While the relational demography literature has explored the consequences of surface-level (visible traits like ethnicity and gender) and deep-level (underlying attributes such as work experience) dissimilarity, it remains unclear how being dissimilar across multiple specific dimensions concurrently relates to employees' perceived inclusion. This study addresses that gap by examining how dissimilarity on specific dimensions relates to perceived inclusion and whether employees perceive less inclusion when they perceive dissimilarity on more dimensions. Additionally, we investigate whether these relationships are contingent on perceptions of organizational climate for inclusion.
We conducted a large-scale survey study comprising 6,319 employees from a Dutch public service organization.
Respondents most frequently reported perceiving dissimilarity from their coworkers in terms of personality, followed by ethnicity/culture, age, work experience, religion, sexual orientation, disability, education level, political orientation and gender (in descending order). Although dissimilarity on some dimensions was unrelated to perceived inclusion, we observed that dissimilarity on a greater number of dimensions was negatively related to perceived inclusion. Moreover, we found that the negative relationships between perceived dissimilarity and perceived inclusion were often mitigated by a positive perception of the organizational climate for inclusion.
This is the first study to assess how the multifaceted nature of dissimilarity on a broad range of dimensions interplays with perceptions of inclusion and organizational climate, highlighting key factors influencing workplace inequality and employee well-being.
As organizations become increasingly diverse, new challenges emerge. Within these changing workplaces, a growing number of employees are or feel dissimilar to their coworkers. This matters, as dissimilar employees often report lower attachment, creativity, job satisfaction and group fit, and higher turnover, absenteeism, stress and relationship conflict (Choi, 2007; Guillaume et al., 2012; Jackson et al., 1991; Jansen et al., 2017; Jehn et al., 1997; Kirchmeyer, 1995; Tsui et al., 1992). A crucial factor in these relationships appears to be employees' perception of social inclusion, namely, the feeling that the group provides them with a sense of belonging and ability to be themselves in the organization (Jansen et al., 2014, 2017; Şahin et al., 2019). These perceptions of inclusion are inherently relational, shaped by how individuals are treated, acknowledged and approached by others at work. We refer to this construct as “perceived inclusion.” This study investigates how perceived dissimilarity on specific dimensions relates to perceived inclusion, whether dissimilarity across multiple dimensions is more strongly associated with perceived inclusion than dissimilarity on a single dimension, and whether perceived organizational climate for inclusion moderates these relationships.
The current body of literature on dissimilarity has taken a narrow focus, leaving some important questions unaddressed. First, research has traditionally focused on broad distinctions between surface-level (e.g. gender and ethnicity) and deep-level (e.g. beliefs and preferences) characteristics (Guillaume et al., 2012; Jackson and Joshi, 2011; Şahin et al., 2019). While useful, this dichotomy obscures meaningful variation between dimensions. Some are task-oriented (e.g. work experience and education level), others are relationship-oriented (e.g. personality and gender; Jackson and Joshi, 2011), and some reflect historically stigmatized categories (e.g. ethnicity and gender; Nkomo et al., 2019). Although we do not empirically examine these distinctions, one open question is whether dissimilarity on various dimensions (e.g. age, ethnicity and personality) relates differently to perceived inclusion.
Second, prior work suggests that perceiving both surface-level and deep-level dissimilarity does not necessarily relate to lower perceptions of inclusion compared to perceiving only one type (Şahin et al., 2019). This raises the question of whether perceiving dissimilarity on more dimensions is related to lower perceptions of inclusion, or whether it is the perception of any dissimilarity, regardless of the number of dimensions, that matters most.
Third, previous studies have demonstrated that a positive organizational climate for inclusion can buffer the negative relationship between gender (Jansen et al., 2017) and deep-level dissimilarity (Şahin et al., 2019) and perceived inclusion. The question remains whether this moderating role extends to a broader range of specific dissimilarity dimensions.
This study aims to address these questions by (1) examining the unique relationships between dissimilarity on specific dimensions and perceived inclusion, (2) testing whether perceived dissimilarity across multiple dimensions relates to less perceived inclusion compared to dissimilarity on a single dimension and (3) testing whether perceived organizational climate for inclusion moderates the relationship between dissimilarity on various dimensions and perceived inclusion.
Effects of dissimilarity at work
According to the relational demography approach, perceptions of similarity or dissimilarity shape workplace relationships (Kaur and Ren, 2022). The underlying processes are often explained through Self-Categorization Theory (SCT; Turner et al., 1987), which suggests that individuals classify themselves and others into in-groups (similar) and out-groups (dissimilar). Such categorization helps individuals make sense of their environment but may also foster preferential treatment of fellow in-group members (Ellemers et al., 1999). Employees who are dissimilar to most of their colleagues—that is, those categorized as out-group members—are therefore likely to perceive less inclusion.
Initial categorization often relies on salient, surface-level dimensions, but over time deep-level differences can also become bases for categorization as colleagues get to know each other. Even when concealed, deep-level dissimilarity has been related to inclusion and job satisfaction (Ellemers and Barreto, 2006; Newheiser et al., 2017), likely because concealment both hinders subgroup formation and imposes a mental toll. Thus, both surface-level and deep-level dissimilarity may undermine perceived inclusion.
While the SCT provides a valuable framework for understanding how dissimilarity shapes relationships, workplace contexts introduce added complexities. Employees share a superordinate organizational identity while also belonging to multiple subgroups (e.g. based on their ethnicity, sexual orientation or team). The Ingroup Projection Model (IPM; Mummendey and Wenzel, 1999) extends SCT by proposing that the prototype of the superordinate group often reflects the dominant subgroup. Because dominant groups typically hold higher status and numbers, their characteristics become the normative benchmark (Waldzus et al., 2004; Wang et al., 2018). Employees who deviate from this prototype risk being seen as inferior and are more vulnerable to negative attitudes and mistreatment, including bullying (Glambek et al., 2020; Waldzus et al., 2004; Wang et al., 2018).
Based on SCT, IPM and prior findings, we expect both perceived surface-level and deep-level dissimilarity to be negatively related to perceived inclusion:
Perceived surface-level and deep-level dissimilarity are negatively related to perceived inclusion at work.
Moving beyond the distinction between surface-level and deep-level dissimilarity
Although distinguishing between surface-level and deep-level categories is useful, these categories lump together diverse dimensions (e.g. ethnicity, gender and age as surface-level dimensions and sexual orientation, work experience and personality as deep-level dimensions; Guillaume et al., 2012; Jackson and Joshi, 2011). Yet people may experience them differently. For instance, sexual orientation can be concealed or signaled by wearing queer symbols, complicating its classification.
The dichotomy also overlooks other key differences. Some dimensions, such as sexual orientation or skin color, are stigmatized, reflecting broader societal disadvantage (Nkomo et al., 2019). Others, such as personality or work experience, are not typically stigmatized, so employees dissimilar on these dimensions may not face the same disadvantages.
Thus, the dissimilarity–inclusion relationship may depend on the specific dimensions involved. To capture this variation, we examine dissimilarity across ten dimensions: personality, ethnicity/culture, age, work experience, religion, sexual orientation, disability, education level, political orientation and gender. We then test how dissimilarity on each relates to perceived inclusion, and whether perceiving dissimilarity on more dimensions is related to lower inclusion perceptions.
The additive relationship between dissimilarity and inclusion
Beyond examining dissimilarity on specific dimensions, we also investigate whether the number of dimensions on which an employee perceives dissimilarity relates to perceptions of inclusion. We adopt a multidimensional perspective (Liu et al., 2019), recognizing that individuals may differ from colleagues in several ways simultaneously.
At first glance, this resembles fault line theory (Lau and Murnighan, 1998), which also considers multiple dimensions. Faultline theory focuses on how dimensions such as gender, age and ethnicity align across individuals, potentially dividing a team into subgroups when these traits cluster. In contrast, our focus is not on team composition but on the individual's dissimilarity from the group. Rather than assuming subgroup splits, we examine how the accumulation of perceived dissimilarity across dimensions relates to perceptions of inclusion.
Multidimensionality has also been explored through the lens of stigma and intersectionality (Crenshaw, 1989). These approaches show that embodying multiple marginalized identities, such as those linked to gender, race or sexual orientation, is associated with greater job insecurity, workplace harassment, incivility, unfair treatment, stereotype threat and feelings of invisibility (Berdahl and Moore, 2006; Lavaysse et al., 2018; Remedios and Snyder, 2018; Zurbrügg and Miner, 2016). While informative, they differ fundamentally from our approach. Intersectionality highlights the societal meaning of identities (i.e. privilege or oppression) as the basis for workplace inequality, while our study focuses on the organizational meaning of difference—how individuals' identities or other dimensions compare to those of colleagues. By including both stigmatized and non-stigmatized dimensions (e.g. personality or work experience), we offer a complementary view on multidimensionality in organizations.
Despite these conceptual differences, we expect our findings to align with those of prior research:
Perceiving dissimilarity on a greater number of dimensions is negatively related to perceptions of inclusion.
The moderating effect of climate for inclusion
While dissimilarity often negatively relates to perceived inclusion, an important question is how organizations can mitigate this effect. Research shows that a positive climate for inclusion benefits diverse teams and dissimilar employees by fostering appreciation of diversity, reducing conflicts and enhancing job satisfaction (Jansen et al., 2017; Mor Barak et al., 2016; Nishii, 2013; Shore et al., 2018).
We define organizational climate for inclusion as employees' perceptions of how dissimilar employees are treated within the organization (Boezeman et al., 2025). This includes whether dissimilar employees are treated fairly, whether their interests are considered, and whether they are included in decision-making processes. The difference between perceived inclusion and organizational climate for inclusion is that perceived inclusion captures how included an individual personally feels in their team – in terms of belonging and authenticity – regardless of their own (dis)similarity, whereas organizational climate for inclusion reflects how that same individual perceives the organization's broader treatment of visibly or invisibly dissimilar employees.
Prior studies show that a positive climate for inclusion buffers the negative relationship between dissimilarity and perceived inclusion: in such environments, employees who perceive dissimilarity feel as included as those who do not (Jansen et al., 2017; Şahin et al., 2019). Moreover, the climate for inclusion is directly related to perceived inclusion: the more positive the climate is perceived, the more included employees feel, regardless of their own dissimilarity.
It is possible that a climate for inclusion is more beneficial for some dimensions of dissimilarity than others. Employees who perceive dissimilarity on certain dimensions may still feel overlooked even when the climate for inclusion is perceived as positive. Understanding these variations is therefore crucial.
We expect to replicate previous findings, predicting a moderating role of perceived organizational climate for inclusion on the relationship between dissimilarity and perceived inclusion. We also explore whether this moderation extends to separate dissimilarity dimensions:
Organizational climate for inclusion directly relates to perceived inclusion and moderates the relationship between dissimilarity and perceived inclusion, such that the negative relationship is diminished among employees who perceive a more positive organizational climate for inclusion.
Method
Participants
This study was approved by the Ethics Review Committee of the Faculty of Social and Behavioural Siences at Utrecht University. All employees of a Dutch governmental organization, approximately 16,000 people, were invited to participate. Of the workforce, 50.5% were women and 49.5% men.
Informed consent was obtained from all participants. In total, 6,319 participants completed the survey (2,991 men, 47.4%; 3,249 women, 51.5%; 70 identified neither as man nor woman, 1.1%; 9 missing, 0.1%; Mage = 47.18, SDage = 11.36). Descriptive statistics on participants' educational background, tenure, working hours and employment characteristics (e.g. contract type and supervisory role) are provided in Table 1.
Descriptive statistics of participants' educational background, tenure, working hours and employment characteristics
| Variable | Number of participants (% of all participants) |
|---|---|
| Education level | |
| Elementary education | 15 (0.24%) |
| High school | 217 (3.43%) |
| Vocational school | 1,055 (16.70%) |
| University of Applied Sciences | 2,660 (42.10%) |
| Research University | 2,366 (37.44%) |
| Missing | 6 (0.09%) |
| Supervisory position | |
| Yes | 748 (11.84%) |
| No | 5,563 (88.04%) |
| Missing | 8 (0.13%) |
| Position | |
| Indefinite | 5,090 (80.55%) |
| Temporary | 391 (6.19%) |
| Contractor | 781 (12.36%) |
| Intern | 51 (0.81%) |
| Missing | 6 (0.09%) |
| Variable | Mean and standard deviation |
|---|---|
| Tenure (in years) | 12.51 (11.29) |
| Work hours per week | 33.47 (5.13) |
Procedure and measures
Employees received an email with a survey link. Those without computer access (N = 225) received a paper-based questionnaire. After providing informed consent, participants first reported demographic information, including their gender, age, education level, years of tenure, number of hours worked per week, whether they held a supervisory position, and their contract status (temporary/indefinite/contractor/intern; see Table 1). Next, they completed the following measures in the order in which they are listed:
Perceived organizational climate for inclusion
We operationalized the perceived organizational climate for inclusion as employees' perception of how those who are surface-level or deep-level dissimilar are generally viewed and treated within the organization (Boezeman et al., 2025). It was measured with six items from the Netherlands Inclusiveness Screener (Boezeman et al., 2025), assessing how people in the organization talk and think about “people who are visibly or invisibly dissimilar than most others.” Items were presented on a bipolar 7-point scale (1 = agree most with the left statement, 7 = agree most with the right statement), with higher scores indicating a more positive climate for inclusion. Examples items include: “They are being disadvantaged at work when making decisions about tasks, salary, etc. – They are being taken into account when making decisions about tasks, salary, etc.,” “They are being seen as an inconvenience – They are being seen as an asset,” and “They are being treated worse than others – They are being treated as people that are valuable” (α = 0.92).
Perceived dissimilarity
Perceived surface-level and deep-level dissimilarity were each measured with one item, based on how Hobman et al. (2003) measured dissimilarity. First, participants indicated whether they perceived themselves as visibly (surface-level) dissimilar to others at work: “In terms of visible characteristics (e.g. age, sex, ethnicity), I think I am different from most colleagues at work.” Response options were “yes” and “no.” Those answering “yes” were asked to specify the basis of their dissimilarity by selecting one or more of eleven dimensions, presented in randomized order: Sexual orientation, personality, political beliefs, religion, education level, work experience, gender, age, ethnicity/cultural background, disability and other (with the option to give an answer in a text field).
Next, perceived invisible (deep-level) dissimilarity was assessed in a similar fashion. The item read: “In terms of invisible characteristics (e.g. beliefs and preferences), I am different than most others at work.” Those who answered “yes” again indicated the relevant dimension(s) from the same list. This design allowed participants to identify themselves as both surface-level and deep-level dissimilar, only one of the two, or neither [1].
Perceived inclusion
Perceived inclusion at work was measured with a 12-item version of the Perceived Group Inclusion Scale (PGIS; Jansen et al., 2014). The scale includes two subscales (belonging and authenticity), each comprising two components (group membership and group affection; room for authenticity and value in authenticity), with three items per component. Response ranged from 1 (completely disagree) to 7 (completely agree; α = 0.97). Example items are: “[organizational unit] gives me the feeling that I belong” and “… encourages me to be who I am.” In this study, we used perceived inclusion as a single construct, as an exploratory factor analysis (EFA) with oblique (Direct Oblimin) rotation indicated that all items loaded strongly on one factor (all > 0.81) [2].
Results
Plan of analysis
Analyses were conducted in R 4.5.1 (R Core Team, 2025), using the car (v. 3.1–3; Fox and Weisberg, 2019), cocor (v1.1-4; Diedenhofen and Musch, 2015), emmeans (v1.11.2; Lenth, 2025), Hmisc (v. 5.2–3; Harrell, 2025), interactions (v1.2.0; Long, 2024) and sjstats (v0.19.1; Lüdecke, 2024) packages. Full code and output are available at https://osf.io/z6avr/.
To test Hypothesis 1, stating that perceived surface-level and deep-level dissimilarity negatively relate to perceived inclusion, we conducted a 2 (surface-level dissimilarity: yes/no) X 2 (deep-level dissimilarity: yes/no) between-subjects ANOVA.
To explore which specific dissimilarity dimensions relate to perceived inclusion, we conducted a multiple regression analysis. Beforehand, we conducted statistical comparisons of dependent overlapping correlations to test whether the same dimensions had different relationships with perceived inclusion depending on whether they were listed after indicating surface-level dissimilarity or deep-level dissimilarity. Based on these results, we either distinguished between the two or collapsed them into a single predictor in subsequent analyses. We had no specific hypothesis regarding which specific dimensions would predict perceived inclusion, nor their relative magnitude.
We also conducted multiple regression analyses to test Hypothesis 2, stating that perceiving dissimilarity on a greater number of dimensions is negatively related to perceived inclusion, and Hypothesis 3, stating that the negative relationship between dissimilarity and perceived inclusion diminishes as employees perceive a more positive climate for inclusion.
Descriptive statistics
Zero-order correlations for all study variables can be found in the supplementary material. Frequencies of specific dissimilarity dimension are shown in Table 2. In total, 1,289 participants (20.40% of the total N) perceived themselves as surface-level dissimilar, 1,365 (21.60%) as deep-level dissimilar and 474 (7.50%) as both surface-level and deep-level dissimilar. When listing specific dimensions, participants most often indicated perceiving dissimilarity from coworkers in terms of personality, followed by (in descending order) ethnicity/culture, age, work experience, religion, sexual orientation, disability, education level, political orientation and gender (see Table 2 and Figure A in supplementary material).
The number and percentage of participants who perceived only surface-level dissimilarity, only deep-level dissimilarity or both surface-level and deep-level dissimilarity per specific dimension
| Dissimilarity dimension | Perceived as only surface-level dissimilar: n (%) | Perceived as only deep-level dissimilar: n (%) | Perceived as both surface-level and deep-level dissimilar: n (%) | Total perceived dissimilarity per Dimension: n (%) |
|---|---|---|---|---|
| Personality | 305 (37.6%) | 404 (49.8%) | 102 (12.6%) | 811 (100%) |
| Ethnicity/Culture | 478 (64.5%) | 114 (15.4%) | 149 (20.1%) | 741 (100%) |
| Age | 347 (77.8%) | 55 (12.3%) | 44 (9.9%) | 446 (100%) |
| Work experience | 201 (47.4%) | 184 (43.4%) | 39 (9.2%) | 424 (100%) |
| Religion | 130 (31.9%) | 224 (54.9%) | 54 (13.2%) | 408 (100%) |
| Sexual orientation | 54 (19.4%) | 200 (71.9%) | 24 (13.2%) | 279 (100%) |
| Disability | 66 (24.3%) | 182 (66.9%) | 24 (8.8%) | 272 (100%) |
| Education level | 97 (42.7%) | 103 (45.4%) | 27 (11.9%) | 227 (100%) |
| Political orientation | 25 (11.6%) | 171 (79.2%) | 20 (9.3%) | 216 (100%) |
| Gender | 113 (77.9%) | 18 (12.4%) | 14 (9.7%) | 145 (100%) |
Note(s): Participants could indicate dissimilarity on multiple dimensions. Percentages are calculated based on the total number of participants perceiving dissimilarity on the specific dimension. Participants could also indicate dissimilarity on “other” dimensions (n = 222), but this option was excluded from further analyses due to its heterogeneity
Some dimensions were reported with roughly equal frequency after participants indicated surface-level or deep-level dissimilarity (e.g. personality, work experience, religion and education level). Others were more often reported after indicating only surface-level dissimilarity (e.g. ethnicity/culture, age and gender) or only deep-level dissimilarity (e.g. religion, sexual orientation, disability and political orientation). Interestingly, ethnicity/culture was an exception: more participants reported it after indicating both surface-level and deep-level dissimilarity than after deep-level alone, whereas this “both” classification was uncommon for most other dimensions.
It was also common for participants to perceive dissimilarity across two or more dimensions concurrently. See supplementary material for detailed frequencies of these dyadic combinations of perceived dissimilarity dimensions.
Confirmatory and exploratory analyses
Hypothesis 1: Surface-level and deep-level dissimilarity and inclusion
We conducted a 2 (surface-level dissimilarity: yes vs. no) X 2 (deep-level dissimilarity: yes vs. no) between-subjects ANOVA to test Hypothesis 1, stating that surface-level and deep-level dissimilarity negatively relate to perceived inclusion. We obtained a main effect of both surface-level dissimilarity, F(1, 6,289) = 95.28, p < 0.001, ηp2 = 0.02, and deep-level dissimilarity, F(1, 6,289) = 177.28, p < 0.001, ηp2 = 0.03, on perceived inclusion. These results indicated that participants who perceived surface-level dissimilarity perceived less inclusion (M = 4.55, SE = 0.03) than participants who did not perceive surface-level dissimilarity (M = 4.93, SE = 0.02). Furthermore, participants who perceived deep-level dissimilarity perceived less inclusion (M = 4.47, SE = 0.03) than those who did not (M = 5.00, SE = 0.02). We did not obtain a significant interaction between surface-level dissimilarity and deep-level dissimilarity on perceived inclusion, F(1, 6,289) = 1.31, p = 0.252. Simple slopes analyses using Tukey's HSD procedure indicated that participants who perceived only surface-level dissimilarity scored lower on perceived inclusion (M = 4.79, SE = 0.04) than participants who did not perceive dissimilarity at all (M = 5.22, SE = 0.02), t(6,289) = 9.73, p < 0.001, but did not differ from those who perceived only deep-level dissimilarity (M = 4.65, SE = 0.04), t(6,289) = 2.51, p = 0.058. Furthermore, participants who perceived only deep-level dissimilarity also scored lower on perceived inclusion than participants who did not perceive dissimilarity at all, t(6,289) = 13.43, p < 0.001. Participants who perceived both surface-level and deep-level dissimilarity perceived less inclusion (M = 4.30, SE = 0.05) than participants who did not perceive surface-level or deep-level dissimilarity, t(6,289) = 16.33, p < 0.001, and also scored lower than participants who indicated only surface-level, t(6,289) = 7.22, p < 0.001, or only deep-level dissimilarity, t(6,289) = 5.20, p < 0.001. These results support our hypothesis (H1), as both surface-level and deep-level dissimilarity were negatively related to perceived inclusion [3].
Explorative analyses: separate effects of dissimilarity dimensions
After participants indicated whether they perceived themselves to be different from their coworkers on surface-level and deep-level dimensions, they reported on which specific dimensions they experienced dissimilarity. To determine whether the specific surface-level and deep-level dimensions should be treated as separate variables in subsequent analyses, or could be collapsed, we conducted statistical comparisons of dependent overlapping correlations (Diedenhofen and Musch, 2015). This way, we tested whether the dimensions that the participants listed after indicating surface-level dissimilarity had a different relationship with perceived inclusion than the same dimensions listed after indicating deep-level dissimilarity. The results showed that some dimensions had different correlations with perceived inclusion, depending on whether they were listed after indicating surface-level or deep-level dissimilarity. These dimensions are personality (surface-level r = −0.11 vs. deep-level r = −0.18, p < 0.001), work experience (surface-level r = −0.06 vs. deep-level r = −0.11, p = 0.002) and political orientation (surface-level r = −0.03, deep-level r = −0.07, p = 0.012; See supplementary material for all correlations between variables). For the remaining dimensions, their correlations with perceived inclusion did not depend on whether they were listed after indicating surface-level or deep-level dissimilarity. Therefore, in the following analyses, we included separate variables for surface-level and deep-level personality, work experience and political orientation, but did not make a distinction based on dissimilarity type for the remaining dimensions.
To examine the relationships between dissimilarity on specific dimensions and perceived inclusion, we then conducted a multiple regression analysis with the specific dissimilarity dimensions as predictors and perceived inclusion as the dependent variable. Results indicated an average significant relationship between the dissimilarity dimensions and perceived inclusion, F(13, 6,299) = 36.06, R2 = 0.07, p < 0.001. Closer inspection of the individual predictors indicated that dissimilarity on personality (both surface-level and deep-level), ethnicity/culture, age, work experience (deep-level), disability, and education level were negatively related to perceived inclusion (See Table 3 for the coefficients) [4].
Regression coefficients for the multiple regression analysis predicting perceived inclusion
| Inclusion | |||||
|---|---|---|---|---|---|
| b | SE b | β | SE β | p | |
| Intercept | 5.17 | 0.02 | 0.00 | 0.01 | |
| Surface-level personality | −0.20 | 0.07 | −0.04 | 0.01 | 0.009* |
| Deep-level personality | −0.53 | 0.06 | −0.12 | 0.01 | <0.001** |
| Ethnicity/Culture | −0.38 | 0.05 | −0.10 | 0.01 | <0.001** |
| Age | −0.16 | 0.06 | −0.03 | 0.01 | 0.018* |
| Surface-level work experience | 0.12 | 0.09 | 0.02 | 0.01 | 0.185 |
| Deep-level work experience | −0.20 | 0.09 | −0.03 | 0.01 | 0.031* |
| Religion | −0.09 | 0.06 | −0.02 | 0.01 | 0.195 |
| Sexual orientation | 0.13 | 0.07 | 0.02 | 0.01 | 0.099 |
| Disability | −0.56 | 0.07 | −0.10 | 0.01 | <0.001** |
| Education level | −0.35 | 0.09 | −0.05 | 0.01 | <0.001** |
| Surface-level political orientation | 0.27 | 0.18 | 0.02 | 0.01 | 0.180 |
| Deep-level political orientation | −0.10 | 0.09 | −0.01 | 0.01 | 0.291 |
| Gender | −0.16 | 0.10 | −0.02 | 0.01 | 0.179 |
Note(s): b represents unstandardized regression weights. β indicates the standardized regression weights
*p < 0.050, **p < 0.001
Hypothesis 2: Additive relationship between dissimilarity dimensions and perceived inclusion
The results above show that certain dissimilarity dimensions significantly predict perceived inclusion. Hypothesis 2 stated that perceiving dissimilarity on more dimensions is negatively related to perceived inclusion. We conducted a regression analysis to test this.
We counted the number of dimensions participants perceived dissimilarity and used this number to predict perceived inclusion. The result supported Hypothesis 2, demonstrating that the number of perceived dissimilarity dimensions was negatively related to perceived inclusion (F(1, 6,311) = 340.10, R2 = 0.05, p < 0.001, β = −0.23, SE = 0.01; see Figure 1) [5].
The scatterplot displays individual data points grouped by eight distinct values (0 through 7). The horizontal axis is labeled “Number of Dissimilarity Dimensions” and ranges from 0 to 7 in increments of 1 unit. The vertical axis is labeled “Perceived Inclusion (mean plus 95 percent C I)” and ranges from 1 to 7 in increments of 1 unit. For each dissimilarity value, a cluster of points is shown along with a mean marker and vertical 95 percent C I bar. A downward-sloping fitted regression line runs across the plot with a shaded confidence band. Mean values decrease gradually from about 5.20 at 0 dimensions to about 4.04 at 7 dimensions, representing a steady downward trend. The highest visible mean occurs at 0 dimensions, and the lowest mean occurs at 7 dimensions. Note: All numerical data values are approximated.The relationship between the number of dissimilarity dimensions and perceived inclusion. Dots represent individual data points
The scatterplot displays individual data points grouped by eight distinct values (0 through 7). The horizontal axis is labeled “Number of Dissimilarity Dimensions” and ranges from 0 to 7 in increments of 1 unit. The vertical axis is labeled “Perceived Inclusion (mean plus 95 percent C I)” and ranges from 1 to 7 in increments of 1 unit. For each dissimilarity value, a cluster of points is shown along with a mean marker and vertical 95 percent C I bar. A downward-sloping fitted regression line runs across the plot with a shaded confidence band. Mean values decrease gradually from about 5.20 at 0 dimensions to about 4.04 at 7 dimensions, representing a steady downward trend. The highest visible mean occurs at 0 dimensions, and the lowest mean occurs at 7 dimensions. Note: All numerical data values are approximated.The relationship between the number of dissimilarity dimensions and perceived inclusion. Dots represent individual data points
Hypothesis 3: Perceived organizational climate for inclusion
To test Hypothesis 3, stating that perceived organizational climate for inclusion is positively related to perceived inclusion and moderates the relationship between dissimilarity and perceived inclusion, we conducted a series of moderation analyses using multiple regression analyses. We first conducted an analysis predicting perceived inclusion from surface-level dissimilarity, deep-level dissimilarity, perceived climate for inclusion and the two-way interactions between climate for inclusion and the two types of dissimilarity, respectively (F(5, 6,259) = 403.50, R2 = 0.24, p < 0.001).
Supporting our hypothesis, we obtained a significant main effect of climate for inclusion, β = 0.39, SE = 0.01, p < 0.001, indicating that regardless of their perceived dissimilarity from coworkers, participants perceived more inclusion to the extent that they perceived a more positive organizational climate for inclusion. We also obtained a significant interaction effect between surface-level dissimilarity and perceived organizational climate for inclusion on perceived inclusion, β = 0.06, SE = 0.01, p < 0.001.
Simple slopes analysis revealed that surface-level dissimilarity had a negative relationship with perceived inclusion when participants perceived a negative organizational climate for inclusion (b−1SD = −0.41, p < 0.001). This relationship became weaker when participants perceived an average (bmean = −0.22, p < 0.001) or positive organizational climate for inclusion (b+1SD = −0.03, p = 0.502), with the relationship disappearing in the latter case (See Figure B in supplementary material).
The same pattern of results was found for the relationship between deep-level dissimilarity and perceived inclusion. We obtained a significant interaction effect between deep-level dissimilarity and organizational climate for inclusion as well, β = 0.05, SE = 0.01, p < 0.001. The relationship between deep-level dissimilarity and perceived inclusion became weaker as participants perceived a more positive organizational climate for inclusion, but did not completely disappear (b−1SD = −0.54; bmean = −0.38; b+1SD = −0.22, all p's < 0.001; See Figure C in supplementary material).
Next, we tested whether the relationships between the specific dissimilarity dimensions that were significantly related to perceived inclusion and perceived inclusion were moderated by organizational climate for inclusion by conducting a multiple regression analysis, predicting perceived inclusion from dissimilarity on personality (both surface-level and deep-level), ethnicity/culture, age, work experience (deep-level), disability, education level and the two-way interactions between climate for inclusion and the specific dissimilarity dimensions (F(15, 6,262) = 134.60, R2 = 0.24, p < 0.001). We obtained significant interaction effects between organizational climate for inclusion and personality (deep-level), β = 0.03, SE = 0.01, p = 0.006; climate for inclusion and ethnicity/culture, β = 0.06, SE = 0.01, p < 0.001; and climate for inclusion and disability, β = 0.04, SE = 0.01, p < 0.001.
Simple slope analyses revealed that as participants perceived the organizational climate for inclusion as more positive, the negative relationships between dissimilarity on personality (deep-level) (b−1SD = −0.52; bmean = −0.40; b+1SD = −0.28; all p's < 0.001), ethnicity/culture (b−1SD = −0.38, p < 0.001; bmean = −0.15, p = 0.001; b+1SD = −0.28, p = 0.176), disability (b−1SD = −0.52, p < 0.001; bmean = −0.26, p < 0.001; b+1SD = −0.01, p = 0.951) and perceived inclusion became weaker; however, they did not always completely disappear when a positive climate for inclusion was perceived. These results are generally supportive of Hypothesis 3.
Discussion
Employees may perceive dissimilarity from coworkers across multiple dimensions, prompting our research into how these perceptions relate to perceived inclusion. Our study contributes in four key ways:
First, both surface-level and deep-level dissimilarity were negatively related to perceived inclusion, replicating and extending earlier work (Guillaume et al., 2012; Şahin et al., 2019). Unlike prior studies, we found that surface-level dissimilarity, though weaker than deep-level dissimilarity, was also associated with perceived inclusion. This may reflect our larger sample, which provided power to detect smaller effects. Moreover, our results show that the same dimensions can be experienced as either surface-level and/or deep-level. Suggesting that researchers should be cautious about fixing this classification in advance.
Second, dissimilarity on personality (surface-level and deep-level), ethnicity/culture, age, work experience (deep-level), education level and disability were negatively related to perceived inclusion. These findings highlight the importance of considering a wide range of dimensions in fostering workplace inclusion, while also acknowledging the persistent disadvantages tied to historically marginalized identities (Koch et al., 2016).
Third, perceiving dissimilarity across multiple dimensions was negatively related to perceived inclusion, suggesting an added burden for employees dissimilar on more than one dimension. This aligns with research on multiple stigmatized identities (Berdahl and Moore, 2006; Lavaysse et al., 2018; Remedios and Snyder, 2018; Zurbrügg and Miner, 2016). However, rather than inferring stigma from demographic coding, our approach relied on employees' own perceptions, showing that these effects extend beyond stigmatized identities. Although relational demography and the IPM offer valuable frameworks, neither explicitly addresses multidimensional dissimilarity or deviation from a group prototype. Our findings point to this gap and suggest that these frameworks should evolve to account for dissimilarity across multiple dimensions.
Fourth, perceived organizational climate for inclusion buffered negative relationships between dissimilarity and perceived inclusion, replicating earlier findings (Jansen et al., 2017; Şahin et al., 2019). This buffering varied across dimensions: for dissimilarity on age, education level and work experience, the relationship was not affected; for personality, it weakened but persisted; and for ethnicity/culture and disability, it disappeared. This suggests that, in this organizational context, not all forms of differences are valued and considered equally.
Specific dissimilarity dimensions
Descriptive findings
Participants most often reported dissimilarity in terms of personality and ethnicity/culture, followed by age, work experience, religion, sexual orientation, disability, education level and political orientation. Gender was least often reported, which is striking given prior work emphasizing its importance (e.g. Jansen et al., 2017). A likely explanation is the relatively balanced gender representation in the studied organization, reducing its salience. By contrast, in settings with stronger gender imbalances, such as STEM or care professions (Statistics Netherlands, 2024), gender may be a more prominent source of dissimilarity.
At the other end of the spectrum, personality was the most frequently reported dimension, appearing in about one-third of all cases where participants perceived dissimilarity. It also often co-occurred with other dimensions. Over half of those reporting dissimilarity on education level, political orientation or work experience, and more than one-third of those on ethnicity/culture, gender or religion, also reported personality. This suggests that differences on several dimensions may be interpreted as reflecting underlying personality traits or values. Personality is likely a broad concept open to multiple interpretations, which may explain why it is the most frequently reported dimension.
Surface-level and deep-level classifications
The literature typically distinguishes surface-level dimensions (almost immediately observable, e.g. gender, age, ethnicity and department membership) from deep-level dimensions (emerging through interaction, e.g. personality, attitudes, sexual orientation; Guillaume et al., 2012; Jackson and Joshi, 2011). Our findings, however, suggest that participants did not apply this distinction consistently. Some dimensions were reported as surface-level by some and as deep-level by others. For example, 19.4% of the participants who perceived dissimilarity on sexual orientation reported this after indicating only surface-level dissimilarity, and 13.2% reported this after indicating both surface-level and deep-level dissimilarity, despite its typical categorization as deep-level. Personality, too, was divided almost equally between surface-level and deep-level. This implies that participants interpret “visible” more broadly, perhaps as something deducible from behavior, dress or other expressions, rather than observable at a glance.
Theoretical classifications, therefore, may not fully capture the lived experiences of dissimilarity, as employees rely on their own perceptions of observability. Still, our results suggest the distinction retains some meaning. For personality, work experience and political orientation, the relationship with perceived inclusion differed depending on whether they were reported as surface- or deep-level. For other dimensions, the distinction mattered less.
Relationship with perceived inclusion
Our findings revealed distinct patterns regarding how dissimilarity dimensions relate to perceived inclusion. Dissimilarity on dimensions frequently examined in diversity literature (e.g. Jansen et al., 2017; Lloren and Parini, 2017; Schneider et al., 2022), such as religion, gender and sexual orientation, were not significantly related to perceived inclusion in our study. This discrepancy may reflect the organizational context, where these dimensions could be normalized or even valued positively. More broadly, it highlights that context influences how dissimilarity shapes workplace experiences.
Conversely, dissimilarity dimensions significantly linked to perceived inclusion, namely personality, ethnicity/culture, work experience and disability, vary widely and do not align clearly with traditional surface-level and deep-level, or task-related and relationship-oriented distinctions. This suggests they operate through different psychological and social mechanisms. To interpret these findings, we utilized Harrison and Klein's (2007) typology of diversity: separation (differences in attitudes or values), variety (categorical differences) and disparity (differences in power, status or resources).
For example, dissimilarity in personality may reflect separation, affecting how individuals interact or align on communication styles and work values. This may influence inclusion through mismatched expectations around visibility, assertiveness, or participation. In such cases, reduced inclusion may not stem from interactional misfit but rather from stigma.
Dissimilarity on ethnicity/culture may reflect variety, through in- and outgroup categorization, and disparity, through historical and institutional bias. Disability is most clearly linked to disparity, since assumptions about competence and the need for accommodations can mark employees as different. Despite different pathways, these mechanisms can yield similar outcomes, for example, exclusion from informal conversations, team activities or collaboration opportunities.
Classifying each dimension into a single type is not always straightforward, and meanings are often context dependent. For instance, non-heterosexual sexual orientation may represent disparity in environments where it is stigmatized, but as a variety in inclusive ones, perhaps similar to the organization studied here. This underscores the need to consider how dissimilarity is experienced within its specific context.
Practical implications. This study illuminates how dissimilarity on various dimensions relates to perceptions of social inclusion. Beyond commonly acknowledged dimensions like ethnicity/culture, age and disability, we found that dissimilarity on personality, work experience, and education level also negatively relates to perceived inclusion. Addressing these dimensions is important, and must go hand in hand with efforts to support marginalized groups. For instance, organizations that include nondemographic dimensions (e.g. personality and work experience) in diversity statements, while neglecting demographic dimensions (e.g. ethnicity and sexual orientation), are perceived as less appealing by minoritized employees than those that address both (Kirby et al., 2023). Effective inclusion policies, therefore, require affirming marginalized identities and appreciating the unique differences among all employees (Russell Pascual et al., 2024).
Our findings also highlight that employees may perceive dissimilarity on multiple dimensions simultaneously, going beyond those typically associated with historical disadvantage or stigma. Because the majority group members are often seen as the “prototypical employee” (Mummendey and Wenzel, 1999), organizations must broaden this prototype to reflect all staff. Research shows that more complex and inclusive representations of superordinate categories allow both majority and minority groups to claim prototypicality more equally (Alexandre et al., 2016). Practical approaches include openly recognizing the workforce diversity and making visible the range of differences within the organization.
Strengths and limitations
Conducting a survey study in a large organization allowed us to capture real-life workplace experiences. By letting employees indicate dimensions of perceived dissimilarity, we focused on their lived experiences rather than researcher-coded identities, which may not be salient or meaningful to them. The large sample also provided sufficient power to test relationships between many dissimilarity dimensions and perceived inclusion, and the moderating role of perceived organizational climate for inclusion.
These strengths should be weighed against several limitations. First, the correlational design allowed us to identify relationships, but not causality. Experimental or longitudinal designs are needed to test causal pathways. Still, we took steps to rule out common method variance as an alternative explanation for the observed associations.
Second, dissimilarity was measured in a binary way. Participants indicated whether they perceived dissimilarity on several dimensions, but not its extent. This allowed broad coverage, but missed intensity or salience. Future research should incorporate continuous measures to assess degrees of perceived dissimilarity more precisely.
Third, the study was limited to a single organization, constraining generalizability. The prevalence of perceived dissimilarity likely reflects the organization’s workforce composition. Nonetheless, the relationship between dissimilarity and perceived inclusion mirrors findings from other contexts (Jansen et al., 2017; Şahin et al., 2019), suggesting broader relevance.
Fourth, while our approach drew on the relational demography framework, it did not consider broader societal structures and histories that shape organizational norms and inequalities (Dacin et al., 1999). Perspectives emphasizing stigma and power, such as intersectionality, could enrich understanding of why certain forms of dissimilarity matter more than others, and for whom, in specific contexts.
Conclusion
This research demonstrates that employees can perceive dissimilarity on multiple dimensions simultaneously. The dissimilarity dimensions related to perceived inclusion vary widely, including both those traditionally associated with stigma and disadvantage and those that are not. Importantly, as employees perceive dissimilarity on more dimensions, they perceive less inclusion from colleagues.
Although perceived dissimilarity is a subjective and complex phenomenon, the implication is clear: organizations must ensure that all employees feel included and valued, regardless of the dimensions on which they differ from others.
Author contributions
Onur Şahin: Writing – original draft preparation; Formal analyses; conceptualization; methodology.
Jojanneke van der Toorn: Writing – review and editing; supervision; conceptualization; methodology.
Wiebren S. Jansen: Writing – review and editing; supervision; conceptualization; methodology.
Naomi Ellemers: Supervision; conceptualization.
The data were collected by the Nederlandse InclusiviteitsMonitor (Netherlands InclusivityMonitor).
This study was carried out in accordance with the recommendations of the Faculty Ethics Review Board (FERB) at Utrecht University.
Notes
We did not have team-level composition data for dimensions such as gender, sexual orientation or disability, which prevented us from calculating objective dissimilarity scores. However, prior research shows that perceived dissimilarity correlates with actual demographic differences and is often a stronger predictor of workplace outcomes (Cunningham, 2007). In our data, some overlap was evident where information was available – for example, between participants who identified as part of the “LGBTI+” group and those reporting dissimilarity based on sexual orientation (see supplementary material). While identifying as part of the “LGBTI+” group is not a direct measure of dissimilarity, it can reasonably serve as a proxy, given that LGBTI + individuals are a numerical minority within most teams.
In the supplementary material, we present the results of exploratory and confirmatory factor analyses, which support the distinction between the four theoretical components of perceived inclusion as well as the appropriateness of using perceived inclusion as a single variable. The analyses further confirm that perceived organizational climate for inclusion and perceived inclusion are distinct constructs.
We replicated these findings using a Structural Equation Model (SEM) that accounted for common method bias. The results, which align with the findings reported here, are available in the supplementary materials. However, due to the complexity of the models required to test our remaining hypotheses, it was not feasible to conduct equivalent SEMs while also controlling for common method bias. As such, we report the regression analyses in the main text, as they provide the most comprehensive and interpretable summary of results across all models.
As a robustness check, we estimated a model that did not distinguish between any surface-level and deep-level dimensions, instead using single predictors for personality, work experience and political orientation. Results were largely consistent with the main analysis, but in this model, dissimilarity on age and work experience no longer predicted perceived inclusion (see supplementary material). This suggests that distinguishing between surface-level and deep-level dimensions adds meaningful nuance, as the predictive value of deep-level work experience was lost when the two were merged.
We additionally tested whether the result would differ depending on whether the counted dissimilarity dimensions were previously found to be related or unrelated to perceived inclusion in our earlier analyses. As detailed in the supplementary material, the effect is primarily driven by dissimilarity on dimensions previously associated with perceived inclusion.
The supplementary material for this article can be found online.
