Subordinate moqi reflects a state in which subordinates understand supervisors or have silent consensuses with supervisors without verbal communication (Zheng et al., 2019a). The stronger subordinate moqi means the better subordinates understand their supervisors. When supervisors imply something in a non-verbal way, subordinates can tacitly perceive and understand it and then conduct consistent actions in their supervisors’ expectations (Zheng et al., 2019a). During the moqi generating process, supervisors are information transmitters, while subordinates are information receivers and interpreters. Why can one party tacitly comprehend the other party without speaking? Some scholars believe that traditions, norms, and assumptions of socio-cultural language are the sources of the consistent actions between two parties (Yao and Chen, 2017).

Zhang and Zhang (2011) believe that keeping consistent with supervisors in intrinsic ideology is good for subordinates to achieve consistent understanding and actions toward the goals. Some studies explore the antecedents of subordinate moqi from the perspective of subordinate-center. Recent empirical research found that subordinates’ explicit and implicit feedback-seeking were two significant predictors of subordinate moqi (Zheng et al., 2019a). Besides, Wang et al. (2018) found that employees’ perceived supervisory status, as one antecedent of subordinate moqi, positively influenced subordinate moqi via feedback-seeking. Additionally, from the perspective of “leader-center,” Zhong et al. (2021) provided evidence that leader humility can significantly predict subordinate moqi by an empirical research. Subordinate moqi links employees and their supervisors. Accordingly, we infer that from the perspective of the link between subordinate and supervisor, factors related to subordinate-supervisor traits may affect subordinate moqi, such as employees’ deep-level similarity with supervisors. The similarity is one important term when we discuss the interaction between subordinate and supervisor, especially in discussing self-categorization (Hogg and Terry, 2000; Zhao et al., 2014). However, it is not clear whether or how employees’ deep-level similarity with supervisors triggers subordinate moqi. Thus, this study aims to discuss the relationships between deep-level similarity and subordinate moqi.

The self-categorization theory developed from the social identity theory assumes that the comparison between one party and other parties in related social activities triggers the functioning of the social self-concept (Oldmeadow et al., 2003). Generally, there are three comparison levels of self-categorization. The theory believes that individuals are human beings that differ from other creatures. Subsequently, individuals can be insiders or outsiders of groups according to their race, nation, or others. Finally, an individual is a unique unit that differs from others in groups because of the differences in personality or other forms (Haslam and Reicher, 2015). The comparison makes individuals form into different clusters, and they can be in-group or out-group members (Hornsey, 2008). People will perceive others with similarities as insiders or others with distinctions as outsiders (Hogg and Terry, 2000). Similarities affect individuals’ cognitive consistency (Morry, 2005) and subsequent behaviors (Li et al., 2017). Through self-categorizing, individuals’ similarity will stimulate individuals’ self-classification actions and inspire individuals to act as consistently as others in the same groups. Furthermore, self-categorizing will satisfy individuals’ self-enhancement and reduce uncertainty (Hogg and Terry, 2000).

Individuals not only expect to keep consistency with persons who belong to the same categorization but also try to reach consensus on issues (Haslam and Reicher, 2015). Researchers have found that agreeableness was related to cooperation and helping behavior (Graziano et al., 2007). In other words, individuals in the same category will achieve consistency with in-group members, cooperate with and help each other. According to the above theory, we argue that deep-level similarity between subordinates and supervisors may affect subordinate moqi.

Similarly, anything that is shared among persons includes surface-level similarity and deep-level similarity (Ndiaye, 2011). Harrison et al. (1998) assert that demographic variables can characterize the surface-level similarity. However, Ndiaye (2011) believes that the deep-level similarity is a common characteristic existing in behaviors or interactions (e.g., attitude, belief, and value similarity). Previous research found that deep-level similarity was a predictor of individuals’ behaviors and work outcomes. When employees are deep-level similarity with their colleagues, they will actively seek feedback (Kammeyer-Mueller et al., 2011) which is related to subordinate moqi (Zheng et al., 2019a). We assume that subordinates’ deep-level similarity with supervisors may induce subordinate moqi. Deep-level similarity reveals that individuals’ consistencies in belief, value, and attitude significantly affect individuals’ judgment, decision, and action on issues or goals (Ndiaye, 2011). When subordinates have deep-level similarities with their supervisors, they will more tend to maintain consistency in intrinsic ideology and show consistent understanding and action of the goals (Zhang and Zhang, 2011). Therefore, when supervisors convey implicit clues to subordinates, their similarities promote subordinates’ capability to accurately explain these clues, correctly understand the real intentions of supervisors beyond the implicit information, and conduct the same behaviors that satisfy supervisors’ expectations. Additionally, previous research found employees who have deep-level similarity with their colleagues engaged in more proactive feedback-seeking (Kammeyer-Mueller et al., 2011). Existing study found that proactive feedback-seeking is positively related to subordinate moqi (Zheng et al., 2019a). According to the above, we infer that the more subordinates are deep-level similarity with supervisors, the more they achieve subordinate moqi with supervisors. Thus, we hypothesize:

Leader-member exchange (LMX) reflects the quality of exchange in relationships between leaders and members (Gomez and Rosen, 2001). Subordinates with low-level LMX relationships generally feel that they are out-group people, while employees with high-level LMX consider themselves as in-group people (Zhao et al., 2014). In-group employees can attain more resources (e.g., information, confidence, and involvement) and achieve good work outcomes (Li et al., 2017), whereas out-group employees have few opportunities (Gomez and Rosen, 2001). In this study, we adopt the self-categorization theory to discuss the relationships between deep-level similarity and LMX.

In this study, deep-level similarity emphasizes the homogeneity in value belief, work type, and work attitude between subordinates and supervisors (Ensher et al., 2002). The self-categorization theory proposes that similarity is a prominent indicator of classification (Hogg and Terry, 2000; Haslam and Reicher, 2015). We assume that deep-level similarity, as an indicator of classification, may stimulate employees with a high level of deep-level similarity with supervisors to classify themselves into the same class as their supervisors, and further increase the interpersonal attractiveness (Zuo and Gao, 2008). As the interpersonal attractiveness increases, employees’ cognitive factors related to the LMX relationship (Dulebohn et al., 2012), such as positive affect, employees’ perceived insider status, and psychological security, will increase (Gonzaga et al., 2007; Li et al., 2017). Moreover, previous studies have provided supportive empirical evidence that actual similarity (e.g., competence, personality, and expectations) was directly associated with LMX (Montoya et al., 2008; Erdogan and Bauer, 2015; Liao and Hui, 2021). According to these clues, we infer that subordinates may build a high-quality LMX relationship when they are deep-level similarity with their supervisors. Thus, we hypothesize:

Both LMX and subordinate moqi are relational constructs that reflect the interactive relationships between subordinates and supervisors. However, LMX is mainly based on trust, respect, and obligation (Graen and Uhl-Bien, 1995) and emphasizes an exchange relationship between subordinates and supervisors based on social exchange (Graen and Scandura, 1987). In this relationship, both supervisors and subordinates can be initiators of exchange behaviors, and they can get benefits through LMX. Differently, subordinate moqi is mainly based on cognitive consistency (Zhang and Zhang, 2011) and reflects a state of subordinates’ understanding and cooperating with their supervisors when they do not speak with their supervisors (Zheng et al., 2019a). In the subordinate moqi context, supervisors are implicit information senders, while subordinates are implicit information capturers and interpreters who can explain implicit information accurately. In a word, subordinate moqi emphasizes that subordinates have reached a consistent cognition with their supervisors on information or actions (Zhang and Zhang, 2011).

We believe that LMX may positively help subordinates cultivate a state of moqi with their supervisor. The emergence of subordinate moqi requires that subordinates are familiar with their supervisors and clearly understand their supervisors. Employees with subordinate moqi can correctly understand supervisors’ concealed expectations, intentions, or desires based on implicit information conveyed by supervisors (Zheng et al., 2019a). Subordinates may go through three steps to achieve the state of moqi with supervisors. Firstly, they should walk close to supervisors to interact. Then, they try to interact with supervisors to be familiar. Finally, they will be familiar to understand implicit information from supervisors (Zheng et al., 2019a). During this process, interacting with supervisors, as a form of LMX, may determine the success of establishing a state of moqi with their supervisor.

When subordinates have high-level LMX, they will acquire a series of positive psychological resources, such as trust, respect, and obligation (Graen and Uhl-Bien, 1995), which will lead subordinates to be more willing to walk close to and support supervisors (Choy et al., 2016; Liao and Hui, 2021). The more subordinates exchange with their supervisors, the more benefits they get, which narrows the psychological distance between subordinates and supervisors, and increases their further interactions (Gomez and Rosen, 2001; Liao and Hui, 2021). As the frequency of interaction increases, subordinates and supervisors become more familiar with each other and “match” with each other. It creates preconditions for subordinates in comprehending supervisors’ ideas unspoken. Furthermore, since employees with a high-quality LMX often have more opportunities to access information related to their jobs (Graen and Scandura, 1987), they may have advantages in forming consistent cognition with supervisors because of the supports of the availability of adequate information for helping them comprehend, judge supervisors’ real intentions, and achieve subordinate moqi (Zhang and Zhang, 2011). Finally, subordinates with high-level LMX have stronger self-efficacy and are more likely to explore unknown things (Zhao et al., 2014); therefore, they may be motivated to interpret implicit information conveyed from supervisors and to achieve a state of moqi with supervisors (Zheng et al., 2019a). Thus, we hypothesize:

According to the above discussion, as the degree of subordinates’ deep-level similarity with their supervisor increase, the sense of being insiders of subordinates with supervisors will increase. This sense will enhance the interactions between subordinates and supervisors and the quality of the LMX relationships (Hypothesis 2). The frequent interactions between subordinates and supervisors will make subordinates familiar with their supervisors. As the two parties (i.e., subordinates and supervisors) know each other, subordinates may understand supervisors even if they do not talk to each other. If two parties can keep achieving tacit understanding and agreements in the workplace context, they can achieve, namely moqi (Hypothesis 3). Thus, we propose that the influence of deep-level similarity on subordinate moqi will be indirectly transmitted via LMX. We hypothesize:

We collected the data from employees of 9 firms in Shanghai, Beijing, and Chongqing municipalities, and Jiangsu, Zhejiang, and Guangdong Provinces. The fields of these firms are electrical appliance sales, information technology, and plastic product production. In each company, we introduced the research content to a manager and invited this manager to assist in distributing the online questionnaire among their colleagues. Every participant is distributed a unique code to match their completed questionnaires at time 1 and at time 2. To control the negative effect of common method variance (Podsakoff et al., 2003), we surveyed two times. In period 1, we invited 523 participants to report their deep-level similarities with their supervisors. One month later, we asked these 523 participants to report LMX and subordinate moqi and collected 451 questionnaires. Each participant was rewarded with a red envelope after completing the survey and their responses to the questionnaire were kept confidential. Since some participants did not write correctly their unique code at time 1 and time 2, namely, the code of the second time was not the same as the one of the first time. These questionnaires did not be matched correctly, thus can’t be used. Finally, we got 316 valid data for the study.

Among the final participants, 68% of the participants were males. 35.4% were between 26 and 30 years old, and 46.5% were between 30 and 40 years old. While only 3.8% of respondents were beyond 40, and 14.2% were between 18 and 25 years old. In terms of the time of working with supervisors, 44.6 percent of subordinates worked with their supervisors for 1–3 years. In contrast, 23.1 percent of subordinates worked with their supervisors for less than 1 year, and 32.3 percent of subordinates worked with their supervisors beyond three years.

Additionally, 17.1% respondents have worked less than 1 year, 40.8% for 1–3 years, 12.7% for 3–5 years, 16.1% for 5–8 years and 13.3% for more than 8 years. Finally, 25.9 percent of respondents had Associate’s degrees, 35.1 percent had Bachelor’s degrees, and 34.2 percent of respondents had high school degrees. However, a small group of participants had a Master’s qualification (4.7%).

This study used the seven-point Likert scales (from 1 = strongly disagree to 7 = strongly agree) to measure the variables.

First, we adopted SPSS18 software to conduct descriptive analysis and correlation coefficient analysis among study variables. Then, we used MPLUS7 software to conduct confirmatory factor analyses (CFA) to test the measurements’ discriminant validity. Finally, we applied the Structural Equation Modeling and the bootstrapping approach in MPLUS 7 to test the hypotheses.

The means, standard deviations and correlations of variables exhibit in Table 1. It shows that subordinate moqi is positively correlated with deep-level similarity (r = 0.52, p < 0.001), and leader-member exchange (r = 0.73, p < 0.001), respectively. Additionally, Table 1 reveals that the leader-member exchange is positively correlated to deep-level similarity (r = 0.54, p < 0.001).

We adopted two methods to examine the common method bias following the suggestions of Podsakoff et al. (2003). First, this study performed Harman’s single-factor test to diagnose the common method variance. Results showed that three factors were extracted to account for 65.76% of the variance, and the first factor accounted for 47.96% of the variance (less than 50%). Second, we applied the controlling for the effects of an unmeasured latent factor method to ascertain the influence of common method variance. Result showed that the fitness of the model with a latent common methods variance factor (CFI = 0.94; TLI = 0.93; RMSEA = 0.06; SRMR = 0.04) did not obviously outperform the three-factor model without a latent common methods variance factor (CFI = 0.92; TLI = 0.92; RMSEA = 0.07; SRMR = 0.05). Since the change of RMSEA is 0.01, the change of CFI is 0.02, and TLI and SRMR are 0.01, we believe that the common method bias is not serious (Xie and Long, 2008).

We adopted CFA to check the degree of constructs’ distinctiveness. Table 2 shows that three-factor model has a superior fit and good discriminant validity (χ² _{df =227, N = 316} = 534, p < 0.001, CFI = 0.92, TLI = 0.92, RMSEA = 0.07, SRMR = 0.05). The three-factor model (M1) better fits the data than alternative models, including two-factor model (M2) that combines leader-member exchange and subordinate moqi into one factor (χ² _{df =229, N = 316} = 750, p < 0.001, Δχ²(2) = 216, p < 0.01); two-factor model (M3) that integrates deep-level similarity and LMX into one factor (χ²_{df =229, N = 316} = 1,085, p < 0.001, Δχ²(2) = 551, p < 0.01); two-factor model (M4) that combines deep-level similarity and subordinate moqi into one factor (χ² _{df =229, N = 316} = 1,298, p < 0.001, Δχ²(2) = 764, p < 0.01); and one-factor model (M5) that merges deep-level similarity, leader-member exchange and subordinate moqi into one factor (χ²_{df =230, N = 316} = 1,494, p < 0.001, Δχ²(3) = 960, p < 0.01). Results offer evidence for the good discriminant validity of the variables.

Hypothesis 1 infers that deep-level similarity will significantly predict subordinate moqi. Figure 1 shows that the path coefficient between deep-level similarity and subordinate moqi is positive and significant (γ = 0.15, p = 0.014). Thus, it supports Hypothesis 1.

Hypothesis 2 assumes that deep-level similarity will significantly predict the leader-member exchange. Figure 1 shows that deep-level similarity (γ = 0.41, p < 0.001) is significantly correlated to the leader-member exchange. Therefore, it supports Hypothesis 2.

Hypothesis 3 proposes that the LMX is positively associated with subordinate moqi. As shown in Figure 1, the path coefficient between the leader-member exchange and subordinate moqi is significant (γ = 1.05, p < 0.001). Thus, it supports Hypothesis 3.

Hypothesis 4 predicts that LMX mediates the influence of deep-level similarity on subordinate moqi. We used a bootstrapping approach to generate the indirect effect coefficient and the confidence intervals of the indirect effects. Table 3 reveals that the indirect influence of deep-level similarity on subordinate moqi through the leader-member exchange is significant (γ = 0.43, p < 0.001). The 95% confidence interval of the mediating effect is (0.331, 0.567), and correspondingly zero is excluded. Thus, hypothesis 4 is supported.

First, this study provides researchers and practitioners with a new angle to understand the antecedents of subordinate moqi. Previous researchers have discussed the antecedents of subordinate moqi from the perspective of subordinates and found that subordinate feedback-seeking (Zheng et al., 2019a) and perceived supervisory status (Wang et al., 2018) could predict subordinate moqi. From the perspective of leader, Zhong et al. (2021) examined the effect of leader humility on subordinate moqi. However, not enough papers have discussed other antecedents of subordinate moqi from the similar traits between subordinates and supervisors’ perspectives. Given that subordinates’ characteristics can influence subordinate moqi, similar traits (i.e., deep-level similarity), as a kind of characteristics between subordinates and supervisors, may affect subordinate moqi. According to the self-categorization theory, we investigate the influence of deep-level similarity on subordinate moqi from the perspective of the connection between subordinates and supervisors. This study provides evidence that the traits linking subordinates and supervisors can affect subordinate moqi.

Second, our study provides researchers with a possible path between deep-level similarities on subordinate moqi, which serves as a modest spur to induce other researchers to come forward with more contributions. Previous studies have not investigated how the traits linking subordinates and supervisors promote or discourage subordinate moqi. From the perspective of self-categorization, when subordinates and supervisors have more deep-level similarities, they may regard themselves as insiders of the same clusters (Haslam and Reicher, 2015). This sense of being in the same groups will enhance the communication between subordinates and supervisors to achieve high-quality LMX relationships (Liao and Hui, 2021), and indirectly promote subordinate moqi. Additionally, this study found that deep-level similarities and LMX are positively related to subordinate moqi. These findings extend our understanding of subordinate moqi from the points of similarities and LMX.

Third, the current research also contributes to followership literature. Previous scholars mainly explore the influencing factors of followers’ behavior from the perspective of a leader or leadership; however, few explore the effect of traits linking subordinates and supervisors on people’s behaviors. This study found that employees’ deep-level similarity with supervisors can induce subordinates’ understandings and cooperation with their supervisors. From the perspective of the traits linking subordinates and supervisors, our findings provide important evidence that subordinates’ similar characteristics (deep-level similarity) with supervisors can play a vital role in influencing subordinates’ behaviors toward their supervisors.

This study proposes several vital implications for organizations. Existing research has provided evidence that subordinate moqi can positively influence employees’ task performance and knowledge sharing behavior (Zheng et al., 2019a,b). To adopt the subordinate moqi as a tool in management and increase employees’ positive behaviors and performance, organizations may concern the deep-level similarity with supervisors, a predictor of LMX and subordinate moqi, and hire employees who have deep-level similarity with supervisors (Kammeyer-Mueller et al., 2011). When HR managers interview candidates, they may evaluate candidates’ values, and working methods through surveys to judge the matching degree of the deep similarity between departments and candidates. Additionally, organizations should cultivate open communication cultures between employees and their supervisors, and provide them with convenient communication channels to help them learn relevant work experience or knowledge from each other. These behaviors can help subordinates and supervisors form deep-level similarities, inspire subordinate moqi, and further generate positive outcomes (Zheng et al., 2019a,b; Zhou and Zheng, 2019).

Our study finds that LMX plays an essential role in explaining how employees’ deep-level similarities with their supervisors build subordinate moqi. These findings support that if subordinates have high-quality LMX relationships, they will have high-level subordinate moqi. Thus, managers should concern more about how to inspire LMX. Firstly, managers should actively seek sufficient resources from organizations for the initiation of LMX. These resources can help supervisors lay the foundation for establishing high-quality LMX (Dulebohn et al., 2012). Secondly, organizations should provide a fair reward system in which employees can receive rewards based on their contribution, which accordingly fosters high-quality LMX relationships (Liao and Hui, 2021). Thirdly, organizations should enhance effective leadership programs training to managers to inspire line managers’ perceptions of LMX and moqi.

Several limitations in this study should be concerned. First, although we used the time-lag approach to control common method variance rising in the self-reported data in this study, the effect of common method variance cannot be utterly avoided. To better measure the variables and understand the cause-effect of the phenomenon, scholars should collect data from multiple sources and use other research designs (e.g., between-subject design experiment) to investigate the underlying mechanism. Second, although the current study has examined LMX as a mediator in the effects of deep-level similarity on subordinate moqi, other variables may also play mediating roles in such relationships. For example, trust-in-supervisor, perceived insider status, and role identity may be possible variables that explain the effect of deep-level similarity on subordinate moqi. Thus, future research should test the effects of other mediators to comprehend further how and why deep-level similarity leads to subordinate moqi. Third, we collect data only from Chinese firms in this study. Considered high context, face, and high power distance culture are salient in China and are related to fostering moqi as well (Zheng et al., 2019a), our findings remain further discussions in generalization in wide cultural contests and industrial contexts.