You are viewing:


Information released online from January 20, 2009 to January 20, 2017.
Note: Content in this archive site is not updated, and links may not function. External links to other Internet sites should not be construed as an endorsement of the views contained therein.

printable banner

U.S. Department of State - Great Seal

U.S. Department of State

Diplomacy in Action

Polling: What the Numbers Tell Us

Peter Brown, Assistant Director, Quinnipiac University Polling Institute
New York, NY
October 7, 2016


MODERATOR: Good morning everyone, and thank you for joining the New York Foreign Press Center teleconference today. We have us – with us today Mr. Peter Brown, the Assistant Director of the Quinnipiac Polling Institute. He serves as a chief spokesperson for the Florida, Iowa, Ohio, and Virginia polls, and is instrumental in developing, analyzing, and presenting the results of Quinnipiac’s poll. Mr. Brown has more than 30 years of experience as a political journalist and editor in Washington, D.C., New England, and Florida, and he’s here with us today to discuss the latest Quinnipiac election poll results released this week that were conducted in swing states. Today’s briefing is on the record. And so with that, I’ll turn it over to Peter.

MR BROWN: Good morning. I thought I’d start the discussion just talking about where we are in the presidential race right now with only a little bit more than four weeks to go. As you know, there’s a debate Sunday night, and debates are obviously the kind of events that make it possible to have election-changing influence. Right now, it’s a pretty fair consensus of our polls and other people’s polls as a group that Secretary Clinton seems to have a small lead, maybe three, four or five points nationally. Now, as you know, this is not a national election; it is an election of the 50 states. So – but measuring on a national basis, just to give you a feel for where we are, it’s probably fair to say that Mrs. Clinton has a 3-to-5-point lead over Mr. Trump nationally.

In some of the big states, what we call the swing states because they swing back and forth from one party to the other party from election to election, there are some results that are just plain different. For instance, we at Quinnipiac have a product that we produce of four key swing states – frankly, the four biggest swing states – that’s Florida, North Carolina, Ohio, and Pennsylvania. And this week we came out with a new swing state poll covering those four states, and here’s what we found.

In Florida, Mrs. Clinton has a 5-point lead over Donald Trump. Now, that’s a big change from the last poll before this which was done previous to the first debate, and that has shown that Mr. Trump and Secretary Clinton were dead even in Florida. In Ohio, this week the poll found that Mr. Trump had a 5-point lead on Mrs. Clinton, and that’s a little bit appointed to better for Mr. Trump than the previous poll had been. In Pennsylvania, Mrs. Clinton had a 4-point lead over Mr. Trump. Again, that’s a point or so better than two weeks ago but not statistically meaningful. And in North Carolina, Mr. Trump – excuse me – in North Carolina, Secretary Clinton’s 4-point lead has become a 3-point lead. So we enter the next debate with Mrs. Clinton slightly ahead but not gigantically ahead.

And the other interesting question is how the debates will affect the presidential race. As most of you I’m sure are aware, following the first debate there seemed to be unanimous sense among the media and pundits that Mrs. Clinton had won the first debate easily, and in fact, by margins of more than 2 to 1 in these four states. We found that voters thought that Mrs. Clinton had won the debate, but even though they had felt she had clearly won the debate, her leads were basically not very different except in Florida. And again, in Ohio, she went – fell further behind.

So we look at Florida, Ohio, and Pennsylvania because since 1960 no President of the United States has been elected without carrying two of those three states. And this year – excuse me – this year we added in North Carolina because North Carolina has now become very, very split politically. It’s not – it’s interesting to look at North Carolina, not only is there a presidential race there this year but there’s a race for governor and a race for the U.S. Senate in North Carolina and they’re all very close.

This should serve as an outline. I’ll be happy to take as many questions as you gentlemen or ladies might have.

OPERATOR: Thank you. And ladies and gentlemen, if you have questions, you may queue up by pressing * followed by 1. Once again, for your questions you may queue up by pressing * followed by 1 at this time. And our first question will come from Sanna Bjorling with Dagens. Please go ahead.

QUESTION: Hi. My name is Sanna. I'm a Swedish reporter; we talked the other day. I just wondered, do you – I can see the samples are – you have (inaudible) in Pennsylvania (inaudible). I just wondered, do you have any idea of different counties or are – is everything in statewide?

MR BROWN: Well, obviously, some counties are more typical of others.


MR BROWN: One – in Pennsylvania one of the things that has tended to be true the last many elections is that the Philadelphia suburbs have become a critical battleground between the two parties. It didn’t used to be that way, but those suburbs have grown dramatically and they’ve become a little bit more Democratic over time. And so both campaigns concentrate heavily in Pennsylvania on those suburbs. Now, they concentrate in other parts around the state also, but the Philadelphia suburbs are thought as to be a key area.

QUESTION: Do you follow them in surveys? Do you poll them specifically by county?

MR BROWN: Not by county, no.




MR BROWN: No, we do not.


MR BROWN: That would be almost impossible to do that, just because in the normal 1,000-person sample, those counties are only a small part. And you just wouldn’t have enough money to adequately analyze.

QUESTION: That’s what I thought. Thank you.

MR BROWN: You’re welcome.

OPERATOR: Thank you. Again, for your questions, you may queue up by pressing * followed by 1. Next in queue is Mladen Petkov with Bulgarian National Radio. Please go ahead.

QUESTION: Hello. Thank you so much for doing this teleconference. And I have a couple of questions. First of all, I wanted to clarify whether you are speculating that winning a debate does not guarantee a lead in the polls. And second of all, what kind of statements have the potential to change opinion in this election cycle? Because we see Trump’s comments usually don’t hurt him. And do you have any – during your research, can you talk about statement that change opinion polls? Thank you.

MR BROWN: Well – okay. Let’s start with the first question, which is: Do debates make a difference in the actual horserace, the voting? What we found, again, after the first debate, was that Secretary Clinton was universally thought to have won that debate and won it easily. Yet the poll taken right after that debate happened – and it was a weeklong poll – found that there had been relatively small increase in support for Ms. Clinton, but not a big increase, not nearly as large as one might expect compared to the overwhelming consensus that Secretary Clinton had actually won that first debate.

Can you repeat the second question? I’m sorry, I --

QUESTION: Yeah. The second question is, in terms of do you – when you make the polls, when you do your research, do you have any idea of the statements from both candidates that have the potential to change opinion, especially in the swing states?

MR BROWN: You mean --

QUESTION: What kind of statements can actually change the opinion in – but – for voters?

MR BROWN: Are you asking me – are you asking me whether we use those statements as part of the questionnaire?

QUESTION: Yes. Do you --

MR BROWN: The answer to that is no.

QUESTION: Yeah, no?

MR BROWN: No. I mean, that would be what they – that would be a no-no in polling land.


MR BROWN: That would be --

QUESTION: Just to follow up, do you have idea of the issues that are very important in this presidential campaign that you think that might influence voters?

MR BROWN: Well, I think it’s pretty clear that the Clinton campaign is basing its efforts on trying to convince average voters that Mr. Trump is not fit to be president. Mr. Trump is running on two basic themes. One is Secretary Clinton is a crook. He uses the line, obviously. And number two, there’s enormous sense of support in this country for change. The question is: What kind of change? It’s not hard to see Mr. Trump as being a perceived – a potential, a potential change agent. He’s running against the Washington establishment. He’s not only running against Democrats in Washington, he’s running against Republicans in Washington. And so that’s kind of what we have.

The economy is always the biggest issue, so to speak. And what you saw in the first part of the debate – in the first debate was they tangled over the issue of the economy. And most people who – most of the pundits who watched the first debate thought that Mr. Trump’s best part of the debate was on that section on the economy. There are others who think he didn’t do very well at all.

But again, voters don’t always vote based on what the candidate – the issues they agree or don’t agree with. Much of it is also a sense of what they represent. And in that case, that Mr. Trump, although clearly slightly behind, has an opening, if he can do it – present himself as the candidate of change. This is an election where roughly 7 out of 10 voters say they’re not happy with the direction of the United States. And so both Mr. Trump and Secretary Clinton are trying to tailor their messages to that point of view.

QUESTION: Thank you.

OPERATOR: Thank you. Our next question will come from Melissa Sim with The Straits Times. Please go ahead.

QUESTION: Hi, thanks for doing this. I just want to ask: Is it typical for a candidate to win the debate and us not see any change in the polling? And is this likely to happen for the debate coming on Sunday – if she clearly wins again, is it likely to happen again that we won’t see a change in the polling?

And my second question is on North Carolina. Are there issues there that you feel North Carolina voters have been polled on and say that those are their top issues?

MR BROWN: At this stage we have not – we’re not doing issue polling anymore. It’s just too close to the election and it’s too hard to find a large enough sample of voters on specific issues.

And your first question – I’m sorry.

QUESTION: Is it typical for a candidate to sort of win the debate but not see a change in the polls?

MR BROWN: It’s not like in the past there have been debates and one candidate won by humongous margins in terms of public perception of the debate performance and then didn’t do anything at all. But in this election, because it’s such a split electorate – as I’m sure you’re aware, there are just very strong divides in this country on both cultural and political issues – that it’s not clear to me there’s a relationship. And again, this is – one can argue this is less likely than many other elections to be an election on the issues. Change is the big deal here, and that’s what the Trump people are pushing.

On the other side, the Clinton people, understanding that they’re on the wrong side of the trade issue – excuse me, on the wrong side on the change issue --

QUESTION: Right. So can I ask --

MR BROWN: -- they are spending – hold on, let me finish. They are spending a great deal of effort to convince voters that even if they don’t – even if they want change from the Democrats, that Mr. Trump is just not, in their view, fit to be – to sit in the Oval Office. And that’s what you’ve got going on.

QUESTION: Okay. I think what I’m trying to find out is whether you’re surprised that there was little change in Mrs. Clinton’s polling numbers even though there was what you said – a clear indication that she had won the debate – and if this would continue in the next debate, if you think that that’s going to continue (inaudible).

MR BROWN: The uneven results of the debate is expressed in the horse race after the first debate. In other words, Florida, Mrs. Clinton did better; Ohio, Mr. Trump did better; and in North Carolina and Pennsylvania, they’re close, although Mrs. Clinton has a small lead. It’s not that nothing happened out of that debate, but it wasn’t very much compared to what many had expected, because there was a consensus – it does tell you something about the consensus of pundits. But there was a consensus that Mrs. Trump – that Mrs. Clinton had won that first debate, and in the Quinnipiac poll that we’re talking about, by a 2 1/2-to-1 margin, roughly, voters said Mrs. Clinton won the debate.

OPERATOR: Thank you. Our next question in queue – that will come from Tim Rahmann with the leading business German weekly. Please, go ahead.

QUESTION: Thank you very much. Peter, I was wondering, how convinced are you that your poll numbers will be close to the actual results? I mean, we have seen in the past years that poll numbers do not necessarily reflect the actual results. We have seen that in a spectacular way in the United Kingdom with the Brexit referendum, and experts are saying it’s tough to get a clear picture with telephone interviews, and it’s especially tough to catch anti-establishment voters. Can you say a few words about that?

MR BROWN: Sure. We obviously think we do a good job – excuse me – and history shows we do a good job. If you compare how our last poll, for instance, four and eight years ago, against what happened, we were very close. You are correct, however, that there are increasing obstacles to good polling, because there are different methodologies used. Much of them have to do with what technology you use. We at Quinnipiac and the major TV networks and newspapers – New York Times, CBS, Washington Post/ABC, Wall Street Journal, NBC – use what’s called a random digit dial message. We use human beings to make our calls, and so we’re very comfortable. But other people use other methods. I don’t want to get into discussing other pollsters, but there are, for instance, pollsters that have computers make the calls. We obviously don’t do that even though it’s much less expensive a methodology, but we think what we’ve done has borne the test of time.

And as for Brexit, we poll in the United States. We just poll in the United States. We know what we know and we know what we don’t know. And obviously, I understand your point about the polls and Brexit, and there have been some other international polls on different issues that perhaps haven’t been as accurate as some would have liked. I can’t comment on those. But on our stuff, I feel very confident on our stuff. I mean, you need to remember the concept of margin of error. Each poll has a margin of error depending on the size of its interview base; and so if we have a poll that says that candidate X is ahead of candidate Y by three points and the margin of error is two points, then that’s a different result than if the margin of error is four points. You understand what I’m saying. So that also plays into it.

And the other big deal in the United States polling – I don’t know what it’s like overseas – is that we at Quinnipiac and most of the other major organizations – the ones that, again, use human beings to make the calls – we call cell phones, and that’s enormously important in this kind of day and age in which fewer and fewer people have landlines. So one of the things we try to do is match the number of our cell phones to the percentage of the electorate that has cell phones.

OPERATOR: All right, thank you. Our next question in queue, that will come from David Smith with The Guardian. Please, go ahead.

QUESTION: Hello, I just wondered how important you think early voting is and whether that combined with if Trump performs poorly again in the debates, could that really make this unstoppable for him in terms of losing? I mean, is it that crucial on Sunday?

MR BROWN: Well, the debate’s obviously very important because there’ll be a very large number of people will be watching. You only get so many opportunities to talk directly to the electorate when they’re paying attention and the – I think the numbers out of the first debate were in the mid-80 million. That’s a lot of eyeballs. So obviously, the debates provide opportunities to the two candidates.

I’m sorry, there was another part of your question. It slipped my mind.

QUESTION: The first – it was just about early voting.

MR BROWN: Oh, early voting. Many states have early voting. Some do not. I live in Florida, for instance, where there’s early voting.


MR BROWN: And a large percentage of the electorate will have voted by Election Day. I typically vote a couple weeks in advance. I just, when my ballot comes in, I fill it out and mail it back. And so there are, again, states like Florida that have large numbers of early voting. So does Ohio have a reasonably large number of early voters. And that’s just a fact of life you have to deal with these days in politics. It’s just the process has changed somewhat for a variety of reasons, and the early voting is certainly part of that.

OPERATOR: All right, thank you. As a reminder for questions, you may queue up by pressing * followed by 1. Next in queue is Gretel Johnston with the German Press Agency. Please, go ahead.

QUESTION: Hi, thanks very much. My question is actually back to the pool of people that you poll. Is there any way of saying what portion of that pool voted in previous elections? That’s one question I have.

Another question is basically about the --

MR BROWN: The simple answer to that is we asked them if they voted – have voted in past presidential or gubernatorial elections, depending on the election.

QUESTION: So do I understand then you would have the result for that? You would actually know?

MR BROWN: We – no, we ask people. In other words, part of our – part of our screening method and one of the ways we determine who’s a likely voter is we ask them their voting history.

QUESTION: And so you’re releasing the poll as people who are likely to vote. Do you also include whether they voted in previous elections so that you can say whether they – what they’re – whether they’re new voters for this time around or not?

MR BROWN: Okay. No, what we do when we do a poll is we ask the people we contact whether – about their voting history. I don’t want to get into the specifics of it. And we use that as a way of screening that part of the electorate to find out whether we consider them likely to vote. We don’t ask someone – in other words, we’re asking informational questions of them about their voting history, just as we’re ask them informational questions about who they’re going to vote for this time. And we – but we use the informational history on their past voting behavior as one of the factors in whether they – we think they’re going to be a likely voter.

QUESTION: Mm-hmm. Well, if I can ask it a slightly different way, I’m just curious about, among pollsters generally, is there concern that the data you’re getting is not going to be as accurate as you think because you’re not tapping into a large number of voters who just haven’t been on the lists in the past or they’re not people who are --

MR BROWN: But we don’t use lists.

QUESTION: So that’s going back to what you said about random, randomly selected numbers.

MR BROWN: Right. Well, without getting into methodological details that would bore you to sleep, we buy cell – we buy phone packages of phone numbers that have – both cell phone and landline that adequately represent the geography and the demography of the state of – whatever state we’re polling. So we ask people – for instance, on party ID we – some pollsters – not very many, but some pollsters only, for instance, will call a certain percentage, a preset percentage of Republicans, Democrats, and Independents based on past state-wide voting. We don’t. We ask someone when we get them on the phone. Among our questions is, “Are you registered to vote in the jurisdiction we’re calling about?” Number one. And we also – I don’t – let’s just leave it there, it’s getting – it gets very complicated.

QUESTION: (Laughter.) All right, all right. Well, the other thing I was wondering is about the debate on Sunday. I just would like to know if you could say what you think Clinton would need to do in order to change some of her – to change her numbers in Ohio?

MR BROWN: Pretty much what she needs to do in most – in some of the other swing states. I mean, her – as I mentioned earlier to one of the other people, the thrust of Mrs. Clinton’s campaign is to convince voters in Ohio and Iowa and Kentucky – any place that – to convince voters that Mr. Trump is not fit to be President. The tone of her television ads make that a point as many times as they can. And I don’t think that’s hugely different. Ohio seems to be better for Mr. Trump than a couple of the other swing states, and the theory is that because there are a sizeable percentage of white male voters without college degrees, which that’s a bigger – that’s a group that Trump has been doing well among, that Ohio might be an especially friendly electorate (inaudible). But it’s also true in other – in some other states.

OPERATOR: And sir, are you ready for the next question?


OPERATOR: Thank you. That will come from Gregory Ho with Radio Free Asia. Please go ahead.

QUESTION: Hi Peter. My question is about – based on the national surveys that Mrs. Clinton has a three to five-point lead, so is it fair to say that for those four battle ground states that Mrs. Clinton, for her, she just need to win only two, but for Donald Trump – Mr. Trump need to have to win those five states to secure a presidency at this moment?

MR BROWN: The way the Electoral College stacks up and which states are locked for – with one candidate or the other, it is true that it’s an easier climb for Mrs. Clinton. If she could win Ohio and – excuse me, if she can win Florida and one of those other three states we mentioned, the chances are very, very good she’ll win the presidency. It’s not 100 percent locked, but it’s very good.

QUESTION: And my – okay. Go, go. Yeah, please.

MR BROWN: But – but the thing about trying to superimpose national polls on state contests is that it only takes a movement of a couple points in the national polls to change many, many states. Actually a couple who I want – again, it would only take – if she’s ahead three to five points – let’s say for the sake of argument, Mr. Trump can get himself back to equal on – in national polls, which is where he was before the first debate. They were just flat out in a dead heat before the first debate, again, only a little more than a week ago. So if you can move the numbers nationally, they in and of themselves then move the state numbers. Am I being clear about what I’m trying to say?

QUESTION: Yeah. Mm-hmm.

MR BROWN: So it doesn’t take a gigantic shift. It takes a consistent shift.

QUESTION: My second question is about the undecided voter. How big the portion has been show on your result?

MR BROWN: It depends. But on state by state, it’s in the mid to high single digits in most cases is the undecided. And that doesn’t – but that doesn’t count the people who say they’ll vote for Jill Stein or the people who say they’ll vote for Libertarian Gary Johnson.

And so the quest – they’re a bit of an unknown. We don’t know what percentage of them will actually vote, and if they do, who they’ll vote for.

QUESTION: And the third would be: Would the third party become an issue in this election or are we just ignore the third party? Ignore the libertarian and --

MR BROWN: There is no – there’s no third party as such there. I mean, I – Ms. – Dr. Stein is representative of the Green Party and Gary Johnson’s representative of the Libertarian Party, but those are tiny parties in and of themselves. Certainly in the case of Mr. Johnson, he’s going to run better than the Libertarian Party would – will do in other offices in some – in many localities. But what – all I’m trying to say is when you start taking on undecided voters in the presidential election, there’s a group that is not officially undecided but history tells us will shrink as we get closer to the election and are less likely to stay with their third-party candidate than shift at the end.

Again, Johnson got 1 percent of the vote four years ago nationally – 1 percent. He’s now getting 5 to 7 in a lot of polls, and in some states he’s doing very well. He got 24 percent in a New Mexico poll. Now, he used to be governor of New Mexico, but still.

OPERATOR: Thank you very much. The next question will come from Asaf Liberman with Galei Tzahal Radio. Please go ahead.

QUESTION: Yes, thank you very much. I wanted to ask exactly about that, about the differences in the scores between the two-way race and the four-way race, being asked about Stein and Johnson. Which of the two different polls is more important?

MR BROWN: Well, we now consider the four-way race the most important, and that’s closest to what the ballot would look like when a voter walks into the voting booth. There are some states where – well, let me think. There are some states where Jill Stein is not on the ballot, but she’s on most of them. So that when a voter walks in, the choice is not going to be just Hillary Clinton and Donald Trump. It’s going to be Hillary Clinton, Donald Trump, Gary Johnson, and in many, many states, Jill Green (sic), and then there’ll be all kinds of other candidates you’ve never heard of that have gotten on the ballot in various states. So that we at Quinnipiac have decided that the four-way is the most accurate at this point.

OPERATOR: Thank you very much. The next question will come from Guan Wang with CCTV. Please go ahead.

QUESTION: Hi there. My question is: We heard Trump campaign saying that the GOP-leaning voters are less likely to respond to polls than they would like to respond to – than the Democratic-leaning voters are responding to polls. So do we have any statistically significant numbers suggesting it that way?

MR BROWN: Well, that’s their theory and they’re – we’ll find out. There’s one poll that matters and everybody in the polling business will wind up comparing their final poll against the actual results, and that’ll tell you who’s been right and who’s been wrong and which methodology has been the most accurate.

QUESTION: But in terms of – I’m sure you’ve been in this business for a while. Do you think the conservative – more conservative voters, people, psychologically have a less – have lesser of a tendency to respond to polls? Do we know that?

MR BROWN: I don’t know that.

QUESTION: Okay. Thank you.

OPERATOR: Thank you. Once again, for questions, you may queue up by pressing * followed by 1. Again, *1 for your questions. The next question comes from Mladen Petkov with Bulgarian National Radio. Please go ahead.

QUESTION: Hello, can you hear me?


QUESTION: Yes. So I wanted to clarify something. When you – when we look at the polls, which indicators do you think we should pay more attention to? And also from the swing states, where do you expect to see very close results? Is it North Carolina or Ohio?

MR BROWN: Oh, I think all four of those are going to be very close. I don’t think one is inherently more likely to be closer than the other. Florida is obviously the one with the history. I mean, not only do you have the 2000 situation where less than a thousand votes decided Florida and the presidency, in the last couple of Florida elections both for president and for governor, all of those elections have been extremely close with a margin of roughly 1 percent, all well within the margin of error.

QUESTION: And again, when we look at polls, which indicators do you think we should pay more attention to? Because there’s so many results now on TV, on radio, and newspapers, and they have different data. Which indicator do you think is very important at this point?

MR BROWN: I don’t – I can’t – I, as a matter of policy, don’t comment on other polls, but I do comment on methodology. And I would give the strongest consideration to those polls that use human beings to make the telephone calls, not computers. I would use one that has a representation of cell phones roughly equal to that of the population you’re surveying. And those are the two really big things.

Obviously, sample size is important. The larger the sample, the better.

QUESTION: Okay. Thank you.

MR BROWN: You’re welcome.

OPERATOR: Thank you very much. Once again, for any questions, you may queue up by pressing *1 at this time. Again, *1 for your questions. And on a few moments here, I’m showing no additional questions in the queue. Please continue.

And presenters, if you have any closing comments?

MR BROWN: Thank you. It’s been a pleasure.

MODERATOR: This is Daphne from the Foreign Press Center. I want to thank you, Peter, for participating in today’s call and the journalists that were on the line. The transcript will be posted to our website and sent to everyone who participated in today’s call as soon as it is ready. The call was on-the-record, and thank you once again. Have a good day.