Search as an election predictor

image from

Adam Mathes has a series of posts at MLKSHK (1, 2, 3, 4, 5) showing searches from Iowa users for each candidate in the last month as well as the 2008 search results for a similar time period (and search was a good predictor for final results in the Iowa caucus).

Long story short: Adam predicts Ron Paul will win in Iowa but not much after and spoil the election for any other Republican candidate. I don't know if Adam is right or wrong, but it's an interesting approach and I could see it coming true.


  • I sure hope it happens!

  • Turned out to be a pretty terrible predictor.
    Search is more like information gathering than decision-making, I think.
    It would be interesting to do this analysis on groups of undecideds over time, to see what candidates relative position may change up or down as voters learn more about them.

  • Yeah, I was going to tell Adam that I think any Google Search listing for Ron Paul is going to be skewed since that’s his own followers slogan for the past several years (“Google Ron Paul!!!”). I felt that search directive was going to unnecessarily influence this type of analysis.

Comments are closed.