The U.S. economy suffers from low growth and high and stable unemployment. The debt crisis is the worst in the history of the Republic. ’Bumps along the road’ signal that U.S. foreign policy is in a shambles. Any incumbent president should be trailing by several points under such indicators of an unacceptable first-term achievement.
Yet recent polls suggest that President Obama will win re-election on November 6, 2012 in a landslide reminiscent of FDR in 1936. Several recent polls indicate that 60 per cent of the electorate believes that Obama will win the race. The Intrade prediction places the odds for a second term in excess of 70 per cent.
Now these poll statistics may be accurate predictors of the election outcome. We shall know for sure if Obama picks up 45 or more states and a huge majority in the Electoral College. With coat-tails, that would imply the House moving under Democratic Party control and the Senate close to a filibuster-proof Democratic majority.
Alternatively, there may be serious problems with the polls, in which case a Romney victory may be enhanced by complacent stay-at-home Democrats, who become over-confident of victory. It would not be the first time that the pollsters got it wrong. The clear-cut, but unanticipated 1980 victory of Ronald Reagan over Jimmy Carter stands as a warning against over-reliance on poll statistics.
A significant weakness in 2012 polling has been over-sampling of pro-Obama constituencies. Many polls include more responses from Democrats than Republicans, based on turnout in the unrepresentative 2008 election. Such bias in the sample significantly and inaccurately inflates Obama’s final score.
Another potential weakness in 2012 polling, especially those conducted by telephone, is the perceived unwillingness of many Americans to expose themselves to criticism by openly disfavoring a non-white candidate. What they say over the telephone may well not be how they vote in the privacy of a booth.
A third potential weakness in 2012 polling may be the result of bias in the framing of questions. The large majority of the media is strongly pro-Obama and it is easy to coax desired responses by the rhetorical phrasing of questions.
In any event, the conservative QStarNEWS has set up a website called ’Unskewed Polls’ which recalculates major poll results on numbers reflecting the current more balanced partisan breakdown. The site shows Republican contender Mitt Romney leading in re-weighted polls by 5 per cent or more. The latest QStarNews Daily tracking Poll has Romney ahead of Obama by 53 per cent to 45 per cent.
Of course, the re-weighting itself may demonstrate Republican bias. Once again, November 6 will tell us everything that we need to know.
In the meantime, be wary of swallowing poll results and writing off Mitt Romney at this stage in the race. Remember that there are lies, damned lies and statistics!
Hat Tip: Editorial, ‘Rigging the polls’, The Washington Times, September 26, 2012
Tags: over-sampling among democrats, pro-Obama polls may be biased, question-framing
September 28, 2012 at 10:54 am |
Charles, there are a lot of people who have been writing about this subject. The problem with the polls is that they are weighted in favour of the Democrats by up to +10 in some states, including Florida. This weighting is based upon 2008 results, and not even on 2010 mid-term elections when there was a tsunami.
From my understanding some of the polls had been using registered voters as opposed to likely voters. Again this had a weighting that favoured the Democrats, and on top of that it was based upon 2008 election results.
The assumption of these polls is that there will be a greater Democrat turnout than Republican, but this is by no means certain. Also, the assumption appears to be that the number of registered Democrat voters has increased, but this is not accurate. Florida is a good example because there has been a rise in registered Republicans as compared to registered Democrats.
Basically it looks like the results of these polls are very skewed. Rasmussen has the polling at even. If Rasmussen polls are using similar weighting to 2008 then there needs to be adjustments based upon the changes in registerd voters in each of the States.
Believe it or not, we have the same problem in Australia where we have a Marxist Prime Minister, and a really terrible government (the woman is extremely corrupt), Most of the polls have the Opposition leading by 54/45 2 party preferred yet there is one poll that came up with an even result. We are all calling BS on that poll because it could not possibly accurately reflect the mood of the electorate as a whole. If there is any movement in those other polls that have been steady for at least 10 months, then I will eat my words.
There is something definitely wrong with the methodology and I think this is the first time that I have seen people openly querying these polls.