If you are reading this article, you are, like me, most likely a poll junkie. After all, we all visit Pollster.com to get our horse race polling fix. Each number we can get our hands gives us that little bit of extra energy to get us through the day. But as my mama once said, even too much of a good thing can be bad for you.
In the past few days, more than a fair share of 2012 presidential matchups have come up on my computer screen. Without getting into too many specifics, one poll by Public Policy Polling (PPP) showed President Barack Obama leading by only 2% over Fmr. Gov. Sarah Palin (R-AK). Another one by Rasmussen Reports, that Matt Drudge linked to, gave Barack Obama only a 1% advantage over Rep. Ron Paul (R-TX). Even Nate Silver felt obligated to publish a response to the Paul poll that corrected for house effects for 2012 polls conducted up to this point.
Let us all take a deep breath. A dip into the polling archive demonstrates why these early horse race numbers are about as useful as Montana Militia money. A December 2002 (even closer to the election than we are now) Fox News/Opinion Dynamics poll of registered voters had then President George Bush defeating Senator John Kerry (D-MA) by 28% in a 2004 Presidential matchup. Bush won the national vote by 2.5%, a difference of 25.5%. What about 2008? A November 2006 Rasmussen Reports poll of likely voters had Senator John McCain (R-AZ) defeating then Senator Barack Obama (D-IL) by 8%. Obama went on to win the national vote by 7.3%, a difference of 15.3%.
If these two examples do not convince you, I leave it to a great map put together by SurveyUSA, the most accurate pollster of 2008. The map shows the results of 50 statewide and the District of Columbia polls conducted in October 2006. Obama won only 28 electoral votes on the strength of D.C., Hawaii, and Illinois.
What actually happened?
Obama won 365 electoral votes by carrying 28 states, the second Congressional District of Nebraska, and the District of Columbia on election night. Need I say anything more?