The workforce has been becoming increasingly culturally diverse (Moore et al., 2020). Although diversity management has been a topic of interest in theory and practice for more than 40 years (Chung et al., 2020), attention has shifted more recently in terminology and intent from “diversity” to “inclusion” (Mor Barak et al., 2016; Roberson et al., 2017; Shore et al., 2018; Ashikali et al., 2020; Chung et al., 2020), moving scholars and practitioners from focusing solely on the “what” and “who” to looking deeper into the “how” and “why” (Mor Barak et al., 2016; Shore et al., 2018; Rice et al., 2021).

Inclusion refers to “the degree to which an employee perceives that he or she is an esteemed member of the work group through experiencing treatment that satisfies his or her needs for belongingness and uniqueness” (Shore et al., 2011, p. 1265). The appeal of inclusion lies in its positive stance in terms of appreciating and valuing individual differences (Ferdman and Deane, 2014), enabling everyone to experience acceptance of their identities and ideas, to feel a part of the system in both formal and informal ways, and to sense that their voices and opinions are welcomed at every level of decision-making (Mor Barak et al., 2016; Chung et al., 2020).

Organizational practices, such as providing mechanisms for members to voice, participate, communicate, and develop (Offerman and Basford, 2014; Sabharwal, 2014; Tang et al., 2015), as well as supporting from their supervisors (Tang et al., 2015) and peers (Offerman and Basford, 2014), are thought to cultivate an individual’s perception of inclusion in the workplace. Additionally, demographic factors, such as gender, age, ethnicity, religious, education, and regional origin, are also found to be related to an individual’s perception of inclusion (Mor Barak et al., 2001; Cho and Mor Barak, 2008; Mousa et al., 2020). While inclusion has been regarded as the root of psychosocial wellbeing (Mor Barak and Cherin, 1998), comprising low exhaustion and high engagement (Bakker, 2014). It has been found to increase employee commitment (Hwang and Hopkins, 2015; Shore et al., 2018), job satisfaction (Brimhall et al., 2014; Hwang and Hopkins, 2015; Shore et al., 2018), organizational citizenship behavior (Ashikali and Groeneveld, 2015; Shore et al., 2018), creativity (Li et al., 2017; Chung et al., 2020), meaningfulness of work (Mousa et al., 2021), and job performance (Chung et al., 2020), while decrease interpersonal conflict (Nishii, 2013), turnover (Brimhall et al., 2014; Hwang and Hopkins, 2015; Shore et al., 2018), and discrimination (Dwertmann and Boehm, 2016).

Although the importance of inclusion has been consistently supported by empirical research (e.g., Nishii, 2013; Dwertmann and Boehm, 2016; Chung et al., 2020) and meta-analyses (Mor Barak et al., 2016; Holmes et al., 2021), the way to foster this sense of inclusion and cultivate an inclusive workplace remains under-examined (Shore et al., 2018; Ashikali et al., 2020; Rice et al., 2021). Research has thus begun to probe into the factors that may contribute to one’s sense of inclusion. In particular, the evidence-based findings regarding which subgroups feel more or less included in an organization are quite insightful and can guide us on where our efforts should be invested for the purpose of achieving the goal of inclusion (Brimhall and Saastamoinen, 2020).

Latent profile or class analysis (LPA/LCA) is a person-centered classification analytical tool used to identify homogeneous latent classes or subgroups within a large heterogeneous sample (Hagenaars and McCutcheon, 2002; Gabriel et al., 2015). In contrast to a variable-centered approach, LPA/LCA allows us to detect profiles that differ qualitatively or quantitatively in combinations of variables (Wang and Hanges, 2011; Woo et al., 2018), and identify how the different subgroups are differentially linked to antecedents and outcomes (Wang and Hanges, 2011). This is a useful approach in identifying meaningful subtypes that are more homogeneous smaller groups within the larger heterogeneous group (da Silva et al., 2019; Vaziri et al., 2020; Wang et al., 2020).

Based on the Inclusion–Exclusion Scale (Mor Barak, 2016), Brimhall and Saastamoinen (2020) explored the topologies of inclusion with LPA using a sample from the public departments in the United States and identified two subgroups (i.e., those who feel less valued vs. those who feel more valued). Although these findings are meaningful and provocative, however, the inclusion typologies are needed to be investigated further as the number and meaning of profiles depend greatly upon sample characteristics and assessment tools (Jason and Glenwick, 2016; Wang et al., 2020). The tool used in their study was the 15-item scale developed by Mor Barak (2016), which measures the decision-making process, information networks, and one’s level of participation, and is more indicative of the causes of inclusion rather than evaluating one’s sense of inclusion per se. Meanwhile, the sample was small with only 213 participants, and from a single source focusing only on public organizations. The lack of representative sample in their study limited the validation of the profiles.

To address the limitations of the extant research, the current study aimed to use LPA to identify the inclusion profiles using the Work Group Inclusion Scale (Chung et al., 2020) and drawing on a larger sample with more participants coming from diverse organizations and multiple professions in a non-Western context. The 10-item Work Group Inclusion Scale, with five items measuring belongingness and five items measuring uniqueness, perfectly aligns with the established definition of inclusion comprising belongingness and uniqueness (Shore et al., 2011). This measure is expected to demonstrate the profiles of inclusion better. Furthermore, a representative sample is crucial to the validation of the profiles obtained. An LPA using larger sample with participants coming from various organizations and engaging in different professions is more likely to generate a more robust classification. Importantly, the Chinese culture is characterized by high collectivism (Hofstede, 1980), conformity (Goncalo and Staw, 2006), and harmony orientation derived from the dominant Confucian values (Leung et al., 2002; Kang et al., 2017; Horak and Yang, 2019). It is not thought to be as diverse as the Western contexts (Cho and Mor Barak, 2008). In this regard, conducing an LPA in such a situation can potentially extend the previous studies grounded in the Western context and identify unique profiles pertinent to the Chinese culture.

Therefore, our pilot study contributes to the existing research by identifying the patterns of inclusion in a non-Western context with a more representative sample and providing a more in-depth understanding of inclusion profiles by using a more valid measure of inclusion, subsequently shedding new light on how we can establish an inclusive workplace.

The CFA results (see Table 2) show that the expected six-factor model involving inclusion, exhaustion, engagement, opportunity for development, support from colleagues, and support from supervisor, fitted the data best (χ² = 7356.722, df = 545, CFI = 0.927, TLI = 0.912, SMR = 0.008), supporting the discriminant validity of our measures.

Table 3 provides descriptive information for study variables, as well as the Pearson correlations between them.

Table 4 presents the LPA model fit indices for the one- to five-profile models in the sample (N = 1,168). However, AIC, BIC, SSA-BIC, Entropy values, and LMR and BLRT results of each model suggest a good fit, thus we cannot determine an appropriate classification. Following the suggestions of Vaziri et al. (2020), we then graphed the elbow plots of the BIC of the five models to identify the ideal profile solution.

As can be seen in Figure 1, the elbow plot of the BIC shows that the slope of the curve flattens around the three-class point. The average probabilities of class membership for this classification were higher than 0.90 (ranging from 0.994 to 0.999, as seen in Table 4). We thus retained the three-class model as the optimal one.

Figure 2 depicts a graphic description of the three-class solution in our sample. Means and Standard Errors of the inclusion measures for each class are reported in Table 5. The majority of our sample (69.5%) belonged to a profile with moderate scores on all items of inclusion, and we therefore named the class the moderate inclusion group (Class 2). Class 3 (22.7%) was characterized by relatively high scores on each item, was thus named the high inclusion group. Although the remaining class (Class 1) accounted for only a small proportion of our sample (7.8%), it was a meaningful class with scores on all items being the lowest, thus was correspondingly named the low inclusion group.

Table 5 summarizes the DCAT and modified BCH results of the overall chi-square tests and chi-square statistics for pairwise comparisons between classes (N = 1,168) for the categorical and continuous outcome variables. As shown in Table 5, all comparisons on the external variables involved across the three subgroups were significant, with the exception of gender. Specifically, those in the high inclusion group (Class 3) tended to be more aged (M = 29.801, SE = 0.328) and tenured (M = 4.215, SE = 0.163), have obtained a master’s degree (23.4%, SE = 0.026) or above (56.5%, SE = 0.031), work in the high-tech industry (98.5%, SE = 0.007), and come from developed regions (44.6%, SE = 0.031). They also had the highest levels of engagement (M = 4.597, SE = 0.085), perceived opportunity for development (M = 4.922, SE = 0.036), support from their colleagues (M = 4.105, SE = 0.056) and supervisor (M = 3.633, SD = 0.045) while the lowest levels of exhaustion (M = 2.558, SD = 0.079). Those in the low inclusion group (Class 1) demonstrated almost the opposite patten. The profile of this group tended to be that they were relatively young (M = 21.582, SE = 0.190) and junior (M = 1.351, SE = 0.077), hold an undergraduate degree (45.0%, SE = 0.052) or below (50.6%, SE = 0.052), work in the manufacturing industry (76.9%, SE = 0.044), come from the less developed areas (64.8%, SE = 0.050), They had the lowest levels of engagement (M = 2.910, SE = 0.100), least likely to perceive the opportunity for development (M = 3.971, SE = 0.056), support from colleagues (M = 1.959, SE = 0.115) and supervisor (M = 1.965, SE = 0.094) while most likely to be exhaustion (M = 4.576, SE = 0.115). Finally, those in the moderate inclusion group (Class 2) tended to work in the service industry (52.4%, SE = 0.018) and come from moderately developed areas (26.4%, SE = 0.016). They exhibited a moderate level on all other foci variables.

Overall, the three-class profiles differ significantly on all distal outcome variables excluding gender. The pattens of the three classes regarding the links to these relevant variables demonstrate that a greater sense of inclusion in a group relates to more beneficial outcomes while fewer harmful consequences, suggesting the homogeneity within the classes and heterogeneity across classes.

To obtain evidence-based information regarding employees’ sense of inclusion, an LPA based on the scores of the Work Group Inclusion Scale (Chung et al., 2020) was conducted in a diverse sample of 1,168 Chinese employees. Three latent profiles were identified: the moderate inclusion group (Class 2) comprised the majority (69.5%) of the total sample, with moderate scores on all items of inclusion; the high inclusion group (Class 3) accounted for a smaller proportion of the sample (22.7%), scoring highest on all items; while the low inclusion group (Class 1) were only a tiny percentage of the sample (7.8%), scoring lowest on all items. The three subgroups were found to differ on the key variables, revealing more details regarding the patterns of inclusion experienced by Chinese employees.

Interestingly, our results showed that members of the high inclusion group tended to be highly educated with a master’s degree or above, work in the high-tech industry, and come from developed areas; members of the moderate inclusion group tended to work in the service industry, and come from moderately developed areas; while members of the low inclusion group tended to have an educational level of an undergraduate degree or below, work in the manufacturing industry, and come from less developed areas. These differentiations pertinent to educational attainment, industry, and regional origins across the three profiles could be due to the inequality resulting from the household registration system in China (the Hukou system). As a dominant means of managing population mobility, the Hukou system has caused the “rural–urban” divisions, determining one’s eligibility for certain state-provided services and welfare programs pertaining to education and employment, among others (Liu, 2005; Chan, 2010; Afridi et al., 2015). Those who are well educated are then more likely to get better jobs in prosperous industries. In such an institutional arrangement, members from less developed areas usually hold a rural status with fewer educational and employment opportunities, thus leading to a lower level of perceived inclusion in the workplace. Compared to them, members from moderately or highly developed areas are more likely to hold an urban status with better education opportunities and more decent jobs, thus leading them to perceiving a relatively higher level of inclusion in the workplace.

Counterintuitively, the current study did not find any differences on gender across the three profiles, contrary to the results of Brimhall and Saastamoinen (2020). Their findings revealed that the group felt less valued is more likely to be female than to be male. However, there were no such differences found across the three profiles in the current study. This could be attributed to the Women’s Liberation Movement in China began in the early 20th century (Zuo, 2013) and has had ongoing supported by the Chinese Communist Party. As a result, gender inequality has been to some extent eliminated (Zuo, 2013) with women having more equal rights and opportunities to education, political participation, and employment than before (Lee, 1995). Those women who participated in the current study all had access to decent job opportunities—some had even been hired by high-tech companies that require high levels of competence and talent. These women may thus have a sense of fair, equal, accepted, and recognized as male employees do in the workplace. Therefore, the three subgroups did not differ on gender.

Noticeably, the patterns of inclusion identified in the current research were characterized by the majority of participants feeling a moderate level of inclusion or below (Class 2 and Class 1, 77.3%), and only a smaller group feeling a high level of inclusion (Class 3, 22.7%). This could be ascribed to the seniority-based system embedded in Confucian ideals in China (Kang et al., 2017; Horak and Yang, 2019). In this type of social setting, juniors (inferiors) are expected to follow, obey, and respect their seniors (superiors; Cho and Mor Barak, 2008), and these seniors are thus more likely to feel higher inclusion in their organizations. In our study, the most majority of participants were juniors with younger ages (M = 26.77, SD = 3.09) and shorter job tenures (M = 4.71, SD = 2.33). As such, it could be that they currently are less likely to feel as respected and valued as their seniors, experiencing fewer wage increases and/or less career growth as their older, more tenured colleagues (Horak and Yang, 2019). Accordingly, they may not feel a higher sense of inclusion that comes through seniority, even though the majority do already experience a moderate sense of being included in their organizations. Similarly, those perceiving a high degree of inclusion (i.e., Class 3) tended to be relatively older and hold longer tenure in our study.

As expected, we verified that there were substantial and regular differences on distal outcome variables across the three profiles by showing that the group with a higher level of inclusion engaged more in work, perceived more opportunity for development, support from colleagues, and support from supervisor, while also experiencing less exhaustion. This result is consistent with previous studies that have suggested that a greater sense of inclusion is associated with more positive outcomes (Nishii, 2013; Hwang and Hopkins, 2015; Dwertmann and Boehm, 2016; Shore et al., 2018) and less negative outcomes (Nishii, 2013; Brimhall et al., 2014; Dwertmann and Boehm, 2016). The findings of the current study therefore support the significance of inclusion in the workplace.

The current study conducted an LPA on inclusion in the Chinese context as an initial exploration into the profiles of inclusion in a non-Western context. Using the 10-item Work Group Scale that fits well with the definition of inclusion comprising belongingness and uniqueness, our results identified three profiles of inclusion: a high inclusion group (22.7%), a moderate inclusion group (69.5%), and a low inclusion group (7.8%). Our findings suggest the uniqueness of these inclusion profiles pertaining to the Chinese context and confirm the importance of inclusion by revealing substantial differences across the subgroups on key variables. Through the use of the Work Group Inclusion Scale, this study also offers insights into how we can cultivate an inclusive workplace and ensure all member are included, that is, by striving to fulfill individuals’ needs for belongingness and uniqueness. Future research could advance our knowledge of the inclusion typologies by enhancing the representativeness of the current study sample, considering pertinent context-related factors and key external variables, and conducting cross-cultural comparative studies.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by the Institutional Review Board of Guangzhou University. The patients/participants provided their written informed consent to participate in this study.

JQ and MW conceptualized the paper and MW contributed to the investigation and helped to perform the revision of the manuscript, and provided final approval for the manuscript, JQ contributed to the analysis of the data, and drafted the manuscript. All authors contributed to the article and approved the submitted version.

This paper is supported by the National Natural Science Foundation of China (NSFC; 71972139; 71802119), the Soft Science Research Projects of Shanxi Province (2018041059–6), the Philosophy and Social Science Research Projects in Colleges and Universities of Shanxi Province (201803007).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.