So if Obama wins, that's democracy at work. If McCain wins, the fix is in, right?

[KrazyHorse]

Every campaign guy I have spoken too says that internal polling is significantly more accurate that public news polls.

[Darius871]

Originally posted by Lefty Scaevola
A poll does not publish the average of all its results. It filters them first by camparing the demogrphics of the sample with the PREDICTED demographics of the turnout.

Are you suggesting that internal polls don't do this? If they do, wouldn't they suffer from the same inaccuracy?

[Boris Godunov]

Originally posted by Lefty Scaevola
Every campaign guy I have spoken too says that internal polling is significantly more accurate that public news polls.

As KH said:

As for the error comesin, (and oviously there is a lot, just by comapring the large variance in polls measring the same thing) is the weighing of the sample. A poll does not publish the average of all its results. It filters them first by camparing the demogrphics of the sample with the PREDICTED demographics of the turnout. Eg, every poll you see published of O vs M is modified by the polster's prediction of how many 20-25 years left handed urban Circassain professionals, etc, are going to vote. EG if the % of x responding to their poll does match the % of x that their predict to be voting, they adjust the weight of the x response in their poll. So if their prediction of x voting is incorrect, their poll is incorrect. The predictions come from other data in the polls and other polls and a lot of guesswork.

Yeah, I know how polling works.

But the point is internal polling is done via the same overall methods as public polling.

[KrazyHorse]

There is no particular reason to believe that internal polls are systematically better than public polls. In fact, since they are done by partisans they may be systematically worse. They probably focus in better on crucial states and demographics, so are more precise in these groups, but for headline national numbers are not particularly favoured.

[Lefty Scaevola]

No reason to be seen from the outside, never the less, the campaigns mostly appear convinced they have the better numbers.

[KrazyHorse]

I'm not surprised that when you ask people in a campaign they think that they're doing a better job than everybody else.

[Darius871]

Methodology aside, is there any reason to think that they at least draw from larger samples, considering they have more money to spare?

[KrazyHorse]

Originally posted by Darius871
Methodology aside, is there any reason to think that they at least draw from larger samples, considering they have more money to spare?

The better public polls have big enough samples that they've reached a real level of quickly diminishing returns. The campaigns probably burn more money on polling more often rather than bigger samples.

(Most polls have big enough samples that the systematics are at least as important as the statistics).

[Lefty Scaevola]

I rather expect they have smaller samples, but are banking on their poltical experience and knowlege of what the electorate is. Like Krazy Horse says, it may just be egotism about their skill and value.

[KrazyHorse]

are banking on their poltical experience and knowlege of what the electorate is

This is precisely where it's easy to start fooling yourself into biasing the polls to tell you what you'd like to hear.

[-Jrabbit]

Pollsters know what results will keep them employed.

[Darius871]

Originally posted by KrazyHorse
This is precisely where it's easy to start fooling yourself into biasing the polls to tell you what you'd like to hear.

Couldn't it just as easily be said that pollsters would tend to bias polls toward a less rosy picture, to make sure nobody in the campaign starts getting lazy?

[KrazyHorse]

Originally posted by Darius871


Couldn't it just as easily be said that pollsters would tend to bias polls toward a less rosy picture, to make sure nobody in the campaign starts getting lazy?

That's not what generally happens. People tend to have an OPTIMISM bias, not the opposite (this is well documented).

[tinyp3nis]

Originally posted by Aeson
The fix is in both ways

[Darius871]

Originally posted by KrazyHorse
That's not what generally happens. People tend to have an OPTIMISM bias, not the opposite (this is well documented).

People be stupid.