As Obama and Hillary move their campaigns into the final week before the Pennsylvania primary, it's useful to examine some past polling phenomenons. If you look at the tracking poll averages from just about every highly contested state to date, you'll notice strikingly similar trends.
And New Hampshire:
These graphs all show a similar pattern. While the spreads between Obama and Clinton a few weeks or months out were relatively large - 20 to 30 points in some states - as the campaigns headed into election day, the spreads universally shrank.
Now take a look at Pennsylvania's trends:
Again we see the same shape we've been seeing in state after state. So what's happening here?
First, the obvious. As it gets closer to voting day in these states, the two campaigns spend more time and money there and opinions change. But I think the real dynamics lie a bit deeper.
As voting day approaches, people think more deeply about their choices. While someone might have supported a candidate earlier in the race simply because that candidate had higher name recognition or a voter happened to catch one of their speeches on television, when finally forced to make a real choice, voters switch support. Most support in the early stages of the primary in each of these states was "soft" support, based on little information or contact with the campaigns. As people actually begin to make up their minds and move from "soft" support to "hard" support, there is a lot of churn in the poll numbers.
While this idea is fairly obvious, media coverage and general discussion around these primaries has tended to ignore these trends. For example, headlines like "Polls: Obama gains on Clinton in Pa." are extremely misleading. Obama isn't really "gaining" support - if by gaining support you mean convincing Clinton supporters to vote for Obama. Instead, the information people are giving pollsters months before the election is misleading. "Soft" supporters along with independents are likely telling pollsters who they support without much thought. As the race gets down to the wire, people actually sit down and make up their minds.
(As a side note, I would say the same thing is happening with McCain's recentsurge in the polls. With Democratic voters split between Clinton and Obama, McCain has seen a surge that will likely evaporate once the general election race gets started in earnest.)
This is not to say, of course, that campaigns don't matter. If you look closely at the above graphs, you'll notice Hillary Clinton almost always experiences an uptick in polling towards the end of the race in each state, even if her general trend was downward. Clearly, events that happen on the campaign trail are having an effect on voters.
Of course, both campaigns are rightly trying to play the expectations game with these numbers, but we keep falling for it. This trend is happening in every single state every single time. As voting day approaches, the polling narrows. Why are we still surprised at this phenomenon? And why do most media outlets think this is still news?
The views expressed by J Ro are his and his alone.