The democratic party is polling tomorrow in Pennsylvania. The conventional wisdom says that there is a number of percentage points reflecting Clinton’s expected win above which this would truly count as a win for her, and below which it could be perceived as a victory for Obama. If Obama “wins” in this election, things could change quickly with Clinton being seen as heading for the door. However, a win is a win, and even a small margin for Clinton may be seen by the Clinton campaign as, well, what it would be, to be fair: Winning Pennsylvania. Were Obama to actually, numerically, win the primary, that would be a huge victory for him.
Well, I’ve analyzed the data and I have a prediction or two to make.
Most of my polling data comes from RealClearPolitics.com, which in turn assembles data from other sources.
Certain polls they report I’m ignoring because they are outliers and otherwise look funny. Two polls look funny but I won’t ingore: There is a Democratic party biased poll and a Republican party biased poll. The Dem-symp poll predicts a Clinton victory of negative 3 (i.e., Obama wins) and the Rep-symp poll predicts a Clinton victory of 7. So, these polls predict, on average, a Clinton victory of 2%
A 2% victory by Clinton would be a blow to her campaign (but still a win, technically).
The other polls that I accept as valid average out to a Clinton win of 5.5%.
So, we have two hypotheses, two numbers well within each other’s margin of error, and thus not scientifically testable. But the polling is tomorrow, so we have no time for science! We must go with some dead reckoning on this one.
Now, here’s the deal. There are about 8,250,000 people expected to vote in this primary tomorrow. There are undecided between 7 and 8 percent, and a few more percent of people who are “maybe going to switch.” So the slop factor is actually fairly large.
There are also approximately 350,000 young people with cell phones (which confounds polsters) who are likely to break towards Obama.
So, I”m going to present two models, one moderately favoring a shift towards Obama, and the other strongly favoring a shift towards Obama.
In the first model, the undecided voters split evenly, pro-rated, across the candidates, and the cell phone kids split 60-40 in favor of Obama. In the second model, the undecides split at 60-40, and the cell phone kids flock to Obama. What the heck, let’s make it 100%.
In the Moderate-Mo model, Clinton gets 52.7 percent to Obama’s 47.3 percent, and thus a split of slightly under 5.5%, no big change.
In the Major-Mo model, Clinton gets 51.9 precent to Obama’s 48.1, and thus a spread of 3.75%.
So, I predict that the outcome will be between these two numbers.
A bigger question is this: Does this range include the magic “Clinton Looses/Wins” number, or is that number above or below that range? (Since we are apparently not talking about Zero any more as the key threshold.)
My answer to this is: No.
There is not a magic number above which a Clinton victory is a real win, and below which it is not. There are two. Above X, she kicked ass, below Y, she bombed, such that X is greater than Y by art least a few percentage points. Indeed, I think X is around 6 and Y is around 3, bracketing the outcome I predict.
In other words, it’s ain’t over, and I don’t see no fat lady a commin’, neither. (That was for all you hicks that James Carville says live in Pennsylvania.)