The Worthless 'Predictive Power' of Early Primaries

“You and I are aware that for the last 28 years, the winner of the South Carolina primary has been the nominee of our party.”  

So spoke John McCain Saturday night in the speech commemorating his 33% “victory” in the South Carolina Republican primary.  

Likewise, in Nebraska and New Hampshire, the victors and a bevy of analysts have marched out and reminded everyone the winner of the contest in that particular state, in nearly every year since Jesus was born, has gone on to be the eventual nominee 93% of the time — in both parties, no less!

It sounds impressive and certain, like hearing Tim McCarver assure everyone that only 1 of 5 times in the last 73 years has an American league team been down by 6 points in the (top of the) 8th inning on a Saturday second half of a double header in August and come back to lose by less than 3.  But here’s the problem with primary prediction power: it doesn’t really exist in a year like this one.

Let’s look at McCain’s grand claim of inevitability in detail to see just how unimpressive it really is.  In the last 28 years there have been seven presidential elections:
Year    Winner
1980    Reagan
1984    Reagan
1988    Bush I
1992    Bush I

1996    Dole
2000    Bush II
2004    Bush II

As you can see in the table above, in 4 of the 7 referenced elections, a sitting President or Vice President was running for re-election (rows in red).  When this occurs, you don’t exactly have to be Nostradamus or even just Michael Barone to get the call right about a year in advance.  The nomination is the incumbent’s for the asking, and unless he molests school kids on live TV two weeks in a row, no challenger has a chance in heck of succeeding.  So already, the majority of McCain’s seven-election streak can be seen as having no predictive power at all for this year.

Only three times in the last 28 years has the winner of the Republican primary been in any doubt at all, and in all three of those cases someone won South Carolina definitively.  In 1980, Ronald Reagan soundly defeated John Connally by 25 percentage points.  In 1996, Bob Dole was victorious over Pat Buchanan by 16 points.  And in 2000, George W. Bush beat John McCain by 9 points, taking an overall majority of the votes.  (Strangely, John McCain was silent about the predictive power of South Carolina in 2000.)

Contrast this with McCain’s victory Saturday.  McCain won only 33% of the vote and had Mike Huckabee nipping at his heels just 3 points back.  Two thirds of Republican voters preferred a candidate other than McCain.

You and I may be aware, as Sen. McCain said, that for the last 28 years, the winner of the South Carolina primary has been the nominee of the Republican party.  But the simple fact is that for the last 28 years there has never been a contest like this in the Republican Party.  So what we really should be aware of is that for the last 28 years the South Carolina primary has predicted exactly 0/0 races like this one.

Until someone can start to win a clear majority of the vote, the term “winner” is a bit semantic anyway.  2008 remains a year without a frontrunner.

Add to this the fact that many of the other states have moved up their primaries anyway, and that South Carolina, like New Hampshire and Nebraska, is not the isolated early contest it once was, and it becomes increasingly clear that the predictive power of the early contests in on the wane in general.  

Factor in the crowded and fairly even field of this year in particular and the probable predictive power of the early primaries is nothing to brag about.

Go put the names “McCain,” “Romney, and “Huckabee” on a dart board (maybe make a smaller section called “Thompson”), have a few beers and toss a dart at it.  You’ll have about as much power to predict as South Carolina.  You might get the right answer, but it’ll be pure luck.

It’s a heckuva year, and don’t let anyone tell you the decision has been made already — not even a self-declared straight talker.