I spend a lot of my life looking at large sets of numbers, or more often, graphical representations thereof. The goal is to discern patterns from noise.
Humans are inherently a lot worse at this than you may think. More specifically, we are tuned to err on the side of pattern completion, rather than accepting chaos for what it is. We see patterns where there is only noise, and we over-generalize our experiences as representative of the whole. Therein lies the human root of superstition, religion (but I repeat myself), prejudice, racism, nationalism, and in extreme cases, paranoid schizophrenia.
So what to make of the data above? Can we extract any predictive value from this incredibly complex data set, given that it represents the aggregate whims of hundreds of thousands of people each with their own complex prejudices, priorities, and media consumption habits?
Obviously, the people who conduct these polls think so.
Here is what I see:
1--Four candidates who are essentially running as Tea Party/outsider/Not Romney (Bachmann, Perry, Cain, and Gingrich) have each had a surge of support, lasting about a month (we are in the ascending phase of Gingrich's right now--I predict he will peak in about a week). There is nothing here to predict that any will get a second surge.
2--Paul, Santorum, and Huntsman have held completely steady for the better part of the year. Though individually different, what they have in common is that they all have very strong, very principled, and very limited constituencies. (Paul: paleoconservative/libertarianish, Santorum: family values are EVERYTHING, Huntsman: "just because I'm a Republican doesn't mean I'm a crazy person). I don't see any of them breaking out.
3--The only candidate whose numbers rise and fall nearly perfectly out of phase with each surge among the first group of candidates is Romney. If his pattern holds, his next peak will be just as the primaries get started.
Mitt Romney will be the nominee.
OK, tell me what's wrong with this.