07
Dec
2004
Last week, the River Cities' Reader sat down with Peter Hart, a national pollster with the firm of Peter D. Hart Research Associates who was in the Quad Cities as Woodrow Wilson Visiting Fellow at Augustana College. Hart talked about the lessons learned from the 2004 election, as well as about how polls are used (and misused) by the media and the public.
River Cities' Reader: What can we learn from the 2004 election?
Peter Hart: First and foremost, we talked about this election in some respects as transformational. It wasn't transformational at all. You compare 2004 to 2000, 47 of the 50 states voted exactly the same way. And essentially three states that changed - New Mexico, New Hampshire, and Iowa - if you take 10,000 votes, all 50 states would have ended up with exactly the same result, and we spent how many of billions of dollars on this election? I mean it was just an astronomically large amount, and we had a tremendous turnout, which meant interest was good. But in terms of it being a transformational election, it certainly was not. What was important is the balance of power in terms of the United States Senate clearly changed. The Republicans went from 51 to 55. They control the House by a slightly larger margin, so they have all instruments of government at this stage of the game, and that gives them exceptional amount of leverage.
The second thing is there were groups within the electorate that became important and showed us differences. Married women changed it in terms of how they voted. They became more Republican. The second group would be Hispanic voters. They were overwhelmingly Democratic four years ago. In this election, 53-44, not overwhelming, and that is a fast-growing group. And the one group that is exceptionally important are the religious voters, and those are people that go to church on a weekly basis. And when I say religious voters, I don't mean white evangelicals. Forty percent of Americans go to church on a weekly basis. Those people voted by a 22-point margin for Bush over Kerry. They account for one-half of all the votes that George Bush received. The president got about 44 percent of the rest of the electorate. That group becomes exceptionally important, and that's a huge change in terms of where things are and what's important.
Essentially the president lost this election on the issues and won the election on the basis of personal values and personal characteristics. Voters saw him as consistent; they saw him a religious person, a man of faith; and they saw him as easy-going and likable. Those are the three elements that worked for him the best and in addition the issue of terrorism.
If there were no terrorism issue, the president would have lost. Seventy-six percent of the voters say that I'm concerned a great deal or quite a bit that there will be a terrorist attack. Among all voters, 57 percent say I have a great deal of confidence in terms of Bush and his ability to handle terrorism. The like number for Kerry was 40 percent. That essentially became the election.
And the final elements obviously were social issues and moral values, and I think the president benefited from that.
One of the things we've been talking about is the importance of the Mississippi River, and if you look at the states along the Mississippi River, you get a sense of the changing nature of American politics. Missouri was a toss-up state last time, solidly Republican this time. Iowa was slightly Democratic and has been Democratic since 1988, and this time flips over. And now Minnesota and Wisconsin are very close to the edge.
RCR: There's a huge cultural difference between Missouri and Minnesota and Wisconsin - .
PH: If you go from Minnesota, the top of the Mississippi River, all the way down to Louisiana in the Gulf of Mexico, essentially what I call this is the lifeblood of this country. Others have said it's the spine of America. It's 2,350 miles. You look at this group of 10 states, and since 1912 a plurality have always voted with the winner. That's pretty indicative. Louisiana and Minnesota are pretty far from one another, but when you look and recognize that Jimmy Carter carried nine out of 10 of those states in 1976 and I think Clinton did about the same in 1992, and Reagan carried nine out of 10 of them in 1980, these are states that obviously do switch and follow the trend in terms of American politics.
You could say some of them are red and some of them are blue; certainly they're much closer to what I call the middle. If you start to look at Minnesota and the rural areas, which used to be heavily DFL [Democrat Farmer Labor], they're a lot closer to looking like Missouri today than they are to looking like New York or California. So I think there's a changing nature, and I think a lot of it has to do with values.
RCR: A lot of the Democratic disappointment with results had to do with the early exit polling. We had two elections, one where the exit polling was wrong in the calling of Florida, and this year when the early results weren't indicative of what actually came through. What went wrong with those?
PH: I started my professional life in 1964, 40 years ago, working on the first exit polls that were ever done for a network. It was called the Voter Profile Analysis, and I went out and personally picked the precincts that would be a part of the model of the sample, and so I've watched the evolution of this and what has happened over the course of 40 years.
I believe that exit polling is one of the great tools that was ever invented, but it never should have been invented for calling elections. Its greatness is in terms of analyzing elections and understanding. The reason it is so great is that you can get somebody right after an event and before they know the outcome of the event and they'll tell you exactly why they voted the way they did and what was involved. Nothing can be more valuable than that.
It's different from watching the debate where people will listen to commentators and then say to you, "You know, I thought that so-and-so failed to fully answer that question." Well, they never saw that, but they are reacting to other external stimuli. Here there are no external stimuli. They come out and say, "This is what I did and this is why I did it." And that is exceptionally helpful both in the scholarly way and also in a reportorial manner.
The problem is, obviously, networks are not into scholarly things. They want to know first who won. So all the networks then built models after the way CBS did it. There was all this competition. And by and large networks got things right; there were always mistakes that were made, but they were made in a state race or in an inconsequential situation and never made in a major way.
And so then the networks said, "This is silly. We're each building our own float to bring it down the street. Let's all pool altogether, and then we can analyze it separately but there's no reason we have to send Peter Hart out to look at one set of precincts and Jeff Ignatius down to look at another set of precincts." So they all pool together.
The disadvantage of that, obviously, is that if there's a mistake made, it's made everywhere. That's what happened in 2000. They had a bad sample in the state of Florida, and nobody recognized it. They all accepted it, and they all made the mistake and called Florida wrong.
Does it happen? Yes. Does it happen in such a critical state in such a critical way? Usually not. But that's what happened.
The second thing that you need to understand. You're a restaurant reviewer. You go in and eat the appetizer and then rush out and give a review of the restaurant. Chances are you are not going to give a very complete and very good review. That's what early exit polls are. They're the appetizer. And at 12 o'clock or 1 o'clock, when they collected this data and put it out, it's a lot of garbage in that.
That's never been a problem because there were a handful of people, such as myself, on the inside of the network, who would look at all of this stuff and say, "Boy, it looks like a Reagan landslide today," or they'd say, "This stuff looks screwy." But nothing happened between 12 o'clock and 5 o'clock except for a handful of people on the inside. Dan Rather would come to me or Tom Brokaw would come to me and say, "What do you make out of this?" I'd say, "Sit on it Dan. There's nothing." Or: "Something is breaking and it is breaking in a very unusual way."
It's like the family. Same thing. When you all eat the appetizer at the table, and you're the reviewer, and you say, "What do you think?" "Boy, I'll tell you. It doesn't seem like they put out a very good Caesar salad, and they don't make their croutons. Then out comes the main course, and you four have a knockout main course. By the end of the meal, you say to yourself, "this is quite a good restaurant but the appetizers aren't so good." Fine. Nobody but the four of you have really discussed it.
Suddenly we have the advent of blogs and the advent of all of the Internet access. This family of four around the table becomes a family of 40,000 or 400,000. Suddenly they think they know the election results.
When Brokaw said to me, "What do you make out of this?" I said to him, "Be careful. It's early." I've seen senators go down in flames at the 1 o'clock level, but by 5 o'clock they're re-elected. I knew my polling was correct, but the people who were there early in the morning weren't an accurate sample. So that's number-one element. The early stuff is always wrong. It's just that everybody found out about it.
But there was the second problem, which was a more significant problem that was neither corrected nor anything else. They were way too female in their samples. There were 55 and 57 percent. I was sitting there with Bill McInturff, who is my co-partner on the NBC/Wall Street Journal poll, and we're sitting there with Russert and Brokaw. We didn't have a problem. We just looked at this stuff and said, "Ain't gonna happen. Too female." Males were plus-11 for Bush; females were minus-3 for Bush.
So then you have to ask how they got the gender wrong. I have no idea, but they did something wrong. My guess is that we probably have seen the end of exit polls.
RCR: Most people think of polling as this very exact science, but I think we've learned the last two elections that it is an art as much as anything. Or is it?
PH: Exit polling is nothing more than an open-book test. There's no way you should get it wrong. People are telling you what they finished doing.
Now there's one problem today that didn't exist 40 years ago and 30 years ago. Forty years ago, everybody went to the polls. Today, one-third of the people are no longer going to the polls. They're voting before the election.
And now all of a sudden I'm doing an election-day poll, I've left out one-third of the people and I say, "Oh, well I'm going to make up for that. I'm going to phone a certain number of people who voted pre-election, and we'll find out what they did." Well, you've lost the value of election-day polling, because they're no longer the same people they were when they exited the polls. So you've got that difficulty.
Now, pre-election polling is much more difficult. Always been more difficult. Why? Because you don't predict. All you're doing when you take a poll is understanding how people are at a point in time.
It is equivalent to me talking about your behavior. On your way to work you may sideswipe the car. You're a totally different person after that event. Same with an election. Events happen. People change their minds.
That's why we do polling. Polling is about telling a story, and where people are and what's happening. But the public always thinks a poll's a prediction. It's not a prediction. Only reflects.
Now when you look at the number of polls that were taken over this election period, and you look at the end results of all of them, they were amazingly accurate within the margin of error. What we're saying is: We can talk to 1,000 people in this country and within plus-or-minus 3 percent tell you what 120 million people are going to do. That's an amazing fact.
And our final poll had for NBC and the Wall Street Journal had Bush ahead 48-47. New York Times had it, I think, 50-47. Somebody else had it similarly. Most of the polls all showed the same thing. This was an amazingly close race, and in most cases they had Bush maybe ahead.
Let's suppose Kerry had won the 70,000 votes that switched in Ohio. Would I be wrong? No. All I can tell you is it's a close election.
Now, when you look at the work that was done by the Gallup organization in individual states, and they miss seven out of seven states, you have to question what is happening there. Essentially they drew a screen that was incorrect. A screen meaning: We looked at our respondents and we said certain people are going to be likely voters; we're going to eliminate people who we think are unlikely voters. Their basis was on a small turnout. It was a big turnout. Consequently they were screening out people who should have been counted in. So states that they thought were going to go Republican went Democratic. There were states that they thought would go Democratic that went Republican. They just did a bad job.
RCR: To determine likely voters, you ask questions such as "Did you vote in the last presidential election?" or other questions that you think will be an accurate indicator of future voting patterns.
PH: If you look at every major good polling organization, when they talk about "all registered voters" we're all talking about the same thing. Those are apples to apples. So you can go across all the polling organizations and, essentially, they should have the same thing. But when you talk about "likely voters," every organization has a way of defining likely voters somewhat differently than every other organization. Look at the difference between X and Y organizations, and in reality their differences are in the way in which they define likely voters, and since nobody knows who is correct or incorrect here, it turns out to be more of an art than a science.
And in the end the Gallup organization, I believe, has a major flaw. They have questions that they've used since 1950, including, "Do you know where your polling place is?" My son has no idea where his polling place is, but he knows he isn't coming home for Thanksgiving dinner unless he votes. And so he's apoplectic when he finds out that he hasn't gotten his absentee ballot, as he has no intention of going to the polling place. But to the Gallup organization, they say, "Oh, he's not a likely voter." Just an antiquated way of looking at the electorate.
RCR: Do news organizations use polling properly and does the public interpret it properly or understand it properly?
PH: Do news organizations use it properly? No. News organizations look at their job as promoting the news or being able to develop a new story. And therefore their tendency is to use polling to try and go in and get something at the highest high or the lowest low. And that's abnormal.
It would be equivalent to going out and measuring the water table of the Quad Cities area the day after a major thunderstorm and then rushing out reporting the new water table. Well, we rush in right after the convention or rush in immediately after a debate; we're not letting opinions settle and come back to the proper position. And therefore we set up something that is totally artificial, which is: "Look, the water table is way up." And then we come in a week and a half later and say, "Oh, this guy has dropped like mad." They never dropped. It was an artificial expansion, and it dropped back to its original level.
We did polling this year as we have done for the last 15 years for NBC and the Wall Street Journal. They have a different way of approaching it. We go in, and we poll every five to six weeks. We purposely avoid the major events. We look for the calmest period.
If you take all of our polls from the middle of March through the end of October, we never had Bush with more than a five-point lead, and we never had him less than dead even. So, did the election go into these plus-12 and minus-3s and whatever else it did? Sure it did. For days or hours but never as a whole. I think that news organizations in that way failed the readers.
The second way in which they failed the readers is we get caught up in horse-race journalism, and we really failed to explain the dynamics. The New York Times does a very good job, the Pew organization, the Wall Street Journal does a very good job, but an awful lot of organizations just are interested in putting out whatever that sort of odd little figure is that sells newspapers.
RCR: You mentioned the way NBC and Wall Street Journal do their polling. Is that your choice, theirs, or a combination?
PH: Ultimately it becomes their choice. But I think it says something about the standards.
And one of the things that we've done for them over a period of time is these huge quarterly polls where we discussed one issue in-depth, and the Wall Street Journal took out eight pages of the paper to discuss the issue of values, the issue of race, etc.
RCR: You mentioned horse-race journalism. Is polling more prevalent now than it was 20 years ago?
PH: Tremendously more. I mean what I call it is poll proliferation. Every organization to get their own bona fides wants to go out as to do a poll.
Newsweek is one of the worst . They try and figure out how they get their poll out in a way that hypes their newsstand sales for Monday. They put it out on Saturday, and they look for that spike in order to be able to help them at the newsstand.
RCR: Is there anything inherently bad about polling? Or is it the way it is used by the politicians and the media?
PH: You can't spend a lifetime doing what I do and say it's bad. It's a marvelous tool, and it can be used correctly or incorrectly. But as a way of telling our story, and a way of understanding where the consensus of the public is, it becomes exceptionally important.
Take two of the major events in my lifetime; one is Viet Nam and the other is Watergate. In both of these instances, the public was way of the politicians, and by having public-opinion polling, I think it helped those in public life understand that we wanted to be in a different place than we were. And from that point of view it was exceptionally helpful. I believe it's helpful during the Iraq war. I think it's telling the story of a very divided America, and from that point of view I think it's going to provide guidance for the administration to recognize they have to find a solution and a way out here. They do not have an unending amount of time from the public.
RCR: Most of your polling looks at a nationwide number, when of course that's electorally irrelevant. I assume those polls are looked at on a state-by-state basis.
PH: No. All we're trying to do is to reflect what's happening in America. We're not trying to predict the Electoral College.
RCR: You're not even breaking those down on the state level then?
PH: You can't. The sample's too small. With 1,000 people, California is the largest portion, and that's 10 percent, and that would be 100 bodies. With Iowa we may have seven people.
River Cities' Reader: What can we learn from the 2004 election?
Peter Hart: First and foremost, we talked about this election in some respects as transformational. It wasn't transformational at all. You compare 2004 to 2000, 47 of the 50 states voted exactly the same way. And essentially three states that changed - New Mexico, New Hampshire, and Iowa - if you take 10,000 votes, all 50 states would have ended up with exactly the same result, and we spent how many of billions of dollars on this election? I mean it was just an astronomically large amount, and we had a tremendous turnout, which meant interest was good. But in terms of it being a transformational election, it certainly was not. What was important is the balance of power in terms of the United States Senate clearly changed. The Republicans went from 51 to 55. They control the House by a slightly larger margin, so they have all instruments of government at this stage of the game, and that gives them exceptional amount of leverage.
The second thing is there were groups within the electorate that became important and showed us differences. Married women changed it in terms of how they voted. They became more Republican. The second group would be Hispanic voters. They were overwhelmingly Democratic four years ago. In this election, 53-44, not overwhelming, and that is a fast-growing group. And the one group that is exceptionally important are the religious voters, and those are people that go to church on a weekly basis. And when I say religious voters, I don't mean white evangelicals. Forty percent of Americans go to church on a weekly basis. Those people voted by a 22-point margin for Bush over Kerry. They account for one-half of all the votes that George Bush received. The president got about 44 percent of the rest of the electorate. That group becomes exceptionally important, and that's a huge change in terms of where things are and what's important.
Essentially the president lost this election on the issues and won the election on the basis of personal values and personal characteristics. Voters saw him as consistent; they saw him a religious person, a man of faith; and they saw him as easy-going and likable. Those are the three elements that worked for him the best and in addition the issue of terrorism.
If there were no terrorism issue, the president would have lost. Seventy-six percent of the voters say that I'm concerned a great deal or quite a bit that there will be a terrorist attack. Among all voters, 57 percent say I have a great deal of confidence in terms of Bush and his ability to handle terrorism. The like number for Kerry was 40 percent. That essentially became the election.
And the final elements obviously were social issues and moral values, and I think the president benefited from that.
One of the things we've been talking about is the importance of the Mississippi River, and if you look at the states along the Mississippi River, you get a sense of the changing nature of American politics. Missouri was a toss-up state last time, solidly Republican this time. Iowa was slightly Democratic and has been Democratic since 1988, and this time flips over. And now Minnesota and Wisconsin are very close to the edge.
RCR: There's a huge cultural difference between Missouri and Minnesota and Wisconsin - .
PH: If you go from Minnesota, the top of the Mississippi River, all the way down to Louisiana in the Gulf of Mexico, essentially what I call this is the lifeblood of this country. Others have said it's the spine of America. It's 2,350 miles. You look at this group of 10 states, and since 1912 a plurality have always voted with the winner. That's pretty indicative. Louisiana and Minnesota are pretty far from one another, but when you look and recognize that Jimmy Carter carried nine out of 10 of those states in 1976 and I think Clinton did about the same in 1992, and Reagan carried nine out of 10 of them in 1980, these are states that obviously do switch and follow the trend in terms of American politics.
You could say some of them are red and some of them are blue; certainly they're much closer to what I call the middle. If you start to look at Minnesota and the rural areas, which used to be heavily DFL [Democrat Farmer Labor], they're a lot closer to looking like Missouri today than they are to looking like New York or California. So I think there's a changing nature, and I think a lot of it has to do with values.
RCR: A lot of the Democratic disappointment with results had to do with the early exit polling. We had two elections, one where the exit polling was wrong in the calling of Florida, and this year when the early results weren't indicative of what actually came through. What went wrong with those?
PH: I started my professional life in 1964, 40 years ago, working on the first exit polls that were ever done for a network. It was called the Voter Profile Analysis, and I went out and personally picked the precincts that would be a part of the model of the sample, and so I've watched the evolution of this and what has happened over the course of 40 years.
I believe that exit polling is one of the great tools that was ever invented, but it never should have been invented for calling elections. Its greatness is in terms of analyzing elections and understanding. The reason it is so great is that you can get somebody right after an event and before they know the outcome of the event and they'll tell you exactly why they voted the way they did and what was involved. Nothing can be more valuable than that.
It's different from watching the debate where people will listen to commentators and then say to you, "You know, I thought that so-and-so failed to fully answer that question." Well, they never saw that, but they are reacting to other external stimuli. Here there are no external stimuli. They come out and say, "This is what I did and this is why I did it." And that is exceptionally helpful both in the scholarly way and also in a reportorial manner.
The problem is, obviously, networks are not into scholarly things. They want to know first who won. So all the networks then built models after the way CBS did it. There was all this competition. And by and large networks got things right; there were always mistakes that were made, but they were made in a state race or in an inconsequential situation and never made in a major way.
And so then the networks said, "This is silly. We're each building our own float to bring it down the street. Let's all pool altogether, and then we can analyze it separately but there's no reason we have to send Peter Hart out to look at one set of precincts and Jeff Ignatius down to look at another set of precincts." So they all pool together.
The disadvantage of that, obviously, is that if there's a mistake made, it's made everywhere. That's what happened in 2000. They had a bad sample in the state of Florida, and nobody recognized it. They all accepted it, and they all made the mistake and called Florida wrong.
Does it happen? Yes. Does it happen in such a critical state in such a critical way? Usually not. But that's what happened.
The second thing that you need to understand. You're a restaurant reviewer. You go in and eat the appetizer and then rush out and give a review of the restaurant. Chances are you are not going to give a very complete and very good review. That's what early exit polls are. They're the appetizer. And at 12 o'clock or 1 o'clock, when they collected this data and put it out, it's a lot of garbage in that.
That's never been a problem because there were a handful of people, such as myself, on the inside of the network, who would look at all of this stuff and say, "Boy, it looks like a Reagan landslide today," or they'd say, "This stuff looks screwy." But nothing happened between 12 o'clock and 5 o'clock except for a handful of people on the inside. Dan Rather would come to me or Tom Brokaw would come to me and say, "What do you make out of this?" I'd say, "Sit on it Dan. There's nothing." Or: "Something is breaking and it is breaking in a very unusual way."
It's like the family. Same thing. When you all eat the appetizer at the table, and you're the reviewer, and you say, "What do you think?" "Boy, I'll tell you. It doesn't seem like they put out a very good Caesar salad, and they don't make their croutons. Then out comes the main course, and you four have a knockout main course. By the end of the meal, you say to yourself, "this is quite a good restaurant but the appetizers aren't so good." Fine. Nobody but the four of you have really discussed it.
Suddenly we have the advent of blogs and the advent of all of the Internet access. This family of four around the table becomes a family of 40,000 or 400,000. Suddenly they think they know the election results.
When Brokaw said to me, "What do you make out of this?" I said to him, "Be careful. It's early." I've seen senators go down in flames at the 1 o'clock level, but by 5 o'clock they're re-elected. I knew my polling was correct, but the people who were there early in the morning weren't an accurate sample. So that's number-one element. The early stuff is always wrong. It's just that everybody found out about it.
But there was the second problem, which was a more significant problem that was neither corrected nor anything else. They were way too female in their samples. There were 55 and 57 percent. I was sitting there with Bill McInturff, who is my co-partner on the NBC/Wall Street Journal poll, and we're sitting there with Russert and Brokaw. We didn't have a problem. We just looked at this stuff and said, "Ain't gonna happen. Too female." Males were plus-11 for Bush; females were minus-3 for Bush.
So then you have to ask how they got the gender wrong. I have no idea, but they did something wrong. My guess is that we probably have seen the end of exit polls.
RCR: Most people think of polling as this very exact science, but I think we've learned the last two elections that it is an art as much as anything. Or is it?
PH: Exit polling is nothing more than an open-book test. There's no way you should get it wrong. People are telling you what they finished doing.
Now there's one problem today that didn't exist 40 years ago and 30 years ago. Forty years ago, everybody went to the polls. Today, one-third of the people are no longer going to the polls. They're voting before the election.
And now all of a sudden I'm doing an election-day poll, I've left out one-third of the people and I say, "Oh, well I'm going to make up for that. I'm going to phone a certain number of people who voted pre-election, and we'll find out what they did." Well, you've lost the value of election-day polling, because they're no longer the same people they were when they exited the polls. So you've got that difficulty.
Now, pre-election polling is much more difficult. Always been more difficult. Why? Because you don't predict. All you're doing when you take a poll is understanding how people are at a point in time.
It is equivalent to me talking about your behavior. On your way to work you may sideswipe the car. You're a totally different person after that event. Same with an election. Events happen. People change their minds.
That's why we do polling. Polling is about telling a story, and where people are and what's happening. But the public always thinks a poll's a prediction. It's not a prediction. Only reflects.
Now when you look at the number of polls that were taken over this election period, and you look at the end results of all of them, they were amazingly accurate within the margin of error. What we're saying is: We can talk to 1,000 people in this country and within plus-or-minus 3 percent tell you what 120 million people are going to do. That's an amazing fact.
And our final poll had for NBC and the Wall Street Journal had Bush ahead 48-47. New York Times had it, I think, 50-47. Somebody else had it similarly. Most of the polls all showed the same thing. This was an amazingly close race, and in most cases they had Bush maybe ahead.
Let's suppose Kerry had won the 70,000 votes that switched in Ohio. Would I be wrong? No. All I can tell you is it's a close election.
Now, when you look at the work that was done by the Gallup organization in individual states, and they miss seven out of seven states, you have to question what is happening there. Essentially they drew a screen that was incorrect. A screen meaning: We looked at our respondents and we said certain people are going to be likely voters; we're going to eliminate people who we think are unlikely voters. Their basis was on a small turnout. It was a big turnout. Consequently they were screening out people who should have been counted in. So states that they thought were going to go Republican went Democratic. There were states that they thought would go Democratic that went Republican. They just did a bad job.
RCR: To determine likely voters, you ask questions such as "Did you vote in the last presidential election?" or other questions that you think will be an accurate indicator of future voting patterns.
PH: If you look at every major good polling organization, when they talk about "all registered voters" we're all talking about the same thing. Those are apples to apples. So you can go across all the polling organizations and, essentially, they should have the same thing. But when you talk about "likely voters," every organization has a way of defining likely voters somewhat differently than every other organization. Look at the difference between X and Y organizations, and in reality their differences are in the way in which they define likely voters, and since nobody knows who is correct or incorrect here, it turns out to be more of an art than a science.
And in the end the Gallup organization, I believe, has a major flaw. They have questions that they've used since 1950, including, "Do you know where your polling place is?" My son has no idea where his polling place is, but he knows he isn't coming home for Thanksgiving dinner unless he votes. And so he's apoplectic when he finds out that he hasn't gotten his absentee ballot, as he has no intention of going to the polling place. But to the Gallup organization, they say, "Oh, he's not a likely voter." Just an antiquated way of looking at the electorate.
RCR: Do news organizations use polling properly and does the public interpret it properly or understand it properly?
PH: Do news organizations use it properly? No. News organizations look at their job as promoting the news or being able to develop a new story. And therefore their tendency is to use polling to try and go in and get something at the highest high or the lowest low. And that's abnormal.
It would be equivalent to going out and measuring the water table of the Quad Cities area the day after a major thunderstorm and then rushing out reporting the new water table. Well, we rush in right after the convention or rush in immediately after a debate; we're not letting opinions settle and come back to the proper position. And therefore we set up something that is totally artificial, which is: "Look, the water table is way up." And then we come in a week and a half later and say, "Oh, this guy has dropped like mad." They never dropped. It was an artificial expansion, and it dropped back to its original level.
We did polling this year as we have done for the last 15 years for NBC and the Wall Street Journal. They have a different way of approaching it. We go in, and we poll every five to six weeks. We purposely avoid the major events. We look for the calmest period.
If you take all of our polls from the middle of March through the end of October, we never had Bush with more than a five-point lead, and we never had him less than dead even. So, did the election go into these plus-12 and minus-3s and whatever else it did? Sure it did. For days or hours but never as a whole. I think that news organizations in that way failed the readers.
The second way in which they failed the readers is we get caught up in horse-race journalism, and we really failed to explain the dynamics. The New York Times does a very good job, the Pew organization, the Wall Street Journal does a very good job, but an awful lot of organizations just are interested in putting out whatever that sort of odd little figure is that sells newspapers.
RCR: You mentioned the way NBC and Wall Street Journal do their polling. Is that your choice, theirs, or a combination?
PH: Ultimately it becomes their choice. But I think it says something about the standards.
And one of the things that we've done for them over a period of time is these huge quarterly polls where we discussed one issue in-depth, and the Wall Street Journal took out eight pages of the paper to discuss the issue of values, the issue of race, etc.
RCR: You mentioned horse-race journalism. Is polling more prevalent now than it was 20 years ago?
PH: Tremendously more. I mean what I call it is poll proliferation. Every organization to get their own bona fides wants to go out as to do a poll.
Newsweek is one of the worst . They try and figure out how they get their poll out in a way that hypes their newsstand sales for Monday. They put it out on Saturday, and they look for that spike in order to be able to help them at the newsstand.
RCR: Is there anything inherently bad about polling? Or is it the way it is used by the politicians and the media?
PH: You can't spend a lifetime doing what I do and say it's bad. It's a marvelous tool, and it can be used correctly or incorrectly. But as a way of telling our story, and a way of understanding where the consensus of the public is, it becomes exceptionally important.
Take two of the major events in my lifetime; one is Viet Nam and the other is Watergate. In both of these instances, the public was way of the politicians, and by having public-opinion polling, I think it helped those in public life understand that we wanted to be in a different place than we were. And from that point of view it was exceptionally helpful. I believe it's helpful during the Iraq war. I think it's telling the story of a very divided America, and from that point of view I think it's going to provide guidance for the administration to recognize they have to find a solution and a way out here. They do not have an unending amount of time from the public.
RCR: Most of your polling looks at a nationwide number, when of course that's electorally irrelevant. I assume those polls are looked at on a state-by-state basis.
PH: No. All we're trying to do is to reflect what's happening in America. We're not trying to predict the Electoral College.
RCR: You're not even breaking those down on the state level then?
PH: You can't. The sample's too small. With 1,000 people, California is the largest portion, and that's 10 percent, and that would be 100 bodies. With Iowa we may have seven people.
