 search engine by Netcreations |
 |
 |
|
 |
|
 |
by Mac MacArthur
| Bush | |||||
| Forbes | |||||
| Dole | |||||
| Quayle | |||||
| Buchanan | |||||
| Keyes | |||||
| Bauer | |||||
| Alexander | |||||
| McCain | |||||
| Hatch | |||||
| Undecided/Pushed | |||||
| Undecided | |||||
| CANDIDATE |
Tuesday, August 17, 1999 --- New York (APJP) -- Take a look at the above numbers. Let me tell you what they really mean.
In addition to American Politics Journal, another group, PSI -- a commercial polling firm -- also polled likely-to-vote Iowans -- 300 of them -- trying to predict who they'd vote for at the "Straw Cash Poll."
There are several problems with the PSI poll.
First of all, their sample was too small -- especially when you are dealing with a group of "voters" who are essentially being paid to vote. American Politics Journal had trouble with this factor as well.
Second, we didn't give people a chance to be "undecided" last Thursday evening -- although we knew that approximately 15% of Iowa's might be "confused" from historical knowledge.
So we beefed up our sample size to over 2,000 to try and get a better handle on things.
In most ways, however, both of us PSI and APJ were in error by almost the same amount, despite the fact that we pushed respondents to choose and that PSI poll showed nearly 18% undecided.
We're glad they had that undecided figure, because it allows us to try and guess what happened between both polls and "election"day.
Since both of us were equally or near equally "optimistic" about Bush, it is clear that Bush lost 6-7% of Iowans on the last day.
But who was the beneficiary?
Without exit poll data we can suppose:
(1) that Forbes or Dole took the Bush 6-7%
(2) that Forbes and Dole split the Bush losses and picked up an additional 2-3% from someone else,
(3) that Forbes got all the Bush switchers,
(4) that Dole got them all.
We first thought that Forbes got most of the Iowa switchers because we couldn't see Dole appealing to Bush voters. However, Dole is more closely aligned to Bush ideologically ,therefore one might think that Bush switchers would move to Dole.
One factor we looked at was the gender of our poll respondents. That told the story.
The Bush "switchers" went largely to Dole among women -- and the rest went to Forbes. It is our opinion that Dole picked up 3-3.5% of the Bush switchers who were women and that Forbes got the male Bush switcher vote.
How ironic can you get!
So, in the end, it was a split.
We, of course, tracked the media over two days following our poll. With the information we have on the split - we can assume media coverage cost Bush the 6-7% he should have had otherwise.
And this bodes poorly for Bush, because the only semantic difference during those last days was that Bush came "out of the closet" and participated in front of the camera. His appearances were obviously a negative for him. People didn't like him in as great a number as they had before they saw him in action.
Forbes did 4-7% better than both the PSI poll and APJ predicted. We are certain that he got at least 3% from Bush, but where did the other 1-4% come from?
We think from Quayle. There are no other candidates with large field operations and lots of money that would be attracted to Forbes as an alternative -- despite Forbes's being a "born-again" abortion foe. Pro-life voters went to Bauer, that much is obvious. And churches -- particularly Evangelical-leaning churches -- were preaching "pro-life" from the pulpit the prior Sunday.
We know -- because we checked. This is an old ruse used by Republicans for decades. They enlist parish priests and ministers to preach pro-life, anti-gay sermons the Sunday before any election.
Bauer's stronger-than-expected showing was a surprise to both the PSI poll and to us. I must tell you that Bauer is one of the most repulsive human beings -- on a personal level -- that I've ever encountered. Most people who have had to deal with him find him to be just that; thus, there is a tendency for poll analysts to downgrade Bauer's outside chances.
This is a good time to discuss polling in general.
Polling, despite what people say, is really not a science, nor is it an art. It is a combination of both -- a kind of alchemy born from deep knowledge on current voter attitudes and a finely-honed historical perspective.
When the raw data from a poll is analyzed, things are never crystal clear. That's why polls need analysis. While every pollster presents the raw data to his academic colleagues, the actual prediction to the public or the client is generally a mix of science and perception by the polling firm's top analysts. This is called, by many firms, the Executive Summary. These perceptions include propensity to lie, what breaking news may do or not do to raw results, and what open-ended questions in the same poll reveal about "trend" -- and all sorts of other conclusions.
We don't know what PSI did, but I can tell you that we did not award Bauer any points from a pool of votes that sort of "coagulates" once the raw data is in. That was a mistake. Had we ascribed some of this "sludge" to Bauer we would have been within 1 point of his actual total. It appears that the PSI poll could have done the same, but instead chose to allow its respondents to respond "undecided" to its poll. They had no way to predict where the undecided votes would go -- save that historically, the majority of the undecided move to the front runner (in this case Bush). Had the PSI poll analysts done this, they would have been off by more than 15% on Bush, we think -- and so would have we.
Looking at that large number -- 17% undecided, just over 1 out of 6 -- we think that these people were evenly distributed among all the candidates, most likely in equal ratio to the final result.
One more thing: polling the Straw Poll voters is a dangerous concept. As a matter of fact, it's lunacy because these so-called "voters" were either ardent supporters of specific candidates, or people who wanted a couple of $25.00 free dinners, booze, entertainment and a fun day in Ames, or folks who wanted to play a part in history.
The fact that both the PSI poll and our poll were so wrong about Bush -- and therefore so wrong about Dole and Forbes -- proves that. But with all the other candidates, we at APJ were within normal plus or minus 3% predictability. The PSI poll didn't do as well -- but we ascribe this to its small panel and its choice to allow respondents to remain "undecided." We pushed our respondents to decide.
One thing's for sure: the Iowa caucuses will be easier to poll and easier to predict.
In addition to American Politics Journal, another group, PSI -- a commercial polling firm -- also polled likely-to-vote Iowans -- 300 of them -- trying to predict who they'd vote for at the "Straw Cash Poll."
There are several problems with the PSI poll.
First of all, their sample was too small -- especially when you are dealing with a group of "voters" who are essentially being paid to vote. American Politics Journal had trouble with this factor as well.
Second, we didn't give people a chance to be "undecided" last Thursday evening -- although we knew that approximately 15% of Iowa's might be "confused" from historical knowledge.
So we beefed up our sample size to over 2,000 to try and get a better handle on things.
In most ways, however, both of us PSI and APJ were in error by almost the same amount, despite the fact that we pushed respondents to choose and that PSI poll showed nearly 18% undecided.
We're glad they had that undecided figure, because it allows us to try and guess what happened between both polls and "election"day.
Since both of us were equally or near equally "optimistic" about Bush, it is clear that Bush lost 6-7% of Iowans on the last day.
But who was the beneficiary?
Without exit poll data we can suppose:
(1) that Forbes or Dole took the Bush 6-7%
(2) that Forbes and Dole split the Bush losses and picked up an additional 2-3% from someone else,
(3) that Forbes got all the Bush switchers,
(4) that Dole got them all.
We first thought that Forbes got most of the Iowa switchers because we couldn't see Dole appealing to Bush voters. However, Dole is more closely aligned to Bush ideologically ,therefore one might think that Bush switchers would move to Dole.
One factor we looked at was the gender of our poll respondents. That told the story.
The Bush "switchers" went largely to Dole among women -- and the rest went to Forbes. It is our opinion that Dole picked up 3-3.5% of the Bush switchers who were women and that Forbes got the male Bush switcher vote.
How ironic can you get!
So, in the end, it was a split.
We, of course, tracked the media over two days following our poll. With the information we have on the split - we can assume media coverage cost Bush the 6-7% he should have had otherwise.
And this bodes poorly for Bush, because the only semantic difference during those last days was that Bush came "out of the closet" and participated in front of the camera. His appearances were obviously a negative for him. People didn't like him in as great a number as they had before they saw him in action.
Forbes did 4-7% better than both the PSI poll and APJ predicted. We are certain that he got at least 3% from Bush, but where did the other 1-4% come from?
We think from Quayle. There are no other candidates with large field operations and lots of money that would be attracted to Forbes as an alternative -- despite Forbes's being a "born-again" abortion foe. Pro-life voters went to Bauer, that much is obvious. And churches - particularly Evangelical-leaning churches -- were preaching "pro-life" from the pulpit the prior Sunday.
We know -- because we checked. This is an old ruse used by Republicans for decades. They enlist parish priests and ministers to preach pro-life, anti-gay sermons the Sunday before any election.
Bauer's stronger-than-expected showing was a surprise to both the PSI poll and to us. I must tell you that Bauer is one of the most repulsive human beings -- on a personal level -- that I've ever encountered. Most people who have had to deal with him find him to be just that; thus, there is a tendency for poll analysts to downgrade Bauer's outside chances.
This is a good time to discuss polling in general.
Polling, despite what people say, is really not a science, nor is it an art. It is a combination of both -- a kind of alchemy born from deep knowledge on current voter attitudes and a finely-honed historical perspective.
When the raw data from a poll is analyzed, things are never crystal clear. That's why polls need analysis. While every pollster presents the raw data to his academic colleagues, the actual prediction to the public or the client is generally a mix of science and perception by the polling firm's top analysts. This is called, by many firms, the Executive Summary. These perceptions include propensity to lie, what breaking news may do or not do to raw results, and what open-ended questions in the same poll reveal about "trend" -- and all sorts of other conclusions.
We don't know what PSI did, but I can tell you that we did not award Bauer any points from a pool of votes that sort of "coagulates" once the raw data is in. That was a mistake. Had we ascribed some of this "sludge" to Bauer we would have been within 1 point of his actual total. It appears that the PSI poll could have done the same, but instead chose to allow its respondents to respond "undecided" to its poll. They had no way to predict where the undecided votes would go -- save that historically, the majority of the undecided move to the front runner (in this case Bush). Had the PSI poll analysts done this, they would have been off by more than 15% on Bush, we think -- and so would have we.
Looking at that large number -- 17% undecided, just over 1 out of 6 -- we think that these people were evenly distributed among all the candidates, most likely in equal ratio to the final result.
One more thing: polling the Straw Poll voters is a dangerous concept. As a matter of fact, it's lunacy because these so-called "voters" were either ardent supporters of specific candidates, or people who wanted a couple of $25.00 free dinners, booze, entertainment and a fun day in Ames, or folks who wanted to play a part in history.
The fact that both the PSI poll and our poll were so wrong about Bush -- and therefore so wrong about Dole and Forbes -- proves that. But with all the other candidates, we at APJ were within normal plus or minus 3% predictability. The PSI poll didn't do as well -- but we ascribe this to its small panel and its choice to allow respondents to remain "undecided." We pushed our respondents to decide.
One thing's for sure: the Iowa caucuses will be easier to poll and easier to predict.
Copyright © 1999, 1998, 1997, 1996, American Politics Journal Publications.
All rights reserved.
ISSN No. 1523-1690