Inside the Alluring Power of Public Opinion Polls From Elections Past | History
Under a pair of bushy eyebrows and a receding salt-and-pepper hairline, George H. Gallup, the father of public opinion, peers out with a neutral expression on the 1948 cover of TIME magazine. Today, the illustration is also seen online in a new context—as the fitting avatar for the Twitter account @HistOpinion. Run by Peter A. Shulman, an associate history professor at Case Western Reserve University, @HistOpinion does exactly what its handle’s name suggests: it tweets out public opinion polls from the past.
Shulman has tweeted nearly 1,500 times from the account. The surveys he curates range from silly to serious. One from a 1997 National Pasta Association survey asked, “Which of the following kinds of pasta best describes your personality?” The responses could be: spaghetti, elbow macaroni, rotini or corkscrew shape, none of the above, don’t know or refuse.
US Jul ’97: Which one of the following kinds of pasta do you think best describes your personality? pic.twitter.com/u8G1PVW0Yq
— Historical Opinion (@HistOpinion) January 29, 2016
“I should probably refuse,” Shulman says when asked which he’d pick. “But I don’t know who refuses pasta. I’ll go with none of the above. I really love fettuccine.”
Another poll, from a 1969 Gallup survey, speaks more directly to today’s political times. It asked, “If your party nominated a woman for President, would you vote for her if she qualified for the job?” Fifty-four percent of responders replied in the affirmative. “I was surprised [it] was as high as it was,” Shulman says. He has posted a few versions of that question, which dates back to at least the 1940s. The responses were all higher than he expected.
US Mar 27 ’69: If your party nominated a woman for President, would you vote for her if she qualified for the job? pic.twitter.com/Qq9wpqKZzw
— Historical Opinion (@HistOpinion) October 15, 2014
Before scientific polling, there was straw polling, usually conducted by news reporters who went out and gathered a large but unrepresentative sample of the population. Straw polls, which many say gets it name from people throwing stalks of straws in the air to see which way they blew, had been around in United States politics since the first presidential poll was published in 1824. The result, which ran in the Harrisburg Pennsylvanian, picked Andrew Jackson over John Quincy Adams. The poll may have called it right that time (Jackson won the popular vote, and Adams won the electoral vote and the presidency), but that was just luck. Like any straw poll, the Pennsylvanian’s methodology was flawed as it only counted respondents that were accessible, leaving out sections of the population, usually poor or working class, who were harder to track down.
By the turn of the 20th century, polling methodology had started evolving. In 1896, W. E. B. Du Bois famously created one of the first empirical social scientific studies when he surveyed 5,000 residents for his groundbreaking work, the Philadelphia Negro. By Gallup’s time, the 1930s, he and other social scientists were starting to argue the merits for surveying a smaller but more representative sample of a population as opposed to gathering a large, homogenous pull.
Gallup along with other early polling pioneers like Elmo Roper and Hadley Cantril were key not just starting up the field of research, but for drawing public attention toward it. (Pollster wasn’t coined until 1949, and it was initially used by in a derogatory way by a critic of the practice.)
Gallup and Roper came from marketing backgrounds and started doing public polls in a bid to increase publicity for their private businesses. Their public opinion polls were first done by people (mostly women) going out with a stack of forms to find a quota of people who seemed working class or middle class or upper class. This, of course, was a flawed methodology. So much, Shulman says, that MIT recently went back and re-weighed the earliest public opinion polls based on the national demographics of the day.
But for the time, their surveys were revolutionary—and popular. Both Gallup and Roper became syndicated columnists, and while they used their fame to help their personal businesses, they also used it for public benefit. Though Gallup was content to reflect where public opinion was in his columns and let the readers draw their own conclusions, Roper actively looked to shape public opinion through polls, commenting on surveys in his columns.
Straw polling was still king when they began conducting their surveys though. The most popular straw poll was published in The Literary Digest, which had been predicting the presidential race for years.
Gallup faced off against the magazine in the 1936 election. The Digest, which had mailed out 10 million ballots to take the temperature of the 1936 election, had predicted Kansas Republican Alf Landon would prevail with 57 percent of the vote. But Gallup, polling a much smaller, but more representative sample size, predicted President Franklin Delano Roosevelt would win his reelection bid. Though Gallup’s margin turned out to be several points off, he had correctly called the winner. That shifted the way presidential polling was done, and scientific polling has continuously evolved since.
Shulman’s first experience digging up an old poll came about when he was looking up a reference to a book by historian David Kennedy something about public opinion related to World War II. He ended up on the JSTOR database, which has a collection of the early polls via the Public Opinion Quarterly. Shulman started flipping through them, reading questions like: How long until you think the war is going to be over? What should be done with the leaders? Should we capture Hitler and Mussolini?
The responses weren’t what he expected. In the post-Pearl Harbor haze, the general public couldn’t have known that the war would end in the summer of 1945. People were predicting the war would end in six months or a year to two years or three years to more than 10 years. “That was really jarring to see the variation in thinking about what was their future,” says Shulman. It wasn’t just their uncertainty, but their points of view that surprised him, like the strong sentiment that existed calling for Germany to be greatly punished following the war, essentially repeating World War I’s mistake.
Shulman read The Averaged American, Sarah E. Igo’s quintessential book on the subject, which goes into the creation of the field of study. He also acquired a copy of the reference book Public Opinion, 1935-1946 by Cantril. But it, and his interest, mostly sat on his desk as he worked through his first book, Coal and Empire: The Birth of Energy Security in Industrial America.
When he finished his manuscript, though, he suddenly found himself needing something to fill the time again. He had been using Twitter a lot, mostly reading other’s posts. Though he had first logged on in 2011, he didn’t start using the medium in earnest until after the night of the 2012 election. Like many that evening, he kept reloading The New York Times’ homepage, impatient for updates. He had his Twitter feed queued up as well. He watched, hooked, as information on the feed came in faster than it was reported on at The Times.
He began to give some thought as to what he could do on the medium himself. He noticed accounts tweeting out historical images and photographs. He wondered if there was some kind of public history he could tweet. That was when he literally just looked at the Cantril book, and thought, maybe polls?
When he opened the Cantril book, as he puts it, “I was just drawn right back into the weirdness of public opinion in the ’30s and ’40s. I just started really opening a page and seeing what’s interesting there that I could fit into 140 characters.”
He became part of the #twitterstorians web—the term which has been around since historian and blogger Katrina Gulliver began compiling a list of historians on Twitter in 2007. Kevin Kruse, a professor of history at Princeton University and active twitterstorian, himself was a reluctant adopter of the platform but has since embraced it. “The strength for academics on Twitter is not that just so many of them are there, but so many active people engaged in public policy and politics and reporting are there,” he says.
As long as history continues to repeat itself, comparing past opinion with contemporary opinion has value. Yet Shulman’s historic pulls find their audience especially because it’s so rare to see such polls contextualized today. This void isn’t because of a lack of data. As Michael Traugott, a Gallup senior scientist who served as George Gallup’s personal research assistant during the 1964 election, points out, extensive data archives and the way polling organizations keep track of information makes historical takes available. “Data resources are there to support that kind of writing,” says Traugott. “It’s just not commonly done.”
At first, Shulman’s selection of polls to run on @HistOpinion was more random. There was a whole period where he tweeted out different questions from a 1971 poll that asked college graduates about their expectations for life, career, social issues of the day. It’s the only time, he says with a chuckle, that he noticed a steady decrease in his number of followers. But he was totally mesmerized by the information coming from the poll because it was given the exact year his parents graduated college.
Going through old polls has its challenges. Because scientific polling data began in 1935, there’s a limited scope of historic data available. (“I would love to know public opinion on the Spanish-American War—’Should American troops be doing the equivalent waterboarding in the Philippines?’” Shulman says.) Most historic polls also skew to white interests, and in the case of the Jim Crow South, because African Americans couldn’t vote, Gallup simply didn’t poll them, excluding their opinions all together.
Shulman used to tweet out three polls a day from the account, but he’s since scaled back. When he does tweet, he often pegs polls to the news of the day. Now, with the election in full swing, he says one prescient poll he’s come across came from August 1942. It asked, “If the question of national prohibition should come up again, would you vote wet or dry?”
US Aug 25 ’42: If the question of national prohibition should come up again, would you vote wet or dry?
— Historical Opinion (@HistOpinion) December 6, 2013
The question itself was a moot point. The 21st Amendment had officially repealed federal Prohibition almost 10 years earlier. What Shulman found interesting about the poll’s result was that it showed a surprisingly large minority—38 percent—said they’d still vote for a Prohibition amendment.
“That goes against what we usually think, that Americans didn’t want Prohibition, it was a huge mistake and they got rid of it,” says Shulman. “Maybe the majority of the country did, but a substantial minority really had a different vision of the direction that the country should be taking in the 1930s.”
It reveals how a significant percent of the country might have a very different view of the nation’s status that differs from where the country should be heading from the perspective recorded by history books or newspapers. In a way, it helps explains Donald Trump’s rise this election season. “It’s easy to forget that you can have sizable minorities who share a view and can coalesce around a particular candidate and party that might be unexpected,” Shulman says.
Shulman’s account is best known for a series of tweets he did last year, which touched on American attitudes toward Jewish refugees during World War II. One of them, which he has pinned to the top of his account, is a pull from Gallup’s American Institute of Public Opinion in January 1939. It asked, “Should the US government permit 10,000 mostly Jewish refugee children to come in from Germany?”
US Jan 20 ’39: Should the US government permit 10,000 mostly Jewish refugee children to come in from Germany? pic.twitter.com/5cFs5RabQn
— Historical Opinion (@HistOpinion) November 17, 2015
The response from the American public was, overwhelmingly, no. Only 30 percent of respondents were in favor of admitting the child refugees, just two months after Kristallnacht.
The tweet—on the nose for an American public that continues to push back against providing asylum to fleeing Syrian refugees today—went viral. Politico ran a piece contextualizing the poll, writing: “Yes, It’s Fair to Compare the Plight of the Syrians to the Plight of the Jews. Here’s Why.” The New York Times even weighed in, and Shulman himself made a case for its relevance in Fortune, writing, “Unquestionably, the two situations have their differences. Yet perhaps the biggest difference is simply that most Jews seeking safety from Nazis could not escape, while today, it is not too late to help those most desperate for security.”
The impact of the tweet, and the conversation Shulman continues on in his account today. One of his latest tweets from a Gallup poll in 1945, asks, “Should we permit more persons from Europe to come to this country each year than we did before the war, should we keep the number about the same, or should we reduce the number?”
US Jan 14 ’45: Should we permit more, fewer, or the same number of persons from Europe to come here each year than we did before the war? pic.twitter.com/pzutY9TEuC
— Historical Opinion (@HistOpinion) September 20, 2016
The results, with only five percent calling for more, should seem less surprising now.