摘要
ABSTRACTTreating relative leader-member exchange (i.e., RLMX) as the objective and leader-member exchange social comparison (i.e., LMXSC) as the subjective operation of LMX comparison (i.e., LMXC) within a team, we integrate the RLMX and LMXSC literature and examine when higher LMXC is not always effective in employees’ workplace outcomes (namely, self-efficacy, task performance, and creativity). Revisiting social comparison theory, we propose that LMXC has positive relationships with team members’ task performance and creativity via the role of self-efficacy. Furthermore, team-level LMX disparity and team members’ neuroticism affect the above direct and indirect relationships. Specifically, we hypothesize that amid low LMX disparity, members are more likely to perform assimilation rather than comparison that weakens the positive impacts of LMXC. Team members with high neuroticism are prone to make upward rather than downward social comparisons, which also mitigates the positive impacts of LMXC. We test our hypotheses in a field (using RLMX as the objective LMXC measure; n = 559, N = 71) and an experimental study (using LMXSC as the subjective LMXC measure; n = 176). Generally, we find support for our hypotheses. Our study thus deepens the understanding of the effectiveness of LMXC within team contexts.KEYWORDS: RLMXLMXSCLMX disparityneuroticismsocial comparison Disclosure statementNo potential conflict of interest was reported by the author(s).The data that support the findings of this study are available on request from the corresponding author.Ethical approvalAll participants were fully informed that this study was anonymized. This study was organized in non-interventional way and the data were only used for a study focusing on leadership effectiveness. There were no risks during completing the survey.Informed Consent StatementWe have got consent from all participants during the survey. All participants were informed that they could skip any question he or she wished or quit the survey at any time. All participants were informed that this study was anonymized and the conclusions were only used for scientific studies.Notes1. Although LMXSC studies have demonstrated that LMXSC captures one focal employee’s subjective comparison of LMX with that of other team co-workers rather than average team LMX level (Vidyarthi et al., Citation2010), we suggest that the reference in LMXSC can be shifted based on the research question. For instance, Pan et al. (Citation2021) have shifted the reference of LMXSC to one particular co-worker. Hence, the reference in LMXSC can be shifted to team average LMX level, rendering LMXSC, basically, the subjective measure of LMXC.2. Although Vidyarthi et al. (Citation2010) have proposed that RLMX is the antecedent of LMXSC, they have found that LMXSC has a relatively high and positive relationship with RLMX (.79, p < .01). In this sense, we consider these two concepts compatible.3. We also applied LMX separation to test our hypothesized model. According to Harrison and Klein (Citation2007), LMX separation is calculated using the standard deviation of LMX within each team (as also seen in Tremblay et al., Citation2021). The results remained virtually the same. These results are available upon request from the first author.4. Because RLMX is calculated via the group-mean centre approach, we did not calculate the group-mean centre for it again. However, we defined it as the within-group level variable to reduce model miscalculations. Similar treatment was given to LMX, which was only defined as a within-group level variable and did not undergo group-mean centring (as RLMX is essentially calculated by the group-mean centring of LMX).5. As creativity and task performance were both obtained from supervisors, we added a common method factor with items from task performance and creativity, loading it to control for any potential common method effect. Although LMX and neuroticism were collected at the same time from the same source, we focused on the role of RLMX (LMX minus the team LMX mean); hence, we did not perform this common method test. The covariance between the common method factor and other latent factors was fixed at zero because the method effect concerns only the measures, not the constructs indicated thereby.6. In our regression analysis, we added a common method factor comprising five factors and specified this common method factor to have zero correlations with the focal variables (Podsakoff et al., Citation2003). This approach facilitated our interpretation of the findings by taking the CMV into consideration.7. We further tested the three-way interaction of RLMX, LMX disparity, and neuroticism on self-efficacy. The interaction effect among RLMX, LMX disparity and neuroticism on self-efficacy was significant (estimate = 1.19, SE = .58, p = .038). We followed Walker et al. (Citation2014) to examine the three-way interactions. Under low neuroticism, RLMX had a strong main effect, while the moderating role of LMX disparity was nonsignificant (slope difference = .05, t = 2.50, p > .05). Individuals with high neuroticism are thus more likely to make comparisons based on both RLMX and LMX disparities because of their insecure traits. Hence, the two-way interaction between RLMX and LMX disparity was significant, with a slope difference of .68 (t = 48.57, p < .001). Such findings affirm our statement suggesting that highly neurotic group members tend to engage in upward comparison.Additional informationFundingThis study was funded by the National Science Foundation of China under Grant [71902061], Grant [71902023] and Grant [72072116].