Pages

Friday, December 18, 2020

Good-enough interventions

From the Workshop on  “News and Information Disorder in the 2020 US Presidential Election.” 

Lisa K. Fazio

We live in a culture where the truth has been devalued where politicians who repeat falsehoods win elections, where lies spread further and faster than the truth, and where misinformation, conspiracy theories, and junk science run rampant on YouTube, Facebook, Twitter, and many other online spaces.

This is a huge problem. And an overwhelming one. It’s often easier to simply document the problem, wring our hands about how difficult it is to solve, and do nothing. In fact, over the past few years, as researchers and the media have done a lot to document the problem, social media companies and politicians have done very little to solve it.

Twitter and Facebook’s actions during the recent U.S. election of more aggressively labeling false claims, promoting quality news sources, and preemptively debunking voter misinformation are great first steps. But many of the interventions are being removed after the election, and there’s a complete lack of transparency. Overall, social media companies are still allowing, and in some cases encouraging, the spread of false information on their platforms.

The good news is that simple, easy-to-implement solutions can help solve this problem. The issue is that none of them, on their own, will fix it. Instead of holding out for the perfect solution, we need to start implementing numerous small changes to our political system, social media platforms, and our own behavior. We need to start implementing “good-enough interventions.”

Rather than viewing misinformation solutions as walls and barricades, we should start viewing them as Swiss cheese. Any individual solution will have holes and weak spots, but by stacking multiple interventions on top of each other we can prevent the spread of misinformation. During the current COVID-19 pandemic, public health experts have encouraged this metaphor of disease prevention. Actions such as hand washing, wearing masks, and keeping physical distance are all flawed and no single precaution is 100% effective; all the slices of cheese have holes. However, the more slices of cheese, the less likely there will be a hole through the entire stack. By stacking multiple interventions we can achieve excellent protection.




 

 

Take the problem of lying politicians. Spouting untruths is an ingrained feature of the trade. But, it turns out that fact-checking organizations can be effective at curtailing it. Fact-checking organizations such as PolitiFact have many roles, including informing the public about what’s true or false. But their most important role is to provide a check against politicians and to encourage them to be truthful.

The threat of fact-checking can change politicians’ behaviors. In one study, a random selection of U.S. state legislators across nine states were sent a series of letters reminding them that their campaign was being fact-checked and of the possible reputational and electoral consequences of their questionable claims being exposed. Legislators who received the letters were less likely to make false claims during their campaigns.

However, this process only works when there are consequences to being proven wrong. Recent studies in the U.S. have shown that those consequences are often missing. When people read fact-checks that contradicted politicians’ statements, the fact-check helped to correct false beliefs, but it didn’t alter readers’ attitudes towards the politician or their voting intentions.

This isn’t a universal quirk, however: When Australian voters are presented with similar evidence, they decrease their belief in the false statement and reduce their support of the politician. There are many differences between the U.S. and Australia political system, but these results suggest deep flaws in the U.S. system. Due to our winner-takes-all system and strong political polarization, it is currently preferable to vote for a liar than to vote for the other party. Electoral changes such as ranked voting may help to increase voters’ choices and improve politicians’ accountability to the truth.

Results such as these make it easy to think that the public does not value accurate information. But that is not the case.

The public does value accuracy and truth in the abstract. In a 2017 poll, only 18% of U.S. adults agreed with the statement “Truth is overrated, lying is the American way.” And people consciously shape their media intake to avoid lies and misinformation. Half of the social media news consumers in a 2019 Pew survey reported that they had stopped following a person because they were posting “made-up news and information.”

The problem is that when we first hear a claim, we often rely on our emotions and how it makes us feel, rather than our prior knowledge and how we know whether the information is true. Thus, the posts that are most likely to spread on social media are those that are emotionally arousing.

In particular, messages that contain emotional words with a moral implication (fight, war, greed, evil, punish, shame, etc.) spread further than messages on similar topics without those words. Across three studies, adding a single moral-emotional word to a tweet increased its expected retweet rate by 20%.

But it doesn’t have to be this way. Companies can help promote information that is useful, informative, and accurate, rather than just information that makes us feel good or bad. Think of how Amazon reviews include a “helpful” button to indicate which reviews are most useful and imagine a similar feature for YouTube videos. Rather than simply marking whether they “liked” or “disliked” videos, viewers could rate the accuracy of the information and those ratings could guide the recommendation algorithm.

Similarly, platforms can encourage users to think about the accuracy of what they are posting. When we rely on our gut feelings, we are likely to judge truth based on unreliable signals such as how many times we’ve heard the claim. However, numerous research studies have shown that when people pause and think about the accuracy of what they’re reading, they are better able to notice false headlines, less likely to share false information, and be less impacted by the effects of repetition.

Simple prompts such as “Are you sure this is true?” that pop up when users try to share a post, may be effective in reducing the spread of false information. Instagram has been implementing a similar intervention aimed at reducing cyberbullying. When users post something that the platform thinks might be harmful or offensive, the user sees a pop-up asking, “Are you sure you want to post this?” A similar friction prompt from Twitter increased the proportion of users who opened articles before retweeting them by 33%.

Valuing accuracy can also help spread truthful information on social media. Many people avoid correcting false information that is posted on social media because they don’t believe that their correction will change the opinion of the original poster. However, those corrections serve a powerful secondary purpose. While they may not be helpful to the original poster, research has demonstrated that they are helpful for other observers who view the interaction. These user corrections are particularly effective when they link to an expert source such as the CDC or the American Medical Association. So, while you might not change the mind of your cranky uncle, by responding with accurate information you will help inform other family members who view the interaction.

Lisa K. Fazio is an assistant professor of psychology and human development at Vanderbilt University.


Cross posted at the Knight Foundation