Predicting Retweeting Behavior on Breast Cancer Social Networks: Network and Content Characteristics

Abstract

This study explored how social media, especially Twitter, serves as a viable place for communicating about cancer. Using a 2-step analytic method that combined social network analysis and computer-aided content analysis, this study investigated (a) how different types of network structures explain retweeting behavior and (b) which types of tweets are retweeted and why some messages generate more interaction among users. The analysis revealed that messages written by users who had a higher number of followers, a higher level of personal influence over the interaction, and closer relationships and similarities with other users were retweeted. In addition, a tweet with a higher level of positive emotion was more likely to be retweeted, whereas a tweet with a higher level of tentative words was less likely to be retweeted. These findings imply that Twitter can be an effective tool for the dissemination of health information. Theoretical and practical implications for psychosocial interventions for people with health concerns are discussed.

Retweeting Risk Communication: The Role of Threat and Efficacy

“…The size of follower networks has been shown to positively influence message passing (Kim, Hou, Han, & Himelboim, ; Suh et al., ). In general, larger networks predict increased message passing.…”

An Influence Prediction Model for Microblog Entries on Public Health Emergencies

“…The attributes of the forwarding behavior consist of forwarding quantity, forwarding time, and forwarding intention according to the research purpose (Palovics, Daroczy, & Benczur, 2013). Researchers predict microblog forwarding behavior through a variety of algorithms and models, including the factor graph model (Yang, Guo, Cai, Tang, Li, Zhang, & Su, 2010), social network analysis and content analysis (Kim, Hou, Han, & Himelboim, 2016), the classification method (Hong, Dan, & Davison, 2011), the probabilistic collaborative filtering model (Zaman, Herbrich, Gael, & Stern, 2010), improved passive-aggressive algorithm (Petrovic, Osborne, & Lavrenko, 2011), the predictive model for retweeting based on conditional random fields (CRFs) (Peng, Zhu, Piao, & Yan, 2011), and so forth. It has been found that users' competition or cooperative interaction (Kong, Mao, & Liu, 2016), characteristics of the forwarding agent (Maleewong, 2016), and network structure (Gao, Ma, & Chen, 2014) all influence their forwarding behavior.…”

How Social Networks May Influence Cancer Patients' Situated Identity and Illness-Related Behaviors

“…Online or internet interactions not based on a listserv or online support group can sometimes represent a less personal set of interactions, but they undoubtedly play a role in how people decide to approach decisions about cancer treatment. In a social network analysis study on Twitter and cancer messages, retweeting about breast cancer was predicted by in-degree centrality or the number of followers, betweenness centrality, and closeness centrality—especially the latter—but only around 7% of messages were retweeted ( 33 ). These scholars pose the interesting idea that the common conceptualization of opinion leaders as always followed by other consumers of information may need revision in order to capture the kinds of interactions involved in consuming and reproducing and revising information and in order to understand the diversity of roles in online dissemination that do not necessary fit a unidirectional flow or a hierarchical structure.…”