Current approaches for mitigating racial prejudice in organizational settings have shown limited effectiveness, underscoring a critical need for new strategies (Hsieh et al., 2021; Onyeador et al., 2024; Paluck et al., 2021). Furthermore, there are very few empirically validated tools for reducing racial prejudice and discrimination in organizational settings. This is a severe limitation in the field, and scholars stress that prejudice reduction interventions need to be tested in actual workplace contexts and preferably with controlled designs (Paluck & Green, 2009; Hsieh et al., 2021). In addition, most anti-discrimination interventions in organizations consist of lectures, which research suggests is less efficient than more behaviorally oriented methods (Paluck et al., 2021). There is also a general lack of prejudice reduction interventions that address implicit prejudices; that is, the less conscious kind of prejudiced thinking which is understood as a core process of modern forms of racism and discrimination (Akrami & Ekehammar, 2005; Paluck & Green, 2009). To further compound the issue, the persistence of racial prejudice and discrimination in the workplace is linked to numerous negative outcomes, including heightened stress, lower job satisfaction, and increased exposure to workplace bullying for minority employees. These conditions not only harm individual well-being but also contribute to long-term physical and mental health problems, such as cardiovascular disease and depression. Without more effective and empirically validated interventions, these negative outcomes will continue to undermine both employee well-being, organizational productivity and social equality. Therefore, addressing these challenges is essential for fostering inclusive, healthy, and high-performing workplaces.
Emerging research suggests that third-wave cognitive behavioral methods, such as Acceptance and Commitment Therapy (ACT), compassion, and mindfulness training, offer promising results in various settings. However, these previous approaches primarily focused on empathy, perspective-taking, or mindfulness alone. The present study builds on this by integrating these components with an explicit emphasis on cultivating psychological flexibility and compassionate action simultaneously, which - to our knowledge - has not been extensively tested in organizational settings. Furthermore, studies on occupational stress reduction and self-efficacy enhancement have validated the efficiency of such approaches in organizational contexts (Brassey et al., 2020; Unruh et al., 2022). These methods aim to shift the relationship individuals have with their thoughts and feelings. Rather than altering the content of thoughts (biases and prejudices) directly, participants learn to be mindful and accept that these thoughts exist, but recognize that they do not need to act on them (Lillis & Hayes, 2007). This awareness fosters greater flexibility in choosing actions guided by their values (Lillis & Hayes, 2007). In summary, previous research has identified some promising tools for reducing prejudice and discrimination based on ACT and CFT that potentially can be used in organizational contexts, but there is a need to test these methods outside the laboratory and in other populations than students.
Modern racism, as it is marked by subtle and implicit prejudice, poses challenges both when it comes to detection and intervention. Indeed, such prejudices may coexist with a professed adherence to humanistic values of equality and justice, complicating efforts to address it (Akrami et al., 2000; McConahay, 1986). Modern racism comprises the belief that racism and discrimination are no longer significant societal problems and that efforts to promote equality are unnecessary or even unjust toward the majority population (Bonilla-Silva, 2017). This underscores the need for interventions that address these subtle forms of prejudice through approaches like ACT and CFT, which can help individuals recognize and respond to these biases more effectively. Furthermore, modern forms of prejudice are not only linked to covert behavior and resistance to anti-discrimination efforts but are also challenging to address due to the cognitive processes underlying them. Prejudice often arises from social categorization and stereotyping, which are automatic ways of managing complex social interactions (Cuddy et al., 2009; Fiske et al., 2002; Kite & Whitley, 2016). These automatically formed mental categories enable us to simplify and navigate social complexities by categorizing individuals into groups, often leading to stereotypical judgments (Fiske & Taylor, 1991).
When it comes to reducing prejudices, approaches such as thought inhibition and suppression have proven largely ineffective and sometimes counterproductive (Hayes & Wilson, 1996; Rehfeldt & Hayes, 2000; Looper et al., 2010). Indeed, research suggests that implicit biases are stronger predictors of behavior than the extent to which one deliberately attempts to eliminate prejudice (Hussey et al., 2015). Consequently, alternative methods that go beyond mere instructions or thought suppression are likely important for achieving long-term reductions in prejudice.
So, what does research tell us regarding the effects of different types of intervention when it comes to reducing prejudice? The most frequently studied method is intergroup contact, in which different racial or ethnic groups work together toward a common goal (Dovidio et al., 2017; Hsieh et al., 2021; Pettigrew & Tropp, 2006). However, the results of this method vary: If Allport’s optimal conditions are met (equal status, common goal, cooperation, and norm support), prejudice reduction from intergroup contact shows moderate effect sizes (Pettigrew & Tropp, 2006), whereas under less optimal conditions – for instance, when the contact is experienced as negative – it leads to increased biases (Hewstone et al., 2002; Kotzur & Wagner, 2021). In Hsieh et al.’s (2021) meta-analysis, they compared the effects of five additional types of interventions: awareness (cognitive training combined with lectures), categorization (refocusing on superordinate categories instead of narrow ingroups), perspective taking (putting oneself in others’ shoes), social norms (finding role models), and perceived variability (increasing awareness of differences between outgroup members). Here, they found the greatest effect size for contact-based prejudice reduction (moderate effect sizes), followed by awareness-based (also moderate effect sizes). However, methods using a combination of different approaches demonstrated stronger effects. Additionally, a meta-analysis by Hellebrand (2017) showed that participation in experiential interventions targeting awareness of prejudice was more effective than lectures. Given this, the present study offers a novel integration of ACT and CFT, focusing on experiential exercises that foster both psychological flexibility and compassion. This combined approach aims to differentiate itself from previous empathy-based or mindfulness interventions by addressing both the emotional and cognitive components of prejudice while guiding behavior according to values of equality, compassion, and inclusion.
As described above, changing negative stereotypes and attitudes is challenging, but research indicates that methods involving perspective taking, psychological flexibility, and compassion might be a promising way of dealing with prejudice-driven behavior without directly engaging in a struggle to change prejudiced thought patterns (Lillis & Hayes, 2007; Brassey et al., 2020; Unruh et al., 2022). The present study will further explore the idea put forth by Lillis and Hayes (2007) that such interventions might be useful when attempting to tackle prejudice-related behaviors, especially since prejudice contains a cognitive element. They argued that, whereas the cognitive aspects of psychopathology may appear superficially different from prejudice, there is a deep-rooted similarity between the two: they both involve deeply ingrained cognitions which influence behaviors. For example, a racist thought is similar to depressiogenic thoughts. Moreover, since behavior therapies teach skills to deal with problematic cognitions in people with psychological suffering or disorders, these therapies might also be relevant to dealing with prejudiced thinking. In line with this suggestion, Levin et al. (2016) found that low empathic concern, low perspective-taking, and psychological inflexibility independently predict generalized prejudice, and interventions enhancing these traits have proven effective in reducing outgroup biases (Christofi & Michael-Grigoriou, 2017; Finlay & Stephan, 2000; Galinsky & Moskowitz, 2000; Todd et al., 2011).
The intervention in the present study uniquely combines ACT and CFT techniques to address prejudiced thoughts. Unlike traditional empathy or perspective-taking approaches, this intervention specifically aims to create internal contextual shifts that promote awareness and acceptance of prejudiced cognitions while simultaneously encouraging compassionate thinking and action, making it distinct from other existing methods. The intervention is inspired by both Acceptance and Commitment Therapy (ACT) and Compassion-Focused Therapy (CFT), which are newer forms of behavior therapy that take acceptance, mindfulness, and compassion-based approaches to behavior change. ACT was developed by Steven Hayes and colleagues (Hayes et al., 1999), and focuses on training one’s attention through mindfulness skills and increasing acceptance through experiential exercises in which individuals observe, become aware of and accept their thoughts, emotions, memories, and physical reactions. Furthermore, participants are encouraged to identify their values and how to live and behave in accordance with those values, also in the presence of aversive emotions and cognitions.
CFT was developed by Gilbert (2009) and focuses on mindfulness skills, but not expressly to increase attentional control. Instead, mindfulness is used to train a compassionate mindset regarding negative thoughts and feelings directed at others or oneself, which in turn helps to transform problematic patterns of emotion and cognition. CFT teaches inner warmth, safety, and soothing to help train the mind to develop compassion for others and oneself (Gilbert, 2009). Compassion increases the ability to feel connected with humanity as a whole (Gilbert, 2017) and when people are in compassionate states, their tendency to view others as a threat decreases (Mikulincer & Shaver, 2017) and solidarity and positive emotions towards outgroups increase (Chow et al., 2023; Zheng et al., 2023; Özkan et al., 2024). Compassion decreases the drive to compete with others and the need to defend social ranking (Kolts, 2016).
In the realm of prejudice reduction, Lillis and Hayes (2007) conducted a pilot study that examined the impact of psychological flexibility training on enhancing bias awareness and fostering positive behavioral intentions towards inclusion. Their findings from a group of 32 students demonstrated that a 75-minute intervention effectively increased awareness of prejudice and positive behavioral intentions immediately after and one week following the intervention. There are also studies with mindfulness and compassionate interventions that have proved effective in reducing prejudice (e.g., Berger et al., 2018; Kang et al., 2014; Parks et al., 2014) and improving intergroup relations (Chow et al., 2023; Oyler et al., 2022; Özkan et al., 2024), which led the present study to add compassion exercises.
Thus, the present study builds on the model suggested by Lillis and Hayes (2007) but extends and develops it by integrating the methods from Acceptance and Commitment Therapy (ACT), with Compassion-Focused Therapy (CFT). This integration fosters mindful awareness and acceptance of existing prejudiced cognitions while promoting compassionate thinking and action. These interventions, rooted in acceptance, mindfulness, and compassion-based methodologies, aim to cultivate psychological flexibility and compassion. The rationale for combining ACT and CFT is that ACT helps individuals become more open and flexible with their thoughts and feelings, while CFT fosters a sense of safety and compassion towards oneself and others. Together, they create a comprehensive approach to behavior change. This dual focus on psychological flexibility and compassion is central to the intervention, as it aims to address both the cognitive and emotional components necessary for comprehensive personal and organizational growth, as has been requested to improve diversity training (Onyeador et al., 2024). Overall, the combined use of ACT and CFT in this context aims to support individual growth while also enhancing collective organizational culture, promoting more inclusive and respectful workplaces.
This study aimed to explore the effects of interventions drawing from both ACT and CFT on participants working in both public and private organizations. In doing so, the study was designed to test several hypotheses of the effects of the psychological flexibility and compassion training compared to a comparison group receiving lectures on discrimination, diversity and inclusion. The first hypothesis under investigation was that the flexibility and compassion training would lead to a greater increase in compassion toward others (H1). This hypothesis was based on findings that compassion training, particularly CFT, has been shown to improve compassionate responses and decrease defensive reactions (Gilbert, 2017; Mikulincer & Shaver, 2017), which is crucial in fostering more inclusive workplace behaviors. Furthermore, given that previous research has highlighted that perspective-taking exercises are effective in reducing biases and improving empathy, which the present intervention integrates with ACT to enhance these effects (Finlay & Stephan, 2000; Todd et al., 2011), we also expected the intervention to lead to a greater increase in empathic perspective taking (H2). In addition, since ACT aims to increase psychological flexibility, allowing individuals to respond to prejudiced thoughts without acting on them (Lillis & Hayes, 2007; Levin et al., 2016), and since psychological flexibility is essential for promoting behavioral changes aligned with organizational values of equality, we also expected the intervention to lead to a greater decrease in psychological inflexibility in relation to stigmatizing and prejudiced thoughts (H3). Furthermore, given that the intervention aims to target more subtle forms of racism that are resistant to traditional training (Bonilla-Silva, 2017; Lillis & Hayes, 2007) by fostering acceptance of prejudiced cognitions and promoting compassion, we also expected the intervention to lead to a greater decrease in modern racist attitudes (H4). Additionally, we expected that the psychological flexibility and compassion training would lead to a greater increase in awareness of and commitment to organizational values concerning equality and anti-discrimination. This hypothesis was based on the assumption that the integration of ACT and CFT aims to align personal behavior with organizational values, thus promoting a culture of inclusion and equality (Brassey et al., 2020; Unruh et al., 2022). Lastly, we also aimed to evaluate the participants’ experiences of the intervention and the comparison lecture.
Method
Participants
The study encompassed 123 employees (96 women; 26 men; 1 “Other/do not want to say”) drawn from a range of organizations from four distinct sectors: telecommunications, furniture manufacturing, municipal administration, and agricultural products (see Table 1 for details). The target sample size, determined through a power analysis, was set at 130 participants to achieve an 80% power level, with the goal of detecting a small interaction effect in a 2 × 2 mixed ANOVA.
Table 1
Gender, job position, and job sector of participants
Distribution | All | Intervention | Lecture |
|---|---|---|---|
n = 123 | n = 64 | n = 59 | |
Gender | |||
Female | 96 | 52 | 44 |
Male | 26 | 12 | 14 |
Other | 1 | 0 | 1 |
Position | |||
Leader | 69 | 38 | 31 |
Non-leader, office | 52 | 25 | 27 |
Non-leader, non-office | 2 | 1 | 1 |
Sector | |||
Agricultural | 29 | 16 | 13 |
Municipal | 27 | 13 | 14 |
Furniture | 31 | 18 | 13 |
Telecoms | 36 | 17 | 19 |
The organizations participating in the study were carefully selected through strategic outreach to HR division leaders via telephone. These organizations were chosen because they are large organizations that routinely conduct training on equal rights for their employees, making them suitable candidates for the focus of the present study.
All participants were informed about the purpose and design of the study in advance, of their right to withdraw, that no benefits were given, and they all gave informed consent. The study was approved by the Swedish Ethics Review Agency (dnr 2018/567).
Procedure
Before integrating Acceptance and Commitment Therapy (ACT) and Compassion-Focused Therapy (CFT) into a unified intervention, we conducted two initial pilot RCT studies to assess the efficacy of each approach independently. The pilot studies were conducted in student samples. The results from the first pilot study suggested that simply engaging in acceptance-based practices and pursuing actions aligned with valued directions were not effective. The intervention did not significantly influence participants’ responses to stigmatizing thoughts, nor did it modify their attitudes towards modern racism, but it did significantly increase empathy, indicating a medium effect size compared to a comparison group receiving a lecture on stereotypes, prejudices, and discrimination that also included discussions. It is important to note that the flexibility training primarily focused on recognizing and accepting prejudiced thoughts without engaging in discriminatory behavior.
The second pilot study aimed to investigate the effect of a compassion-focused intervention, where a mindfulness and compassionate approach was taught and practiced. The pilot study found that the compassion intervention increased participants’ compassion to others compared to the comparison group (participating in lectures about prejudices and discrimination), indicating a large effect-size. The compassion intervention also increased empathy, compared to the comparison group, with a medium effect size.
Based on the two pilot studies, the flexibility and compassion intervention was developed by combining the approaches from the pilot studies. In the flexibility and compassion intervention, the experiential exercises were intertwined with psychoeducation about prejudice and discrimination, and group discussions. The intervention consisted of two sessions. In these sessions, the participants were guided through mindfulness training, key aspects of psychological flexibility, and CFT-exercises to increase a sense of compassion. In addition, the intervention included exercises and discussion aimed at increasing the participants’ awareness of their organizations’ values with regard to inclusion and equal treatment, which could serve as guides for committed action. An overview of the intervention and its parts is presented in Table 2.
Table 2
Combined mindfulness, flexibility and compassion training content for the two Conditions
Intervention group | Comparison group (Lecture) | |||
|---|---|---|---|---|
Session 1 | LECTURE 1 content | |||
INTERVENTION Contents | Example of the intervention | Purpose of the intervention | ||
Introduction conduct codes for the intervention | Introduction conduct codes for the lecture | |||
Exercise 1 | Exercises in mindfulness (CFT/ACT) | Mindful breathing Imagining Eye contact | Attention control, maintaining of attention, observation of inner state and acceptance | Theme 1: Definition of discrimination, racism, and the discrimination act |
Exercise 2 | Diffusion of thoughts and feelings: Understand the role of language with regards to prejudiced thoughts, and how these thoughts can often be triggered without our awareness. (ACT) | Completing unfinished statements Observing the pen exercise | Revealing implicit stereotypes Defusion to stereotypes | Discussion on theme 1: Experiences of terms and definitions |
Exercise 3 | Self as Context: Get a bird’s eye perspective of ourselves and our thoughts with regards to biases in the workplace. (ACT) | I had a thought that… | Self as Context | Theme 2: Stereotyping, prejudice, emotions and perceived threat |
Exercise 4 | Acceptance and Compassion: Acceptance of uncomfortable feelings (such as biases or defensiveness) and how to move past them to create a more open workplace. Acceptance of compassion towards ourselves and practicing compassion towards others. (CFT/ACT) | Don´t think of an apple The lie-detector Mindful visualization of hostility vs. kindness in and kindness at workplace | Acceptance of thoughts and emotions as they are and compassion to others | Discussion on theme 2: Experiences in everyday life and on workplace |
Exercise 5 | Identify and foster antiracist-based values within the organization (ACT) | The bus metaphor and the Bulls’ eye exercise, reframed as organizational values to work with | Identifying values within organization that promote prosocial behaviors | Theme 3: Classic and Modern prejudice, in Sweden in work environments. Statistics about diversity in Swedish workplaces today |
Summing up and introducing Homework | Trying to practice defusion related to negative stereotypes and acceptance to negative attitudes. Be mindful about prosocial values defined within workplace (ACT). | Participants write down organizational values to work related context | Reinforce the skills learnt during the session | Discussion on theme 3: experiences in everyday life and on workplace. Summing up and conclusions from the themes |
Session 2 | LECTURE 2 content | |||
Introduction of the focus of the session | ||||
Takeaways from the homework | Brief recap from previous lecture (Lecture 1) | |||
Exercise 1 | Psychoeducation and exercises in compassionate and soothing breath Calming down the mind and lower down the defense mechanism. (CFT) | Mindfulness exercise in Compassionate and soothing breath | Calm breathing, stability, openness, warmth, wisdom. | Theme 1: The link between prejudice and discrimination |
Exercise 2 | Revealing that our minds are overloaded with choices and what the consequences are. (ACT) | Pick a letter | Establish the notion that humans are keen to explain and understand social situations, even when not having enough information, which makes us rely on biases | Discussion on theme 1: Experiences in everyday life and on workplace |
Exercise 3 | Understanding that one can take someone else’s perspective even when being annoyed with that person. (CFT) | Think of someone you are annoyed of | Compassionate perspective taking | Theme 2: Norms and norm criticism in the workplace |
Exercise 4 | Mindfulness exercise. Discover and investigate one´s ability to open up and accept compassion from other people. (CFT) | Allowing yourself to accept compassion – a mindful exercise | Openness to compassion | Discussion on theme 2: Experiences in everyday life and on workplace |
Exercise 5 | Distinguish between goals and values. (ACT) | Goals vs. Values an exercise | Goal setting and distinguishing them from values | Theme 3: What is inclusion in a workplace? The benefits of diversity, inclusion. |
Exercise 6 | Identifying values that one keeps precious in relation to colleagues. (ACT) | Retirement party | Awareness of prosocial values and behaviors | Discussion on theme 3: Experiences in everyday life and on workplace |
Exercise 7 | Continue to identify and foster antiracist-based values and relate them to concrete goals within the organization. (ACT) | Committed action – Create an action plan for value congruent behaviors in the workplace | Reflect on how values can be formulated as concrete goals and behaviors | Theme 4: Goal setting. Elaborating on what individuals can do to ensure their workplace is one working towards a more inclusive and diverse workplace |
Participants randomized to the lecture-based comparison condition, received two lectures (of equal length as the intervention) focusing on stereotypes, racism, discrimination and prejudices as well as on workplace diversity and inclusion. The lectures also included discussions on the addressed topics.
Prior to the intervention, participants from each organization filled out pre-tests and were randomized to either the intervention group (the flexibility and compassion training) or the comparison group (lectures and group discussion). The pre-test took place one week before the first intervention/lecture and the post-test at the end of the second session. The interventions were led by three different workshop leaders (two women and one man of mixed ethnic backgrounds). They were all trained in ACT and CFT by the research group before holding the intervention/lecture sessions. Both the flexibility and compassion training and the lecture condition consisted of two sessions (each session was two hours long in both conditions). The second session took place two weeks after the first. The purpose of this was to give participants the opportunity to practice what they had learned in the first session.
Measures
The study employed several measures to assess the various outcome variables. The measures used were:
The Compassion Scale
The Compassion Scale (CS) was used to measure compassion (Pommier et al., 2020) and consists of 16 items rated on a 5-point Likert scale. In the present study we used a validated Swedish version of the measure (Wolgast et al., 2024). In CS, compassion for others is operationalized as a trait encompassing the dimensions of kindness vs. indifference, common humanity vs. separation, and mindfulness vs. disengagement in response to the suffering of others. In this context, kindness is conceptualized as being concerned for and caring towards others who are in pain, as well as having the desire to support those in need. Common humanity is understood as recognizing that it is a common human experience to experience hardship as well as a sense of connection to those who are suffering. Lastly, mindfulness is conceptualized as a form of balanced awareness which neither avoids nor becomes lost in the pain of others; rather, it refers to being willing to pay attention and listen to others when they are suffering (Pommier et al., 2020). In their subsamples reliability was found to vary between α = 0.77 and α = 0.90 and in the present sample α = 0.75, ω = 0.79.
Empathic Perspective Taking (Interpersonal Reactivity Index (IRI))
IRI, developed by Davis (1983), measures various facets of empathy through 28 items distributed across four subscales: Perspective Taking (PT), Fantasy, Empathic Concern, and Personal Distress. In the present study, the focus was exclusively on the Perspective Taking (PT) subscale. This decision was driven by the specific aim of the study to examine cognitive empathy, particularly the tendency to adopt the psychological viewpoints of others. The PT subscale evaluates how individuals understand and process the perspectives and mental states of others in various scenarios, which is central to the study’s objectives. The use of only the PT subscale was further justified by its relevance to the interpersonal dynamics within organizational settings. Understanding and relating to the perspectives of colleagues is crucial in promoting effective communication and collaboration in the workplace. By isolating this component of empathy, the study aimed to generate more targeted insights into cognitive empathy processes that directly influence interpersonal interactions in professional environments. The Swedish version of the IRI, utilized in this study, demonstrated adequate internal consistency for the PT subscale (α = 0.82; ω = 0.83), confirming the reliability of the measure in assessing perspective-taking within the Swedish cultural context.
Acceptance and Action Questionnaire – Stigma, AAQ-S
AAQ-S measures psychological inflexibility in relation to prejudiced and stigmatizing thoughts (Levin et al., 2014) and has 20 items, scored on a 5-point Likert scale. An example of an item is “My biases and prejudices affect how I interact with people from different backgrounds.” Analyses have indicated that the AAQ-S correlate with other measures of psychological flexibility and measures of prejudices (Levin et al., 2014) but it is described as more predictive of inflexibility in relation to prejudiced thoughts than a general measure of psychological flexibility. The internal consistency reported in Levin et al. (2014) was a Cronbach’s alpha of 0.84 and for the present study: α = 0.80; ω = 0,84. The Swedish version was constructed through a back-translation process and in the reliability evaluation of this version, 1 item did not relate to the scale as a whole and was therefore deleted (No 18 in the original). It should be noted that the items of the AAQ-S encompass more than just cognitive inflexibility. It also contains explicit references to behaviors and patterns of social interactions and how these are affected by the presence of prejudiced or stigmatizing thoughts.
Modern Racial Prejudice Scale (MRPS) measures explicit attitudes of modern racism. It is a self-assessment scale that has been tested in Sweden (Akrami et al., 2000) and consists of 9 items scored on a 5-point Likert scale. Examples of statements are “Enough efforts have been made for unemployed immigrants” and “Discrimination against immigrants is no longer a problem in Sweden”. MRPS scores correlate significantly with other measures such as Conservatism, Modern Sexism, Classical Sexism, and Social Dominance Orientation (Akrami et al., 2000). The internal consistency of the modern racism scale indicated α = 0.82. A similar Cronbach alpha was found in the present study (α = 0.81; ω = 0.79).
Organizational Values (OV)
OV was constructed for the present study and was used to determine the level of participants’ (a) knowledge of their organization’s values regarding equal right (e.g., “I am aware of my organization’s values concerning that everyone should receive equal treatment”); (b) whether employees shared these values (“I share my organization’s values regarding equal treatment and non-discrimination”); and (c) whether they tried to comply with them (“I behave in accordance with my organization’s values and see that everyone receives equal treatment”). There were 8 items which were added up to a total score. Responses were given on a 5-point Likert scale and the internal consistency was α = 0.80, ω = 0.89.
The Intervention/Lecture Evaluation
The evaluation consisted of twelve questions determining the participants’ level of satisfaction with the assigned training. Questions concerned, for example, how interesting, engaging, useful, or boring the experience had been, whether they felt that they had learned something, and whether the training had been useful. The items were scored on a 5-point Likert scale ranging from Agree strongly to Disagree strongly. The Cronbach’s α for this measure was 0.84. The average item scores on this measure were high in both groups (Mintervention = 4.50; SD = 0.41; Mlecture = 4.41; SD = 0.62) and the difference between groups was non-significant (t (76) = -0.99; p = 0.33), indicating that the two interventions were experienced equally valuable.
Data Analyses
All analyses were performed using SPSS version 26. The percentage of missing values across the six outcome variables at the post-intervention measurement point varied between 18.7% and 30.9%. In total, 171 out of 567 values (23.2%) were incomplete. Little’s MCAR test indicated that the data were missing completely at random (χ2 (9) = 14.54, p = 0.10). We used multiple imputations to create five multiply imputed datasets, and the pooled values of the imputed variables were used in all analyses. Methodologists presently regard multiple imputation as a state-of-the-art technique because it improves accuracy and statistical power relative to other missing data techniques (Enders, 2010). Incomplete variables were imputed using all pre-test scores and all complete post-test scores as predictors. For comparison, we also performed the analyses on the subset of complete cases.
To test the hypotheses regarding differences between the two conditions, a series of mixed ANOVAs were performed. In these analyses, pre and post-test scores on the different variables were entered as within-subjects factor, and experimental condition was entered as the between-subjects factor. To examine the hypotheses, the interaction effect of time x condition was examined for all dependent variables. Significant interaction effects were further examined using Bonferroni corrected within groups pairwise comparisons to test whether there was a significant change between pre- and posttests in the two conditions.
Results
As described above, all hypotheses were tested using mixed ANOVAs. Descriptive statistics for pre- and post-scores on all measures and statistics for the interaction effects (time x condition) of the ANOVAs on all variables are presented in Table 3.
Table 3
Descriptive statistics on pre- and post-measures on the dependent variables and test statistics for the Interaction Effect (time x group) of the performed mixed ANOVAs
Intervention (n= 64) | Lecture (n= 59) | Time x Condition interaction | ||||
|---|---|---|---|---|---|---|
Pre M (SD) | Post M (SD) | Pre M(SD) | Post M (SD) | F(1, 121) | Part. η2 | |
CS | 66.90 (5.95) | 69.61 (7.88) | 67.95 (6.26) | 66.30 (9.29) | 10.59** | 0.08 |
IRI-PT | 25.69 (4.04) | 26.41 (4.68) | 26.71 (3.42) | 25.90 (4.20) | 2.80 | 0.02 |
AAQ-S | 55.33 (13.04) | 48.88 (15.45) | 54.08 (12.63) | 54.20 (15.92) | 6.00* | 0.05 |
MRPS | 20.98 (4.13) | 21.22 (4.35) | 21.52 (4.12) | 21.87 (5.21) | 0.16 | < 0.01 |
OV | 34.68 (4.95) | 34.74 (5.12) | 35.24 (4.71) | 33.68 (5.79) | 6.02* | 0.05 |
As seen in Table 3, there were significant interaction effects (time x condition) on the scores for compassion (CS), response to stigmatizing thought (AAQ-S) and commitment to organizational values (OV). The significant interaction effects were further examined using Bonferroni corrected within groups pairwise comparisons. For compassion (CS), there was a significant increase between pretest and posttest in the intervention group (MD = 2.71; p < 0.01) but not in the lecture group (MD = -1.65; p = 0.09). When it comes to psychological inflexibility in relation to prejudiced thoughts (AAQ-S) there was a significant decrease from pretest to posttest in the intervention group (MD = -6.44; p < 0.01) but not in the lecture group (MD = 0.13; p = 0.95). For commitment to organizational values (OV) on the other hand, the significant interaction effect was driven by a significant decrease in the lecture condition (MD = -1.56; p < 0.01), whereas there was no change from pretest to posttest in the intervention condition (MD = 0.05; p = 0.91). Hence, the results were in line with our hypotheses for compassion (H1) and psychological inflexibility in relation to prejudiced thoughts (H3), which means that the flexibility and compassion training seemed to increase the participants’ scores for compassion and decrease their psychological inflexibility in relation to prejudiced thoughts.
Regarding organizational values however, the interaction effect was driven by a decrease in the score of the lecture-based comparison condition, and not – as hypothesized (H5) – by an increase in the average score in the group that had received the flexibility and compassion training.
In contrast to our expectations (as stated in H2 and H4), however, there was no significant interaction effect between time and group regarding empathic perspective taking (IRI-PT), nor for modern racist prejudices (MRPS).
When re-running the analyses using complete cases (i.e., without imputed data), the interaction effects of time x condition were still significant for psychological inflexibility in relation to prejudiced thoughts (F (1, 94) = 6.79, p = 0.01, part. η2 = 0.07), compassion (F (1, 92) = 5.11, p = 0.03, part. η2 = 0.05), and organizational values (F (1, 82) = 5.12, p = 0.03, part. η2 = 0.06). All results that were non-significant when running the analyses on the dataset that included imputed values were non-significant when analyzing only complete cases.
Finally, regarding the risk of Type I error due to multiple testing, it should be noted that the results identified as significant in the analyses above would also be significant when employing Benjamini and Hochberg’s (1995) false discovery rate procedure to correct for multiple testing, and that all follow-up tests of significant interaction effects were Bonferroni corrected.
Discussion
The purpose of the present study was to evaluate the effects of an intervention utilizing experiential exercises derived from ACT and CFT, to address psychological processes in promoting inclusive and equal workplaces concerning race/ethnicity. The exercises were designed to increase psychological flexibility in relation to prejudiced thoughts (i.e. decrease the extent to which one’s behaviors are affected by prejudiced and stigmatizing thoughts), compassion, and empathic perspective-taking while enhancing commitment to and awareness of organizational values. Given the focus on mitigating the impact of prejudice within organizations, modern racist attitudes were also assessed as an outcome measure.
Hypothesis 1 posited that the training would increase compassion for others more than a lecture-based comparison condition. This hypothesis was supported by a medium-sized effect. This finding is consistent with Gilbert’s (2017) work on compassion-focused interventions, which have been shown to enhance compassionate responses and decrease defensive reactions in various contexts. Hence, the results suggest that the participants in the flexibility and compassion condition increased their self-reported tendency to care about the suffering of others and to act to relieve such suffering to a larger extent than those taking part in traditional diversity and inclusion lectures. In the present study, this increase in compassion is intended to encompass an increase in compassion toward the victims of discrimination and racism, although this more specific form of compassion was not directly measured. It should be noted, however, that although our study resulted in increases in compassion, no corresponding changes in prejudice levels were observed, warranting further discussion (see below).
Hypothesis 2 suggested that the flexibility and compassion intervention would enhance empathic perspective-taking. This hypothesis was not supported. In the meta-analysis by Hsieh et al. (2021), there were only a few included studies on perspective taking as the main approach, and some interventions combined it with other approaches. For instance, Christofi and Michael-Grigoriou (2017) found that a game-like training situation, where participants were placed in the role of stigmatized individuals, significantly increased their perspective-taking abilities. In most of these studies, all four of Davis’ dimensions of empathy were measured, while our study used only the cognitive dimension, perspective taking, as we had the Compassion Scale to capture emotional and behavioral aspects. It is important to acknowledge that perspective taking was not explicitly addressed or trained in the flexibility and compassion intervention. Despite this, we expected to see an effect on this outcome variable based on the distinct focus on the common human experience of struggling with difficulties and on openness to one’s own cognitions and emotions entailed in the included CFT and ACT exercises. A possible explanation as to why we did not see this effect, however, is that this aspect of empathy might require a more direct targeting in order to facilitate changes than what was included in the intervention evaluated in the present study. This aligns with findings by Christofi and Michael-Grigoriou (2017), which suggest that more direct, experiential components are often necessary to achieve meaningful changes in perspective-taking abilities.
Hypothesis 3 stated that the flexibility and compassion training would reduce psychological inflexibility in relation to prejudiced thoughts more than the participants in the lecture condition. The results supported this hypothesis with a medium-sized effect. The results from this part of the study thus suggest that the participants in the flexibility and compassion condition experienced decreases in the extent to which their behaviors and patterns of social interaction are affected by prejudiced and stigmatizing thoughts. The measure for this variable contains the most explicit references to actual behaviors and social interactions. The results from this part of the study hence expand the findings from Lillis and Hayes’s (2007) pilot study by using validated measures to assess psychological inflexibility related to prejudiced thoughts, supporting Hayes et al. (1999) in highlighting the importance of psychological flexibility in reducing prejudiced behaviors. In addition, the reduction in psychological inflexibility observed in the intervention group suggests that methods derived from ACT and CFT may offer a different pathway for addressing cognitive processes related to prejudiced thoughts in organizational settings. As noted in the introduction, traditional diversity lectures tend to focus on explicit attitudes, often overlooking the more subtle, ingrained cognitive patterns that contribute to biased behaviors (Akrami & Ekehammar, 2005; Paluck et al., 2021). While we did not measure implicit biases directly, the reduction in psychological inflexibility could indicate that participants became more aware of their prejudiced thoughts and better equipped to manage them in a flexible way. This is consistent with the literature on ACT, which posits that increased psychological flexibility helps individuals refrain from acting on distressing thoughts (Lillis & Hayes, 2007).
Hypothesis 4, which anticipated a greater reduction in modern racist attitudes through the flexibility and compassion training, was not supported. A possible explanation is that altering racist attitudes, particularly entrenched stereotypes and covert prejudices, may require more time. In line with this interpretation, Paluck et al. (2021) emphasize that attitude change often necessitates prolonged engagement, suggesting that more extensive interventions may be needed to address modern racist attitudes effectively. In addition, it is crucial to note that the ACT-inspired exercises did not aim to change the content of thoughts but to foster flexibility in relation to prejudiced thoughts, thereby weakening their effects on behaviors. In this context, it should be noted that the measure used to operationalize modern racist attitudes does not contain references to behaviors, measuring only attitudes, whereas the flexibility and compassion intervention primarily stressed the importance of one’s behaviors. The hypothesis however was that the decreased inflexibility and in particular increases in compassion would also be associated with reductions in modern racist attitudes, but no such effect was found. Importantly, the introduction highlighted concerns with traditional diversity programs, which have been critiqued for focusing too narrowly on explicit attitudes (Paluck & Green, 2009). Our findings suggest that, while explicit prejudiced attitudes may not have shifted significantly, interventions targeting the flexibility of thought may still play a meaningful role in helping participants manage biased thinking patterns, which could support behavior change over time. However, further research is needed to assess whether these cognitive changes translate into long-term behavioral differences.
Hypothesis 5 suggested that the flexibility and compassion training would lead to a greater awareness of – and commitment to – organizational values regarding equal right and anti-discrimination compared to the lecture-based comparison condition. As hypothesized, the performed analyses indicated a significant time x condition interaction effect. However, as mentioned in the “Results” section, this effect was driven by a decrease in the score for the lecture-based comparison condition, and not by an increase in the scores for the group that received the flexibility and compassion training. Hence, the outcome on this variable did not provide support for the hypothesis. A possible explanation for these results is that the participants in the lecture condition received lectures partly focusing on the prevalence and pervasive nature of racism and discrimination, which might lead to decreased scores on a measure that included items regarding the extent to which participants believed that employees in their organization shared values regarding equal right and inclusion. Additionally, the organizational values measure was constructed for the present study, as we have found no other studies that have used similar measures.
Limitations and Future Directions
There are important limitations that should be considered when interpreting the results of the study. First, the absence of a long-term follow-up precludes any definitive conclusions regarding the enduring effects of the training. Future research should incorporate extended follow-up periods, such as three months post-intervention, to assess the sustainability of changes. Another limitation was the brief duration of the intervention, consisting of only two two-hour sessions. This limited exposure may have been insufficient to effect a change in modern racism scores, possibly explaining the lack of significant findings for this variable. Future studies could benefit from increasing the number of sessions and possibly integrating booster sessions to examine if modern racist attitudes can be altered over time.
Furthermore, as highlighted in previous research, altering prejudice at the content level is challenging, and the effects of such changes on behaviors are unclear (Paluck & Green, 2009; Devine & Ash, 2022). This underscores the importance of developing measures that not only assess changes in prejudiced behavior but also evaluate motivational aspects of behavior change, such as individuals’ intentions to act toward equality, and their awareness of inequalities and the dynamics of discrimination (e.g., Chang et al., 2019; Lillis & Hayes, 2007).
Moreover, the intervention aimed to increase the extent to which employees align their actions with explicit organizational values of equality. Unfortunately, the study did not find evidence to support this. Future research should continue to develop interventions and measures that can effectively capture changes in adherence to organizational values as has been identified as an important goal of diversity training (Onyeador et al., 2024).
Although direct behavioral outcomes were not measured in this study, prior research has indicated that higher self-rated compassion generally predicts both lower prejudice and reduced discriminatory behavior (Galinsky & Moskowitz, 2000; Kite & Whitley, 2016; Levin et al., 2016), that empathy and compassion can foster prosocial behavior (Oyler et al., 2022; Preckel et al., 2018) and the measure of psychological inflexibility contains direct references to the extent to which one’s behaviors are affected by prejudiced and stigmatizing thoughts. Despite this, it is a limitation of the present study that all outcome variables are dependent on self-report and self-assessed behaviors. In future research it would be valuable to include more direct measures of behaviors and organizational outcomes.
The introductory review also noted a lack of controlled trials outside laboratory settings in prejudice reduction research (Paluck et al., 2021; Devine & Ash, 2022; Hsieh et al., 2021). Indeed, the present study is one of the few studies employing an experimental design in organizational contexts, making its results, which indicated positive effects on two of five measures, noteworthy despite the relatively small sample size. We believe that the current study underscores the importance of these kinds of field experiments, and the need for more studies evaluating the effects of different approaches to diversity and inclusion training remains great.
Finally, future research should explore the concept of “compassionate flexibility,” as described by Gilbert (2017) and Trich et al. (2014), which refers to the capacity to not be governed by one’s automatic reactions, such as unpleasant feelings, but to engage with others in a compassionate manner. This study did not investigate compassionate flexibility due to the absence of a validated measure, presenting a valuable area for further research.
In conclusion, the present study found that flexibility and compassion training had significant and favorable effects on compassion for others and psychological inflexibility concerning prejudiced thoughts when compared to a lecture-based comparison condition. Hence, when compared to participants taking part in traditional diversity and inclusion lectures, the results suggest that the participants in the flexibility and compassion condition showed an increase in the self-reported tendency to care about the suffering of others and to act to relieve such suffering, and a decrease in the extent to which they experience that their behaviors and social interactions are affected by prejudiced and stigmatizing thoughts. One of the key findings of this study is the significant reduction in psychological inflexibility among participants in the intervention group. This result highlights the potential of ACT and CFT-based interventions to foster greater cognitive flexibility, which may help individuals respond more effectively to prejudiced thoughts. While we cannot claim that this directly impacts implicit biases, as noted in the introduction, cognitive flexibility is an important factor that can influence how individuals engage with challenging thoughts and situations, potentially reducing the likelihood of acting on prejudices (Lillis & Hayes, 2007). This suggests that targeting thought flexibility could be an important complement to existing diversity and inclusion approaches, which often focus more on explicit attitudes. In addition, the significant increase in self-reported compassion found in the intervention group supports the idea that experiential, mindfulness-based approaches may contribute to a more inclusive organizational climate. As noted in the introduction, compassion-focused interventions are gaining attention for their potential to enhance empathy and reduce social distance between individuals from different racial and ethnic backgrounds (Chow et al., 2023). It remains to be seen however, whether such increases in self-reported compassion and psychological flexibility leads to sustained behavioral changes in workplace settings. Being one of few randomized controlled studies performed in actual organizations showing at least moderate effects on relevant outcome measures, the results are interesting and promising in relation to future attempts to develop, spread, and scale up diversity training which can be used in organizational contexts.
Declarations
Ethics Statement
The study was approved by the Swedish Ethics Review Agency (dnr 2018/567). In Sweden Ethical Approval is not granted by universities but by the Swedish Ethics Review Agency.
Informed Consent
All participants gave their informed consent and were informed that they could withdraw from the study at any time.
Use of Artificial Intelligence Statement
AI was not used in any way.
Conflict of Interest
There is no conflict of interest to report.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
