Abstract: As a form of constructive, change-oriented communication, voice behavior contributes to the achievement of organizational goal. However, such behavior will bring potential risks to the speaker, resulting team members refuse to voice when they facing team problems. Thus, how to effectively motivate voice behavior of team members to promote team performance has become an important issue for both managers and researchers. Previous scholars have discussed the mechanism of voice behavior from three levels, among which individual psychological cognition has attracted a lot of attention. Sense of power, as a psychological state of individuals, significantly affects individual behavior. Previous studies about sense of power and voice behavior were mainly based on the situated focus theory, but barely explored from the perspective of internal self-cognition mechanism.    Therefore, based on approach-inhibition theory, our research aims to explore how and when sense of power influences voice behavior. Several hypotheses are proposed: (1) team members′ sense of power was positively related to constructive change of sense of responsibility; (2) constructive change of sense of responsibility mediated the relationship between sense of power and voice behavior; (3) team legitimacy moderated the relationship between sense of power and constructive change of sense of responsibility; (4) team legitimacy moderated the indirect effect of sense of power on voice behavior through constructive change of sense of responsibility.The participants of this study were 311 full-time employees from manufacturing enterprises in Zhejiang province. In order to reduce the common method bias, our research collected data in three waves with one-month interval, involving 69 teams. At time 1, we invited all 478 team members to rate sense of power, constructive change of sense of responsibility. At the same time, we collected related demographic information including gender, age, education, and team size. One months later, at time 2, we invited all 401 team members who had returned time-1 survey to rate legitimacy of team power. Another one months later, at time 3, we invited all 345 team members who had returned both time-1 and time-2 surveys to rate their own voice behavior. The final valid sample consisted of 311 employees (for a final response rate of 65.06%). Regression analysis, Monte Carlo analysis, and bootstrapping technique were used to test the mediation, moderation, and moderated-mediation relationships among the study variables. Consistent with our predictions, the results indicated that team members′ sense of power was positively related to constructive change of sense of responsibility. And the relationship between sense of power and voice behavior was mediated by constructive change of sense of responsibility. Moreover, team legitimacy negatively moderated the relationship between sense of power and constructive change of sense of responsibility, and also moderated the whole mediating mechanism.This study offers several theoretical contributions. Firstly, drawing upon approach-inhibition theory, this study revealed the mechanism that sense of power stimulated constructive change of sense of responsibility and then promoted the generation of voice behavior from the perspective of internal cognition mechanism. It not only expanded the research on the internal psychological cognitive mechanism of voice behavior, but also enriched the extension and application field of the approach-inhibition theory of power. Secondly, this study has further painted a more complete picture by exploring the contingent effect of team legitimacy, which to some extent extended previous studies to examine from across-level perspective. In general, there are still some limitations to be improved in the future. Firstly, in order to fully reveal the internal mechanism of voice behavior, formation path based on other theoretical perspective should be tested. Secondly, future research could try to explore other potential antecedents as there are many factors affecting team members′ voice behavior, some uncontrolled factors such as team culture and leadership style may influence the applicability of the research results. Therefore, more situational variables such as team power distance can be further taken into consideration in future studies. Thirdly, the research data of this study were from manufacturing enterprises, such that the universality of the conclusion remains to be tested. Future research should test different types of teams to further examine the generalizability of the conclusions of this study.

Key words: sense of power, team legitimacy, constructive change of sense of responsibility, voice behavior