It has been an exciting debut year for the Engaging News Project. The aim of the project is to research democratically-beneficial and commercially-viable tools and strategies for engaging online news audiences. Below, we detail four take-aways from our work. “Like” is a common button on news websites. You can “like” news organizations, articles, and others’ thoughts in comment sections. The use of the word “like,” however, may exacerbate partisan reactions to news and comments. The word asks people to think in terms of agreeing or disagreeing, approving or disapproving. But not all words inspire the same reaction. Indeed, several organizations have incorporated other buttons onto their sites. The Tampa Bay Times has “Important,” Civic Commons “Informative,” and Huffington Post “Bored,” to name but a few. We tested a new word: “Respect.” Perhaps asking people to “Respect” others’ comments will lead to different behaviors in a comment section compared to “Like,” or another frequently used word, “Recommend.” The results were encouraging. From a democratic angle, “Respect” led people to click on more comments expressing different political viewpoints. From a business angle, “Respect” resulted in more clicks overall, particularly for some topics. We encourage news organizations to consider using this new button. . Online polls are regular features on news sites. They solicit site visitors’ opinions on everything from proposed legislation to how a sports team will fare during an upcoming season. Truth be told, however, the poll results are of limited value. They offer no insight on the sentiments of a broader public. They should not be used to form an opinion or to inform policy. The main purpose of these features is to keep people on a news page longer. We hoped that we could find a way to increase the democratic value of these interactive tools. Instead of polls that ask people to report their opinions, we tested quizzes that ask people about a fact and then provide reliable information. For example, a question could be posed about what percentage of the federal budget is spent on Social Security. Quiz questions also could ask people to predict public opinion about a topic, giving results based on reliably-gathered public opinion data. We tested two types of polls: a multiple-choice poll and a slider poll. Consider the Democracy Fund’s work with the “Oregon Citizens Initiative Review (CIR).” As Peter Levine explained in his earlier blog post, the project involves sending Oregonians pamphlets created by a representative group of 24 citizens that explain ballot measures. Here are examples of the two types of polls:
We first conducted a laboratory test on polls like these. We learned that people spend more time with slider polls and that these interactive tools improve learning compared to just telling people a fact (e.g. “nearly half of Oregon voters were aware of the CIR’s explanations.”). Next, we partnered with a news organization that let us include polls on their site. The code randomized whether people saw a multiple-choice or slider poll. Results showed that including more than one poll, and at least one slider poll, can increase the amount of time people spend on a webpage. In visiting with news staff across the country as part of this project, we were struck by the tremendous variability in comment section practices. Some news organizations actively cultivate a vibrant online community. Others essentially ignore the section, writing it off as a wasteland that is included on a site because having people argue, the theory goes, could increase their time on site. Calls by political leaders from President Obama to Texas Governor Perry to improve the civility of public discourse inspired us to examine strategies for combatting incivility in comment sections. We worked with a local television station to randomize what took place following 70 different political posts to their active Facebook page. For some posts, the station’s popular political reporter interacted with commenters. For other posts, the station interacted with commenters. And for yet other posts, no one interacted with commenters. Results showed that having a political reporter interact with commenters can improve civility in the comment section. Our final project analyzed how hyperlinks are labeled on news sites. Right now, they are commonly labeled as “Top Stories” or as “Most Popular Stories.” We wondered whether other phrases, developed based on research in psychology, communication, and political science, could affect people’s on-site behavior. For example, we analyzed whether including the phrase “Follow the issues that worry you.” would influence people’s on-site behavior and attitudes about politics. We conducted a lab experiment where we compared how people behaved on a news site that was identical in all ways except one: a single phrase included on the site. Results were decidedly mixed. All of the six different phrases that we evaluated had effects, but none of them had uniformly positive business and democratic outcomes. For example, the phrase “Follow the issues that worry you” resulted in some respondents displaying less polarized political views, but it had no discernible business effect. As a result of our study, we can’t recommend any of the phrases that we tested. But we can report with confidence that news organization should tread cautiously in adding new phrases to their sites. A single phrase can affect people’s attitudes and behaviors. For the Engaging News Project these results are the beginning of the project’s work to help newsrooms. Many new tools and strategies remain to be discovered and evaluated and we look forward to continuing our work in the coming years. If you’re interested in getting involved, you can follow us on Twitter, like us on Facebook, Sign up for our newsletter, and Email us with any suggestions you might have!