Saturday, August 19, 2006

How to know when its a push poll and what you can do

The talk today around the Arizona blogosphere has to do with 'push-polling.' Push polling, which is certainly unethical, though legal is the process by which a person (usually either a known activist, a consistent voter or a member of some identifiable 'swing' group) is called and asked to participate in a short 'poll.' It usually takes about twenty minutes, and at first the person is asked a number of questions that relate to a particular race. Then at the end of the push poll, they are asked some 'questions' that are very damaging to one of the candidates (i.e. 'would you be more or less likely to vote for candidate X if you knew....')

There are several reasons for these polls. One is to gauge the relative strength of candidates as perceived by voters, through 'test' marketing negative messages, to actually bringing these smear tactics right into a voter's home via what sounds like an innocent 'poll.'. Sometimes it is to test the power of a negative message (this being the reason why sometimes known activists, such as myself, are called-- they figure if something will weaken our faith in the candidate then it will certainly score points with marginal or undecided voters.) Often many, many thousands of people are called, while in fact it is rare that any legitimate poll would ever need a sample size anywhere near this big. It is a new way to run a smear campaign, pure and simple.

And today, voters down in CD 8 have all been discussing push polls. It seems that a push poll targetting Patty Weiss, one of the two front runners for the Democratic nomination, was run, and at first the assumption was that it had been paid for by the campaign of Gabrille Giffords, the other front runner. But now, it appears to have been commissioned by the National Republican Congressional Committee. Why would the NRCC go to those lengths to influence a Democratic primary? Well, first they are apparently trying to cause a split in the ranks. Secondly, they obviously have a 'favorite' candidate (though we won't know yet whether it is Weiss or Giffords.)

I would like to tell you how to identify push polls. First of all, unlike a real sample, their objective is not randomness or accuracy, it is a campaign tool. As such, if they are targetting you, then they will call your household and ask for you by name. There is your difference. Push polls, in order to hit the most likely voters, or at least the most persuadable voters, start with a list of names first. A real poll would never do this.

Second, once someone has asked for you by name and it turns out to be a pollster, so that you know that it is a push-poll, pay very close attention and get the name of the polling organization that they represent. The law says that they have to provide this information. Then search for information on the company. For example, I got a push-poll call on the A.G. race, in which a company wanted to smear Terry Goddard. So when they identified themselves as 'FMA,' I went online. I found that 'FMA' is a Republican polling service based in Virgnia (Fabrizio, McLaughlin and Associates. And that is what to do. Learn who is sponsoring the poll, and then trace it back.


Anonymous said...

Hmmmm.... that's interesting....

Anonymous said...

Ok- First things first- you my friend are a moron. NO push poll in the HISTORY of the world would take 20 minutes- what you are referencing is a real good old fashioned poll. The point of a poll (I'll use small words so you understand) is to test messages and attacks not just against generic candidates but also against specific
Let me be more plain-you ask questions that mimic the arguments and attacks that you think will be beneficial to you, and vice versa i.e. you test negative attacks against your opponent as well as negative attacks against yourself. You test positives on your opponent and positives on yourself. This also is frequently preceded by a head to head question up front and followed by what is known as a reballot or an additional head-to-head at the end, so that the various arguments can then be analyzed for effectiveness.
A push poll is always under 5 minutes, is usually under 3 and can be under 2. The point is to call a person, say you are conducting a survey and then ask really nasty questions like- how would you feel about candidate x if you knew he was an alcoholic? Then they get off the phone. They won't tell you who they are calling on behalf of.
Learn some damn politics before you decide to post.

Anonymous said...

I'm skeptical about push polls taking 20 minutes, Eli, since most surveys only contain ~20 questions which can be gone through in about 5-8 minutes tops. You'll lose your audience if the poll goes longer. I remember once agreeing to participate in a non-political survey which would take "just a few minutes" but dragged on for ~20 minutes, and about 2/3 od the way through I simply started making up answers to get the survey over with. If they got screwy results as a result I could care less, since they lied about how long the survey would take.

Interesting point about asking for you by name - that *would* be a dead giveaway that something's not right with this "random" survey. However, failure to ask for you by name isn't a guarantee of legitimacy. The push-pollers could have correlated their own list of targeted voters and phone numbers and could proceed on the presumption that the adult they reach is in the targeted group.

Also, push polls don't have to be aimed at targeted groups. You could have a push poll call 1000 households, regardless of ideology, and spread the nasty innuendos, then let word of mouth spread about the rumors of an alcoholic, wife-beater, or whatever. This type of thing happened frequently in 2004 as the campaigns were really targeting undecided and soft voters.

Tom & Icy said...

That's like those cable news shows where the guy gave his view of an issue then told us to go vote on a poll at his website. Then at the end of the show the poll showed that most of the people agreed with his view. I just thought, of course, if someone didn't agree, then they would just turn the channel and ignore his website. I had turned the channel until near the end of the show when I turned back to it to see what the results were just out of curiosity.

Anonymous said...

An additional point...interviewers conducting legitimate surveys sometimes ask for a specific member of a household to participate. Likely voter surveys, for example, draw a sample of respondents from voter registration roles. Its a useful technique for determining the current state of election dynamics.

shrimplate said...

"anonymous" is such a helpful person.

Personally, I am overcome with gratitude or his observances.