Traditionally polling is done and political strategy is discussed based on a model that assumes voters are normally distributed (the familiar bell curve) over the political spectrum.
Is this assumption always true and is it true today?
If it isn't true, what are the ramifications for political strategy?
The traditional bell curve says that voter preferences bunch up on either side of the middle, and assumes that undecideds follow a similar distribution (perhaps with some historical adjustments added).
At a point in time like 1860 however, the voter preferences more likely followed a bimodal distribution (think a graph that looks like Pamela Anderson's most notable attributes, and no, I don't mean acting ability).
While there still may be voters in the middle, the larger masses of voters are distributed around peaks at either side of the middle. A Dem strategy that says "you pick up more votes by moving to the right" makes sense for a bell curve, but not for a bimodal distribution. In the latter case, a candidate moving to the center or right is actually moving farther away from "his" voters.
How this would be reflected in poll results, I have no idea. It seems to me when we talk about the left-right polarization (or war/anti-war) we are looking at something that's more bimodal than normally distributed.
Any stats experts have a comment?