How to keep social media reaction in perspective when covering the elections

Flip the channel or the page, and you’ll find it: coverage of the social media reaction to news events — and political events like conventions and debates.

Much of it, however, talks about that reaction as though it represents the entire population. That, or it offers numbers without context (tweets per minute! Number of times the debate is mentioned!), as Stephen Colbert so ably skewered. For many news organizations, Twitter in particular has become a stand-in for public reaction.

The problem with that? Simple — social media users may be a lot like you and me, but they are not like everyone. Only 85 percent of people in the U.S. use the Internet. (Don’t get me started on the fact that we don’t talk about the 15 percent of Americans who don’t have access to the Internet. Fifteen percent!). Of those who do use the Internet, about half are on Facebook and 16 percent are on Twitter. And a smaller number of those folks ever post about politics.

Who are social media users? Thanks to findings from organizations like the Pew Research Center’s Internet & American Life Project, we’re learning a good bit about them. Lee Rainie, director of the project, said by phone that “they skew in several directions, including towards young adults, towards upscale adults, and towards liberals.” The newest study points out that those most likely to use social media to post about politics tend to be those further from the center — liberal Democrats and conservative Republicans, and among those, it’s not an even split.

A chart from the study Pew released last week makes this clearer. While 69 percent of all Internet users are on some kind of social networking service, 63 percent are conservative, 70 percent are moderate and 79 percent are liberals. Looking at age, 92 percent of 18-29 year olds use some kind of social networking site, while 57 percent of 50-64 years olds do.

How does this matter to our coverage? Tom Rosenstiel, director of the Pew Research Center’s Project for Excellence in Journalism, points out that, “If Twitter were a leading indicator, Ron Paul would be the Republican nominee. Ron Paul got far and away the most favorable attention on Twitter during the caucus/primary season.”

These are “interesting somebodys but not everybody,” Rainie says. They are “a subset of a subset.”

A Project for Excellence in Journalism study comparing social media sentiment to mainstream media sentiment during the conventions found that the social media was consistently more negative:

The conversation on Twitter, blogs and Facebook about Mitt Romney and Barack Obama during this key period changed little with events – even during the two candidates’ own nominating conventions. The conversation in all of these platforms was also consistently negative, according to the study by the Pew Research Center’s Project for Excellence in Journalism.

In the mainstream media, by contrast, both Romney and Obama received a version of the traditional convention bounce, with coverage about them becoming more positive during the week of their party’s nationally televised gathering.

The study authors wonder whether social media are making what we hear about politics more negative, and make it harder for those involved in campaigns to “alter the media narrative.”

One problem in finding the answer to this is that measurement tools for social media are still very new. Rainie points out that “we can’t tell if A caused B, if one thing led to another. We can see more conversations than we used to, but it doesn’t stand in for everyone and everything.”

I’m not saying don’t talk about social media; it’s likely that social media users are influencing the discussion (it seems clear they’re influencing campaign finances). But I think we need to be more conscious about how we frame the discussion. With that in mind, I’d like to make a few suggestions on how to cover social media reaction accurately.

Add context

Understanding who social media users are, and telling your audience about them is useful. Adding deeper context is even better. For instance, we know that liberals are more likely to tweet political statements than conservatives are, so comparing the volume of tweets between the Republican National Convention and the Democratic National Convention probably isn’t useful.

But comparing the volume of tweets in the presidential debate to the vice presidential debate can show the difference in interest between the two — as long as you don’t generalize the Twitter reaction to the entire population.

Some projects, like the CNN-Facebook Election Insights or the Project for Excellence in Journalism’s Campaign 2012 in the Media, examine how often a particular candidate is mentioned. Given that Facebook, by sheer numbers of people, is more likely to be representative of the population as a whole, you might be tempted to generalize. Resist the temptation. Rainie points out that while Facebook users are more like the general population, they still skew young, female, and upscale.

CNN doesn’t make the leap from Facebook to everyone, and your coverage shouldn’t either. However, just talking about the numbers probably isn’t useful unless you can talk about changes to a very narrow group of Facebook users over time.

Understand sentiment analysis

Sentiment analysis — the measurement of whether the tone of a post is positive or negative about a topic — is a fairly new, and tricky, process. Some sentiment analysis is aided by humans, while some is done strictly by computers looking for positive or negative words and their proximity to a keyword. But a statement like “I want to love Obama, but I don’t” illustrates how complex this can be. Clearly, it’s not a positive statement, but a computer seeing the word “love” next to “Obama” might rank it as one.

So be cautious when quoting reports of positive or negative statements in social media. Even when sentiment analysis is correct, researchers aren’t sure what impact, if any, it has on wider public opinion.

Rainie suspects that people posting about politics on Facebook may be influencing others, but our current analysis tools don’t let us know for sure. “It’s discourse,” he said. “Discourse sometimes drives opinion, but sometimes makes it muddier.” Rosenstiel thinks Twitter may have influence as well. He said that while only 10 percent of Americans get political news from Twitter, he bets that the percentage is much higher for political reporters:The paradox is that Twitter is a way to influence the thinking of old media.”

Vary your reaction sources

Website polls, person on the street interviews and social media reaction are not representative of the population at large. Still, it’s important to listen to what different groups of people are saying. So make social media coverage just one piece of the public reaction you cover. Pull in a group of undecided voters for a focus group. Interview people on the street in the suburbs. Do a Web survey. And make it clear that none of these are everybody, but they’re all interesting somebodys.

Highlight specific users

Instead of generalizing, stay narrow. One of the great things about social media is being able to hear from people you normally wouldn’t be able to contact for an interview on short notice. If you can find and use quotes from specific people on Facebook and Twitter, it can enrich your coverage. Just be sure that they’re verified accounts, and not spoofs.

Facebook’s public search lets you find public reaction and follow up. When you come across someone with an interesting perspective, you can reach out to individuals for an interview.

Interact with your audience

There’s no better time than an election season to use social media to engage with your audience and get them engaged in the political process. Ask for their opinion, use Facebook polls, or crowdsource interview questions.

Go ahead. Enjoy, and cover, the voices of social media. Just remember, it’s not everyone’s voice.

We have made it easy to comment on posts, however we require civility and encourage full names to that end (first initial, last name is OK). Please read our guidelines here before commenting.