The following is an excerpt of a conversation between History Factory’s managing director and host of the History Factory Plugged In podcast, and Time magazine Senior Editor Lily Rothman. Ahead of Election Day, the pair sat down to discuss the history of the Gallup Poll and what lessons we can learn from history about the electoral polling process. For the full conversation, listen to the latest episode of the podcast.
Jason Dressel: Lily, welcome to History Factory Plugged In. You wrote an article following the 2016 election, and certainly a name synonymous with the industry is the Gallup Poll. Who was George Gallup, and how did he get his start?
Lily Rothman: So by way of background, as an editor at Time, part of my purview is stories using our own archives. And George Gallup was on the cover of Time in 1948. That story really exposed a lot of ways in which the polling industry is the same as it ever was and interesting insight into its development. So I got to take a look back and learn a lot about George Gallup in the process. He was a guy who was born in Iowa in 1901, goes to the University of Iowa and, what is important to his start as a pollster, is that he was the editor of the school paper. One of the things that he wanted to know as the editor was why people read the paper. So he grabbed a couple of copies of the paper and roamed around Iowa City asking people, “What do you like in this paper? What do you not like?” He ended up using the material that he got from those surveys as part of his Ph.D thesis.
After he finished school, he got a job working for other papers doing basically the same kind of survey for them, giving them information about what readers wanted, what worked and what didn’t. He ends up transitioning to working for an ad agency and testing ads. And he’s picking up the skill of analyzing opinion and getting a really accurate picture of the way the American public responds to an idea, whether it’s a newspaper article or an advertisement. Then the 1932 election comes around, and he decides to start thinking about whether he could do the same thing for politics.
JD: If I recall from your article, he realized that the difference between getting an opinion on toothpastes and politics were not necessarily that different. And it seems like the 1936 election was really a game changer for him as an individual and ultimately for his organization. What happened in that 1936 election?
LR: At the time, The Literary Digest was a publication that ran a well-regarded, famous poll, the standard for polling. But Gallup realized not just how to ask people questions but how to make up a representative sample. The way The Literary Digest was doing their poll didn’t make sense, because they were getting data about who to contact from car-ownership and telephone-ownership registration information. This is the 1930s, the Great Depression. A lot of people don’t have cars and telephones. The poll is missing a huge swath of the population. So he makes this public prediction that the famous poll is going to be wrong. He didn’t get it right, either. He was a little off on what the results would be, but he was extremely right about how wrong The Literary Digest poll would be. So after 1936, he’s able to say to people he might sell his poll to, “Hey, look, I was totally right about this thing that the trusted source was wrong about and you should come to me when you want to know what the American people think.”
Well, after 1936, he does all kinds of things. The organization over the next decade really explodes, doing politics, also pre-release testing for movies. By the late 1940s, they have more than 1,000 people going out across America doing polling. They’ve developed a new system that according to Time was called “The Quintamensional Plan of Question Design,” which asks not just yes/no (questions) but also how much do you know about the thing? Why are you for it? Why are you against it? They’ve become really a household name—the polls Gallup conducts appear in papers across the country.
JD: So by 1948, he and the Gallup Organization are really widely known. He’s on the cover of Time leading up to the 1948 presidential election. And Time calls him the Babe Ruth of the polling profession. Then what happens?
LR: So he’s almost synonymous with polling. In fact, there was some concern that the polls were getting too good, that they would skew the political process in some way. That’s the environment in which this 1948 cover appears. We have Harry Truman as the incumbent in the presidential race, and the Republican Thomas Dewey. Everybody thought Dewey was going to win, but Truman did, and that famous photograph of Truman holding up a copy of the Chicago Tribune with the banner headline “Dewey defeats Truman” when in fact it had been the other way around. So the polls had been extremely certain about what was going to happen for a number of reasons. The democratic vote was supposed to split. Truman was unpopular, but they got it wrong.
There’s a big post-mortem: How could these polls, including the Gallup Poll, have made such a big mistake? George Gallup sends his pollsters back out to talk to the same people that they interviewed before to try to find where the error came from. He tries to minimize the problem. One of his theories is that the error came from undecided voters, that he and a bunch of other pollsters figured the undecided vote would split. So they didn’t really follow up with them closer to the election, but in fact, they must have gone for Truman.
And his business is in jeopardy from this. A bunch of papers cancel their contracts with Gallup, and there’s talk of people no longer trusting polling because here’s proof it doesn’t work. But obviously Gallup survived as an organization and the polling industry survived, and here we are many years later with lots of polls. So 1948 did not end up being the death of polling.
JD: It’s interesting that so many aspects of polling put in place in the early 20th century are still relevant today. And obviously all these years later in 2016, yet another election where most of the polls got it wrong. So it’ll be interesting to see how things play out with our upcoming election, and we’ll see just how right they are, right?
LR: Yeah, time will tell.