Background
Explanations for why social media users propagate misinformation include failure of classical reasoning (over-reliance on intuitive heuristics), motivated reasoning (conforming to group opinion), and personality traits (e.g., narcissism). However, there is a lack of consensus on which explanation is most predictive of misinformation spread. Previous work is also limited by not distinguishing between passive (i.e., “liking”) and active (i.e., “retweeting”) propagation behaviors.
Methods
To examine this issue, 858 Twitter users were recruited to engage in a Twitter simulation task in which they were shown real tweets on public health topics (e.g., COVID-19 vaccines) and given the option to “like”, “reply”, “retweet”, “quote”, or select “no engagement”. Survey assessments were then given to measure variables corresponding to explanations for: classical reasoning [cognitive reflective thinking (CRT)], motivated reasoning (religiosity, political conservatism, and trust in medical science), and personality traits (openness to new experiences, conscientiousness, empathy, narcissism).
Results
Cognitive reflective thinking, conscientiousness, openness, and emotional concern empathy were all negatively associated with liking misinformation, but not significantly associated with retweeting it. Trust in medical scientists was negatively associated with retweeting misinformation, while grandiose narcissism and religiosity were positively associated. An exploratory analysis on engagement with misinformation corrections shows that conscientiousness, openness, and CRT were negatively associated with liking corrections while political liberalism, trust in medical scientists, religiosity, and grandiose narcissism were positively associated. Grandiose narcissism was the only factor positively associated with retweeting corrections.
Discussion
Findings support an inhibitory role for classical reasoning in the passive spread of misinformation (e.g., “liking”), and a major role for narcissistic tendencies and motivated reasoning in active propagating behaviors (“retweeting”). Results further suggest differences in passive and active propagation, as multiple factors influence liking behavior while retweeting is primarily influenced by two factors. Implications for ecologically valid study designs are also discussed to account for greater nuance in social media behaviors in experimental research.