Rasmussen Polls Were Biased and Inaccurate

[barfo]

Denny, you might be interested in this, as you've been a big fan of Rasmussen and have often trumpeted their numbers.

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.

Imagine that.

barfo

 

[MrJayremmie]

They have always seemed pretty bias.

 

[BLAZER PROPHET]

Polls, to me, are pretty much worthless.

 

[Denny Crane]

http://www.rasmussenreports.com/pub...10/election_night/election_2010_how_did_we_do

Silver talks about how Rasmussen missed a 40 point win for Inouye in Hawaii (by 23 points including margin of error). Funny things happen when your polls close 3 hours after the west coast ones do and 6 hours after the east coast ones do - and after it's already widely reported by the networks that the house is hugely republican gains and the senate is held by the democrats.

You might come to the conclusion that republicans didn't bother to vote because the outcome was meaningless. Yet in the governor's race, which does affect people locally on a day to day basis, Rasmussen predicted 58%-32% for Abercrombie and the outcome was 58.2%-41.1%.

I'm also not seeing the bias toward republicans:

http://www.rasmussenreports.com/pub..._rasmussen_poll_results_2010_senate_elections

 

[44Thrilla]

Rasmussen's polls were, by far, the most accurate in my state

 

[barfo]

Silver talks about how Rasmussen missed a 40 point win for Inouye in Hawaii (by 23 points including margin of error). Funny things happen when your polls close 3 hours after the west coast ones do and 6 hours after the east coast ones do - and after it's already widely reported by the networks that the house is hugely republican gains and the senate is held by the democrats.

You might come to the conclusion that republicans didn't bother to vote because the outcome was meaningless. Yet in the governor's race, which does affect people locally on a day to day basis, Rasmussen predicted 58%-32% for Abercrombie and the outcome was 58.2%-41.1%.

So your contention is that the poll was right, but republicans didn't bother to vote for senator, but did vote for governor? That's really quite implausible.
There were 14,680 blank votes in HI for senator. Even if you assign all of those to the Republican candidate, the margin of victory would have been reduced by only 3.8 percentage points.

I'm also not seeing the bias toward republicans:

Nevertheless, the statistics say the bias is there, whether your bias allows you to see it or not.

barfo

 

[44Thrilla]

In a way, aren't they actually hurting the Republicans by (supposedly) reporting inflated numbers for them? In the close races, my guess would be they are motivating democrats who may not have voted, into actually voting.

 

[barfo]

In a way, aren't they actually hurting the Republicans by (supposedly) reporting inflated numbers for them? In the close races, my guess would be they are motivating democrats who may not have voted, into actually voting.

It depends on where the race stands. If the republican is (in reality) tied or ahead (as was most often the case in this election) and the republican numbers are inflated, that tends to discourage democrats from voting because it make it look like more of a lost cause.

If on the other hand the democrat is leading in reality, then inflating the republican numbers make the race look more competitive, and that might motivate democratic voters to get out and vote. But that scenario was relatively less common this cycle.

barfo

 

[Denny Crane]

So your contention is that the poll was right, but republicans didn't bother to vote for senator, but did vote for governor? That's really quite implausible.
There were 14,680 blank votes in HI for senator. Even if you assign all of those to the Republican candidate, the margin of victory would have been reduced by only 3.8 percentage points.



Nevertheless, the statistics say the bias is there, whether your bias allows you to see it or not.

barfo

Silver also said the democrats would do worse if they didn't pass health care and he was completely wrong.

 

[barfo]

Silver also said the democrats would do worse if they didn't pass health care and he was completely wrong.

So what? That's an opinion. The bias in Rasmussen polling is a fact.

barfo

 

[Denny Crane]

Just looking at the last rasmussen link...

he understated Shelby's vote %, the alaska result, McCain, Boozman, overstated Fiorina, etc.

It looks to me like it really depended on the race.

 

[Denny Crane]

So what? That's an opinion. The bias in Rasmussen polling is a fact.

barfo

No it isn't a fact.

Though when pollsters do poll, they weight the results based upon what they expect the electorate to look like. So if they think it's 60-40 republicans, they'll weight it that way. And they're not exactly guessing about the 60-40 bit, they try to deduce it by asking a few questions that indicate what party the voter is likely to vote for.

And Gallup called me this morning and I did their survey.

 

[MARIS61]

Rasmussen has long been known to be nothing more than a disinformation tool of the far-right.

That's why Denny loves to post them so frequently.

 

[barfo]

No it isn't a fact.

Right... it's not a fact because... because... because you don't like it.

Though when pollsters do poll, they weight the results based upon what they expect the electorate to look like. So if they think it's 60-40 republicans, they'll weight it that way. And they're not exactly guessing about the 60-40 bit, they try to deduce it by asking a few questions that indicate what party the voter is likely to vote for.

Right. And apparently Rasmussen isn't very good at that, because their numbers were biased towards the Republicans.
I don't think Nate is implying that the bias is intentional. It could just be incompetence, or it could be that Rasmussen biases things subconsciously. But the numbers show that whatever the cause, the result is a bias.

And Gallup called me this morning and I did their survey.

We'll have to look for a sudden spike in the population of libertarians according to Gallup.

barfo

 

[Denny Crane]

Right... it's not a fact because... because... because you don't like it.

It's only a fact because you like the guy who says it's a fact (and he's wrong).

Right. And apparently Rasmussen isn't very good at that, because their numbers were biased towards the Republicans.
I don't think Nate is implying that the bias is intentional. It could just be incompetence, or it could be that Rasmussen biases things subconsciously. But the numbers show that whatever the cause, the result is a bias.



We'll have to look for a sudden spike in the population of libertarians according to Gallup.

barfo

Again, looking at rasmussen's page where he shows his results vs. the actuals, he's not particularly favoring any of the candidates. As I pointed out, he understated a lot of the big winners who were republicans as well as Innouye. And it looks to me like he got worse in the western time zones where the voters knew the Republicans already were assured to win the house and lose the senate.

But really, barfo, weren't the election results heavily biased toward republicans? They gained the 2nd most seats in the house in history, won 2/3 of the senate elections, a huge number of seats in state legislatures, as well as a big gain in governorships.

Rasmussen predicted the right gain in senate seats and a gain of 55 house seats with several too close to call (republicans gained 60 with some outstanding as of election night).

Where is the inaccuracy?

 

[Denny Crane]

LOL

http://blogs.dailymail.com/donsurber/archives/24109

Speaking of accuracy, on Election Day, Nate Silver predicted a pickup of 7 seats in the Senate for Republicans.

They picked up 6.

In the Senate on Election Day, Nate Silver predicted a 53-seat gain for Republicans. They picked up 61.

So let’s see — Republicans went 24-13 in the Senate; he said 25.3-11.6 — he was off by 1.3 seats.

In April I said Republicans would go 23-13 (the West Virginia seat was added in July*) and I was off by 1.

In the House on Election Day, Nate Silver was off by at least 7.8 (with 10 seats to go) while I was off by at least 10.

 

[Denny Crane]

http://www.leftcoastrebel.com/2010/11/nates-recent-rasmussen-bias-i-sure-hope.html

The brilliant Nate Silver may be just about the best in the polling business, but that doesn't mean he's not subject to insufferable hackery. For one reason or another, Nate has is in for Scott Rassmusen lately and Nate's bias has been spilling into the pages of NY Times this election season.

Today's post on Rasmussen accuracy reeks of similar hackedness.

Nate made much ado of Rasmussen's worst poll ever in his analysis. The critique is germane, but not in the statistical sense. It would certainly be, by Nate's own god statistics, an outlier and as such should have been removed from Nate's accuracy model.

An outlier in statistics is a data point that is outside the norm and that certainly describes the worst poll in Nate's ten years. That's not to say that we ignore it. By all means, Scott Rasmussen ought to take the poll very seriously and find a reason for its inaccuracy. His methodology could be flawed, it could be poor polling questions or polling staff, he should look into why no one noticed it until after released, and it could even be a bad roll of the statistical die. Nate ought to do the same when considering the poll, but if Nate's just objectively looking at the data, he can't just arbitrarily ignore the fact that it is radically outside the norm for Rasmussen. Doing so skews his own model.

 

[barfo]

http://www.leftcoastrebel.com/2010/11/nates-recent-rasmussen-bias-i-sure-hope.html

The brilliant Nate Silver may be just about the best in the polling business, but that doesn't mean he's not subject to insufferable hackery. For one reason or another, Nate has is in for Scott Rassmusen lately and Nate's bias has been spilling into the pages of NY Times this election season.

Today's post on Rasmussen accuracy reeks of similar hackedness.

Nate made much ado of Rasmussen's worst poll ever in his analysis. The critique is germane, but not in the statistical sense. It would certainly be, by Nate's own god statistics, an outlier and as such should have been removed from Nate's accuracy model.

An outlier in statistics is a data point that is outside the norm and that certainly describes the worst poll in Nate's ten years. That's not to say that we ignore it. By all means, Scott Rasmussen ought to take the poll very seriously and find a reason for its inaccuracy. His methodology could be flawed, it could be poor polling questions or polling staff, he should look into why no one noticed it until after released, and it could even be a bad roll of the statistical die. Nate ought to do the same when considering the poll, but if Nate's just objectively looking at the data, he can't just arbitrarily ignore the fact that it is radically outside the norm for Rasmussen. Doing so skews his own model.

I like the bit about god statistics.

Unfortunately, this guys objection is not terribly meaningful. Take out the HI results, and Rasmussen still missed by 5.5 points instead of 5.8. And that's still the worst performance among the major polling firms.

barfo

 

[barfo]

It's only a fact because you like the guy who says it's a fact (and he's wrong).

Show me some evidence of that. All you are doing is expressing your opinion that he's wrong. Frankly, I've seen enough Nate Silver math and enough DennyMath (TM) to know that he's infinitely more careful about numbers than you are.

Again, looking at rasmussen's page where he shows his results vs. the actuals, he's not particularly favoring any of the candidates. As I pointed out, he understated a lot of the big winners who were republicans as well as Innouye. And it looks to me like he got worse in the western time zones where the voters knew the Republicans already were assured to win the house and lose the senate.

The difference here is that Nate is actually crunching the numbers, and you are just eyeballing them.

But really, barfo, weren't the election results heavily biased toward republicans? They gained the 2nd most seats in the house in history, won 2/3 of the senate elections, a huge number of seats in state legislatures, as well as a big gain in governorships.

In the sense that the word "bias" is being used here, no. The election results are unbiased. They are what they are. Of course it is true that Republicans won lots of races.
Bias here means deviation from reality, not deviation from a 50-50 republican/democrat split.

Rasmussen predicted the right gain in senate seats and a gain of 55 house seats with several too close to call (republicans gained 60 with some outstanding as of election night).

Where is the inaccuracy?

In the margin of victory in individual races.

barfo

 

[BGrantFan]

Yes, Rasmussen sure is biased!

In 2008, Obama won 53%-46% and our final poll showed Obama winning 52% to 46%. While we were pleased with the final result, Rasmussen Reports was especially pleased with the stability of our results. On every single day for the last six weeks of the campaign, our daily tracking showed Obama with a stable and solid lead attracting more than 50% of the vote.

We also have provided a summary of our 2008 state-by-state presidential results for your review.

In 2004 George W. Bush received 50.7% of the vote while John Kerry earned 48.3%. Rasmussen Reports polling projected that Bush would win 50.2% to 48.5%. We were the only firm to project both candidates' totals within half a percentage point by (see our 2004 results).

Stupid criticism; stupid thread. What is the point of it? Every pollster has their misses.

 

[agoo]

Yes, Rasmussen sure is biased!



Stupid criticism; stupid thread. What is the point of it? Every pollster has their misses.

Look at how severely they underrated Kerry in 2004!!!!!!

Also, Denny, how does one get called by Gallup? I get plenty of credit card offers, but never anything fun like that.

 

[julius]

Yes, Rasmussen sure is biased!



Stupid criticism; stupid thread. What is the point of it? Every pollster has their misses.

I bet some have mistresses too.

 

[barfo]

Yes, Rasmussen sure is biased!



Stupid criticism; stupid thread. What is the point of it? Every pollster has their misses.

The point was to discuss 2010 results, not the 2008 and 2004 results. Because, you know, it's 2010 now.
Yes, every pollster has their misses. Some have more misses than others. It's interesting to know which ones are more accurate.

barfo

 

[BGrantFan]

The point was to discuss 2010 results, not the 2008 and 2004 results. Because, you know, it's 2010 now.
Yes, every pollster has their misses. Some have more misses than others. It's interesting to know which ones are more accurate.

barfo

Historically, Rasmussen has been the most accurate. Were you posting threads when that was the case? You seem like the biggest partisan hack on this board, IMO.

 

[barfo]

Historically, Rasmussen has been the most accurate. Were you posting threads when that was the case? You seem like the biggest partisan hack on this board, IMO.

I posted this thread in response to about 10,000 threads of Denny's trumpeting various Rasmussen polls this year. Had he not posted those, I would not be posting this.

As for your opinion, it's amusing.

barfo

 

[BGrantFan]

I posted this thread in response to about 10,000 threads of Denny's trumpeting various Rasmussen polls this year. Had he not posted those, I would not be posting this.

As for your opinion, it's amusing.
barfo

The most funny things are based in reality.

 

[Denny Crane]

Look at how severely they underrated Kerry in 2004!!!!!!

Also, Denny, how does one get called by Gallup? I get plenty of credit card offers, but never anything fun like that.

They called me out of the blue.

They have to use some random phone number generator for the poll to be reasonable.

 

[Denny Crane]

I like the bit about god statistics.

Unfortunately, this guys objection is not terribly meaningful. Take out the HI results, and Rasmussen still missed by 5.5 points instead of 5.8. And that's still the worst performance among the major polling firms.

barfo

LOL

It's not meaningful because you don't agree with it.

 

[barfo]

LOL

It's not meaningful because you don't agree with it.

No, it's not numerically meaningful. If you take his suggestion and remove HI as an outlier, the conclusion is still that Rasmussen had the worst record.

As for whether I agree with him, I don't think I do, unless there is evidence that the HI result came from some glitch - like, for instance, someone made a typo. In that case, it definitely should be thrown out. But if it is just an infrequent failure of the method applied as intended, it seems to me that it should be kept.

barfo

 

[Denny Crane]

No, it's not numerically meaningful. If you take his suggestion and remove HI as an outlier, the conclusion is still that Rasmussen had the worst record.

As for whether I agree with him, I don't think I do, unless there is evidence that the HI result came from some glitch - like, for instance, someone made a typo. In that case, it definitely should be thrown out. But if it is just an infrequent failure of the method applied as intended, it seems to me that it should be kept.

barfo

I didn't read his article to have anything to do with HI being an outlier.

Maybe you want to read it again.