How Much Do Re-shares on Social Media Change Beliefs or Opinions?
Article: Andrew M. Guess et al., “Re-shares on social media amplify political news but do not detectably affect beliefs or opinions,” Science 381:6656 (July 27, 2023), pp. 404-408.
https://www.science.org/doi/10.1126/science.add8424.
Re-posts, re-shares, and even re-truths are ubiquitous across social media platforms. This content can sometimes go viral, the shared material may be harmful, and it can sometimes catalyze the spread of misinformation and disinformation. Re-posting activity may change a person’s political views or beliefs during critical periods before elections. However, a large group of social media analysts have examined this phenomenon and found that re-posting articles and items in itself does not often go viral and change opinions as much as expected.
Introduction
On Facebook alone, an estimated 25 percent of posts shown to people on their feed have been “re-shared,” according to a recent study by Andrew M. Guess, Neil Malhotra, Jennifer Pan, and 26 other researchers who were granted access to Facebook user data during the 2020 U.S. elections. The study from Guess et al. is a product of the U.S. 2020 Facebook & Instagram Election Study, which aims to give the public new “insights and perspectives” into digital democracy.
Guess et al. found that even though Facebook posts can go viral at times through re-sharing, “the kinds of cascades typically associated with viral memes and emotionally engaging stories are rare.” What often happens is that re-shared content, especially from friends, may result in more exposure and receive more “reactions,” but does not go viral in a way that significantly changes political beliefs or opinions.
Guess et al. interviewed 7,730 random Facebook users who agreed to answer a survey about their behavior on the platform during a three-month period around the election from Sept. 24, 2020 to Dec. 23, 2020. The researchers then set up a series of experiments whereby the users belonged to a control group that received no changes to their Facebook feed and were subjected to the usual number of re-shares from friends, Groups, or Pages. Then an experimental group was not allowed to see re-shares and were deemed the “No Re-share” group.
Results
The researchers found that “respondents without re-shared content in their feeds (the experimental group) did not express significantly lower levels of affective or issue polarization than those in the control group.”
The experiment also determined that the No Re-share group did not have “statistically distinguishable effects on perceived accuracy of various factual claims, trust in media (either traditional or social), confidence in political institutions, perceptions of political polarization, political efficacy, belief in the legitimacy of the election, or support for political violence.”
These results were not in line with expectations. As described above, the experimental group did not have the predicted amount of issue polarization. The experimental group also did not have the expected levels of accuracy, trust, and confidence in facts, the media, and political institutions. However, Guess et al. did find that respondents with re-shares saw a greater proportion of their feed replete with political content and political news.
Discussion
One of the difficulties for the study was achieving a critical mass of respondents. Over 14 million users were presented with an initial invitation to participate, but ultimately, only 7,730 people became full respondents and shared data about their Facebook use in the survey. Admittedly, the researchers were asking for a heavy lift from participants and the respondents were compensated an undisclosed amount of money for their usage data. But only 0.05 percent of the people who received the initial invite chose to become part of the experiment. Obviously, it is better to have a higher representative sample, but time and resources may have limited the researchers when it came to analyzing the survey despite the high number of analysts who conducted the experiments. Nevertheless, the low number of participants is a weakness of this research.
The researchers also did not find evidence of significant amounts of re-shares that went viral during the study. One interesting finding was that “suppressing re-shares cut the share of content from untrustworthy sources by nearly a third relative to the control group.” This low incidence of untrustworthy sources is significant because it shows that re-shares can create a situation in which Facebook users see questionable content that could include election misinformation and disinformation.
One other problem of the study is that the assumptions and hypotheses seem to paint right-leaning social media users with a broad brush. The authors seemed to insinuate that alleged viral content from conservatives leads to more political polarization and decreases in knowledge—divisions that are damaging to democracy. The authors refrained from acknowledging this type of bias after many of their political polarization hypotheses were not supported. Perhaps right-leaning social media users are similar to left-leaning groups when they re-share or do not re-share content, but this assertion is not clear in the article.
Interestingly, the authors did not find that conservatives spread more misinformation and disinformation or threaten violence. Users who are members of both political ideologies disseminate politically charged content or share feeds from untrustworthy sources and engage in uncivil or harmful posting. Divulging such insights of partisan and ideological behavior would have been beneficial and would allow readers to see the authors acknowledge those findings about political ideologies more clearly.
This raises an important question for researchers who study right-leaning social media users and their dissemination of content. There has been various research on whether conservatives have more reach in their posts. One study found that the right has distinct “amplification” abilities and conservatives have high visibility of content from their social media posts.
Moreover, a study by New York University’s (NYU) Stern Center for Business and Human Rights suggested that conservatives thrive on social media when it comes to spreading their messages. The authors debunked the notion that right-of-center users are victims of anti-conservative bias or are somehow censored on social media platforms such as Facebook and X. The NYU Stern Center determined that conservative sites such as Fox News, Breitbart, and The Daily Caller dominated other news organizations and were the top three most-visited pages on Facebook from Jan. 1, 2020 until Election Day.
This does not mean that re-shared conservative social media content often goes viral as Guess et al. concluded, but it does have ramifications for elections in 2024. Many conservatives have villainized social media platforms for allegations that right-wing posts are minimized or somehow censored, leading to unfounded calls for more social media regulation. For example, some Republicans in certain states want to ban censorship outright on social media platforms, which could lead to increased forms of hateful or harmful speech.
Conclusion
While many of their hypotheses did not turn out as expected, the Guess et al. group executed an important study during a pivotal election that turned out to be fiercely contested by then-President Donald Trump. Investigating whether re-shares lead to political polarization or more partisan news clicks is a worthy endeavor. This is because such investigations grant more understanding into online behavior that leads to viral content and misleading or untrue posts that can encourage some users to change their minds and vote for one candidate over another.
It is worth further examining whether conservatives had advantages or disadvantages on propagating their messages on social media. Some researchers have determined that right-of-center news sites are not censored and instead they are somewhat dominant on platforms such as Facebook. It will be interesting to see the social media landscape in the 2024 election cycle and whether social media will have the same significance for Trump and conservatives that it did in 2016.