March 14, 2013

Account for Social Media Distortion

By Dr. Ken Broda-Bahm: 


The way we create, locate, and measure public opinion is changing, and fast. It used to be that we treated the body public as a passive audience waiting to be canvased, surveyed, or focus-grouped. Today, however, people in all demographics are active in public and semipublic internet spaces creating, sharing, and liking their way to an electronic body of evidence that’s unparalleled in history. The existence and continued growth of that body is starting to affect the way we research, including in the litigation world. In what I called the “best trial idea of 2011,” I previously discussed the concept of “social listening.” Originally a technique used by market research firms in order to keep track of online statements about a product, a company, or an issue, social listening stands to become a tool promoted and used by some consultants to monitor and react to large-scale discussions about high profile court battles like the Casey Anthony trial.[1]

Of course, at least some of that promotion may be jumping the gun a bit. But still, when there is a pool of searchable information out there, it is only inevitable that social scientists will be drawn to it. And what social scientists use, lawyers, litigation consultants, and even the jurors themselves are bound to use as well. But even as we cautiously approach this new source of data, we need to be aware of a distorting effect. If social media is “a mirror up to society,” it is more like a funhouse mirror, magnifying some areas and diminishing others. As interactive Web 2.0 features are increasingly used as a guide to public opinion, this post looks at some new research on the extent of social media’s distortion and identifies a few practical places where we should account for that distortion. 

New Research: Social Media Views Society Through a Glass, Darkly

Twitter has a lot of advantages as a data source: It is widespread, decentralized, immediate, searchable, and free. But despite all of that, and some earlier research showing some level of correspondence to public opinion, it isn’t quite the perfect data source. Earlier this month, Pew Research released the conclusions of a one-year study comparing the results of national polls to the content and tone of tweets in response to major news events. Looking at reactions to the presidential election, the debates, and to major speeches by the candidates, Pew found that the tweet content was not a reliable indication of public opinion. And worse, the skew itself was not predictable because sometimes the reaction from the “Twitterverse” was more liberal than measured public opinion, and at other times it was more conservative. On each end of the scale, the differences can be stark. Pew reports that after the first presidential debate (the one we’ve previously discussed as somewhat of a nonverbal disaster for the President), 59 percent of those on Twitter felt the President did a better job, while in sampled public opinion data only 20 percent felt that way. Or in the President’s second inaugural (right, that first debate wasn’t fatal), 13 percent of comments on Twitter were positive, while the survey data showed 48 percent support. 

Pew has also shown that this failure to track public opinion isn’t unique to Twitter. Factoring in blogging and Facebook comments, Pew showed a similar wide gulf between the hot and hip world of social media, and the traditional and stodgy (yet mostly accurate) world of public opinion research.

Of course, it isn’t too much of a surprise that there would be these differences. Social media users are not yet a random cross section of the populations, and there are still rather extreme differences between individuals regarding degree and purpose of participation. It is also possible to say that both public opinion data and social media activity are measuring something, just not the same thing. A spike in tweets or Facebook comments, for example, could easily be seen as a measure not necessarily of attitude content, but of attitude intensity. After all, answering a survey is a passive act, but choosing to compose and send a message is active and behavioral, and each mode has its advantages. 

My position is that litigators and trial consultants should still be moving toward the cautious analysis of social media content, as long as they remain aware of the distortions in a few settings. 

Account for Distortion in High Profile Cases

Even when the social media world has a laser-like focus on a particular case, that audience is not a proxy for your jury, but neither is public opinion data. The Casey Anthony trial is probably the best example of that. With social media virulently against the accused, and public opinion nearly unanimous in condemnation, a jury still found insufficient evidence to convict. That means that neither should be used as a tool of overall case assessment, nor as a more specific measure of whether particular arguments, witnesses, or pieces of evidence are working or not. At the same time, the tone and content of online comments could still serve a heuristic function. As a working tool for coming up with a large number of possible themes and reactions, it can serve the role of a large and unruly focus group. Subject to the same necessary lack of representativeness, it can still be a good sounding board to spark ideas. 

Account for Distortion All Cases Using Social Listening

Though the research and our understanding of what actually happened in the Casey Anthony trial [1] has improved since its writing, my earlier post on the role of social listening in a variety of cases still has some relevance. Using the tools described in that post by New York litigation and social media analytics consultant Christine Martin, it still makes sense to look at the ways a company, party, or issue is trending in social media. At the same time, the Pew Research data give new reasons to be cautious of the representativeness of those views. It is worthwhile to see this activity as a source of ideas, and a behavioral indication of intensity of views, but not as a reliable mirror. 

Account for Distortion in Jurors’ Own Views of Public Opinion

It isn’t just researchers who use social media in order to assess public opinion. The public does it as well. We no longer live in a world where most everyone in town reads the same newspaper or gets their knowledge from the same three television channels. Increasingly, people learn about the world and form opinions not through common sources, but through their network. A group of friends and contacts, like the one depicted in the “FriendCircle” above, is a unique and highly personalized source of information, and being within that circle can create a kind of “filter bubble” influencing what we take to be normal or common. In the current gun control discussion, for example, it is likely that active social media users would see either stricter gun control or unfettered gun rights as incredibly popular, depending on who their friends are. In both cases, the perception of public opinion is distorted by the echoes within one’s own personal network. These perceptions of where attitudes lie on a number of common issues — corporations, the government, personal responsibility, privacy, etc. —  can matter in a variety of cases. So as always, the best advice for knowing how your panelists see the world is to avoid assumptions and ask individually targeted questions. 


[1] Since the publication of the Palm Beach Post article about the role of social listening in the Casey Anthony trial, there has been reason to doubt how central the strategy was to the Defense. During a presentation at the American Society of Trial Consultants’ annual convention in 2012, Anthony’s attorney Jose Baez, told attendees that it was not true that the team was tracking or using social media views in the manner described in the article. 


Other Posts on Social Media:


Image Credit: Miss_Rogue, Flickr Creative Commons (Facebook ‘Friendwheel’)

Related Posts Plugin for WordPress, Blogger...