Detecting the 'Artful Dodge'
IRA FLATOW, HOST:
This is SCIENCE FRIDAY. I'm Ira Flatow.
Henry Kissinger once joked at a press conference: Does anyone have any questions for my answers? If politicians had their way, they might just write their own questions for the press, but, of course, politicians can't write all the questions. So instead, they're coached on the art of question-dodging, taught how to segue from the question they're asked to the question they wish they had been asked and are prepared to answer.
Come October, how often will politicians pull that trick in their news conferences and debates? Will they be able to do it without us noticing it? And more importantly, is there a way to prevent this dodging and keep the debates honest? My next guest has a few ideas. Todd Rogers is a social psychologist at Harvard Kennedy School in Cambridge and the assistant professor of public policy there. He joins us from a studio on campus.
Welcome to SCIENCE FRIDAY, Dr. Rogers.
TODD ROGERS: Hi, Ira.
FLATOW: Now, you did a study looking at three different debate questions, but the politician gives the same answer to all three questions. Tell us about that.
ROGERS: That's right. So the objective of the study was to try to understand: How is it the politicians manage to get away with answering different questions than the ones they're asked without being punished? And so what we did was we recorded a politician being asked a question. The question was about: What will you do about the health care problem in America? And then we have the politician answer: Well, I'm glad you asked me that. There are so many important problems facing America today. We need universal health care because - and then he proceeds to ramble for a few minutes about universal health care.
So what we did was we spliced that video, and people watching the video were randomly assigned to three conditions. One condition watched health care question, health care answer. Another group watched illegal drug use question - what will you do about illegal drug use - health care answer: I'm glad you asked me that. We need universal health care.
ROGERS: And then a third group - yeah. And then a third group gets: What will you do about the war on terror? I'm glad you asked me that. We need universal health care. And so what we asked, we had participants view these videos, and then at the end we say: How honest, likeable and trustworthy are these candidates? And what question were they asked?
And so not surprisingly, they thought the person who answered the question about health care with an answer about health care, they thought he was honest, likeable, trustworthy, and they could remember the question he was asked.
But then when he was asked about illegal drug use, they thought he was just as honest, likeable and trustworthy, and they had no idea what question he was asked. But then when he was asked about war on terror, they were capable of detecting dodges. They knew what question he was asked - war on terror - but they thought he was a jerk.
And what that - that's a basic study. We had a series of these that look at different versions, but the basic takeaway is that the most - even egregiously not even acknowledging the question you're asked, but offering an answer that is vaguely related, people don't notice. They don't detect. They don't punish.
FLATOW: So you get in the ballpark, so to speak, and people don't detect the dodge there.
ROGERS: Right, but...
FLATOW: What about how skillful you answer a question, whether it's the right one or the wrong one?
ROGERS: Yeah. So in another study, we have the same health care question, health care answer, fluently answered. And then we have the illegal drug use question, health care answer, fluently answered. And then we add another condition now where health care question, and we have the direct health care answer where the guy stutters.
And the guy says well, um, I'm glad you, you know, I'm glad you asked me that. And he dis-fluently answers it. And what we find is that viewers think that that person is less honest, less likeable, less trustworthy than the person who dodged the question fluently. So, in this way, it's better to answer the wrong question well than the right question poorly.
FLATOW: Talking with Todd Rogers of Harvard Kennedy School in Cambridge. Our number is 1-800-989-8255 if you want to talk to Todd Rogers and discuss dodging the question, or you can also tweet us @scifri. And it can - Dr. Rogers, can you get somebody to answer a question who's really skilled and knows how to dodge it? Is there any way you can do that?
ROGERS: So I can think of two strategies. One strategy is to have a moderator that is actually committed to getting an answer. And so the moderator could just stay on the person. But the downside to that is what the quote you - the clip you played a moment ago, where Mitt Romney dodged the question, and the moderator said I'm sorry, Governor Romney, you didn't answer the question, and then Governor Romney just replies with: I get to offer whatever answer I want.
FLATOW: We have a good - we have a very - a clip from a famous interview I'm sure you're familiar with that illustrates your first point. It's the BBC's Jeremy Paxman and British Conservative Leader Michael Howard, and where Paxman, the interviewer, continually repeats this question - now, it's very involved so I'll just give you - listen now for the question. He's going to ask it over and over again: Did you threaten to overrule him? Let's hear how he kept answering it.
(SOUNDBITE OF BBC BROADCAST)
JEREMY PAXMAN: Did you threaten to overrule him?
MICHAEL HOWARD: I did not overrule Derek Lewis.
PAXMAN: Did you threaten to overrule him?
HOWARD: I took advice on what I could or could not do, and I acted...
PAXMAN: Did you threaten to overrule him, Mr. Howard?
HOWARD: ...scrupulously in accordance with that advice. I did not overrule Derek Lewis.
PAXMAN: Did you threaten to overrule him?
HOWARD: Mr. Marriott was not suspended.
PAXMAN: Did you threaten to overrule him?
HOWARD: I have accounted for my decision to dismiss Derek Lewis in great...
PAXMAN: Did you threaten to overrule him?
HOWARD: ...in great detail before the House of Commons.
PAXMAN: I note you're not answering the question, whether you threatened to overrule him.
FLATOW: You're familiar with that famous question, line of reasoning?
ROGERS: These guys are amazing. And what's problematic about that is the interviewer, many people thought the interviewer was being a jerk. You know, so he was actually trying to get the voters an answer to the question that, at the time, was very important, and he looks like a jerk. But they both look like jerks, truthfully.
FLATOW: Yeah. Is there a way for we, as an audience - now you're taking - the person who's dodging the question wants us to forget what the question is, right? Because - so we know the answer doesn't connect with it. Is there a way for us to pay attention to the question and say, hey, you're dodging it?
ROGERS: Yes, there is. So this was - the whole point of this research was not to describe how can people dodge, and then to eventually start a consulting business teaching people to evade questions.
FLATOW: There are plenty of those, I'm sure.
ROGERS: Yeah, exactly. This - the point of this was to figure out how to - why did it happen, and how can we prevent it? And so what we found is that there's a very small tweak that can eliminate politicians' capacity to get away with it, which is posting the verbatim question on the screen while the answer is being offered.
Even the subtlest dodges, even the most artful dodges get detected, and people think that, you know, judge the politicians accordingly. So this, we're advocating very strongly that on the debates on October 3rd, October 16th and October 22nd, if the broadcast stations and whoever else is broadcasting these debates just posts the question during the answer, and maybe in the short run, these guys will still dodge. But they'll be judged, and eventually, there will be an account for it.
FLATOW: Wouldn't the candidates have to agree to have the question posted on the screen during the debate?
ROGERS: Ira, is it that fixed?
ROGERS: Is it that rigged?
FLATOW: Res ipsa loquitur, you know.
FLATOW: I don't know how the inner workings are, you know, but I would imagine that if I were - you know, if I wanted to keep control - and they're not real debates, you know, in a certainly back-and-forth sense. They're really more just fancy press conferences asking each one a question. As you say, it makes it easy to dodge them.
ROGERS: Yeah, absolutely. I mean, I do think that the important thing is eventually, they will start answering questions if they know that they are being judged for it, and it's very easy to just post the question. One thing that networks have started doing, though, is posting vague summaries of the question.
So, Ira, what will you do about what's going on in Libya? And then if I ask you that, it gets posted on the screen: foreign policy, which is, like, license to answer whatever you want.
FLATOW: What about in this social media age, if you had people ask them do they think the candidates are answering the question truthfully, or have they dodged it? And you have a score up there on the screen.
ROGERS: That - so there were innovations in this in January, during - I think it was the 1,032nd Republican debate of this presidential primary.
ROGERS: They, tweeter - or Twitter, excuse me. Twitter had people tweet hashtag #dodge or hashtag #answer during one of the debates. And then, in fact, the moderators brought that - the responses from Twitter. So they had, like, a moving dial showing these guys are dodging or not, and the moderator mentioned it during the debate and said, look. The audience thinks that you have dodged that question. Can you please answer it? I mean, that's a fancy, fun, engaging way to do it. An easier way is just to post the question.
FLATOW: Let me bring on another guest, because political debates really don't matter much if people don't vote. And in the 2008 presidential election, only about two-thirds of eligible voters actually cast a ballot. And in 2010 midterms, you're up to maybe only 40 percent. So how do you motivate more people to vote? Speaking of the social networks, is that a possibility?
Well, there's a new study that says it could help. A simple message on Facebook actually sent a third of an extra - a million extra votes, a third of a million extra voters to the polls in the 2010 election. That study appears in the journal Nature. James Fowler is author of "Connected: How Your Friends' Friends' Friends Affect Everything You Feel, Think, and Do."
Welcome to SCIENCE FRIDAY.
JAMES FOWLER: Thank you.
FLATOW: You've been listening in on our talk, here. Was - is Facebook really able to change how people go to the polls or not?
FOWLER: They are, but I think a key finding from our study was that Facebook matters a little bit, but friends matter a lot. When we parsed out how much of this message actually got people to change their behavior, we found that the direct effect of the message was to mobilize about 60,000 people. But those people who had a friend who saw the message were actually also more likely to vote. And this indirect effect of the behavior spreading through the network actually got an extra 280,000 people to vote, four for one.
FLATOW: Mm-hmm. Is that enough to sway an election?
FOWLER: It can be. You know, in Florida it was, what, 537 votes in 2000. So it certainly could be if it was focused. In this particular case, we don't have evidence that it actually made a difference for outcomes, because liberals and conservatives were equally susceptible to the treatment of this message.
FLATOW: Do you think this is a technique that's going to be used in this campaign season?
FOWLER: Absolutely. And, you know, what's really heartening is I think both the campaigns are really paying attention to social science, and they're doing experiments, like the experiment we did here, where they - you know, just like a drug, they send a message to some people, and they randomly have other people not see the message, and then they see how it changes their behavior.
But what this study, I think, calls attention to and what they might not be doing fully is that the real impact of these messages is not on the person who receives them. It's on how it spreads through the network. And I think that's going to be really key going forward.
FLATOW: 1-800-989-8255 is our number. We're going to talk about social networking and how it might affect the outcome of your voting. Let us know: Would you vote if you heard your friends are voting? Would you rather go out and vote knowing that your friends are voting and maybe, you know, go out yourself, thinking that's motivational enough? Talking with James Fowler and Todd Rogers. Our number: 1-800-989-8255.
Do you want to see - how do you think the debate should be handled? Would you like to figure out when they're dodging the questions? Would you like to participate in judging that? Let us know: 1-800-989-8255. Also, you can tweet us @scifri and on our Facebook page and on our homepage at sciencefriday.com. Stay with us. We'll be right back after this break.
(SOUNDBITE OF MUSIC)
FLATOW: This is SCIENCE FRIDAY. I'm Ira Flatow. We're talking this hour about the science of persuasion, in this case getting out the vote, persuading people to vote, and with my guests: Todd Rogers is a social psychologist at Harvard's Kennedy School in Cambridge and assistant professor of public policy there; James Fowler, author of "Connected: How Your Friends' Friends' Friends Affect Everything You Feel, Think, and Do." He's also professor of medical genetics and political science at the University of California, San Diego.
Let me get a reaction to what you were saying. Todd, were you surprised to see that it's really the viral effect that multiplies and goes out there, not the first, direct effect?
ROGERS: I was. I'm a big fan of the study, and I'm very impressed they managed to pull it off. It involved 61 million people, and with that scale, they were able to look at spillover to friends' networks, and that was really, I think, the - I think that's what James was highlighting, that really the most surprising is that the effect on friends was much larger than the effect on the target.
FLATOW: 1-800-989-8255. Where you think that indirect viral effect comes from, James?
FOWLER: Well, one of the - the other thing that we found in the study was that it wasn't any old friend on Facebook that mattered. The average person on Facebook today has about 150 friends, you know, Facebook friends. But we also were able to use this separate study to try to figure out which of those people were likely to have a close, face-to-face relationship in real life.
And we looked at just those 10 closest friends, those people you're mostly likely to see every day. Those are the ones that were driving the whole effect. The other 140 people, they had no effect whatsoever. So Facebook is really getting at us not these new ties online but via these real-world social networks we've always had.
FLATOW: Do you think we could use the social networks in this political season? Do you agree that the candidates are going to be relying on them heavily?
FOWLER: I do, and I think that the real-world social network is really important. You want to be able to activate that. You want your messages to become organic because people are much more likely to do something from someone they trust than from a stranger when they hear these messages.
But I also think that there's a lot of opportunity for innovation when it comes to spreading information, and this is where, you know, Todd's studies are ingenious. There's just little things you can do in the debates with Twitter, with collection of information (unintelligible) social networks that can help voters to make better decisions.
FLATOW: Todd? Tell us about that.
ROGERS: You know, these innovations, I guess, there's a lot of them. One thing that I took from this study is the power of the face-to-face social networks. It seems like a lot of this probably happens in the world anyway and that now with social networks we're able to amplify it but also finally measure it. Is that - I mean, is that the way you think about it, James?
FOWLER: It is absolutely. I mean, I think, you know, there's two groups of people out there that I always contend with. There's one group that says online social networks don't matter, that's just for playing Farmville. And then there's another group that says oh my God, everything is different, our brains are different, we can make as many friends as we want now.
And neither one of those things is true. What this study really shows is that online social networks matter, but they matter because they're a way of accessing these real-world social networks that we've always had, and they're allowing us to see not just our, you know, the people we're directly connected to but the people that we're indirectly connected to, as well.
FLATOW: Let's get some calls in from out there where the folks are voting. Gary(ph) in Washington, D.C., hi Gary.
GARY: Hi, I'm calling from the District of Columbia, and, you know, you sort of addressed my question about following up on evasive answers when you said, you know, when you noted that Mitt Romney answered - you know, said I can answer the question any way I want to. But whom do you think would have the power or the, you know, the gravitas or the persuasive ability to change the way political debates, the so-called political debates, are carried out in this country so that we do actually get answers to questions?
Maybe questions could be more clearly asked. They often have a preamble, a beginning, a middle and then finally the question. They're so long, sometimes you're not quite sure what the question was asked. But whom do you think could do that, the League of Women Voters or maybe the former presidents or...
FLATOW: All right, let me get an answer because we're running out of time.
ROGERS: So I think that this dodging is most important for - because it is specifically the questions that candidates dodge that they suspect will affect voter views of them. So it's as if the most important questions - they are helping us identify the most important questions by failing to answer them.
But I also - I ask you, Ira, and other listeners: What question did Gary ask?
FOWLER: It's too bad we don't have, you know, a mechanism here like you suggest to put the question up on the screen. I don't see why we can't do that. I mean, I think that we should tell the people who are organizing this year's presidential debates - how hard is it to just get a direct transcript of the question and put it at the bottom?
I mean, we're watching these news channels all the time anyway. They've got tons of information at the bottom. We're used to doing that. Just put the question up there. Your research shows that this works.
FLATOW: Yeah, we can't even get them to ask a question about science, you know, so good luck.
FLATOW: Good luck in changing the format.
FLATOW: And speaking of science, would that be some way - if a candidate, if there is a section where one candidate wants to ask the other candidate or gets the opportunity for a real debate and come up with a candidate for the - a question for the other candidate, might not choosing a question that they're not prepared for, you know, there are just so many topics you can choose and prepare for, might that not be a way of getting a truthful answer out of somebody?
ROGERS: You could certainly get some stuttering.
ROGERS: I mean, what they are coached to do is nod - I mean, we took the most extreme version which is don't even acknowledge the question and just give some weird, rambling, I'm glad you asked that, America, America, America. But what they're coached to do is just make some bridge, which is in as few words as possible - it's almost like a game.
How many - we should do this, count how many words does it take for them to answer a different question, right? So they'll say well, I'm glad you asked me about Libya. You know, there are so many problems that I think the military matters for, and military health care is terrible. We need to get the health care problem - boom, another topic.
FLATOW: Todd, you...
FOWLER: Can I just...
FLATOW: Yeah, please, go ahead.
FOWLER: I'm sorry, let me just jump in. I just want to also emphasize that one thing that's in common with our research is this idea that people are persuaded by things other than information. But I just want to emphasize the fact that there might be reasons for that. It could be that it's actually more important to think about the character of a candidate than the actual informational content of what they're saying.
And by the same token, it might be more important for us to pay attention to our friends than it is to pay attention to information. And so once we find out what these biases are, we can figure out how to correct them, but we need to I think be a little bit careful because it could be the reason why we have these biases is because we're already programmed to really pay attention to those things that are most important.
FLATOW: Todd, you also studied different ways to increase voter turnout, including one where you asked people about their election day plans because voter turnout, what was it, 37 percent in the last election? It was amazing. What are different ways - and there are some countries that have mandatory, right, you get penalized if you don't go to vote. How do these work? Do they work any better? Are they things that we thought about?
ROGERS: Right, so there - for the last five to 15 years, researchers have been able to run randomized experiments like these pharmaceutical drug trials, like the way James was describing, on voter turnout. And I've been involved in this, and there are several psychological tools that people have begun to use based on this research to increase turnout.
One thing that me and a collaborator, David Nickerson, we studied was having people making a concrete plan: What time will you vote? How will you get there? Where will you be coming from? It makes voter mobilization efforts twice as effective.
We've also, and this is very related to the social networks study, we've also - Alan Gerber and I ran a series of studies where we convinced people turnout was going to be high, a lot of other people, the sort of social system, social perception, turnout's going to be high versus turnout's going to be low.
And when - rationally, if turnout is high, it should make their vote matter less, but in fact because we've evolved to conform to the behavior of others, emphasizing high turnout makes people more likely to vote or more motivated to vote, and that's completely consistent with what James found.
FOWLER: And that's consistent with a wide variety of other behaviors, too. I mean, you look on campus now, and you see social norming campaigns. You know, they don't tell you oh, look at these, you know, five percent of people who binge drink. They say: Did you know that 95 percent of people on this campus don't binge drink?
And that kind of social norming, that positive emphasis, can really make a difference.
FLATOW: One study, which had the biggest effect on sending voters to the polls, used peer pressure on voters in Michigan. Tell us about that, Todd.
ROGERS: It's genius, and it frankly reminds me a lot - James' is sort of a social media version, it feels like. This - they send people a mailing, this is Alan Gerber, Don Green and Christian Larimer. They send people a mailing that says here's a list of all your neighbors who voted and didn't vote and their vote histories for the last two elections.
And we sent it to all your neighbors, and after the election, we'll update it and send it to everyone else. So you'll all know who voted and who didn't. And so it creates this social pressure to vote. It's incredibly creepy...
ROGERS: But outrageously effective.
FLATOW: But effective, yeah.
ROGERS: Exactly. And so it's proven to be many multiples more effective than the next best thing, and so it feels a lot, the social media version - although it's not validated vote that they put on Facebook, James, it feels very similar.
FOWLER: Oh, no, it is. We actually matched six million of the 61 million people to publicly available voter records. So these are people who actually went to the polls.
ROGERS: Right. I mean, on Election Day, when you show pictures of your friends, you don't have a validated...
ROGERS: ...vote on that day. But it's a very similar feel.
FOWLER: That's right.
FLATOW: Is the Internet polarizing the political views people have, you know, by allowing them to associate only with people whose views are similar to their own?
FOWLER: Well, you know, it's funny. Some people who've studied this - Pew is among them - find that people will de-friend one another during campaigns. In fact, it's one of the three top reasons for de-friending, along with politics and people who share too much. And this process means that we eventually end up surrounding ourselves with people who have exactly our same beliefs. So that's one problem.
But another problem is, because it's so easy to access anyone in the world, you can find the person who has exactly the same idiosyncratic point of view as you. And I really worry about this. I worry about this online tendency - this combination of the online world and our real-world tendency to seek out other people like us. It might exacerbate this problem of pulling us apart.
FLATOW: We had a tweet come in from Bob Rains(ph), who says: In addition to the questions printed on the TV screen, I'd like a fact-checker scale to run on the TV at the same time.
ROGERS: Yeah, good luck getting that past...
FOWLER: Fact shmact(ph).
FLATOW: Well, but it does look like, you know, that social networking is becoming a major part of the campaigns and now actually the political debates. And maybe there is a feedback mechanism, you know, in real time, to the political debates that could be - some creative people could think about.
ROGERS: It's going to have to be - the moderator, in the end, is going to have to be the arbiter or mediator of this whole process, although there are Twitter debate - you know, there was a YouTube debate where people got to ask questions via YouTube. And that's a nice innovation.
But in the end, they can answer whatever they can get away with answering. And so how we hold them to that is going to have to be a human or some structure that compares their answer to what's being said, whether it's the fact checker or the question posted.
FLATOW: We have somebody who got - Tab(ph) here suggests: Why not ask candidates to swear in under oath before a debate? That would be...
FOWLER: I think we've seen lots of examples where that probably wouldn't work. (Unintelligible) hold you to the truth.
FLATOW: I'm Ira Flatow, and this is SCIENCE FRIDAY from NPR, talking with Todd Rogers and James Fowler. James Fowler is author of "Connected: How Your Friends' Friends' Friends Affect Everything You Feel, Think, and Do." Does it go that far, James, the effect - friends' friends' friends?
FOWLER: We saw it go out to the friends in the real-world behavior. In the online behavior we saw it go out to the friends of friends. And one limitation we had is it's really expensive computationally to calculate the friends of friends of friends. You're talking about almost a trillion sort of relationships. So it's possible it goes out that far, but we didn't check that far.
FLATOW: Mm-hmm. 1-800-989-8255. So what - are there any creative ways that you were thinking of that - I'll give you the blank check that I - you said it's very expensive. If you had a blank check to study this, what would you like to know, and how would you like to study it? What would you do? Design an experiment for me.
FOWLER: Yeah. So I think the next step now is we want to know what kinds of messages are most influential and what kinds of people are most influential. So we haven't sort of figured out if we're going to do another experiment yet, but as we're discussing this, I'm continuing to work with Facebook, and those are the kinds of questions we'd like to ask.
And that would involve trying out different messages and also trying to figure out what are the attributes of people that seem to be most effective in getting their friends to change their behavior.
FLATOW: What about Twitter? Are there different kinds of people on Twitter than are on Facebook? Sometimes when I ask people on Twitter to go to a Facebook page, they jump up and down and say, I don't want to go anywhere near that thing.
FOWLER: No, they are complements for some people, and they're substitutes for other people, and so you have kind of both ties. One big difference with Twitter is it tends to be more promotional, tends to be more professional, tends to be more like self-promotion in particular. But it's also about weaker ties.
It's hard to define the real friends on there. People have - you know, some people have millions of ties on there, and no one has a million close friends. And so it's kind of a different beast when you're looking at trying to change real-world behaviors because the research suggests you have to get those real-world ties to be able to get that to happen.
FLATOW: Well, as the baby boomers are aging and they're becoming the oldest part of the population - and they didn't grow up with social media - is there data - is it going to be hard to find the real views of those people if you're on social networks because they may not be participating the way the younger people are.
FOWLER: Yeah. No, I disagree with that because if you look at the profile of users on Facebook, you find all kinds of people who are older are using it now; there are over five million users in the United States that are over the age of 65, for example.
And so it's still true that it skews young, but these people are going to age, and gradually the people who are older are coming to use it more. And, you know, I think 10 years from now, a lot of this excitement about being a new media, we're all going to be sort of ho-hum. It's going to be like the telephone.
FLATOW: Ten years from now, who knows where we're going to be?
FLATOW: Look how fast this changes.
FOWLER: That's right. It's really become representative at light speed.
FLATOW: Yeah. Do you think that the - we might be using social networking in the voting process within 10 years?
FOWLER: I don't know about that. I think - you know, people have not reacted well to electronic voting. So I have a feeling that it's - you know, it would be a ways off because of the concern about privacy.
FLATOW: Yeah. Todd, you agree?
ROGERS: We certainly - most people think that they don't know that they're - whether they vote or not is public, and it is. But many people think that who they vote for and how they vote is public, which it is not. And I think that using social media to use public data to mobilize people makes a ton of sense. I don't know that we're going to get to using social media to cast votes in a social way for things other than "American Idol." I do think that in 10 years, you're right, it's hard to know where we'll be. But given your previous hour, I suspect we'll have at least 50 more varieties of fungi.
FLATOW: And we will be there for all...
FLATOW: Believe it, for all 50 varieties. Gentlemen, I wish you good luck, and we'll have you back as the season comes along, you know, and see what's going on. Thanks a lot for joining us today.
ROGERS: Thank you, Ira.
FLATOW: You're welcome. Todd Rogers is social psychologist at the Harvard Kennedy School in Cambridge and assistant professor of public policy there. James Fowler, author of "Connected: How Your Friends' Friends' Friends Affect Everything You Feel, Think and Do." He's also professor of medical genetics and political science at University of California at San Diego. Transcript provided by NPR, Copyright NPR.