Examining why information ‘goes viral’ on social media

Background

Research has shown that people are sensitive to threats (Nairne et al., 2007; Stubbersfield et al., 2015) and other negative stimuli (Bebbington et al., 2017), resulting in a tendency to transmit these types of information during interpersonal communication. Despite some compelling evidence, extant findings are not easily accounted for by the major theories of social transmission, with null and contrary results in the literature (for examples, see Leeuwen et al., 2018; Stubbersfield et al., 2015). This study explores whether there are additional social and motivational factors that provide explanatory dimensions to related sharing decisions. Namely the social environment in which the information is received (during the COVID-19 pandemic) and the motivational framing of the information – whether messages are framed in terms of promotion (gains) or prevention (the avoidance of loss; see regulatory focus theory, Higgins, 1997).    More specifically, this study (the first of two parts) was designed to test the influence of the COVID-19 pandemic on the transmission (online sharing) of threat-related survival information. We developed a protocol to investigate online sharing behaviour. In this protocol, respondents were presented with a simulated Twitter feed and were instructed to choose which messages they would retweet. This procedure will be repeated by two samples. The first sample (Time 1) was collected during the first peak of the coronavirus pandemic (between April 27 and May 31, 2020) and the second sample (Time 2) will be collected after national case numbers drop (timing is to be decided). This method permits a quantitative analysis of sharing as a function of both threat environment and message content.

Research Questions / Hypotheses

We made the following central predictions:    Hypothesis 1. In line with both the negative information bias and the survival information bias, tweets containing survival-related information would be transmitted more than tweets containing no Survival information at Time 1 and Time 2.    As threat-survival might be a prime motive underlying social transmission, we hypothesised that information aimed at threat avoidance (prevention focus) would have a transmission advantage, particularly within an environment of threat. Avoidant information might be advantageous when the sender/receiver is under a highly salient threat. Arguably, reactive threat avoidance has an immediate payoff in terms of survival. This benefit might be exacerbated in the face of a novel threat where there is less common ground regarding methods of mitigation. In contrast, a promotion-focus on hopes and accomplishments requires proactive effort. Thus, to the extent that a threat is novel or imminent, proactive (vs. reactive) action is arguably riskier in terms of outcome. We made the following predictions:    Hypothesis 2. The transmission advantage of Covid-related information will be more marked when the information is survival-related and is framed using prevention language, this will hold at Time 1 but not at Time 2.    Hypothesis 3. The transmission advantage of general threat-related information (no Covid) will be more marked when the information is framed using promotion-focused language.

Participants

Two hundred and nineteen students enrolled in MBB1 were recruited from the REP. Thirty nine participants were excluded for one of the following reasons: failing to complete the study, failing at least two attention checks, failing the speed check, or recording unusual or suspicious answers. The final sample included 180 students (76% female, Mean age = 19.76).

Methods

We created a set of 16 tweets. Tweet (message) content was balanced on three dimensions, creating a 2 (COVID-19 related: no vs. yes) × 2 (survival-related: no vs. yes) × 2 (framing: prevention vs. promotion) design.    In addition to the Twitter sharing task, questionnaires collected data on the tweet’s familiarity, its relevance to the reader, and its relevance to social interactions. Participants were also asked about their motivation for sharing the tweet, their target audience, and their emotional response while reading the tweet. We also collected demographics, including political orientation, concern about the coronavirus, and personality (Big Five Aspect Scale: DeYoung, Quilty, and Peterson, 2007).    After providing informed consent, participants completed demographic questions followed by the sharing task. Participants then completed the tweet questions followed by our measure of COVID-19 concern and the BFAS.

Results

To examine the influence of content type, the number of tweets shared by each participant will be subjected to a 2 (Survival: no vs. yes) × 2 (Covid: no vs. yes) × 2 (Framing: prevention vs. promotion) repeated measures Poisson regression using Generalised Estimating Equations (GEEs).    Multiple Poisson regression will be used to examine the effects of personality, social media use, and COVID-19 concern on sharing. Specifically, share counts for each message type will be regressed on COVID-19 concern and BFAS, in order to identify predictors of sharing behaviour.

Implications

Examining the influence of a global pandemic on the spread of information will give researchers insight into the mechanisms of online information sharing during a crisis. Ultimately, uncovering the mechanisms of social transmission will help authorities, researchers, lay people, and the media to protect against misinformation, to better disseminate important and potentially life-saving information, and to design more effective public information campaigns.