Abstract:     In the field of channels, the application of social media among enterprises has begun to attract the attention of scholars. However, existing studies only focus on the positive side of the influence of social media on the relationship between enterprises, but ignore the negative side of the influence of social media. So far, no one has discussed the different impacts of social media on enterprises, which restrict our comprehensive understanding of social media. Therefore, it is necessary to test the duality of social media through quantitative empirical analysis to help enterprises better understand and make use of social media. More importantly, does the policy orientation adopted by enterprises work in the different influences of social media? It is not clear whether different policy orientation of enterprises change the effectiveness of interactive behaviors on the consequences of behaviors. Exploring this issue can provide theoretical support for enterprises to formulate effective social media policies, which are worth further study. Based on this, this paper based on the social exchange theory, explores the different impacts of the use of social media by boundary spanners on the performance of enterprises and the opportunistic behavior of partners, so as to verify the duality of the influences of social media. Furthermore, this paper further studies of the moderating effect of enterprise policy orientation on the relationship between social media interactive behaviors and its consequences, and proposes a conceptual model for the study.
    In the research design, we first put forward research hypotheses by theoretical and logical derivation aiming at studying the relationship between different variables. It is assumed that the social media interactive behavior of enterprise boundary spanners affects the performance of firms and the opportunistic behavior of partners, as well as the moderating effect of encouragement orientation and control orientation on their relationship. This study was undertaken by questionnaire. According to the purpose of the study, manufacturing enterprises were selected as the survey samples in many regions in China (including Beijing, Guangzhou, Shenyang, Dalian, Changchun, Harbin and Zhengzhou, etc.). Each sample selected respondent who is market operations to answer questions about the cooperation of one of the distributor that accounts for the largest portion of the company′s sales. Data were collected by research companies, and the survey of 603 enterprises was finally completed. During the data review, we carried out a preliminary examination and screening, and removed 53 invalid questionnaires, because the respondents of these questionnaires did not understand some specific information or provided contradictory information. We ended up with 550 valid questionnaires. During the scale design process of the questionnaire, measurement items of main variables are mainly based on the results of existing studies and combining with the answers that are selected and extracted from the field interviews of enterprise executives.
    Based on the research purpose and the survey data finally collected from 550 domestic manufacturing enterprises, we carried out the empirical analysis. First, to ensure that the study′s measurements are valid, we conducted a reliability and validity test. The results of reliability and validity test showed that all constructs in the study had good internal consistency, good construct reliability, good measurement reliability, good convergent validity, and the scale had good discriminant validity. All test results show that the measurement of the scale in this study has good validity and reliability. The measurement is effective and reliable. Then, we utilize multiple hierarchical regression analysis to test the hypothesis. We performed multivariate hierarchical regression analysis by taking task-based and relational-based interactive behaviors as independent variables, encouragement orientation and control orientation as moderating variables, and the performance of firms and the opportunistic behavior of partners as outcome variables. The results of the hypothesis test show that: (1) more task-based and relational-based interactive behaviors can bring higher performance to firms; more task-based interactive behavior can reduce the opportunistic behavior of partners, but more relational-based interactive behavior can promote the opportunism. (2) The encouragement orientation of firms could strengthen the relationship between relational-based interactive behavior and the performance of firms. The control orientation could weaken the relationship between relational-based interactive behavior and the performance. However, the influence of encouragement or control orientation on the relationship between task-based interactive behavior and firm′s performance is inconspicuous. (3) The encouragement orientation of firms can respectively strengthen the relationship between task-based or relational-based interactive behavior and the opportunistic behavior of partners, while the control orientation weakens the relationship between them. 
     The research has certain significance for the management practice of enterprises using social media. On the one hand, this study verifies the duality of social media influence on enterprises, and finds that relational-based interactive behavior is the key to the opportunistic behavior of partners. The results provide theoretical guidance for enterprises to better understand and apply social media. Enterprises needs to decide whether to choose social media for interaction based on their own interaction need. For example, enterprises with a high degree of tolerance for partner opportunism may consider allowing boundary spanners to interact with social media to help enterprises achieve higher performance. However, enterprises with low tolerance for partner opportunism should avoid the risks associated with relational-based interactive behavior of boundary spanners. On the other hand, we explored the moderating effect of enterprise policy orientation on the effect of social media interactive behavior. The results provide theoretical support for enterprises to formulate effective social media policies and provide a new ideal for enterprises to avoid the adverse effects of relational-based interactive behavior. For instance, when enterprises believe that boundary spanners are likely to collude with partners to harm the interests of enterprises, they can adopt control orientation to reduce the opportunism. However, enterprises may choose to use encouragement orientation to help enterprises achieve better performance when they believe that boundary spanners do not have satisfactory business contacts with partners or complete work tasks.

Key words:  social media, policy orientation, interactive behavior, performance, opportunism