One of the thing that makes it difficult to forecast primaries (as opposed to general elections) is that people tend to make last second decisions. This problem is not because primary voters are more indecisive than general election voters, but mostly because they have more choices. In the general, 96% of the voters know well in advance whether they will be voting Republican or Democrat. In a primary, voters have to choose which Republican or which Democrat will best represent them and their party in the general election. That choice involves every voter deciding what is more important — pragmatism or ideology.
While most people when they think of pragmatism think of the concept of which candidate is best able to win the general election (i.e. who is the most conservative or most liberal candidate who is likely to win the general election), there is also (particularly in a large field) the question of which candidate that I like has the best chance of winning the nomination. Voters do not like to cast votes to simply make a statement, they want to have a candidate that thinks like them as their party’s nominee.
This dilemma is particularly true on the Republican side. The folks at 538 like to refer to the Republican Party as a five-ring circus (the moderates, the establishment, the Christian conservatives, the libertarians, and the Tea Party) using those Venn diagrams that you might remember from math class to show where each candidate lies. If, as a voter, you fall in the Christian conservative circle and you like Governor Huckabee (currently polling at around 2%), you have a tough choice to make. Do you stick with Governor Huckabee (knowing that he will probably withdraw by next Wednesday) or do you go with Senator Cruz or Senator Rubio, knowing that they both have a fighting chance to win the nomination and keep it from that RINO from New York. Whether the support collapses for the walking dead could determine who wins. If you are a supporter of Governor Bush or Governor Kasich, do you hold out hope for a strong showing in New Hampshire or do you decide to put your support behind Senator Rubio.
With only three candidates, there is likely to be less last minute decisions on the Democratic side. The potential for a surprising outcome comes from the way that the results are reported (as estimated state convention delegates). Aside from the fact that, in most precincts, candidates will win fractional delegates (and thus the rounding rules will lead to percent of delegates that differs from the percent of attendees), there is also the fact that the results from the second round of alignment (the results that are used to calculate delegates) are likely to differ from the first round of alignment (the real level of support) due to the threshold to get any delegates.
As things stood on Saturday, the numbers look bad for Governor O’Malley, and there is a substantial likelihood that — after the first round of alignment — Governor O’Malley will not have 15% of caucus attendees in many precincts. But that will be when things get interesting. In some precincts, Governor O’Malley will not have enough supporters to make a difference in delegate allocation. In those precincts, you will not see a lot of effort to swing the O’Malley supporters. In some precincts, Governor O’Malley will be just short of the 15%, and it may make sense to one of the two candidates teams in the precinct to loan a supporter or two to the O’Malley team (e.g. to deny the other candidate of a delegate). In other precincts, Governor O’Malley will be far enough short, but still be in a position to swing a delegate or two in the allocation, and it may make sense to promise (although the promise is not binding) that your candidate’s caucus will elect a county convention delegate proposed by the O’Malley group (to get the additional delegate for your candidate when the results are submitted to the state party). How this dance plays out will impact the reported results but has nothing to do with the actual support for the three candidates. (Additionally, how this plays out will depend in part on how well organized the precinct teams are for each of the candidates. Well organized campaigns tend to do a better job at seeing the numbers and making the deals.)
As this is posting, the registration is ending and the last second speeches and voting are about to begin. The guessing and speculating is done, and it is now in the hands of real people.