1 00:00:15,396 --> 00:00:22,236 Speaker 1: Pushkin from Pushkin Industries. This is Deep Background, the show 2 00:00:22,276 --> 00:00:25,996 Speaker 1: where we explore the stories behind the stories in the news. 3 00:00:26,476 --> 00:00:31,316 Speaker 1: I'm Noah Feldman. On Saturday, we'll be continuing our special 4 00:00:31,356 --> 00:00:34,636 Speaker 1: series Deep Bench, about the Right Word turn of the 5 00:00:34,676 --> 00:00:38,636 Speaker 1: Supreme Court. But today, in the midst of an election season, 6 00:00:38,796 --> 00:00:42,396 Speaker 1: we thought we should still deliver a regular Deep Background 7 00:00:42,436 --> 00:00:45,996 Speaker 1: episode to you. We're about two weeks out from the 8 00:00:45,996 --> 00:00:50,676 Speaker 1: presidential election now, and in most polls it looks like 9 00:00:50,956 --> 00:00:55,236 Speaker 1: Biden has a serious lead. I don't know about you, 10 00:00:55,316 --> 00:00:57,836 Speaker 1: but I don't pay much attention to the polls until 11 00:00:58,076 --> 00:01:00,676 Speaker 1: I do, and I've entered that phase of the election 12 00:01:00,756 --> 00:01:03,916 Speaker 1: where I can't quite help myself, like a guilty pleasure 13 00:01:04,036 --> 00:01:07,036 Speaker 1: or maybe a non pleasure. I keep stealing back to 14 00:01:07,076 --> 00:01:10,636 Speaker 1: the websites to see what the polls say. Can we 15 00:01:10,756 --> 00:01:14,196 Speaker 1: trust the polls that are out there? What have posters 16 00:01:14,316 --> 00:01:19,476 Speaker 1: learned from recent experiences, including the twenty sixteen election, and 17 00:01:19,636 --> 00:01:23,156 Speaker 1: what methods do they use to get the results that 18 00:01:23,236 --> 00:01:26,916 Speaker 1: we see when we look online. Here to discuss these 19 00:01:26,996 --> 00:01:30,876 Speaker 1: questions with us is Anthony Salvanto. He is the Elections 20 00:01:30,916 --> 00:01:34,396 Speaker 1: and Surveys Director at CBS News, which puts him in 21 00:01:34,556 --> 00:01:38,516 Speaker 1: charge of one of the most significant operations doing polling 22 00:01:38,876 --> 00:01:42,716 Speaker 1: and calling elections in the country. He's also the author 23 00:01:42,756 --> 00:01:45,236 Speaker 1: of the book Where did You get This Number? A 24 00:01:45,396 --> 00:01:54,956 Speaker 1: polsters Guide to making sense of the world. Anthony, thank 25 00:01:54,996 --> 00:01:57,716 Speaker 1: you for being here. I sometimes feel like the rest 26 00:01:57,716 --> 00:02:01,116 Speaker 1: of us are being cruel to posters because three and 27 00:02:01,196 --> 00:02:04,276 Speaker 1: three quarters years we're interested in what you're saying, we're 28 00:02:04,316 --> 00:02:07,756 Speaker 1: intrigued with what you're saying, and then for a tiny 29 00:02:07,956 --> 00:02:11,956 Speaker 1: run up to a election, we suddenly blitz you, demand 30 00:02:11,956 --> 00:02:15,636 Speaker 1: that you perfectly predict the future and predictively complain if 31 00:02:15,676 --> 00:02:18,156 Speaker 1: you get it even a little bit wrong. So a 32 00:02:18,236 --> 00:02:22,116 Speaker 1: collective apology in advance of my doing exactly those things 33 00:02:22,116 --> 00:02:26,396 Speaker 1: in our conversation. You know what, It's all good. The 34 00:02:26,516 --> 00:02:30,836 Speaker 1: history of polling is one such that elections were used 35 00:02:31,076 --> 00:02:35,356 Speaker 1: as a benchmark. It was never really intended to be 36 00:02:35,836 --> 00:02:40,516 Speaker 1: a tool to just quote unquote predict elections. They were 37 00:02:40,596 --> 00:02:43,356 Speaker 1: used as a benchmark by posters in the earlier part 38 00:02:43,396 --> 00:02:46,756 Speaker 1: of the twentieth century to basically show people, yes, this works, 39 00:02:47,036 --> 00:02:50,476 Speaker 1: Yes we can do a scientific example of the population 40 00:02:50,516 --> 00:02:54,796 Speaker 1: and then get something external to validate that. People tend 41 00:02:54,836 --> 00:02:59,676 Speaker 1: to conflate the ideas of predicting with understanding. It kind 42 00:02:59,676 --> 00:03:02,036 Speaker 1: of went from there, and that just somewhat comes with 43 00:03:02,076 --> 00:03:05,396 Speaker 1: a territory. But it's not at all cruel if it 44 00:03:05,436 --> 00:03:08,516 Speaker 1: brings attention to what we do, because I think the 45 00:03:08,676 --> 00:03:11,716 Speaker 1: large or point of what we do is to try 46 00:03:11,716 --> 00:03:14,836 Speaker 1: to understand the public mind and to try to understand 47 00:03:14,916 --> 00:03:18,476 Speaker 1: people through the lens of looking at aggregate data to 48 00:03:18,636 --> 00:03:22,796 Speaker 1: what they do as mass behavior. And if that brings 49 00:03:22,876 --> 00:03:25,356 Speaker 1: us attention, and then if people say, hey, you know what, 50 00:03:25,476 --> 00:03:27,636 Speaker 1: these polls are kind of interesting, maybe I will pay 51 00:03:27,676 --> 00:03:30,636 Speaker 1: attention to them in twenty twenty one or twenty twenty 52 00:03:30,636 --> 00:03:33,076 Speaker 1: two and see what people think of public policy and 53 00:03:33,156 --> 00:03:37,316 Speaker 1: other important issues, then it's all good. Anthony, you said 54 00:03:37,356 --> 00:03:39,236 Speaker 1: something kind of profound right out of the box, which 55 00:03:39,276 --> 00:03:41,996 Speaker 1: is impressive at eight in the morning. You were just 56 00:03:42,036 --> 00:03:45,396 Speaker 1: saying that we tend to conflate our ability to predict 57 00:03:45,436 --> 00:03:49,676 Speaker 1: something convincingly with our ability to understand that same thing. 58 00:03:50,156 --> 00:03:51,596 Speaker 1: And when you put it that way, you're making the 59 00:03:51,636 --> 00:03:54,156 Speaker 1: point that actually prediction isn't the same thing as understanding. 60 00:03:54,156 --> 00:03:56,276 Speaker 1: You know, understanding. We actually want to know why people 61 00:03:56,356 --> 00:04:01,036 Speaker 1: do things, whereas prediction in theory could be blind to 62 00:04:01,076 --> 00:04:03,556 Speaker 1: the question of why as long as we know what 63 00:04:03,676 --> 00:04:06,476 Speaker 1: people are going to do. And I guess I want 64 00:04:06,476 --> 00:04:08,716 Speaker 1: to ask you more about this idea that you're seeking 65 00:04:08,716 --> 00:04:10,916 Speaker 1: to under stand people. Is it do you see when 66 00:04:10,916 --> 00:04:13,276 Speaker 1: you wake up every morning your goal to be I 67 00:04:13,316 --> 00:04:15,356 Speaker 1: want to make the best predictions that I can, or 68 00:04:15,396 --> 00:04:17,236 Speaker 1: do you see it as I want to understand why 69 00:04:17,396 --> 00:04:19,516 Speaker 1: people are going to do the things that you're going 70 00:04:19,596 --> 00:04:21,436 Speaker 1: to do, including voting the way they're going to vote. 71 00:04:21,556 --> 00:04:24,316 Speaker 1: It's the latter, It's entirely the latter. And I've told 72 00:04:24,356 --> 00:04:27,356 Speaker 1: people you should judge me on whether I help you 73 00:04:27,556 --> 00:04:31,196 Speaker 1: understand what is going on in the world around you. 74 00:04:31,876 --> 00:04:35,116 Speaker 1: Here's an example. You always see around this time of year, 75 00:04:35,596 --> 00:04:38,236 Speaker 1: people saying, well, I can't talk to friends and family 76 00:04:38,316 --> 00:04:41,876 Speaker 1: because they're for the other side. You know, one person's 77 00:04:41,876 --> 00:04:44,956 Speaker 1: a Democrat, one person is a Republican. Let's provide for Trump. 78 00:04:46,516 --> 00:04:49,356 Speaker 1: We don't get wrong anymore. We don't know what they're thinking. Well, 79 00:04:49,436 --> 00:04:52,156 Speaker 1: suppose you had a tool that could help you understand 80 00:04:52,156 --> 00:04:55,636 Speaker 1: a little bit about what the other side is thinking, 81 00:04:55,956 --> 00:05:01,076 Speaker 1: then maybe that could facilitate a conversation. Maybe basic level. 82 00:05:01,156 --> 00:05:04,476 Speaker 1: You just have a better time at dinner. You have 83 00:05:04,596 --> 00:05:08,836 Speaker 1: that ability if you read and understand a good pull. 84 00:05:09,396 --> 00:05:14,356 Speaker 1: We posters often try to avoid the name posters, and 85 00:05:14,436 --> 00:05:17,076 Speaker 1: we prefer to be called survey researchers because we talk 86 00:05:17,156 --> 00:05:20,876 Speaker 1: to people and that idea can be carried through to 87 00:05:20,956 --> 00:05:25,396 Speaker 1: the user as well. So I like to think all 88 00:05:25,396 --> 00:05:28,516 Speaker 1: the way back to the start of your question, if 89 00:05:28,556 --> 00:05:32,836 Speaker 1: we can deliver that understanding, then that's really the goal, 90 00:05:33,036 --> 00:05:36,276 Speaker 1: that's really the job. Let me ask you a few 91 00:05:36,356 --> 00:05:40,956 Speaker 1: questions about key concepts that you and all other survey 92 00:05:40,996 --> 00:05:43,596 Speaker 1: researchers use but that at least for me, aren't as 93 00:05:43,596 --> 00:05:44,876 Speaker 1: clear as I would like them to be, and I 94 00:05:44,956 --> 00:05:48,156 Speaker 1: might be not alone in this. Let's start with likely voter, 95 00:05:48,636 --> 00:05:51,836 Speaker 1: which is obviously hugely important concept. It's great to get 96 00:05:51,876 --> 00:05:54,356 Speaker 1: a sense of how the average person would vote, but 97 00:05:54,516 --> 00:05:57,236 Speaker 1: the average person isn't the person necessarily who was going 98 00:05:57,236 --> 00:06:00,076 Speaker 1: to vote, because we have turnout levels in the United 99 00:06:00,076 --> 00:06:01,836 Speaker 1: States that are much lower than they are in some 100 00:06:01,916 --> 00:06:05,676 Speaker 1: other countries, and so a lot rests on your prediction 101 00:06:05,716 --> 00:06:08,196 Speaker 1: of whether someone is likely to vote, and then you 102 00:06:08,196 --> 00:06:12,516 Speaker 1: have to weigh that person's opinion in your overall analysis 103 00:06:12,556 --> 00:06:16,436 Speaker 1: of your research data, balanced by the probability that you 104 00:06:16,516 --> 00:06:20,196 Speaker 1: assigned the likely of their voting. So how do you 105 00:06:20,276 --> 00:06:23,356 Speaker 1: know what are the measures that you consider most reliable 106 00:06:23,556 --> 00:06:26,596 Speaker 1: when you're asking if someone is a likely voter. So 107 00:06:26,996 --> 00:06:30,756 Speaker 1: the likely voter model, which we used in a survey, 108 00:06:31,036 --> 00:06:36,836 Speaker 1: basically says, how can we combine things you the respondent, 109 00:06:37,396 --> 00:06:42,116 Speaker 1: tell us you're going to do, and whether or not 110 00:06:42,276 --> 00:06:45,876 Speaker 1: we believe, for lack of a better word, that you're 111 00:06:45,916 --> 00:06:48,356 Speaker 1: going to do what you say you're going to do. 112 00:06:48,796 --> 00:06:52,836 Speaker 1: This is the behavioral component of a poll. We know 113 00:06:52,916 --> 00:06:55,676 Speaker 1: the attitudinal component. I think the economy is good. I 114 00:06:55,756 --> 00:06:57,716 Speaker 1: like the president and only the president. This is the 115 00:06:57,756 --> 00:07:01,996 Speaker 1: behavioral stuff, which frankly is a lot harder. Well. On 116 00:07:01,996 --> 00:07:06,076 Speaker 1: one hand, we can take everyone who in a previous 117 00:07:06,156 --> 00:07:08,676 Speaker 1: poll told us they were going to vote, and we 118 00:07:08,756 --> 00:07:11,276 Speaker 1: can go call them back after the election and see 119 00:07:11,276 --> 00:07:13,996 Speaker 1: how many of them tell us that they actually did. Now, 120 00:07:14,036 --> 00:07:16,716 Speaker 1: there may be some overreporting in that, but for the 121 00:07:16,756 --> 00:07:19,796 Speaker 1: most part, this is a pretty good measure, meaning that 122 00:07:19,796 --> 00:07:21,996 Speaker 1: people don't necessarily tell you the truth. If they did vote, 123 00:07:21,996 --> 00:07:23,436 Speaker 1: they're happy to tell you, But if they didn't vote, 124 00:07:23,436 --> 00:07:25,556 Speaker 1: they might be ashamed to say, yeah, I didn't actually 125 00:07:25,596 --> 00:07:28,116 Speaker 1: turn out. That's right, And we know that there is 126 00:07:28,156 --> 00:07:31,916 Speaker 1: some overreporting of that. It is not large by the way, 127 00:07:31,956 --> 00:07:35,076 Speaker 1: Just as a quick aside, I don't mix that with 128 00:07:35,156 --> 00:07:39,236 Speaker 1: people lying. This is usually what happens. They say they're 129 00:07:39,236 --> 00:07:42,516 Speaker 1: going to vote, they intend to vote, and then things 130 00:07:42,516 --> 00:07:44,516 Speaker 1: get in the way. Absolutely, we do that, and we 131 00:07:44,556 --> 00:07:47,716 Speaker 1: also know from lots of other economic data that we 132 00:07:47,836 --> 00:07:51,796 Speaker 1: do all kinds of predictions about ourselves which overstate our 133 00:07:51,836 --> 00:07:56,436 Speaker 1: ability to stick to our word exactly. So we've got 134 00:07:56,596 --> 00:07:58,516 Speaker 1: some rough measure, and we can go back and look 135 00:07:58,556 --> 00:08:01,196 Speaker 1: at that from past polls, and let's say it is 136 00:08:01,236 --> 00:08:03,676 Speaker 1: the case about ninety percent of people tell us they 137 00:08:03,756 --> 00:08:08,276 Speaker 1: vote actually vote. Well, then we've got other measures, like 138 00:08:09,476 --> 00:08:13,476 Speaker 1: from the social science literature, from political science literature. If 139 00:08:13,556 --> 00:08:17,156 Speaker 1: people feel like they're part of a community, they are 140 00:08:17,236 --> 00:08:21,716 Speaker 1: increasingly likely to vote. So someone who is a homeowner 141 00:08:22,236 --> 00:08:25,916 Speaker 1: who's lived in a place for a long time is 142 00:08:25,996 --> 00:08:28,556 Speaker 1: more likely to vote. Now, how do you put that 143 00:08:29,036 --> 00:08:31,876 Speaker 1: into a pull In our case, you build an aggression 144 00:08:31,956 --> 00:08:34,836 Speaker 1: model that would have worked in the past, or you 145 00:08:34,836 --> 00:08:39,196 Speaker 1: can back test your data on and you assign everybody 146 00:08:39,236 --> 00:08:44,756 Speaker 1: in the poll a likelihood of voting based on their 147 00:08:45,076 --> 00:08:49,556 Speaker 1: characteristics such as they're known in the aggregate. If we 148 00:08:49,756 --> 00:08:53,356 Speaker 1: know that ninety five percent of people who have voted 149 00:08:53,436 --> 00:08:57,476 Speaker 1: in every single election, vote in the next one. So 150 00:08:57,556 --> 00:08:59,916 Speaker 1: and then so we have a respondent who's voted in 151 00:08:59,996 --> 00:09:04,556 Speaker 1: every single election, give them a point nine five probability 152 00:09:04,596 --> 00:09:08,596 Speaker 1: score of voting in this one. And then the final 153 00:09:08,636 --> 00:09:13,076 Speaker 1: technical side of that poll is that you take everybody 154 00:09:13,156 --> 00:09:16,196 Speaker 1: with a weight assigned to them, a likely voter score, 155 00:09:16,756 --> 00:09:20,476 Speaker 1: and you sum those up to get the overall likely 156 00:09:20,556 --> 00:09:24,476 Speaker 1: voter estimate. As one way of doing a likely voter model. 157 00:09:24,756 --> 00:09:28,036 Speaker 1: There are others, but it is taking what you see 158 00:09:28,076 --> 00:09:31,636 Speaker 1: in the aggregate and behavior in the aggregate and trying 159 00:09:31,676 --> 00:09:35,756 Speaker 1: to apply it to an individual, which for anybody, whether 160 00:09:35,796 --> 00:09:38,516 Speaker 1: you're trying to gauge voting or whether somebody will buy 161 00:09:38,556 --> 00:09:42,916 Speaker 1: your product, can be a tricky business. There are factors 162 00:09:42,956 --> 00:09:45,756 Speaker 1: that are particular to any given election which are much 163 00:09:45,796 --> 00:09:49,876 Speaker 1: harder to incorporate. So in a given election, somebody might 164 00:09:49,916 --> 00:09:53,076 Speaker 1: be really motivated to vote against or for an incumbent, 165 00:09:53,676 --> 00:09:57,596 Speaker 1: or there might be closed polling places and it might 166 00:09:57,636 --> 00:10:00,956 Speaker 1: be harder to vote, or it might be that you've 167 00:10:00,996 --> 00:10:03,916 Speaker 1: recently moved and don't know where you're going to vote. 168 00:10:04,116 --> 00:10:07,156 Speaker 1: All of those things are variables that are much harder 169 00:10:07,196 --> 00:10:10,276 Speaker 1: to gauge. I want to ask you and drill down 170 00:10:10,316 --> 00:10:13,396 Speaker 1: on those elections specific features because they seem to, at 171 00:10:13,476 --> 00:10:16,396 Speaker 1: least in my generalist perception. So I've had a big 172 00:10:16,396 --> 00:10:19,156 Speaker 1: impact on the twenty sixteen election, and one imagines they 173 00:10:19,236 --> 00:10:21,996 Speaker 1: might on the twenty twenty election. So are you and 174 00:10:22,076 --> 00:10:26,556 Speaker 1: our other posters trying now to make strong predictive answers 175 00:10:26,556 --> 00:10:29,836 Speaker 1: to the following question, our Biden voters going to be 176 00:10:29,916 --> 00:10:32,236 Speaker 1: more likely to really want to turn out to vote 177 00:10:32,276 --> 00:10:35,476 Speaker 1: against Trump, or our Trump voters are going to really 178 00:10:35,556 --> 00:10:38,796 Speaker 1: be likely to turn out in high percentages to defend 179 00:10:38,836 --> 00:10:42,156 Speaker 1: the president as opposed to what people do and let's 180 00:10:42,156 --> 00:10:44,756 Speaker 1: say the immedian election, or as opposed to what they 181 00:10:44,756 --> 00:10:47,596 Speaker 1: did in twenty sixteen. I mean, that seems to be 182 00:10:48,276 --> 00:10:50,316 Speaker 1: certainly from the standpoint of the way the campaigns talk 183 00:10:50,676 --> 00:10:53,476 Speaker 1: maybe the whole ball of wax here right that if 184 00:10:53,876 --> 00:10:56,196 Speaker 1: you know, if one side can really motivate its base, 185 00:10:56,796 --> 00:10:58,716 Speaker 1: it's going to have a huge advantage in winning. So 186 00:10:58,756 --> 00:11:01,476 Speaker 1: I guess what I'm wondering is, are you and our 187 00:11:01,556 --> 00:11:06,676 Speaker 1: other polls actually trying to call that feature in? So 188 00:11:06,716 --> 00:11:08,756 Speaker 1: tell us how you bake it in. Yeah, what we're 189 00:11:08,796 --> 00:11:14,276 Speaker 1: doing is, first we ask attitudinal questions, are you motivated 190 00:11:14,316 --> 00:11:17,716 Speaker 1: to vote? Are you motivated more than you were? I 191 00:11:17,756 --> 00:11:20,236 Speaker 1: have questions like how long would you stand in line? 192 00:11:20,436 --> 00:11:22,436 Speaker 1: And the majority of people tell us they'll stand in 193 00:11:22,476 --> 00:11:25,636 Speaker 1: line quote as long as it takes. Is that the case? 194 00:11:26,156 --> 00:11:28,156 Speaker 1: Probably the case for most of them, but things get 195 00:11:28,196 --> 00:11:30,716 Speaker 1: in the way. But I also see people who say 196 00:11:30,716 --> 00:11:33,956 Speaker 1: they'll only stand in line for half an hour. That's 197 00:11:34,036 --> 00:11:40,276 Speaker 1: useful information we have and turnout model in every poll 198 00:11:40,356 --> 00:11:44,276 Speaker 1: number that you see from US. Joe Biden's fortunes right now, frankly, 199 00:11:44,276 --> 00:11:46,916 Speaker 1: including in all the polls that I put out, is 200 00:11:46,956 --> 00:11:49,476 Speaker 1: that he is heavily dependent on people who tell us 201 00:11:49,516 --> 00:11:51,596 Speaker 1: that they're going to vote for the first time, so 202 00:11:51,636 --> 00:11:56,436 Speaker 1: they're included in the poll. But does past behavior predict 203 00:11:56,476 --> 00:12:00,196 Speaker 1: future behavior? That's harder to know. And if they don't 204 00:12:00,236 --> 00:12:03,556 Speaker 1: show up, this election will be much tighter. How much 205 00:12:03,596 --> 00:12:05,636 Speaker 1: of a discount factor are you applying? I mean, just 206 00:12:05,676 --> 00:12:08,876 Speaker 1: to make it as practical as possible. If the first 207 00:12:08,916 --> 00:12:13,236 Speaker 1: time TI voter on average did not turn up, how 208 00:12:13,276 --> 00:12:15,436 Speaker 1: bad is that for Biden? Whereas if the first time 209 00:12:15,516 --> 00:12:17,476 Speaker 1: voter who says, yes, I'm voting, but I've never done 210 00:12:17,476 --> 00:12:20,316 Speaker 1: it before does turn out and vote, how good is 211 00:12:20,316 --> 00:12:22,876 Speaker 1: that for Biden. Oh, it varies state by state, but 212 00:12:22,916 --> 00:12:26,676 Speaker 1: if the first time voter turns out, then what you 213 00:12:26,716 --> 00:12:31,556 Speaker 1: see in the polling will will manifest itself, meaning Biden's 214 00:12:31,596 --> 00:12:34,676 Speaker 1: big lead will be carried out. He'll win by a lot. Yeah, 215 00:12:34,716 --> 00:12:37,036 Speaker 1: I mean I shy away from saying anybody will win, 216 00:12:37,196 --> 00:12:40,796 Speaker 1: but it would greatly advantage him. However, if those folks 217 00:12:40,916 --> 00:12:45,836 Speaker 1: don't show up, that's a much much tighter race. And look, 218 00:12:46,076 --> 00:12:50,356 Speaker 1: I will add this because it really applies here. The 219 00:12:50,476 --> 00:12:53,436 Speaker 1: uncertainty in this election, in my mind, is something that 220 00:12:53,476 --> 00:12:56,636 Speaker 1: I have a great deal of difficulty quantifying, and that 221 00:12:56,796 --> 00:13:00,356 Speaker 1: is the mechanism of voting. It is the balloting process. 222 00:13:00,796 --> 00:13:03,356 Speaker 1: We talk a lot about turnout as though it is 223 00:13:03,556 --> 00:13:08,236 Speaker 1: all based on motivation, because in much money part parts 224 00:13:08,236 --> 00:13:12,436 Speaker 1: it is, there's a motivation, etc. But in this case, 225 00:13:12,596 --> 00:13:16,476 Speaker 1: you have people transitioning from old habits to new ones. 226 00:13:16,516 --> 00:13:19,716 Speaker 1: They stood in line at their local polling place, their 227 00:13:19,836 --> 00:13:23,876 Speaker 1: local school, which is two blocks away for twenty five years, 228 00:13:24,156 --> 00:13:28,236 Speaker 1: and now that polling place has been closed because counties 229 00:13:28,276 --> 00:13:31,316 Speaker 1: are consolidating due to COVID, and now they have to 230 00:13:31,396 --> 00:13:35,116 Speaker 1: drive across the county to a massive voting center where 231 00:13:35,156 --> 00:13:37,636 Speaker 1: they need to find parking and where they need to 232 00:13:37,676 --> 00:13:39,676 Speaker 1: stand in longer line, and where they need to know 233 00:13:39,676 --> 00:13:43,316 Speaker 1: where it is in the first place. Or they're requesting 234 00:13:43,316 --> 00:13:46,156 Speaker 1: a mail ballot, Well, okay, this big envelope comes in 235 00:13:46,236 --> 00:13:49,876 Speaker 1: the mail, and it's not that difficult. But now you've 236 00:13:49,876 --> 00:13:52,276 Speaker 1: opened up these new forms and you have to sign 237 00:13:52,276 --> 00:13:53,996 Speaker 1: it in the right place, and in some places you 238 00:13:54,036 --> 00:13:57,156 Speaker 1: have to stick it inside another envelope. All of those 239 00:13:57,236 --> 00:14:02,316 Speaker 1: changes and how they affect people are really unknown to us, 240 00:14:02,396 --> 00:14:05,876 Speaker 1: and I think that's a great deal of uncertainty in 241 00:14:06,036 --> 00:14:11,636 Speaker 1: this election. And I suspect if turnout patterns vary a lot, 242 00:14:11,876 --> 00:14:14,596 Speaker 1: or by more than we think that they are, that 243 00:14:14,596 --> 00:14:17,836 Speaker 1: that's going to be one explanation why that person in 244 00:14:17,836 --> 00:14:20,276 Speaker 1: the example I just used, Well, they decided they couldn't 245 00:14:20,276 --> 00:14:23,676 Speaker 1: find parking at Dodgers Stadium, which they turned into a big, 246 00:14:23,716 --> 00:14:26,996 Speaker 1: giant polling place. Maybe no surprise there, Or they didn't 247 00:14:27,036 --> 00:14:30,876 Speaker 1: quite navigate the mail ballot process correctly and their ballot 248 00:14:30,876 --> 00:14:33,476 Speaker 1: didn't get there or was rejected or what have you. 249 00:14:33,916 --> 00:14:37,956 Speaker 1: So all of that is incredibly difficult to quantify. I 250 00:14:38,036 --> 00:14:41,436 Speaker 1: suspect we'll get some help in that in the aggregate 251 00:14:41,836 --> 00:14:44,316 Speaker 1: as we get closer to the election, because we we'll 252 00:14:44,356 --> 00:14:49,116 Speaker 1: see is the return rate on a lot of mail votes, 253 00:14:49,396 --> 00:14:53,156 Speaker 1: and we'll also have the total early vote from all 254 00:14:53,236 --> 00:14:56,396 Speaker 1: these early voting locations, so we'll start to have a 255 00:14:56,476 --> 00:14:59,476 Speaker 1: sense of who's done that, how much of it has 256 00:14:59,516 --> 00:15:02,556 Speaker 1: sort of been cannibalized from what would have been election 257 00:15:02,676 --> 00:15:05,996 Speaker 1: day vote, and then we'll narrow down to what remains 258 00:15:06,196 --> 00:15:08,196 Speaker 1: the people who haven't voted yet when we wake up 259 00:15:08,236 --> 00:15:11,756 Speaker 1: on the morning of November third. That'll be a help, 260 00:15:12,236 --> 00:15:15,756 Speaker 1: but from a polling standpoint, that is all very, very 261 00:15:15,796 --> 00:15:30,716 Speaker 1: hard to quantify. We'll be right back. You mentioned aggregation, 262 00:15:30,796 --> 00:15:33,716 Speaker 1: and that's also something that I'm totally fascinated by, and 263 00:15:33,716 --> 00:15:36,116 Speaker 1: it's something that in recent years has become more and 264 00:15:36,196 --> 00:15:39,196 Speaker 1: more salient to anybody who reads polls in the newspapers 265 00:15:39,316 --> 00:15:42,636 Speaker 1: or watches them on television. I'm specifically thinking of the 266 00:15:42,676 --> 00:15:46,076 Speaker 1: aggregation of lots of polls. There are lots of different polls. 267 00:15:46,196 --> 00:15:49,196 Speaker 1: They have different methodologies. Some are local, some are national. 268 00:15:49,756 --> 00:15:53,596 Speaker 1: But there are sites and experts, and you're one of 269 00:15:53,636 --> 00:15:56,916 Speaker 1: them who say, look, here's what my polling data shows, 270 00:15:56,916 --> 00:16:01,316 Speaker 1: but here's what my aggregation of all everybody's polling data shows. 271 00:16:01,876 --> 00:16:04,076 Speaker 1: And there's some implication there, a kind of wisdom of 272 00:16:04,116 --> 00:16:08,636 Speaker 1: crowds idea that it's better to have as many different polls, 273 00:16:08,636 --> 00:16:10,716 Speaker 1: even if they different methodology, some of which might not 274 00:16:10,716 --> 00:16:14,596 Speaker 1: be your favorite methodology, than just to rely on one poll. 275 00:16:14,596 --> 00:16:16,276 Speaker 1: What's your sense of that. Are you one of those 276 00:16:16,316 --> 00:16:19,756 Speaker 1: people who believes that will have better information by aggregating 277 00:16:19,796 --> 00:16:22,076 Speaker 1: lots of polls even if some of them have a 278 00:16:22,116 --> 00:16:24,996 Speaker 1: methodology that you wouldn't be crazy about, or are you 279 00:16:24,996 --> 00:16:28,036 Speaker 1: somebody who thinks, no, we're better off relying on carefully done, 280 00:16:28,396 --> 00:16:31,836 Speaker 1: well done polls and only those polls. The second, I'm 281 00:16:31,876 --> 00:16:35,396 Speaker 1: not a fan of aggregation. I understand why it's done, 282 00:16:35,676 --> 00:16:39,596 Speaker 1: and in general it's not unreasonable. The comparison I make 283 00:16:39,756 --> 00:16:42,396 Speaker 1: is do you want a beat you can dance to 284 00:16:42,756 --> 00:16:45,636 Speaker 1: or do you want to hear the song? If you 285 00:16:45,716 --> 00:16:49,476 Speaker 1: go to a nightclub and the DJ is combining a 286 00:16:49,476 --> 00:16:52,916 Speaker 1: lot of different songs together to give you a really 287 00:16:52,956 --> 00:16:56,556 Speaker 1: good beat, or blending one after another, that's really fun 288 00:16:56,796 --> 00:16:59,596 Speaker 1: and that's really useful because you just want to dance 289 00:16:59,596 --> 00:17:03,156 Speaker 1: and have a good time. If you really want to 290 00:17:03,196 --> 00:17:06,836 Speaker 1: appreciate music and you want to dive into what an 291 00:17:06,876 --> 00:17:10,356 Speaker 1: artist put out as their song, and you listen to 292 00:17:10,396 --> 00:17:14,116 Speaker 1: a song, and that to me is a little closer 293 00:17:14,156 --> 00:17:17,556 Speaker 1: to what the polster is trying to deliver. Any good 294 00:17:17,596 --> 00:17:21,636 Speaker 1: poll is trying to tell you what people think and 295 00:17:21,796 --> 00:17:26,116 Speaker 1: explain why people think it. And they've offered you a 296 00:17:26,156 --> 00:17:30,236 Speaker 1: model of the electorate. They've implicitly or directly offered you 297 00:17:30,436 --> 00:17:34,276 Speaker 1: a model of behavior because of the questions that they asked, 298 00:17:34,676 --> 00:17:37,516 Speaker 1: because of the things they tried to test or examine 299 00:17:37,796 --> 00:17:41,676 Speaker 1: by asking those questions. Is it about the candidates personalities, 300 00:17:41,756 --> 00:17:45,876 Speaker 1: is it about public policies, etc. So they're offering that 301 00:17:46,036 --> 00:17:48,156 Speaker 1: and you can take it for what it's worth. But 302 00:17:48,836 --> 00:17:52,196 Speaker 1: if it's done well, it should offer you a good 303 00:17:52,356 --> 00:17:56,756 Speaker 1: study of the electorate, and you can compare it. If 304 00:17:56,756 --> 00:17:59,556 Speaker 1: you have the time or the inclination, you can compare 305 00:17:59,636 --> 00:18:03,356 Speaker 1: that study to another one that's a really comprehensive and 306 00:18:03,436 --> 00:18:06,836 Speaker 1: I would dare say a little more sophisticated view of 307 00:18:06,996 --> 00:18:10,316 Speaker 1: how to approach getting information. But again it depends on 308 00:18:10,436 --> 00:18:13,076 Speaker 1: what you want, how deep you want to go. So 309 00:18:13,476 --> 00:18:17,476 Speaker 1: the aggregate is for perhaps a shorthand who's winning? I 310 00:18:17,516 --> 00:18:19,636 Speaker 1: can see it from this, Okay, God, move on with 311 00:18:19,676 --> 00:18:22,116 Speaker 1: my life. Do I want to understand what's going on. 312 00:18:22,516 --> 00:18:25,116 Speaker 1: I don't think you get that with aggregation at all. 313 00:18:25,396 --> 00:18:28,036 Speaker 1: I love the analogy that what you do is you're 314 00:18:28,076 --> 00:18:30,996 Speaker 1: a musician. You know, you're a soloist. You're singing your 315 00:18:31,036 --> 00:18:34,876 Speaker 1: song or you're playing your piece of music, and we 316 00:18:34,996 --> 00:18:37,276 Speaker 1: get a certain kind of depth and appreciation out of that. 317 00:18:37,356 --> 00:18:39,916 Speaker 1: And what the aggregators are doing, you know, on five 318 00:18:40,116 --> 00:18:41,996 Speaker 1: thirty eight or you know what the New York Times 319 00:18:41,996 --> 00:18:45,196 Speaker 1: aggregation is doing is they're DJs. You know, they're they're 320 00:18:45,236 --> 00:18:48,116 Speaker 1: playing it all. They're mixing and matching and there's a 321 00:18:48,116 --> 00:18:50,156 Speaker 1: beat that emerges, and you know that can be as 322 00:18:50,196 --> 00:18:51,956 Speaker 1: you say, that can be great too. It sounds like 323 00:18:51,956 --> 00:18:54,236 Speaker 1: it depends on what you want to listen to in 324 00:18:54,276 --> 00:18:57,716 Speaker 1: a given moment or circumstance. I think that's a brilliant metaphor. So, 325 00:18:58,156 --> 00:19:01,396 Speaker 1: but let me ask you, what song are you playing 326 00:19:01,476 --> 00:19:06,596 Speaker 1: right now? What's your feeling at a musical interpretive level 327 00:19:07,196 --> 00:19:12,556 Speaker 1: of the information you're gathering. That's a great question for 328 00:19:12,676 --> 00:19:16,876 Speaker 1: many many people, it is a referendum on the president. Now. 329 00:19:16,916 --> 00:19:20,716 Speaker 1: In some sense that's not unusual because there's an incumbent 330 00:19:20,836 --> 00:19:23,996 Speaker 1: on the ballot, but it is very particularly the case 331 00:19:24,156 --> 00:19:27,916 Speaker 1: now and for Democrats They say that they are voting 332 00:19:28,716 --> 00:19:32,236 Speaker 1: as much, if not more, to vote against the president 333 00:19:32,476 --> 00:19:37,276 Speaker 1: than for Joe Biden. They say that they are motivated 334 00:19:37,356 --> 00:19:41,836 Speaker 1: to vote. At the same time, you have the president 335 00:19:41,876 --> 00:19:46,356 Speaker 1: who has as solid and a core base of supporters 336 00:19:46,796 --> 00:19:53,436 Speaker 1: as we've ever seen, and that kind of allegiance is 337 00:19:53,836 --> 00:19:58,196 Speaker 1: something unlike we've seen. We did a study right before 338 00:19:58,236 --> 00:20:02,076 Speaker 1: the Republican Convention where we ask does Donald Trump deserve 339 00:20:02,156 --> 00:20:05,036 Speaker 1: your loyalty? And many Republicans said yes. In fact, I 340 00:20:05,036 --> 00:20:08,516 Speaker 1: even tried this. We said what's more important to you 341 00:20:09,116 --> 00:20:14,436 Speaker 1: being a Republican or being a Trump supporter? And Moore 342 00:20:14,596 --> 00:20:17,796 Speaker 1: said it was being a Trump supporter. That kind of 343 00:20:17,836 --> 00:20:22,916 Speaker 1: connection is strong. So you have deeply held, very emotional 344 00:20:23,396 --> 00:20:27,956 Speaker 1: connections to their positions on each side. And it's one 345 00:20:27,996 --> 00:20:31,276 Speaker 1: of the reasons that you can explain in the aggregate 346 00:20:31,916 --> 00:20:36,116 Speaker 1: why you've seen the president's approval ratings hold so historically 347 00:20:36,116 --> 00:20:40,036 Speaker 1: incredibly steady throughout his presidency, where the economy has gone 348 00:20:40,076 --> 00:20:41,996 Speaker 1: up or whether it's gone down, no matter what he said, 349 00:20:42,036 --> 00:20:45,596 Speaker 1: no matter what he's done. And it's also one reason 350 00:20:45,676 --> 00:20:49,236 Speaker 1: why you find remarkable stability in the polls this year 351 00:20:49,636 --> 00:20:52,796 Speaker 1: that Joe Biden edge has been about what it's been 352 00:20:52,876 --> 00:20:55,196 Speaker 1: and it's not like twenty sixteen at all, where we 353 00:20:55,236 --> 00:20:59,756 Speaker 1: saw a large fluctuations in the polling. That is on 354 00:20:59,796 --> 00:21:02,516 Speaker 1: the personal level. On the individual level, part of the 355 00:21:02,516 --> 00:21:05,836 Speaker 1: explanation for what you see in the aggregate that is 356 00:21:05,916 --> 00:21:10,276 Speaker 1: the song of twenty twenty. It is those deeply held 357 00:21:10,916 --> 00:21:14,916 Speaker 1: convictions about what people are seeing. And the final maybe 358 00:21:14,916 --> 00:21:18,676 Speaker 1: a coda if you will, on this is that there 359 00:21:18,716 --> 00:21:21,516 Speaker 1: are very different views of what kind of shape the 360 00:21:21,596 --> 00:21:26,716 Speaker 1: country is in, not just differences on how to solve 361 00:21:26,716 --> 00:21:32,116 Speaker 1: the problems. The Republicans feel that the coronavirus outbreak is 362 00:21:32,156 --> 00:21:37,356 Speaker 1: not as bad as the medical folks say that it is. 363 00:21:37,796 --> 00:21:43,556 Speaker 1: They believe that the death toll is overreported. They are 364 00:21:43,676 --> 00:21:48,516 Speaker 1: less concerned about the virus themselves. Democrats, by contrasts, believe 365 00:21:48,636 --> 00:21:53,156 Speaker 1: that the death toll is underreported. They are themselves much 366 00:21:53,196 --> 00:21:57,196 Speaker 1: more deeply concerned about the virus, and there's some personal 367 00:21:57,196 --> 00:22:00,076 Speaker 1: connection and experience in that, and that they tell us 368 00:22:00,076 --> 00:22:04,316 Speaker 1: it's affecting their communities more. The Democrats live in cities, etc. 369 00:22:04,836 --> 00:22:07,756 Speaker 1: That may be changing as they outbreak unfortunately spreads to 370 00:22:07,796 --> 00:22:11,676 Speaker 1: other places. So there's a very different view. And that's 371 00:22:11,676 --> 00:22:15,956 Speaker 1: just one example of what is happening in the country 372 00:22:15,996 --> 00:22:19,236 Speaker 1: that we don't even agree on what the facts are, 373 00:22:19,796 --> 00:22:24,636 Speaker 1: let alone the more traditional or conventional arguments around politics 374 00:22:24,676 --> 00:22:27,876 Speaker 1: of what do we do to solve this? And that 375 00:22:27,996 --> 00:22:31,276 Speaker 1: is something supposed We're still kind of wrestling with what 376 00:22:31,436 --> 00:22:34,956 Speaker 1: happens in that environment. That remains a question for me. 377 00:22:37,276 --> 00:22:41,156 Speaker 1: Exit pulling. How do you do exit pulling in this 378 00:22:41,236 --> 00:22:45,716 Speaker 1: weird COVID year where many people have mailed in ballots? 379 00:22:46,436 --> 00:22:48,916 Speaker 1: How does that game change and how will you do 380 00:22:48,996 --> 00:22:52,196 Speaker 1: the exit pulling this time around? Is it easier because 381 00:22:52,196 --> 00:22:55,236 Speaker 1: of the mail in votes or is it harder because 382 00:22:55,276 --> 00:22:57,276 Speaker 1: of the disparities in the different ways that votes are 383 00:22:57,276 --> 00:23:00,396 Speaker 1: being cast in different places. Well, the mail in votes 384 00:23:00,596 --> 00:23:04,876 Speaker 1: will have to be interviewed by phone, so that'll either 385 00:23:04,956 --> 00:23:08,356 Speaker 1: be a random digit dial phone looking for somebody voted 386 00:23:08,356 --> 00:23:12,476 Speaker 1: by mail, or where possible, you take the voter rolls, 387 00:23:12,596 --> 00:23:16,196 Speaker 1: you know the folks who returned mail ballots, and you 388 00:23:16,236 --> 00:23:18,836 Speaker 1: attach a phone number to that you call them. So 389 00:23:19,036 --> 00:23:21,676 Speaker 1: that's a change. We've always been able to reach out 390 00:23:21,716 --> 00:23:23,636 Speaker 1: and we have reached out to people who voted by 391 00:23:23,676 --> 00:23:26,476 Speaker 1: mail by telephone. That's always been a component of the 392 00:23:26,516 --> 00:23:28,796 Speaker 1: exit polls. However, this year obviously is going to be 393 00:23:28,836 --> 00:23:32,876 Speaker 1: a much larger one in the early vote. What we're 394 00:23:32,916 --> 00:23:36,516 Speaker 1: doing this year, and this is changed or it's rather 395 00:23:37,036 --> 00:23:41,916 Speaker 1: enlarged or expanded, is positioning exit poll interviewers at the 396 00:23:41,996 --> 00:23:46,956 Speaker 1: early voting sites, so they're out there now and they'll 397 00:23:46,996 --> 00:23:50,716 Speaker 1: be interviewing people who are lined up for early voting 398 00:23:50,756 --> 00:23:53,436 Speaker 1: the same way they would have been on election day 399 00:23:53,476 --> 00:23:56,996 Speaker 1: as people are leaving the polling place. And then the 400 00:23:57,836 --> 00:24:01,316 Speaker 1: final part of that, the actual election day exit poll 401 00:24:02,236 --> 00:24:05,876 Speaker 1: is going to have to deal with the fact that 402 00:24:05,956 --> 00:24:09,436 Speaker 1: polling places are consolidated, there may be more people at 403 00:24:09,476 --> 00:24:14,396 Speaker 1: those polling places, and there will be COVID protections in place, 404 00:24:14,516 --> 00:24:17,796 Speaker 1: so the interviewers will be wearing a mask, they'll be 405 00:24:17,876 --> 00:24:21,956 Speaker 1: hand sanitizer at the table. All of those are sort 406 00:24:21,956 --> 00:24:25,316 Speaker 1: of added this year. As far as the actual sampling 407 00:24:25,436 --> 00:24:29,116 Speaker 1: is concerned, people sometimes think I sort of take every 408 00:24:29,196 --> 00:24:31,836 Speaker 1: chance I can to address this. People sometimes think that 409 00:24:31,876 --> 00:24:36,596 Speaker 1: the exit poll goes to a set of Bellweather precincts. 410 00:24:36,596 --> 00:24:39,756 Speaker 1: That is not the case. It is a randomly sampled 411 00:24:39,836 --> 00:24:43,316 Speaker 1: set of precincts, and so that sampling will have to 412 00:24:43,356 --> 00:24:46,836 Speaker 1: be based on where they open up polling places, and 413 00:24:46,876 --> 00:24:50,516 Speaker 1: then people will be deployed out to a sample of them. 414 00:24:51,076 --> 00:24:56,276 Speaker 1: In the more conventional way, decision desk. You sit on 415 00:24:56,396 --> 00:24:59,716 Speaker 1: the CBS decision desk, And these decision desks are fascinating 416 00:24:59,756 --> 00:25:02,556 Speaker 1: institutions to me because I would say, and my day 417 00:25:02,596 --> 00:25:05,436 Speaker 1: job as constitutional law professor, that the decision desks, though 418 00:25:05,516 --> 00:25:09,076 Speaker 1: unmentioned in our written constitution, have come to be crucial 419 00:25:09,356 --> 00:25:13,476 Speaker 1: elements in how we in fact do the constitutional practice 420 00:25:13,476 --> 00:25:15,436 Speaker 1: of deciding who won elections. We don't have in the 421 00:25:15,476 --> 00:25:18,316 Speaker 1: United States, but a lot of countries have a central 422 00:25:18,556 --> 00:25:21,956 Speaker 1: electoral Commission for the whole country that says, here are 423 00:25:21,956 --> 00:25:25,116 Speaker 1: the results. We have fifty states which each have to 424 00:25:25,116 --> 00:25:27,876 Speaker 1: do their processing on their own, and we never wait 425 00:25:27,956 --> 00:25:29,836 Speaker 1: for all of those to count all of the votes 426 00:25:29,876 --> 00:25:33,316 Speaker 1: before we quote unquote call the election. You call the election. 427 00:25:33,396 --> 00:25:36,876 Speaker 1: And I don't just mean UCBS, I mean you, the 428 00:25:36,916 --> 00:25:40,076 Speaker 1: person sitting in charge of the decision desk with your team. 429 00:25:40,996 --> 00:25:45,836 Speaker 1: So how does it work and how different do you 430 00:25:45,876 --> 00:25:49,996 Speaker 1: think it's going to be in this strange year. It's 431 00:25:50,036 --> 00:25:54,116 Speaker 1: going to be different. People may need to be patient. However, 432 00:25:54,196 --> 00:25:57,276 Speaker 1: I don't think if folks pay attention and they listen 433 00:25:57,316 --> 00:25:59,396 Speaker 1: to us and they watch, I'll do a plug here 434 00:25:59,436 --> 00:26:01,716 Speaker 1: if they watch CBS News, but really, if they watch 435 00:26:02,156 --> 00:26:07,556 Speaker 1: a good network broadcast, they won't be confused because and 436 00:26:07,676 --> 00:26:10,036 Speaker 1: this goes to a question on how does it work. 437 00:26:10,996 --> 00:26:16,356 Speaker 1: What we do is we report what we see unfolding 438 00:26:17,156 --> 00:26:22,076 Speaker 1: from an event, which frankly has already happened. Pretty unusual 439 00:26:22,116 --> 00:26:24,316 Speaker 1: from a pulling perspective, right for once in your lives, 440 00:26:24,356 --> 00:26:28,276 Speaker 1: you're not predicting a future event. You're sort of predicting 441 00:26:28,276 --> 00:26:32,076 Speaker 1: a past event. That's exactly right, that's exactly right. We 442 00:26:32,116 --> 00:26:37,076 Speaker 1: are like a puzzle being revealed piece by piece. There's 443 00:26:37,156 --> 00:26:40,356 Speaker 1: votes in a bank somewhere, there's votes at a voting 444 00:26:40,356 --> 00:26:43,196 Speaker 1: center that are being reported, and the puzzle pieces you 445 00:26:43,236 --> 00:26:45,316 Speaker 1: could think of them as sort of county by county. 446 00:26:45,876 --> 00:26:49,196 Speaker 1: As that's revealed, we start to see a picture, and 447 00:26:49,276 --> 00:26:52,636 Speaker 1: what we're doing our best to do is report what 448 00:26:52,676 --> 00:26:56,356 Speaker 1: that picture looks like, perhaps a little ahead of anybody 449 00:26:56,396 --> 00:26:59,156 Speaker 1: else that could see it that doesn't necessarily have the 450 00:26:59,156 --> 00:27:02,076 Speaker 1: tools that we have at our disposal, and then telling 451 00:27:02,116 --> 00:27:05,516 Speaker 1: you this is what we believe has been revealed. So 452 00:27:05,996 --> 00:27:09,076 Speaker 1: that's the big picture of how it works. What we do, 453 00:27:09,316 --> 00:27:13,236 Speaker 1: specifically is we combine a bunch of different kinds of 454 00:27:13,316 --> 00:27:16,396 Speaker 1: data that we're collecting. One of them is the exit polls. 455 00:27:16,396 --> 00:27:19,276 Speaker 1: So we've got a sense from talking to voters who've 456 00:27:19,316 --> 00:27:23,476 Speaker 1: left their polling place. Now, sometimes in a state that's 457 00:27:23,476 --> 00:27:28,396 Speaker 1: completely lopsided, that's enough to make a projection about what's happened. Well, 458 00:27:28,436 --> 00:27:30,316 Speaker 1: we've heard from voters and so and so has a 459 00:27:30,356 --> 00:27:34,836 Speaker 1: fifty point lead. Okay, that person has one after the 460 00:27:34,916 --> 00:27:39,956 Speaker 1: polls have closed. But in most cases, and in certainly 461 00:27:39,996 --> 00:27:43,956 Speaker 1: in battleground tight states, that's not enough. So now you 462 00:27:43,956 --> 00:27:48,076 Speaker 1: start to get county vote coming in, and what we 463 00:27:48,196 --> 00:27:50,596 Speaker 1: do with that is we look for patterns in it. 464 00:27:51,516 --> 00:27:53,996 Speaker 1: You can look for patterns and you can model data 465 00:27:54,036 --> 00:27:56,676 Speaker 1: in a bunch of different ways. So one would be 466 00:27:56,996 --> 00:28:00,636 Speaker 1: to compare it to pass vote. Let's suppose in county 467 00:28:00,676 --> 00:28:05,756 Speaker 1: after county you see Joe Biden running five points better 468 00:28:05,836 --> 00:28:10,076 Speaker 1: than any Democrat has than the last time, typical Democrats 469 00:28:10,156 --> 00:28:13,356 Speaker 1: or whatever. Well, if you get twenty counties and in 470 00:28:13,436 --> 00:28:16,916 Speaker 1: every single one he's in sagurated example, every single one 471 00:28:17,036 --> 00:28:20,596 Speaker 1: he's five points better than past Democrats, well you can 472 00:28:20,636 --> 00:28:23,596 Speaker 1: make a pretty good inference about what he'll do in 473 00:28:23,796 --> 00:28:26,756 Speaker 1: the remaining counties, maybe five points better than what if 474 00:28:26,756 --> 00:28:29,956 Speaker 1: the Democrat did there, and you can extrapolate that out 475 00:28:30,476 --> 00:28:34,156 Speaker 1: to what he might get statewide, and off you go 476 00:28:34,476 --> 00:28:38,356 Speaker 1: to get a statewide estimate and maybe a projection. But 477 00:28:38,996 --> 00:28:41,716 Speaker 1: where it gets more difficult is, let's suppose you get 478 00:28:41,756 --> 00:28:44,956 Speaker 1: county by county and there's a lot of variants in that. 479 00:28:45,116 --> 00:28:47,796 Speaker 1: So Joe Biden in some places is doing five points better, 480 00:28:47,916 --> 00:28:51,516 Speaker 1: someplaces fifteen points better, and someplaces ten points worse and 481 00:28:51,716 --> 00:28:55,196 Speaker 1: some places twenty points worse. Well, there's no clear pattern 482 00:28:55,236 --> 00:28:57,356 Speaker 1: in that. So what do you do with it? Well, 483 00:28:57,836 --> 00:29:00,996 Speaker 1: not much. You wait before you make a projection. What 484 00:29:01,116 --> 00:29:04,396 Speaker 1: I would add after all that is if you see 485 00:29:04,476 --> 00:29:07,556 Speaker 1: a network say oh we're waiting for vote someplace or 486 00:29:07,596 --> 00:29:10,556 Speaker 1: it's too close to call, that's not always the case. 487 00:29:10,716 --> 00:29:12,836 Speaker 1: What I will tell you, when I try to tell 488 00:29:12,916 --> 00:29:17,036 Speaker 1: you is the patterns here are uncertain. We can't get 489 00:29:17,036 --> 00:29:19,596 Speaker 1: a read on this just yet. It's not that we 490 00:29:19,796 --> 00:29:24,596 Speaker 1: don't have any information. It's that that information looks like 491 00:29:24,636 --> 00:29:28,396 Speaker 1: a big cloud instead of a straight line. And that 492 00:29:28,516 --> 00:29:32,476 Speaker 1: I think, I hope is helpful to people as we 493 00:29:32,556 --> 00:29:36,996 Speaker 1: are transparently trying to describe for you what we see 494 00:29:37,596 --> 00:29:42,396 Speaker 1: being revealed as it's revealed. I don't think of myself 495 00:29:42,396 --> 00:29:47,076 Speaker 1: as filling a constitutional duty, and I remind people we 496 00:29:47,196 --> 00:29:51,036 Speaker 1: do not as networks seat anyone in office. We don't 497 00:29:51,076 --> 00:29:55,076 Speaker 1: certify any votes. We are reporting, I would dare say, 498 00:29:55,156 --> 00:29:59,276 Speaker 1: merely reporting what it is that the voters have done. 499 00:29:59,396 --> 00:30:02,636 Speaker 1: And that's not a sort of false humility. It is, 500 00:30:02,676 --> 00:30:05,756 Speaker 1: in fact the case, as much as the bright lights 501 00:30:05,756 --> 00:30:08,716 Speaker 1: and TV cameras make it seem like it's very, very 502 00:30:08,756 --> 00:30:12,356 Speaker 1: important well, and the fact is that the candidates also 503 00:30:12,396 --> 00:30:15,036 Speaker 1: act on that basis of information. At least back in 504 00:30:15,076 --> 00:30:18,396 Speaker 1: the day, when there were concessions and declarations of victory 505 00:30:18,436 --> 00:30:21,556 Speaker 1: that were credible, they were often based on precisely the 506 00:30:21,636 --> 00:30:24,876 Speaker 1: data that you were aggregating. Thank you for taking the time, Anthony, 507 00:30:24,876 --> 00:30:26,876 Speaker 1: really thank you. I really appreciate it. No, thank you, 508 00:30:26,916 --> 00:30:35,916 Speaker 1: and I hope we get to talk more soon. Talking 509 00:30:35,956 --> 00:30:40,196 Speaker 1: to Anthony Salvanto led to a few powerful takeaways for me. 510 00:30:41,316 --> 00:30:44,996 Speaker 1: Most significantly, Anthony made the point that a lot of 511 00:30:45,036 --> 00:30:49,316 Speaker 1: the reliability of the existing polls depends on people who 512 00:30:49,436 --> 00:30:52,356 Speaker 1: say they are first time voters and that they will 513 00:30:52,436 --> 00:30:56,196 Speaker 1: vote for Joe Biden. The lead that Biden has in 514 00:30:56,276 --> 00:30:59,476 Speaker 1: so many polls is, he says, in important ways, a 515 00:30:59,596 --> 00:31:04,676 Speaker 1: product of believing those voters what they do will have 516 00:31:04,876 --> 00:31:09,076 Speaker 1: a large impact on the outcome of the election. Another 517 00:31:09,156 --> 00:31:12,116 Speaker 1: their takeaway is just how different this time around may 518 00:31:12,156 --> 00:31:15,356 Speaker 1: be from previous times. People will be voting in different 519 00:31:15,356 --> 00:31:19,316 Speaker 1: places than usual, and that means that exit polling also 520 00:31:19,436 --> 00:31:22,356 Speaker 1: has to follow a different approach. What's more, there will 521 00:31:22,396 --> 00:31:25,316 Speaker 1: be mail in votes, potentially in very large numbers, and 522 00:31:25,436 --> 00:31:31,036 Speaker 1: those two need to be incorporated into changed models. Last, 523 00:31:31,156 --> 00:31:35,876 Speaker 1: but by no means least, there's Anthony's idea that individual polls, 524 00:31:35,996 --> 00:31:39,316 Speaker 1: like the kind he conducts, give you the music of 525 00:31:39,356 --> 00:31:43,196 Speaker 1: the election, as opposed to the aggregated beat that comes 526 00:31:43,236 --> 00:31:46,716 Speaker 1: from adding lots of polls together, and for him, the 527 00:31:46,876 --> 00:31:50,276 Speaker 1: music of the election. The song of this election is 528 00:31:50,356 --> 00:31:53,756 Speaker 1: Democratic voters very eager to get rid of Donald Trump 529 00:31:53,836 --> 00:31:57,676 Speaker 1: on the one side, and deeply loyal Trump voters, for 530 00:31:57,756 --> 00:32:00,156 Speaker 1: whom being a Trump supporter is even more important than 531 00:32:00,196 --> 00:32:04,436 Speaker 1: being a Republican on the other. In other words, the 532 00:32:04,476 --> 00:32:10,036 Speaker 1: future of our republic is genuinely at stake. Remember to 533 00:32:10,116 --> 00:32:14,356 Speaker 1: tune into our special series Deep Bench this Saturday. Until 534 00:32:14,396 --> 00:32:17,476 Speaker 1: the next time I speak to you, be careful, be safe, 535 00:32:17,676 --> 00:32:21,596 Speaker 1: and be well. Deep Background is brought to you by 536 00:32:21,636 --> 00:32:25,556 Speaker 1: Pushkin Industries. Our producer is Lydia Gencott, our engineer is 537 00:32:25,596 --> 00:32:29,276 Speaker 1: Martin Gonzalez, and our showrunner is Sophie Crane mckibbon. Theme 538 00:32:29,356 --> 00:32:32,636 Speaker 1: music by Luis Guerra at Pushkin. Thanks to Mia Lobell, 539 00:32:32,836 --> 00:32:36,676 Speaker 1: Julia Barton, Heather Faine, and Carlie mcliori, Mackie Taylor, Eric Sandler, 540 00:32:36,676 --> 00:32:39,436 Speaker 1: and Jacob Weisberg. You can find me on Twitter at 541 00:32:39,436 --> 00:32:42,596 Speaker 1: Noah R. Feldman. I also write a column for Bloomberg Opinion, 542 00:32:42,716 --> 00:32:45,316 Speaker 1: which you can find at Bloomberg dot com slash Feldman. 543 00:32:45,876 --> 00:32:49,236 Speaker 1: To discover Bloomberg's original slate of podcasts, go to Bloomberg 544 00:32:49,276 --> 00:32:52,116 Speaker 1: dot com slash podcasts, and if you liked what you 545 00:32:52,236 --> 00:32:55,676 Speaker 1: heard today, please write a review or telefrat. This is 546 00:32:55,716 --> 00:32:56,476 Speaker 1: deep background