What the frock is wrong with Rasmussen?

[Ava]

Rasmussen is consistently the outlier in these polls, sometimes by a wide margin, as they are now. How do they manage that? How do they defend that?

Absolutely nothing is wrong with their poll, if you look at their strongly approve/disapprove poll. That matches all the others about perfectly at -11%. The numbers used in the table you have is from the total approve. I am not sure what this means. You are right though, the Rasmussen number is an outlier.

Daily Presidential Tracking Poll - Rasmussen Reports?

That's interesting about how different their "Presidential Approval Index" is from the result used in the RCP table.

I've been suspicious of them for months now, unless they are polling different people and groups...

 

[RandomVariable]

Gallup Daily: Obama Job Approval
Each result is based on a three-day rolling average
Gallup Daily: Obama Job Approval

 

[Amelia]

Everyone seems to understand that Rassmussen is usually off to the right.

That's what makes their recent trend so mystifying.

They appear to have swung sharply in the opposite direction.

I guess I should look at their results for specific candidates to see if those have also veered left in the last several months.

Stat is correct that Rasmussen's steering has a hard pull to the right and yes, you are right that is an anomaly for them.

What usually happens when they get out of step is that they do drop their rightward bias in the last 2 weeks immediately prior to an election in order to be "accurate" as far as the results are concerned. Since they are known to do things like this it would not be surprising if they screwed up this result by getting a decimal point in the wrong place or forgetting to add instead of subtract.

The problem with being a liar is that they have to be both consistent and never forget the lies that they have told in the past. It is when liars catch themselves out with contradictions that you know that they are untrustworthy.

So when it comes to Rasmussen I usually just ignore everything they publish because there is no way to be sure exactly how much bias they are using at any point in time. And yes, even venerable Gallup has become suspect of late too so it isn't just Rasmussen.

The indexes of the polls are a much better measure in my opinion and I ignore the current snapshot unless it is right before an election and focus instead on the trends because they are the best indicator of all in my opinion. I am sure that [MENTION=46168]Statistikhengst[/MENTION] can do the math but my experience tells me that the trend predicts the winner/loser far more accurately than any specific pollster.

Just my 2 cents since you were kind enough to invite me.

Peace
DT

I question the use of the word "anomaly". Their hard right pull appears to have vanished some time during the past year. They've consistently had the best numbers for Obama for months now.

It's been the same in the generic congressional vote. When trends are toward the Democrats, their result is most emphatically D. When other polls start favoring Republicans, they're the last to finally get to R, if they even do.

 

[Statistikhengst]

Rasmussen samples what they define as Likely Voters...seems very out of sync with the actual sentiments of the public.

I did a study on this in 2008.

The LV polls were too restrictive.

The RV polls were too generous.

The averages were spot-on.

In the 2008, I predicted Obama with 52.77%, based purely on the averages of polls, 1/2 of the LV, the other half RV. He won with 52.87%.

I predicted his margin at +7.54. He won with +7.26.

So, LV is not all that terrible. It serves a purpose, in tandem with RV polls, imo.

 

[Statistikhengst]

The problem is that the Ds are likely to be in very big trouble on several fronts so Rasmussen is carrying water for the country club Rs to avoid a party split.

If that is true, then what Ras is doing is illegal.

 

[Statistikhengst]

Everyone seems to understand that Rassmussen is usually off to the right.

That's what makes their recent trend so mystifying.

They appear to have swung sharply in the opposite direction.

I guess I should look at their results for specific candidates to see if those have also veered left in the last several months.

Stat is correct that Rasmussen's steering has a hard pull to the right and yes, you are right that is an anomaly for them.

What usually happens when they get out of step is that they do drop their rightward bias in the last 2 weeks immediately prior to an election in order to be "accurate" as far as the results are concerned. Since they are known to do things like this it would not be surprising if they screwed up this result by getting a decimal point in the wrong place or forgetting to add instead of subtract.

The problem with being a liar is that they have to be both consistent and never forget the lies that they have told in the past. It is when liars catch themselves out with contradictions that you know that they are untrustworthy.

So when it comes to Rasmussen I usually just ignore everything they publish because there is no way to be sure exactly how much bias they are using at any point in time. And yes, even venerable Gallup has become suspect of late too so it isn't just Rasmussen.

The indexes of the polls are a much better measure in my opinion and I ignore the current snapshot unless it is right before an election and focus instead on the trends because they are the best indicator of all in my opinion. I am sure that [MENTION=46168]Statistikhengst[/MENTION] can do the math but my experience tells me that the trend predicts the winner/loser far more accurately than any specific pollster.

Just my 2 cents since you were kind enough to invite me.

Peace
DT

Trends. Yes.

Polling averages, dealing with margins, can and should look like sinus curves. They will open and close some and then open again, and so forth.

You can spot a winning candidate a mile away when the curve for him opens more than it closes, from sinus cycle to the next sinus cycle.

Example:

Candidate A in the first week. An average of +3.2 over candidate B.
Candidate A in the second week: an average of +2.7 over candidate B.

Shrinkage: 0.5

Candidate A in the third week: an average of +4.1 over candidate B.
Candidate A in the fourth week: an averag of +3.7 over candidate B.

Shrinkage: 0.4, but the jump between week 2 and week 3 was 1.4...

And so on and so on.

This is why when you see one candidate slightly down and people are shouting "he's goin down" - not so fast. He might be at a shrinkage point in a sinus curve, assuming that the polling from all sources is kosher.



Hope that info helps.

 

[william the wie]

The problem is that the Ds are likely to be in very big trouble on several fronts so Rasmussen is carrying water for the country club Rs to avoid a party split.

If that is true, then what Ras is doing is illegal.

Try proving that in court. If, as some columnists at Real Clear Politics claim, the Ds appear likely to lose a net 6-12 senate seats then an R/TEA party split is much more likely.

 

[WelfareQueen]

Rasmussen samples what they define as Likely Voters...seems very out of sync with the actual sentiments of the public.

I did a study on this in 2008.

The LV polls were too restrictive.

The RV polls were too generous.

The averages were spot-on.

In the 2008, I predicted Obama with 52.77%, based purely on the averages of polls, 1/2 of the LV, the other half RV. He won with 52.87%.

I predicted his margin at +7.54. He won with +7.26.

So, LV is not all that terrible. It serves a purpose, in tandem with RV polls, imo.

In 2012 Rasmussen under sampled likely Democrat voters relative to most other polling agencies by 2 or 3 points. I think Dems were +8 in 2008 and +7 in 2012. Rassmussen was consistently under representing those Democrat likely voter samples.

I do not know if they have changed their methodology. My guess is they have. I would think the most difficult guess for any polling agency is what the actual turnout by party affiliation will be and sampling correctly. Rassmussen did not do a good job in 2012.

 

[WelfareQueen]

Everyone seems to understand that Rassmussen is usually off to the right.

That's what makes their recent trend so mystifying.

They appear to have swung sharply in the opposite direction.

I guess I should look at their results for specific candidates to see if those have also veered left in the last several months.

Stat is correct that Rasmussen's steering has a hard pull to the right and yes, you are right that is an anomaly for them.

What usually happens when they get out of step is that they do drop their rightward bias in the last 2 weeks immediately prior to an election in order to be "accurate" as far as the results are concerned. Since they are known to do things like this it would not be surprising if they screwed up this result by getting a decimal point in the wrong place or forgetting to add instead of subtract.

The problem with being a liar is that they have to be both consistent and never forget the lies that they have told in the past. It is when liars catch themselves out with contradictions that you know that they are untrustworthy.

So when it comes to Rasmussen I usually just ignore everything they publish because there is no way to be sure exactly how much bias they are using at any point in time. And yes, even venerable Gallup has become suspect of late too so it isn't just Rasmussen.

The indexes of the polls are a much better measure in my opinion and I ignore the current snapshot unless it is right before an election and focus instead on the trends because they are the best indicator of all in my opinion. I am sure that [MENTION=46168]Statistikhengst[/MENTION] can do the math but my experience tells me that the trend predicts the winner/loser far more accurately than any specific pollster.

Just my 2 cents since you were kind enough to invite me.

Peace
DT

I don't think Rasmussen was intentionally lying the last election cycle....they just got it wrong. I think it is very difficult for any polling group to get a handle on voter intensity. Most thought voter turnout for Obama's core voting bloc would be down significantly in 2012. It was not. Blacks stilled turned out very heavily. Hispanics and Asians were down, but only slightly. Overall, it was a drop of about 1%. It made the election a little closer, but not much.

I agree with you that aggregate polling data is the best. It evens out the outliers. But to me, it seems like accurate polling is as much art as science. For example, the speculation is Republican voter intensity will be much higher in the 2014 mid-terms than it was in 2012. There is some data to support this conclusion, but who knows? No one can truly say who will show up to vote on election day. I have not researched how the various polling agencies are weighting Democrat versus Republican leaning voters in the currect election cycle. I think it would be interesting to know.

 

[WelfareQueen]

I just finished some research. I think I have an answer. Gallup rated voter intensity for Democrat or lean Democrat on election day 2012 at + 7.5% The final margin was somewhere around 7%...so they were very close.

As of 1-8-14 they put the Democrat or lean Democrat at +5%. This data is exactly in line with other analysis I have seen recently. To put this in perspective....if the Romney/Obama election were held today it would be a virtual tie.

I do not know how each polling agency is weighting their data by party affiliation other than Gallup. Rassmussen will not tell you unless you pay them a fee. I am not going to do that.

Right now Gallup has Obama approve/disapprove at 39% - 53%. CNN is weighted around 5.5% Dem Lean Dem as of a month ago. They have Obama at 41% -56%

CNN also has the generic Congressional Ballot at Republican +5%. My guess is Gallup and CNN are pretty close and Rasmussen is a significant outlier.


RealClearPolitics - Election Other - 2014 Generic Congressional Vote

 

[Mad_Cabbie]

I just finished some research. I think I have an answer. Gallup rated voter intensity for Democrat or lean Democrat on election day 2012 at + 7.5% The final margin was somewhere around 7%...so they were very close.

As of 1-8-14 they put the Democrat or lean Democrat at +5%. This data is exactly in line with other analysis I have seen recently. To put this in perspective....if the Romney/Obama election were held today it would be a virtual tie.

I do not know how each polling agency is weighting their data by party affiliation other than Gallup. Rassmussen will not tell you unless you pay them a fee. I am not going to do that.

Right now Gallup has Obama approve/disapprove at 39% - 53%. CNN is weighted around 5.5% Dem Lean Dem as of a month ago. They have Obama at 41% -56%

CNN also has the generic Congressional Ballot at Republican +5%. My guess is Gallup and CNN are pretty close and Rasmussen is a significant outlier.


RealClearPolitics - Election Other - 2014 Generic Congressional Vote

I don't really think that he would have gotten the nod, even today. The problem with Romney is that he just did not inspire the public enough.

You can't just "agree" with your party's talking points - you have to inspire them.

I liked Romney as a person, he just was not very interesting.

Repealing Obamacare was not a good platform - he should have offered up something better.

 

[Statistikhengst]

I just finished some research. I think I have an answer. Gallup rated voter intensity for Democrat or lean Democrat on election day 2012 at + 7.5% The final margin was somewhere around 7%...so they were very close.

As of 1-8-14 they put the Democrat or lean Democrat at +5%. This data is exactly in line with other analysis I have seen recently. To put this in perspective....if the Romney/Obama election were held today it would be a virtual tie.

I do not know how each polling agency is weighting their data by party affiliation other than Gallup. Rassmussen will not tell you unless you pay them a fee. I am not going to do that.

Right now Gallup has Obama approve/disapprove at 39% - 53%. CNN is weighted around 5.5% Dem Lean Dem as of a month ago. They have Obama at 41% -56%

CNN also has the generic Congressional Ballot at Republican +5%. My guess is Gallup and CNN are pretty close and Rasmussen is a significant outlier.


RealClearPolitics - Election Other - 2014 Generic Congressional Vote

Thank you for adding this data. Good work.

I do want to point out some things for you all to chew on:

1.) Congressional ballot measurement is not the same as approval/disapproval, which, in turn, is not the same as favorability/disfavorability, so it is entirely possible to come up very different values for all three. Even with very high fav numbers for him personally, Eisenhower's Republican party suffered massive losses in the 1958 mid-terms.

2.) Your summation about the 2012 is correct, which only impugnes Gallup's reputation even more. If D +4 would have been enough to make 2012 a virtual tie, and Gallup was showing for weeks on end that Romney was up by as much as +6, then this can only mean that not only within the Gallup Model, which was based on 78% white vote instead of the actual 72% that came out, it must have used a party identification of at least R +1, which is about as false as false can get. Because of that 6% shift in the white vote, Obama still would have gotten 2.5% of that 6%, had it been 78% of the electorate.

3.) Just to show how unbelievably wrong Gallup also was in 2010, in the last mid-terms, on the eve of the Mid-terms, it showed R +15 in the congressional ballot (55-40)

Gallup.Com - 2010 Congressional Midterm Elections


Actual results?

IN the House of Representatives, it was R +6.8

United States House of Representatives elections, 2010 - Wikipedia, the free encyclopedia

In the Senate, it was R +5.7

http://uselectionatlas.org/RESULTS/data.php?year=2010&datatype=national&def=1&f=1&off=3&elect=0

Combined, it was a total of 153,397,897 votes cast.

R total: 78,009,816 (50.85%)

D total: 68,476,144 (44.64%)

Combined margin: R +6.21

So, essentially, Gallup was off by NINE FULL points in 2010. That is just plain old criminal.

4.) In the year of a possible GOP wave, expect party identification to shift more toward the GOP and also expect the generic to move to at least a tie, if not R+1 to R+5. We are not seeing that yet because most people are not like the political junkies here in USMB, they don't even care about the elections yet.

Hope that data helps.

Once again, nice job, [MENTION=45104]WelfareQueen[/MENTION]

 

[Statistikhengst]

Rasmussen samples what they define as Likely Voters...seems very out of sync with the actual sentiments of the public.

I did a study on this in 2008.

The LV polls were too restrictive.

The RV polls were too generous.

The averages were spot-on.

In the 2008, I predicted Obama with 52.77%, based purely on the averages of polls, 1/2 of the LV, the other half RV. He won with 52.87%.

I predicted his margin at +7.54. He won with +7.26.

So, LV is not all that terrible. It serves a purpose, in tandem with RV polls, imo.

In 2012 Rasmussen under sampled likely Democrat voters relative to most other polling agencies by 2 or 3 points. I think Dems were +8 in 2008 and +7 in 2012. Rassmussen was consistently under representing those Democrat likely voter samples.

I do not know if they have changed their methodology. My guess is they have. I would think the most difficult guess for any polling agency is what the actual turnout by party affiliation will be and sampling correctly. Rassmussen did not do a good job in 2012.

I'm not so sure about that. Most all other pollsters who allow self-identification were showing D+6 to D+8 for just about every battleground state and exactly those kinds of figures showed up in the exit polls. In other words, a lot of people were clearly indicating their preference in enough polls in order to get a good picture of how the electorate was shaping up.

And on my politics blog, where I have a tradition of doing battleground reports every single night for 42 days (6 weeks) before a presidential GE, I kept noting that D +6 to D +8 kept showing up. So, I don't think it's that hard to guess.

You can get a taste of those battleground reports starting here:

http://rosenthalswelt.blogspot.de/2012/09/preface-to-battleground-reports.html

(BTW, in the preface report, notice the mention of Ohio and it's polling average 42 days away from the election and notice how it almost perfectly matches where Obama landed nationally on election day)

Statistikhengst's ELECTORAL POLITICS - 2013 and beyond: Battleground Report 09/25/2012 - T-minus 42 days



And, based on the 1,900 state polls, it was obvious to me that, regardless of the whacky national polling disparities, Obama was winning in the battleground states where it counted. He carried Michigan by +9.54 on election night. Michigan was never really in play. He carried Minnesota (now a 10 for 10 DEM state on the presidential level) by +7.69. It too the networks almost three hours to call Minnesota, but Romney won Georgia by 7.80%, just 0.11% more than Obama's win in Minnesota, and yet, Georgia was called almost instantly for Romney. There were just too many states where Obama was above +5 in the polling averages, meaning that he only needed 1 or 2 states that were below the +5 mark in order to win. So, predicting a win for him was not that hard a call. And showing many pollsters the door for bad polling at the national level was called for.

 

[TooTall]

The problem with being a liar is that they have to be both consistent and never forget the lies that they have told in the past. It is when liars catch themselves out with contradictions that you know that they are untrustworthy.

So when it comes to Rasmussen I usually just ignore everything they publish because
DT

I use the same criteria for Obama.

 

[Statistikhengst]

The problem with being a liar is that they have to be both consistent and never forget the lies that they have told in the past. It is when liars catch themselves out with contradictions that you know that they are untrustworthy.

So when it comes to Rasmussen I usually just ignore everything they publish because
DT

I use the same criteria for Obama.

Actually, until you showed up, people have not been flinging poo here.

We have been sticking to the numbers, and Gallup and Rasmussen's reputations can be proven mathematically.

So, as you see, this has nothing to do with any words that may have come out of the President's mouth that you think are lies. You are capable of understanding the difference, right?

Amelia, as far as I can tell, is a Rightie, and she doesn't act this way. So is WelfareQueen. Right now, in this thread, Righties and Lefties are having a meaningful conversation. Please don't fuck it up with mud. Thank you.

 

[Sallow]

they only polled liberal democrats

Funny thing is..population wise..there are more Liberal Democrats in this country than there are Conservative Republicans.

Which is why you folks cling to the Electoral College.

You'd never see a conservative President again if it were just left to the popular vote.

 

[JakeStarkey]

Daily Presidential Tracking Poll - Rasmussen Reports?

Thursday, January 16, 2014

The Rasmussen Reports daily Presidential Tracking Poll for Thursday shows that 51% of Likely U.S. Voters approve of President Obama's job performance. Forty-seven percent (47%) disapprove (see trends).

The latest figures include 26% who Strongly Approve of the way Obama is performing as president and 37% who Strongly Disapprove. This gives him a Presidential Approval Index rating of -11.

This is the president’s highest daily job approval rating since late October.

 

[Statistikhengst]

they only polled liberal democrats

Funny thing is..population wise..there are more Liberal Democrats in this country than there are Conservative Republicans.

Which is why you folks cling to the Electoral College.

You'd never see a conservative President again if it were just left to the popular vote.

With all respect to you, I am not so sure about that.

Most partisan identification polling shows Liberal Dems and Conservative Republicans pretty evenly split, with the hardest of the hard core going more for the Conservatives, actually. But when you factor in moderates, the Ds indeed have an edge, but not a majority, which means that the so-called "Independents" pretty much decide elections, which is not such a bad thing. But we will never know for sure.

If you check my voter registration statistics stuff, it is also pretty inconclusive:

Statistikhengst's ELECTORAL POLITICS - 2013 and beyond: Complete Voter Registration Statistics (USA), End of 2013

(this was just updated for Ohio on January 4th, so much of it is still "fresh")

 

[WelfareQueen]

I just finished some research. I think I have an answer. Gallup rated voter intensity for Democrat or lean Democrat on election day 2012 at + 7.5% The final margin was somewhere around 7%...so they were very close.

As of 1-8-14 they put the Democrat or lean Democrat at +5%. This data is exactly in line with other analysis I have seen recently. To put this in perspective....if the Romney/Obama election were held today it would be a virtual tie.

I do not know how each polling agency is weighting their data by party affiliation other than Gallup. Rassmussen will not tell you unless you pay them a fee. I am not going to do that.

Right now Gallup has Obama approve/disapprove at 39% - 53%. CNN is weighted around 5.5% Dem Lean Dem as of a month ago. They have Obama at 41% -56%

CNN also has the generic Congressional Ballot at Republican +5%. My guess is Gallup and CNN are pretty close and Rasmussen is a significant outlier.


RealClearPolitics - Election Other - 2014 Generic Congressional Vote

I don't really think that he would have gotten the nod, even today. The problem with Romney is that he just did not inspire the public enough.

You can't just "agree" with your party's talking points - you have to inspire them.

I liked Romney as a person, he just was not very interesting.

Repealing Obamacare was not a good platform - he should have offered up something better.

I completely agree, and your point is validated by voter intensity. Democrat or lean Democrat voters turned out at about 7% higher rate than the opposition in 2012. You can fairly say they were more inspired to vote for Obama than the Romney folks.

 

[william the wie]

This is not deliberate poo throwing but given the nature of the Washington-Boston echo chamber and outliers the thing that surprised me about Obama's 2008 election was that he was from Chicago and still got nominated much less elected. This, in my opinion, was solely due to having an NY senator primary opponent for the nomination and Sarah Palin on the opposition ticket in the GE. Any numbers on this sort of regional antipathy?