Wednesday, April 16, 2008

The US Elections: Predicting Pennsylvania

Later today, ABC News will be broadcasting the Clinton/Obama debate in Philadelphia. It'll be interesting to see how well - or indeed, how poorly - Obama does. He's certainly gotten better at these debates as the campaign moved on; on the other hand, there is sure to be some emphasis on the recent "Bittergate" controversy.

Meanwhile, though, predictions are always fun to read, especially if they're specific and clearly well thought-out.

Predictions of the outcome of the Pennsylvania primary have been made in the past months. Here's an example from Real Clear Politics, written by Jay Cost. However, these have tended to be a bit unspecific. The RCP article doesn't, for instance, actually state a likely outcome when it comes to the one thing that matters above all else: the delegate tally.

Two other articles, on the other hand, provide just such a tally, based on a district by district breakdown of the state. The first is from CQ Politics. The second can be found over at Daily Kos.

To understand all the numbers, you have to first understand that Pennsylvania's pool of delegates amounts to a grand total of 187. However, that includes 29 superdelegates, who aren't going to be subject to the vote at all. That leaves 158 delegates that are actually going to be up for grabs during the primary.

Of these, 55 will be allocated on the basis of the statewide popular vote. If, for example, Clinton wins by 55% to 45% (the same margin of victory she had in Ohio), these 55 delegates will be split up as follows: 30 to Clinton, 25 to Obama.

The remaining 103 delegates will be allocated on the basis of the outcome of the vote on a district by district basis. There are 19 congressional districts in total. Each district has a certain number of delegates, the "largest" district (the 2nd) having 9 delegates and the "smallest" (the 9th) having three. 6 districts have 5 delegates to allocate; 5 districts have 4 delegates. (Strange, there almost seems to be a patern of numerical reflections here...)

Given the peculiarities of the district system, the outcome per district will either substantially augment the victor's win or substantially decrease it. The winner of the popular vote in a district with an uneven number of delegates will generally walk away with a relatively high number of pledged delegates; the loser, on the other hand, benefits when the number of delegates is even. If one candidate were to win the popular vote by, say 5%, in a district with 5 delegates, he or she would get 3 delegates, leaving 2 for the opposing candidate. The 5% margin in the popular vote, in other words, translates into a 20% margin in the delegate tally. Conversely, if that district had only 4 delegates on offer, those delegates would be split evenly. In fact, if the popular vote margin went up to, say, 20% the delegates would still be split down the middle (if there were only 4 of them to begin with).

The consequence of this system is that, when it come to the district delegates, it makes very little sense to campaign hard in a district with 4 delegates if your chances of sweeping it aren't very high; on the other hand, getting that slight edge in a 5-delegate district could be well be worth your while. The campaigns, therefore, are waging an on-the-ground war aimed at key districts; the statewide polls of the popular vote are meaningful, but not necessarily as decisive as many people think.

Now back to the two predictions. In both cases, the outcome is not really good news for Clinton. In the CQ projection, the 103 district delegates will be split as follows: 53 will go to Clinton and 50 to Obama. That's a Clinton victory of just 3 delegates. Unfortunately, CQ doesn't say anything about the expected outcome of the popular vote, so it remains unclear how they think the other 55 delegates would be allocated. Assuming, though, that Clinton "wins" Pennsyslvania as a whole by 10%, that would mean her net gain in the state would amount to (5 + 3 =) 8 delegates.

Over at Daily Kos, a statewide Clinton win of 9% is assumed, with Clinton gaining a net advantage in the district war of 5 delegates. That would put her total gain in the state at (5 + 5 =) 10 delegates.

The predictions are, therefore, effectively the same. Interestingly enough, Clinton won Ohio by 9 delegates. So Jay Cost at RCP, who wasn't explicit when it came to delegate numbers but who did say that Pennsylvania might well go the way of Ohio, may be even more correct than he imagined.

Oh, and to put it all into some perspective: Obama currently leads Clinton in the pledged delegate tally some 164 delegates. Let's say that does, indeed, decrease to 154 or thereabouts.

After six weeks, some bowling, a few shots of whiskey and all the rest, that's not really much to talk about, is it? More than anything else, I feel this proves the silliness of the entire "Bittergate" thing. Which is why I'll not say more about it.

No comments: