)
(0.022)
(0.019)
-0.055** 0.045*
0.091*** 0.114*** 0.016
0.046**
0.002
0.079***
(0.021)
(0.021)
(0.019)
(0.018)
(0.018)
(0.017)
(0.019)
(0.016)
-0.028
0.007
0.005
0.007
0.002
-0.011
0.011
0.007
(0.017)
(0.018)
(0.016)
(0.015)
(0.015)
(0.014)
(0.015)
(0.014)
0.224
-0.463
-0.295
-0.516
-1.192*** -0.628*
-0.847*
-0.406
(0.392)
(0.400)
(0.368)
(0.341)
(0.349)
(0.351)
(0.310)
-0.356
0.010
-1.578***-1.757***-2.428*** -2.611*** -2.351*** -2.011***
(0.434)
(0.442)
(0.407)
1.589*
-0.783
(0.778)
(0.793)
0.170
(0.675)
(0.387)
(0.344)
-4.472***-2.942***-2.566*** -1.416*
-0.842
-0.611
(0.730)
(0.676)
(0.694)
(0.615)
-0.025
1.475*
1.919** -2.043*** -2.363*** -2.079*** -1.521**
(0.688)
(0.633)
(0.587)
(0.600)
(0.542)
-0.354
-0.838
-1.117
-0.990
-0.502
-2.054*** -2.691*** -1.263*
(0.691)
(0.704)
(0.648)
(0.600)
(0.615)
(0.555)
1.293
-1.411
-1.210
0.031
0.671
2.864*** 2.585*** 1.318*
(0.836)
(0.853)
(0.784)
(0.726)
(0.744)
(0.673)
(0.747)
(0.662)
0.695*** -0.838***0.134
-0.376***0.122
-0.198
-0.178
-0.306**
(0.129)
(0.112)
(0.115)
(0.104)
(0.115)
(0.102)
0.040*** -0.046***-0.010*
-0.001
-0.016*** -0.003
-0.016*** 0.015***
-(0.005)
-(0.005)
-(0.004)
-(0.005)
-(0.005)
0.010*
-0.018***0.004
-0.011** -0.013*** -0.008*
-0.013*** -0.010**
(0.004)
(0.004)
(0.004)
(0.003)
(0.004)
0.165
-0.712*
2.195*** 2.908*** 2.688*** 2.925*** 2.932*** 3.172***
(0.347)
(0.357)
(0.325)
(0.303)
(0.309)
(0.279)
(0.310)
(0.275)
0.868
0.873
0.820
0.859
0.860
0.882
0.852
0.886
(0.131)
(0.121)
-(0.005)
(0.376)
(0.386)
(0.315)
(0.692)
(0.004)
(0.349)
(0.625)
-(0.004)
(0.003)
(0.603)
(0.617)
(0.534)
(0.547)
-(0.004)
(0.003)
Note: Data shown are coefficients from OLS regression with standard errors in parentheses. The analyses include
establishment and year fixed effects (parameters for 30 binary variables for the years 1972-2001 are not shown,
1971 is the omitted year and 2002 is included only for calculating the outcome variable). All independent variables
are lagged by one year, excluding proportion of managerial jobs. Number of parameters is 81. N (organization-year;
organizations)= 16,570; 805.
** p<0.01; * p<0.05; (two tailed test)
31
References
Amodio, D. M., E. Harmon-Jones, and P. G. Devine. 2003. “Individual Differences in the
Activation and Control of Affective Race Bias as Assessed by Startle Eyeblink
Responses and Self-Report.” Journal of Personality and Social Psychology 84:738–53.
Anand, Rohini and Mary-Frances Winters. 2008. “A Retrospective View of Corporate Diversity
Training From 1964 to the Present.” Academy of Management Learning & Education
7:356–72.
Anderson, Bernard E. 1996. “The Ebb and Flow of Enforcing Executive Order 11246.” The
American Economic Review 86:298–301.
Barley, Stephen R., Meyer Gordon, and Debra Gash. 1988. “Cultures of Culture: Academics,
Practitioners and the Pragmatics of Normative Control.” Administrative Science
Quarterly 33:24–61.
Baron, James N., Brian S. Mittman, and Andrew E. Newman. 1991. “Targets of Opportunity:
Organizational and Environmental Determinants of Gender Integration within the
California Civil Services, 1976-1985.” American Journal of Sociology 96:1362–1401.
Blau, Francine D. and Andrea Beller. 1992. “Black White Earnings over the 1970s and 1980s:
Gender Differences in Trends.” The Review of Economics and Statistics 74:276–86.
Brehm, S. S. and J. W. Brehm. 1981. Psychological Reactance: A Theory of Freedom and
Control. New York, NY: Academic Press.
Castilla, Emilio J. 2008. “Gender, Race, and Meritocracy in Organizational Careers.” American
Journal of Sociology 113:1479–1526.
Castilla, Emilio J. and Stephen Benard. 2010. “The Paradox of Meritocracy in Organizations.”
Administrative Science Quarterly 55(4):543–76.
Cohen, Lisa E., Joseph P. Broschak, and Heather A. Haveman. 1998. “And Then There Were
More? The Effect of Organizational Sex Composition on the Hiring and Promoting of
Managers.” American Sociological Review 63:711–27.
Cottrell, Nicholas. 1972. “Social Facilitation.” Pp. 185–236 in Experimental Social Psychology,
edited by C. McClintock. New York: Holt, Rinehart & Winston.
deCharms, Richard. 1968. Personal Causation: The Internal Affective Determinants of Behavior.
Hillsdale, New Jersey: Erlbaum.
Deci, Edward L. and Richard M. Ryan. 1985. Intrinsic Motivation and Self-Determination in
Human Behavior. New York, NY: Plenum Press.
32
Deci, Edward L. and Richard M. Ryan. 2002. Handbook of Self-Determination Research.
Rochester, NY: University of Rochester Press.
Dencker, John C. 2008. “Corporate Restructuring and Sex Differences in Managerial
Promotion.” American Sociological Review 73:455–76.
Devine, P. G., E. A. Plant, D. M. Amodio, E. Harmon-Jones, and S. L. Vance. 2002. “The
Regulation of Explicit and Implicit Race Bias: The Role of Motivations to Respond
Without Prejudice.” Journal of Personality and Social Psychology 82:835–48.
Dobbin, Frank. 2009. Inventing Equal Opportunity. Princeton, NJ: Princeton University Press.
Dobbin, Frank, Alexandra Kalev, and Erin Kelly. 2007. “Diversity Management in Corporate
America.” Contexts 6:21–28.
Dobbin, Frank, Soohan Kim, and Alexandra Kalev. 2011. “You Can’t Always Get What You
Need: Why Diverse Firms Adopt Diversity Programs.” American Sociological Review
76:386–411.
Edelman, Lauren B. 1992. “Legal Ambiguity and Symbolic Structures: Organizational
Mediation of Civil Rights Law.” American Journal of Sociology 97:1531–76.
Edelman, Lauren B. and Stephen M. Petterson. 1999. “Symbols and Substance in Organizations
Response to Civil Rights Law.” Research in Social Stratification and Mobility 17:107–
35.
EEOC. 2004. “Instructions for Standard Form 100 (EEO-1).” Retrieved
(http://www.eeoc.gov/stats/jobpat/e1instruct.html).
EEOC. 2015. “Best Practices for Employers.” Retrieved March 4, 2015
(http://www.eeoc.gov/eeoc/initiatives/e-race/bestpractices-employers.cfm).
Egan, Mary Lou and Marc Jr. Bendick. 2008. “Combining Multicultural Management and
Diversity Into One Course on Cultural Competence.” Academy of Management Learning
& Education 7:387–93.
Ely, Robin J. and David A. Thomas. 2001. “Cultural Diversity at Work: The Effects of Diversity
Perspectives on Work Group Processes and Outcomes.” Administrative Science
Quarterly 46(2):229–73.
Esen, Evren. 2005. Workplace Diversity Practices Survey Report. Alexandria, VA: Society for
Human Resource Management.
Fox, John. 1997. Applied Regression Analysis, Linear Models, and Related Methods. Thousand
Oaks, CA: Sage.
33
Galinsky, Adam D. and Gordon B. Moskowitz. 2000a. “Perspective-Taking: Decreasing
Stereotype Expression, Stereotype Accessibility, and in-Group Favoritism.” Journal of
Personality and Social Psychology 78(4):708–24.
Galinsky, Adam D. and Gordon B. Moskowitz. 2000b. “Perspective Taking: Decreasing
Stereotype Expression Stereotype Accessibility, and In-Group Favoritism.” Journal of
Personality and Social Psychology 78:708–24.
Goldstein, Irwin L. 1991. “Training in Work Organizations.” Pp. 507–620 in Handbook of
Industrial and Organizational Psychology, vol. 2, edited by J.D. Dunnette and L.M.
Hough. Palo Alto, CA: Consulting Psychologists Press.
Goldstein, Morris and Robert Smith. 1976. “The Estimated Impact of the Antidiscrimination
Program Aimed at Federal Contractors.” Industrial and Labor Relations Review 29:523–
43.
Gordon, Randall A., Richard M. Rozelle, and James C. Baxter. 1988. “The Effect of Applicant
Age, Job Level, and Accountability on the Evaluation of Job Applicants.” Organizational
Behavior and Human Decision Processes 41:20–33.
Gordon, R. A., R. M. Rozelle, and J. C. Baxter. 1989. “The Effect of Applicant Age, Job Level,
and Accountability on Perceptions of Female Job Applicants.” The Journal of psychology
123(1):59–68.
Heckman, James J. and V. J. Hotz. 1989. “Choosing Among Alternative Nonexperimental
Methods for Estimating The Impact of Social Programs.” Journal of the American
Statistical Association 84:862–74.
Hegewisch, Ariane, Cynthia H. Deitch, and Evelyn F. Murphy. 2011. Ending Sex and Race
Discrimination in the Workplace: Legal Interventions That Push the Envelope.
Washington, DC: Institute for Women’s Policy Research.
Herman, Alexis M. et al. 2003. Second Annual Report of the Coca Cola Task Force. Prepared for
the United States District Court for the Northern District of Georgia pursuant to a
Settlement Agreement in Ingram, et al. v. The Coca-Cola Company (Case No. 1-98-CV3679 (RWS)).
Hirsh, Elizabeth C. 2009. “The Strength of Weak Enforcement: The Impact of Discrimination
Charges on Sex and Race Segregation in the Workplace.” American Sociological Review
74:245–71.
Hirsh, Elizabeth and Julie A. Kmec. 2009. “Human Resource Structures: Reducing
Discrimination or Raising Rights Awareness?” Industrial Relations: A Journal of
Economy and Society 48(3):512–32.
Hodson, Randy. 1996. “Dignity in the Workplace Under Participative Management: Alienation
and Freedom Revisited.” American Sociological Review 61(5):719–38.
34
Hodson, Randy and Robert L. Kaufman. 1982. “Economic Dualism: A Critical Review.”
American Sociological Review 47(6):727–39.
Holzer, Harry J. and David Neumark. 2000. “What Does Affirmative Action Do?” Industrial and
Labor Relations Review 53(2):240–71.
Johnson, C. Douglas. 2008. “It’s More Than the Five To Do’s: Insights on Diversity Education
and Training From Roosevelt Thomas, a Pioneer and Thought Leader in the Field.”
Academy of Management Learning & Education 7:406–17.
Johnston, William B. and Arnold Packer. 1987. Workforce 2000: Work and Workers for the
Twenty-First Century. Indianapolis, IN: Hudston Institute.
Kaiser, Cheryl R. et al. 2013. “Presumed Fair: Ironic Effects of Organizational Diversity
Structures.” Journal of Personality and Social Psychology 104:504–19.
Kalev, Alexandra. 2014. “How You Downsize Is Who You Downsize: Biased Formalization,
Accountability and Managerial Diversity.” American Sociolgical Review 79(1):109–35.
Kalev, Alexandra and Frank Dobbin. 2006. “Enforcement of Civil Rights Law in Private
Workplaces: The Effects of Compliance Reviews and Lawsuits over Time.” Law and
Social Inquiry 31:855–79.
Kalev, Alexandra, Frank Dobbin, and Erin Kelly. 2006. “Best Practices or Best Guesses?
Assessing the Efficacy of Corporate Affirmative Action and Diversity Policies.”
American Sociological Review 71:589–617.
Kalleberg, Arne L., David Knoke, Peter V. Marsden, and Joe L. Spaeth. 1996. Organizations in
America: Analyzing Their Structures and Human Resource Practices. Thousand Oaks,
CA: Sage Publications.
Kelly, Erin. 2000. “Corporate Family Policies in U.S. Organizations, 1965-1997.” Dissertation,
Princeton University, Princeton, NJ.
Kelly, Erin A. 2003. “The Strange History of Employer-Sponsored Childcare: Interested Actors,
Uncertainty, and the Transformation of Law in Organizational Fields.” American Journal
of Sociology 109:606–49.
Kelly, Erin and Frank Dobbin. 1998. “How Affirmative Action Became Diversity Management:
Employer Response to Antidiscrimination Law, 1961-1996.” American Behavioral
Scientist 41:960–84.
Kochan, Thomas A. et al. 2003. “The Effect of Diversity on Business Performance: Report of the
Diversity Research Network.” Human Resources Management 42:3–21.
Kruglanski, Arie W. and Tallie Freund. 1983. “The Freezing and Unfreezing of Lay-Inferences:
Effects on Impressional Primacy, Ethnic Stereotyping, and Numerical Anchoring.”
Journal of Experimental Social Psychology 19:448–68.
35
Kulik, Carol T., Elissa L. Perry, and Anne C. Bourhis. 2000a. “Ironic Evaluation Processes:
Effects of Thought Suppression on Evaluations of Older Job Applicants.” Journal of
Organizational Behavior 21(6):689–711.
Kulik, Carol T., Elissa L. Perry, and Anne C. Bourhis. 2000b. “Ironic Evaluation Processes:
Effects of Thought Suppression on Evaluations of Older Job Applicants.” Journal of
Organizational Behavior 24:289–711.
Legault, Lisa, Isabelle Green-Demers, and Allison L. Eadie. 2009. “When Internalization Leads
to Automatization: The Role of Self-Determination in Automatic Stereotype Suppression
and Implicit Prejudice Regulation.” Motivation and Emotion 33:10–24.
Legault, Lisa, Isabelle Green-Demers, Protius Grant, and Joyce Chung. 2007. “On the SelfRegulation of Implicit and Explicit Prejudice: A Self-Determination Perspective.”
Personality and Social Psychology Bulletin 33:732–49.
Legault, Lisa, Jennifer N. Gutsell, and Michael Inzlicht. 2011. “Ironic Effects of Antiprejudice
Messages How Motivational Interventions Can Reduce (but Also Increase) Prejudice.”
Psychological Science 22:1472–77.
Leonard, Jonathan S. 1984a. “Employment and Occupational Advance Under Affirmative
Action.” The Review of Economics and Statistics 66:377–85.
Leonard, Jonathan S. 1984b. “The Impact of Affirmative Action on Employment.” Journal of
Labor Economics 2:439–63.
Leonard, Jonathan S. 1984c. “The Impact of Affirmative Action Regulations and Equal
Employment Law on Black Employment.” Journal of Economic Perspectives 4:47–64.
Leonard, Jonathan S. 1985. “What Promises Are Worth: The Impact of Affirmative Action
Goals.” The Journal of Human Resources 20:3–20.
Leonard, Jonathan S. 1989. “Women and Affirmative Action.” The Journal of Economic
Perspectives 3:61–75.
Leonard, Jonathan S. 1990. “The Impact of Affirmative Action Regulation and Equal
Employment Opportunity Law on Black Employment.” The Journal of Economic
Perspectives 4:47–63.
Lerner, Jennifer S. and Philip E. Tetlock. 1999. “Accounting for the Effects of Accountability.”
Pyschological Bulletin 125:255–75.
Malhotra, Yogesh, Dennis F. Galletta, and Laurie J. Kirsch. 2008. “How Endogenous
Motivations Influence User Intentions: Beyond the Dichotomy of Extrinsic and Intrinsic
User Motivations.” Journal of Management Information Systems 25(1):267–99.
36
Manjoo, Farhad. 2014. “Exposing Hidden Bias at Google.” The New York Times, September 24.
Retrieved March 11, 2015 (http://www.nytimes.com/2014/09/25/technology/exposinghidden-biases-at-google-to-improve-diversity.html).
Milkman, Ruth. 1985. “Women Workers, Feminism, and the Labor Movement since the 1960s.”
in Women, Work, and Protest: A Century of Women’s Labor History, edited by Ruth
Milkman. Boston, MA: Routledge and Kegan Paul.
Noe, Raymond A. and J. Kevin Ford. 1991. Emerging Issues and New Directions in Training
Research. Greenwich, CT: JAI Press.
Osterman, Paul. 1994. “How Common Is Workplace Transformation and Who Adopts It?”
Industrial and Labor Relations Review 47:173–88.
Paluck, Elizabeth Levy and Donald P. Green. 2009. “Prejudice Reduction: What Works? A
Review and Assessment of Research and Practice.” Annual Review of Psychology
60:339–67.
Paluck, Elizabeth L. and Donald P. Green. 2009. “Prejudice Reduction: What Works? A Critical
Look at Evidence from the Field and the Laboratory.” Annual Review of Psychology
60:339–67.
Plant, E. Ashby and Patricia G. Devine. 1998. “Internal and External Motivation to Respond
Without Prejudice.” Journal of Personality and Social Psychology 75:811–32.
Plant, E. Ashby and Patricia G. Devine. 2001. “Responses to Other-Imposed Pro-Black Pressure:
Acceptance or Backlash?” Journal of Experimental Social Psychology 37:486–501.
Plant, E. Ashby, Patricia G. Devine, and Michelle B. Peruche. 2010. “Routes to Positive
Interracial Interactions: Approaching Egalitarianism or Avoiding Prejudice.” Personality
and Social Psychology Bulletin 36:1135–47.
Plaut, Victoria C., Flannery G. Garnett, Laura E. Buffardi, and Jeffrey Sanchez-Burks. 2011.
“‘What about Me?’ Perceptions of Exclusion and Whites’ Reactions to
Multiculturalism.” Journal of Personality and Social Psychology 101(2):337–53.
Reskin, Barbara F. and Debra B. McBrier. 2000. “Why Not Ascription? Organizations’
Employment of Male and Female Managers.” American Sociological Review 65:210–33.
Robertson, Nan. 1971. “‘Race Awareness’ Drive Stirs Capital Dispute.” New York Times,
November 29, 1857-Current File.
Robinson, Corre, Tiffany Taylor, Donald Tomaskovic-Devey, Catherine Zimmer, and Matthew
W. Irvine. 2005. “Studying Race/Ethnic and Sex Segregation at the Establishment-Level:
Methodological Issues and Substantive Opportunities Using EEO-1 Reports.” Work and
Occupations 32:5–38.
37
Ryan, Richard M. and James P. Connell. 1989. “Perceived Locus of Causality and
Internalization: Examining Reasons for Acting in Two Domains.” Journal of personality
and social psychology 57(5):749.
Rynes, Sara and Benson Rosen. 1995. “A Field Survey of Factors Affecting the Adoption and
Perceived Success of Diversity Training.” Personnel Psychology 48:247–70.
Schlanger, Margo and Pauline T. Kim. 2013. “The Equal Employment Opportunity Commission
and Structural Reform of the American Workplace.” Retrieved November 4, 2014
(http://repository.law.umich.edu/law_econ_current/89/?utm_source=repository.law.umic
h.edu%2Flaw_econ_current%2F89&utm_medium=PDF&utm_campaign=PDFCoverPag
es).
Shenhav, Yehouda A. and Yitchak Haberfeld. 1992. “Organizational Demography and
Inequality.” Social Forces 71:123–43.
Sidanius, Jim, Erik Devereux, and Felicia Pratto. 2001. “A Comparison of Symbolic Racism
Theory and Social Dominance Theory as Explanations for Racial Policy Attitudes.”
Journal of Social Psychology 132:377–95.
Skaggs, Sheryl L. 2001. “Discrimination Litigation: Implications for Women and Minorities in
Retail Supermarket Management.” Ph.D. Dissertation, North Carolina State University,
Raleigh, NC.
Skaggs, Sheryl L. 2008. “Producing Change or Bagging Opportunity? The Effects of
Discrimination Litigation on Women in Supermarket Management.” American Journal of
Sociology 113:1148–82.
Society for Human Resources Management. 2004. “What Are The Components of a Successful
Diversity Initiative?”
Sturm, Susan. 2001. “Second Generation Employment Discrimination: A Structural Approach.”
Columbia Law Review 101:459–568.
Tetlock, Philip E. 1992. “The Impact of Accountability on Judgment and Choice: Toward a
Social Contingency Model.” Pp. 331–76 in Advances in Experimental Social Psychology,
vol. 25, edited by Mark P. Zanna. New York: Academic Press.
Thomas, David A. and Robin J. Ely. 1996. “Making Differences Matter: A New Paradigm for
Managing Diversity.” Harvard Business Review, September, 79–90.
Wiggenhorn, William. 1990. “When Training Becomes an Education.” Harvard Business Review
68:71–83.
Williams, Joan. 2000. Unbending Gender: Why Family Work Conflict and What to Do about It?.
New York: Oxford University Press.
38
Williamson, Thomas S. Jr. et al. 2002. Fifth Annual Report of the Equality and Fairness Task
Force for the Year Ending June 20, 2002. Texaco Task Force on Equality and Fairness.
Prepared for the United States District Court for the Southern District of New York.
39