Why some pollsters got it so wrong on election day - "We blew it," Zogby says now

Minneapolis Star-Tribune ^ | 11-10-02

[Oldeconomybuyer]

Just before election day, the St. Louis Post-Dispatch published a stunning poll about the governor's race in Illinois: GOP candidate Jim Ryan was ahead.

The Republican's 43.5 to 43.2 percent edge over Democrat Rod Blagojevich, though tiny and statistically insignificant, landed with a bang in the world of Illinois politics because many earlier surveys had shown Ryan to be far behind. Moreover, the poll was conducted by nationally known pollster John Zogby, of Zogby International in Utica, N.Y., acclaimed for his accuracy in the past two presidential campaigns. Zogby told the Post-Dispatch that he had personally affirmed its accuracy.

[ReleaseTheHounds]

Zogby's face is covered with his special sauce... Gee, how would Bill Clinton govern with unreliable polling now the norm?

[SamAdams76]

Does this mean that at the next election, we don't have to suffer through several thousand meaningless "Zogby is God" threads here at Free Republic?

[Always Right]

Polling is not a science but an art. The +/-2% they claim to be should be multiplied by about 4. Zogby does pretty good, but there are just too many factors that effect elections results. It is impossible to predict how millions of people are gonna act.

[G.Mason]

"We blew it," Zogby says now."

"So, it appears, did many other pollsters. The reasons may be as various as the recent popularity of caller ID and cell phones, which hamper efforts to reach voters, and the nation's increasing ethnic diversity, which makes it harder to get an accurate statistical sampling of the electorate."

In a few words?

Adapt or die!

[Foxphile]

Gee, I guess politicians have to learn how to lead again (like W) instead of using polls (like x42). </sarcasm off>

[rstevens]

Biggest problem with polling now is that the pollsters have an agenda. They carefully word the questions to get the result they want to purposedly mislead the opposition.

These turds like Zogby should try some truthful and honest polls before they start blaming the American citizenry.

[Tom D.]

One thing that is never explained to the populice is what that +/-2% means. It does not mean that the sample is absoloutely within 2% of the universe; it means that the sample size is such that there is usually a 90% chance that it is within that 2% margin of error.

Further complicating matters is the fact that the mathematical formulae used in making these margin of error calculations work best for such things as sampling widgets comming off an assembly line to determine what proportion of them are defective. They are not designed for dealing with people who sometimes lie or are ashamed to admit that they do not know something or whose mind might be changed the next time they discuss the issue with a friend or see a TV ad that slams the other side.

Another issue is whether the questions used in the poll either intentionally or unintentionally lead to a given response. We have the push-polling, but often the wording or the delivery of questions unintentionally cause the answers to be skewed.

Still more problematic is the issue addressed in the article: How can you be certain that you have a representative sample?

The media are either ignorant about polling results and techniques or they wilfully hide the questionable results of various polls. The New York Times can whine about Drudge dealing in rumor rather than in news, but the Gray Strumpet has no qualms of hailing what is of little more value than the arrangement of tea leaves, particularly when they say what the NY Times wants to hear.

[ErnBatavia]

Pollsters are assuming honest answers. Twice here in Kalifornia, Tom Bradley (black, liberal demoncrat) had double digit leads in his gubernatorial races - right up until election day. Twice, he lost.

IMO,people being good demoncrats told pollsters they'd vote for the black candidate, but getting in that private booth, they played the race card....in the truest sense.

[Steve Eisenberg]

Twenty years ago, two-thirds or more of Americans were willing to accept calls from interviewers and linger on the phone for the 20 minutes or so that it takes to conduct the survey, Wirthlin says. Today, that proportion has been cut in half.

Good news. At this rate, in 20 more years, these fellows are out of business.

[Capn America]

Your observation is correct, the pollsters DO have an agenda. It is the same as the Establishment's agenda.

The agenda is Big Government, Gun Control, Internationalism, Anti-Christian, Globalism, etc.

Moreover, the High Cabal uses these polls to indoctrinate the populace, as in, "Hey, everyone else thinks such-and-such, and so should YOU!"

But not only do they use polls to further their agenda, they also "spike" polls when the outcome doesn't match their desired message.

Never stop questioning the things the Media tell us, or wondering if they are telling us the WHOLE story.

Kinda reminds me of the debate on NAFTA, the entire Establishment was pushing NAFTA, the leaders of both major parties, nearly the entire media was pushing for NAFTA too.... but there were so many things they didn't tell us during that time, one thing in particular, they didn't talk much about the fact that the overwhelming majority of Americans did NOT want NAFTA!

God Bless America!

[Gorzaloon]

To compensate, some research firms are moving toward alternative means of polling, such as the Internet. That has been hindered by the fact that Internet users aren't representative of the general population. Harris Interactive, in Rochester, N.Y., argues that it has developed "weighting" methods to overcome that problem.

Yes, Harris relies on spam, and sued one of the blackhole lists over having their spam blocked.

A really reputable organization. *spit*

[umgud]

The article does a good job of explaining the mechanics of why polls are not accurate.

It does not explain why the results usually favor the democrats.

[Hurricane]

I am one of those who factor into the half in the last 20 years who choose not to respond to pollsters. Prior to the election I had no less than five calls from pollsters. None would tell me who was paying for the poll. I told them all to go pound sand.

[marajade]

The reason why Zogby did so bad is because he'll use the most previous election percentages and weights them into his new polling. Nobody predicted that the Reps would win by 6-7 pts in the congressional seats.

[Always Right]

Further complicating matters is the fact that the mathematical formulae used in making these margin of error calculations work best for such things as sampling widgets comming off an assembly line to determine what proportion of them are defective.

Really these sampling methods work well when there is a definite outcome. When they sample people for a poll, you really don't know the outcome. Will that person actually vote? Maybe, maybe not. Will that person change their mind? Maybe, maybe not. Pollsters are not really sampling but guessing.

[Semper Paratus]

Following the election night was much better this cycle than in 2000.
VNS almost croaked Bush, with a lot of help from the 3 networks.
This time you could kick back and enjoy the slow build up hour by hour watching Woodruf and Rather get long faces as the evening wore on.

[gaspar]

Ruch Limbaugh, who used Zogby's info on Illinois to warn his listeners of a potential upset, must now feel like a perfect fool. Perhaps it is time for him to admit that the Zogby of 2002 is not the Zogby of the mid-1990s.

[Balding_Eagle]

The +/-2% they claim to be should be multiplied by about 4.

Think about that a bit..............what that means is that a flip of a coin will usually be as accurate as the poll.

BTW, I do agree with you. Just pointing out the obvious conclusion.

[Psycho_Runner]

Absolutely right. The knucklehead pollsters need to jack up the sample count to get a better forecast. Instead of 90 try 95. The other observation that goes unnoticed is talk radio and the internet. Why isn't talk radio growth ever discussed? And finally the internet, where I can choose the news source. Drudge is actually a good source.