Party affiliation

[Dr.House]

Really? So there are always the exact same number of Dems and Reps in the US?

Other than you, who said that?

If you suggest that an equal number of Dems and Reps should be sampled, you are suggesting that there are the same number of both. I can't be held accountable if you don't understand the impact of your own words.

I'm helping you out with what is considered over sampling... This seems to be your issue and you're just not getting it... Not my problem, though...

The goal of the poll is to find out how the election would turn out were the election held at that point in time... If you are polling registered voters, then you are asking people who may or may not show up... Not all registered voters vote... You want a venn diagram to help you out? This is why RV polls are "meh"...

You keep changing the subject. We're discussing the party affiliation of those polled, not their likelihood to vote

Then why poll if you don't care how they will vote?

You are losing it, boy...

Any RANDOM sample of RVs that contains more than one party is considered over sampled for that party...

No, that's a fundamental misunderstanding of statistics.

Not true for RANDOM polls where the demographics are unknown...

You're floundering now...

Let's say suddenly wake up and see the light in your mind. Therefore, millions of people drop the Dems and register as Reps. In fact, a later census confirms that 90% of people are now Republicans.

You do a poll about voter intent, and in doing so you don't random sample - instead you weight the responses so that 50% are Dems and 50% are Reps.

Would that give you a fair representation?

You're talking about a different more expensive poll than just a random sample of RVs...

You're all over the place now...

No wonder you're confused...

 

[8537]

Registered Democrats still dominate the political playing field with more than 42 million voters, compared to 30 million Republicans and 24 million independents. But Democrats have lost the most — 1.7 million, or 3.9%, from 2008.


The numbers themselves shows to make it fair you have to sample MORE dems than Rs

We have literally dozens of polls about party affiliation in recent months - from Fox to Pew to Gallup to CNN and the WSJ.

In virtually every case, we end up with a breakdown that shows a 2 to 12 point lead for Dems. In virtually every case, the 1000+ person surveys have overlapping MoE's. Yet Republicans want us to believe that a fair sample should include the same number of Reps as Dems.

Either these folks failed basic statistics or they are frauds.

 

[commonsensenow]

going back to the original question ... who gives a shit? .... if things continue on as they're headed, both houses of cards collapse under the weight of the debt anyway ...

 

[8537]

Other than you, who said that?

If you suggest that an equal number of Dems and Reps should be sampled, you are suggesting that there are the same number of both. I can't be held accountable if you don't understand the impact of your own words.

I'm helping you out with what is considered over sampling... This seems to be your issue and you're just not getting it... Not my problem, though...

No, you're continuing to pass false information. If a Population is 73% white and 27% black, a proper sample of the population would contain 73 whites and 27 blacks - not 50 of each.

Not true for RANDOM polls where the demographics are unknown...

You're floundering now.

..
The demographics are known because the poll asks.

You're talking about a different more expensive poll than just a random sample of RVs...

No, I'm not. That's how polls are done, and it's why they ask party affiliation.

 

[Dr.House]

If you want a true representation on how people will vote, likely voters makes more sense..

If you're just interested in making your party look better than it is, then registered voters is the way to go...

 

[MeBelle]

Suppose you wanted to find out what the breakdown of Dems, Reps and I's in the US is right now.

You can't afford a census, so that method is out.

How would you go about determining the most accurate possible breakdown?

<I'll be back in an hour to find out what we've come up with...>

You conduct a random phone sampling using your local phone book....

 

[8537]

If you want a true representation on how people will vote, likely voters makes more sense..

If you're just interested in making your party look better than it is, then registered voters is the way to go...

That has nothing to do with party affiliation data.

 

[MeBelle]

If you want a true representation on how people will vote, likely voters makes more sense..

If<snip> registered voters is the way to go...

Agree-there is a difference.

 

[Dr.House]

If you want a true representation on how people will vote, likely voters makes more sense..

If<snip> registered voters is the way to go...

Agree-there is a difference.

Party affiliation in a sample of RVs only tells us how skewed it's going to be for one or the other, as the important "will you vote" is not known...

 

[Dr.House]

If you want a true representation on how people will vote, likely voters makes more sense..

If you're just interested in making your party look better than it is, then registered voters is the way to go...

That has nothing to do with party affiliation data.

party affiliation in a sample of RVs is "meh"...

 

[8537]

If you want a true representation on how people will vote, likely voters makes more sense..

If you're just interested in making your party look better than it is, then registered voters is the way to go...

That has nothing to do with party affiliation data.

party affiliation in a sample of RVs is "meh"...

Then why do you continuously claim that the sample should be just as many Reps as Dems?

 

[Truthmatters]

Registered Democrats still dominate the political playing field with more than 42 million voters, compared to 30 million Republicans and 24 million independents. But Democrats have lost the most — 1.7 million, or 3.9%, from 2008.


The numbers themselves shows to make it fair you have to sample MORE dems than Rs

We have literally dozens of polls about party affiliation in recent months - from Fox to Pew to Gallup to CNN and the WSJ.

In virtually every case, we end up with a breakdown that shows a 2 to 12 point lead for Dems. In virtually every case, the 1000+ person surveys have overlapping MoE's. Yet Republicans want us to believe that a fair sample should include the same number of Reps as Dems.

Either these folks failed basic statistics or they are frauds.

they dont want everyone to vote and work to keep democrats from the polls in any way they can.

The court is full of the records of it.

They dont care about democracy

 

[MeBelle]

If you want a true representation on how people will vote, likely voters makes more sense..

If<snip> registered voters is the way to go...

Agree-there is a difference.

Party affiliation in a sample of RVs only tells us how skewed it's going to be for one or the other, as the important "will you vote" is not known...

^^^^ that and the history of registered voters who actually voted.

 

[Dr.House]

That has nothing to do with party affiliation data.

party affiliation in a sample of RVs is "meh"...

Then why do you continuously claim that the sample should be just as many Reps as Dems?

Why not?

Without that you see a bias one way or the other way...

In an even split, you don't have that bias... Makes sense when you're only dealing with RVs....

 

[emptystep]

I am not sure I am following but perhaps I am. From the link I pasted there were 55M Republicans, 72M Democrats, and 42M Independants/others Nov 2nd (I assume 2008). If the numbers are the same and a poll asks 100 people and ~40 are Republicans and ~60 are Democrat and ~30 are Independant than the poll has not oversampled any particular group because the proprotions are accrurate. Is that what you are asking/saying? What you want to know is what are the numbers currently?

 

[Leweman]

Voter registration isn't a great way to look at things. In many of the southern states people are registered democrat but are very conservative and haven't voted for a Dem in forever. They are Dems in name only. Pew estimates that number to be around 15 percent of registered voters. If these Dems were asked how they lean in a poll many are going to say strongly Repub even if they are registered Dems.

 

[8537]

party affiliation in a sample of RVs is "meh"...

Then why do you continuously claim that the sample should be just as many Reps as Dems?

Why not?

Because it doesn't reflect the actual breakdown. It biases the poll towards whoever has a smaller number of members.

 

[8537]

I am not sure I am following but perhaps I am. From the link I pasted there were 55M Republicans, 72M Democrats, and 42M Independants/others Nov 2nd (I assume 2008). If the numbers are the same and a poll asks 100 people and ~40 are Republicans and ~60 are Democrat and ~30 are Independant than the poll has not oversampled any particular group because the proprotions are accrurate. Is that what you are asking/saying? What you want to know is what are the numbers currently?

Well, if we used that 2008 data then a poll would sample approximately as you say - certainly not half and half as others are suggesting.

If we want to compensate for changes over the past four years, we'd do a statistically significant, scientifically administered poll asking people which party they are in. Then we'd use that as our basis.

 

[Dr.House]

Then why do you continuously claim that the sample should be just as many Reps as Dems?

Why not?

Without that you see a bias one way or the other way...

In an even split, you don't have that bias... Makes sense when you're only dealing with RVs....

Because it doesn't reflect the actual breakdown. It biases the poll towards whoever has a smaller number of members.

My full post explains.... You're just being a dickhead now....

 

[8537]

Why not?

Without that you see a bias one way or the other way...

In an even split, you don't have that bias... Makes sense when you're only dealing with RVs....

Because it doesn't reflect the actual breakdown. It biases the poll towards whoever has a smaller number of members.

My full post explains.... You're just being a dickhead now....

I read you full post. It remains wrong. The poll sample should be an accurate reflection of the population - not a 50-50 breakdown.

The people who do polls realize this and that's why they were almost all within the MoE in 2008.