Diversity Training and Intergroup Contact: A Call to Action Research

Journal of Social Issues, Vol. 62, No. 3, 2006, pp. 577--595
Diversity Training and Intergroup Contact: A Call
to Action Research
Elizabeth Levy Paluck∗
Yale University
Diversity training is a type of prejudice reduction and social inclusion intervention
in need of “action research”—an integration of research and theory with practice
(Lewin, 1946). Hundreds of workplaces and schools use some form of diversity
training, but most interventions are not grounded in theory and there is little
evidence of program impact. A recent study of a school diversity training program
illustrates how action research can address theoretical issues using experimental
methods and unobtrusive outcome measures. For future research, the literature
on intergroup contact (Pettigrew, 1998) can provide theoretical guidance while
testing and refining its principles in the application and investigation of diversity
training. Action research will benefit diversity training and the broader theoretical
and applied project of prejudice reduction and the promotion of social inclusion.
“Diversity Day”
A group of office workers shuffle into a fluorescent-lit room where two rows
of chairs face a television stand. Standing in front of a “Diversity Day—Take 2!”
banner, their supervisor Michael urges them into their seats: “Let’s have fun, ev-
erybody!” After the group watches a video of a man speaking about the importance
of diversity, a South Asian woman rises and heads for the door. She pauses in front
of Michael to explain she has a customer meeting. “If you leave, we’ll only have
two left—er, yes, enjoy!” Michael blurts. Turning back to the group, composed of
eight White men and women, one Hispanic man and one Black man, he introduces
himself and the exercise of the day.
Michael instructs each person to pick an index card from a pile and put the card
on his or her forehead without seeing what is written on the other side. The various
∗Correspondence concerning this article should be addressed to Elizabeth Levy Paluck,
Department of Psychology, Yale University, PO Box 208205, New Haven CT 06520 [e-mail:
elizabeth.paluck@yale.edu].
577
C 2006 The Society for the Psychological Study of Social Issues

---

578
Paluck
cards say “JEWISH,” “ASIAN,” “ITALIAN,” and “BLACK.” “I want you to treat
other people like the race that is on their forehead, okay? . . .Nobody knows what
their race is.” As the men and women eye the small billboards on their partners’
heads, Michael encourages them to mingle and “let the sparks fly! . . .Let’s get
real!”
A woman with “JEWISH” on her head stands with the Black man who has by
chance chosen the “BLACK” card. Carefully, he offers, “I admire your culture’s
success in America.” Michael appears between the two of them “Good! Olympics
of suffering right here! Slavery versus the Holocaust, come on!” The Black man
frowns and pulls off his index card: “Who am I supposed to be?”
The “JEWISH” woman moves on to speak to a man wearing an “ASIAN” card.
He hails her with a “Shalom” and asks her for a loan. Grimly, she compliments
him on his culture’s cuisine. Once again Michael jumps in: “Come on, stir the pot.
Stir the melting pot!” She draws a breath. “Okay. If I have to do this, based on
stereotypes that are totally untrue, that I do not agree with, you would maybe. . .
not be a very good driver.” The “Asian” man grimaces. “Oh man! Am I a woman?”
The South Asian woman returns to the room, and Michael strides to meet
her, motioning all to watch. Using an absurdly exaggerated Indian English accent,
he pretends that he is a convenience store owner inviting her into his store. The
woman’s expression shifts from bewilderment to anger. As Michael’s voice reaches
a fever pitch, she reaches back and slaps him. The group stands in a dead silence.
Michael straightens up and declares, “Yes, that’s it! NOW she really knows what
it’s like to be a minority!”1
Diversity Training as an Intergroup Relations Intervention
“The goal of diversity training programs is to increase awareness of racial, ethnic, and
cultural differences and help [people] to value these differences” (Hollister, Day, & Jesaitis,
1993; cited in Stephan & Stephan, 2001 p 80).
Diversity training is an intergroup relations program that often triggers a
heated and politically sensitive public debate. The “Diversity Day” story above
is actually a scene from the American television comedy “The Office,” a show
that lampoons workplace culture. Diversity training’s critics might agree with
the office’s portrayal of a diversity training gone terribly wrong. Journalists and
academics have asserted that diversity trainings perpetuate racial tension, heighten
stereotypes, and foster new sensitivities and anxieties, while relying on pseudo-
scientific theories and no supporting evidence (see Day, 1995; Ford, 2000; Lasch-
Quinn, 2001; Lynch, 1997). Proponents of diversity training might respond that
the show is a satire and it portrays one ill-advised approach out of a wide spectrum
of diversity training programs.
1 The Office, c 2005 NBC Studios, Inc., and Universal Network Television LLC. All rights
reserved. Special thanks to Richard Eibach for bringing this episode to my attention.

---

Diversity Training and Intergroup Contact
579
But like all good satire, “Diversity Day” exposes and exaggerates some very
real problems and questions that are recognized by those who believe in the value
of diversity training. For example: Do representatives of minority groups need to be
present? Should diversity trainings encourage people to openly express their stereo-
types and prejudices? How do people’s emotional reactions inhibit or facilitate the
goals of diversity training?
A review of the professional and scholarly literature about diversity training
reveals that after 30 years and thousands of workplace interventions, the most
accurate answer to these and other questions is “we aren’t sure.” Previous re-
views recognize that diversity training’s critics are correct in some respects—
oftentimes programs are not designed on established theory or empirical evidence,
and there is a serious lack of rigorous evaluation and follow-up to gauge program
impact (Stephan & Stephan, 2001; Weithoff, 2004). With exceptions (Alderfer,
1992; Katz, 1977; Stockdale & Crosby, 2003; Weithoff, 2004), academics who
are interested in prejudice reduction and social inclusion have left diversity train-
ing to a corps of professional nonacademic consultants. Meanwhile, unanswered
questions about diversity training leave it exposed to polemical attacks (e.g.,
Feder, 1994; Lubove, 1997) and even to lawsuits (e.g., Stender v. Lucky Stores,
1992).
Yet diversity training sells well; it has become a fixture of the American
workplace, where in 2005 66% of U.S. employers used diversity training despite the
fact that training is not required by federal equal opportunity law (Compensation
and Benefits for Law Offices, 2006). Diversity training is positioned to impact
thousands of people and workplaces in a positive way. But by and large, scholars
and practitioners have passed up the opportunity for a collaborative project that
could harness this widespread intervention to improve the theory and practice of
prejudice reduction and social inclusion. A comprehensive project would have two
goals:
1. A clear theoretical rationale for predictions about the implementation and
outcomes of diversity training: for whom, when, for how long, with which
methods, and to what ends.
2. Evaluation research to determine the impact of different types of diversity
training in different contexts, prioritizing randomized controlled field experi-
ments and unobtrusively measured outcomes.
These two goals reflect the Lewinian (1946) “action research” approach high-
lighted in this issue, where social interventions based on theory are refined and con-
tinuously retested in real world contexts. This article calls for a coordinated effort
among theoreticians, researchers, and practitioners to understand diversity train-
ing’s processes and potential impact in real world settings. Because intergroup re-
lations scholars and diversity training practitioners wrestle with similar questions,

---

580
Paluck
a logical starting place is to use existing theory and research that has immediate
relevance for the questions surrounding diversity training.
The literature on intergroup contact (Pettigrew, 1998) is one body of academic
work where such collaboration could begin. Academic research on intergroup
contact probes the conditions under which prejudice decreases among groups and
individuals, even in situations when there is no direct intergroup contact. Diversity
training and intergroup contact both stand to gain from a collaborative action
research project; diversity training programs could use the relevant insights of the
intergroup contact literature, while providing externally valid tests of its hypotheses
and generating new insights and further research questions. This particular example
of action research demonstrates how this kind of collaboration is important for the
broad project of intergroup relations theory and application.
Roadmap
First, I present a brief history and some popular forms of diversity training,
and review the sparse literature on its impact. Next, I provide methodological
recommendations for future action research on diversity training, highlighting the
importance of field experimental methods and unobtrusive outcome measurement.
I use one recent study of a school diversity training program as an example. I then
review the intergroup contact research and how its common concerns overlap
with many questions regarding the purposes, methods, and outcomes of diversity
training. These areas of common concern are potential spaces where action research
could benefit intergroup contact theory and the practice of diversity training.
Diversity Training: An Overview
The invention of diversity training followed affirmative action efforts in the
1960s and 1970s that changed the demographic composition of many organizations.
These early “sensitivity trainings” were often responses to or preventative mea-
sures against discrimination lawsuits. In the late 1980s a think tank report called
Workforce 2000 (Johnson & Packer, 1987) jumpstarted a “diversity craze” (Judy
& D’Amico, 1997, cited in Hays-Thomas, 2004, p 13) with its projection that the
percentage of non-Whites and females would rise significantly in the workplace
over the coming decade.2 Diversity consultants recommended “diversity manage-
ment”—a comprehensive approach ranging from one-day diversity trainings to
institutional reforms for hiring and retention—as a labor market imperative, not
2 Interestingly, part of the rush to implement diversity programs came out of a widespread mis-
interpretation of the report’s graph of a projected 15% net percentage decrease among White male
workers. The media incorrectly reported that by 2000, only 15% of the workforce would be made
up of White males. The actual projected percentage of White males in the workforce was 41%, one
percentage point above the actual figure in 2000.

---

Diversity Training and Intergroup Contact
581
just a social good. Diversity training experienced another surge in demand after
the 9/11 attacks, even as companies cut their budgets for other kinds of training
(cited in Leonard, 2002).
Diversity training is a catchall title that encompasses many types of activities,
from lectures to movies to role-plays. Some researchers classify diversity training
activities along a continuum from instructional to experiential training methods
(Gudykunst & Hammer, 1983; Lindsay, 1994; Stephan & Stephan, 2001).
Instructional methods of diversity training supply information and raise aware-
ness of the problems associated with misunderstanding or mishandling diversity,
or conversely, the benefits of “diversity friendly” behaviors and policies. Instruc-
tional activities include lectures, videos, fact sheets, and group discussions led by
a diversity trainer (usually an outside consultant). Presentations may cover such
topics as equal opportunity laws, policies against gender harassment, the nature of
a region or country’s demographic changes, or information about underrepresented
groups meant to replace myths and stereotypes (Ferdman & Brody, 1995; Gannon
& Poon, 1997; Holvino, Ferdman, & Merrill-Sands, 2004; Roberson, Kulik &
Pepper, 2001). Some diversity trainings lecture about different “cultural styles”
of communication and social interaction (Griggs & Louw, 1995), or different
personality profiles associated with different groups (www.DiscProfile.comTM).
A brand new development in instructional diversity training is the use of the
Implicit Attitude Test to teach members of organizations about the prevalence
and implications of unconscious bias in everyday behavior (IAT Corporation,
2005).
Experiential methods of diversity training take a personalized and partici-
patory approach to building skills that promote harmonious and productive in-
teraction across group lines. Participants in these trainings may travel to neigh-
borhoods of a different economic or ethnic background than their own, practice
communication techniques, or observe one another’s style of intergroup interac-
tions (e.g., Hanover & Cellar, 1998). Sometimes experiential methods have the
flavor of a group therapy session, in which participants are invited to disclose
their feelings toward diversity, or to describe their personal backgrounds or past
experiences with prejudice. Group discussions and dialogues about “difference”
of all sorts comprise another popular method (Walker & Hanson, 1992) as do
role playing exercises featuring work or social situations with characters from a
variety of backgrounds (see Alderfer, 1992). Recently a city council encouraged
employees to attend the Hollywood movie “Crash,” which deals with the topic
of race relations, as part of their annual diversity training requirement (Frazier,
2006).
Whether from an instructional or experiential approach, most diversity train-
ings are based on implicit assumptions about the value of overcoming ignorance,
expressing one’s hidden assumptions, or feeling empathy for an oppressed group
or individual. Fewer trainings are explicitly based on established theories about
prejudice reduction or social inclusion (c.f. Nagda, this issue). Moreover, programs

---

582
Paluck
take different views on what exactly constitutes diversity. Some programs focus
on traditionally recognized group cleavages like race, ethnicity, gender, disability,
religion, and sexual orientation, while others expand the meaning of diversity to
include ability, philosophical or political views, working style, and so forth. The
number of activities in a diversity training varies—some programs have one sig-
nature exercise, while other trainings use an assortment of exercises, discussions,
and videotapes over the course of the training session. The majority of diversity
trainings take place in one day, but some diversity consultants contract with an
organization to conduct courses across weeks or months.
Real “Diversity Days”: Two Illustrations
An example of an instructional diversity training is a video used in thou-
sands of organizations, including major corporations like Hewlett Packard and
Proctor and Gamble. A Tale of “O” features animated X’s and one O, who sym-
bolizes the odd person out. O works among X’s, walks by X’s on the street, and
often feels different and confused about how to behave. When the X’s are not
actively prejudging O, they still feel uncomfortable and hesitant in their inter-
actions with O. The video encourages participants to relate to O by thinking
of a time they felt like a minority for any number of reasons—whether they
were a woman among men, wearing the “wrong” clothes at a party, or travel-
ing in a foreign country. Based on findings about minority status from a semi-
nal study of gender in organizations (Kanter, 1977), the video’s objective is to
“. . .help both X’s and O’s understand what is happening to them, and [to] help each
learn to feel more comfortable dealing with the other” (www.goodmeasure.com,
2006).
The Blue Eyes/Brown Eyes technique is a type of experiential training made
famous by Jane Elliot, who divides trainees into two arbitrary groups (most
famously, according to eye color) and for the next several hours favors one group
while verbally harassing and denying privileges to the other. To highlight group
differences she pins collars on the “inferior” group members, expounds on
pseudo-scientific theories that justify their lesser status, and singles out individ-
ual group members for humiliation. Her website quotes a review of her train-
ing method: “Even we, the spectators in BLUE EYED, can’t get rid of this
feeling of uneasiness, embarrassment, anxiety, and utterly helpless hatred when
she starts keeping people down.” The purpose of the training is to demonstrate
that “prejudice. . .is in irrational class system based on purely arbitrary factors”
(http://www.janeelliott.com/, 2006). Trainees are supposed to develop empathy
and awareness through their own personal experience of discrimination.
Organizations request a live trainer, including Elliot herself, or they watch a video
of the program.

---

Diversity Training and Intergroup Contact
583
The Measured Impact of Diversity Training
“The preferred solution to the problems of [outcome] measurement and description is to
declare them irrelevant and proceed on faith alone. “One of the problems corporations have
with diversity is that they like things in boxes,” says [a diversity consultant]. . .“It is. . . an
evolutionary process that is very difficult to be specific about . . .you need to work five or
ten years before you can say you’re into a diversity process. It never ends.” And at [the
consultant’s fee of] two grand a day, there’s no reason to think it will.” (MacDonald, 1993,
p. 22)
What rigorous field studies demonstrate the impact of the various types of di-
versity training? Unfortunately there are very few studies that meet social scientific
standards for measuring the casual effects of an intervention. Even if one is sym-
pathetic to practitioners’ concerns that diversity programs need time to transform
an organization, the lack of evaluation has given rise to cynics who see diversity
training as a modern day medicine show.
A large number of posttraining surveys measure participants’ opinions of the
training exercises or of their own consequential change in attitudes or behaviors
(e.g., Morris, Romero, & Tan, 1996). These surveys cannot provide accounts of
program impact because it is impossible to rule out well-known sources of bias
like self-presentation and social desirability concerns that plague self-report survey
studies. Without an equivalent comparison group of nonparticipants, investigators
cannot know whether enthusiastic answers regarding the value of diversity do not
come from participants’ desire to satisfy the diversity trainer or their workplace
supervisor (a self-presentation bias), or from their need to conform to social norms
of political correctness (a social desirability bias). Even in the absence of these
concerns, people are not especially insightful when it comes to evaluating outside
influences on their behavior (Nisbett & Bellows, 1977), thus surveys asking par-
ticipants to judge their own level of prejudice reduction cannot be weighed heavily,
even when they are confidential.
Excluding the posttraining surveys, a limited group of prepost or controlled
studies has posed some interesting and important questions about the impact of
diversity training. These questions include: what is the impact of isolated one-shot
trainings compared to those that are part of comprehensive diversity management
programs? Does employee feedback increase their managers’ sensitivity to diver-
sity concerns posttraining? What are the short versus long-term effects of diversity
training? Unfortunately, just a few isolated studies are devoted to each question, and
correlational designs or small sample sizes circumscribe these studies’ disparate
conclusions.
One example is a study by the American Management Association that corre-
lated the investigators’ own opinions of companies’ “training effectiveness” with
the structure and the duration of the diversity training. From these biased data, they
concluded that isolated instances of diversity training were less effective compared

---

584
Paluck
to those that are part of diversity management programs (cited in Caudron & Hayes,
1997).
Investigators have probed other questions with better methods, but with a
small number of studies from which it is difficult to generalize. One evaluation
of a diversity training program for White middle managers investigated whether
a diversity training plus anonymous feedback from their subordinates would im-
prove managers’ diversity-sensitive behaviors such as discouraging jokes based on
stereotypes (Hanover & Cellar, 1998). Managers who participated in the training
rated diversity-sensitive behaviors as more important and reported engaging in
those behaviors more than a control group of managers. Two problems with the
conclusions of this study are that there is no way of telling which component or
components of the treatment (e.g., the diversity training activities or the anony-
mous feedback) were responsible for the outcome, and if the managers actually
behaved in the way that they reported in the final questionnaires.
An evaluation of a diversity training for supervisors at a U.S. military in-
stallation attempted to determine whether the impact of a stereotype-reduction
training would endure over time by comparing trained supervisors to those who
were waiting to be trained (Tansik & Driskell, 1977). Initial decreases in stereo-
typing (measured by a questionnaire in which trainees rated “American Indians”,
“Blacks,” “Whites,” and “Latinos” as successful vs. unsuccessful, unpleasant vs.
pleasant, honest vs. dishonest, etc.) rebounded after 3 months had passed, but
stereotyping continued to fluctuate over the course of repeated follow-up surveys.
It is unclear whether the investigators would have observed the same outcomes
with a less transparent measure of stereotyping.
One field experiment examined whether the demographic composition of
diversity training groups would change trainees’ reactions, using a sample of grad-
uate instructors at a business school (Roberson et al., 2001). For the 4-hour vol-
untary and instructional training, 127 graduate students were randomly assigned
to a racially homogenous or a racially heterogeneous training group. At the end of
the training the investigators measured the trainees’ knowledge, attitudes toward
diversity, and their intentions to promote intergroup understanding. Group com-
position did not make a difference for any of these outcomes; however, graduate
instructors who reported prior experience with diversity training scored better on
the knowledge and intention outcomes when they were in homogeneous groups.
Roberson et al. hypothesized that homogeneous groups may be best for a social
modeling process, in which people are better able to imitate those who are similar
to themselves. However, this was a post hoc explanation rather than a theoretically
driven prediction.
Thus, many of the important questions investigators pose about diversity train-
ing are stranded by the small number of studies and the methods used in each
one; meanwhile consultants make many recommendations in advance of the
accumulation of solid evidence. The most important lesson from the current
assertment of evidence is that future research needs to overcome their

---

Diversity Training and Intergroup Contact
585
methodological limitations in order to fully address important practical and theo-
retical questions.
Methods for Investigating Diversity Training
The key strategies for overcoming past methodological limitations of diversity
training research are: (1) establish the causal effect of the program, (2) use
unobtrusive outcome measurement that goes beyond self-report, and (3) conduct
the research in relevant populations and settings. Accomplishing these goals will
help to ensure that studies are able to answer the questions they are designed to
ask.
Previous research on diversity training has been unable to establish the causal
effect of training by using random assignment of participants to training groups
and to a no-training control group. In a review of the prejudice reduction literature,
my colleague Donald Green and I only found eight studies of diversity training
activities that used randomized treatment and control groups (Paluck & Green, in
press). Experimental designs could also help to solve the “kitchen sink” problem
faced by diversity training evaluations—that is, which component of a multifaceted
diversity training is the “active ingredient” of its success? Alternately, what is the
added benefit of each component: lectures, videos, role-plays, and the like? Exper-
iments can address these questions by randomly assigning participants to different
training groups, i.e., lecture-only versus video-only versus lecture plus video.
Future research should also develop more unobtrusive measures for diver-
sity training outcomes. Moving beyond transparent questionnaires that communi-
cate the “correct” response (e.g., “Do you discourage biased statements in your
workplace,” or, Rate the honesty of “blacks,” “women,” “Asians,” etc.) is espe-
cially important in the context of diversity training. Participants’ concerns about
self-presentation may be magnified in workplace or school settings by the presence
of a research investigator in addition to a work supervisor or teacher. Measuring
behavior is a challenging task of any evaluation, but it is an outcome measure
that is generally less subject to self-presentation bias and is an important piece
of evidence for diversity training’s impact. A creative and underutilized solution
to the difficulties with direct observation of behavior is to collect “third party re-
ports” from outside sources like colleagues and supervisors, or to examine physical
evidence (if accessible) such as minutes from meetings and employee or student
evaluations. Because each one of these sources of information is unreliable on its
own, evaluators should try to collect multiple measures of participants’ statements,
work habits, and interactions with others.3
3 Many diversity trainers value qualitative feedback from participants, which can still be collected in
the context of an experiment with all of the other unobtrusive measures mentioned above. Participants’
impressions and reactions can inform certain aspects of the training design, however, this feedback
should be collected in a separate session, to avoid “contaminating” the less-transparent impact questions
with straightforward feedback questions. Feedback should never be confused with evidence of impact.

---

586
Paluck
The populations studied in diversity training evaluations are not always similar
to participants in nonresearch diversity trainings. For example, the Blue
Eyes/Brown Eyes training has only been evaluated in a college student population,
though it is regularly used with adults in professional organizations. Researchers
should strive to study diversity training in the contexts where they most often occur
in order to generalize their findings.
One understudied setting is organizations that are forced to conduct diversity
training under threat of or following a discrimination lawsuit. These types of orga-
nizations are usually reluctant to allow researchers access, but researchers could
make the case that an evaluation would send a signal that the organization is com-
mitted to finding out “what works” to change their hostile workplace environment.
Whatever the arrangement, the pay off would be large in terms of the ability to
extend theoretical insights and intervention methods to a wider range of settings
and populations.
A Randomized Controlled Evaluation of a Diversity Training Program
To illustrate how future studies of diversity training can use rigorous methodol-
ogy to answer theoretically driven questions about prejudice reduction and social
inclusion, I offer an example from my own work, a study of a school diversity
training I recently conducted with Donald Green (Paluck & Green, 2006). We
investigated the impact of a Peer Training Program run by the Anti-Defamation
League (ADL) in several U.S. high schools.
The Peer Training Program aims to reduce prejudice in individual students
and in the general school culture by training a representative sample of student
“Peer Trainers” to be voices of tolerance in the hallways and classrooms of their
schools. Peer trainers are expected to engage fellow students in discussions about
prejudice and social inclusion, and to stand up for targets of prejudice in their school
community. We were interested in first, the impact of the training program on the
knowledge and attitudes of student Peer Trainers, and second, the influence of the
Peer Trainers on their friends and classmates. To study these two processes, we
selected a sample of 10 schools, matched similar schools into pairs, and randomly
assigned one school in each pair to begin the Peer Training Program (treatment
condition), and the other to a waiting list (control condition). Before the program
started, future Peer Trainers in both the treatment and the control schools wrote
down the names of two close friends and eight classmates, in a supposedly unrelated
general questionnaire administered by the ADL.
After Peer Trainers in the treatment school completed their training, we
conducted a telephone survey of students in both treatment and control schools:
Peer Trainers as well as the close friends and classmates they listed in the presurvey.
The telephone survey was designed to minimize self presentation and social
desirability concerns with a number of different techniques. For one, interviewers

---