The last time my Avast Antivirus updated itself, it asked an odd thing, “Would you like to take a political survey?” Slightly bemused, I thought, “Sure, why not?”, and proceeded to answer their few questions. They have since published their poll that attempts to predict early voting results.
Czech Republic-based Avast’s goal is not partisan. They desire to use the power of Big Data to attempt to crowdsource an election prediction from their 11 million USA-based Avast users. According to National Defense Magazine, the “ability to crowdsource predictions has been the subject of much research and experimentation in the past few years.” In fact, 100,000 of these Avast users responded to the political survey, which is 100 times the sampling size of Gallup’s own polling technique! This might even be more reliable than some early voting results.
According to the blog post on Avast, here are some interesting results:
- 48.9 percent of users who are registered voters said they will definitely vote for Romney, compared to 46.1 percent who will vote for Obama. Interpreted by electoral votes, this is 290 votes for Romney and 230 for Obama.
- As to who they think will actually win the election, however, 47.6 percent think Obama compared to 39.6 percent in Romney’s favor. Interpreted electorally, this gives 349 to Obama and only 189 to Romney.
- New Mexico and Virginia are dead heats—46.9 percent for each candidate in New Mexico and 47.6 percent in Virginia.
- The other tightly contested states in the survey are Florida with 49.2 percent for Romney and 47.5 percent for Obama; Ohio with 48.8 percent for Romney and 46.3 percent for Obama; and New Hampshire with 48.2 percent for Obama and 45.3 percent for Romney.
Attempts to predict elections with crowdsourcing can be mixed, according to crowdsourcing.org. There were some odd results based upon the Avast political survey, like Obama winning Alaska even though no Democrat has done so in many years. The most interesting insight is that a marked majority of Avast users prefer that Romney win the 2012 election, but, at the same time, a majority seem resigned to Obama actually winning. I don’t think I’ve seen a poll collect data like that before. As Avast puts it, either way they are right. “If Romney wins, the survey preference of our users is proven. If Obama wins, it shows the ability to accurately crowd source the correct result.”