CSNbbs

Why pollsters don't 'weight' surveys for Dem-GOP mix

Quote:“Party ID is not a demographic quality like age, sex, income or education. It’s an attitude,” said Andrew Kohut, president of the Pew Research Center. “And it’s an attitude that varies with preferences, so generally when a Republican wins you will see a boost in Republican identification and when a Democrat wins you will see a boost in Democratic identification. If you try to standardize the party ID number, you standardize out some of that change.”

Kohut notes, for example, that Pew successfully called the Republican surge in congressional seats in 2010, but would have missed the change if it had insisted on weighting its results to reflect the Democrat’s pronounced party identification advantage after Obama’s 2008 victory.

Quote:"If a pollster weights by party ID, they are substituting their own judgment as to what the electorate is going to look like. It's not scientific," said Doug Schwartz, the director of the Quinnipiac University Polling Institute

Quote:"There are more people who want to identify with the Democratic Party right now than the Republican Party," Schwartz said.

Part of what might explain the tempest over the polls is the Republican conviction that Americans would be driven, in big numbers, into the party’s arms in 2012 because of the sputtering economy. Unemployment stays above 8% and GOP registration (or, at least, identification) has to go up; so the thinking went.

Many surveys have shown, instead, that — though voters continue to have deep misgivings about the economy — Americans don’t have a warm feeling for the GOP standard-bearer. If they don’t take to Mitt Romney, they’re less likely to identify themselves with his party.

Quote:Kohut said that reputable pollsters are constantly checking to make sure that they have sampled correctly. If the “horse race” result suddenly is way off what other surveys have found, they go back to check all the variables.

“We will go back to check how many interviews were done, was it a reasonable sample size in terms of age and education and other factors,” Kohut said. But party registration is going to vary and can’t be controlled in advance, he said.

Quote:Lee Miringoff, director of the Marist College Institute for Public Opinion in New York, is conducting state polls for NBC News and the Wall Street Journal. He also does not weight results for a pre-supposed party ID mix. Miringoff told the Atlantic that the party focus is too narrow but easily “jumped on” by those who think they know what the Dem-GOP mix will be this Nov. 6.

He rejected the notion that pollsters would have something to gain by fixing the results. "Why,” Miringoff said, “would pollsters want to look inaccurate?"

Weighting by party ID and representative sample aren't the same thing.

Since you are ignorant take your own advice and STFU.

You'll never convince them until election day. And even then, they'll blame the results on the librul meteor depressing turnout with their skewed polls.

What a shock. Maxi is as ignorant as Tom is about polls.

Weighting and sampling are not the same thing ladies. The criticisms of the polls aren't based on weighting, they're based on sampling. Ergo an article talking about why polls don't weight based on party ID has nothing to do with the argument being made about these polls.

What a shock. Liberals making an argument that isn't being made to deflect from the argument that is being made and then working to disprove the former.

What legitimate sampling concerns have been made? Most of the more respected (long-running) polls have similar outcomes from what I've seen and all have shown similar dips at similar times, so I'm wondering where the concern comes from?

(09-28-2012 03:39 PM)UCF08 Wrote: [ -> ]What legitimate sampling concerns have been made? Most of the more respected (long-running) polls have similar outcomes from what I've seen and all have shown similar dips at similar times, so I'm wondering where the concern comes from?

The main point of contention is that Dems are being sampled based on 2008, or greater, turnout models. A poll that samples dems 7 to 10 points more than republicans is obviously going to show big leads for Obama.

The question is, in a bad economy, with unpopular polices, the emotion of Hope and Change gone and novelty of the first African American presidential nominee no longer a factor, does Obama really get a 2008 or greater turnout as most of the polls suggest?

I for one believe that to be highly unlikely. I believe the turnout model will be between 2004 and 2008.

Listen Boys. I passed a 200 level stats class in college. Then another 400 level in one grad school. I can see both sides in this argument. It's too easy to get caught up in the mumbo jumbo. That's why I go to Intrade. Barack is approaching 80% there. That's all you need to know. Take that sampling, weighted party z score multiplied by the derivative of the sample then shove it where the sun don't shine. The people who do understand the stuff is putting their money where their mouth is. Barack is an 80% favorite right now. Quit arguing.

(09-28-2012 03:53 PM)Machiavelli Wrote: [ -> ]Listen Boys. I passed a 200 level stats class in college. Then another 400 level in one grad school. I can see both sides in this argument. It's too easy to get caught up in the mumbo jumbo. That's why I go to Intrade. Barack is approaching 80% there. That's all you need to know. Take that sampling, weighted party z score multiplied by the derivative of the sample then shove it where the sun don't shine. The people who do understand the stuff is putting their money where their mouth is. Barack is an 80% favorite right now. Quit arguing.

How did Intrade do on the Obamacare ruling? The striking down of the mandate was at 77%.

They also blew who'd control the Senate in 2006, Howard Dean's Iowa primary and before the first presidential debate in 2004 they were predicting Bush would blow Kerry away with Bush shares trading at 75 cents and Kerry's trading around 30 cents.

"The wisdom of crowds is highly overstated."

'Unskewed Polls' Critics Miss Basics Of Party Identification

Quote:This focus on party identification is as misguided as it was eight years ago. Here are five tips to avoid some common misconceptions at the heart of the controversy.

1. Party identification is an attitude, not a demographic characteristic. "Generally speaking, do you consider yourself a Republican, a Democrat, an independent or what" is basically the question most pollsters ask. People can change their minds about which party they consider themselves closer to. They cannot change their age, gender or race (at least not easily).

2. Party identification does not equal party registration. Many (but not all) states ask voters to affiliate with a party when they register to vote, usually to enable voting in party primary elections, and most of those states publish statistics on the number of Democrats and Republicans.

As Nate Cohn of The New Republic has noted, however, this week's much-criticized Florida survey conducted by CBS News, The New York Times and Quinnipiac University asked questions about both and, not surprisingly, found much inconsistency. Specifically, the poll found twice as many voters who said they consider themselves independent or part of a third party (36 percent) as voters who said they were registered with no affiliation to the Democratic or Republican parties (18 percent). Notably, registered Republicans were a little more likely than registered Democrats to report an independent identification.

3. Party identification can change slightly during a presidential campaign, as the 2004 experience demonstrates. Although the vast majority of voters have a sense of party identification that rarely alters, some voters -- particularly those who just "lean" to one of the parties -- may shift back and forth between categories depending on recent events. When pollsters ask the party ID question and the context in which they ask it can also affect the answers.

4. Claims that media polls "assume" a specific partisan or demographic composition of the electorate are mostly false. The pollsters behind most of the national media surveys, including those who conduct the CBS/New York Times/Quinnipiac, NBC/Wall Street Journal/Marist and Washington Post polls, all use the same general approach: They do not directly set the partisan or demographic composition of their likely voter samples. They first sample adults in each state, weighting the demographics of the full adult sample (for characteristics such as gender, age, race and education) to match U.S. Census estimates for the full population. They then select "likely voters" based on questions answered by the respondents, without making any further adjustments to the sample's demographics or partisanship.

There are pollsters that weight the subset of "likely voters" by party or to match very specific assumptions about the demographics of those they expect to vote. However, such practices are generally shunned by the national media surveys whose recent results have drawn most of the "skewed poll" criticism.

5. Weighting a new survey to match the party ID results of an old exit poll is a bad idea; "unskewing" polls by weighting to Rasmussen Reports' party ID results is even worse. Exit polls provide a helpful guide to the composition of past electorates, but have their own potential for random error and, as recent experience shows, sometimes make for poor predictors of the future.

But if weighting current surveys to match past exit polls is a bad idea, the new idea of 2012 -- "unskewing" polls by reweighting their results to match the party results produced by the Rasmussen polls -- is galactically stupid.

Rasmussen does report party affiliation results for its adult samples, but it asks a very different question and asks it using an automated, recorded voice -- rather than live interviewers. Moreover, some believe Rasmussen's method reaches a less-than-representative sample of adults.

Put the debate over Rasmussen's methodology aside, however. Even Scott Rasmussen thinks weighting other surveys to his measurements is a bad idea. "Different firms ask about partisan affiliation in different ways," he told BuzzFeed. "You cannot compare partisan weighting from one polling firm to another."

In Nate I trust!

Poll Averages Have No History of Consistent Party Bias

Maxipad, CommieTom, ucf, I have some advice for you, drugfree.org

SuckOnAllRears, I have some advice for you. Turn off Fox News! Nate Silver knows a little more on this matter than Stuart Varney and Sean Hannity.

Although it's probably better for you to just stand, heavily sedated, in the corner

(09-29-2012 11:48 PM)Redwingtom Wrote: [ -> ]In Nate I trust!

Poll Averages Have No History of Consistent Party Bias

Of course you do. He says what you believe but can apply the reasoning ability you can't.

If you weren't ignorant you'd have noted a few things. First his sample wasn't apples to apples. He took polls from 21 days out. We are a little less than twice that now. If he had been really interested in debunking the claim he'd have looked at polls at the same point in the cycle. Since he's not I'll do that for him below.

Second he doesn't even address the main criticism about using a turnout model equal to, or greater, than 2008. That addresses not only the large spread between D's and R's in the sample it also deals with the weighting applied to demographic groups. There were record turnouts by African Americans and Hispanics in the last election, well above their historic averages. In a depressed economy with no Hope and Change is it realistic to believe those same numbers will turnout? I would say no. But it's always possible I guess. If that turnout model doesn't hold then the polls are over weighting those demographic groups which are overwhelmingly democratic.

Let's look at key item from Nate's blog. I'm about to employ what they call the thinking skills Tom, so naturally what I say won't be readily understood by you.

Quote: In 2004, Democratic Web sites were convinced that the polls were biased toward George W. Bush, asserting that they showed an implausible gain in the number of voters identifying as Republicans. But in fact, the polls were very near the actual result. Mr. Bush defeated John Kerry by 2.5 percentage points, close to (in fact just slightly better than) the 1- or 2-point lead that he had on average in the final polls.

Let's break it down. Nate references Dems being convinced polls were biased toward GWB. His evidence? A post, written recently, on your favorite site that tells you what to think, the Huffpo. What is Huffpo referencing? A blog written on September 18th, 2004. This essentially serves as Nate's kicking off point. "Dems claimed the same in 2004 and were wrong." But context is always key. Were they really wrong?

The blog it references is over 40 days out from the 2004 election. Why is that important? Because he doesn't reference the polls at the time of the post as being right, he references the ones much closer to the election. The polls being addressed by the blog had Bush up 49% to Kerry's 42%. A full 7% difference. But as Nate acknowledges, Bush won by 2.5. So in fact, the premise Nate is using to begin his defense of polls is flawed. Ergo his entire analysis, as it pertains to the case being made, is also flawed.

Nate argues Dems concerns 40 days out were incorrect because polls 21 days out or on election day showed Bush winning by the margin he did.

Quote:The analysis that follows is quite simple. I’ll be taking a simple average of polls conducted each year in the final 21 days of the campaign and comparing it against the actual results.

Hence the point, he's not comparing apples to apples. If he had really wanted to do a legitimate comparison he'd have looked at the sampling and weighting of polls at least 50 days out and see how they compared to the actual results. But as we can see, he didn't. And the polls that he uses to begin his claims are at least 40 days out but he compares them to polls 21 days out or on election day. He is doing what Maxi likes to do. Take a snapshot of data that tells the story he wants to tell, ignoring all context, and passes it off as the truth. It's why Owl b!tch slaps Maxi on a routine basis because Owl can look at data and form a conclusion. He doesn't have to be told what conclusion to reach because he he doesn't lack the ability to intelligently reach one himself, unlike Maxi.

See Tom this once again highlights your intellectual deficiencies. Nate is making an argument that doesn't compare apples to apples and doesn't address the major criticism being offered. This entire thread has seen you do that. That's the problem when people like you lack critical thinking skills. You go to a source you trust, a liberal site or writer, post their thoughts on the matter and accept them as the gospel. Whereas if you were a remotely intellectually honest, or capable, person you'd examine the claims they are making to see if they are sound. It's the difference between thinking and being told what to think. You are firmly entrenched as someone who falls into the latter.

One interesting thing though. Did you see where Nate acknowledged the polls of registered voters favor Dems? Remember when I said that exact thing a few weeks ago and you told me I didn't know what I was talking about? Well suck it. He in who you trust has confirmed it for you. Now that he's told you what to think I'd look for a retraction on your part, but we both know you don't have the character for that.

(09-30-2012 07:09 AM)Ninerfan1 Wrote: [ -> ]

(09-29-2012 11:48 PM)Redwingtom Wrote: [ -> ]In Nate I trust!

Poll Averages Have No History of Consistent Party Bias

Of course you do. He says what you believe but can apply the reasoning ability you can't.

If you weren't ignorant you'd have noted a few things. First his sample wasn't apples to apples. He took polls from 21 days out. We are a little less than twice that now. If he had been really interested in debunking the claim he'd have looked at polls at the same point in the cycle. Since he's not I'll do that for him below.

Second he doesn't even address the main criticism about using a turnout model equal to, or greater, than 2008. That addresses not only the large spread between D's and R's in the sample it also deals with the weighting applied to demographic groups. There were record turnouts by African Americans and Hispanics in the last election, well above their historic averages. In a depressed economy with no Hope and Change is it realistic to believe those same numbers will turnout? I would say no. But it's always possible I guess. If that turnout model doesn't hold then the polls are over weighting those demographic groups which are overwhelmingly democratic.

Let's look at key item from Nate's blog. I'm about to employ what they call the thinking skills Tom, so naturally what I say won't be readily understood by you.

Quote: In 2004, Democratic Web sites were convinced that the polls were biased toward George W. Bush, asserting that they showed an implausible gain in the number of voters identifying as Republicans. But in fact, the polls were very near the actual result. Mr. Bush defeated John Kerry by 2.5 percentage points, close to (in fact just slightly better than) the 1- or 2-point lead that he had on average in the final polls.

Let's break it down. Nate references Dems being convinced polls were biased toward GWB. His evidence? A post, written recently, on your favorite site that tells you what to think, the Huffpo. What is Huffpo referencing? A blog written on September 18th, 2004. This essentially serves as Nate's kicking off point. "Dems claimed the same in 2004 and were wrong." But context is always key. Were they really wrong?

The blog it references is over 40 days out from the 2004 election. Why is that important? Because he doesn't reference the polls at the time of the post as being right, he references the ones much closer to the election. The polls being addressed by the blog had Bush up 49% to Kerry's 42%. A full 7% difference. But as Nate acknowledges, Bush won by 2.5. So in fact, the premise Nate is using to begin his defense of polls is flawed. Ergo his entire analysis, as it pertains to the case being made, is also flawed.

Nate argues Dems concerns 40 days out were incorrect because polls 21 days out or on election day showed Bush winning by the margin he did.

Quote:The analysis that follows is quite simple. I’ll be taking a simple average of polls conducted each year in the final 21 days of the campaign and comparing it against the actual results.

Hence the point, he's not comparing apples to apples. If he had really wanted to do a legitimate comparison he'd have looked at the sampling and weighting of polls at least 50 days out and see how they compared to the actual results. But as we can see, he didn't. And the polls that he uses to begin his claims are at least 40 days out but he compares them to polls 21 days out or on election day. He is doing what Maxi likes to do. Take a snapshot of data that tells the story he wants to tell, ignoring all context, and passes it off as the truth. It's why Owl b!tch slaps Maxi on a routine basis because Owl can look at data and form a conclusion. He doesn't have to be told what conclusion to reach because he he doesn't lack the ability to intelligently reach one himself, unlike Maxi.

See Tom this once again highlights your intellectual deficiencies. Nate is making an argument that doesn't compare apples to apples and doesn't address the major criticism being offered. This entire thread has seen you do that. That's the problem when people like you lack critical thinking skills. You go to a source you trust, a liberal site or writer, post their thoughts on the matter and accept them as the gospel. Whereas if you were a remotely intellectually honest, or capable, person you'd examine the claims they are making to see if they are sound. It's the difference between thinking and being told what to think. You are firmly entrenched as someone who falls into the latter.

One interesting thing though. Did you see where Nate acknowledged the polls of registered voters favor Dems? Remember when I said that exact thing a few weeks ago and you told me I didn't know what I was talking about? Well suck it. He in who you trust has confirmed it for you. Now that he's told you what to think I'd look for a retraction on your part, but we both know you don't have the character for that.

Bump...Tom, anything?

Nice little rouse once again Niner. But that still doesn’t change the fact that you are wrong about the polls using this supposed 2008 turnout model.

Quote: But pollsters, at least if they are following the industry’s standard guidelines, do not choose how many Democrats, Republicans or independent voters to put into their samples — any more than they choose the number of voters for Mr. Obama or Mitt Romney. Instead, this is determined by the responses of the voters that they reach after calling random numbers from telephone directories or registered voter lists.
Pollsters will re-weight their numbers if the demographics of their sample diverge from Census Bureau data. For instance, it is typically more challenging to get younger voters on the phone, so most pollsters weight their samples by age to remedy this problem.

They simply call the people. Get their results. Then ask if they are R, D or I. If out of every 100 people they get 38 D’s, 29 I’s and 33 R’s, THAT’S THE SAMPLE!

And the article isn’t comparing the two sets of data! He only links the huffpo story to point out that Democrats did the same type of unfounded bitching, at one point in time (40+ days out) 2004. Nothing more!

He then proceeds to conduct his analysis using polls 21 days out. Yes, in a perfect world he could have perhaps used something a little closer. But that doesn’t change the facts of the 21 day out poll analysis. Something tells me that you will still be bitching about this 21 days out anyway!

And sadly, no, you aren’t going to get any retraction from me. I wasn’t implying awhile back that the polls were not reporting more D’s than R’s, I was implying as I proved above, that the polls are not oversampling democrats on purpose.

Although I will give you props for continuing to defend this indefensible position that flies in the face of very reputable polling firm in the country…lol…I think I see now why you aren’t in polling anymore

Conservatives can just ignore facts they don't like. It's actually quite remarkable.

Am I misunderstanding this situation here, or were there not multiple polls by multiple polling entities which showed similar levels of support for each candidate? Meaning that the drop in support for Romney was seen in multiple different samples?

Where's the issue then?

(10-01-2012 11:22 AM)Redwingtom Wrote: [ -> ]Nice little rouse once again Niner. But that still doesn’t change the fact that you are wrong about the polls using this supposed 2008 turnout model.

Um, no.

This:

Quote: But pollsters, at least if they are following the industry’s standard guidelines, do not choose how many Democrats, Republicans or independent voters to put into their samples — any more than they choose the number of voters for Mr. Obama or Mitt Romney. Instead, this is determined by the responses of the voters that they reach after calling random numbers from telephone directories or registered voter lists.
Pollsters will re-weight their numbers if the demographics of their sample diverge from Census Bureau data. For instance, it is typically more challenging to get younger voters on the phone, so most pollsters weight their samples by age to remedy this problem.

Does not mean this:

Quote:They simply call the people. Get their results. Then ask if they are R, D or I. If out of every 100 people they get 38 D’s, 29 I’s and 33 R’s, THAT’S THE SAMPLE!

Pollsters put together models that will hopefully get them a "representative" sample. They continue making calls until they reach that representative sample. That representative sample is based on their assumptions. They determine "likely" voters largely based on...wait for it...turnout models. So they call until they get a representative sample, which means when they get a Dem sample of +7 they stop calling. Because their underlying methodology determines their representative sample. And the best indicator of that is past elections.

Quote:And the article isn’t comparing the two sets of data!

I know it's NOT. The problem is it makes the case that it IS.

Quote:He only links the huffpo story to point out that Democrats did the same type of unfounded bitching, at one point in time (40+ days out) 2004. Nothing more!

Was it unfounded Tom? Did Bush win by 7%? That'd be no. Your latest "brain" Nate is making a case based on two sets of data that are not consistent. If he wanted to make an accurate case he'd have used the same time frame. He didn't. Therefore his methodology flawed, therefore the conclusions he reaches are flawed. This is a complicated subject, one you are not capable of grasping. It's why you post other people's views and make no effort to analyze them.

Quote:He then proceeds to conduct his analysis using polls 21 days out. Yes, in a perfect world he could have perhaps used something a little closer. But that doesn’t change the facts of the 21 day out poll analysis.

Wow, you really are dumb. No one is disputing the facts 21 days out. The dispute is that HE is disputing facts about polls 40 days out with data of polls 21 days out. He is using the latter to bolster is point undercutting the facts of the former. I'm sorry Tom, but it doesn't work that way. This is what we refer to as a straw man. Something you engage in frequently yet appear ignorant of how to recognize them.

Quote:Something tells me that you will still be bitching about this 21 days out anyway!

The number of days out doesn't change the reality of an underlying poll Tom. The methodology is the methodology, no matter how many days out you are. Since you're stupid you don't understand that.

Quote:And sadly, no, you aren’t going to get any retraction from me. I wasn’t implying awhile back that the polls were not reporting more D’s than R’s, I was implying as I proved above, that the polls are not oversampling democrats on purpose.

No you weren't boo boo. I made a specific statement that polls of registered voters favor dems. You said I didn't know what I was talking about. Nice effort to lie and now claim you were saying something different.

Quote:Although I will give you props for continuing to defend this indefensible position that flies in the face of very reputable polling firm in the country…lol…I think I see now why you aren’t in polling anymore

Tom you still haven't grasped that the case the people you read to tell you what to think isn't refuting the specific case being made. Again, it's a straw man on their part. They are refuting a case not being made in this case.

I realize your intellectual limitations preclude you from ever grasping this. It's not your fault really. But the damnable thing is you are acting like you do know what you're talking about when you don't.

(10-01-2012 11:31 AM)UCF08 Wrote: [ -> ]Am I misunderstanding this situation here, or were there not multiple polls by multiple polling entities which showed similar levels of support for each candidate? Meaning that the drop in support for Romney was seen in multiple different samples?

Where's the issue then?

Yes they did. However they were all using samples anywhere from Dem +7 to Dem +10 in their samples.

No one disputes, or at least shouldn't, that Romney lost some support and was losing to Obama. But the polls drive the narrative the media puts out, and that narrative right now appears to be the race is nearly over for Romney when the reality is that's not the case.

When a poll finds Rommey and Obama pulling essentially equal support from each of their parties, and Romney up among independents, the only way you explain Obama leading is the model showing more Dems coming out to vote than Republicans or Independents. And when a poll shows an enthusiasm gap of R's 64% and Dems 48% a reasonable person wonders how an 8 to 10 point lead for Obama can really be logical.

That's the underlying point.

Ninerfan1 Wrote:However they were all using samples anywhere from Dem +7 to Dem +10 in their samples.

Niner, you can keep shouting it as much as you want, but that doesn't change the fact that basically EVERY ******* POLL is showing more Democrats than Republicans. EVEN THE FOX ONES! The country is not 50/50.

You also seem to ignore the FACT that party affiliation is a fluid item. That's why no reputable polling firm uses it as a sound demographic item.