"The starting points for cascade models are Mark Granovetter's simple and abstract Threshold Models of Collective Behaviour and Thomas Schelling's similar ideas in "Macromotives and Microbehaviour". If a population of individuals is presented with a choice between A and B, and if each will choose A only if some number of other people (their "threshold") also choose A, then very small differences in the distribution of thresholds can lead to very different results. In some cases, everyone chooses A, in others everyone chooses B, in others, the population is split. One spark sputters; an identical spark starts a prairie fire."This is why I don't buy the "Republican Pollsters got it wrong" story. I don't believe for a minute they believed their own polls. They were trying to cause the very small change needed to start a large shift in the vote towards Romney.
Remember about six weeks prior to the election? There was a conservative outcry about how the left-wing media was slanting their polls to help swing the election to Obama by giving the impression Obama was ahead. I think the R's were the ones really doing that. They tend to do a lot of the things they accuse the left of doing and I'll bet this was the case here. They thought they could give the impression that Romney was winning and that masses of voters, wanting to be on the winning side, would jump on the R-boat.
The problem is, it didn't work and afterwards, it was obvious something was wrong with their polls. They couldn't admit they were up to something fishy, especially after accusing the left, so they just made up the story that their polls were wrong.
'via Blog this'