The Accuracy (or not) of primary polling:

[PINN4CL3]

Does primary polling accurately reflect public opinion? One 20 year veteran pollster has reason to believe it doesn't. While historically, general election polling can be quite accurate, primary polling is not so much, and it bothers me the emphasis placed on it by the mainstream media. They treat the numbers as if they are infallible, when in fact they are not.

Here is an article I stumbled upon while researching the topic. Note especially the part when talking about the primary election of 1984:


In a nationwide poll that CBS News and The New York Times conducted just before the 1984 New Hampshire Democratic primary, Walter Mondale held what CBS News and The Times characterized as the largest lead ever seen in national polls in the race for a nomination. Fifty-seven percent of Democratic primary voters in that poll chose Mondale, and only 7 percent chose Gary Hart. However, the day after that poll was reported, Gary Hart beat Mondale handily in the New Hampshire primary.

 

[alexlcameron]

Interesting... I think the polling methods need to be revisited to reflect technological trends ie cell phones

 

[a_european]

You may disagree with me, but i think polls are usually bullshit. I call such things "vodoo science".

 

[PINN4CL3]

Interesting... I think the polling methods need to be revisited to reflect technological trends ie cell phones

I read another article where Gallup has actually taken this into consideration. They polled people with cell phone only , with land & cell, and with land only to compare the results. At the time, the results were so similar that I'm not sure if they've continued the trend.

I think the biggest disconnect right now is that the big pollsters are polling those "most likely to vote" in a primary. What does this mean? Who are these people, and how are they determined? If these are strictly year 2000 Bushies, it could explain alot. Futher, there is a HUGE amount of uncertainty among republican voters. There is a larger portion of uncertain voters than there are for any one political candidate.

What this says is that people may have a guy in mind, but are open to a message. With the money Ron has raised over the last year or so, it's very possible alot of these voters could flip. Ron Paul has the most powerful message of all the candidates imo. It just needs to get out.

 

[PINN4CL3]

You may disagree with me, but i think polls are usually bullshit. I call such things "vodoo science".

That may very well be true. In General Elections though, political scientists feel they have it on lockdown - able to predict outcomes 2 months in advance. This may very well be the case, but the methods are different. In a general you can poll HUGE portions of the populations, and ask "Candidate A?" or "Candidate B?".

Primaries are just not that accurate, and have historically been WAY off the mark.

 

[dspectre]

There's nothing wrong with the polls themselves. It's what people do with them and how they interpret them is the problem.

Doing the math for a poll isn't that hard. It's usually just basic algebra. For some reason though, it is revered as science. Which we know now and days science = magic.

The poll is only as good as the data used. Even with that data, you have to look at the results to see what they mean. This is where all the propaganda comes in. The results are meaningful, but they have to be taken in context.

Without making a more rigid and rational argument(yeah I have things to do), if the polls are so accurate, why even have an election? Just take a poll and call it a done deal.

 

[dircha]

Polling is a statistical science.

Polls have a confidence interval and a margin of error.

Polls are not perfect, but they are the most effective means we have of rapidly sampling public opinion short of a comprehensive survey of every member of the population.

The fact that polls are not perfect, and do not purport to be, is precisely why citing anecdotal accounts of poll inaccuracy means very little.

Studies have been done on the effects of excluding households having a cell phone but no landline phone. They are a small percentage of the population. When comparing the combined results vs the landline only results, on most issues the differences are statistically insignificant.

There are some differences. Households with a cell phone only tend to be younger, tend to reflect more liberal stances on social issues, and tend to be significantly less likely to follow politics and to vote.

I think it's very unlikely many of them would be considered likely Republican voters even if polled. I think its likely that the majority of them, like the majority of all Americans, still haven't heard of Ron Paul and don't know where he stands on the issues.

It would be nice to know, yes, but I assure you, it isn't the difference between 5% and 25%. Ron Paul has a very long way to go to have any chance at winning even a single state.

 

[PINN4CL3]

Polling is a statistical science.

Polls have a confidence interval and a margin of error.

Polls are not perfect, but they are the most effective means we have of rapidly sampling public opinion short of a comprehensive survey of every member of the population.

The fact that polls are not perfect, and do not purport to be, is precisely why citing anecdotal accounts of poll inaccuracy means very little.

Studies have been done on the effects of excluding households having a cell phone but no landline phone. They are a small percentage of the population. When comparing the combined results vs the landline only results, on most issues the differences are statistically insignificant.

There are some differences. Households with a cell phone only tend to be younger, tend to reflect more liberal stances on social issues, and tend to be significantly less likely to follow politics and to vote.

I think it's very unlikely many of them would be considered likely Republican voters even if polled. I think its likely that the majority of them, like the majority of all Americans, still haven't heard of Ron Paul and don't know where he stands on the issues.

It would be nice to know, yes, but I assure you, it isn't the difference between 5% and 25%. Ron Paul has a very long way to go to have any chance at winning even a single state.

Nevertheless I still don't understand then the disconnect between his online, meetup, and grassroots support and his polling. I can name you 10 people off the top of my head who are voting in primary for Ron Paul, and not a single one of them has been polled.

So who are these samples? They are not random. They are aimed at those who someone feels are most likely to vote. I would agree with nearly everything you said when it comes to general election polling, but I think there is far too large a discrepancy when it comes to primaries.

So imo, they should not be using this as the infallible truth in media coverage, and I would not be surprised at all if Ron Paul even at his current numbers wins an early state.

 

[Alawn]

I have heard they base it on who voted in the last republican primary. Nobody voted in the Republican primary last time because Bush was the incumbent. Only people with really strong feelings would have bothered. The sample would be very biased.

 

[arkitekt]

There are two aspects to consider when looking at the accuracy of a poll. First is the methods used in polling, and second is the group which conducted the poll.

I cannot speak all that well upon the first subject. My knowledge of polling has come from hearsay sources, and while they are sources I trust, hearsay is always something to be considered with caution.

As for the second, never trust any - and I mean NEVER TRUST ANY - poll done by an organization which stands to gain or lose greatly as a result of the event it is polling. As quite a few polls are done by the media, or funded by media syndicates, I put nearly no stock in them.

Check out my blog below for an essay on the mainstream media and why they have been censoring Ron Paul, it ties in directly with how the MSM as a whole is a special interest group with HUGE amounts of money riding on this election. None of that conspiracy theory crap, I make all of my arguments from a sound basis using my knowledge of economics and journalism.

 

[Chickensoup]

I only respect polls that put the exact question asked with them because ultimately thats the most telling part of the poll. You can get a poll to say virtually anything semi-respectable depanding on what question you ask, how you say it, when you say it, and how you verbalize it (different inflections during the question).