Two very different polling organizations have taken their lumps since Election Day now that voters have actually spoken and pre-election forecasts can be judged by the results at the polls. One is the venerable Gallup organization whose name, after 75 years in the business, is synonymous with polling. The other is Rasmussen Reports
, founded in 2003, which has made itself known by the sheer volume of national and state surveys it churns out during each election cycle.
Gallup is a traditional pollster which, like the Pew Research Center and most major news organizations, conducts its surveys based on telephone interviews, (including cell phone-only households, which it added in 2008). Rasmussen is what is known as an IVR pollster
(for "interactive voice response") which relies on automated calls that ask those being surveyed to give their responses by selecting a number on their keypad. Rasmussen does not reach cell phone-only households.
Gallup's headache was that its final "generic" congressional ballot poll
predicted a whopping 55 percent to 40 percent advantage for Republicans, a result that surprised many because of the size of the margin compared to similar surveys by other organizations. By comparison, polls by Pew, New York Times/CBS News, Wall Street Journal/NBC News and Washington Post/ABC News had Republicans ahead by margins ranging from four to six points.
True, when it came to winning control of the House, the Republican victory was sweeping, netting a gain of 60 seats, more than the 1994 landslide that returned the GOP to power in that year. But the actual popular-vote margin for the congressional elections was much closer to the predictions of the other major pollsters and ended up being about half of what Gallup forecast.
Writing on the Huffington Post's Pollster.com
, political scientist Alan Abramowitz said
, "Not only did Gallup miss the actual vote margin by a mile, but their projections about the composition of the midterm electorate were also way off the mark. Based on the exit poll results, it appears that the actual electorate was not nearly as male, old, Republican, or conservative as Gallup's final likely voter sample."
We asked Frank Newport, Gallup's editor-in-chief, about the outcome and he conceded, "It does appear that Gallup's 2010 final likely voter estimate of the House vote did in fact overestimate the Republican share of the vote and underestimate the Democratic vote – although the Republicans obviously won big, as the model predicted."
The extraordinary Republican advantage in voting enthusiasm evident throughout the year may have proved a challenge for Gallup's likely voter model and thus its estimate of the 2010 vote was more Republican than it turned out to be, perhaps due to the very the large Republican advantage in the vote among those who had the highest possible score on Gallup's likely voter scale. It is not unusual for Republicans to lead among those with the highest score on the likely voter scale, but the size of the lead in 2010 was larger than is typical. At any rate, we'll be reviewing in the months ahead.
As for Rasmussen, it took a hit from polling analyst and statistical guru Nate Silver who just came out with a rating of pollster performances on his FiveThirtyEight.com blog
, headlined in part: "Rasmussen Polls Were Biased and Inaccurate
Rasmussen ranked last in accuracy among a list of eight pollsters who released at least 10 surveys of Senate and gubernatorial races in the last three weeks of the campaign, according to Silver.
The 105 polls released in Senate and gubernatorial races by Rasmussen Reports and its subsidiary, Pulse Opinion Research (which was used for polls issued by Fox News), missed the final margin between the candidates by 5.8 points, a considerably higher figure than that achieved by most other pollsters. Some 13 of its polls missed by 10 or more points.
Silver also asserted that Rasmussen's polls consistently overestimating the standings of Republican candidates and criticized the pollster for its methodology, saying:
Rasmussen ... generally conducts all of its interviews during a single, 4-hour window; speaks with the first person it reaches on the phone rather than using a random selection process; does not call cellphones; does not call back respondents whom it misses initially; and uses a computer script rather than live interviewers to conduct its surveys. These are cost-saving measures which contribute to very low response rates and may lead to biased samples.
Rasmussen's own post-mortem
-- and it said it is working on a more detailed one -- did not address the kinds of questions, Silver raised. We've asked them for a response and will post it when we get it.
A spokesman for Ramussen said, "We have no comment other than to say that we're quite comfortable letting the facts speak for themselves. "
Follow Poll Watch on Twitter
Visit the Poll Watch Home Page and see all the latest polls in one place