« Is GAAP on its way out? | Main | Doing business in China: Cutting paths on uncertain terrain. »

April 22, 2008

Predicting Pennsylvania--and the rest of the primaries.

Pollster and consultant Mike O'Neil of O'Neil Associates is a friend of ours. We liked the following article by Mike--he prepared it especially for the Pennsylvania primary today--enough to print it entirely and word-for-word:

The Rest of the Story: Predicting the Outcome of All of the Remaining Democratic Primaries.

Michael J. O'Neil, Ph.D

"It is difficult to predict, especially the future."

--Yogi Berra

Last month, I sent out a missive "The Myth of Momentum in the Democratic Primaries" in which I argued that the pattern of Obama and Clinton victories could be explained primarily by the demographics of the various states, rather than by any of the "momentum" that has been widely discussed in the press.

The beauty of any such argument is that of ex post facto logic: if you explain after the fact, you will always be "right" in your "prediction" since you can alter the theory to fit the known facts precisely.

If the theory is right, however, it should predict the future. Of course, as the philosopher Yogi noted above, this is harder to do with accuracy.

Those of us who conduct opinion research, however, are often (incorrectly) presumed to be in the prediction business. We are not, but (with appropriate caveats), it can be fun.

So, for fun only, here it is: if the demographics relationships observed throughout the election so far hold (that is the "catch", to cover my posterior if I am "wrong"), it would mean the following:

Clinton wins Pennsylvania tomorrow.

But the following Tuesday, Obama wins North Carolina by an even bigger margin (completely erasing Clinton's delegate gains in PA).

Indiana is closer, but probably Clinton territory (like PA and Ohio, but a chunk is in the Chicago media market, and that gives Obama a boost, but probably not enough to win (this is the closest of the remaining states).

Back to tomorrow: Should Clinton win by less than 5%, the confetti will drop and she will declare a big victory. But every politico (and super-delegate) will know that her campaign is toast. The only question will be how long it will take her to get the message. A 10% win will be real, but still not enough to get her many delegates. (5% to 10% will be ambiguous; well within expectations-and spin-meisters on both sides will make arguments, all of them fallacious). A 20% blowout, on the other hand, would be a real loss for Obama.

Poll numbers: these average a 6% Clinton win, down from 20% a couple of weeks ago. Will this momentum continue? I doubt it, for two reasons. First, historical patterns: Obama has typically gained until the weekend before the election when Clinton gets back a few points. Second, most of the "undecideds" look (demographically) like Clinton people. If they vote, she wins-and maybe big. If they stay home, it gets close.

Expectations will be reversed in North Carolina, though Obama currently is winning there by more than Clinton in PA.

And the Rest of the Primaries?

Following demographic and geographical patterns we would predict Clinton wins in West Virginia (5/13), then Obama wins in Oregon while losing Kentucky (both 5/20), Obama ends primary season by winning both Montana and South Dakota (6/03).

In between, we have votes in two non-states, Guam (4 delegates) and Puerto Rico (55 delegates, more than many states!). All of the prognosticators have said that Puerto Rico should be Clinton territory, since she has won Hispanics to date. I don't buy this argument, since I see no reason to assume that native Puerto Ricans necessarily resemble Mexican Americans or Cuban Americans. They are different. But, I am NOT saying that Obama will PR. I just reject the logic behind the assumption that Clinton will win there. And there have been no reported polls in PR. Bottom line, like Sgt. ("I know NOTHING!") Shultz, we really know nothing about PR. The demography of PR does not resemble that of any state.

There you have it. A scorecard against which to measure the results of the rest of the primaries.

What does it mean?

The Obama lead is almost certainly intact on June 3. If so, he continues to trickle in super-delegates (he has been gaining about one a day for the last six weeks). Then the only question will be whether enough super-delegates declare early enough to forestall a convention fight. (Watch to see if a major Clinton insider, someone like PA Governor Ed Rendell, bolts. Nothing less may be required to get her to realize it is over). The speed of super-delegate decisions may depend on whether they enjoy being courted more than they hate being threatened/cajoled. (Is a 3am call from Bill a good thing or a bad thing?)

What could change this story?

A major Obama disaster-something much bigger than the stuff we have seen thus far. Don't hold your breath: it's not likely.
A Clinton loss in any of the states I have indicated she will win. This would signal a collapse of her campaign. Up until now, her campaign has looked viable, even if it is now a long-shot. A loss in a "must win" state would make the campaign look futile, and would likely signal the end.

Michael J. O'Neil PhD

Posted by JD Hull at April 22, 2008 12:35 AM


Post a comment

Thanks for signing in, . Now you can comment. (sign out)

(If you haven't left a comment here before, you may need to be approved by the site owner before your comment will appear. Until then, it won't appear on the entry. Thanks for waiting.)

Remember me?