Gallup - Bush +3 wo/Nader, +4 w/Nader

[The Vorlon]

A tad higher for Bush than I expected, but still a dead heat given the +/- 4% MOE

http://www.gallup.com/content/?ci=11260



Likely Voters



Registered Voters

[angus]

very briefly:  registered voters vs likely voters?

[agcatter]

statistical tie for sure - better for Bush than I thought given the week we've had in Iraq.

[agcatter]

seems Republicans seem to run 2 - 3 pts better among likely voters.

[The Vorlon]

very briefly:  registered voters vs likely voters?

the 45ish % who rarely vote trend heavily democratic.  The news is bad (iraq) hence those who rarely vote are voting more so for Kerry, if they actually voted.

The "gap" between likely (+3 Bush) and registered (Kerry +2) of 5 points is, historically quite high - usually it's 3 points or so....

[angus]

very briefly:  registered voters vs likely voters?

the 45ish % who rarely vote trend heavily democratic.  The news is bad (iraq) hence those who rarely vote are voting more so for Kerry, if they actually voted.

The "gap" between likely (+3 Bush) and registered (Kerry +2) of 5 points is, historically quite high - usually it's 3 points or so....

how one goes about making the call may require a longer answer.  I suppose it's determined with the initial questioins, "did you vote last time?" "do you usually vote?"  I always wonder if there's a way to determine whether the respondent is being completely honest.

[The Vorlon]

very briefly:  registered voters vs likely voters?

the 45ish % who rarely vote trend heavily democratic.  The news is bad (iraq) hence those who rarely vote are voting more so for Kerry, if they actually voted.

The "gap" between likely (+3 Bush) and registered (Kerry +2) of 5 points is, historically quite high - usually it's 3 points or so....

how one goes about making the call may require a longer answer.  I suppose it's determined with the initial questioins, "did you vote last time?" "do you usually vote?"  I always wonder if there's a way to determine whether the respondent is being completely honest.

3 things to remember when asking people about voting intentions..

1)  They lie
2)  They lie
3)  They lie

The Gallup screen is 13 questions long... Terrance Group's is 12, as is Teeter/Hart.

Most firms use 3 to 5...

Gallup, and the other real firms, ask for actual specific answers..

Are you registered to vote? (yes/no)

Did you vote in the last Presidential Race (yes/no)

For whom did you vote? (get actual name)

Did you vote in the last US Senate Race (yes/no)

For whom did you vote? (get actual name)

Did you vote in the last US House of Representatives race(yes/no)

For whom did you vote? (get actual name)

Did you vote in the last state Governors Race(yes/no)

For whom did you vote? (get actual name)

Do you know where you polling place is? (yes/no)

Where is your polling place is located (get an actual location)

a few more..

How closely paying attention...
How firm is your support...
Not Excited/Somewhat/Very/Extreamly excited about candidate..

......

Another twist at Gallup is that their operators have discretion to classify a "likely" voter based on tone of voice, oveall impression, special circumstance, etc...

Gallups operators make $55K a year BTW versus $6.50 an hour at a lot of firms..

All of the above is part of why Gallup is, well, Gallup, and the others are not...

Hope that helps!

[Fritz]

Vorlon- please see the last 2 pages of the Prediction thread.  I posted a map in there, and then started comparing it to yours, would appreciate your feedback.  Thanks.

[Gustaf]

very briefly:  registered voters vs likely voters?

the 45ish % who rarely vote trend heavily democratic.  The news is bad (iraq) hence those who rarely vote are voting more so for Kerry, if they actually voted.

The "gap" between likely (+3 Bush) and registered (Kerry +2) of 5 points is, historically quite high - usually it's 3 points or so....

how one goes about making the call may require a longer answer.  I suppose it's determined with the initial questioins, "did you vote last time?" "do you usually vote?"  I always wonder if there's a way to determine whether the respondent is being completely honest.

3 things to remember when asking people about voting intentions..

1)  They lie
2)  They lie
3)  They lie

The Gallup screen is 13 questions long... Terrance Group's is 12, as is Teeter/Hart.

Most firms use 3 to 5...

Gallup, and the other real firms, ask for actual specific answers..

Are you registered to vote? (yes/no)

Did you vote in the last Presidential Race (yes/no)

For whom did you vote? (get actual name)

Did you vote in the last US Senate Race (yes/no)

For whom did you vote? (get actual name)

Did you vote in the last US House of Representatives race(yes/no)

For whom did you vote? (get actual name)

Did you vote in the last state Governors Race(yes/no)

For whom did you vote? (get actual name)

Do you know where you polling place is? (yes/no)

Where is your polling place is located (get an actual location)

a few more..

How closely paying attention...
How firm is your support...
Not Excited/Somewhat/Very/Extreamly excited about candidate..

......

Another twist at Gallup is that their operators have discretion to classify a "likely" voter based on tone of voice, oveall impression, special circumstance, etc...

Gallups operators make $55K a year BTW versus $6.50 an hour at a lot of firms..

All of the above is part of why Gallup is, well, Gallup, and the others are not...

Hope that helps!

Vorlon,

I been meaning to ask you whether you think it's possible that due to the ciontroversy surrounding Bush, it is possible that people wh didn't vote in 2000 might do it now, and, of course, whether this might be significant enough to actually affect the race in a close race?

[The Vorlon]

Gustaf:

There are two scenarios:

The GOP base is absolutely dialed in.  I can't see this dying down.

IF the Democratic base stayed dialed in as well, it will be a very close election. likely with a modest edge to Kerry.

If Bush gets up a fairly solid say +5 in the polls, it could very quickly turn into +10 for Bush as the Democratic base is more easily discouraged than the GOP base.  

If that happens I expect a lot of Democrats will stay home or go Nader, in which case Bush could win by 10 or so.

[Gustaf]

Gustaf:

There are two scenarios:

The GOP base is absolutely dialed in.  I can't see this dying down.

IF the Democratic base stayed dialed in as well, it will be a very close election. likely with a modest edge to Kerry.

If Bush gets up a fairly solid say +5 in the polls, it could very quickly turn into +10 for Bush as the Democratic base is more easily discouraged than the GOP base.  

If that happens I expect a lot of Democrats will stay home or go Nader, in which case Bush could win by 10 or so.

OK, sounds reasonable. I am not sure whether you understand my exact question, I was wondering whether the 'likely voter' stats might be skewered some by the above mentioned reasons?

[JNB]

 I used to think that non voters tended to be heavily Democratic, but I have read and even seen to the contrary in many cases. For one, so called Conservative Christians only had a 54% turn out in 2000, and also, younger white males, a group that is small l libertarian leaning has a very poo rturnout as well, far lower than young white females.

   Other notes was that the 94 and 02 mid term elections, elections where the GOP did well, had a bit higher turn out than the 98, 90 and 86 mid terms where the Democrats did well.

[The Vorlon]

Gustaf:

There are two scenarios:

The GOP base is absolutely dialed in.  I can't see this dying down.

IF the Democratic base stayed dialed in as well, it will be a very close election. likely with a modest edge to Kerry.

If Bush gets up a fairly solid say +5 in the polls, it could very quickly turn into +10 for Bush as the Democratic base is more easily discouraged than the GOP base.  

If that happens I expect a lot of Democrats will stay home or go Nader, in which case Bush could win by 10 or so.

OK, sounds reasonable. I am not sure whether you understand my exact question, I was wondering whether the 'likely voter' stats might be skewered some by the above mentioned reasons?

Yes, I did misunderstand your question..

Turnout was about 54.7% in 2000.

Gallup, for example, says things are tracking to about 59% or so this year.

Gallup's methodology is "self adjusting" in the sense they make no assuptions about turnout, but yes the Gallup way of doing things will definitely catch any changes in voter turnout, as will the models of all the better firms.   Terrance group project 61%, Teeter/Hart are saying 58%, so the folks who know what they are doing have things well under control.

A lot of the "second tier" (or worse) firms simply employ some crude weights to compensate for bad fundemental poll design, a lot of these firms are going to get really embarrased this year.

Survey USA seems to be having a very poor start to the season, as is ARG.

I'll bet you a nickle that a ton of College, Newspaper, and TV Station polls are going to be looking utterly absolutely silly after the election.

[Gustaf]

Gustaf:

There are two scenarios:

The GOP base is absolutely dialed in.  I can't see this dying down.

IF the Democratic base stayed dialed in as well, it will be a very close election. likely with a modest edge to Kerry.

If Bush gets up a fairly solid say +5 in the polls, it could very quickly turn into +10 for Bush as the Democratic base is more easily discouraged than the GOP base.  

If that happens I expect a lot of Democrats will stay home or go Nader, in which case Bush could win by 10 or so.

OK, sounds reasonable. I am not sure whether you understand my exact question, I was wondering whether the 'likely voter' stats might be skewered some by the above mentioned reasons?

Yes, I did misunderstand your question..

Turnout was about 53% in 2000.

Gallup, for example, says things are tracking to about 59% or so this year.

Gallup's methodology is "self adjusting" in the sense they make no assuptions about turnout, but yes the Gallup way of doing things will definitely catch any changes in voter turnout, as will the models of all the better firms.   Terrance groups project 61%, Teetr/Hart are say 58%, so the folk who know what they are doing have things well under control.

A lot of the "second tier" (or worse) firms simply employ some crude weights to compensate for bad fundemental poll design, a lot of these firms are going to get really embarased this year.

Survey USA seems to be having a very poor start to the season, as is ARG.

I'll bet you a nickle that a ton of College, Newspaper, and TV Station polls are going to be utterly absolutely silly after the election.

OK, I should have guessed that they'd be prepared...I just figured that since a lot of firms seemed to base their stats on whether people voted previousyly, and I thought some people might not have voted before but were damned sure about doing it now, either b/c they H-A-T-E Bush, or L-O-V-E him.

[Fmr. Gov. NickG]

very briefly:  registered voters vs likely voters?

the 45ish % who rarely vote trend heavily democratic.  The news is bad (iraq) hence those who rarely vote are voting more so for Kerry, if they actually voted.

The "gap" between likely (+3 Bush) and registered (Kerry +2) of 5 points is, historically quite high - usually it's 3 points or so....

how one goes about making the call may require a longer answer.  I suppose it's determined with the initial questioins, "did you vote last time?" "do you usually vote?"  I always wonder if there's a way to determine whether the respondent is being completely honest.

3 things to remember when asking people about voting intentions..

1)  They lie
2)  They lie
3)  They lie

The Gallup screen is 13 questions long... Terrance Group's is 12, as is Teeter/Hart.

Most firms use 3 to 5...

Gallup, and the other real firms, ask for actual specific answers..

Are you registered to vote? (yes/no)

Did you vote in the last Presidential Race (yes/no)

For whom did you vote? (get actual name)

Did you vote in the last US Senate Race (yes/no)

For whom did you vote? (get actual name)

Did you vote in the last US House of Representatives race(yes/no)

For whom did you vote? (get actual name)

Did you vote in the last state Governors Race(yes/no)

For whom did you vote? (get actual name)

Do you know where you polling place is? (yes/no)

Where is your polling place is located (get an actual location)

a few more..

How closely paying attention...
How firm is your support...
Not Excited/Somewhat/Very/Extreamly excited about candidate..

......

Another twist at Gallup is that their operators have discretion to classify a "likely" voter based on tone of voice, oveall impression, special circumstance, etc...

Gallups operators make $55K a year BTW versus $6.50 an hour at a lot of firms..

All of the above is part of why Gallup is, well, Gallup, and the others are not...

Hope that helps!

So does Gallup only survey people who answer yes to all these questions?  There are an awful lot of people who will vote in 2004 who didn't vote in 2002, and who can't name their congressman.  What do they do with all those people?  Or are they just trying to create a "definite voter" sub-category to contrast with the "likely voters"?

[The Vorlon]

So does Gallup only survey people who answer yes to all these questions?  There are an awful lot of people who will vote in 2004 who didn't vote in 2002, and who can't name their congressman.  What do they do with all those people?  Or are they just trying to create a "definite voter" sub-category to contrast with the "likely voters"?

Gosh no...

The reason they ask for specific answers is that it cuts down on the BS factor...

If you can say... I voted for Mr X, my polling station is at 123 anystreet... you are almost certainly a likely voter,..

But Gallup weights in other questions as well.

Almost nobody would answer all the questions correctly...

Gallup assigns all votes a "score" and those above a certain threshold are deemed likely. Takes about 7 out of 13 to be "likely" in Gallup's book.

Gallup (as do others BTW) give operators the ability to make a judgement call as well.

If somebody is adament they will vote and very high in conviction, a Gallup operator would override them into the "likely" catagory

The thing about Rasmussen that worries me is there is no human in the loop.

(the Rasmussen poll is behaving oddly BTW)

Hope that helps!

When it comes to nailing voter turnout, Gallup is as good as it gets.  They have lterally millions of interviews to draw an historic comparison from..

[Gustaf]

So does Gallup only survey people who answer yes to all these questions?  There are an awful lot of people who will vote in 2004 who didn't vote in 2002, and who can't name their congressman.  What do they do with all those people?  Or are they just trying to create a "definite voter" sub-category to contrast with the "likely voters"?

Gosh no...

The reason they ask for specific answers is that it cuts down on the BS factor...

If you can say... I voted for Mr X, my polling station is at 123 anystreet... you are almost certainly a likely voter,..

But Gallup weights in other questions as well.

Almost nobody would answer all the questions correctly...

Gallup assigns all votes a "score" and those above a certain threshold are deemed likely. Takes about 7 out of 13 to be "likely" in Gallup's book.

Gallup (as do others BTW) give operators the ability to make a judgement call as well.

If somebody is adament they will vote and very high in conviction, a Gallup operator would override them into the "likely" catagory

The thing about Rasmussen that worries me is there is no human in the loop.

(the Rasmussen poll is behaving oddly BTW)

Hope that helps!

When it comes to nailing voter turnout, Gallup is as good as it gets.  They have lterally millions of interviews to draw an historic comparison from..

That was kind of what I was getting at... Too bad, the hopes of hidden Deaniacs is then compltetely gone.

[classical liberal]

If the Dems had nominated Edwards, they would have won easily.  But they nominated Kerry, so they will have to endure the fact that having such a dull candidate doesn't inspire record turnout, although I think that we'll all be surprised when the election rolls around and turnout is in the mid to upper sixties.

[zachman]

If the Dems had nominated Edwards, they would have won easily.  But they nominated Kerry, so they will have to endure the fact that having such a dull candidate doesn't inspire record turnout, although I think that we'll all be surprised when the election rolls around and turnout is in the mid to upper sixties.

Dean might have been doing well with the war in Iraq as it currently is. I think he could have pulled off some surprising magic.

[muon2]

If the Dems had nominated Edwards, they would have won easily.  But they nominated Kerry, so they will have to endure the fact that having such a dull candidate doesn't inspire record turnout, although I think that we'll all be surprised when the election rolls around and turnout is in the mid to upper sixties.

There hasn't been a presidential turnout over 55.2% in voting age popualtion since 1968. That corresponds to the reduction of the voting age to 18. Following the work of M. McDonald at GMU, if one compensates for ineligible voters (noncitizens and felons) and overseas eligible voters every presidential election in the last 30 years has had between 52 and 61% turnout of the eligible population. There is noting that indicates a change from this long-standing trend.