Even if they aren't fully aware of it consciously.
A new study to be published in Science this week has found it can predict the decision of "undecided" people with 70% accuracy:
The team tested the "automatic mental associations" of 129 residents of Vicenza, Italy, about a controversial expansion of an American military base in their community.
In a computer test, participants were asked to rapidly rate a series of words, interspersed with pictures of the base, as either positive or negative. One week later, when participants made a conscious choice to support or oppose the base expansion, researchers found that those choices invariably reflected the initial automatic associations.
Comments inside.
Update - The sample only included 33 undecided voters (the article I read did not convey that, but the original from Science does. My scepticism goes up a bit (though I think a bigger study would find a similar result). Science is highly respected, but that doesn't preclude them from running with an unconvincing study. Hopefully someone else repeats this research.
129 residents isn't all that big a sample, but then 70% accuracy is much much higher than any kind of fluke one could reasonably expect for even this moderate sample size (feel free to do the math if you're statistically inclined). Another independent study confirming this result would be needed for greater certainty.
I'm fairly sure they're on to something as it makes a lot of intuitive sense to my thinking. What we often call "ideology" is often a kind of fairly deeply seated psychological heuristic that is present in most people. This is probably why so many people who aren't particularly politically aware or highly educated will consistently take certain positions (which we usually label conservative and liberal) on sets of seemingly unrelated issues.
So your preset heuristic for making these kinds of decisions will already bias you to one way even if you don't know the formal ideological debate club rationale for it.
Also of course, there are other factors - if you stand to benefit or lose personally from the larger base, if you had some kind of traumatic experience with a soldier from the base, if a person you respect has taken a side and so on.
As to what this means for elections: My early thought is that the goal with undecided voters is not to convert them to your side so much as find the ones already disposed to support you and motivate them to do so. It might come down to finding a trigger to force them to consciously decide so that they stop playing a self-imposed role of "objective observer" and take your side.
Finally:
There are some technical difficulties to be overcome before the research can be applied to public opinion polling . But Gawronski said it could eventually make it easier for pollsters to predict how the undecided will ultimately vote and, consequently, to predict the outcome of elections with greater accuracy.
He said some researchers in the United States are already using automatic association tests to determine why undecided voters seem to be flocking to Republican presidential nominee John McCain, rather than Democrat Barack Obama. The results could prove controversial.
McCain may have stumbled onto this already (or this kind of thing was already known through private research to the marketing gurus in the Republican camp).
