Tweets = public opinion? New data suggests we should think twice on this

Are the number of tweets per television show the new Nielsen ratings? Or are tweets on election day the same as exit polls? Twitter might be moving in that direction, but new data from the Pew Center should have you think twice before trusting Twitter as a barometer of public opinion.

It’s not all that surprising that tweets aren’t a perfect indicator of public opinion, considering that only 13 percent of Americans are currently using the service. But the discrepancy is worth noting as Twitter continues its push for journalists and the public to consider its data a legitimate source of news and an accurate indicator of the national discussion.

Pew explained why evaluating tweets can be useful, but not necessarily definitive:

“Overall, the reaction to political events on Twitter reflects a combination of the unique profile of active Twitter users and the extent to which events engage different communities and draw the comments of active users. While this provides an interesting look into how communities of interest respond to different circumstances, it does not reliably correlate with the overall reaction of adults nationwide.”

The Pew data, released Monday, found that during the 2012 presidential election (when Twitter was quick to note the 31 million tweets on the big night), Twitter user reactions to President Obama and Mitt Romney weren’t exactly representative of American opinion at large. Twitter users were much more critical of Romney in the first debate and more positive about Obama’s electoral victory than the public as a whole. And while both candidates faced high levels of criticism on the site, Romney faced more than Obama did through the fall campaign.

However, Twitter doesn’t just lean left. Reaction to Obama’s inaugural and State of the Union addresses on Twitter were more critical than overal reactions.

As my colleague Derrick Harris has written before, there’s huge value in using tweets as part of a larger set of data to evaluate situations, but there are plenty of statistical reasons why measuring tweets is challenging. While Twitter is already having a significant impact on how we consume news and information and is relatively mainstream at this point, it’s important to remember that people who voice their opinion on the service are still just a tiny percentage, and who might have different reasons for wanting to share.