Previous research has found that messages with moral-emotional words are more likely to go viral on social media — a phenomenon dubbed the moral contagion effect. Now, a new study on the diffusion of political messages on Twitter has found that conservatives — including Donald Trump — enjoyed a greater moral contagion effect than liberals in 2016.
The new research appears in the Journal of Experimental Psychology: General.
“I became interested in what factors make moralized content more likely to spread online based on the observation that moral and political discourse are becoming so prevalent on social media,” said study author William J. Brady of Yale University.
“Recent polls suggest that around 90% of a typical user’s social media feeds contain at least some political content. Besides ordinary citizens discussing moral and political issues, we’ve seen social media used extensively as a tool to gain political leverage whether it be content produced by politicians or disinformation campaigns seeking to sway elections.”
The researchers analyzed 9,505 tweets sent by presidential candidates Donald Trump and Hillary Clinton, 99,750 tweets sent by members of the Senate, and 177,000 tweets sent by members of the House of Representatives in the year leading up to the 2016 U.S. election. They relied on previously established language dictionaries to identify emotional and moral terms.
Brady and his colleagues found that tweets that included a combination of moral and emotional words tended to have more retweets. Emotional language by itself, however, was not associated with increased retweets, suggesting that the inclusion of moral language was key.
Conservative politicians also appeared to have an advantage, as their moral-emotional content tended to result in a greater diffusion on the social media site compared to liberals’ messages.
“One of the key takeaways is that politicians’ use of moral-emotional language (e.g. ‘hate’, ‘shame’, ‘destroy’, ‘faith’) in their messages was associated with increased exposure in social media networks even when accounting for other factors such as the use of media, URLs, and the number of followers each politician had,” Brady told PsyPost.
“This is an effect we’ve seen in our previous work examining ordinary citizens’ language use so it appears to be very robust. Simply put, crafting political messages with the use of moral-emotional language may help political messages go viral.”
“Interestingly, we also found that conservative politicians exhibited this effect to a relatively greater extent than liberal politicians. Regarding Clinton and Trump in the 2016 election, it’s interesting to note that Trump’s use of moral-emotional language was associated with large increases in diffusion of these messages, but this was not the case for Clinton even though she used the same amount of moral and emotional language,” Brady explained.
“There are various reasons why this could have been the case, but from our analyses of U.S. Congress members it appears that ideology, rather than gender, played a bigger role.”
The conservative advantage appeared to be driven in part by positive moral-emotional language related to patriotism and religion. But it is still unclear why the asymmetry exists.
“It’s hard to tell from our work whether the asymmetry we found in the effect of moral-emotional language on diffusion is explained by conservatives users being more drawn to moral-emotional content than liberals, or if conservative politicians are just relatively better at crafting such messages,” Brady said.
“Our previous work examining ordinary citizens suggests that it is likely an interaction of both. Our results are also from a very specific time period — the 2016 U.S. presidential election — and it remains an open question whether the ideological asymmetry we found will replicate in future elections.”
“For instance, it could be the case that whichever political party is under more threat (i.e., the party that does not have presidential power) can benefit more from the use of certain types of moral-emotional language in their social media messages. If this is true, then in the case of the 2020 U.S. election, for example, we might expect to see the asymmetry change since liberals are ostensibly under greater threat,” Brady explained.
“With methods for measuring psychological constructs on social media evolving rapidly, I think it’s important to continue the study of moral and political psychology in the context of online social networks. I’m glad to see many psychologists now using their training to test interesting questions and theories that can shed light on behaviors such as political polarization, online activism, and political propaganda as they are unfolding in the digital age,” Brady added.
“There are many data science tools now available that can help us collect online data, but psychologists and behavioral scientists can help provide theory-driven constraints on data collection and analyses that can push this area of research forward in a focused manner.”
The study, “An ideological asymmetry in the diffusion of moralized content on social media among political leaders“, was authored by William J. Brady, Julian A. Wills, Dominic Burkart, John T. Jost, and Jay J. Van Bavel.