Rasmussen Polls Were Biased and Inaccurate; Quinnipiac, SurveyUSA Performed Strongly

Every election cycle has its winners and losers: not just the among the candidates, but also the pollsters.

On Tuesday, polls conducted by the firm Rasmussen Reports — which released more than 100 surveys in the final three weeks of the campaign, including some commissioned under a subsidiary on behalf of Fox News — badly missed the margin in many states, and also exhibited a considerable bias toward Republican candidates.

Other polling firms, like SurveyUSA and Quinnipiac University, produced more reliable results in Senate and gubernatorial races. A firm that conducts surveys by Internet, YouGov, also performed relatively well.

What follows is a preliminary analysis of polls released to the public in the final 21 days of the campaign. Our process here is quite simple: we’ve taken all such polls in our database, and assessed how accurate they were, on average, in predicting the margin separating the two leading candidates in each race. For instance, a poll that had the Democrat winning by 2 percentage points in a race where the Republican actually won by 4 would have an error of 6 points.

We’ve also assessed whether a company’s polls consistently missed in either a Democratic or Republican direction — that is, whether they were biased. The hypothetical poll I just described would have had a 6 point Democratic bias, for instance.

The analysis covers all polls issued by firms in the final three weeks of the campaign, even if a company surveyed a particular state multiple times. In our view, this provides for a more comprehensive analysis than focusing solely on a firm’s final poll in each state, since polling has a tendency to converge in the final days of the campaign, perhaps because some firms fear that their results are an outlier and adjust them accordingly.

If it’s a hit piece, it is definitely backed up by the election results. Rasmussen overestimated the Republican performance in just about every race.

Going backward from Ras/POR's latest polls:

WA-Sen 49-47 Murray - almost exact
OH-Gov 48-44 Kasigh - actual 49.4-46.7 - almost exact
NV-Sen 48-45 Angle - every public poll got this wrong
WV-Sen 50-46 Manchin - nearly identical to PPP result, both slightly underestimated Manchin
CO-Sen 50-46 Buck - every public poll in the final week had Buck leading
CO-Gov 47-44-6 Hickenlooper - every public poll overestimated Tancredo vs Maes, and many votes were cast well before the poll was even taken, likely favoring Maes
IL-Sen 46-42 Kirk - actual 48.2-46.3, right on target
IL-Gov 44-38 Brady - same result as every other public poll, all underestimated the final result of 46.1-46.6 Quinn
CT-Gov 48-46 Foley - very close to actual result of 49.0-49.5. No poll showed Malloy winning, the only one showing a tie had 10% undecided/3rd party.
CT-Sen 53-46 Blumenthal - actual 54-43, right on target

How much more do you need to see? In every single case, Rasmussen's polling was either exact, extremely close, or a bit off but in agreement with every other public poll.

If you know of a superior prediction tool, please advise.

It's not "predictive", period. It just reflects conventional wisdom made by others. No value added whatsoever, and major potential downside since it's not a well-functioning market.

Cook Political Report, FR's own Key House Races, FiveThirtyEight (whose race ratings are massively driven by inputs from Cook Political / Rothenberg), RealClearPolitics, even an idiot like Larry Sabato are all better than Intrade.

All those sources use a synthesis of information, without being subject to market manipulation.

1. I’m comparing Rasmussen to actual results - not to PPP
2. You selectively omitted many races which don’t support your case
3. Many of your numbers are just plain wrong. see this chart:
http://www.rasmussenreports.com/public_content/politics/elections/election_2010/election_2010_senate_elections/final_rasmussen_poll_results_2010_senate_elections

I stand by my comment.

1. I’m comparing Rasmussen to actual results - not to PPP

I indicated each race where Rasmussen -- and all other public polls -- got the result wrong, which suggests if Rasmussen was to blame, they were no more than any others.

If you are arguing Rasmussen is biased, then to be consistent you should at least admit that all the other public polls were biased for those races. Or, more likely, there were non-polling reasons for the polls to diverge vs the actuals.

2. You selectively omitted many races which don’t support your case

I included every single poll released the last two days before the election by Rasmussen Reports using Public Opinion Research methodology or Fox News using Public Opinion Research methodology.

3. Many of your numbers are just plain wrong. see this chart:

As noted, I used all polls with POR methodology, whether conducted for Rasmussen or Fox News.

If you don't understand why, do some reading and be minimally informed before trying to reply again.

If you want to include other polls in your analysis, you need to do it in every race - not in just the races where it supports your hypothesis.

Again, your numbers don’t match Rasmussen’s own Senate analysis page, comparing their last poll to the actual result.

On top of this, Rasmussen estimated a 12-point margin on the generic ballot, and it was only 6.

They are biased by a few point to Republicans, everybody sees it now.

If you want to include other polls in your analysis, you need to do it in every race - not in just the races where it supports your hypothesis.

Completely ignoring my response. Pick one of those races and show me where Rasmussen (POR) methodology was more Republican-friendly than other polls, and thus got the race wrong. It's only the 10 races polled most closely to Election Day, surely it's not too taxing to find ONE of the ten to support your hypothesis?

You have no argument and can't counter mine.

Again, your numbers don’t match Rasmussen’s own Senate analysis page, comparing their last poll to the actual result.

You have no idea what POR (Pulse Opinion Research) is, do you?

I'll even add the seven Rasmussen/POR polls released on Sunday and Saturday before the election, as well:

Released Monday before Election

WA-Sen 49-47 Murray - almost exact
OH-Gov 48-44 Kasigh - actual 49.4-46.7 - almost exact
NV-Sen 48-45 Angle - every public poll got this wrong
WV-Sen 50-46 Manchin - nearly identical to PPP result (only other public poll), both slightly underestimated Manchin
CO-Sen 50-46 Buck - every public poll in the final week had Buck leading
CO-Gov 47-44-6 Hickenlooper - every public poll overestimated Tancredo vs Maes
IL-Sen 46-42 Kirk - actual 48.2-46.3, right on target
IL-Gov 44-38 Brady - same result as every other public poll, all underestimated the final result of 46.1-46.6 Quinn
CT-Gov 48-46 Foley - very close to actual result of 49.0-49.5. No poll had Malloy up, one showed tie w 10% other/undec
CT-Sen 53-46 Blumenthal - actual 54-43, right on target

Released Sunday before Election
NH-Gov 51-45 Lynch - actual 52.6-45.1, almost exact
PA-Gov 52-43 Corbett - actual 54.5-45.5, almost exact
VT-Gov 50-45 Shumlin - actual 49.4-47.9 Shumlin: poll biased toward Democrat

Released Saturday before Election
NH-Sen 56-41 Ayotte - actual 61-36, poll biased toward Democrat
OH-Sen 57-33 Portman - actual 57.3-39.0, slightly underestimating D in a blowout
AZ-Sen 52-32 McCain - actual 59-35, slightly underestimating R in a blowout
AR-Gov 60-38 Beebe - actual 64.6-33.6, slightly underestimating D in a blowout

I’m not trying to be difficult, but don’t understand why you aren’t using the numbers straight from Rasmussen:
http://www.rasmussenreports.com/public_content/politics/elections/election_2010/election_2010_senate_elections/final_rasmussen_poll_results_2010_senate_elections

For instance, this link says the final poll had Rossi up 1, while the actual is Murray by 3. That’s overestimating the Republican by 4, and you say they are “almost exact”.
This link says the final poll had Blumenthal up 7, the actual was 11, also overestimating the Republican by 4, and you say “right on target”.

Did they factor in the cheating? Where was dingy Harry in the polls before he started giving out free food and gift certificates? How many votes did he really have before that “unexpected” power outage.

Is there any way they can poll and factor in the Democrat fraud?

Didn’t think so.

I’m not trying to be difficult, but don’t understand why you aren’t using the numbers straight from Rasmussen:

Polls using Rasmussen's methodology via subsidiary POR were sponsored by two organizations: Rasmussen Reports and FoxNews. I'm using a 3rd party source that includes both. Obviously Rasmussen Reports only includes those done for them.

For instance, this link says the final poll had Rossi up 1, while the actual is Murray by 3. That’s overestimating the Republican by 4, and you say they are “almost exact”.

The final poll with Rasmussen methodology was commissioned by FoxNews, and showed a 49-47 Murray lead. CNN shows an actual result of 51-49 with 97% reporting. I call that almost exact.

Now, polls will fluctuate within (and occasionally outside of) the margin of sampling error. POR polls since October showed the following margins for Murray (oldest to newest): -3, -1, +3, -1, +2, all of which are consistent with an extremely close race, as it proved to be. (Compare to CNN, the Washington Poll, and the Elway Poll, all of which showed margins for Murray of over 5%). While the final poll nailed the correct winner and margin, the poll from a few days previous for Rasmussen Reports showing Rossi 48-47 was perfectly consistent with the actual result and well within the margin of error.

This link says the final poll had Blumenthal up 7, the actual was 11, also overestimating the Republican by 4, and you say “right on target”.

53-46 is well within the margin of error of the actual result of 55-43 (91% reported). As such, the poll was correct, right winner, very close to the margin. POR polls since October (oldest to newest): +10, +11, +6, +5, +13, +7. All of these results are within the margin of error of the actual result. Additionally, indications from Connecticut are that Democrats exceeded their expected results.

Disclaimer: Opinions posted on Free Republic are those of the individual posters and do not necessarily represent the opinion of Free Republic or its management. All materials posted herein are protected by copyright law and the exemption for fair use of copyrighted works.