A systematic review of diversity, equity, and inclusion and antiracism training studies: Findings and future directions

Abstract

A growing number of organizations are prioritizing diversity, equity, and inclusion (DEI) and antiracism in the workplace, including investing resources in DEI or antiracism training. However, such trainings vary widely in curriculum, objectives, delivery, and evaluation, with little known about the efficacy of existing trainings. The aim of this systematic review is to evaluate training characteristics, measures, and results of peer-reviewed studies (published between 2000 and 2022) testing DEI or antiracism trainings. Studies were identified using Google Scholar, JSTOR, and a university library database. Key search terms included “diversity, equity, and inclusion training”; “antiracism training”; and “effect,” “impact,” “outcome,” or “evaluation.” The search yielded N = 15 DEI training studies and N = 8 antiracism training studies. The majority of studies (75% of antiracism training; 66.6% of DEI training) utilized a one-time training session. Content, objectives, measures, and impact varied widely across studies. Randomized designs were uncommon (13%), and over 70% of studies had majority female participants. Findings highlight several strategies to advance the field of DEI and antiracism training, such as shifting curriculum from targeting individual knowledge to supporting behavioral and organizational change, providing longitudinal training, standardizing outcomes of interest, and implementing rigorous evaluation methods.

Lay Summary

Despite the increased number of organizations dedicating resources to diversity, equity, and inclusion training and/or antiracism training, little is known about which strategies yield successful results. This systematic review synthesizes research findings on diversity, equity, and inclusion and antiracism training studies over the past two decades and presents research and practice-based recommendations for how to move the field forward.

Implications

Practice: Diversity, equity, and inclusion (DEI) and antiracism trainings may benefit from including longitudinal training design, targeting skill building and organizational development as core parts of the curriculum, and prioritizing behavioral and organizational change as outcomes of interest.

Policy: Investments in research on DEI and antiracism trainings are needed to identify best practices for cultivating equitable and inclusive work environments.

Research: Future research is needed that integrates community-engaged approaches in the design, implementation, evaluation, and dissemination of DEI and antiracism trainings. Additionally, the use of more rigorous study design methods and longer follow-up assessments can help identify efficacious training strategies.

Introduction

Advancing diversity, equity, and inclusion (DEI) and antiracism in the workforce has increased in priority across industries and sectors, including health care and higher education, over the past few decades and in particular over recent years. Following the racial reckoning of 2020 in the USA and globally, the number of DEI-related job positions increased by 60% in the USA [1] and 86% of large financial companies surveyed stated intention to increase investments in DEI training [2]. In 2021, the USA spent nearly $3.8 billion dollars in DEI-related training, accounting for 45.5% of the total global spending on DEI [3]. By 2026, global DEI-related spending is expected to reach $15.4 billion [3].

Despite the growing number of organizations expressing commitment and dedicating resources to DEI and/or antiracism training, little is known about the efficacy of such trainings (T2 of the translational research spectrum) [4]. Increasing our understanding of what works and what does not will allow us to better identify best practices, avoid investing in inefficacious strategies, and innovate new, community-engaged research approaches that are better positioned to make an impact [5]. The need for novel and impactful DEI and antiracism trainings is great, as inequities in the workplace continue to persist. Across industries, pay disparity continues to affect employees based on race [6], gender [7], age [8], disability [9], and sexual orientation [10]. A recent Gallup poll results revealed that 61% of Latine employees and 75% of Black employees in the USA experienced ethnicity or race-based discrimination in the last year [11]. In 2021, over 47 million employees voluntarily quit their jobs, and subsequent analysis revealed that toxic culture was the primary driver of attrition, with the failure of organizations to promote DEI among the top characteristics of an unhealthy work culture [12].

Studies to date evaluating such trainings have yielded inconsistent findings, with some demonstrating change in knowledge and attitudes associated with training participation [13, 14] while others show null results [15, 16]. The mixed findings may partially be attributed to the wide heterogeneity of training characteristics (e.g. goals, curriculum, duration), study participants and settings, and outcomes measured. Given the wide range of training and study characteristics, research to date on the efficacy of DEI and antiracism trainings has been unable to reach a consensus regarding how such trainings ought to be designed, measured, implemented, and evaluated to advance equity in the workplace.

While the definition and purpose of DEI and antiracism trainings has varied over time and context, we define DEI training as the intentional provision and application of education (in various forms) to “recognize and address how power, privilege, and how society affects our personal identities,” including race, socioeconomic status, sexual orientation, gender, etc [17]. We defined antiracism training as the intentional provision and application of education to help individuals and organizations “understand systemic racism, along with the practices or structures that perpetuate it” [17]. The goal of this systematic review was to evaluate the training content and strategies, theoretical frameworks, delivery modality and duration, participant characteristics, measures, and results of peer-reviewed research studies (published between 2000 and 2022) of DEI and antiracism trainings tested in work settings. As DEI and antiracism trainings are needed and applicable across all industries and sectors, including medicine, public health, and higher education, we did not limit our search to studies conducted in health care or health-related settings.

Methods

Data sources, study selection, and data abstraction

The systematic review was conducted using three online databases: Google Scholar, JSTOR, and a university libraries database, which included relevant databases such as PsycINFO, Wiley Online Library, SpringerLink, ScienceDirect, and others. Key search terms for DEI training studies included “diversity, equity, and inclusion training” and [“effect,” “impact,” “outcome,” or “evaluation”]. Key search terms for antiracism training studies included “antiracism training” and [“effect,” “impact,” “outcome,” or “evaluation”].

Inclusion criteria for the studies of interest were: (i) original, empirical study that tested and measured the effect(s) of a DEI or an antiracism training; (ii) peer-reviewed publication between 1 January 2000 and 31 December 2022; and (iii) published in the English language. One coauthor conducted a search for DEI training studies and a separate search for antiracism training studies based on the search terms and engines specified above. Article titles and abstracts were screened based on inclusion criteria. Articles that met inclusion criteria based on the abstract review were selected for a full-text review. Articles that met all criteria were included for review.

Data abstracted included sample characteristics (race, gender, and other reported sociodemographics of participants, number of participants), training characteristics [theoretical frameworks that informed the training (if any), training format, delivery, topics/curriculum, duration, and setting], study design (e.g. quasi-experimental vs. randomized, use of a comparison group, follow-up duration), outcomes assessed (e.g. knowledge, attitudes, skills, behavior change) and survey instruments used, and training effect(s) on measured outcomes of interest. One reviewer first abstracted data of interest for each article and an additional reviewer conducted independent data validations. The training topics were gleaned from the descriptions of the interventions in the studies’ methods section; two independent reviewers reviewed the methods section of all the studies and described the training topics. Discrepancies were reviewed and resolved by consensus with a third reviewer.

Results

As DEI and antiracism training are conceptually distinct, we present the results of DEI and antiracism training studies reviewed separately and with patterns observed across both types of studies. Studies included for the systematic review based on the DEI training search methods did not overlap with studies identified via the antiracism training search methods.

DEI training search results

The literature search for DEI training studies yielded a total of 251 nonduplicate articles (Figure 1). Of those, 198 articles were excluded based on the title review. N = 31 studies were excluded during the abstract review process, and N = 7 studies were excluded during the full-text review, resulting in a total of 15 articles eligible for inclusion. Table 1 presents a summary of characteristics across DEI training studies included in this review.

Table 1

Characteristics and results of DEI training studies published between 2000 and 2022 (N = 15)

Study article	Sample and setting characteristics	Training characteristics	Measures	Analysis and results
1. Castillo et al., 2011 [18]	N = 84 participants Demographics: - Gender: 80% female - Race: 77.4% White, 3.6% Black, 6% Asian, 6% Latino, 7% biracial Setting: higher education	Theory: adverse racism theory, feminist theory Modality: in-person Duration: weekly 3-hour course for 15 weeks Facilitator characteristics: all were professors; 2 were Latina, 3 were White (1 female, 2 male) Topics covered: history, cultural customs, traditions, and values, feminism	Study design: course evaluations analyzed from 10 courses (5 multicultural, 5 counseling foundations) Outcomes measured: knowledge, attitudes, skills, self-awareness Instruments used: Multicultural Counseling Inventory, Race Implicit Association Test, study survey	Assessment time points: Pre and post-training Analysis: multilevel modeling Effect(s): null improvements in knowledge and skills, statistically significant improvement in awareness
2. Cavaleros et al. [19]	N = 382 participants Demographics: - Gender: 174 women, 208 men - Race: 78 non-white, 304 white Setting: corporate	Theory: cognitive theory Modality: in-person Duration: One 2-day session Facilitator characteristics: NR Topics covered: racism, sexism, classism, prejudice, discrimination, and bias	Study design: Solomon four-group design (see page 55 of Cavaleros et al.) Outcomes measured: awareness, job satisfaction Instruments used: study survey, Minnesota Job Satisfaction Questionnaire	Assessment time points: pre- and 1 month post-training Analysis: 2 × 2 ANOVA Effect(s): null improvements in both awareness of diversity and job satisfaction
3. Celik et al. [20]	N = 30 participants Demographics: - Gender: 60% female - Age range: 25–60 Setting: health care	Theory: Deming cycle Modality: in-person Duration: Four 4-hour sessions Facilitator characteristics: NR Topics covered: levels of diversity (individual, symbolic, institutional), advantages of diversity for professionals, applying diversity in the institution	Study design: Quantitative one-group pretest–posttest and qualitative post interviews Outcomes measured: awareness, knowledge, behavior Instruments used: study survey, interviews	Assessment time points: pre- and post-training; 3 years after (interview) Analysis: statistical comparisons measured using Wilcoxon signed rank test Effect(s): significant improvements in awareness; null effects on knowledge, attitudes, and behavior
4. Chang et al. [21]	N = 3016 participants Demographics: - Gender: 61.5% male Setting: corporate	Theory: NR Modality: online Duration: One 1-hour session Facilitator characteristics: NR Topics covered: - Gender-bias training: gender-bias, stereotyping - General-bias training: gender, race, and sexual orientation bias, stereotyping	Study design: Random assignment to 1 of 3 experimental conditions: gender-bias training, general-bias training, or a control training Outcomes measured: attitudes and behavior (workplace decisions) Instruments used: Implicit Association Test, study survey (validated)	Assessment time points: pre- and post-training, 3 weeks, 6 weeks, 14 weeks (follow-ups for gender-bias group only) Analysis: controlled ordinary least-squares regressions with interactions terms Effect(s): significant effects on attitudes toward women, acknowledgment of gender bias, and intentions to engage in inclusive workplace behaviors toward women (non-US participants only). Null effects at follow-up. Significant effects in acknowledgement of racial bias and number of racial minorities recognized for excellence (US participants)
5. Davis et al. [22]	N = 615 participants Demographics: NR. All participants were first-year medical students at the University of California, San Francisco Medical School Setting: higher education	Theory: NR Modality: in-person Duration: One 2-day session Facilitator characteristics: received 1-day training to prepare Topics covered: health disparities, communication skills, privilege, power, socio-cultural history, and microaggressions	Study design: Mandatory one-group posttest Outcomes measured: knowledge, skills, attitudes Instruments used: study survey	Assessment time points: 4 weeks post-training Analysis: qualitative data analyzed by 2 authors Effects: Null results in improvement in knowledge
6. Ehrke et al. [23]	N = 62 participants Demographics: - Gender: 84% female [Experiment 1]; 77% female [Experiment 2] - Age: ages 19–31 Setting: higher education	Theory: ingroup projection model Modality: in-person Duration: 2 hours [Experiment 1]; 1 day [Experiment 2] Facilitator characteristics: NR [Experiment 1]; 2 experienced trainers (1 male, 1 female), prepared with training manual and “train the trainer” session [Experiment 2] Topics covered: gender stereotypes [Experiment 1], stereotypes, social categorization, discrimination, general diversity [Experiment 2]	Study design: 2 (condition: diversity or control) × 3 (time: 1, 2, 3) design [Experiment 1]; cross-over design with three measurement timepoints [Experiment 2] Outcomes measured: perceived diversity, self-typicality attitudes Instruments used: study survey	Assessment time points: pre- and post-training, 1 month [Experiment 1] Analysis: ANOVAs Effect(s): significant effect in perceived diversity of students in diversity-intervention condition, self-typicality, and positive attitudes toward gender outgroup; significant increase in perceived diversity and reduced ambivalent sexism, age bias, sexist attitudes. Null improvements in anti-gay attitudes.
7. Fischer [24]	N = 49 participants Demographics: - Average age: 22 - Gender: 78.4% female - Race: 44.7% non-white Setting: higher education	Theory: training theory and cross-cultural training models Modality: in-person Duration: weekly 2-hour lectures for 6 weeks Facilitator characteristics: trained facilitators Topics covered: cultural diversity, culture theory, dimensions of culture, cultural competence, intercultural training	Study design: Two-group pretest–posttest Outcomes measured: cultural intelligence, open-mindedness, cultural essentialism Instruments used: Multicultural personality Questionnaire, study survey	Assessment time points: pre- and post-training Analysis: within-subject ANOVA and regression analysis Effect(s): significant improvement in cultural essentialism, null effect in open-mindedness
8. Harrison-Bernard et al. [25]	N = 55 participants Demographics: - Gender: 67% female, 31% male, 2% no response - Race: 12% Black, 14% Asian, 7% Hispanic/Latinx, 5% no response, 62% White - Age: 5% 25–34, 29% 35–44, 38% 45–54, 19% 55–64, 7% 65+, 2% NR Setting: higher education	Theory: NR Modality: in-person Duration: One 3-hour workshop Facilitator characteristics: facilitators were required to use foundational materials to inform the training (see Table 1 on pg. 287 of Harrison-Bernard et al) Topics covered: importance of a diverse workforce, inclusion, stereotypes, implicit bias, cultural competency	Study design: one-group pretest–posttest Outcomes measured: awareness, knowledge, self-perceptions Instruments used: study survey	Assessment time points: pre- and post-training Analysis: paired t test Effects: Statistically significant increase in self-perceptions regarding the importance of improving diversity, recognizing bias and stereotypes, and cultural awareness. Statistically significant improvement in knowledge of terms (color-blind racial attitudes, tokenism, failure to differentiate, status leveling).
9. Holladay and Quinones [26]	N = 191 participants Demographics: - Gender: 84 men, 107 women - Race: 108 white, 8 Black, 21 Hispanic, 37 Asian, 17 other Setting: higher education	Theory: NR Modality: online Duration: One 1-hour session Facilitator characteristics: facilitators ranged in sex and race (White men, White women, Black men, Black women) Topics covered: benefits of diversity	Study design: quasi-experimental with 6 conditions; 2 × 2 × 2 focus (training focus: similarities vs. differences) × (trainer sex) × (trainer race: White vs. Black) Outcomes measured: reactions, knowledge, attitudes, self-efficacy Instruments used: study survey	Assessment time points: pre- and post-training, no data reported for follow-ups Analysis: mediation analysis Effect(s): Trainees who participated in a diversity training program focused on similiarities expressed significantly lower levels of expected backlash and performed better on the situational judgment test of conflict resolution skills, similarity-focused program significantly mitigated the effects of trainer race and sex on trainee reactions.
10. Hostager and De Meuse [27]	N = 302 participants Demographics: - Gender: 47% female, 53% male - Race: 94% white, 3% Asian or Pacific Islander, 2% Hispanic, 1% American Indian or Alaskan Native Setting: higher education	Theory: NR Modality: in-person Duration: one semester course Facilitator characteristics: NR Topics covered: - Course 1 (Diversity in the Workplace): workforce demographics and diversity, prejudice, discrimination, racism, legislation, gender diversity and harassment, diversity trends - Course 2 (Organizational Change and Development): strategic forms of organizational change, organizational change theory, employee empowerment, personality types - Course 3 (Strategic Management in Global Business Environment): maximizing company performance through knowledge of organizational structure, cultural, politics, and change, theories of ethics, stakeholder management, market trends.	Study design: Quasi-experiment with three conditions and pretest and posttest. Participants enrolled in one of three courses. Outcomes measured: reactions (emotional and behavioral), judgements, personal consequences, organizational outcomes Instruments used: study survey, Reaction-to-Diversity Inventory	Assessment time points: pre- and post-training Analysis: ANOVA, MANCOVA Effect(s): significant effect of courses describing both positive and negative aspects of diversity
11. Lindsey et al. [28]	N = 118 participants Demographics: - Gender: 61% female, 39% male - Race: 62% White, 12% Asian, 10% Black, 6% Hispanic, 10% other All incoming freshman at a southern university in 2006. Setting: higher education	Theory: diversity management theory and social identity theory Modality: in-person Duration: One 2-hour training Facilitator characteristics: NR Topics covered: varied dependent on group; - Goal setting: setting goals related to diversity training - Stereotype: discrediting stereotypes - Perspective taking: challenges faced by marginalized groups	Study design: 3x2 design. Participants randomly assigned to 1 of 3 diversity training methods (perspective taking vs. goal setting vs. stereotype discrediting) and 1 of 2 groups (African–Americans vs. LGBT individuals). Outcomes measured: attitudes, motivation, empathy Instruments used: 7-item Modern Racism Scale; 10-item Attitudes Toward Lesbians and Gay Male Scale; 10-item scale from International Personality Item Pool; study survey	Assessment time points: Pre and post-training; 0, 3, 6 months Analysis: regression analyses and SPSS macros Effect(s): Null results in the stereotype discrediting and goal setting conditions for both African-American and LGBT groups; statistically significant results in the perspective taking condition for the LGBT group.
12. O’Connor et al. [29]	N = 44 participants Demographics: Not reported (NR); all participants were faculty and staff at University of Washington School of Nursing Setting: higher education	Theory: NR Modality: in-person Duration: One 3-day training Facilitator characteristics: professional social justice workers (no specific training given) and DEI faculty mentors who received a 5-hour training on inclusive teaching practices Topics covered: awareness building, creating inclusive environments for diverse learners, managing difficult conversations on racism, incorporating DEI into curriculum, normalizing and handling mistakes with humility	Study design: one-group pretest–posttest Outcomes measured: behavior change Instruments used: study survey, University of Washington course evaluation form	Assessment time points: pre- and post-training Analysis: nonparametric Kolmogorov-Smirnov test for equality of distributions, paired t tests Effects: statistically significanta increases in DEI-related teaching self-efficacy
13. Powell et al. [30]	N = 74 participants Demographics: - Gender: 65% female, 34% male, 1% prefer to self-describe (queer) - Race/ethnicity: 1% American Indian/Alaskan Native, 31% Asian/Asian American, 22% Black, 20% Hispanic/Latino/Spanish origin, 22% non-Hispanic White, 5% other Setting: higher education	Theory: self-determination theory Modality: in-person Duration: monthly 2-hour sessions over the course of 10 months Facilitator characteristics: no training received, facilitators were selected from an applicant pool Topics covered: importance of diversity, implicit bias, microaggressions, emotional intelligence, growth and fixed mindsets, privilege and allyship in the workplace, promoting underrepresented trainees in leadership, principles of compassionate leadership, “inclusive diversity” culture, structural racism in medicine, imposter syndrome, social justice in medicine	Study design: LEAD program (longitudinal, single institution program) over 3 cohorts, from 2017 to 2018, 2018 to 2019, and 2019 to 2020. Outcomes measured: self-efficacy, actual and planned behavior change Instruments used: IRB-approved, de-identified retrospective pre- and post-surveys	Assessment time points: Pre and post-training; annual follow-ups (no data reported) Analysis: descriptive statistics, 2-tailed paired t tests and qualitative analysis. Hand-coded using content analysis Effect(s): statistically significant improvement in self-efficacy for all LEAD sessions’ learning objectives
14. Rawski and Conroy [31]	N = 194 (N = 53 in Experiment 1 and N = 141 in Experiment 2) Experiment 1 demographics: - Gender: 74% female - Race: 92% White - Average age: 41.5 Experiment 2 demographics: - Gender: 75% female - Race: 81% White - Average age: 43.3 Setting: higher education	Theory: social identity theory Modality: in-person Duration: - Experiment 1: voluntary attendance of 14, one 4-hour diversity training workshops - Experiment 2: one 1-hour session Facilitator characteristics: described as subject matter experts Topics covered: Experiment 1: cultural competency, microaggressions, unconscious bias, power, privilege, and oppression; Experiment 2: sexual harassment	Study design: Quasi-experiment with two conditions. Experiment 1 was a multi-session, general diversity training; Experiment 2 was a mandatory, one-session sexual harassment training. Outcomes measured: knowledge, motivation Instruments used: study survey, Organizational Identification Scale	Assessment time points: pre- and post-training Analysis: logistic regression analysis Effect(s): statistically significant higher effect of learning in mandatory training [Experiment 2] but better reactions in voluntary training [Experiment 1]
15. Tillman et al. [32]	N = 21 participants Demographics: - Gender: 23.8% male, 71.4% female, 4.8% nonbinary - Race: 14.3% Asian, 19.1% Black, 4.8% Pacific Islander, 57.1% White, 4.8% multiracial - Age: 4.8% 18-24, 90.48% 25–29, 0% 30–34, 4.8% 35+ Setting: health care	Theory: NR Modality: NR Duration: monthly 1-hour sessions from August 2020 to June 2022 Facilitator characteristics: no training received, session speakers were sought out on the basis of lived experience Topics covered: Race and ethnicity (identity, implicit and explicit bias, race and economics, racial representation in health professions, micro and macro aggressions, racial equity), gender/gender identity (LGBTQI+ definintions, safe zone project training, health disparities for trans individuals, gender differences in healthcare experiences and health professional training), religion (myths and truths around religion-driven medical decision making); social determinants of health.	Study design: longitudinal DEI curriculum Outcomes measured: knowledge, awareness, satisfaction Instruments used: study survey	Assessment time points: post-training only Analysis: NR Effect(s): null improvements in knowledge and awareness

NR, not reported.

^aStatistically significant refers to P < .05.

Figure 1

Flow diagram of study article inclusion and exclusion: DEI training studies.

DEI training studies: Participant, setting, and training characteristics

The number of participants in DEI training studies in our review varied widely from 30 to 3016 with 7 studies (46.7%) having sample sizes of 100 participants or less. Most studies (66.7%) had participants that were majority (60% or more) female, and 46.7% of studies had participant samples that were majority (60% or more) White. The majority of DEI training studies took place in higher education (73.4%), followed by corporate (13.3%) and health care (13.3%) settings. The majority of trainings (80%) were conducted in-person by trained facilitators, with two studies’ training activities occurring online. No studies utilized a hybrid delivery approach.

Studies varied widely in the duration, frequency, and curriculum of DEI training. More than half of the studies (53.3%) used only one-time training sessions that ranged from 1-hour to 3-day sessions. The remaining studies (46.7%) implemented multiple training sessions spaced out over time, ranging from 6 weeks to 10 months. Two-thirds of DEI training studies (66.7%) identified one or more theories to inform their training, including social identity theory [33], self-determination theory [34], and diversity management theory [35]. See Supplementary Appendix 1 for a complete list and summary of theories. Training topics varied, with the most common topic being bias (discussed in 42.9% of DEI training studies), stereotypes (26.7%), and cultural competency (20%), among others. Nine of the 15 DEI studies (60%) included training on topics other than race and racism, most commonly sexism, classism, homophobia and transphobia.

DEI training study design and measures

Very few studies (20%) utilized randomized conditions, and even fewer (6.7%) utilized a control group. All studies administered a pre and immediate post-training survey to measure outcomes. Of the four studies that collected longer-term follow-up data, only two reported results (follow-up ranged from 1 month to 3 years). Measured outcomes varied widely across studies. The most frequently measured outcome was content knowledge (53.3%), with most studies assessing knowledge, awareness, and attitudes (e.g. motivation, self-efficacy). Seven out of the 15 studies also measured participants’ skills or actionable change, such as conflict resolution or decisions. Over half of the studies (53.3%) developed their own study survey for assessing outcomes, with the other half using established inventories including the Race Implicit Association Test [34] and the Organizational Identification Scale [36].

DEI training impact

Twelve out of the 15 DEI training studies (80%) reported statistically significant results for one or more measured outcomes. In general, several studies reported significant short-term changes in knowledge and awareness, such as improvements in knowledge of specific terms [25], cultural knowledge [18], cultural essentialism [24], enhanced self-awareness [20], and perceived importance of recognizing biases and improving diversity [25, 27]. Four studies (26.7%) reported significant changes in participants’ attitudes, such as increased participant self-efficacy (confidence) to act in accordance with, amplify, and teach DEI principles [29, 30], and improved positive feelings toward gender outgroups [21, 23]. Three studies (20%) reported significant differences in participants’ skills or behavioral changes, including improved conflict resolution [26], increased familiarity with skills needed to work in culturally and racially diverse teams [18] and fostered inclusive environments [25], and being less likely to exhibit less racial prejudice [18].

Antiracism training search results

The literature search for antiracism training studies yielded a total of 149 nonduplicate articles (Figure 2). Of those, N = 94 articles were excluded based on the title review, N = 39 articles were excluded during abstract review, and N = 8 articles were excluded during full-text review, resulting in a total of eight articles eligible for inclusion. Table 2 presents a summary of characteristics across antiracism training studies included in this review.

Table 2

Characteristics and results of antiracism training studies published between 2000 and 2022 (N = 8)

Study article	Sample and setting characteristics	Training characteristics	Measures	Analysis and results
1. Abramovitz and Blitz [37]	N = 875 participants Demographics: - Gender: 77.7% female, 22.3% male - Race: 57.8% White, 38.8% person of color, 3.5% other - Age: 31.3% 18–35, 36.8% 36–55, 31.9% 56+ Setting: social work	Theory: participatory action model Modality: in-person Duration: one 2.5 day workshop Facilitator characteristics: facilitated by multicultural team Topics covered: history of racism in USA, individual attitudes and knowledge, internalized oppression and privilege, institutional racism	Study design: one-group pretest and follow-up Outcomes measured: knowledge and attitudes, engagement in job-related racial equity activities, view of organizations’ progress toward racial equity Instruments used: study survey	Assessment time points: post-training (2–7 years after) Analysis: frequency distributions reported, nonparametric tests used for analysis Effect(s): statistically significant improvements in knowledge, attitudes, and engagement.
2. Johnson et al. [38]	N = 462 participants Demographics: - Gender: 78% women, 22% men - Race: 57% White, 36% Black, 6% other - Age: range of 21–83, average 42 Setting: community service providers	Theory: cultural competency training theory Modality: in-person Duration: 1 training over 2.5 days Facilitator characteristics: 20+ years experience in antiracism education and organizing Topics covered: foundations of race and racism, impact of racism, approaches to community organizing, leadership development, accountability to communities, creation of networks, learning from history, combating racism methods	Study design: one-group, pretest–posttest Outcomes measured: anticipated transfer of knowledge and skills, knowledge of key concepts, attitudes toward race Instruments used: study survey, Color-Blind Racial Attitudes Scale	Assessment time points: pre- and post-training Analysis: independent samples t-tests and one-way ANOVAs, paired samples t-tests and repeated measures ANOVAs, bivariate correlations Effect(s): significantly higher satisfaction among participants of color than white participants; significant increase in knowledge; significant decrease in color-blind racial attitudes.
3. Kingett et al. [39]	N = 109 participants Demographics: - Gender: 74% male, 26% female Setting: secondary education	Theory: NR Modality: online Duration: One day (# hours not specified) Facilitator characteristics: training facilitated by Show Racism the Red Card, antiracism educational charity Topics covered: history and effects of racism, stereotypes, racism prevention, prejudice, discrimination	Study design: one-group pretest–posttest Outcomes measured: knowledge (recognizing racism, perceptions of racial attitudes), awareness, confidence, acceptance/rejection of discrimination, response to racist incidents Instruments used: study survey	Assessment time points: pre- and post-training Analysis: NR Effect(s): statistically significant improvement in knowledge, awareness, confidence, and acceptance of discrimination. Null improvement in response to racist incidents
4. Loe et al. [40]	N = 77 participants Demographics: NR Setting: health care	Theory: Kern’s 6 steps of curriculum development, self-determination theory Modality: in-person Duration: One 3-hour workshop Facilitator characteristics: NR Topics covered: racial awareness, child development, impact of racism on child development, behavior, and health, addressing racism	Study design: one-group pretest–posttest Outcomes measured: knowledge, attitudes, skills (comfort and preparation for counseling patients, offering resources, etc.) Instruments used: study survey	Assessment time points: pre- and post-training Analysis: paired t test Effect(s): significant improvement in all measured outcomes
5. Martin and Baxter [41]	N = 50 participants Demographics: - Race: 100% White - Gender: 24% male, 76% female - Age: range from 30 to 60, average age of 47 Setting: secondary education	Theory: 5 stage model of White Racial Identity Development Modality: in-person Duration: 3 hours/week over 14 weeks Facilitator characteristics: NR Topics covered: principles of antiracism, applying antiracism	Study design: one-group pretest–posttest Outcomes measured: stages of the White racial identity development model (contact, disintegration, reintegration, pseudo-independent, autonomy) Instruments used: White Racial Identity Attitude Survey, study survey	Assessment time points: pre- and post- training Analysis: confirmed reliability with difference approach Effect(s): significant decrease in disintegration (positive result), significant improvement in pseudo-independence and autonomy stages. No changes in contact and reintegration stages.
6. Santoro et al. [42]	N = 42 participants Demographics: - Race: 55% White, 32% Black, 9% Asian, 4% other - Gender: 91% female - Age: average age 30 Setting: healthcare (nutrition)	Theory: cultural humility Modality: in-person Duration: one, 3-hour training Facilitator characteristics: NR, led by Afro-Latinx antiracism educator and White midwife with public health research experience Topics covered: racism, privilege, white supremacy, implicit bias and cultural humility, socio-ecological models of behavior change	Study design: one-group pretest–posttest Outcomes measured: awareness (of systematic racism, bias, privilege), confidence, skills (identifying and addressing racism) Instruments used: study survey	Assessment time points: pre- and post-training, 6 months post-training Analysis: 2-sided z-tests Effect(s): significant and sustained improvements in awareness and skills, significant but unsustained improvements in confidence
7. Webb and Sergison [43]	N = 48 participants Demographics: NR Setting: child health services	Theory: NR Modality: in-person Duration: NR Facilitator characteristics: NR Topics covered: stereotyping, empathy, communication, attitudes, cultural conditioning, how racism affects social services	Study design: one-group pretest–posttest Outcomes measured: cultural competence Instruments used: study survey	Assessment time points: pre-training; 2–7 years Analysis: NR Effect(s): null improvements
8. Yang et al. [44]	N = 9 participants Demographics: of initial 12 applicants, there were 7 White, 1 Black, 3 Hispanic Setting: education	Theory: Dialectical Behavior Therapy (DBT) Modality: online Duration: one 3-hour training over 2 days Facilitator characteristics: NR Topics covered: history of racism, impact of racism on students, antiracism, DBT skills	Study design: one-group pretest–posttest Outcomes measured: confidence, knowledge, ability to teach antiracism Instruments used: study survey, Black-White Implicit Association Test (optional)	Assessment time points: Pre and post-training, 6 months (no data reported) Analysis: NR Effect(s): null improvements in confidence, knowledge, ability to teach antiracism.

NR, not reported.

^aStatistically significant refers to P < 005.

Figure 2

Flow diagram of antiracism training study article inclusion and exclusion.

Antiracism training: Participant, setting, and training characteristics

The number of participants in antiracism training studies in this review varied from 9 to 875, and 6 of the 7 studies reported having at least 55% White participants. Of the five studies that reported gender demographics, four had participant samples that were 75% or more female participants. Studies of antiracism training took place in the following settings: secondary education (N = 3 studies, 37.5%), health care (N = 3 studies, 37.5%), social work (N = 1 study, 12.5%), and community service providers (N = 1 study, 12.5%). Trainings were conducted in-person (N = 6 studies, 75%) or online (N = 2 studies, 50%). No studies utilized a hybrid approach to delivering trainings.

Seven out of the eight antiracism training studies utilized a one-time training session. Antiracism trainings lasted 1 day on average and ranged from 3-hour to 3-day sessions. The trainings covered a variety of topics such as bias (N = 4 studies, 50%), stereotypes (N = 2 studies, 25%), and cultural competency (N = 2 studies, 25%), among others. Five studies explicitly identified theories or frameworks as informing the antiracism training, including cultural competency training theory [45], Kern’s six-Step Approach to Curriculum Development for Medical Education [46], dialectical behavioral therapy [47], the five-stage model of white racial identity development [48], cultural humility [49], and self-determination theory [34].

Antiracism training study design and measures

All studies of antiracism trainings utilized a single group pretest/posttest study design with participants self-selecting to participate. None utilized a randomized design or included control or comparison groups. Participants in all studies completed self-reported, self-administered surveys. Five out of the eight antiracism training studies used a nonvalidated study survey to assess outcomes, with the remaining three studies using the Race Implicit Association Test [50], the White Racial Identity Attitudes Scale [51], and the Color-Blind Racial Attitudes Scale [52].

The most frequently measured outcome across studies was attitudes (N = 5 studies, 62.5%), followed by knowledge (N = 4 studies, 50%), behavior (N = 3 studies, 37.5%), and cultural competence (N = 1 study, 12.5%). Of the two studies that collected and reported follow-up data, one reported data at 6 months post-training [38] and one reported data from participants 2–7 years post-training [37].

Antiracism training impact

Five out of seven of the antiracism training studies reported statistically significant results for at least one measured outcome, including improvements in knowledge and awareness [37–40,42] or attitudes (e.g. decreased color-blind attitudes, more open attitudes about racial issues, privilege, and institutional discrimination, increased confidence in applying antiracism knowledge in the workplace and teaching antiracism awareness to others [42, 44]) following completion of trainings. Three studies reported significant changes in skills, such as improved professional development skills [38] and cultural competence in serving diverse communities [43]. Only one study reported significant differences in behaviors or actions in the workplace (e.g. becoming familiar with the customs of local Black communities, changing pediatric posters to be more welcoming to diverse clients) [43].

Patterns across DEI and antiracism training studies

Of the DEI studies that used multi-session trainings, 85.7% reported significant improvements in one or more measured outcomes compared with 62.5% of DEI training studies that used one-time sessions. With respect to antiracism training studies, 100% of those that used multi-session trainings reported significant improvements in one or more measured outcomes, compared with 83.3 of studies that used one-session trainings. Of the 14 studies that were theory informed, 85.7% reported improvement in one or more measured outcomes compared to 66.6% of the nine studies that did not report being informed by a specific theory. Nine of the 23 studies (39.1%) had long-term follow-up timepoints (longer than one month), ranging from 6 weeks to 7 years. Of these studies, only six (66.7%) reported follow-up data and three (33.3%) reported continued improvement at follow-up. All four of the studies that used randomized designs showed significant improvements in at least one measured outcome [21, 23, 26, 53].

Discussion

This study systematically reviewed the characteristics, measures, and outcomes of DEI and antiracism training studies published over the past two decades. Of the 23 studies included in this review, 15 articles (65.2%) utilized a one-time training session and 19 were delivered in-person (82.6%). Over 70% of studies had majority female participants. Sample sizes ranged widely from 5 to 3016 participants, with less than half (N = 9 studies, 39.1%) having 100 or more participants. Training content, objectives, theoretical framework applied (if any), measures, and impact varied widely across studies. Few studies utilized randomized designs to test the training (N = 4 studies, 17.4%) or collected follow-up data (N = 6 studies, 26.1%). The majority of studies took place in education (N = 14 studies, 60.9%) or health care (N = 4 studies, 17.4%). Over half (60.9%) of studies in this review reported statistically significant changes in outcomes of interest, with improvements most often observed in participants’ self-reported knowledge, attitudes, and awareness, and less frequently observed in participants’ skills or behavior change. We noted certain trends, such as studies that used multi-session trainings, were theory informed, and utilized randomized designs had higher rates of yielding improvements in one or more outcomes; however, our sample size of studies was too small to make further analytic comparisons. Findings are consistent with prior research demonstrating wide variation in characteristics and effects of DEI and antiracism trainings [13–16, 54].

Results also highlight that the majority of research studies on DEI and antiracism trainings in this review are in the earlier phases of the translational research spectrum (T1: intervention development research phase and T2: efficacy research to determine intervention impact) [4, 5]. This review also included eight studies (five DEI and three antiracism trainings) published in the aftermath of the racial reckoning in the USA, a period when attention to community engagement and community-centered approaches was emphasized. Of these eight recent studies, only three mentioned community engagement to varying degrees. One study mentioned that facilitators who were knowledgeable about community engagement [25]; another described participants’ interest in the community post-training [30]; and a third described a community partnership with the local school system and discussed the community need for antiracism training [44]. It is worth noting that due to the timeline of academic peer review and pandemic-related delays, these studies may have been initiated before the summer of 2020, and there may be relevant studies in progress or currently under review that have utilized community-centered approaches for the research.

Overall, few studies in this review regardless of publication date reported utilizing community-engaged or community-centered approaches, indicating that much opportunity exists to integrate communities into all aspects of the research, including research question conceptualization, intervention design and delivery, participant recruitment and retention, data collection measures and methods, data analysis and interpretation, dissemination of findings, and adapting and scaling interventions. Based on our review, we propose several strategies to advance the field of DEI and antiracism training through research and practice.

Practice recommendation 1

Develop and deliver longitudinal trainings so that concepts and skills are able to be built upon over time. To enhance efficacy, both the frequency and duration of trainings can be extended beyond one-time sessions.

Practice recommendation 2

Deepen the curriculum focus beyond individual knowledge and attitudes to include skill building and organizational development to support and sustain individual, interpersonal, and organizational change [55].

Practice recommendation 3

Prioritize skills and behavioral and organizational change as outcomes of interest, using validated or established instruments where possible. While demonstrating improvements in individual knowledge, awareness, and attitudes can be useful in assessing immediate gains from training participation, changes in these domains alone need to be accompanied by changes in skills and actual behavior change (individual, interpersonal, or organizational) in order to advance DEI and antiracism in the workplace.

Practice recommendation 4

Use validated measures of DEI and antiracism where relevant. Studies in our review that reported using validated instruments in their outcome assessments include the Reaction-to-Diversity Inventory [56], Ambivalent Sexism Inventory [53], Race Implicit Association Test [34], and the Modern Racism Scale [57]. These four instruments have demonstrated reliability and validity in the studies included in our review as well as others [53,58–60]. Other validated instruments that practitioners and researchers can consider using include the Anti-Racism Behavioral Inventory [61], the Diversity Engagement Survey [62], and the Political Skill Inventory [63].

Research recommendation 1

To enhance equity, integrate community-engaged approaches into the research design, implementation, evaluation, and dissemination. Examples include inviting community experts as co-leads or co-investigators on the research and including communities most impacted by discrimination to provide input and guidance.

Research recommendation 2

To improve generalizability, future studies can recruit more representative samples (e.g. participants reflect organizational, sector, or regional demographics) and test trainings across a variety of sectors. The majority of studies in this review were conducted in higher education or health care, where female employees are overrepresented [64].

Research recommendation 3

Studies of DEI and antiracism trainings can implement more rigorous study design methods to better identify best practices. Specifically, using a randomized controlled design as utilized by Ehrke et al., Chang et al., Holladay and Quinones, and Lindsey [21, 23, 26, 28] allows researchers and practitioners to better isolate the effect(s) of the training being tested, as this kind of design is the strongest for controlling for measured and unmeasured confounding. Additionally, utilizing comparison groups that receive some kind of DEI and/or antiracism training may be more ethical and equitable, so that all participants have some exposure to DEI and/or antiracist content (vs. an experimental group receiving training and a control group receiving no training or resources at all). For example, one study [21] randomly assigned participants to one of three experimental conditions (gender-bias training, general-bias training, or control). Another study used a 2 × 2 × 2 design, where participants were randomly assigned to three conditions: training focus (similarities or differences), trainer gender (male or female), and trainer race (White or Black) [26]. Another study [28] used a 3 × 2 design, where participants were randomly assigned to one of three diversity training methods (perspective taking, goal setting, or stereotype discrediting) and one of two groups of focus (African–Americans or the LGBT community). These methods can be implemented during the pilot phase of new training programs.

Research recommendation 4

To measure change over time, studies of DEI and antiracism trainings can include intermediate and long-term follow-up assessments. As only six studies in this review (26.1%) collected and reported follow-up data, and of those, only one study focused assessing program satisfaction rather than intervention impact at the long-term follow-up [20], longer-term effects associated with such trainings remain not well understood. To address this gap, studies recommend an implementation of longitudinal design in diversity training interventions, which would improve data collection and allow researchers to ensure such interventions meet their goals and sustain improvements over time [65–67].

The above recommendations may facilitate the movement of DEI and antiracism training studies further along the research translational continuum. Study findings and recommendations should be interpreted within the context of the following limitations. The relatively small number of studies in our review limited our ability to conduct statistical comparisons of efficacy across study characteristics (e.g. multi-session vs. one-time training sessions, theory-informed vs. non-theory-informed interventions, randomized vs. non-randomized design). Of the studies that reported participant demographic data, several included majority female, majority White participants, which may limit the generalizability of findings to other populations. Additionally, studies did not include information on their overall workforce population; thus, participant representation and study findings may not necessarily be representative of the demographics of the organizations and industries from which the studies were conducted.

Study designs were moderate-to-weak with respect to causal inference. Combined with high variation in training content, duration, instruments, and outcomes assessed, it was not possible to determine which training characteristics are critical for improving outcomes. Studies also lacked uniformity in reporting the duration of training sessions (e.g. some only reported day-long sessions while others reported duration in hours) and in defining terminology. This study was limited to peer-reviewed research available through search engines and does not include white reports or publicly unavailable research. Depending on the organization, some recommendations may be more feasible than others. Lengthening the duration or frequency of training can represent a challenge in terms of time and resources. Organizations may have limited diversity among employees, thus limiting how representative participants may be with respect to demographics. More rigorous evaluation designs such as randomized conditions or comparison groups may be difficult to implement.

In conclusion, findings from this systematic review highlighted conceptual and methodological directions to enhance the delivery and efficacy of DEI and antiracism training interventions in the workplace. Areas for future research include exploring in-depth the intervention design elements that are critical to facilitate improved outcomes, such as examining the mechanisms and efficacy of different training durations, application of various theories and frameworks, and topics and skill sets addressed using strong evaluation designs and longer follow-up time periods. Future community-engaged research across the spectrum of translational research stages is needed to understand the optimal characteristics of DEI and antiracism trainings and how to best integrate these trainings within organizations for impact, sustainability, and scalability.

Acknowledgements

M.L.W is supported in part by National Institutes of Health (NIH) National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) Grant # R01DK120713 (PI: Wang). The study funder did not have any role over any of the following activities: study design, collection, analysis, or interpretation of data; writing the report; the decision to submit the report for publication. The content of this manuscript is solely the responsibility of the authors and does not necessarily represent the views of the NIH.

Conflict of Interest Statement

The authors have no conflicts of interest to disclose.

Author Contributions

M.W. conceived of the research questions and study design. A.G. and M.S. conducted the literature searches and synthesized data. All authors contributed to drafting and revising this manuscript and approved the final version.

Ethical Approval

This article does not contain any studies with human participants performed by any of the authors. This article does not contain any studies with animals performed by any of the authors.

Informed Consent

This study does not involve human participants and informed consent was therefore not required.

Transparency Statements

This study was a systematic review of peer-reviewed studies; formal registration of the study and analysis plan was not applicable. There is no analytic code associated with this study. De-identified data is not applicable for this study. Methods and materials used to conduct this systematic review are stated in the manuscript.