Skip to main content

tv   Fred Yang Discusses Polling in Campaign 2016  CSPAN  December 28, 2016 9:55am-10:43am EST

9:55 am
and how race influences the american soul and journalists and editor of washington post david maranais. watch live sunday noon-3:00 p.m. eastern on c-span two. democratic pollster stoke about polling data during the 2016 election cycle at an event hosted by new york city law school. his remarks were p part of a long day long congress foe cushion on components of a system. this portion is about 45 minutes.
9:56 am
[ applause ] plauz. good afternoon everyone. i apologize for sitting down. it's been a long year. there are other parts of my body that need rest. thank you very much. it's an honor to be here and to help walk us all through what happens on november 8, and what that means going forward. if i am a little distracted, i apologize. the only thing that is more important to me than this is my daughter's basketball game. they are playing right now. i am the unofficial statistician for the team. i told him i want going to be there today because i was doing this, they said oh yeah, you're the guy who said hillary
9:57 am
clinton would win. maybe we will get someone else to do our stats from now on. a couple of things and then i would average all of you -- heard all of you to speak when the mood strikes if you have questions. number one, i come here not to praise pulling per se but also not to condemn on public polling. the national polls for the most part got the numbers almost right. at the end of the day mrs. clinton will win the popular vote by two points. an aggregation of all the publicly released polls before the november 8 election had the average clinton margin of 3.4%. number two, those same polls had the average for johnson-stein at 7%. they ended up getting 4.2% in real life. the third factor
9:58 am
that was part of life and polling is that the national polls, even the ones at the end had an average undecided of 5%. that is the one advantage that an actual election has over polls. when you go to a polling booth or when you sell out your ballot at home, you have two options, voting for one of the candidates or not voting in that race. in polls we offer the undecided option. so in election day there was no undecided on election day. there was 5% average undecided in a national poll. if you take that plus the 3% or more than johnson and stein got on election day and
9:59 am
added to the volatility of the national polls and in particular if you take it one level down, obviously we don't elect presidents by the popular vote, they are elected by electoral college, if you factor, if you think about those dynamics in the national polls, they were even more prevalent in the battleground polls, especially in wisconsin, michigan and pennsylvan pennsylvania. number two, when you look at the three states where the presidential campaign was one, or audio the side lost. wisconsin, michigan and pennsylvania. in addition to the volatility of public opinion research, there was volatility in real life. one statistic among the many that i've tried to remember is that in the exit
10:00 am
polls, and the state of wisconsin, 10% of voters who were surveyed in days of polls said they had made up their minds on who to vote for president in the last couple of days in the election. so in the last couple of days in the election, it could've been tuesday, it could have been monday, the second coming letter hit on friday afternoon. in the state of wisconsin, among the 10% who said they voted in the last couple days of the election, i would wager that most of those people were after the second coming letter. trump was ahead of clinton by 27 points. one of the understated questions i'm answering is what happened with the polling. what happened with the polling is that the margin of error, the undecided, the fact that johnson and stein got 7% and 4% in real life. it is the normal
10:01 am
volatility in a public opinion survey. and the volatility of fluidity in real life. i would say that we would be having a slightly different conversation, regardless of what party you are for if 52,000 or so has switched between wisconsin, michigan and pennsylvania. mrs. clinton lost those three states by roughly 106,000 votes total. if 53,000 had changed hands, maybe we would be talking about how close the election was but the election that she had been ahead. a couple of other things about how to read what happened on election day. i will argue that analysts from both parties will spend time going through the numbers and the results because i really believe that what happened on november 8
10:02 am
isn't something that was short term. it didn't happen in the last three or four days. it didn't happen in september, it didn't happen in 2016, arguably, a lot of what happened or didn't happen on november 8, 2016 was a result of long trends. we were part of the polling team for nbc wall street polling firm. our national surveys which do with the republican pulling from show that the wrong track was in the 70% range. when asked if they wanted to elect a president who would make major changes, even if they don't know what those changes may be, or would you like to elect someone who will make steady progress? five to four in the last poll, they
10:03 am
wanted major changes. they are biased against people like me. i am toward the end. we are getting 60 or 65% of americans. i am saying that these -- system does not represent me anymore. through the same surveys, that suggested that ms. clinton had the advantage electoral. you can see signs of discontent within the electorate. i think one of the things, i will speak for myself and not necessarily for my republican colleagues, one of the things that we are really trying to reckon with is that i do believe there are some things that polls got right. one of the things that polls got right is that i do think donald trump was
10:04 am
the most unpopular person to run for president as a major party nominee. i believe that is true. exit polls will show that a lot of trump voters voted for him not because they were voting for him but were voting against somebody. that is the thing. how can someone with numbers and and his negatives were 50 8, 50 nine, 60. those are tremendous negatives. the fact that he could win despite that is something that i think we are still trying to figure out. you want to give the trump campaign and donald trump there do. this was not just people voting against something. there was clearly something they might have been voting for. that is one of the things that we are trying to figure out. the other interesting thing about the election is that there are some
10:05 am
states where it is not clear that there was a so-called trump phenomenon. state of california, nevada where we picked up a u.s. senate seat and two congressional seats, my home state of maryland, in maryland in 2016, chris van hollen one that senate seat by 25 points. there are areas in maryland where a lot of blue-collar voters lived for example, baltimore county. baltimore county in 2014 voted for the republican candidate for governor by 20 points. so i was thinking that if there was a trump phenomenon nationally, you might see it in a place like baltimore county which voted for the republican candidate for governor by 20 points in 2014. a
10:06 am
month ago, all the more county, that same place that voted for the republican candidate by 20 points voted for hillary clinton over donald trump by 17 points. as a pundit and analysts and as people who are interested in politics, you want to see uniformity or you want to see a trend. i do think that it is one thing to look at the numbers nationally. you really do need to look at individual localities. again, there is not a uniform trend as to what happened in -- on november. one thing that i think is something to look for is what a lot of the analysts and pundits have noticed since november is the
10:07 am
increase or higher turnout in small-town, rural america. read america getting reder. i do think that we need to be careful about applying a uniform theory to the entire country but there are parts of the country where areas that romney, mccain, other republicans did well, trump did even better. turnout in those areas were if not the same in 2012 a little higher. one of the areas like that that comes to mind to make, i think i will close by saying this really speaks to what happened to democrats in 2016 is elliott county in kentucky. it is in
10:08 am
eastern kentucky number 1. elliott county kentucky is very democratic. there is party registration and kentucky. some states don't have that. and elliott county, for every one republican, there are 10.5 democrats. third, elliott county, kentucky had voted for every democratic candidate for president from 1868 to 2012. so every single presidential election, elliott county kentucky had voted for the democratic candidate. in 2008, barack obama defeated john mccain in elliott county, kentucky 61-39%. in 2012, barack obama beat mitt romney 50-47. in
10:09 am
2016, donald trump defeated other clinton 70%-26%. when you get his summers in elliott county kentucky, it feels like montgomery county, maryland. they are 48 hundred registered voters. is a small sample size. in closing, people ask me in particular as a democratic strategist what happened. what happened in michigan. what happened in wisconsin? i would argue that we are still looking through all of the results. i think that the anecdote of elliott county, kentucky and i know this because i studied pennsylvania was replicated in elliott county, kentucky, michigan and wisconsin. 4800 votes by themselves doesn't sound like a lot but there are
10:10 am
30 counties like that, it does make a difference. what i would argue again was a historically close election. i think that one of the things on our side of it aisle that democrats will look at is not only the lower turnout in some of the urban areas of the country which also hurt democrats but also the issue of a place that voted for barack obama by 33 points, just eight years ago, what happened to make them turn around and vote for donald trump by 44 points one month ago? [ applause ] >> thank you very much. >> we have time for a few questions before the reception. i thought that if you would like
10:11 am
to put a question over, we have to tear on the side who could take some questions and bring them to us. i'm going to start by asking you fred if i could. i will encourage anyone if i understood you correctly, that is the analysis. how could it occur? i don't want to hold you to your own theory about that. what are some of the competing theories about that? >> another we are talking specifically about elliott county, kentucky. >> what might have driven the politics of that kind of electorate? >> elliott county is in eastern kentucky. it is not close to the region that pours west virginia.
10:12 am
i think a lot of it was economic. i think a lot of people saw job loss over the last few decades. jobs that had not come back and look, i think there is a famous thing from 1992 that the economy is stupid. i think that a part of what happened in the elliott county, kentucky's of the world was economy. that is not the only thing. one of the interesting questions from the national exit polls was that the exit polls asked the electorate what the most important issue was to you. they looked at how people voted who picked those issues. the top issue in 2016 nationally was the economy. the 2% of americans said that the economy was the most important issue. among those people who picked the economy, they voted for
10:13 am
hillary clinton by 52-42. that is a good margin. if the economy number one issue -- you kind of had -- she knew to win the economy more. we would have hoped that the advantage on the economy would have been greater. they asked another question. which one of the following characteristics or approaches do you want the next president of the united states to have? the number one answer was changed washington. among those voters who said the top quality was change, they voted for donald trump by 83-84%. there are competing variables out there. issue was, it was about the economy, jobs and in particular,
10:14 am
job loss. there was another strain out there that i was referencing at the beginning that what happened on november 8 was not just november 8, it was a combination of things. every poll shows this. the dissatisfaction with washington. not just democrats, not just republicans but both parties. the sense that the political and economic system was not working for people like them. look, if i said, and i believe this is true that donald trump was the most unpopular president to run for president, he also clearly was one of the most unique people to run for president and if there is a strain out there of voters, americans want dramatic change. what better way to express that than voting for someone who was as different from anybody else
10:15 am
who had run from president regardless of how popular or unpopular he was. >> thank you, we are getting a lot of questions. he will be here until about 10:00 p.m. without knowing which order to begin, assuming that the polls were more or less close, there was an impression that people received. the secretary clinton was on the road to a certain victory. is there an issue of concern that you have and other pollsters would have about polling data and how it is communicated to the public? >> that is a great question. since i have been relieved of my duties as official statistician to the middle school team, i have a lot of time to talk about
10:16 am
this. which came first question the chicken or the egg. it is true that if you're a political analyst, if you're a party of -- broadcast or newsprint, you look at a lot of different factors to help you basically understand was going on and look ahead to what will happen. polling clearly plays a role in that. i think the 10 national polls that i looked at to come up with these numbers, there were 10 national polls that i'm referring to. only one had
10:17 am
trump ahead. of the last 10 polls that were released, nine of them showed hillary clinton ahead. you're going to say that hillary clinton is probably going to win. now, i do think there was one poll who had trump ahead. again, i am not saying that rerun the election. i'm not saying us to a do over. that poll that had trump ahead was actually wrong. ms. clinton did win the popular vote. i think that bob and everyone, the one critique i would make of polling and those who analyze polls is that when people see a number that is 44-40, to think that 44-40 means-four -- means 44-40,
10:18 am
the margin of error -- the other way of answering the question is, what if we had no polls? don't everyone clap. if we didn't have any polling, where would that leave us? i am not as old as i look but i've been doing this for 30 years. it is amazing when i started out with peter and jeff 30 years ago. when there is a poll, remember i'm a poster. i got excited. i said there is a poll. it is in illinois. let's look at it. in 2016, there was a poll every day. i think that not treating every poll the same, it will
10:19 am
help. what of the track records in the various states? how long were the polls in the field? i do believe that there was volatility with voters. even to the end. some of the reason why the polls missed the election. most major polls stopped polling saturday, sunday in the election. i know, that is late. i know people are thinking, saturday, sunday, the election was tuesday. as we know, public opinion can move very quickly. i don't think is the only reason why but i do think that the fact that the second comey letter came out on friday afternoon and then there was a saturday,
10:20 am
sunday, monday, tuesday, i think that did have some impact in changing. i bet if we pull tuesday morning, we would have found a slightly different result. >> as turned up, that is the next one. the comey letter. how do you see the impact of that letter? what are his strengths and with the letter. >> i think that there are going to be lots of theories about what happened. some of them may
10:21 am
even be true. i think -- look here. we did the nbc wall street journal poll. we had mrs. clinton ahead by four points, if i recall saturday. we got out saturday and it was seasoned morning. it looked like she was the internal and subgroups. it was a national poll and not a poll of all 50 states. it looked like she did not have it locked up, but no -- it looked like she was headed to win. but even in that poll, she had dropped from a previous poll. the other thing in polls is not just what are the results now, but what were the results previously? the trend for her was the race was narrowing, but at that point, not that trump
10:22 am
was gaining, but that some of her vote was going to johnson, stein, or undecided. a second comey letter was exculpatory. but in campaigns, especially when you are a candidate with a strong headwind, 70% in the wrong direction, the majority wanting a president who would make major changes, two thirds or so saying the political system was biased against people like them, conditions that would favor an insurgent candidate. she was in the incumbent, per se but she was. trump was the insurgent. you could see all of those things making life difficult for the clinton
10:23 am
campaign. this is something all of us will be thinking about and analyzing for the weeks to come. i do think the second comey letter put the whole issue back on the public consciousness. if you think about it, it is like a courtroom. the last argument voters heard from both sides was email. she didn't do anything wrong, but the email, once again. this is something where people should check the record. i think one thing i noticed was how someone seems to have taken away his twitter account. i don't think he commented personally on the second comey letter. instead, the trump
10:24 am
campaign was he was going to wisconsin and pennsylvania and that was his closing argument. there are a couple of theories and all could be equally true. when you have an election like this, it's really not one thing you can point to. it's a combination of things. i happen to think that's the reality of what happened. >> okay. thank you. next question. this is about kentucky. i assume it's for to be seen more broadly. how would you estimate the role of sexism and the first nominee being a woman. how did this go into the enormous role in the county?
10:25 am
>> i believe at some level, it had to be a factor. i don't think it was a number one issue. i think there was a lot of strength and weaknesses hillary clinton brought to the campaign. if her gender had been reversed, what would have happened in kentucky? i really think the result probably would have been the same. so you change gender but you change the party? do you change the experience and resume? all of the things you could see in the polls, the headwinds she was dealing with,
10:26 am
the dissatisfaction with government, happiness with the status quo, voters wanting change, i think gender might have been an issue but there were other factors that were equally more important. >> thank you. the median income of trump voters is 71,000 voters and rural noncollege educated white voters only made up 17% of the electorate. how does that square with the narrative of how trump won? >> those are broad numbers nationally. in some of these battleground states, there were higher proportions of those groups plus i think the numbers might have changed a little bit.
10:27 am
in the state of wisconsin which mrs. clinton lost by 11,000 votes, there were roughly 30,000 fewer votes in the city of milwaukee in 2016 than in 2012. i think the proportion in wisconsin of that last group, it was 17% nationally, i think it was maybe 21% in wisconsin. take that, take the lower turnout in a strong democratic area like milwaukee and all of those small things add up to an 11,000 vote defeat. when you have a situation where 104,000 votes out of 12 million or 14 million cast in pennsylvania, michigan
10:28 am
and wisconsin, i don't think there's just one thing you can point to. it's not like 100% of wisconsinites were rural, noncollege educated white voters but if they were one out of five lower turnout in urban areas, that adds up to a narrow defeat. >> thank you. i have two questions here. we don't want to hold you longer with how we're planning the day. do you think a general population to urban areas make it less likely for this kind of result to occur and related to that, what about the apparent absence from the levels of
10:29 am
participation among young people? >> i think one of the disappointments on the democratic side of things was how we were hoping and expecting or robust turnout with millennials, for example and it didn't happen. we know they are out there. we know that when they vote, they vote democratic, but one of the challenges of 2016 was they did not vote in the numbers we hoped for. is it a message issue for the democrats? was it a candidate issue for the democrats? was it organizational or realistic? i don't know. again, we talk about a lot of the analysis on
10:30 am
national terms. i think in california, if you look at all the california results, you would not know donald trump had been elected president of the united states. turnout among latino was fairly high and i think as evident by the press, both parties but in particular, democratic parties are going through a period of self-interest section and analysis. we know we have the votes to win elections, especially national elections. what was the mist in 2016 we can avoid in 2018 and especially 2020? >> thank you. exit polls suggest 30% of latinos voted for trump. do you have confidence that
10:31 am
number is within the ballpark? it seemed higher were. >> yeah, we are not quite sure yet of the 30% number if it is quite right. they are the growing demographic group in american politics. 30% would seem to be a high number for me. especially given the things he said during the campaign. i was fortunate enough to work on catherine cortez masters campaign in nevada and i'm sure since she won in nevada and hillary clinton won in nevada that tlooeths he did though the get 30 percent of latinos.
10:32 am
he did not do that well in california. maybe florida is a different issue. maybe texas is a different issue, but 30 percent would seem high to me. we really have to to look at it state by state. >> just combining two more and then there's one after that. thank you for hanging in there. the question is not whether in fact people see the polls nationally, whether what we have is a problem with the numbers or the way the numbers are red. i suppose your answer to that is interpretation has a lot to do with it. >> i would say both. the last 10 national polls had a clinton advantage of plus 3.4%. as of a couple of days, she was ahead in the national popular vote. that
10:33 am
is not that far off. but there was a sense from the polls and analysts that it was an election mrs. clinton was very likely to win. look, if you are a normal human being and see nine out of 10 polls say she is ahead, you are going to draw the not an obvious conclusion that she's probably going to win the election. i also think the other intangible -- and i hate to keep going back to this -- that i'm a pollster and like a good candidate stays on message, one of the things, if you hook all of us up to a lie detector, we would have said yes, we believe
10:34 am
she is going to win. but it's going to be difficult. and there are things where you could see donald trump could win. but the big sticking point to going over the other side, which is where we are now, that he could win, was how unpopular he was. i think one of the things we underestimated as analysts and maybe i should say me and not we so as not to condemn everybody else, but we said this in our analysis, but it's not as provocative as saying he was the most unpopular person to run for president. mrs. clinton was the second most unpopular person to run for president. at the end of the day, the analysts probably should have given that more
10:35 am
weight. if it's a flip of a coin, where -- were there other things that should have and looked at. beyond the 3.4 percent ahead. the right direction and wrong track. the fact that a lot of americans still believe that they had recovered from the recession. approximate the strong majority who said the economy situation of this economy was biassed against them. if we paid for amention to the signs that were in plain view but we were distracted by donald trump's not only the unusual person but the most unpoplar, we could have given a clearer view of what did happen. >> this is the last one and
10:36 am
combining two. whether it's responding to the experience or whether it's the turn out or trying to decide how to accomplish with the number of the voters and the number of cell phones, do you think more generally there will be a movement to find a way to adjust to this kind of issues? >> i know this is the last question but i feel like i need a couch. to lie down on. look the short answer is yes. the polling industry goes through this because as you will recall, in 2012, it was more that the republican pollsters had quote unquote got the race wrong. if one side or the other gets it wrong, we are professionals. we don't like it if anyone gets it
10:37 am
wrong because our job is to get it right whether you are a democrat or republican pollster and we did do a reflection as the industry after that. it wasn't just a republican thing, it was the polling industry. i know some people may not believe this, but every pollster out there, whether you are democrat or republican, we really do want to get this right and one of the challenges for industry is a good poll tries to eliminate bias. a particular bias and who is being selected to take the poll, it was a heck of a lot easier in the good old days 10 or 20 years ago when the predominant indication device of the american public was a landline. it's obviously gotten a lot more complicated now and a
10:38 am
lot more diverse. i think polling still has a role. it is important to know what people think, but not just starting on november 9, we started 2, 4, 6 years ago, what is going to be the best? what's going to be the most efficient and effective way to truly get a random sample of people that insurers if you are talking about a national poll that has as much chance of getting contacted if they lived in anchorage, alaska as if they lived in maryland? when my mentor and your friend, peter hart, was doing this, a lot of the polls were done door to door. face-to-face. clearly, if you're trying to reach people,
10:39 am
that's probably quote unquote the best way of conducting research because even if you don't have a landline, someone can get you. the issue with that, when i look back at our old polls and it was considered state-of-the-art, a typical statewide survey would take 30 days. if bob bauer is not home, i have to schlep five walks over and talk to someone else. it is a time-consuming process. number one, polling needs to be accurate. polling needs to be effective. polling needs to be efficient. 15 days or 20 days just is not practical given
10:40 am
where we are now in society and we have technology, it sounds like the line from the $6 million man. we just have to figure out what this new polling world looks like. maybe it's a blend of different technologies. maybe it's something else. we are all striving to figure out how to do our jobs better. >> you are one of the nations premiere cultures and we are grateful you could be with us today. thank you very much. [applause] >> we are moving to the reception upstairs. thank you. while congress is on break this week, we're showing american history tv programs normally seen only on the weekends. tonight a look at world war ii. it starts at 8:00 eastern with spies and code breakers and
10:41 am
followed by the fbi vins gags of a spy ring and then world war ii veterans on resistance in paris and then the start of what is now the cia. american history tv prime time tonight here on c-span3. >> sunday in depth is going to feature a live discussion of barack obama. we're taking the phone callings, tweets, e-mails and facebook questions during the conference. it features april ryan and author of the presidency black and white. my up close view of three presidents and race in america. princeton university professor eddie and then surprise winning and editor of the washington post david maraniss. watch in depth live from noon to 3:00 p.m. eastern on sunday on
10:42 am
c-span2. >> the presidential inauguration is friday january 20th. c-span is going to have live coverage of all the days events and ceremonies. watch live on c-span and and listen live on the free c-span radio app. >> the new york university school of of law hosted a day long forum on the u.s. political system. in this panel experts said whether a third party can succeed in a two party system. this is about an


info Stream Only

Uploaded by TV Archive on