Final and extensive analysis by Gallup

[Almighty Colin]

http://gallup.com/poll/content/default.aspx?ci=13873&pg=1

PRINCETON, NJ -- Two days before the election, the final CNN/USA Today/Gallup poll shows a dead heat in the presidential race, with President George W. Bush and Sen. John Kerry each receiving 49% support among likely voters in the final allocated estimate. Independent candidate Ralph Nader garners 1% of the vote, and all other candidates an additional 1%.

The poll was conducted Oct. 29-31 among 2,014 national adults and includes 1,573 likely voters, and was weighted to reflect an estimated voter turnout of 60%. The final numbers also reflect Gallup's judgment of how undecided voters will cast their ballots.

Before allocation of the undecided vote, Gallup's likely voter model shows Bush ahead by two points, 49% to 47%, while the results among all registered voters show Kerry with a two-point lead, 48% to 46%.

The poll also shows that among "high interest" voters -- all Americans who express a high verbal commitment to voting, regardless of whether they have actually voted in previous elections -- Kerry enjoys a two-point lead, 49% to 47%.

In its traditional likely voter model, Gallup screens out older people if their past voting performance does not reinforce their stated intentions to vote. (Younger people who could not have voted in 2000 are included in the likely voter model, based solely on the intensity of their expressed commitment.) Some observers have suggested that in this election year, the intensity of the public's interest will stimulate a large number of older people to vote for the first time. If that is the case, then the "high interest" voter model could come closer to the final outcome than the traditional likely voter model.

Among all registered voters, 13% say they will be voting in a presidential election for the first time. Among Gallup's "likely voters," just 7% say they will be first-time voters.

Still, all of the differences in support between Kerry and Bush -- among traditional likely voters, all registered voters, and the new category of "high interest" voters -- are well within the poll's margin of error, confirming the general conclusion that the presidential race is too close to call.

The allocation of undecided voters is part of the tradition started in 1936 by Dr. George Gallup, who wanted to provide the public with the pollster's best estimate of what the data indicate. This year, the allocation of the undecided vote is based on Gallup's experience in previous presidential elections, showing that in election contests with an incumbent, virtually all of the undecided vote among likely voters will break for the challenger(s). Thus, in this case, with 3% undecided, 2% is allocated to Kerry and 1% to the Nader/other group, resulting in the estimated tie.

One in 11 Voters Could "Swing"

The poll finds that 9% of likely voters can be classified as potentially "swing" voters -- people whose commitment to a major candidate is not firm. Among all likely voters, 91% say their votes for one of the two major candidates are "certain" -- 46% for Bush and 45% for Kerry. The 9% who do not express a firm intention to vote for either major candidate include 3% who support Bush but could change their minds, 2% who support Kerry and could change their minds, and 4% who indicate no preference for either major candidate.

MORE: http://gallup.com/poll/content/default.aspx?ci=13873&pg=1

[DrGuile]

Isnt 2000 individuals a REALLY small population sample? How accuratly can it represents the different groups in the USA... (what's the US population now? 300mil?)

I know you can get pretty accurate with a moderate sized sample in modern statistical analysis, but that seems a bit of a stretch... eh, we'll see...

[Mike AI]

Things do not look good for Bush right now...... mostly due to the huge voter registration campaigns, which I fear is going to lead to thousands of fraudulent votes in swing states.

[Winetalk.com]

Originally posted by Mike AI@Nov 1 2004, 11:11 AM
Things do not look good for Bush right now...... mostly due to the huge voter registration campaigns, which I fear is going to lead to thousands of fraudulent votes in swing states.
Mike,
regardless who wins this election,
in such a close race we ALL have to live with:
The NEXT USA President is elected mostly by...
out of their mind senior citizens who don't even remember the next day whom they have voted for
;-))

This is whom I'll blame if my candidate loses
;-)))

[Almighty Colin]

Originally posted by DrGuile@Nov 1 2004, 11:10 AM
Isnt 2000 individuals a REALLY small population sample? How accuratly can it represents the different groups in the USA... (what's the US population now? 300mil?)

I know you can get pretty accurate with a moderate sized sample in modern statistical analysis, but that seems a bit of a stretch... eh, we'll see...
Remarkably, the sampling error decreases very rapidly with the size of the sample. It is proportional to the square root of the sample size.

A truly random sample of 2000 will represent a population of 300 million to within the square root of 2000 or to within 2.2%.

In this case ...

1 standard deviation means the sample represents the total population to within 2.2% 67% of the time.

2 standard deviations means the sample represents the population to within 4.4% 95% of the time.

3 standard deviations means the sample represents the population to within 6.6% 99.7% of the time.

[Mike AI]

Originally posted by Serge_Oprano+Nov 1 2004, 11:16 AM--></div><table border='0' align='center' width='95%' cellpadding='3' cellspacing='1'><tr><td>QUOTE (Serge_Oprano @ Nov 1 2004, 11:16 AM)</td></tr><tr><td id='QUOTE'> <!--QuoteBegin-Mike AI@Nov 1 2004, 11:11 AM
Things do not look good for Bush right now...... mostly due to the huge voter registration campaigns, which I fear is going to lead to thousands of fraudulent votes in swing states.
Mike,
regardless who wins this election,
in such a close race we ALL have to live with:
The NEXT USA President is elected mostly by...
out of their mind senior citizens who don't even remember the next day whom they have voted for
;-))

This is whom I'll blame if my candidate loses
;-))) [/b][/quote]


Serge, you pretty much described Kerry supporters!

I will support whoever is elected, even if its a victory by one vote. Many people hate Nixon, but he showed remarkable leadership skills when he let Kennedy slide into office and did not pull a "Gore".

The last thing we need is a Constitutional crisis or some long protracted court battles to determine the President.

[Evil Chris]

I agree with Mike... It won't be a doomsday no matter who gets elected. But it should be a clear election, and not another Supreme Court decision.

It will be enjoyable to watch it all unfold tomorrow.

[DrGuile]

Originally posted by Colin+Nov 1 2004, 11:26 AM--></div><table border='0' align='center' width='95%' cellpadding='3' cellspacing='1'><tr><td>QUOTE (Colin @ Nov 1 2004, 11:26 AM)</td></tr><tr><td id='QUOTE'> <!--QuoteBegin-DrGuile@Nov 1 2004, 11:10 AM
Isnt 2000 individuals a REALLY small population sample? How accuratly can it represents the different groups in the USA... (what's the US population now? 300mil?)

I know you can get pretty accurate with a moderate sized sample in modern statistical analysis, but that seems a bit of a stretch... eh, we'll see...
Remarkably, the sampling error decreases very rapidly with the size of the sample. It is proportional to the square root of the sample size.

A truly random sample of 2000 will represent a population of 300 million to within the square root of 2000 or to within 2.2%.
[/b][/quote]
Ive forgotten most of my stats classes ;)


but how truly random can this sample really be?

I mean, it certainly doesnt include people who dont answer polls on the phone... or who do not have phones..

is the 2000 divided in a representative % of age groups, ethnies, geographical locations, religions, education level....

etc...

theoritically, within 2.2%, 2/3 of the time sounds good, but 2% is probably what will decide your election...

[Almighty Colin]

Originally posted by DrGuile+Nov 1 2004, 12:23 PM--></div><table border='0' align='center' width='95%' cellpadding='3' cellspacing='1'><tr><td>QUOTE (DrGuile @ Nov 1 2004, 12:23 PM)</td></tr><tr><td id='QUOTE'> Originally posted by Colin@Nov 1 2004, 11:26 AM
<!--QuoteBegin-DrGuile@Nov 1 2004, 11:10 AM
Isnt 2000 individuals a REALLY small population sample? How accuratly can it represents the different groups in the USA... (what's the US population now? 300mil?)

I know you can get pretty accurate with a moderate sized sample in modern statistical analysis, but that seems a bit of a stretch... eh, we'll see...
Remarkably, the sampling error decreases very rapidly with the size of the sample. It is proportional to the square root of the sample size.

A truly random sample of 2000 will represent a population of 300 million to within the square root of 2000 or to within 2.2%.

Ive forgotten most of my stats classes ;)


but how truly random can this sample really be?

I mean, it certainly doesnt include people who dont answer polls on the phone... or who do not have phones..

is the 2000 divided in a representative % of age groups, ethnies, geographical locations, religions, education level....

etc...

theoritically, within 2.2%, 2/3 of the time sounds good, but 2% is probably what will decide your election... [/b][/quote]
It's not truly random, of course. There are methods by which the various polling companies use to compensate for the shortcomings of their methodology. That makes for a good debate in and of itself. For example, will 60% of the population really vote? (they seem to think so). Have the polls done a good job of including the new voters? (they have tried).

I think the overall record of the polls is pretty good - about the error one would expect over time.

Agreed on your last sentence. In this case the difference is within the polling error no matter how you look at it. Most years it is not so close. Gallup has called the election correctly in all but three elections since World War II with the biggest error being the infamous Truman/Dewey election which prompted changes in methodology. The other two were elections decided by 2% or less.

[RawAlex]

Realistically, they need to sample about 500 - 1000 people PER state, and break up the electoral college that way. There is no way that 2000 people sampled total would be enough to represent all of the american people and their individual powers to vote.

With the electoral college system, the total votes (percentage) is a mostly meaningless number. Heck, Gore had more votes than Bush, but Bush has the connections and ended up with the job.

Sampling each of the states would give a MUCH better indication of who is going to get elected, not who is going to get the most votes. Gallup is producing a poll that gives people the mistaken impression that each vote counts (and it doesn't!).

Bush is toast. The question is "lightly done" or "burnt".

Alex

[Almighty Colin]

Originally posted by RawAlex@Nov 1 2004, 01:13 PM
Realistically, they need to sample about 500 - 1000 people PER state, and break up the electoral college that way. There is no way that 2000 people sampled total would be enough to represent all of the american people and their individual powers to vote.

With the electoral college system, the total votes (percentage) is a mostly meaningless number. Heck, Gore had more votes than Bush, but Bush has the connections and ended up with the job.

Sampling each of the states would give a MUCH better indication of who is going to get elected, not who is going to get the most votes. Gallup is producing a poll that gives people the mistaken impression that each vote counts (and it doesn't!).

Bush is toast. The question is "lightly done" or "burnt".

Alex
Alex, I have no idea why you think a poll of the total popular vote is confusing everyone. Only you seem confused. Gallup does both a popular vote sample AND a state by state sample.

In the state-by-state polling they survey well more than the 500-1000 number per state that you mentioned - about 1300 per state. Results? Same as everywhere else. Too close to call. You can find the data on their site.

That Gallup data above is just what it says. It is a sample of the popular vote. How is that meaningless? It's meant to be exactly what it is. Not a prediction of who will win the election but a sample of the popular vote. If you want state-by-state polls they are on the Gallup site and their sample sizes are larger than you stated would be needed.

[PornoDoggy]

Pugh Foundation had it too close to call today, according to NPR - something like a 1 point lead overall for Kerry, a 3+ lead in the so-called "most likely to vote" for W.

Wait .... wait .... my psychic abilities are kicking into high gear.

THEY are going to try to keep US from voting.

Certain pigment enriched groups will accuse "THE MAN" of trying to prevent them from voting - and in some cases they will be right.

Certain individuals with crimson between their shoulders and their heads will barf back up the "fried chicken and free cigarettes" routine - and in some cases they will be right.

I can't quite discern whether the dead Democrats of Cook County will outweigh the dead Republicans of downstate Illinois.

Lawyers will be featured on television.

The moron who gets .00000004% more than the other guy will talk about his mandate.

The other side will be guaranteed 4 years of howling at the indignaties dumped upon them.

[Almighty Colin]

PD, that was :pearl: worthy and all true too.

[RawAlex]

Colin, it is confusing for the general public because they get a simplistic number (48% to 46%! 52% to 41% ) jammed at them daily, but that number does not actually reflect who will end up in the whitehouse, rather a popular vote trend, which feeds into the electoral college. The electoral college is where the real magic occurs, and it is rarely talked about in detail with these polls - they just give "a number".

Because the vote gets broken up 50+ ways the effective error rate in any single state is very high (2000 people spread over 50 states, that is 40 people per state average, which is statistically too low of a sample to be respresentitive, especially in states like California, New York, and Florida). They would have to sample something like 1000 people in EACH of those states to get numbers that even begin to express the voter direction in each state, which would then give a much clearer picture of the overall electoral college (and therefore election) situation.

There is a site posted, http://www.electoral-vote.com/pred/index.html - I don't agree with their methods, but this to me is a better indication of WHO will end up in the whitehouse.

The straight 49%-48% suggest that each american gets an even vote, and we know that just isn't true. Overvotes in one state do not diminish or take down over votes in another. So measuring it overall isn't a true indication of the outcome either (Ask Gore).

Alex

[wig]

Obviously, it comes down to who is connected. :lol:

[Almighty Colin]

Originally posted by RawAlex@Nov 1 2004, 03:00 PM
They would have to sample something like 1000 people in EACH of those states to get numbers that even begin to express the voter direction in each state, which would then give a much clearer picture of the overall electoral college (and therefore election) situation.
Alex, pay attention this time. Gallup ALSO does state-by-state polls and they do poll more than 1000 in EACH state.

[Almighty Colin]

Originally posted by RawAlex@Nov 1 2004, 03:00 PM
There is a site posted, http://www.electoral-vote.com/pred/index.html - I don't agree with their methods, but this to me is a better indication of WHO will end up in the whitehouse.

Yes, there are many sites like this and they all have different results because the polls are so close in the swing states. RealClearPolitics has Bush 227, Kerry 207. They use an average of many state-by-state polls. Rasmussen has Bush 222, Kerry 186. They conduct their own polls. The site you mentioned also uses an average of polls including the Rasmussen poll (Kerry 298, Bush 231 there). Now what do you conclude?

[RawAlex]

Colin, pay attention: I know Gallup does a state by state. BRAVO!

What I am saying is that because the media wants "sound bites" rather than detail, the single poll numbers from the overall poll are put out there as "the truth". Viewers and readers then take that SINGLE set of numbers and judge the condition of the race (and potentially adjust their vote) based on that single number.

Overall polling of such small groups of people might be a good indicator of what the total vote will look like, but does nothing to explain who will end up in the whitehouse.

As for the different sites, well, many of them appear to have a particular tilt or leaning in the way the place the undecided / unknowns. As an eample, the Electoral-vote site gives the undecided to Kerry at 2:1. Not entirely sure why the heck they do that, but there ya go. That lack of transparancy on many of these sites makes them hard to take seriously.

I have been interested that many of these sites have shown the same trend in the last 7 to 10 days, with Kerry catching up to or passing Bush and pulling ahead by EC, without there being a major shift in the popular vote polls. I think of this as a clear indication of the fact that the EC gives more power to some votes, and less to others.

Alex

[Almighty Colin]

Originally posted by RawAlex@Nov 1 2004, 07:35 PM
As an eample, the Electoral-vote site gives the undecided to Kerry at 2:1. Not entirely sure why the heck they do that, but there ya go. That lack of transparancy on many of these sites makes them hard to take seriously.

I have been interested that many of these sites have shown the same trend in the last 7 to 10 days, with Kerry catching up to or passing Bush and pulling ahead by EC, without there being a major shift in the popular vote polls. I think of this as a clear indication of the fact that the EC gives more power to some votes, and less to others.

Alex
EC is all over the place. They just uses the latest poll regardless of where it came from.
So they might use a poll showing an 5% advantage for Kerry in a state and then a new poll from Survey USA will come out showing a 4% advantage to Bush in the same state and the result swings.

EC's current electoral predictor is now ... Kerry 262 Bush 261 because of just such a thing.

BTW, Everyone uses that same 2/3 factor. Traditionally, 2/3 of the undecided voters in the last days will vote for the challenger.