A Call for a Science of Engagement: Comment on Rus and Cameron

We applaud Rus and Cameron for their pioneering work describing predictors of engagement in online patient communities [1]. Their piece advances our understanding of engagement in online patient communities and underscores the need for a science of engagement that informs the development of social media delivered interventions [2]. Interdisciplinary teams of behavioral scientists, social media marketing experts, and data scientists are uniquely positioned to lead this science. A recent systematic review of social media-delivered behavioral interventions concluded that many suffer from low engagement—meaning that participants interacted less than hoped [3]. This is in contrast with the highly engaged online patient communities described by Rus and Cameron [1] and elsewhere [4, 5]. Rus and Cameron found that specific message features stimulate different forms of engagement. Some messages are effective at eliciting agreement (i.e., “likes”), others are effective at starting conversations (i.e., comments), and yet other messages facilitate dissemination of content (i.e., shares). Three findings from the present study challenge our intuitions about how to engage people online and reveal the need for much more science on online engagement. First, we might assume that including images in our intervention posts will draw attention and stimulate engagement, but Rus and Cameron reveal a more nuanced situation, such that images promote engagement in some circumstances, but inhibit it in others. Second, posts that express negative affect (e.g., sadness, fear, hopelessness, or anger) were actually associated with more engagement whereas posts expressing positive affect (e.g., hope, optimism, and humor) were not. Third, posts that included links to articles and media were associated with lower engagement. Although links are heavily circulated in social media, they may be counterproductive as a tool for engaging patients. Studies randomizing participants to different engagement approaches for the same intervention content are an important next step. The use of online social networks to deliver behavioral interventions is an exciting development in behavioral medicine not only by giving us a unique way to interface with patients but also by allowing us to measure the impact of specific intervention components. Measuring engagement in traditional clinic-based interventions at this level of granularity has not been feasible. For example, tracking how often participants speak, ask questions, or nod their head during a treatment session would be enormously time consuming. In online social network studies, behavioral responses to interventions are easy to extract and analyze, given appropriate technical expertise. Online engagement data can provide important insight into how people react to and digest every aspect of the intervention. If a social media-delivered intervention fails to engage participants (i.e., poor treatment receipt), it will likely fail to change behavior. However, the reason for poor outcomes may have more to do with how the intervention was presented rather than the actual content. A science of engagement is necessary to guide the development of online content that effectively attracts participants’ attention, insures content is digested, and generates opportunities for feedback, learning, and support. The time has come to make a greater investment in the science of online engagement. None. None.