How "off" have polls been in the past?

[WilmywoodNCparalegal]

I am looking for data to see what the difference is between polls and actual election results to see if there is any consistency whereby I could say that, because polls are usually off by a * percentage or number, then I can safely predict that today's polls are off by * percentages or numbers...

Do I make any sense?

[WilmywoodNCparalegal]

1. shameless self kick!

:kick:

[DemocratSinceBirth]

2. Most Pre Election Final Polls Since Truman Have Been Within The Margin Of

Error...


Some polling artifacts....


Many polls in 92 and 96 overestimated Clinton's percentage of the vote...

[pmbryant]

3. going on memory

In the week before the 2000 election, most polls showed Bush with a 3-5 point lead. Some showed it tied, others (e.g., Rasmussen) showed an even bigger Bush lead. Gore won by 0.5 percent.

In 1996, most polls showed Clinton winning by 13-15 points. He only won by 8. (Presumed last minute break for the challenger.)

In 1992, Perot did significantly better than the last minute polls indicated. Clinton won by a little less than predicted.

In 1984 and 1988, the polls were about right. :shrug:

In 1980, last minute polls showed it close. Reagan won by 10 points.

This is all from memory based on perusing historical data in recent weeks. The Gallup website has info on their past polls, going back at least to the 1940s, I believe.

--Peter

[DemocratSinceBirth]

4. I Was Referring To Final Pre Election Polls...

Most pollsters have a pretty decent track record...

Really, Scott Rasmussen is a joke but I'll give him major props for being the founder of ESPN...

[pmbryant]

8. Those are the ones I was referring to also.

Final pre-election polls (or at least ones in the final week).

I'm perplexed by how many Democrats tout Rasmussen's polls after his clear bias and dismal performance last time around.

--Peter

[DemocratSinceBirth]

10. Rasmussen Got Ripped For That One...

He had a comments site at his board and the comments were slashing and hilarious...

Also, a lot of freepers were surprised by how they were so misled...

[PATRICK]

5. Caution keeps them honest

They can shake off accusations about early polls before the voting week with any number of excuses as they modify them closer to reality.

This "art" is about the same as any stargazer or crystal ball con artist practices- the confidence game. They are selling confidence- as long as they are not forced by the power in charge to get extra deceitful- which is the case today.

They will risk very lame excuses that spin what we are complaining about.
Outdated or insufficient sampling, new voters, trends changing, etc.

Like any unflappable con. There is no science involved in today's polling samples only informed intuition and masked prejudice. The issue of soft opinions and pushed opinions to name two of the biggest unaddressed criticisms IS a scientific question. The confidence game merely brushes those aside with the fairly honest whine that admitting such considerations would destroy their business.

Since when has science been used so intuitively to replace common sense intuition? Since they needed to get at the edges and wedges that CONTROL the opinion making process, not the opinion monitoring process.

In this criminal endeavor so typical of brash American politics there is no science that is not flimflam and crackpot in its application. The relative validity of statistics and polling has been rendered superfluous, at a hidden level of trust in all polls, and unaccountable.

Trust your instincts and your ears, not your wishes or your fears.

[DemocratSinceBirth]

9. Polling Is Part Art And Part Science...

It's built on a lot of assumptions but if it was all voodoo how have they more or less nailed every election since 1948....

[PATRICK]

16. Because they had to to survive

I guarantee a session with a mind reader would produce the same confidence in you that those accurate polls have done. Trying to dissect how your trust has been tricked is equally difficult and time consuming. We settle for trust and thereby surrender the possible abuse that the "part science" part has achieved. They claim too much, create opinion too much, leave out too much and they always get more accurate as intuition gets closer to the actual date. That change is passed off as voter change of opinion if need be.

[DaveinMD]

6. based on the polling I saw

I didn't think Gore had any shot in 2000. So polling can be wrong. The trend is definitely in our direction right now.

[DemocratSinceBirth]

7. The Final Polls Had Him Within The MOE But Most Showed * Winning

it was a suprise....


go to www.ncpp.org/1936-2000.htm for documentation

[Jack from Charlotte]

13. Zogby was the closest. Most all picked Bush by from 2-5 points (nft)

[DemocratSinceBirth]

14. CBS And Harriss Had It Right...

if you look at that graph Harriss has been the best since 60...

[KayLaw]

11. Probably not what you want

I was just at www.smirkingchimp.com and today's last article is by Arianna Huffington, and it's about polls. This whole " likely voter " thing is a big crock. Recent polls are weighted in Bush's favor because the pollsters apply the notion that next month's voters will be 40% Republican and 33% Democrat. Where they get this idea, I don't know. She says the in the past two presidential elections, the voters were 39% Democrat, 35% Republican.

She also made the point that fake poll numbers like these can demoralize campaigns and decrease the numbers of volunteers and donations.

[DemocratSinceBirth]

12. If The Polls Are Bogus How Come They Always Get It Right

at least in presidential elections...

Doubting the polls is a classic suckers game....

And it's done on the right and the left...

[KayLaw]

15. Well

One point she made is if the polls are all so very different, how can they all be right?

[DemocratSinceBirth]

21. They Get It Right At The End.....

Are they that different?

www.pollingreport.com


Depends...

[PATRICK]

17. Other kinds of prediction

based on experience and assets yet to come into play and psychology are too complex for polling science. Trends of only several decades of elections are not significant in the progress and flux of the modern world and the Bush anomaly.

The assets the energized and wiser Dems with better candidates will bring into play tells me we are in very good shape. Bush's terrible presentation and worse record is also predictable. The polls ignore this analysis for good reason. It ruins their business.

The only imponderables are crises and fraud or some uncharacteristic mistake. Things are looking very good, but I would prefer Kerry gets the kill tonight.

[popstalin]

18. Not considering this years voter reg drive.

Another thing that people aren't taking into consideration is the polls are based on past voter turn out. If the turn out is anything like the mass registration drive, then the polls would be way off.

[Kammer]

19. Go here:

http://www.ncpp.org/1936-2000.htm

[Radical Activist]

20. Funny how you never see that analyzed after the election

The media doesn't do critical analysis of their polls too often. It's too bad we don't see more stories about how wrong and meaningless polls taken a month before the election have always been.

[DemocratSinceBirth]

22. That's Because Often Races Are Fluid...

Jimmy Carter was 33% ahead of Gerald Ford in the Summer of 76 and held on to win by a point or so....

In 80 the lead exchanged hands many times in the waning days of the campaign....

Of course some races are more stable....

Think 64

Think 72

Think 84

[PATRICK]

23. Heavy vote fraud

ruined exit polls reputation, not the other way round. So now they factor in some type of "anomaly"? or just stop polling. Wish we could have gotten them to back so easily on real harm they have caused in the past.

[TexasSissy]

24. Hey, NYCparalegal!......

Same here! Texas paralegal!:hi:

[Nicholas_J]

25. In 2000

Only Zogby predicted the events as they occured. Gore winning popular vote, Bush
winning the electoral college. Even so, Zogby is one of the pollsters most highly criticized
by other pollsters by using a double blind method in order to make sure that the people he polls are really either democrat, republican or independent that other pollsters state is not necessary (for not necessary, read another step that requires more work).

The most accurate pollsters in 2000 were theindepenndent pollsters, like Zogby, ARG,Harris, who did their own polls as well as polls for media clients. The mos unreliable were polls done by polling department set up by media entities themselves.

The lessons of the 2000 election that are being usded by republicans are that polls can be used to effect the outcome of elections be demoralizing the opposition.

This explains why Zogby and a few other independent pollsters quickly attacked the results of polls that showed George Bush gaining enormous leads after the REpublican convention, based on the oversampling of Republicans. They quickly analyzed the data to discover this unrevealed methodology, but in moore simple language. pointed out that the electorate does not switch camps so quickly in such a short period of time. This would be even more true in a nation so deeply divided by an extremely ideologically driven administration. Large percentages of people who one week stated that the war in Iraq has made us less safe from terrorism do not change their minds in a week based on the Republican convention and Zell Miller.

Polls can be accurate, but once the toll has been corrupted by those who figure they can use the tool as a double edged sword, to discourage enough of the opposition into not voting at all because the polls reveal that one candidate is SO far ahead that the opposing candidate cannot possibly win, places pollsters in a position that they must weed out the pollsters who are not using
polls as a toll to reflect public opinion, but as a tool that manipulates public opinion.

Zogby pointed this out very clearly by stating that the only hope that George W. Bush has to win is by getting as few democrats to vote as possible, because all things being equal, if every registered democrat and republican were to vote, and do so for the parties that they usually vote for, with the same percentages of cross party voting occuring, no Republican could ever win the presidency. If the same percentage of registered Republicans and Democrats vote (lets say 50 percent of all registered Republicans and 50 percent of all registered Democrats vote) then Democratis will always win.

In order for Bush to win, he has to maintain a continual 3 percent lead over Kerry in all poll. This is the bare minimum lead Bush needs in order to compensate for the greater percentage of Democrats, as well ash to compensate for the fact that historically, in the last weeksof the campaign for the presidency, the lions share of undecided voters vote against the incument.

Those who remain undecided so long do so because they are not satisfied with the administration in office and are waiting to see more of the opponenet before finally casting their support in the direction of the opposition.


For the past few days, independent pollsters are showing either a dead heat, or a few showing a statistical lead for John Kerry.

This does not bode well for Bush, as he no longer has enough of a lead to make up for Democratic numerical superiority, and historical tendencies for the final undecided voters to oppose the incumbent.