Easy as 1, 2, 3 -- Except for The Maybes

By Libby Copeland
Washington Post Staff Writer
Friday, April 25, 2008

John King on CNN the other night, confusing us all:

"Let's game this out," he says, playing with delegate and superdelegate numbers on the network's Magic Wall, trying to show through the power of bar graphs how much easier it would be for Barack Obama to win the Democratic nomination than it would be for Hillary Clinton.

"Let's just say there are 585, let's give her around 300," King says, and then numbers like 307, 278 and 305 are popping up on the screen, and the bar graphs are growing and shrinking -- which proves the point, King says, circling another number . . .

At which point we are lost, utterly lost, like when Dr. Gomez was explaining calculus and we nodded but we never really got it. This time, we are unashamed. This time, on behalf of all the students, we raise our hand and say, Huh?

"It's pretty simple," says Alan Abramowitz, a political scientist at Emory, author of an article called "Superdelegate Math" filled with charts and predictions and superdelegate voting preferences. "There's no matrix algebra, there's no differential equations."

That does not make us feel better.

The calculations involved in forecasting presidential politics are not themselves so difficult. What's complicated is keeping track of all the assumptions that go into the predictions.

Let's consider the rest of the primary season. Let's look at the polling data, and let's assume Obama nabs North Carolina by 20 points, and Clinton nabs Puerto Rico by 20 points. (At which point you say, "But wait . . . " and we say, "Shhh.") And let's figure that Obama gets Oregon and Clinton gets Kentucky and Guam -- eh, forget Guam -- and blah-blah-blah the rest of the states, and now let's crunch the numbers and voilĂ !

We're in exactly the same spot. Obama is ahead in pledged delegates but he doesn't have the magical number, 2,025. That's because of proportional representation and little states with big margins and soon enough all the students are raising their hands, saying, Huh?

How can this be?! We call our high school math teacher.

Okay, actually, Dr. Gomez has retired and moved to Florida. So instead we call Henry Kepner, the president of the National Council of Teachers of Mathematics. He gets a great kick out of all the math involved in presidential politics. In fact, he says, he thinks it would be cool to have middle school kids game out the Pennsylvania primary so they can learn the principles of proportional representation.

"It's kind of like a baseball player at the end of the season," he says, explaining the Democratic race in terms of batting average.

Which totally makes sense, in a box-score kind of way. So let's go back to predicting the future. Based on his predictions, Abramowitz says, Clinton would have to win two-thirds of the remaining uncommitted superdelegates to take the nomination.

Given everything we know about the superdelegates -- up to and almost including their preference in ice cream flavors -- the likelihood of this happening would be very, very small.

How small?

"I think it's zero," Abramowitz says. "Unless --" Aha! "Unless she demonstrates in the remaining primaries that she's been able to overcome or change the dynamics of the race." Which is possible if something "catastrophic" happens to Obama's campaign, he says, something of "Spitzer" proportions.

Most smart observers agree that the math is most definitely not in Clinton's favor. But everyone watches the race because it's still a race, because politics doesn't always obey the predictions.

Back in 2000, a whole lot of political scientists used models to predict who would win the presidential election, and a whole lot of them picked a guy who went on to win the Nobel Prize in Not-My-Presidency.

That was the year that "everything went up in flames," says the University of Iowa's Michael S. Lewis-Beck, who had Al Gore winning with 56.9 percent of the vote.

The lesson is not to trust the numbers too much.

If math were a guy, math would be a pompous guy, the sort who's absolutely always sure about everything and never apologizes when he's wrong. And the fact is, math isn't actually ever wrong, not technically. Math is a perfectly logical and intelligent guy. He just sometimes makes the wrong assumptions.

"The fact that people use numbers doesn't make it true," says Jim Campbell, a political scientist at SUNY Buffalo.

Campbell has a bet going with Abramowitz, who says whomever the Democratic nominee turns out to be, that person will whup John McCain's butt. Abramowitz and Campbell have dueling statistical models to predict the outcome of the general election. Interestingly enough, their predictions follow their party affiliations.

The winner has to buy beer for the loser.


© 2008 The Washington Post Company