Should You Trust Those Polls? We Talk with Expert Kathleen Frankovic
With months until the 2024 election when our votes will determine, not only who occupies the White House, but which party controls the Senate and the House of Representatives, as well as many offices in state governments, polls dominate the headlines. The latest ones show that President Biden and former President Trump are running neck and neck in many of the battleground states where, because of the Electoral College, the election may be decided. How accurate are these polls? Which ones can be trusted and which are suspect? Who creates these polls? Who participates? And do polls have the power to alter the way people will vote?
For answers, we consulted with Kathleen A. Frankovic, one of the world’s leading experts in public opinion polling. She spent more than three decades as Director of Surveys at CBS News. Since she retired, she has been an election and polling consultant for CBS News, Harvard University and now with the YouGov/Economist Poll and other research organizations. (The following has been edited for clarity and length.)
Tell us about the history of polls and how they have evolved.
You might be surprised to learn that, in the United States at least, there are examples of people trying to figure out who’s going to win an election by conducting polls that go as far back as 1824. And there’s always been news media involvement. In the early days, sometimes it was partisan newspapers, (most newspapers then), trying to show that their candidate might be favored by certain groups of people. But as time went on, polling basically became the landscape of pre-election periods.
In the late 19th century and the early 20th century, they were done by newspapers and news organizations sending out a massive number of questionnaires to registered voters, usually in their community and in neighboring states. They would report this information daily.
We’ve changed a lot in the last 90 years, starting with George Gallup and Elmo Roper and Archibald Crossley in 1935 and 1936 to where we are now. Back then they sent their interviewers to a random sample of households. Later on, polling became done by telephones. We had a huge proliferation of polls in the 1990s. It’s probably been cut back a little bit, but it’s still omnipresent in the media.
Why do we need polls?
The news media likes polls, but I think polls are also quite beneficial to society. When you read a poll, your reaction might be, “well, I don’t think the way most people think, or I don’t think the way the minority thinks.” We are in our little enclaves. Most of us have friends who think like us and I think it’s healthy for people to recognize that there are other people in society who might be very different.
There are reasons why polling is important Polls are the only way, apart from letters or phone calls or demonstrations, that citizens can transmit their opinions to the government. You can get this information to the decision makers. Polls can also assess whether or not a government policy is liked, disliked, or effective with the public overall. There was a lot of polling done around the Affordable Care Act, Obamacare. You could see that, at least at some point, support grew as people received the ACA’s benefits.
If you’re in leadership in government, you really don’t want to be acting on untrue information. A poll is information that you can use. I do think that it serves the purpose of a feedback loop from people to government to people and back again.
How did the approval rating poll begin?
George Gallup, back in the 1930s, invented the approval rating. “Do you approve or disapprove of the way(at that time) Franklin Roosevelt is handling his job as president?” We still ask that question the same way. We may have moved our way of asking it from in person to on the telephone to online. But we do ask the same question. A democratic society needs to have some measure of what the public thinks and this is the way you get it.
Aside from news media, there are organizations, some on the far right, that are doing polls now and getting publicity to reinforce their views. Which polls should we trust?
That’s not a new phenomenon. I would say that goes back about 30 or 40 years. But there are several ways to manage this. The American Association for Public Opinion Research is one of them. It’s been around since 1947 and is a meeting place where people can share information. It doesn’t do it often, but it can take on groups or individuals who are doing the sort of thing you’re talking about. It’s a measure of how important people think polling is that people are out there making it up.
(Photo by Bigstock)
How are the polls conducted now? If over the phone, landlines or cellphones? Is there polling on the internet?
Polls are conducted in a variety of ways. Since the 1990s, mobile phones have been included in most major public opinion surveys. Nowadays there may be more interviews conducted on mobile phones than on landlines because fewer people have landlines and don’t answer their landline if they have one. The telephone is a lovely way of polling: you can feel comfortable about the people you’re interviewing, and it remains possible to draw a random sample of telephones in the United States.You know the existing area codes. Then you then have to draw a sample from that and make the phone calls. That’s one way of polling that’s been around — and working — since the 1960s.
However, I would say the more common method today is online, basically bringing people in somehow to a sample and having them complete a questionnaire on their phones, on their iPads, or on their computer. This model has been around since the 1990s, but there are potential problems that arise. You really cannot find a random sample of internet users without spending a lot of money and resorting to old fashioned ways of conducting random samples.
But online is extremely popular. It’s way more popular than telephone polls. The respondent can manage the time to do it, when you do it. You can start it and finish it later. It basically gives the respondent a lot more control. That’s a good thing, but there is opposition to online polling, as many are not probability samples, true random samples. That is difficult and costly for an online survey to achieve.
What type of person participates in polls?
The goal is to have it be all sorts of people — everyone. But response rates, whatever the mode, whatever way you do it, they’ve always been lower among younger adults, lower among minorities, lower among people whose first language is not English.
There are occasionally political differences in the kinds of people who answer questionnaires. That could be in part because candidates yell and complain about polls, say they’re fake, and that could influence their supporters.
Do polls have the power to influence an election?
I do think that’s true in one particular case. If you discover that your favorite candidate, your first preference, is getting minimal support and is nowhere near the top of the list, you might consider, and I think people do, changing your vote, changing your support to your second favorite candidate who may, in your opinion, from what you’ve seen, have a chance of winning.
This is a problem that’s present with third party candidates in the United States. We’ve seen that in cases of people liking Ross Perot in 1992, or George Wallace, in 1968, and maybe even this year with Robert Kennedy, Jr. Candidates like this, who are neither a Democrat nor a Republican, or not running as a Democrat or a Republican, may start out higher in the polls than they end up.
So yes, polls can influence voters, but I can argue that in this case that’s probably a good thing because the voter is now making a rational choice and a more informed choice. But in terms of the typical two candidates, Republican versus Democratic race, there’s not a lot of evidence that polls make changes in people’s vote.
The two-candidate race does at least allow for an honest comparison and choice between the two candidates. And polls are not going to change that. But people talk a lot about the possibility of voters changing their minds because of polls, even though we don’t have any really big evidence on that.
So basically what you’re saying is that by this time most people have decided whether they’re going to vote for Biden or Trump and that’s not going to change?
Well, it’s only May. I used to say that until you get to Labor Day it’s not real. But this year, there are two candidates that people know very well. People have pretty much fixed opinions about them, so that might not be the case, and maybe we should be paying more attention to polls today.
Historically, these early polls are not perfect. They are not excellent. That’s also true when we’re talking about Senate races, and Governor’s races in states, if we look at polls conducted now. A lot of voters are not yet plugged into those contests, so polls asking about November done in May for those races, should be taken with a grain of salt.
What issues do you see driving voters in the 2024 election from what you’ve seen?
The largest share of people will say that their biggest issues are inflation and immigration. Inflation affects all kinds of people. Immigration is a much bigger issue among Republicans than it is among Democrats.
Economic issues have always been important. A good economy doesn’t always help an incumbent candidate, but it certainly can hurt. At this point, the fact that Biden and Trump are basically neck and neck in most polls is a real statement. It matters because voters have negative feelings towards both men.
Is the fact that our democracy might be in peril an issue?
It’s really hard to motivate voters about a virtue or about something that’s not tangible. We’re going to have to wait and see what happens or how it plays out in the next six months.
How does the Electoral College skew all of this? Are there more polls done in the battleground states than in other states?
There are and they are different from the national polls. Back in 2000, in the Bush versus Gore election, Gore got more votes nationally, half a million more, yet lost in the state of Florida, which determined who became president. We saw it in a much bigger way in 2016 when Hillary Clinton got something like three million more votes nationally than Donald Trump did.
But Donald Trump took enough electoral votes. You realize that the Electoral College, because it’s based on the number of Senators and Representatives gives a slight benefit, a slightly disproportional benefit, to people in small population states, most of which trend Republican. The Electoral College has a slight bias in favor of Republicans who are dominant in the smaller states.
All of this election denial business that we saw in 2020, how do you see that impacting this election?
I don’t know. I think it can motivate both sides. I think it can also limit the effectiveness of the number of Republican voters. If you don’t believe elections are free and fair, if you don’t believe your vote is going to be counted properly, if you think that the 2020 election was stolen, is that going to depress your turnout or is it going to increase your turnout?
I would doubt that I would have a sense of that until the fall. We just don’t know what’s going to happen between now and the fall in terms of how this is going to be used by both sides.
(Photo by Shutterstock)
How about reproductive issues, abortion, and now, you know, Trump said something the other day about birth control, which I think is going to send shock waves again?
Whenever it’s been on the ballot since the Dobbs decision (which overturned Roe v Wade, which had legalized abortion nationwide), the right to abortion has won, even in “red” states.. It’s even helped candidates. We don’t know how many candidates it’s going to help in December. It’s definitely something where Republicans and Democrats disagree, really disagree. (Republicans are more likely to favor restrictions on abortion, while Democrats oppose them.) It’s of more importance now to Democrats because they’re opposed to the Dobbs decision, and it seems that you can get more motivation if you’re fighting against something.
It’s interesting because back in the 1980s, and this was after Roe v. Wade, there really wasn’t a difference between men and women on the issue of abortion. Now there is. Some of that has to do with the differences between men and women when it comes to identifying as a Democrat or a Republican. That’s an interesting thing to look out for in the election.
What about polling approval ratings for some of our institutions? Because everyone’s saying that one reason why Trump has his supporters is because he keeps bashing our institutions. There are low rates of approval for a lot of our institutions, and I’m thinking in particular about the Supreme Court.
Supreme Court approval is very partisan. It’s extraordinarily partisan. It is about two to one approval among Republicans and two to one disapproval among Democrats. It’s huge, it just got huge and it’s stayed that way. Independents are also more disapproving than approving. It’s kind of interesting because you can ask people about individual justices these days, and most people will give you an opinion of many of them. Contentious Senate confirmation hearings for many of them, made them household names.
What percentage of voters do you think are going to remain undecided until right up until the election?
Right up until the election? Certainly under ten percent. In recent polls it has been 10 percent or less. Exit polls have always asked “when did you finally make up your mind who to vote for?” It’s a small number that say they decided on election day. This year, it could be more if something astonishing happens in the last week.
Do you see any effect with this hush money trial now in New York? The other cases aren’t going to go to trial before the election, but if he is convicted next week, any effect?
A poll, and I didn’t do it, so I don’t really know it all that well, says that there’s a chunk of people who say they wouldn’t vote for Trump if he’s convicted, and that it could be enough to turn the election around. It’s one poll done before the trial, so I’m not sure it still holds. Republicans haven’t deserted Trump yet. We don’t have a verdict yet and I think we’ll just have to wait and see.
Are you doing any active polling these days?
I retired from CBS in 2009 and I consulted for them for a few more years and I’ve been consulting for other organizations. One critical one I’ve been working with is YouGov which is based in California, but it’s an international survey research organization. It does do online polling with panels that they manage. Once a week, I get data and I analyze it and look at it and write up short summaries.
It’s an excellent position to have because I get to deal with current data and I try and make sense of it. I do that once a week, and then I work on creating the questionnaire with others at YouGov. I also have some other consulting jobs.
Any revelations in any of the data that you’ve seen?
There’s been so much. It’s been very helpful to be able to track how people are responding to the Russian invasion of Ukraine and how it affects their opinions of possible U.S. involvement. It has been very interesting to look at reactions to the Israel/Hamas war. Even though there are a lot of people who don’t want to increase aid to Ukraine, more people at this moment in time, are likely to say that they would increase military aid to Ukraine than say they would increase military aid to Israel. That was not the case several months ago. Americans are not often entranced by foreign policy issues. But they are paying attention now to these.
How did you get into this line of work?
I started out as an academic. I was a government major at Cornell, interested in Southeast Asian politics. I went to graduate school at a place that had somebody who specialized in the country that I was studying — Burma, now Myanmar. I got my PhD and taught for four years combined at Case Western Reserve and at the University of Vermont. I learned a lot about Vermont, a very small state.
I had met the political director at CBS News Martin Plissner, when I was up in New Hampshire for one of the primaries. He told me about an opening for somebody who was an academic and who specialized in public opinion, and I did. I came down to New York, had an interview, and was hired. I came to CBS News in 1977 and stayed. Later, my husband Hal and I lived in San Francisco and in Hawaii. But we’ve always kept an apartment here and now we’re back here to stay.
How do you keep yourself sane when you’re seeing all of this? And, and how do you keep your own personal feelings about some of this out of it when you’re analyzing all of this?
Yeah, that’s a good question because I guess I always have tried to do that and, and I’m pretty sure I’ve done a reasonably good job of that.
It’s just that you can’t shy away from highlighting something that might be beneficial to somebody you don’t like or harmful to somebody you do. You kind of go, “oh, I hate this story. I really don’t want to write about it.” But sometimes you have to. It’s hard, but you’re just conscious of it. You have to be conscious of it. And my involvement in national and international organizations, survey research organizations, has also helped because I sit on the professional standards committee of an international organization that’s based in the Netherlands. And, I’ve been president of the American Association for Public Opinion Research and the World Association for Public Opinion Research. So I’ve been engaged in a lot of these issues for a long time. If you don’t like writing about politics, you can always find something else to write about.
Besides politics what else do you use polls for?
I started looking at the history of public opinion polling in the United States and that’s been fun. I’ll come back to that again. I also love it when we get to ask questions about sports, entertainment, and Taylor Swift and Caitlyn Clark in particular. There are things that matter to a lot of people that people pay attention to, and we should be measuring them.
As we go into these last few months before the election, which polls should we pay attention to?
I would say that for the most part, all of those associated with any major media will be fine. And I’m going to say this that includes Fox News because the person who is the polling director there is excellent and has guided that poll so it really is fair and balanced.
I pay attention to CBS News, of course. I pay attention to NBC and the Wall Street Journal. I pay attention to YouGov, The Economist, CNN, ABC News,The New York Times. I’d look at the major non-profit foundations like PEW.
The New York Times is partnering with Siena College. We do seem to have a lot of smaller colleges in the tri-state area that are very much into polling. The Marist College Poll in Poughkeepsie, Quinnipiac in Connecticut — these are schools you might not have heard of, except for the fact that they do public opinion polling.
It certainly raised their visibility.
It sure did. It was an investment, but probably a very good one.
Is Gallup still around?
Yes, but Gallup is out of the election polling business. It does measure approval ratings. It is far more involved in market research and business training.
I wonder why.
Polls are out there and they are such an easy target for people when they’re annoyed. You just get used to it. Hell, (former Speaker of the House) Newt Gingrich criticized us back in 1994, and that was okay.
