From debates to Sandy: Things "everyone knows" aren't always true

Daryl Cagle, PoliticalCartoons.com Now that it's all over but the GOP civil war, explaining what happened in this past election is easy, right? Well, maybe not. Here's a great concept to start us off:

Once they know an outcome, people tend to inflate their initial predictions by an average of 15 to 20 percent, Dr. Roese said ' ample wiggle room to retrospectively alter almost any prediction from 'it's going to happen' to 'it probably won't,' be it a tennis match, a legal decision or a presidential race.

In 800 studies during the past few decades, researchers have demonstrated the effect consistently in predicting everything from legal decisions and stock movements to sports contests and world affairs.

Given that that's the case, let's go back and look at some key events using the interesting RAND panel published every day during the election.

RAND American Life Panel, annotated

BY the way, RAND did pretty well, forecasting a 3.3 gap (Obama 49.5-Romney 46.2) when currently we have 2.7 (Obama 50.6-Romney 47.9). In fact, Nate Silver in his latest review, gives RAND good marks along with other internet polls.

Online polls were among the most accurate in this year's elections... http://t.co/...
' @politicalwire via bitly
Still, one can pick aggregates of polls to do the same review of events (for example, Kos used pollster.com for this post).

In any case, check out some of the key indicators, such as the 47% tape release, rise in Romney's numbers even before the first debate, as GOP voters did their inevitable consolidation; the arrested development due to the VP debate (yes, Joe Biden, you rock!); the general equivalency from before debate 1 to after the third debate widely seen as won by Obama; and the lack of data points showing Sandy killed Romney's chances.

Yes, Virginia, debates don't matter that much. We noted that beforehand, so it's not hindsight bias. No one's saying they didn't matter "at all", it's that they didn't matter "that much", and certainly not in the game changing way the pundits claimed. The first debate was worth a few points for Romney, the debates as a whole were close to a wash as Obama gained back his voters with two strong performances and a "kicked his butt" addition from Biden. They were a joy to watch.

A word of praise for political scientists, who said debates don't swing elections, got mocked by jerks like me, but were 100% right.
' @daveweigel via TweetDeck
And don't get me started on idiot reporters pointing at a candidate and yelling "Gaffe!" They wouldn't know one if it bit them in the ass. How relevant was "you didn't build that!" other than revealing the workings of a tactical campaign that had no coherent strategy after the Bain ads smoked Romney all summer?

Now, keep in mind that the 47% tape was far more than a gaffe. It fit into every Bain-ad narrative about Romney and sunk his campaign with cement shoes. See the RAND graph, see the item below on who was liked and who was disliked.

See, also, that Romney's natural trajectory wasn't radically altered by Sandy. Those making the case that "Sandy was why Romney lost" have no firm ground to stand on. Sure, it didn't help him, but there are more important reasons why Romney lost. Let's explore them after the fold, using exit polls to illustrate some points.

2004: Bush 50.7%, Kerry 48.3%. 2012: Obama 50.6%, Romney 47.9%. Turned out those 2004 paralells held up really well.
' @fivethirtyeight via TweetDeck
And before we delve into the exit polls, understand that the fundamentals always favored Obama. And that's not hindsight bias, either.

More below the fold.