Toward an Integrated Model of Public Engagement on Corporate Social Networking Sites: Antecedents, the Process, and Relational Outcomes

Abstract

The importance of social media engagement as an influential factor for cultivating quality organization-public relationships has been increasingly recognized by strategic communication scholars and professionals. However, there exist few theoretical deliberations on the underlying mechanisms of organization-public engagement on social media. This study focused on organization-public engagement and relationship management via social networking sites. We proposed and tested an integrative model that incorporated the antecedents of social media dependency, parasocial interaction, and community identification, as well as the effects of user engagement on relational outcomes of trust, satisfaction, and commitment. Acknowledging the worldwide popularity of social networking sites, the proposed model was tested among users of leading social networking site in China (i.e., Renren, Sina Weibo), the country with the world's largest Internet population and second-largest economy. The study findings well supported the proposed conceptual model. Theoretical contributions and managerial implications are also provided.

Notes

¹Established in 2005, Renren, which literally means “everyone” in Chinese, is widely considered the “Facebook of China” (Geron, 2011) and is the country's “most popular, most open and best-financed social network site” (Lukoff, 2010, para. 16). Sina Weibo, another dominant SNS, is a microblogging site that blends Twitter and Facebook and has more than 300 million users, as of 2012 (Hunt, 2012).

²According to Kline (2005), SEM is a technique that can be applied to both nonexperimental and experimental data to verify a priori models comprised of latent variables or a mix of latent and observable variables. Thus, in the present study, structural SEM was used as the primary statistical method to test the hypothesized model.

³According to Hu and Bentler (1999), a cutoff value close to 0.95 for CFI and Tucker-Lewis index (TLI); a cutoff value close to 0.08 for standardized root mean square residual (SRMR); and a cutoff value close to 0.06 for RMSEA indicate good fit between the hypothesized model and the observed data. Additionally, according to Hu and Bentler's (1999) joint cutoff criteria, an SEM model with CFI, TLI ≥ 0.95 and SRMR < 0.10 or RMSEA ≤ 0.06 and SRMR ≤ 0.10 suggests that the fit between the data and the proposed model is reasonable.

⁴According to Kline (2005), a single-fit index reflects only a particular aspect of model fit and a favorable value of that index does not by itself indicate good fit. There is no single “magic index” that provides a gold standard for all models. The chi-square is the most commonly reported measure of model-data fit. However, it is strongly dependent on the sample size.

⁵According to Byrne (2010), bootstrapping is a procedure in which one takes repeated, smaller random samples of an existing sample to develop empirical estimates of standard errors of any parameter. Bootstrapping is a common procedure used to address multivariate nonnormality issues.

⁶According to Keith (2006), a standardized coefficient (â) of less than 0.05 suggests a negligible effect; a standardized coefficient of 0.05–0.10 suggests a small but meaningful effect; a standardized coefficient of 0.10–0.25 suggests a moderate effect; and a standardized coefficient of above 0.25 suggests a large effect.