03.26.14 8:00 PM ET
You Don’t Need Nate Silver to ‘Predict’ a GOP Win This Fall
When Chicago White Sox star Shoeless Joe Jackson was accused of helping to throw the 1919 World Series, legend has it that a teary-eyed youngster confronted him, pleading, “Say it ain’t so, Joe!”
I would not be at all surprised if, sometime in the next few days, a teary-eyed Democratic campaign operative confronts FiveThirtyEight’s Nate Silver on a Washington street, pleading: “say it ain’t, Nate!”
The seer of the 2012 campaign, whose “I Got Algorithm” approach proved a far more reliable guide to the outcome than the emanations, visions, and psychic insights of his detractors, was a constant source of comfort to Democrats, reassuring them that even the worst blows to the Obama campaign—most notably the self-inflicted wound of the first debate—had not changed the president’s fundamental advantage.
Now, that same Nate Silver has chilled their blue blood; by his measurements—spelled out in detail on his new website, Republicans are a “slight favorite” to take control of the Senate this November. The cautionary words surrounding that projection did little to stop panic; for Democrats, this was as if a parent had told a child at bedtime, “actually, sweetie, there’s a 60-40 chance there IS a monster in the closet.” They’ve chosen to respond by a) noting that Nate Silver has occasionally gotten stuff wrong, and b) sending out fund-raising pitches whose general tone may be summed up thus: “Ow-wee! The dam has burst!”
For me, this dust-up raises two questions:
First, how come so many people mistake probability estimates for predictions?
Second, how come so many otherwise intelligent people who spend their lives in the political arena fall prey to the temptation to predict?
For all of his reliance on national polls, fund-raising numbers, and other metrics (first and last time I will ever use that word in a sentence), Silver has reached pretty much the same conclusion as many of us who were drawn to politics because we were told there would be no math. Longtime incumbents are retiring, or have already done so (Montana, Iowa, Michigan, West Virginia); many Democratic seats are up in states where Obama’s numbers are bad to very bad to terrible (Arkansas, Louisiana, West Virginia, Alaska). The party in the White House historically takes a hit in the sixth year midterms.
Given this terrain, it’s unexceptionable to suggest—with or without numbers—that the Democrats could well lose the Senate.
What this is not…is a prediction.
It’s an estimate that acknowledges that we’re still almost eight months away from election; that party primaries could produce candidates that could turn an eminently winnable race into a loss (Google: 2008 + 2010 + GOP + Delaware, Indiana, Missouri, Colorado); that Obama’s approval ratings could improve between now and the fall, that candidates could do dumb things, like using a Duke basketball celebration in Kentucky, or insult the business of farming in Iowa).
The meaning of a “probability” estimate can be understood by resorting to the last refuge of political metaphors: sports.
If a football team is ahead by three points with five minutes to go, that team will win just about three-quarters of the time. But if the trailing team pulled out a victory with a late touchdown, would anyone be shocked? Would fans and analysts howl that that result was fixed (not counting bettors)? Of course not. The fact that the leading team will win most of the time tells us nothing about what will happen this time. When Silver argued just before the 2012 election that Romney had only a seven percent chance of winning, he was in effect saying, “it’s not likely, but Romney could win.” (His detractors, of course, would have had a field day).
If you want to predict a GOP Senate takeover, you’re saying something much more bold and risky than an estimate. You’re saying, “I don’t care what the primaries produce, I don’t care what gaffes and stumbles happen in the campaign, I don’t care where Obama stands—the Republicans are going to win.” Manifestly, that’s not what most folks—Silver included—are saying.
So why do some people in my line or work assert that they “know” what will happen? Sometimes, it’s a case of the wish being father to the thought. For partisans who saw Obama as a disastrous president, who had been waiting four years for his ouster, any sign that Romney was catching on was likely to be magnified into a near-certainty of his victory. There’s also another reason: all through the campaign season, we’re asked “what’s going to happen?”—as though we have been handed the script, and are keeping it to ourselves the way the writers and producers of The Good Wife kept Will Gardner’s impending death to themselves. When you’re asked enough times, ‘what’s going to happen,” you may wind up thinking, “well, so many people think I know…maybe I do.” You begin to look for signs and portents—a well-received speech, bigger crowds, high-fives among the candidate’s staff.
Sometimes, you can sense movement; anyone who rode John McCain’s Straight Talk Express in 2000 in New Hampshire felt—accurately—that something was afoot.
But sometimes big crowds and enthusiasm are of limited guidance. George McGovern in 1972 and Romney in 2012 drew great crowds. To see them as signs that the country was turning their way was misleading at best.
I learned my lesson early when I co-authored a book that “predicted” a 1972 presidential victory for New York Mayor John Lindsay. Others may have learned their lessons in 2012. Still, get ready for countless, confident assertions of what will happen next November. Unhappily, that’s one prediction on which I’d bet the farm.