Lies, Damned Lies and Statistics
Will a dead Osama bin Laden equal a reelected Obama? Barack's poll numbers areway up
If President Obama is re-elected next November, he's going to have to thank Navy SEAL Team 6 in his victory speech.
Ever since they carried out the successful raid that killed Osama bin Laden on May 1, Obama's poll numbers have been moving steadily upwards, hitting a high this week with a 60 percent approval rating based on an AP/Gfk survey.
The AP survey gives Obama his highest approval rating since May 2009, but it's hard to say whether this is a trend towards a larger bin Laden bump or an outlier. When Real Clear Politics gathered the 10 polls conducted since the death of bin Laden, the approvals range from AP's 60 down to 48, according to right-leaning Rasmussen, but the average shows a majority of Americans — 52 percent — approve of Obama's performance.
When looking at polls from the same firms taken before and after the raid, the numbers show about a 5 point bump, which are on the low end of typical political boosts after similar events, according to FiveThirtyEight's Nate Silver. Silver also predicted on May 2 that Obama's numbers would rise into the low sixties.
But the size of the initial bump is not as important as the long term effect, as The Atlantic shows by charting similar gains by presidents and how long the inflated numbers lasted. Will killing bin Laden give Obama a boost that lasts 40 weeks, like Kennedy's did after the Cuban Missile Crisis? Or will he be back to the starting point in less than two months, just like George W. Bush after the capture of Saddam Hussein?
Stats wizard Silver is postulating that the long-term increase for Obama could be around one percent. While that doesn't sound like much, Silver lays out how important a point can be:
In the 35 presidential elections since the Civil War, there have been 9 elections (1876, 1880, 1884, 1888, 1916, 1948, 1960, 1976, 2000) in which the losing candidate would have won the electoral vote had he received 1 percentage point more of the vote in each state, and his opponent 1 percentage point less. That’s a pretty frequent occurrence — it’s happened 26 percent of the time.
What this means is that if a presidential candidate were spotted an additional 1 percentage point “bonus” in each state at the expense of his opponent, he would win in the Electoral College an additional 13 percent of the time. (Why 13 percent and not 26 percent? Because among the close elections, the candidate is already on the winning side half the time, in which case the bonus votes would be superfluous.)
Now, I don’t mean to suggest that Mr. Obama’s odds of winning re-election have improved by 13 percent. A formal model would need to be far more sophisticated than this. More importantly, the notion that Mr. Obama’s approval ratings will enjoy a permanent one percentage point improvement is just a hypothetical.
But if the killing of Bin Laden produces any lasting impact on Mr. Obama’s approval rating at all — 2 percentage points, 1 point, even half a point — that translates into a relatively material improvement in his re-election odds.
Do you think Obama's numbers are real? Will bin Laden's death be a major factor in the 2012 election? I'd be a lot more skeptical of the benefits of the bump if there seemed to be more life in Obama's Republican challengers.