Affected our National Politics for decades. Campbell spent 20 or centralism before becoming a professor in the Communications Studies program at American University in washington, d. C. He is also a writer, historian, be the clinic and blogger. He has solo authored seven books including getting it wrong, defunct in the greatest myths in american journalism, and 1995, the year the future began. Tonight campbell will talk about is brenda book from university of California Press, lost in a gallup. Following his opening remarks i will moderate a q a session. Those watching us live on crowd cast can submit a question by clicking ask the question at the bottom of the screen at any time during appropriate. Now i would like to welcome w. Joseph campbell to let talks. John, thank you. Great to be a period great to be back in i was a, if virtually. I spent productive period of time but two and half years ago in i was city going through the Gallup Organization papers university of iowa city special collections. It was very useful and revealing visit and i must say i had a good time in iowa city. Wish i could be there in person rather than virtually but this will have to do for now. One of the surprises, mild surprises if you will, about the gallup collection at the university was a small folder of cartoons about polling. These were cartoons that were currently fairly collected by one of george gallops sons pick the dates of some of these polling cartoons was after his death in 1984 sublease its one of one of his sons who collected these cartoons. I have a few i would like to just show at the outset of the program tonight. If we can go to the next slide, john. This is a mildly amusing cartoon that appeared in the wall street journal. Im not sure about the date but nonetheless it is kind of amusing. So was the next one, too. A woman says ive got to see doctor gallup. Ive changed my mind. Presumably a poll respondent. And then the next cartoon is one that appears in my book lost in the gallup early in the book on page nine disappears. The cartoonist richard rice came up with this amusing characterization of rogue pollsters, when good pollsters go bad to ask who cares what you think with their interactions with the poll respondent. Just a small collection of cartoons, nonetheless im using an kind of revealing. Tonight presentation will focus on a few cases although not all but a few cases of polling philly and u. S. President ial elections. We will also take up a few takeaways and reminders about election polls and i will offer a couple of suggestions about what to keep in mind and what to look for this fall. Then we would go to q a. The presentation is drawn from i just published book lost in a gallup polling failure in u. S. President ial elections. And its a book, a handsome book, that was brought up just recently by the university of California Press. A great press to work for. This talk to matt will consider just a few cases. I mention specifically those from 1948, 1980, 2012, and 2016, polling failures in those president ial elections. This is not a complete universe of polling failures but it covers some of the betterknown so well go to the next slide and that when takes up the Dewey Defeats Truman election of 1940. This was an epic poll failure and which the polls got it completely and utterly wrong. George gallup and of the pollsters forecast a certain victory the republican candidate thomas dewey. And dewey ran what i call a light path campaign. He very seldom invoked controversial points of view. He tried to just smoothly run through the fall election, the campaign, and not upset anybody or not take any controversial views or positions. Whereas harry truman ran a very aggressive campaign, and truman was recognized as being behind in the polls pick a fact one poster elmo roper, one of the early figures in Public Opinion research, announced 72 years ago today after september 9, 1948, that he was so certain that thomas do is going to win the election that he was no longer going to be reporting poll results. He would take polls but it just wouldnt report the results because he didnt think it would add too much to the understanding of the race in 1940. That was emblematic of the conference, supreme confidence pollsters had in the outcome of the naked break election. Harry truman won by 4. 5 percentage point. As election as election one of the comics said it was the first time, truman was the first candidate to lose in a gallup but when in a walk. 4. 5 percentage point victory is fairly modest but pretty clear as well. And to think although its hard to measure the shock of 1948 was probably greater than that of just four years ago in 2016 when donald trump wanted unexpectedly. The shock was just really ran deep and it was emblematic of the shock was the probation of the Chicago Tribune, one of the early editions of the day after the election that declared Dewey Defeats Truman and it is one of the most memorable, Iconic Images of american politics. Harry truman was on his way back to washington from missouri, and at a stop in st. Louis and unionization in st. Louis. Some gave him a copy of the front page of the Chicago Tribune and he held it aloft in what is a very memorable photograph. Journalists afterwards really criticize themselves having delegated the responsibilities, their legwork to the pollsters, that they relied too heavily on polls. So what went wrong in 1948 . A number of factors contributed to dewey upset victory. One of the factors was one of the thirdparty candidates, the Democratic Party next 40 by the way split into three factions. It was a mainstream democrats represented by harry truman and then there was a progressive wing that henry wallace, former Vice President under franklin roosevelts, was a leader of, and then the third split in the Democratic Party was the dixiecrat party, the states right party. States rights party was supposed to Harry Trumans civil rights measures and broke off and form a separate party led by Strom Thurmond of south carolina. During the election the support for the Progressive Party dwindled dramatically, and the beneficiary of the loss of support was harry truman and his campaign. Thats one factor. Another explanation for what went wrong in 1948 is the pollsters stop pulling pretty close to the election date, by midtolate october they were done with the polling. Elmo roper who announced in september he was not going to take polling or another polling result in a longer did conduct a poll late in october but it didnt show much in the way of changes so we it didnt reports poll at all. The pollsters figured that nothing much is going to change and it didnt continue folding right up to the end. This is a lesson that pollsters learned and relearn every so often a president ial elections. Another contributing factor probable was Republican Voters were so confident that thomas do is going to win that the polls and the pundits and the press all said he was headed for victory, that many Republican Voters decided not to turn outcome decided not to vote. That overconfidence perhaps translated into a deficit for thomas dewey. Those are some of the factors that explain the loss of dewey 1940, the loss of the pollsters as well. We can move on to the 1980 which was another surprise outcome when news organizations had entered the polling realm in large numbers, New York Times, cbs news among others without doing their own polls or commissioning their own polls, and polling by them 40 years ago was more numerous than ever. The polls indicated that president jimmy carter was locked in a very tight race with republican Ronald Reagan. The polls were consistent in saying so, and yet on election day Ronald Reagan wins in a landslide, almost ten percentage point victory in an outcome that no pollster had anticipated. Afterwards, pollsters bickered and quarreled among themselves as to what went wrong. Their dispute spilled over unusually so into the public realm, as the article from Los Angeles Times suggest, pollsters spent over why they erred so badly. So what went wrong in 19861 of the factors was the fact that the only debate between the two major party candidates, reagan carter, took place very late in the campaign, a week before the election. That seemed to have been a factor in keeping support to Ronald Reagan. The people could see that he conducted himself well on the stage with carter and that he was not really as loud indocentric as many people thought he was. That was reassuring and that probably contributed to reagans sizable victory. Another effect was the pollsters again did not pull up until the very end. To realize they ought to do that but for various reasons did not continue polling over the final weekend of the 1980 election. Also there was a thirdparty candidate that year john addison who is running as an independent, a republican running for an independent for a while in the fall of 1980 and look like use go to great a lot of votes from Ronald Reagan. As election day approached andersons support dwindled and reagan was the beneficiary. Those are some of the factors that explain this unanticipated outcome, this near landslide that no pollster anticipated 40 years ago. We can take on the next slide, john. The next case is that of 2012 when Gallup Organization was essentially alone in calling the election for estimating the election in mitt romney is vivid. Throughout the campaign the gallup polling kept signaling that mitt romney was ahead four, five, six, seven Percentage Points. At the end of the Campaign Gallup suggested it was a very tight race but romney was one point ahead, and in the end obama, president barack obama wins reelection by nearly four Percentage Points, a fivepoint miss, a real embarrassment for the Gallup Organization. It was also the year in which nate silver confirmed his status as an election oracle come if you will. In 2008 he had, through a poll based a physical model that he developed, estimates that become accurately and 49 out of 50 states. That was in 2008. In 2012 accurately forecast the outcome in all 50 states, and he was recognized as this statistics guru, is forecasting guru that help to his reputation and two signals rise of Data Journalism as a way to also interpret polls and Public Opinion. It was an embarrassment for gallup and a confirmation for nate silver. And then the 2016 election inevitably is one we remember quite clearly. It was the night that was not supposed to happen, according to the public editor of New York Times in a very memorable phrase that she used. And the shock randee. As i mentioned earlier it was probably akin to that maybe not as deep as 1948, but nonetheless shocked that night grand prix keep because Hillary Clinton was widely expected to win and perhaps fairly easily. And what happened in 2016 was key polls in battleground states, particularly in upper midwest, wisconsin, michigan, pennsylvania looked as if they were headed as going to give the outcome to Hillary Clinton in those states, and had she won those three states you would have had enough electoral votes to win the election. Instead, donald trump narrowly wins wisconsin, michigan and pennsylvania, and he also takes battleground state such as florida and North Carolina and ohio, and that combination of the states swept into the presidency, and Electoral College victory. It was a split second. Hillary clinton included won the popular vote but trump won the Electoral College vote. What went wrong in 2016 . An argument that are still being discussed and arguments another side are many sides are still being made but nonetheless it appears that holes in these key states either into their polling too early or failed to wait their percentages, failed to wait the results in a statistical adjustment that pollsters inevitably make. They fail to specifically adjust for college, noncollege educated voters who went to trump fairly heavily. That is one interpretation of some of these erratic polls in key midwestern states. And its also pretty clear that trump picked up undecided voters in large numbers toward the end of the race, and yet more undecided swing into him and Hillary Clinton had swinging to her. So the combination of factors was enough to probably tip the Electoral College to donald trump. Its a scenario that some people suggest could happen again, could happen again in 2016. My research into polling fairly president ial elections suggest that just doesnt know to president ial elections are like him know to polling failures are the same either. So its not likely well have a carbon copy of 2016 this year, but we will see in what, eight or nine weeks. So what do these cases tell us . What are some of the takeaways and reminders . Obviously its pretty rare for a president ial election not to be characterized by some sort of polling disputes. Holding controversies are commonplace in president ial elections, and we cant expect to see them this year. The types of polling failure, the variety of polling failures is not the same. We had seen just in this brief presentation format different types of polling failure. There is the epic failure of 1948. There is the landslides that pollsters did not foresee in 1980. There is the venerable pollster who gets it wrong, embarrassment so, they Gallup Organization in 1980. I mean, 201212. And also another type of polling failure are polls in key states that upset the national outcome, as would happen in 2016, again wisconsin, michigan and pennsylvania. Its also interesting that polling to the end of the campaign right up to the last weekend before the voting, before election day anyway, is a lesson that not all pollsters have learned or that there was put into effect. You see this happen time and again. We saw happen in 1948 and was sought in a few key polls in 19 in 2016. Another take away is that polling failures often are correlated to journalistic failures, and in the sense that journalists often take their lead from polls, preelection poll. Polls are central to how journalists understand and interpret president ial campaigns. It is essential that polls are essentials to have a set and fakes and pursue the campaign narrative. So when polls messed up, journalism can falter, too. So journalistic failure is often equated to polling failure and that may not be so surprising but its something we dont always keep in mind. Finally, what might we see in 2020 . What we are likely to see this year . Polls are going to be more numerous than ever. Seems like its hard to escape the polling deluge and were only in september, still eight weeks away. Even now whole results seem to be all over the place. Real clear politics which is an enviable and very evenhanded political aggregation site just today posted several polls that show joe biden is ahead of donald trump by a range of two Percentage Points to 12 Percentage Points. So polls seem to be all over the place and some are suggesting a close race, some are suggesting something less than close. We will be seeing probably polls with some erratic swing to them. And one of the reason for that is polls are being done by a variety of methodologies anymore. Theres phones, cell phones, robocallers, internet panels, even social media platforms are being asked for insights into Public Opinion in president ial elections. There is no single old standard for polling any longer. That Gold Standard used to be random digit dialing telephone calls with a live operator, but the Response Rates to those calls, to those rated digit dialing telephone calls, Response Rates are dropping, dropped into the Single Digits so it makes it very, very difficult and very expensive for pollsters to use this method and try to get a good sample of opinion. So people in the polling business are looking intensively for the next Gold Standard, if you will, the next approach to polling where theres going to be reasonably accurate and reliable and not terribly expensive to do. Pulses been looking at this for nepotistic the first internetbased polling with the 1999 at the end of the 90s, so we are still in this period of a good deal of extreme edition and good deal of churn among pollsters trying to find what the next best standard is. Its also important to keep in mind that polls are not always wrong. They are done by people who are professionals and have a strong stake for the most part in the outcome in being accurate and reliable, in offering the public a good idea, a reasonable, a reasonably accurate clue clue as to whats going on. But pulse of wrong often enough as weve seen just in a very small tonight. Polls have been wrong often enough and they have a checkered record so that i think its really advisable and its not a bad idea to treat them warily, to be a little bit skeptical about polls and the polling numbers. And polls like the wedge is referred to, the deluge of polls showing the race