Public Polls vs Campaign's Internal Polls?

[kennetha]

Anybody here ever work with the polling unit of a campaign? If so, maybe you can answer a question for us. Do campaigns do better polling, somehow, than the public polling organizations? You often hear campaigns say that their internal polls are showing something different than the public polls. Perhaps that's often just spin. But it's a constant enough theme in election after election that I often wonder whether campaigns have better, more sophisticated polls than the public polls. It stands to reason that they might. Since they need polls that are reliable enough and informative enough to guide their decisions about where to spend resources, etc.

But I have no idea whether they poll in a different way than the public polling organizations do. Anybody got any clue? If so, enlighten us.

[McCCain4retirment]

1. McCain campaign called my house

I live in a battleground state and just answered their questions like I was a right wing nut job.


When McCain called me(like 8 times since july)

the voice record would just ask who I'm voting for
and if his VP made a difference
and if I was a clinton supporter
and questions like that.

[Occam Bandage]

2. In the only campaign in which I've ever worked closely enough to see the internals,

they were usually more or less the same as the public polling. As for methodology? The questioning is usually the same, but I can't speak for the algorithms used to determine who to call or how to compile the data.

I can say, however, that good polling firms make a decent amount of money providing the best polls possible in a very competitive industry, so I highly doubt they're using a worse methodology.

[jwirr]

3. We used voters lists of reg. Dems when we did it years ago. We

were right - we got beat.

[mb7588a]

4. Brief explanation from how I've seen it done:

Campaigns first target the number of each type of voter they need to win by one voter and then aim at some point above that. Say you need 5,000 votes to be elected to state assembly by one vote. You need 3,000 women, 2,000 men.

Assume that to meet your vote goals, you need 65% of women, and 40% of men (this is made up).

Periodically, campaigns will poll within the areas that they have targeted. If they are meeting their percentage, their internals are on the positive side. Usually internals are interpreted as "up or down/positive or negative" from the target.

It is not as simple as the Gallup type polls with a perfect sample of the entire electorate. Campaigns don't care about the entire population, only those they need to win.

If they are not meeting the target, then they work on persuasion in the targeted audience. If the poll in my scenario showed winning 71% of women, and only 25% of men, I would begin to persuade men to my side.

In the end, it is about turnout as well, so internals can be about intent to vote

You can surmise that there are many many types of internals. However, you can generally consider it as a measure of whether the campaign is meeting its vote targets.

I infer from what I have heard about Obama's internals that they have set their targets extremely high. That's a good thing for us!

[kennetha]

6. Thanks

Though brief, this helps a lot. I can begin to see why campaigns might often say that their internals show something different than the public polls show. They are playing a rather different game.

[BzaDem]

5. My random guess is

that Obama's internal polling weights younger voters more heavily. All polls have to weight by many factors (including age) to reduce the margin of error to acceptable levels, so the question becomes what weightings to use. I think public polling uses conservative weights (especially on age). Many go by the 2004 election. Others use various forms of a likely voter model, which doesn't count younger voters as much. My guess would be that Obama's campaign feels good about being able to turn younger voters out at higher levels than the 2004 election, and is therefore counting them more heavily.

[maseman]

7. Internal polls are much much deeper in questions

perceptions, trends, emotions, etc.

Internal polling is done in areas where the campaign needs to win. So obviously there will be heavy polling in the "swing" states. The polling will be conducted asking some of the basic general questions like who you will vote for, have you made up your mind, would you possibly change your mind, do you typically vote, did you vote in 2000 or 2004, etc.

Then the polling will get much deeper. It will probe for certain qualities and characteristics of the respondent such as are you a gun owner or belong to the NRA? Do you go to church on a regular basis? Do you work full time? Do you have children in the household? What is your highest level of education? What are your thoughts on abortion? What are your thoughts on the current economic crisis? Have you cut back on driving due to gas prices?

One major aspect of internal polling is to do multiple polls with the same questions in the same geography to see trends. That is what they are doing today to see if it is swinging in certain ways with certain groups of people. They will also use the polling data to see if ad campaign are effective. If not they will pull the ads or tweak the message per the polls and focus groups.

Much of what they do is cross-tabbed with the data within the poll. So they may look at Females with children in the household who also work full time and see how that is trending for Obama or McNuts. They may look at football watching, gun owning middle class men to see how that is trending, etc.