WEBVTT - The Election Pollster’s Song 0:00:15.396 --> 0:00:22.236 Pushkin from Pushkin Industries. This is Deep Background, the show 0:00:22.276 --> 0:00:25.996 where we explore the stories behind the stories in the news. 0:00:26.476 --> 0:00:31.316 I'm Noah Feldman. On Saturday, we'll be continuing our special 0:00:31.356 --> 0:00:34.636 series Deep Bench, about the Right Word turn of the 0:00:34.676 --> 0:00:38.636 Supreme Court. But today, in the midst of an election season, 0:00:38.796 --> 0:00:42.396 we thought we should still deliver a regular Deep Background 0:00:42.436 --> 0:00:45.996 episode to you. We're about two weeks out from the 0:00:45.996 --> 0:00:50.676 presidential election now, and in most polls it looks like 0:00:50.956 --> 0:00:55.236 Biden has a serious lead. I don't know about you, 0:00:55.316 --> 0:00:57.836 but I don't pay much attention to the polls until 0:00:58.076 --> 0:01:00.676 I do, and I've entered that phase of the election 0:01:00.756 --> 0:01:03.916 where I can't quite help myself, like a guilty pleasure 0:01:04.036 --> 0:01:07.036 or maybe a non pleasure. I keep stealing back to 0:01:07.076 --> 0:01:10.636 the websites to see what the polls say. Can we 0:01:10.756 --> 0:01:14.196 trust the polls that are out there? What have posters 0:01:14.316 --> 0:01:19.476 learned from recent experiences, including the twenty sixteen election, and 0:01:19.636 --> 0:01:23.156 what methods do they use to get the results that 0:01:23.236 --> 0:01:26.916 we see when we look online. Here to discuss these 0:01:26.996 --> 0:01:30.876 questions with us is Anthony Salvanto. He is the Elections 0:01:30.916 --> 0:01:34.396 and Surveys Director at CBS News, which puts him in 0:01:34.556 --> 0:01:38.516 charge of one of the most significant operations doing polling 0:01:38.876 --> 0:01:42.716 and calling elections in the country. He's also the author 0:01:42.756 --> 0:01:45.236 of the book Where did You get This Number? A 0:01:45.396 --> 0:01:54.956 polsters Guide to making sense of the world. Anthony, thank 0:01:54.996 --> 0:01:57.716 you for being here. I sometimes feel like the rest 0:01:57.716 --> 0:02:01.116 of us are being cruel to posters because three and 0:02:01.196 --> 0:02:04.276 three quarters years we're interested in what you're saying, we're 0:02:04.316 --> 0:02:07.756 intrigued with what you're saying, and then for a tiny 0:02:07.956 --> 0:02:11.956 run up to a election, we suddenly blitz you, demand 0:02:11.956 --> 0:02:15.636 that you perfectly predict the future and predictively complain if 0:02:15.676 --> 0:02:18.156 you get it even a little bit wrong. So a 0:02:18.236 --> 0:02:22.116 collective apology in advance of my doing exactly those things 0:02:22.116 --> 0:02:26.396 in our conversation. You know what, It's all good. The 0:02:26.516 --> 0:02:30.836 history of polling is one such that elections were used 0:02:31.076 --> 0:02:35.356 as a benchmark. It was never really intended to be 0:02:35.836 --> 0:02:40.516 a tool to just quote unquote predict elections. They were 0:02:40.596 --> 0:02:43.356 used as a benchmark by posters in the earlier part 0:02:43.396 --> 0:02:46.756 of the twentieth century to basically show people, yes, this works, 0:02:47.036 --> 0:02:50.476 Yes we can do a scientific example of the population 0:02:50.516 --> 0:02:54.796 and then get something external to validate that. People tend 0:02:54.836 --> 0:02:59.676 to conflate the ideas of predicting with understanding. It kind 0:02:59.676 --> 0:03:02.036 of went from there, and that just somewhat comes with 0:03:02.076 --> 0:03:05.396 a territory. But it's not at all cruel if it 0:03:05.436 --> 0:03:08.516 brings attention to what we do, because I think the 0:03:08.676 --> 0:03:11.716 large or point of what we do is to try 0:03:11.716 --> 0:03:14.836 to understand the public mind and to try to understand 0:03:14.916 --> 0:03:18.476 people through the lens of looking at aggregate data to 0:03:18.636 --> 0:03:22.796 what they do as mass behavior. And if that brings 0:03:22.876 --> 0:03:25.356 us attention, and then if people say, hey, you know what, 0:03:25.476 --> 0:03:27.636 these polls are kind of interesting, maybe I will pay 0:03:27.676 --> 0:03:30.636 attention to them in twenty twenty one or twenty twenty 0:03:30.636 --> 0:03:33.076 two and see what people think of public policy and 0:03:33.156 --> 0:03:37.316 other important issues, then it's all good. Anthony, you said 0:03:37.356 --> 0:03:39.236 something kind of profound right out of the box, which 0:03:39.276 --> 0:03:41.996 is impressive at eight in the morning. You were just 0:03:42.036 --> 0:03:45.396 saying that we tend to conflate our ability to predict 0:03:45.436 --> 0:03:49.676 something convincingly with our ability to understand that same thing. 0:03:50.156 --> 0:03:51.596 And when you put it that way, you're making the 0:03:51.636 --> 0:03:54.156 point that actually prediction isn't the same thing as understanding. 0:03:54.156 --> 0:03:56.276 You know, understanding. We actually want to know why people 0:03:56.356 --> 0:04:01.036 do things, whereas prediction in theory could be blind to 0:04:01.076 --> 0:04:03.556 the question of why as long as we know what 0:04:03.676 --> 0:04:06.476 people are going to do. And I guess I want 0:04:06.476 --> 0:04:08.716 to ask you more about this idea that you're seeking 0:04:08.716 --> 0:04:10.916 to under stand people. Is it do you see when 0:04:10.916 --> 0:04:13.276 you wake up every morning your goal to be I 0:04:13.316 --> 0:04:15.356 want to make the best predictions that I can, or 0:04:15.396 --> 0:04:17.236 do you see it as I want to understand why 0:04:17.396 --> 0:04:19.516 people are going to do the things that you're going 0:04:19.596 --> 0:04:21.436 to do, including voting the way they're going to vote. 0:04:21.556 --> 0:04:24.316 It's the latter, It's entirely the latter. And I've told 0:04:24.356 --> 0:04:27.356 people you should judge me on whether I help you 0:04:27.556 --> 0:04:31.196 understand what is going on in the world around you. 0:04:31.876 --> 0:04:35.116 Here's an example. You always see around this time of year, 0:04:35.596 --> 0:04:38.236 people saying, well, I can't talk to friends and family 0:04:38.316 --> 0:04:41.876 because they're for the other side. You know, one person's 0:04:41.876 --> 0:04:44.956 a Democrat, one person is a Republican. Let's provide for Trump. 0:04:46.516 --> 0:04:49.356 We don't get wrong anymore. We don't know what they're thinking. Well, 0:04:49.436 --> 0:04:52.156 suppose you had a tool that could help you understand 0:04:52.156 --> 0:04:55.636 a little bit about what the other side is thinking, 0:04:55.956 --> 0:05:01.076 then maybe that could facilitate a conversation. Maybe basic level. 0:05:01.156 --> 0:05:04.476 You just have a better time at dinner. You have 0:05:04.596 --> 0:05:08.836 that ability if you read and understand a good pull. 0:05:09.396 --> 0:05:14.356 We posters often try to avoid the name posters, and 0:05:14.436 --> 0:05:17.076 we prefer to be called survey researchers because we talk 0:05:17.156 --> 0:05:20.876 to people and that idea can be carried through to 0:05:20.956 --> 0:05:25.396 the user as well. So I like to think all 0:05:25.396 --> 0:05:28.516 the way back to the start of your question, if 0:05:28.556 --> 0:05:32.836 we can deliver that understanding, then that's really the goal, 0:05:33.036 --> 0:05:36.276 that's really the job. Let me ask you a few 0:05:36.356 --> 0:05:40.956 questions about key concepts that you and all other survey 0:05:40.996 --> 0:05:43.596 researchers use but that at least for me, aren't as 0:05:43.596 --> 0:05:44.876 clear as I would like them to be, and I 0:05:44.956 --> 0:05:48.156 might be not alone in this. Let's start with likely voter, 0:05:48.636 --> 0:05:51.836 which is obviously hugely important concept. It's great to get 0:05:51.876 --> 0:05:54.356 a sense of how the average person would vote, but 0:05:54.516 --> 0:05:57.236 the average person isn't the person necessarily who was going 0:05:57.236 --> 0:06:00.076 to vote, because we have turnout levels in the United 0:06:00.076 --> 0:06:01.836 States that are much lower than they are in some 0:06:01.916 --> 0:06:05.676 other countries, and so a lot rests on your prediction 0:06:05.716 --> 0:06:08.196 of whether someone is likely to vote, and then you 0:06:08.196 --> 0:06:12.516 have to weigh that person's opinion in your overall analysis 0:06:12.556 --> 0:06:16.436 of your research data, balanced by the probability that you 0:06:16.516 --> 0:06:20.196 assigned the likely of their voting. So how do you 0:06:20.276 --> 0:06:23.356 know what are the measures that you consider most reliable 0:06:23.556 --> 0:06:26.596 when you're asking if someone is a likely voter. So 0:06:26.996 --> 0:06:30.756 the likely voter model, which we used in a survey, 0:06:31.036 --> 0:06:36.836 basically says, how can we combine things you the respondent, 0:06:37.396 --> 0:06:42.116 tell us you're going to do, and whether or not 0:06:42.276 --> 0:06:45.876 we believe, for lack of a better word, that you're 0:06:45.916 --> 0:06:48.356 going to do what you say you're going to do. 0:06:48.796 --> 0:06:52.836 This is the behavioral component of a poll. We know 0:06:52.916 --> 0:06:55.676 the attitudinal component. I think the economy is good. I 0:06:55.756 --> 0:06:57.716 like the president and only the president. This is the 0:06:57.756 --> 0:07:01.996 behavioral stuff, which frankly is a lot harder. Well. On 0:07:01.996 --> 0:07:06.076 one hand, we can take everyone who in a previous 0:07:06.156 --> 0:07:08.676 poll told us they were going to vote, and we 0:07:08.756 --> 0:07:11.276 can go call them back after the election and see 0:07:11.276 --> 0:07:13.996 how many of them tell us that they actually did. Now, 0:07:14.036 --> 0:07:16.716 there may be some overreporting in that, but for the 0:07:16.756 --> 0:07:19.796 most part, this is a pretty good measure, meaning that 0:07:19.796 --> 0:07:21.996 people don't necessarily tell you the truth. If they did vote, 0:07:21.996 --> 0:07:23.436 they're happy to tell you, But if they didn't vote, 0:07:23.436 --> 0:07:25.556 they might be ashamed to say, yeah, I didn't actually 0:07:25.596 --> 0:07:28.116 turn out. That's right, And we know that there is 0:07:28.156 --> 0:07:31.916 some overreporting of that. It is not large by the way, 0:07:31.956 --> 0:07:35.076 Just as a quick aside, I don't mix that with 0:07:35.156 --> 0:07:39.236 people lying. This is usually what happens. They say they're 0:07:39.236 --> 0:07:42.516 going to vote, they intend to vote, and then things 0:07:42.516 --> 0:07:44.516 get in the way. Absolutely, we do that, and we 0:07:44.556 --> 0:07:47.716 also know from lots of other economic data that we 0:07:47.836 --> 0:07:51.796 do all kinds of predictions about ourselves which overstate our 0:07:51.836 --> 0:07:56.436 ability to stick to our word exactly. So we've got 0:07:56.596 --> 0:07:58.516 some rough measure, and we can go back and look 0:07:58.556 --> 0:08:01.196 at that from past polls, and let's say it is 0:08:01.236 --> 0:08:03.676 the case about ninety percent of people tell us they 0:08:03.756 --> 0:08:08.276 vote actually vote. Well, then we've got other measures, like 0:08:09.476 --> 0:08:13.476 from the social science literature, from political science literature. If 0:08:13.556 --> 0:08:17.156 people feel like they're part of a community, they are 0:08:17.236 --> 0:08:21.716 increasingly likely to vote. So someone who is a homeowner 0:08:22.236 --> 0:08:25.916 who's lived in a place for a long time is 0:08:25.996 --> 0:08:28.556 more likely to vote. Now, how do you put that 0:08:29.036 --> 0:08:31.876 into a pull In our case, you build an aggression 0:08:31.956 --> 0:08:34.836 model that would have worked in the past, or you 0:08:34.836 --> 0:08:39.196 can back test your data on and you assign everybody 0:08:39.236 --> 0:08:44.756 in the poll a likelihood of voting based on their 0:08:45.076 --> 0:08:49.556 characteristics such as they're known in the aggregate. If we 0:08:49.756 --> 0:08:53.356 know that ninety five percent of people who have voted 0:08:53.436 --> 0:08:57.476 in every single election, vote in the next one. So 0:08:57.556 --> 0:08:59.916 and then so we have a respondent who's voted in 0:08:59.996 --> 0:09:04.556 every single election, give them a point nine five probability 0:09:04.596 --> 0:09:08.596 score of voting in this one. And then the final 0:09:08.636 --> 0:09:13.076 technical side of that poll is that you take everybody 0:09:13.156 --> 0:09:16.196 with a weight assigned to them, a likely voter score, 0:09:16.756 --> 0:09:20.476 and you sum those up to get the overall likely 0:09:20.556 --> 0:09:24.476 voter estimate. As one way of doing a likely voter model. 0:09:24.756 --> 0:09:28.036 There are others, but it is taking what you see 0:09:28.076 --> 0:09:31.636 in the aggregate and behavior in the aggregate and trying 0:09:31.676 --> 0:09:35.756 to apply it to an individual, which for anybody, whether 0:09:35.796 --> 0:09:38.516 you're trying to gauge voting or whether somebody will buy 0:09:38.556 --> 0:09:42.916 your product, can be a tricky business. There are factors 0:09:42.956 --> 0:09:45.756 that are particular to any given election which are much 0:09:45.796 --> 0:09:49.876 harder to incorporate. So in a given election, somebody might 0:09:49.916 --> 0:09:53.076 be really motivated to vote against or for an incumbent, 0:09:53.676 --> 0:09:57.596 or there might be closed polling places and it might 0:09:57.636 --> 0:10:00.956 be harder to vote, or it might be that you've 0:10:00.996 --> 0:10:03.916 recently moved and don't know where you're going to vote. 0:10:04.116 --> 0:10:07.156 All of those things are variables that are much harder 0:10:07.196 --> 0:10:10.276 to gauge. I want to ask you and drill down 0:10:10.316 --> 0:10:13.396 on those elections specific features because they seem to, at 0:10:13.476 --> 0:10:16.396 least in my generalist perception. So I've had a big 0:10:16.396 --> 0:10:19.156 impact on the twenty sixteen election, and one imagines they 0:10:19.236 --> 0:10:21.996 might on the twenty twenty election. So are you and 0:10:22.076 --> 0:10:26.556 our other posters trying now to make strong predictive answers 0:10:26.556 --> 0:10:29.836 to the following question, our Biden voters going to be 0:10:29.916 --> 0:10:32.236 more likely to really want to turn out to vote 0:10:32.276 --> 0:10:35.476 against Trump, or our Trump voters are going to really 0:10:35.556 --> 0:10:38.796 be likely to turn out in high percentages to defend 0:10:38.836 --> 0:10:42.156 the president as opposed to what people do and let's 0:10:42.156 --> 0:10:44.756 say the immedian election, or as opposed to what they 0:10:44.756 --> 0:10:47.596 did in twenty sixteen. I mean, that seems to be 0:10:48.276 --> 0:10:50.316 certainly from the standpoint of the way the campaigns talk 0:10:50.676 --> 0:10:53.476 maybe the whole ball of wax here right that if 0:10:53.876 --> 0:10:56.196 you know, if one side can really motivate its base, 0:10:56.796 --> 0:10:58.716 it's going to have a huge advantage in winning. So 0:10:58.756 --> 0:11:01.476 I guess what I'm wondering is, are you and our 0:11:01.556 --> 0:11:06.676 other polls actually trying to call that feature in? So 0:11:06.716 --> 0:11:08.756 tell us how you bake it in. Yeah, what we're 0:11:08.796 --> 0:11:14.276 doing is, first we ask attitudinal questions, are you motivated 0:11:14.316 --> 0:11:17.716 to vote? Are you motivated more than you were? I 0:11:17.756 --> 0:11:20.236 have questions like how long would you stand in line? 0:11:20.436 --> 0:11:22.436 And the majority of people tell us they'll stand in 0:11:22.476 --> 0:11:25.636 line quote as long as it takes. Is that the case? 0:11:26.156 --> 0:11:28.156 Probably the case for most of them, but things get 0:11:28.196 --> 0:11:30.716 in the way. But I also see people who say 0:11:30.716 --> 0:11:33.956 they'll only stand in line for half an hour. That's 0:11:34.036 --> 0:11:40.276 useful information we have and turnout model in every poll 0:11:40.356 --> 0:11:44.276 number that you see from US. Joe Biden's fortunes right now, frankly, 0:11:44.276 --> 0:11:46.916 including in all the polls that I put out, is 0:11:46.956 --> 0:11:49.476 that he is heavily dependent on people who tell us 0:11:49.516 --> 0:11:51.596 that they're going to vote for the first time, so 0:11:51.636 --> 0:11:56.436 they're included in the poll. But does past behavior predict 0:11:56.476 --> 0:12:00.196 future behavior? That's harder to know. And if they don't 0:12:00.236 --> 0:12:03.556 show up, this election will be much tighter. How much 0:12:03.596 --> 0:12:05.636 of a discount factor are you applying? I mean, just 0:12:05.676 --> 0:12:08.876 to make it as practical as possible. If the first 0:12:08.916 --> 0:12:13.236 time TI voter on average did not turn up, how 0:12:13.276 --> 0:12:15.436 bad is that for Biden? Whereas if the first time 0:12:15.516 --> 0:12:17.476 voter who says, yes, I'm voting, but I've never done 0:12:17.476 --> 0:12:20.316 it before does turn out and vote, how good is 0:12:20.316 --> 0:12:22.876 that for Biden. Oh, it varies state by state, but 0:12:22.916 --> 0:12:26.676 if the first time voter turns out, then what you 0:12:26.716 --> 0:12:31.556 see in the polling will will manifest itself, meaning Biden's 0:12:31.596 --> 0:12:34.676 big lead will be carried out. He'll win by a lot. Yeah, 0:12:34.716 --> 0:12:37.036 I mean I shy away from saying anybody will win, 0:12:37.196 --> 0:12:40.796 but it would greatly advantage him. However, if those folks 0:12:40.916 --> 0:12:45.836 don't show up, that's a much much tighter race. And look, 0:12:46.076 --> 0:12:50.356 I will add this because it really applies here. The 0:12:50.476 --> 0:12:53.436 uncertainty in this election, in my mind, is something that 0:12:53.476 --> 0:12:56.636 I have a great deal of difficulty quantifying, and that 0:12:56.796 --> 0:13:00.356 is the mechanism of voting. It is the balloting process. 0:13:00.796 --> 0:13:03.356 We talk a lot about turnout as though it is 0:13:03.556 --> 0:13:08.236 all based on motivation, because in much money part parts 0:13:08.236 --> 0:13:12.436 it is, there's a motivation, etc. But in this case, 0:13:12.596 --> 0:13:16.476 you have people transitioning from old habits to new ones. 0:13:16.516 --> 0:13:19.716 They stood in line at their local polling place, their 0:13:19.836 --> 0:13:23.876 local school, which is two blocks away for twenty five years, 0:13:24.156 --> 0:13:28.236 and now that polling place has been closed because counties 0:13:28.276 --> 0:13:31.316 are consolidating due to COVID, and now they have to 0:13:31.396 --> 0:13:35.116 drive across the county to a massive voting center where 0:13:35.156 --> 0:13:37.636 they need to find parking and where they need to 0:13:37.676 --> 0:13:39.676 stand in longer line, and where they need to know 0:13:39.676 --> 0:13:43.316 where it is in the first place. Or they're requesting 0:13:43.316 --> 0:13:46.156 a mail ballot, Well, okay, this big envelope comes in 0:13:46.236 --> 0:13:49.876 the mail, and it's not that difficult. But now you've 0:13:49.876 --> 0:13:52.276 opened up these new forms and you have to sign 0:13:52.276 --> 0:13:53.996 it in the right place, and in some places you 0:13:54.036 --> 0:13:57.156 have to stick it inside another envelope. All of those 0:13:57.236 --> 0:14:02.316 changes and how they affect people are really unknown to us, 0:14:02.396 --> 0:14:05.876 and I think that's a great deal of uncertainty in 0:14:06.036 --> 0:14:11.636 this election. And I suspect if turnout patterns vary a lot, 0:14:11.876 --> 0:14:14.596 or by more than we think that they are, that 0:14:14.596 --> 0:14:17.836 that's going to be one explanation why that person in 0:14:17.836 --> 0:14:20.276 the example I just used, Well, they decided they couldn't 0:14:20.276 --> 0:14:23.676 find parking at Dodgers Stadium, which they turned into a big, 0:14:23.716 --> 0:14:26.996 giant polling place. Maybe no surprise there, Or they didn't 0:14:27.036 --> 0:14:30.876 quite navigate the mail ballot process correctly and their ballot 0:14:30.876 --> 0:14:33.476 didn't get there or was rejected or what have you. 0:14:33.916 --> 0:14:37.956 So all of that is incredibly difficult to quantify. I 0:14:38.036 --> 0:14:41.436 suspect we'll get some help in that in the aggregate 0:14:41.836 --> 0:14:44.316 as we get closer to the election, because we we'll 0:14:44.356 --> 0:14:49.116 see is the return rate on a lot of mail votes, 0:14:49.396 --> 0:14:53.156 and we'll also have the total early vote from all 0:14:53.236 --> 0:14:56.396 these early voting locations, so we'll start to have a 0:14:56.476 --> 0:14:59.476 sense of who's done that, how much of it has 0:14:59.516 --> 0:15:02.556 sort of been cannibalized from what would have been election 0:15:02.676 --> 0:15:05.996 day vote, and then we'll narrow down to what remains 0:15:06.196 --> 0:15:08.196 the people who haven't voted yet when we wake up 0:15:08.236 --> 0:15:11.756 on the morning of November third. That'll be a help, 0:15:12.236 --> 0:15:15.756 but from a polling standpoint, that is all very, very 0:15:15.796 --> 0:15:30.716 hard to quantify. We'll be right back. You mentioned aggregation, 0:15:30.796 --> 0:15:33.716 and that's also something that I'm totally fascinated by, and 0:15:33.716 --> 0:15:36.116 it's something that in recent years has become more and 0:15:36.196 --> 0:15:39.196 more salient to anybody who reads polls in the newspapers 0:15:39.316 --> 0:15:42.636 or watches them on television. I'm specifically thinking of the 0:15:42.676 --> 0:15:46.076 aggregation of lots of polls. There are lots of different polls. 0:15:46.196 --> 0:15:49.196 They have different methodologies. Some are local, some are national. 0:15:49.756 --> 0:15:53.596 But there are sites and experts, and you're one of 0:15:53.636 --> 0:15:56.916 them who say, look, here's what my polling data shows, 0:15:56.916 --> 0:16:01.316 but here's what my aggregation of all everybody's polling data shows. 0:16:01.876 --> 0:16:04.076 And there's some implication there, a kind of wisdom of 0:16:04.116 --> 0:16:08.636 crowds idea that it's better to have as many different polls, 0:16:08.636 --> 0:16:10.716 even if they different methodology, some of which might not 0:16:10.716 --> 0:16:14.596 be your favorite methodology, than just to rely on one poll. 0:16:14.596 --> 0:16:16.276 What's your sense of that. Are you one of those 0:16:16.316 --> 0:16:19.756 people who believes that will have better information by aggregating 0:16:19.796 --> 0:16:22.076 lots of polls even if some of them have a 0:16:22.116 --> 0:16:24.996 methodology that you wouldn't be crazy about, or are you 0:16:24.996 --> 0:16:28.036 somebody who thinks, no, we're better off relying on carefully done, 0:16:28.396 --> 0:16:31.836 well done polls and only those polls. The second, I'm 0:16:31.876 --> 0:16:35.396 not a fan of aggregation. I understand why it's done, 0:16:35.676 --> 0:16:39.596 and in general it's not unreasonable. The comparison I make 0:16:39.756 --> 0:16:42.396 is do you want a beat you can dance to 0:16:42.756 --> 0:16:45.636 or do you want to hear the song? If you 0:16:45.716 --> 0:16:49.476 go to a nightclub and the DJ is combining a 0:16:49.476 --> 0:16:52.916 lot of different songs together to give you a really 0:16:52.956 --> 0:16:56.556 good beat, or blending one after another, that's really fun 0:16:56.796 --> 0:16:59.596 and that's really useful because you just want to dance 0:16:59.596 --> 0:17:03.156 and have a good time. If you really want to 0:17:03.196 --> 0:17:06.836 appreciate music and you want to dive into what an 0:17:06.876 --> 0:17:10.356 artist put out as their song, and you listen to 0:17:10.396 --> 0:17:14.116 a song, and that to me is a little closer 0:17:14.156 --> 0:17:17.556 to what the polster is trying to deliver. Any good 0:17:17.596 --> 0:17:21.636 poll is trying to tell you what people think and 0:17:21.796 --> 0:17:26.116 explain why people think it. And they've offered you a 0:17:26.156 --> 0:17:30.236 model of the electorate. They've implicitly or directly offered you 0:17:30.436 --> 0:17:34.276 a model of behavior because of the questions that they asked, 0:17:34.676 --> 0:17:37.516 because of the things they tried to test or examine 0:17:37.796 --> 0:17:41.676 by asking those questions. Is it about the candidates personalities, 0:17:41.756 --> 0:17:45.876 is it about public policies, etc. So they're offering that 0:17:46.036 --> 0:17:48.156 and you can take it for what it's worth. But 0:17:48.836 --> 0:17:52.196 if it's done well, it should offer you a good 0:17:52.356 --> 0:17:56.756 study of the electorate, and you can compare it. If 0:17:56.756 --> 0:17:59.556 you have the time or the inclination, you can compare 0:17:59.636 --> 0:18:03.356 that study to another one that's a really comprehensive and 0:18:03.436 --> 0:18:06.836 I would dare say a little more sophisticated view of 0:18:06.996 --> 0:18:10.316 how to approach getting information. But again it depends on 0:18:10.436 --> 0:18:13.076 what you want, how deep you want to go. So 0:18:13.476 --> 0:18:17.476 the aggregate is for perhaps a shorthand who's winning? I 0:18:17.516 --> 0:18:19.636 can see it from this, Okay, God, move on with 0:18:19.676 --> 0:18:22.116 my life. Do I want to understand what's going on. 0:18:22.516 --> 0:18:25.116 I don't think you get that with aggregation at all. 0:18:25.396 --> 0:18:28.036 I love the analogy that what you do is you're 0:18:28.076 --> 0:18:30.996 a musician. You know, you're a soloist. You're singing your 0:18:31.036 --> 0:18:34.876 song or you're playing your piece of music, and we 0:18:34.996 --> 0:18:37.276 get a certain kind of depth and appreciation out of that. 0:18:37.356 --> 0:18:39.916 And what the aggregators are doing, you know, on five 0:18:40.116 --> 0:18:41.996 thirty eight or you know what the New York Times 0:18:41.996 --> 0:18:45.196 aggregation is doing is they're DJs. You know, they're they're 0:18:45.236 --> 0:18:48.116 playing it all. They're mixing and matching and there's a 0:18:48.116 --> 0:18:50.156 beat that emerges, and you know that can be as 0:18:50.196 --> 0:18:51.956 you say, that can be great too. It sounds like 0:18:51.956 --> 0:18:54.236 it depends on what you want to listen to in 0:18:54.276 --> 0:18:57.716 a given moment or circumstance. I think that's a brilliant metaphor. So, 0:18:58.156 --> 0:19:01.396 but let me ask you, what song are you playing 0:19:01.476 --> 0:19:06.596 right now? What's your feeling at a musical interpretive level 0:19:07.196 --> 0:19:12.556 of the information you're gathering. That's a great question for 0:19:12.676 --> 0:19:16.876 many many people, it is a referendum on the president. Now. 0:19:16.916 --> 0:19:20.716 In some sense that's not unusual because there's an incumbent 0:19:20.836 --> 0:19:23.996 on the ballot, but it is very particularly the case 0:19:24.156 --> 0:19:27.916 now and for Democrats They say that they are voting 0:19:28.716 --> 0:19:32.236 as much, if not more, to vote against the president 0:19:32.476 --> 0:19:37.276 than for Joe Biden. They say that they are motivated 0:19:37.356 --> 0:19:41.836 to vote. At the same time, you have the president 0:19:41.876 --> 0:19:46.356 who has as solid and a core base of supporters 0:19:46.796 --> 0:19:53.436 as we've ever seen, and that kind of allegiance is 0:19:53.836 --> 0:19:58.196 something unlike we've seen. We did a study right before 0:19:58.236 --> 0:20:02.076 the Republican Convention where we ask does Donald Trump deserve 0:20:02.156 --> 0:20:05.036 your loyalty? And many Republicans said yes. In fact, I 0:20:05.036 --> 0:20:08.516 even tried this. We said what's more important to you 0:20:09.116 --> 0:20:14.436 being a Republican or being a Trump supporter? And Moore 0:20:14.596 --> 0:20:17.796 said it was being a Trump supporter. That kind of 0:20:17.836 --> 0:20:22.916 connection is strong. So you have deeply held, very emotional 0:20:23.396 --> 0:20:27.956 connections to their positions on each side. And it's one 0:20:27.996 --> 0:20:31.276 of the reasons that you can explain in the aggregate 0:20:31.916 --> 0:20:36.116 why you've seen the president's approval ratings hold so historically 0:20:36.116 --> 0:20:40.036 incredibly steady throughout his presidency, where the economy has gone 0:20:40.076 --> 0:20:41.996 up or whether it's gone down, no matter what he said, 0:20:42.036 --> 0:20:45.596 no matter what he's done. And it's also one reason 0:20:45.676 --> 0:20:49.236 why you find remarkable stability in the polls this year 0:20:49.636 --> 0:20:52.796 that Joe Biden edge has been about what it's been 0:20:52.876 --> 0:20:55.196 and it's not like twenty sixteen at all, where we 0:20:55.236 --> 0:20:59.756 saw a large fluctuations in the polling. That is on 0:20:59.796 --> 0:21:02.516 the personal level. On the individual level, part of the 0:21:02.516 --> 0:21:05.836 explanation for what you see in the aggregate that is 0:21:05.916 --> 0:21:10.276 the song of twenty twenty. It is those deeply held 0:21:10.916 --> 0:21:14.916 convictions about what people are seeing. And the final maybe 0:21:14.916 --> 0:21:18.676 a coda if you will, on this is that there 0:21:18.716 --> 0:21:21.516 are very different views of what kind of shape the 0:21:21.596 --> 0:21:26.716 country is in, not just differences on how to solve 0:21:26.716 --> 0:21:32.116 the problems. The Republicans feel that the coronavirus outbreak is 0:21:32.156 --> 0:21:37.356 not as bad as the medical folks say that it is. 0:21:37.796 --> 0:21:43.556 They believe that the death toll is overreported. They are 0:21:43.676 --> 0:21:48.516 less concerned about the virus themselves. Democrats, by contrasts, believe 0:21:48.636 --> 0:21:53.156 that the death toll is underreported. They are themselves much 0:21:53.196 --> 0:21:57.196 more deeply concerned about the virus, and there's some personal 0:21:57.196 --> 0:22:00.076 connection and experience in that, and that they tell us 0:22:00.076 --> 0:22:04.316 it's affecting their communities more. The Democrats live in cities, etc. 0:22:04.836 --> 0:22:07.756 That may be changing as they outbreak unfortunately spreads to 0:22:07.796 --> 0:22:11.676 other places. So there's a very different view. And that's 0:22:11.676 --> 0:22:15.956 just one example of what is happening in the country 0:22:15.996 --> 0:22:19.236 that we don't even agree on what the facts are, 0:22:19.796 --> 0:22:24.636 let alone the more traditional or conventional arguments around politics 0:22:24.676 --> 0:22:27.876 of what do we do to solve this? And that 0:22:27.996 --> 0:22:31.276 is something supposed We're still kind of wrestling with what 0:22:31.436 --> 0:22:34.956 happens in that environment. That remains a question for me. 0:22:37.276 --> 0:22:41.156 Exit pulling. How do you do exit pulling in this 0:22:41.236 --> 0:22:45.716 weird COVID year where many people have mailed in ballots? 0:22:46.436 --> 0:22:48.916 How does that game change and how will you do 0:22:48.996 --> 0:22:52.196 the exit pulling this time around? Is it easier because 0:22:52.196 --> 0:22:55.236 of the mail in votes or is it harder because 0:22:55.276 --> 0:22:57.276 of the disparities in the different ways that votes are 0:22:57.276 --> 0:23:00.396 being cast in different places. Well, the mail in votes 0:23:00.596 --> 0:23:04.876 will have to be interviewed by phone, so that'll either 0:23:04.956 --> 0:23:08.356 be a random digit dial phone looking for somebody voted 0:23:08.356 --> 0:23:12.476 by mail, or where possible, you take the voter rolls, 0:23:12.596 --> 0:23:16.196 you know the folks who returned mail ballots, and you 0:23:16.236 --> 0:23:18.836 attach a phone number to that you call them. So 0:23:19.036 --> 0:23:21.676 that's a change. We've always been able to reach out 0:23:21.716 --> 0:23:23.636 and we have reached out to people who voted by 0:23:23.676 --> 0:23:26.476 mail by telephone. That's always been a component of the 0:23:26.516 --> 0:23:28.796 exit polls. However, this year obviously is going to be 0:23:28.836 --> 0:23:32.876 a much larger one in the early vote. What we're 0:23:32.916 --> 0:23:36.516 doing this year, and this is changed or it's rather 0:23:37.036 --> 0:23:41.916 enlarged or expanded, is positioning exit poll interviewers at the 0:23:41.996 --> 0:23:46.956 early voting sites, so they're out there now and they'll 0:23:46.996 --> 0:23:50.716 be interviewing people who are lined up for early voting 0:23:50.756 --> 0:23:53.436 the same way they would have been on election day 0:23:53.476 --> 0:23:56.996 as people are leaving the polling place. And then the 0:23:57.836 --> 0:24:01.316 final part of that, the actual election day exit poll 0:24:02.236 --> 0:24:05.876