Many forecasted that Mitt Romney would defeat President Obama in the 2012 presidential election. They were wrong. Some of those who were incorrect took it in stride and admitted as much. Others, such as Twitter sensation and National Review blogger Josh Jordan (aka NumbersMuncher), decided that their logic was sound, and it was only because of Superstorm Sandy that Obama won.
At first glance, their logic seems to makes some sense. Obama got good Sandy press, highlighted by nice photos with and words from New Jersey Governor Chris Christie. Polls indicated that Americans approved of Obama’s Sandy response. The national polls were biased against Obama, and even some of the state polls were too pro-Obama compared to the final results.
The question is whether the polls indicate that Obama gained in the final days. We can figure this out by comparing pre- and post-Sandy pre-election surveys.
Examining the data in that manner finds that the Sandy theory for Obama’s rise is fatally flawed. The majority of polls did not show great movement to the president after Sandy.
The two most accurate national polls were consistent in pegging Obama at a mid single-digit victory. Ten days before Sandy, Democracy Corps had Obama leading by 3 points. In the days after Sandy, Obama gained a statistically insignificant 1pt after Sandy, to be up by 4pt. Rand actually had Obama’s lead falling from 5.1pt just prior to Sandy, to 3.3pt in their final tracking poll.
The full national polling also doesn’t support universal movement to Obama after Sandy. Seven pollsters had Obama gaining no points, or actually losing a few, after Sandy. Four pollsters had Obama’s margin expanding only by a point after Sandy. Put together, with an 11 out of 19 majority, pollsters had Obama gaining a point or less post-Sandy.
This confirms the finding I made in an earlier piece. The national surveys were consistently skewed against Obama in the final month. Sandy did not change that, and it wasn’t because of the superstorm that the vast majority of national surveys failed to accurately project the outcome.
You might argue that the national polls don’t matter. After all, the election was won and lost in the swing states. Referencing polls only pre-Sandy, Simon Jackman of HuffPollster had Obama winning every state he would win, except for Florida – the closest state. Given the polling splits in swing states, Jackman estimated that Obama had a 99% chance of winning. Others, from Drew Linzer to Nate Silver, to Sam Wang, all had Obama as a heavy favorite pre-Sandy.
You can see why their odds changed little after Sandy by examining the 12 swing states. In the table below, I controlled for each pollster and averaged the result from those who had polls completed fully post-first debate and post-Sandy (that is, by 29 October). If a pollster had two post-Sandy polls, I took the mean.
It turns out that in only three of the 12 states did Obama see a gain in the majority of post-Sandy swing-state polls. In the 67 situations where the same pollster had a pre- and post-Sandy survey, only 25 resulted in Obama doing better after Sandy. Only in Virginia did Obama pick up more than 1pt. So, like the national polls, the swing-state polls don’t indicate that Obama was better-off electorally thanks to Sandy.
It’s possible, of course, that there was a Sandy effect in the states most affected by the storm. So, what do the polls say of Connecticut, New Jersey, New York and Pennsylvania – the states with the greatest numbers of power outages caused by Sandy?
Like all the other polling, Sandy’s effect seems to be minimal. Obama was at or below his pre-Sandy levels in seven of the ten polls released in this quad-state region. Only in Connecticut did Obama gain in both polls post-Sandy, by an average of 2.5 points. Only one poll was published post-Sandy in New Jersey and New York respectively, and Romney picked up 1pt in both. The only swing state in this group is Pennsylvania, and Obama wasn’t in a better position in five of the six surveys there. In fact, Romney closed the gap between Obama and him by an average of 1.5pt.
Overall, there is very little evidence that Obama added to his lead because of Sandy. Neither the national, nor especially in the state polls, did Obama poll better post-Sandy. Any polling errors were because of longstanding miscalculations of the make-up of the electorate, not Sandy-specific pollster mistakes.
guardian.co.uk © Guardian News & Media Limited 2010