Researchers Suggest Twitter Data Might Replace Polls

    May 11, 2010

The next time there’s a big election or important issue up for debate, think about thanking Twitter’s founders if your phone doesn’t ring.  Researchers at Carnegie Mellon have determined that, at least in some instances, combing Twitter for data can be as good a way of researching opinions as conducting an actual poll.

Twitter LogoBrendan O’Connor, Ramnath Balasubramanyan, Bryan R. Routledge, and Noah A. Smith examined consumer confidence and political opinions that were measured in 2008 and 2009, putting information from the Consumer Board, Gallup, Reuters, the University of Michigan, and Twitter side by side.

The researchers then concluded in a paper, "[W]e find that a relatively simple sentiment detector based on Twitter data replicates consumer confidence and presidential job approval polls.  While the results do not come without caution, it is encouraging that expensive and time-intensive polling can be supplemented or supplanted with the simple-to-gather text data that is generated from online social networking."

More research on the subject will probably occur as a result, and if this finding holds, Twitter is almost sure to receive a lot of exposure.  Given how could and easy analyzing Twitter data should be, the site may get name-checked on news programs and financial channels left and right.

This development is sort of a win for Twitter even if the researcher’s initial conclusion is disproved, too, considering that not long ago, it would have been impossible to imagine Carnegie Mellon taking an interest in the site.