04 October 2012

Debates and Election Outcomes / Approval numbers

People who say that debates generally don't move the needle much are right. They are more right than they may realize.

Nate Silver has done what is probably the first-ever statistical workup over the correlation between debate performance and election outcome.


He pretty much sums it up in one graphic (C = challenger PARTY, I = Incumbent PARTY)

If you look closely, you will see a pattern more often than not:

In 1984, 1988 and 2004, the challenger Party won at least the first debate. The Incumbent party, however, won the election.

In 1992, in a three way debate, the Challenger Party (Clinton), won the biggest debate polling margin of any of them, more likely than not because of this interchange in the town-hall debate:


And 1996 is also an exception - Bill Clinton was already so way ahead the entire race, he was also very way ahead in the debates against Bob Dole. But most would agree that Bill Clinton is hands-down the best debater currently living.

2000 is a weird one: Gore won the outer debates, Bush won Debate no. 2, but the change was more to the challenger Party (Bush 43). However, Gore (D) actually won in the PV, but Bush won in the EC.

So, there is no real good correlation between debate performance and actual outcome. But if you will notice, the SECOND debate played a big role in alot of this: Mondale won the first debate in 1984, but Reagan won the second. Reagan was re-elected. 

Nate Silver's graphic is also not complete. There were TWO debates in 1988, not just one. The debate that Silver scores a statistic for was the first debate from September 25th, 1988. The concensus was that Dukakis won that debate. And then came the Willy Horton thing and the rape question at the second debate, and the concensus is that the debate was won by Bush.

In 2004, according to debate polling, Kerry won all three debates and hammered incessantly at Bush, Jr. I watched all three of these debates. But Bush nailed Kerry on the "international test". This happened in the SECOND debate. Bush won re-election.

2008 did not fit the pattern so well: Obama "won" all three debates against McCain and his debate polling margin continued to increase.


Nate's facit:

"But for what it’s worth, the historical data would project a gain of 2.2 percentage points for Mr. Romney in the head-to-head polls by this time next week."
Nate did not say that the margin would be narrowed by 2.2. He projects a percentage gain for Romney of 2.2, which means the bottomline (Romney's statistic at current) should go up around 2.2.

So, let's watch this starting today. Right now, in national polling:




531Ipsos/Reuters04.10.12536 RV+/-4.8
530YouGov04.10.121000 A+/-4.7
529Gallup *04.10.123050 RV+/-2.0
528Rasmussen04.10.121500 LV+/-3.0
533UPI / CVOTER03.10.121000 RV+/-3.5
530NPR / DemCorps (D) / Resurgent Republic (R)03.10.12800 LV+/-3.5
527United Technologies / National Journal02.10.12
526NBC / WSJ02.10.12832 LV+/-3.4
525PPP (KOS / SEIU)02.10.121100 RV+/-2.8
521Merriman Group01.10.12981 RV+/-3.1
520ARG01.10.12800 LV+/-3.6
519ABC / WaPo01.10.12813 LV+/-4.0
517CNN / ORC01.10.12783 LV+/-3.5
515Zogby (online)01.10.12800 LV+/-3.5
514Politico / GWU / Battleground01.10.121000 RV+/-3.1
505Fox News27.09.121092 LV+/-3.0

Of the four national polls that came in today, 3 are either completely or mostly from people who were questioned before the debate. HOWEVER, the fourth poll, the Ipsos/Reuters poll, released this evening, was conducted mostly after the debate and through the morning of 10/04, and it shows a reduction in margin of 1 point. On the day before, it was Obama 47 / Romney 41. It is now Obama 48 / Romney 43. Please notice that the bottom-line (Romney's statistic) jumped two points.

If we see more polling showing similar movement, then we will know that Silver's prediction was pretty darned good.


There is one more very important factor to consider: the President's job approval rating:

Gallup Daily: Obama Job Approval

As of today, Obama is at 54 approve / 42 disapprove.

This is a three day rolling average, so we will first see complete results all from after the debate on SUNDAY, so this value reflects his approval before the debate.

To note: a 54% approval rating for the President is HIGHER than the ratings he got from Gallup following the killing of OBL in May 2011. In fact, the last time the President was at 54% by Gallup was August 2009, more than 3 years ago. Go look at the graphic yourself.

If the fallout from the debate causes his approval numbers to plunge, then that is a danger sign for him. But if not, then a 54% approval rating is hard for a challenger to beat in a GE. In fact, a challenger has never beaten an incumbent who was at 54% at the time of an election.

So, this number is worth looking at every day from now until the election.

Those are the numbers.

No comments:

Post a Comment

Constructive comments and critique are always welcome. Please keep it polite and respectful.