The Poll Truth: The Facts Behind Poll Fluctuations [LA Times Poll out on Wednesday]

NBC4 ^ | 9-29 | NBC4

[ambrose]

<<back to home | Politics

 |  Print This Story

The Poll Truth: The Facts Behind Poll Fluctuations

Timing, Definitions Could Explain Range In Poll Numbers

POSTED: 5:29 p.m. PDT September 29, 2003

UPDATED: 6:08 p.m. PDT September 29, 2003

LOS ANGELES -- Why do the results of polls vary so widely? Channel 4's Investigator Joel Grover tracks down an explanation.

Frank Newport, the head of the Gallup Poll, was in the hot seat on Monday, trying to explain his organization's latest poll.

The head of the organization said, "We're happy to stand behind our data... What we're offering is a snapshot."

That snapshot shows Arnold Schwarzenegger surging ahead of Cruz Bustamante.

Poll: Californians Widely Favor Recall

Pollsters said it's all about the timing.

Meryn Field, the founder of the Field Poll, told NBC4, "They can do a poll in a week one, and we can do a poll (in) week three. Between week one and week three, a lot of changes have occurred.

 SURVEY

Do you trust polls?

Yes.

No.

I don't know.

Results  |  Disclaimer

The latest Gallup poll was taken right after the last debate, where voters polled gave Schwarzenegger much higher marks than Bustamante.

Newport said, "Clearly the debate, the only debate in which media star Arnold Schwarzenegger participated, could very well have had an effect on how voters feel about the race."

But there are other reasons why the results of this new Gallup poll differ from the rest. Each polling organization surveys people they consider "likely voters," but how each poll defines a "likely voter" differs from poll to poll.

Newport said, "Every polling organization -- the LA Times, the Field Poll in California, polls elsewhere -- use their own techniques for trying to take the registered voter base."

The Gallup organization admits that each and every poll could shift radically in the next few days as voters are bombarded with a new wave of TV ads and news coverage.

In the end, the biggest factor in this recall won't be the polls, but the turnout on election day.

Bob Stern, of the Center for Governmental Studies, told NBC4, "I think all polls are really suspect. The only poll this time that is going to count is on election day."

Before election day, you can expect a few more polls to grab the headlines. The LA Times is expected to release its next poll Wednesday morning.

Copyright 2003 by NBC4.tv.

[ambrose]

ping

[Diogenesis]

Real reasons: sampling error, select questions
and manipulation of the data

[kingu]

Ugh, I dread the Wednesday polls. I think I'll just refrain from any and all recall issues starting Tuesday night until election day.

[DoctorZIn]

Join Us…Your One Thread To All The California Recall News Threads!

Want on our daily or major news ping lists? Freepmail DoctorZin

[Rodney King]

Notice how the media only questions polls when the Republicans are ahead in them.

[ambrose]

So after the LA Times releases its poll, the media spin for the remainder of the campaign will be the "Davis surge"...

[goldstategop]

Liberals are trying to explain the polls away. Let's face it, if Gray Davis were really ahead and Large Breasts looked like he was having a victory lap, the media would trumpet the polls till they're blue in the face. All of a sudden its the other way around and they're pretending the polls don't matter. Life's funny when you're going to lose. Insisting the only poll that matters is the one that counts on Election Day is like whistling past the graveyard. They still don't get it yet but they will next week!

[goldstategop]

If the Slimes manufacture a Davis momentum that doesn't exist, I won't be surprised. We know the liberal media and the Democrats like the back of our hand.

[DrDeb]

"Notice how the media only questions polls when the Republicans are ahead in them."

GREAT INSIGHT -- I didn't notice any of the leftist media outlets questioning Gallup's last poll concerning the President (a poll that placed GWB in a statistical tie with 'Ashley' Clark -- a stat NOT verified by any other poll, including the recent Battleground, Zogby and Newsweek polls)!

[Graewoulf]

Food for thought:

1.) Polls are based on OPINION, but "The Media" presents these opinions as facts.

2.) According to Zogby, 60 percent of those chosen to be polled refuse to be polled. Thus, the margin of error is the usual 3 to 5 percent PLUS 60 percent.

3.) "The Media" uses opinion polls to bypass elections. They try to advance their candidates, and discourage support for candidates that do not fit into the agenda of "The Media."

[irv]

"The Media" uses opinion polls to bypass elections

4) Reporters are not very bright, but are exceedingly lazy. Thus polls give them a way to pump up their own importance (influencing the outcome, as stated above) without having to actually work. And they may even believe that the polls they are "reporting" constitute news.

I say this as a non-news employee of a newspaper.

[Graewoulf]

Excellent point! What is easier? Go out and ask people what their actual opinions are, or ask their opinion on one of the journalist's questions? The former case is called reporting, and the latter case is called journaling.

[kingu]

Davis and Tom surge... Aww, heck, it'll be a surge for everyone except Arnold.

[dpwiener]

Thus, the margin of error is the usual 3 to 5 percent PLUS 60 percent.

I suggest that you take an elementary statistics course before you utter more absurdities like that one. The margin of error is NOT 3 to 5 percent plus 60 percent. The only question pollsters face is whether the characteristics of those who refuse to respond to a poll differ significantly from those who do. We know of only ONE difference -- they are people who choose not to respond. But that tells us nothing about whether their views would systematically diverge from those who do respond. Unless the decision not to respond correlates strongly with political views in one direction or the other, it is quite reasonable to assume that their opinions are accurately represented by the sample which does respond.

[cartoonistx]

Well, I did my part! The Times polled me last night!

[Graewoulf]

"We know of only ONE difference -- ." The "We" in your quote implies that you are a pollster.

" --- it is quite reasonable to assume that their opinions are accurately represented by the sample which does respond." This sounds like a flawed and probably "absurd" assumption to me.

BTW, has the concept of "none of the above" crossed your mind? I submit that 60 percent of the population of voters prefer it to the question-of-the-day posed by the pollster. Until a better term for it can be invented, I shall continue to call it a 60 percent error by the pollsters.

[dpwiener]

"We know of only ONE difference -- ." The "We" in your quote implies that you are a pollster.

It was merely a stylistic "we"; it was not intended to imply that I am a pollster. I am not a professional pollster (although I have done some polling in the past, and I do know what I'm talking about).

[Bonaparte]

"Real reasons: sampling error, select questions and manipulation of the data"

Also, differences in turnout projections.

From the article:

"I think all polls are really suspect. The only poll this time that is going to count is on election day."

This is the only thought the writer wanted to put across. All the rest is fluff. The LAT is upset that all the polling results show Arnold with a comfortable lead, a lead they have been working overtime to eradicate. So they dis all the polls, including their own. The left can live with chronic self-contradiction, but not with Arnold in the governor's office. They can even live with an impoverished, bankrupt California (as their endorsement of Davis proves), but not with Arnold in the governor's office.