Social Media Research: An Assessment of the Domain's Productivity and Intellectual Evolution

AbstractThe purpose of this study is to conduct a bibliographic investigation and meta-analysis of the full body of social media scholarship produced over eight years, since the domain's emergence in 2004. A total of 610 journal and conference papers were carefully reviewed and subjected to bibliometric and meta-analysis techniques. A number of research questions pertaining to country, institutional, and individual productivity, as well as research design and data practices in the social media field, were proposed and answered. Our results reveal two main challenges faced by the field. First, the social media domain displays limited intellectual diversity in terms of productive and impactful actors—individual, institutions, and countries—as well as publications that have hitherto skewed the domain's focus in a limited direction. Second, the research design approaches and data practices characterizing the domain seem to reflect methodological singularity characterized by a strong tendency for cross-sectional, individual-level, survey or case-based studies. Furthermore, speculative and anecdotal evidence, based on personal opinions and armchair hypotheses, is extremely widespread and stand in the way of the domain's methodological and theoretical advancement. These challenges not only help to improve one's understanding of the identity and intellectual core of social media as a distinct scientific field but can also further prompt academic debate and careful (re)examination of the domain's scholarly practices and assumptions to ensure its future advancement in the most productive manner.Keywords: Social MediaSocial Network SitesOnline Social NetworksBibliographic AnalysisResearch ProductivityMeta-analysisLiterature ReviewData PracticesResearch Design Notes1 For a detailed historic overview of influential scholars in the domain of scientometrics, see Garfield (Citation2009).2 From the 610 papers, 604 were journal articles and 6 were conference papers.3 The initial search resulted in 1,516 papers, of which 466 papers were duplicates. Therefore the 1,050 unique scholarly papers is the total count after removing the duplicate papers from our data-set.4 We had access to the e-resources of three large University libraries.5 Cohen's kappa coefficient is a statistical measure of interrater agreement and it is generally considered to be a more robust measure than simple percentage agreement because it takes into account the agreement occurring by chance. A Cohen's kappa coefficient of 0.75 is considered substantial agreement (Landis & Koch, Citation1977). The reported coefficient is a cumulative (omnibus) score computed across the six categories of the coding scheme. The Appendix reports Cohen's kappa coefficients per category of the coding scheme.6 Based on the normalized scores calculated through the equal credit approach of each author associated with a particular institution.7 Mixed method refers to the explicit combination of quantitative and qualitative methods in a single study, not to the combining of various quantitative methods or multiple qualitative methods in a single study.