The day before the election the RCP average has Obama ahead by 0.5%, which means the race is essentially tied. However, the RCP average gives Obama the edge in 8 of its 10 toss-up states. In addition, there were 21 polls of various battleground states released on Saturday, and 16 of them had Obama ahead. Only two put Romney ahead, and the rest were tied. In fact, despite all the criticism from we conservatives of biased polling (and yes there is plenty), if you apply the methodology RCP uses to create its polling average to presidential elections state-by-state since 1988, you’ll find they have correctly predicted the winner 96% of the time.
That is incredible accuracy.
For example, look at the NBC News poll we conservatives routinely mock. In 2008 the final NBC News poll predicted Obama to win by 8 points, and he won by 7.3. In 2004, the final NBC News poll predicted Bush to win by 1 point, and he won by 1.5. This year, the final NBC News poll predicted Obama to win by 1 point as well.
Which name polls have been the most inaccurate the past two cycles? In 2004, Newsweek predicted Bush to win by 6 points, and Fox News predicted Kerry to win by 2 points. In 2008 Gallup and Reuters each predicted Obama to win by 11 points.
But are the polls always right? Well, 96% is pretty darned close, but when they’re wrong they’re really wrong.
The two most glaring examples are the exit polling fiasco in the 2000 election that led to an overhaul of that system, and – get this – 1980. That year all the pre-election polls undervalued Reagan’s actual support, and some did so substantially. For instance, an October 26th CBS/New York Times poll gave Reagan just 39% support, as did an October 29th Washington Post poll. But on November 4th Reagan got 51% of the vote and won by 10 points (third party candidate John Anderson got 7%).
Polling has come a long way since then, and state-by-state polling wasn’t even being done en masse back then. Still, given the documented ideological bias in the mainstream media slanted towards Democrats (Benghazi anyone…anyone…no, really, anyone?), it is worth remembering.
Speaking of state-by-state polls, how accurate has the Real Clear Politics average been with those the past two elections? In 2008 RCP considered 21 states battleground states, and the RCP average correctly predicted the winner in all 21 of them. Granted, that was a blowout for Obama which makes it easier, but what about in the much closer 2004 election? In 2004 RCP considered 18 states battleground states, and correctly predicted the winner in 16 of them. In other words, in the past two presidential elections the RCP average has correctly predicted the winner in 95% of the battleground states.
Misconceptions & Urban Myths
Misconception & Urban Myth #1: Incumbents below 50% always lose
While you’d rather not be an incumbent below 50% in the polls, there just isn’t as much truth to this as people think. Again, comparing this election to 2004, the final RCP average had Bush below 50% in all but 4 of its 18 battleground states. But Bush ended up winning 6 other battleground states where he was polling less than 50%, and that included the decisive battleground states of Florida and Ohio.
Misconception & Urban Myth #2: Independents always break late for the challenger
Actually, late deciders ended up breaking late for Vice President Gore in 2000, which was one of the problems with the exit polling. I still remember Rove on Fox News prior to the polls closing on Election night 2000 predicting Bush to beat Gore by 5 points. Kerry did not get a huge boost on Election Day from undecided voters in 2004 despite Bush being below 50%. In fact, the national and state-by-state polls prior to that election were extremely accurate.
Misconception & Urban Myth #3: The mainstream media polls are biased
They absolutely could be biased this year, but we have already proven the inconvenient truth they’ve been very, very accurate the past two presidential elections. Past history isn’t always indicative of future performance in the ever-changing world of politics, but for conservative claims of rigged polling in 2012 to be true would require a level of media treachery the polling in the past two presidential election cycles shows no evidence of. In fact, in several battleground states in 2008 the RCP polling average underestimated the size of Obama’s victories.
For the past six months I have been saying there is a better chance of Romney winning the popular vote than the Electoral College, and I still believe that to be the case.
Read more from this story HERE.