More proof telehpone polls are not sciencitific

[dude58677]

This is a scientific journal article:

http://poq.oxfordjournals.org/cgi/c...INDEX=0&sortspec=relevance&resourcetype=HWCIT

http://www.csicop.org/doubtandabout/polling/

 

[kylejack]

That was due to different wording of a question. The questions asked about the election are not unfairly biased agaisnt Ron Paul. I believe the Ron Paul poll survey results to be mostly correct and we're deluding ourselves by trying to tear them down. Rather, we should focus ourselves on getting the message out and not pretend that the online polls represent the public at-large.

 

[dude58677]

Zogby and Gallup make millions off their research and their clients are the news media and government. There is a huge conflict of interest. There is no way a News/Gallup/Zogby opinion poll will say that Ron Paul is the leader.

 

[joenaab]

The truth lies in between

kylejack, i agree with you in part. the phone surveys are accurate for what they are, but do not represent accurately Ron Paul's true support.

These surveys only include:

1. People who have listed phone numbers (less than 50% of americans).
2. People who don't hang up on telemarketers (under 5% of Americans).
3. People who don't hang up, but still refuse to participate in a political survey.
4. People who don't screen their calls (dialing machines hang up on answering machines in less than one second).
5. People who still have LAN lines (i.e. many people now only use their cell phone)
6. In the event that a qualified database is used, only people who are in this database (eg. registered republicans who vote in primaries).
7. A very, very small sample group, usually in the range of 500 people, which has a margin of error of 5-7% (i.e. Ron Paul's 2% could actually be 9%).

Adding this all up, phone polling, especially in the case of Ron Paul, is highly suspect.

And I speak from having more than ten years of experience in and around call centers and predictive dialing technology. Hangups and answering machines alone total in excess of 95% of dial attempts. These people, by their actiions, are likely to be skewed toward privacy advocates and, hence, Ron Paul.

 

[Bradley in DC]

Yes, telephone poll science has been constantly evolving since Gallup made his name pointing out his competitors were not making adjustments for the socio-economic factors involved (only wealthier people had phones at the time). Most of the polls at this point reflect the fact that few people, statistically, know who Dr. Paul is. That will change with the debates, as we get closer to the primaries and more people start to pay attention and after Dr. Paul starts running ads (get your contributions in!). Doing what we can with bumper stickers, yard signs, blog postings, letters to the editor of local papers, etc., all help.

 

[dude58677]

"These people, by their actiions, are likely to be skewed toward privacy advocates and, hence, Ron Paul."

Bingo! That's it in a nutshell.

 

[kylejack]

kylejack, i agree with you in part. the phone surveys are accurate for what they are, but do not represent accurately Ron Paul's true support.

These surveys only include:

1. People who have listed phone numbers (less than 50% of americans).

False. Having worked at Gallup, I can tell you that they're using a random dialer and I often spoke with people who were unlisted. One time I got an internal secret number at the CIA. They wanted to know how I got the number, heh.

2. People who don't hang up on telemarketers (under 5% of Americans).
3. People who don't hang up, but still refuse to participate in a political survey.
4. People who don't screen their calls (dialing machines hang up on answering machines in less than one second).

Perhaps, but I'm not sure it can be demonstrated that these people are more likely to support Ron Paul. Indeed, I think Ron Paul voters are more politically aware and would love to participate in a survey to endorse Ron Paul.

5. People who still have LAN lines (i.e. many people now only use their cell phone)

This point has some truth to it, so long as the cell phone users are not living in a house where someone has a land line.

6. In the event that a qualified database is used, only people who are in this database (eg. registered republicans who vote in primaries).

Gallup dials and then uses pre-qualifying questions. For these recent ones, the question has asked if the person is likely to vote in the Republican primary rather than asking if they're a registered Republican.

7. A very, very small sample group, usually in the range of 500 people, which has a margin of error of 5-7% (i.e. Ron Paul's 2% could actually be 9%).

More like 4% usually, but the consistency of the results with Ron at 1-4% indicates that its fairly accurate.

And I speak from having more than ten years of experience in and around call centers and predictive dialing technology. Hangups and answering machines alone total in excess of 95% of dial attempts. These people, by their actiions, are likely to be skewed toward privacy advocates and, hence, Ron Paul.

I'm not sure I accept that conclusion.

 

[dude58677]

"I'm not sure I accept that conclusion."

Then give a reason.

 

[Bradley in DC]

Look, guys, Dr. Paul is confirmed in the June 5th and August 5th debates where poor poll numbers might have kept him out. The only polls that matter are the ones with ballots attached that chose the delegates to the nominating conventions.

 

[joenaab]

Ah, Bradley, let us have some fun anyway! (I read that Menger article yesterday, great stuff, thanks).

Kylejack, I wasn't referring to any one specific polling technique. For example, if they are calling random numbers, then, yes, some of the points I made will change. But Zogby is often employed to call a qualified database of, for example, registered republicans, and hence do not use a randomizer. It depends, as you know, on the campaign (dialing campaign, not political campaign).

Even in a randomized campaign, I would expect that people with unlisted phone numbers would hang up more frequently than those with listed numbers, thus skewing the results.

My call center experience was never in taking political surveys, just to be clear.