Idea: More accurate polls makes it harder to steal elections?

[drfresh]

Hi guys. Some chap recommended that I post this here, originally my post was in reply to this message.

OK, so Zogby and the other pollsters use a sample size of something around 1,000 people, giving their polls a margin of error of 3-5% ... note that in the last election Bush supposedly got 51% and Kerry 48%, a 3% difference.

As long as polls have big enough margins of error, discrepancies between the election results and poll results can be blown off, making it easier to steal elections. But if we could commission our own legit poll with a larger sample size, like 10,000 people, then the MOE would be more like 1% ... making it harder to explain the differences between poll & vote counts.

Now I'm not suggesting DU start their own polling operation, but rather maybe we could make up the cost difference for Gallup/Zogby/whoever to poll 10,000 instead of 1,000 people.

Could this be a tool to help prevent Diebold vote fraud? Any thoughts on this? Thanks.

[melody]

1. sampling size

I think the problem would be taking a sufficient enough sample size to provide for all the data, to create a margin of errorless poll.

[SheilaT]

2. Nice idea,

but I think you might be a tad, I dunno, naive?

Exit polls are simply the most reliable polls out there, short of actually counting every vote. Only now most people are convinced that exit polls aren't that good.

They will continue to steal the elections, regardless of quality of polls, so long as they have black box voting with which to do it. Until we go to a system of simple paper ballots, counted in front of observers from all political parties, I may not bother to vote again.

[drfresh]

3. Yep, paper ballots are a must

My knowledge of stats is from an introductory college course. How/why are exit polls more reliable than any other ones?

I agree we need paper ballots. E-voting has got to go. Either way though, I think you should bother to vote :)

[OnTheOtherHand]

9. in response to your question

It isn't really a stats issue. The main argument for exit polls being more reliable is that they are conducted as people have just voted, so there is (loosely speaking) no argument over whether you have found "likely voters" or whether they have made up their minds. (The sample size is also larger, certainly if you aggregate all the state exit polls.)

On the other hand (gotta be true to my screen name), when some of the exit poll interviewers are college students standing next to MoveOn activists several hundred feet from the polling place, it isn't hard to imagine that some bias could creep in to a greater extent than in telephone surveys.

In short, both pre-election telephone polls and election day exit polls have significant potential sources of non-sampling error, no matter what the sample size is.

[Der Blaue Engel]

4. I agree with everything you say, except not bothering to vote

Even if this country becomes such a complete fascist state that they can stand in the polling place and say, "You will not win. We're in control of the counting and no matter how you vote, you lose," I will still vote as an act of defiance. They can tear it up in front of me and laugh if they want to, but I will exercise my right to vote.

[kster]

5. I won't vote, But I will

damn sure locate the Central tabulater in my state, AND DO EVERYTHING IN MY POWER TO SECURE THE VOTE COUNT!

[SheilaT]

6. I live in Kansas.

It's as red as a state can be. So long as we have the stupid electoral college thing my vote doesn't count. Kerry could not be bothered to campaign in this state last year, and in fact Democratic volunteers crossed the state line and went over to Missouri to campaign for him, taking valuable resources away from local candidates (like myself) who needed all the help we could get. And I didn't see anyone coming from Missouri over to Kansas to help out any candidates here.

As long as my vote is wasted or it's not going to be counted my only means of protesting is to not vote. Last year I was talked out of not voting by the argument that it still somehow mattered in the larger picture, but it didn't. The election was still stolen. They can steal it the next time without my vote to slow them down.

[Land Shark]

11. OK, but there's more than just winning and losing

there's also the matter of how much someone wins by, the strength of the "mandate" and whether minor parties are deemed to have any support at all or not.

In most cases, as long as people are talking and communicating, only relatively close elections can be stolen. Probably more than 10 points and it becomes unbelievable to most people, so long as Osama doesn't appear on television right before the election.

VOting is also a chance to register one's protest to election officials to the technology or whatever issues there may be. It's a chance to talk about it with neighbors, you have a common experience to strike up a conversation. You can vote for write-ins when you have no real choice, and see if it shows up in the published totals or not. There's lots of reasons to vote that go beyond whether or not you will personally cast the decisive vote. This is the same reason many Dem voters don't vote: they believe even in the absence of fraud that the hassle to make the vote is more than the chance it will matter.

[Land Shark]

10. That's a good policy. I like it. Nothing they do takes away our vote.

That also just happens to be the philosophy of the Declaration of Independence and unalienable rights.

[sunshinekathy]

7. Mitofsky's MOE was 1% according to its own published report

Accurate polls do nothing to make it harder to steal elections - unless the candidates stop stupidly conceding when they've obviously won the elections with a mere glance at the data (at least for those of us like me to whom numbers speak).

[kster]

8. I don't understand

please embarrass me.

[Land Shark]

12. Accuracy is in the eye of the beholder

As pollsters start adjusting the weights in their polls based on mythical reluctant bush responders, the polls themselves get more and more biased, because they only have value to their purchasers as PREDICTORS of elections. If there's a hand on the scales of election justice providing a little extra weight on one side, they need to be able to predict that too.

These come in the following form: assumptions about what % of the population is Dem or Rep, male or female, etc. GAllup was using heavily pro-R assumptions last year, and MoveOn called them on it. I don't know how close Gallup came in its predictions of the election, but those with pro-R weights on their polling are the predictive "geniuses" who now are the witchdoctors of politics until the next time around in 2008

I guess I'd want to know how large the sample size needs to be before the weights to adjust for imperfections in the sample can essentially be omitted? Then you might have an accurate poll. Otherwise, there are important assumptions that go into it.