Polling has been a hot topic around here...

[Deadgnome]

I'm an avid reader of Democratic Underground, and have been for years. I may not post often and I don't reply too much, but I appreciate what is discussed in this forum and the way people go about debate. I've noticed recently that there have been many topics about polling, specifically mentioning Gallup seeing how they have been mentioned in the news. I, myself have worked for Gallup for a little over a year now doing interviews and I have tried to offer some of my insights into how things go on at this company. Granted, I in a low level position (being and interviewer peon and all) but I do get to notice some trends I see and I hope I can offer a more in-depth perspective on how things work around here.

What I wanted to bring to light here today was something very interesting I experienced here last night while doing our final election poll.

First, there has been much to be made about how the cell phone only households are vastly under-represented, which is entirely true. But last night I saw Gallup doing something that I have never seen before. We were actually screening for cell-phone only households. While we do call cell-phones in our daily tracking polls, we never screen for cell-phone only. The reason that we can't do so many, or so I've been told by management here, is because the ways the laws are set up we can't put those numbers on an automatic speed dialer. What that dialer does is weed out the bad numbers, disconnects, and mis-dials by working ahead of the interviewer 20 or so calls and dumping the good numbers to the interviewer for a higher contact rate. Apparently when it comes to cell-phones we are only allowed to hand dial.

Last night though, I did 20 surveys on the USA Today upcoming election poll. Of those 20, I did 10 surveys on cell-phone only. And each one of those surveys I made sure to thank them specifically for doing the poll on their cell phone and discussing the changes in our habits when it comes to cell-phone usage versus land line usage. Every one of those people were extremely grateful we were starting to do this.

Second, we obviously know that we have been inundated with this constructed narrative from the Right about how they are going to wipe the floor on the 2nd. The polls have supposedly shown this, including Gallup's, which have been very skewed from my perspective and from the perspective of many others. I wanted to say though, and this is simply from my experience last night (the first night of the USA Today Upcoming Election Poll) I talked to 20 respondents. Of those 20, I only polled 2 people that were planning to vote Republican, the rest, Democrat. That was 18-2 in favor of the Democrats and these were people from all over the country, North, South, East, and West.

Now, the other interesting thing I saw, and this is when I asked about the Tea Party, and the question stated: Do you or do you not support the Tea Party? I received 20 respondents saying the did NOT. I even had one of the Republican respondents and a handful of Democratic respondents expound upon their utter distaste for the group.

So, make what you will of what I have mentioned here, I thought it would be an interesting insight, seeing what is going on in this country. And if you have any questions, please ask them, I would love to discuss what I do and what I see from my position here.

Good day.

[sufrommich]

1. Thanks for that info from the inside. Very interesting. nt

[johnnie]

2. Good info

Thanks

[Cal33]

3. Great information. Do you mind if I pass it on to some, who I know

are feeling depressed because of their low expectations of the Dems. winning?

[Deadgnome]

7. Feel free

I just figured this would be some interesting information to some because I see a lot of talk about the polling companies here.

[Cal33]

12. Many thanks. Since 90% of the news media is Republican-owned,

one never knows if one is hearing the truth or not.

[county worker]

4. What bull shit! Screening for cell phone only households?

You have never studied statistics have you? Samples have to be randomly selected to be statistically accurate! Screening for cell phone households is not random selection!

[RaleighNCDUer]

5. This is deliberate screening for cell-only to redress the imbalance created

by the laws that prohibit autodialing to cells. It is the only way to get a genuine cross-sample.

IOW, 'random' is not really random, but is skewed toward a particular population.

[Deadgnome]

6. How do you figure?

How would you ever even be able to know that you are calling a cell phone only household if you didn't screen for that? I'm assuming from what I see at this company we are operating on a sort of systematic sampling, which I guess is a form of probability sampling. From what I know of statistics if the starting point of the sample is randomized, and I know Gallup has done things like that before, like a survey we take in Iraq, where you talk to every 3rd house on the block or whatever, the survey can be somewhat accurate. Now, this method of survey taking, I am assuming, has all sorts of issues, and I'm not fluent enough in statistical gathering and interpretation to really go more in depth than that. I'm merely saying what this company is doing.

What I can tell you is we do call households that are land line only, we call some that are both cell phone and land line, and apparently they are screening for cell phone only as well. We screen for the person with the most recent birthday and over age 18 in the household, and sometimes we screen for males or females that have the most recent birthday.

I'm just passing on what we are doing, I'm not the one deriving the statistics, compiling, etc. I simply do the interviews so the methodology isn't something that is on me, it is on the company itself. Every polling company

[Gormy Cuss]

8. Here's how it may have been done.

I can't comment directly on how Gallup conducted this survey,but here's a standard approach to identify cell-only households.

Using a sampling frame of likely cellphone numbers, with blocks of numbers randomly selected and assigned, interviewers begin dialing. The first question after the introduction would be a screening question to determine if the cell phone user was part of a cell-only household.

Using RDD sample with screeners is a very old and well established method.


eta to clarify.

[Deadgnome]

9. Yes

These surveys are RDD surveys, granted they have to play with the automated dialer for law purposes. I know for fact that all of the GIK, USA Today, and GPSS studies are done on an RDD server. And that is exactly how we do that, after I read the intro and state the survey will be recorded for internal quality assurance we have the screening question, be it the 18 years old, etc. or the cell-phone only on this new poll. It might help to note we don't screen for recent birthday on the cell-phone surveys on confirm that we are talking to someone of voting age, so 18 plus.

[cilla4progress]

10. If this bears out --

11/2 will be a wonderful, wonderful night!

[CoffeeCat]

11. I'm such a pessimist these day...

...but it's because of the corporatist elites in this country who need angry, unhealthy and divided
citizens--to pull of their heist of our democracy.

I don't know what is going on, but they are up to something. An election is a big event and with
big events come big opportunities to cause additional damage.

I am very skeptical of these coordinated media memes that predict (directly and indirectly) a Republican
slaughter. I watched as a CNN reporter asked, "How big of a majority will the Republicans win next week?".
We've all seen the Newsweek cover with "Speaker Bohner". The collective meme is that Republicans WILL win.

Again, I know I'm a pessimist, but I'm wondering if the kingmakers are trying to cultivate further
division and hatred--by making the public at large suspicion and mistrustful of ANY Dems gains--or God
forbid, significant wins!

They've got everyone believing that the Republicans have it in the bag. There's just no basis for this
meme. There are polls that show Dems gaining, and I personally believe that the Tea Party is detested
by everyone except the radical 20 percent of the Republican party that belongs to it. I've never met
a Dem who does not have a visceral reaction to the tea party and I know many Republicans who hate it
too. The tea party appears to have a faction of the Republicans on board, but this wide-sweeping
popularity seems to be hot air.

I wondered if the "Republican slaughter" meme was being spread to justify Republican wins that would be
gained due to election fraud. While I'm sure there will be election fraud, I am starting to think that
the real reason for the meme is to spread further hate, division--and even violence in our country.

[Deadgnome]

14. I'm with you

I share the same exact sentiments about this "Slaughter" meme. I see it as stoking civil unrest, for what ends I do not know, but it certainly seems that this divisive rhetoric that the public has been subjected to since Mr. Obama has taken office is pushing toward violence.

As far as the Tea Party is concerned, all I see are a lot of loud folks, a small population at that, that is simply getting attention because they are loud. My wife and I were talking last night and I made a comment that I wouldn't doubt that this is a concerted effot in some part, by the Klan to push their belief set into the maintstream. Now, it wouldn't be the whole of that party, but some of it stinks of Klan influence and it was an idea that just popped into my head for some reason. Obviously the Klan can't be effective if they are appearing as they normally do, hooded, etc., but what's to stop them from seeing this loosely organized movement as a way to exert some of their influence. I guess take that idea for what it is worth, it was just something the cropped up in my thoughts.

[Beetwasher]

13. Gallup's Problem Is Their Likely Voter Screen, Which Is A Giant Pile Of Garbage

Meant to skew their polls in favor of Republicans. If they blow these numbers they can blame it on a faulty screen, and point to their poll of registered voters and say "See, we ARE accurate, we just have to tweak our screen!".

[Deadgnome]

16. We don't screen for likely voters

We screen for either the male, 18 years of age or older with the most recent birthday; the female, 18 years of age or older with the most recent birthday, or just the person in the household 18 years of age or older with the most recent birthday... I do all of these political polls and have never seen a likely voter screen, it is a question we ask, but it has no bearing on the screening process.

I am aware of the report that came out recently about this likely voter poll you're probably mentioning and honestly I have no idea how they got the numbers they did there.

[Radical Activist]

18. That screening happens later.

Once they have the raw data they make a lot of other assumptions about who's likely to vote and also make sure their sample is statistically representative of the population. Many very subjective assumptions are made before they calculate the final product that's reported in the news.

[Beetwasher]

19. You Don't, But The Data Collected Goes Through A Screen

at some point that determines who is likely to vote and it's that number that they push.

[Deadgnome]

20. Thanks for to both posts above to clarify that op

On that end I have no idea how it functions because I don't have access to that part of the process. From where I stand I will say I don't trust what we do, just looking for discussion on this.

[Radical Activist]

15. This is very distrubing if they only did this on their final election poll.

Pollsters often justify their work by pointing to the accuracy of their final poll before the election. What you're suggesting is that they're doing something for the final poll to make it more accurate that they didn't do for earlier polls. In other words, they're doing one very good poll that will be used to lend credibility to the shitty polls they usually do the rest of the year.

I wonder if there's a way to find if they're doing this nationally or if this is a common practice in previous elections. If so, it would bring their entire operation into question.

[Deadgnome]

17. That is exactly what I was thinking

When I made this post. I found it odd that we were doing this NOW. Like I mentioned previously I have only been here a year so I have no idea how this worked during the last election cycle, but I do know it has only been a year or so since we first starting hitting any cell phones at all because of the laws.

The poll is a National study, and I see regions pop up on each poll I do. One of the things I was wondering about this and I have since I started was why we were running 2010 election studiees almost 10 months prior to the election, long before the primaries were even held and there was not necessarily any identifiable candidates.