Which Poll to Believe?

AP Newswire Research | Oct. 18, 2004 | PKAJJ

[pkajj]

Which Poll to Believe?

With the current presidential polls in some instances being all over the map, I thought it might prove useful to take a quick review of the various pollsters’ performances for the 2000 presidential election (based on their final published poll prior to election day).

The actual final results for the 2000 race were Gore 48.4, Bush 47.9, and Nader 2.7.

Here are the pollsters final predictions for 2000:

ABC News – Bush 48, Gore 45, Nader 3

CNN/USA Today / Gallup - Bush 48, Gore 46, Nader 4

MSNBC/Zogby - Gore 48, Bush 46, Nader 5

Voter.com / Battleground – Bush 50, Gore 45, Nader 4

CBS News – Gore 45, Bush 44, Nader 4

Pew – Bush 46, Gore 43, Nader 3

TIPP – Bush 48, Gore 46, Nader 4

Harris – Bush 47, Gore 47, Nader 5

Fox News – Bush 43, Gore 43, Nader 3

Marist – Bush 49, Gore 44, Nader 2

NBC / WSJ – Bush 45, Gore 43, Nader 5

Wash. Post – Bush 48, Gore 45, Nader 3

Note that Zogby came closest to the actual result. At the time, he attributed his success to his working right up until the last minute, and thus detected the last minute Gore surge. His late work also fully accounted for the fallout from the last minute revelation of Bush’s DUI charge from 20 years before. Besides Zogby, only CBS showed a Gore lead.

[KJacob]

I do not think how they did in 2000 is any indication of how they will do in 2004. Karl Rove's 72-hour GOTV campaign could be the difference. Look at the 2002 elections. Then Rove used a prototype of his strategy and as a result pollsters underestimated republicans by nearly 8% on average. Since then he has modified his strategy. Many pollsters today are using models from 2000 and ignoring 2002.

[jmstein7]

An interesting point.

[Always Right]

CBS, Fox and Harris were actually more accurate in showing the margin between the candidates. Zogby showed Gore winning by 2 points, when Gore actaully only won the popular vote by 0.5%. So those showing a tie or a 1 point Gore lead were actually more accurate.

[Ingtar]

Whichever one that shows the worst case scenario for the President. That way we will all work hard enough to hopefully overcome the Dim vote fraud.

[Busywhiskers]

The ballots for Oregon arrived in the mail on Saturday, (here in Oregon, we don't vote like normal people). My wife, oldest son and I sat down last night and voted.

Exit polls from our balloting indicated a landslide victory for President Bush. If the results are extrapolated to the whole state, I predict that President Bush will win Oregon with somewhere around 100% of the vote.

Which poll should we believe. How about the certified results of the one taken by election officials on November 2. It's the only one that counts.

[watsonfellow]

Zogby did not come the closest, he underestimated Bush by almost two full points....Fox News came closest in that they assessed the race at a tie.

[AFPhys]

Many of the major pollsters did not release poll figures taken during/after the DUI release ... Zogby did. That's the only reason his figures were "more accurate".

[Josh in PA]

I'm going to be releasing my final CAMEC, Computer Analyis Model of the Electoral College, Monday night, November 1st!!! You can search back through FR for "CAMEC" for previous results

The final results will be complete with prediction of actual vote for each state!!

Stay Tuned!!!

[finnman69]

Keep in mind that the best performing pollster in 2002 was Mason Dixon and the worst was Zogby. 2004 will be much more similar to 2002 than 2000.

[neutrality]

Fox also had undecided at 10% (Buchanan 1%) which is laughable

[Jambe]

It is interesting that almost every polster in 2000 had Gore down by 2 points or more than the final figure.

How was this explained? A Gore surge? Caused by what?

[taxcontrol]

To my mind, what is disconcerting to my quick reveiw is that these polls appeared to get the Bush vote very close - the majority being less than 1% off, but in many cases UNDER estimated the Gore vote.

If there is no change in their weighting, it is quite possible that these polls are not giving Kerry a point or two.....

Food for thought.

That means that every and I mean EVERY Republican needs to make sure that all eligable Republican voters gets to the polls. Here is what you can do:

1) Take a day of vacation from work!
2) Vote early either by absentee or take advantage of early voting.
3) Clean up the car/van and call the local campain HQ and offer to give people rides to the polls
4) Offer to work the call lists to remind Republicans to vote
5) Go door to door in your neighborhood encouraging people to vote for President Bush. You don't have to be fancy, just tell them why you support President Bush.

Got to get out the vote folks. This one could be quite close. This is NOT the time to sit back and expect the vote to happen our way.

[Always Right]

What about Zogby with Nader at 5%, which doubled what he actually got. Buchanan did get .42% of the vote, which is close to .5% which is the roundup point for 1%

[eureka!]

"2004 will be much more similar to 2002 than 2000."

Agreed. The reason: the GOP now has a ground game. And bigger than in 2002....

[Only1choice____Freedom]

Moulder from X-Files:

"I'm left wondering which lie to believe."

[Congressman Billybob]

Notice that in all the final polls for the 2000 election, they had the Nader percentage wrong, by as much as 100% (showing his admittedly small number as twice what it became). I've worked on many third party campaigns, including most notably Gene McCarthy (independent, 1976) and John Anderson (independent, 1980).

I've been writing for months the following black-letter statement: Third party and independent candidates ALWAYS shrink between the last poll and the actual votes cast. The reason is that all but the hardest hard-core supporters really do want their votes to HAVE AN EFFECT ON THE OUTCOME OF THE ELECTION.

Therefore, expect the Nader percentage in this election to drop to two-thirds down to half of what the final polls say. And expect many but not all of those few votes to go to Kerry. (A very small number of these will simply get discouraged, and not vote at all.)

Therefore, the conservative way of approaching the polls in the last few days before the election is to overcount the Nader effect. Add all of the Nader numbers to the Kerry numbers. That will overstate the Kerry position by about 1%. Bush should be ahead of that combined number, and that means there is a sufficient "fraud pad" as well.

Congressman Billybob

Latest column, "Mein Fuhrer, I Can Valk!"

[Semper Paratus]

At this point I look at trends in the polls rather than the numbers themselves and right now the trend is towards the President.

[madison46]

Zogby was the worst in 2002. Mason-Dixon was the best then

[Illinois Rep]

I think it's safe to assume the polling is absolutley OFF and will be. The only poll that is accurate is the final election poll when VOTERS show up to the polls and their votes are tabulated.

We know one thing...in voting...there is a 100% total. All votes will equal 100%. You never get that in polling. Bush + Kerry + Undecideds+ others never seem to equal 100.

Also watch polls where one day Candidate A is up 8 points and a week later Candidate B is up 10 points. The electorate is NOT all over the map like that. There will be no wild swings like that...if they are..discount EVERY poll that pollster has and will do...Just throw it in the garbage.

I for one think Rasmussen is the most accurate. He has shown a consistenly close race. After the conventions...no real significant bounce...NO major movement after the first debate...and has shown Bush consistently 2-4 points ahead of Kerry. I'd assume that is where the election is today. If you think Bush is going to win with 10% cushion then he will get 400+ EV and that isn't happening much to our dismay. A result of Bush 52 Kerry 48 would have a nice EV cushion of around 310 for Bush....that is where I think the race is at.