- December 6, 2011
I’ve long held the suspicion that opinion polls for general elections were fairly useless. Consider that fact that these polls are still conducted with small sample sizes (often about 1000 people) and inconsistent frequency. Why go through the effort of questioning thousands of individuals when you can mine social networks for opinions that are already publicly available?
Two weeks ago, we found some great research from 140Elect that showed peaks in Twitter follow activity for a candidate foreshadowed jumps in the polls, weeks or months later. While not a perfect analog for a number of reasons (demographic bias, new follows from bots, etc), this measure may be called “good enough” for now in making reasonable guesses about a candidates popularity amongst their constituents.
Want to know an even better way to determine if a candidate is bound for victory or an embarrassing defeat? A 2002 study from the US Public Interest Research Groups found that in 2000, 94% of candidates who raised the most money won their general election contests. In 2002 congressional primaries, 90 percent were won by the biggest fundraisers.
One of the New York Time’s current interactive graphics features rolling fundraising totals for different candidates in the current presidential race. I might be going out on a limb here, but based on the data, I’m going to say that the guy with 11.5 million Twitter followers and nearly $100 million fundraised so far (with over $57 million on hand) will probably win come next November.