In recent years several researchers have reported that the volume of Google Trends and Twitter chat over time can be used to predict several kinds of social and consumer metrics. From the success of new movies before their release to the marketability of consumer goods to the prediction of voting results in the recent 2009 German elections, Google Trends and Twitter message volume have been treated as indispensable tools for not only recording current social trends, but even for predicting the future. This is surprising, given the significant differences in the demographics of voters and people who use social networks and Web tools. But is there some underline logic behind these predictions or are they simply a matter of luck? With this work we wanted to test their predictive power against the US elections. One could argue that, following the previous research literature, and given the high utilization that the Web and the social networks have in the US, Google Trends and Twitter volume may be able to predict the outcomes of the US Congressional elections. In this paper we report that Google Trends was, actually, not a good predictor of both the 2008 and 2010 elections, and we offer some explanation on why this may be the case. On a forthcoming paper we report on our analysis on Twitter.
Catherine Lui, Panagiotis T. Metaxas and Eni Mustafaraj (2011). "On the Predictability of the U.S. Elections through Search Volume Activity," Proceedings of the IADIS International Conference on e-Society, Avila, Spain, March, 2011.