Tom Smith, PhD’80, became the director of the General Social Survey in 1980. (Photography by Jason Smith)

Growing numbers

For 40 years, the General Social Survey has cultivated a vast body of knowledge about Americans’ personal attitudes and opinions.

Every two years, a couple hundred survey takers working for NORC at the University of Chicago set out across the country, knocking on doors in urban neighborhoods, suburban subdivisions, rural communities, and remote outposts, talking to thousands of Americans about everything from the nation’s politics to their own personal lives. This is the General Social Survey, or GSS, which for 40 years has measured the country’s attitudes and feelings on a vast range of subjects.

The GSS is how we know, for instance, the percentage of Americans who believe in an afterlife, or the legitimacy of the death penalty, or the competence of the executive branch—and how those numbers have changed over time. It’s how we know what their sexual behavior is like and whether they’re happily married and how often they go to church. It tells us how many Americans belong to unions or own guns, how they feel about physician-­assisted suicide, prayer in school, gay marriage, and mothers working outside the home. How many hours they spend watching television or visiting friends, how often they feel rushed or bored. Racial attitudes, job satisfaction, drug use. Whether people feel afraid walking alone at night in their neighborhoods. Whether they think their children’s standard of living will exceed their own.

In important ways, the GSS tells us who we are.

Tom Smith, PhD’80, the survey’s longtime director, is careful to emphasize that the GSS doesn’t actually measure everything. “I take the word ‘comprehensive’ very seriously,” he says. “Nothing covering something as big as American society could possibly be comprehensive.” Still, the GSS is colossal. Only the US Census is cited more often in the academic and popular press. If you’ve ever read a newspaper story or magazine article about a sociological trend or a long-term shift in American opinion, chances are the data originated with the GSS. Smith gets as many as 15 calls a month from reporters, and every year the survey is cited in hundreds of academic studies. As of early 2012, NORC, which administers the GSS, had counted nearly 20,000 scholarly papers and books that had used GSS research since the survey first launched in 1972.

Duke University professor Mark Chaves, who specializes in the sociology of religion, offers a representative sentiment. “The GSS is by far the most important source of information about religion, especially when it comes to tracking trends in American religion,” says Chaves, who chaired the survey’s board of overseers from 2008 to 2011. “There’s lots of surveys out there now, but there’s no other source that tracks as many aspects of Americans’ basic religiosity—beliefs, practices, affiliation.” Without the GSS, he says, his 2011 book, American Religion: Contemporary Trends (Princeton University Press), “would not have been possible.”

The data is open to everyone. That is the survey’s core principle. Before the GSS, Smith says, most surveys were carried out by scholars with access to funding, and centered around narrow research questions. “And he—I say ‘he,’ because they were almost always men—he’d get the data and analyze it for two or three years, and then he’d write the book,” Smith says. “And maybe the data would eventually get into the public domain, and maybe it wouldn’t.” The GSS’s founder, sociologist James A. Davis, wanted to change that system, with an idea he first called the 20 questions project. “It occurred to me that there could be a program which provided sociologists and social scientists everywhere data that they could work on without having to get individual grants,” says Davis, NORC’s director at the time and now a GSS principal investigator emeritus, who has taught at numerous universities including UChicago. “So that was the germ of the idea, sort of a primitive socialism.”

The other half of the survey’s “birth story,” as Smith calls it, began in the mid-1960s with the social indicators movement, a push to study subjective measures of well-being in addition to the objective economic statistics the government collected: unemployment, cost of living, gross national product, and corporate profits and losses. “Those are important economic indicators,” Smith says. But “society’s a lot more than that.” Harvard sociologist Peter Marsden, AM’75, PhD’79, a GSS principal investigator who in 2012 edited ­Social Trends in American Life: Findings from the General ­Social Survey since 1972 (Princeton University Press), remembers the first year’s data scrolling across his computer screen in 1973, when he was a Dartmouth undergrad in Davis’s class. “It looks much more dramatic now than it did then,” he says. But by the 1980s, “you started to get a glimmer of what could happen if this went on for a while.”

It was always conceived as a decades-long project. “There’s very little meaningful change that happens on a year-to-year basis,” Smith explains. Davis puts a finer point on it. “Certainly we hoped it would go on forever,” he says. “If 200 years from now it was continuing, think what you could go back and trace. What if we’d had one like this during the time of the Civil War, or the Depression?”

And so, for 40 years, GSS interviewers have sat down with Americans in their living rooms and at their kitchen tables, asking for their honest feelings and opinions, compiling snapshot after snapshot of American society. Out by O’Hare International Airport, there’s a warehouse that holds original copies of every GSS questionnaire ever filled out (or, at least, every one up to 2002, when NORC began sending interviewers into the field with laptops instead of paper questionnaires). In all, 55,087 surveys are in storage there, millions upon millions of individual bits of data, personal testimonies that add up to what it means and feels like to be an American at a particular moment and place.

Over the decades, a few social trends have taken Smith by surprise, the arcs of American opinion angling out in ways he wouldn’t have predicted. One example is the marked shift in attitudes toward homosexual behavior, which the GSS has monitored since 1973. A suite of connected questions asks respondents to rate the “wrongness” of four types of sex: teenage, premarital, extramarital, and same-gender. “The attitudes toward these behaviors are all correlated,” Smith says. “Someone who is permissive on one tends to be permissive on the others.” So he would have expected them all to move more or less together. But since the early 1990s, disapproval of gay sex has fallen much faster than the others, to less than 50 percent. Most of that, he says, is driven by “cohort turnover”: as older generations dwindle, they’re being replaced by younger people more open to gay rights. That cohort difference doesn’t hold for the other categories of sexual behavior. “Extramarital sex has actually become slightly less accepted over time,” Smith says. Its disapproval now hovers at around 80 percent, and it’s never been lower than 70 percent. For premarital sex, always the least objectionable to respondents, there has been a gradual but constant wearing away of disapproval.

Smith sees similar nuance and complexity in the issue of free-speech rights for social groups held in suspicion: anti-religionists, communists, gays, militarists, and racists. “We ask a series of questions about civil liberties: Should members of these certain groups be allowed to make a public speech? To teach in a college? To have a book they’ve written in the public library?” From the early 1970s to the present, Smith says, the numbers show a basic increase in support for civil liberties—with one exception: “There’s no greater support for the racist.” Less than 60 percent say racists should be allowed the same free speech as other Americans (in 1976, free speech for racists was more popular than for communists or militarists). For the other groups, those numbers have reached 70, 80, or close to 90 percent. “So what you have here is a general social trend pushing support for civil liberties up,” Smith says. “And then there’s a second social trend that is supporting racial equality, that’s also moving up. And that makes the racist less and less socially acceptable over time.” In 2008 the GSS added Muslim extremists to the list of groups. Only 41 percent of respondents said they’d allow them free speech. That number didn’t budge in the 2010 survey.

Perhaps more than any other subject, religion—one of the GSS elements most intensely analyzed by outside researchers—demonstrates the intricacy of societal change. “We document a huge rise in the number of people with no religious affiliation, a drop in church attendance, very little change in the belief in God, and very little change in belief in an afterlife,” Smith says. In fact, belief in life after death has risen a few percentage points, he adds, while belief in God has dropped slightly, but only to about 90 percent. “There’s a very strong theory in sociology that says of secularization, as societies become more modern, as education is increased, science takes over and the nonscientific mythology of religion dies away.” According to GSS numbers, he says, the answer to that theory is both yes and no. The religiousness of society is changing, he says, but in a complex way that can’t be fully understood without examining multiple facets of belief. If you looked at only one, Smith says, “you would substantially misunderstand the religious profile of America.”

Chaves agrees, although he reads the numbers differently. “We now know, I think, that Americans’ religiosity”—that is, traditional religious belief—“has actually been declining.” But the change has been so slow that “if you have only ten years or 20 years of data, you can’t quite see it,” he says. “Now that we have 40 years of data, it’s like building a more powerful telescope.” And alongside that slow slide in traditional worship, Chaves sees what he calls a “diffuse spirituality” accounting for things like a growing belief in the afterlife.

Smith sums up: “What people want is often a very simple story. ‘We’re all becoming more permissive,’ or, ‘We’re all becoming more selfish.’ But society’s more complex than that. Are we becoming more religious? There isn’t a simple answer.”

In the decade after the GSS first launched, other countries began conducting their own, similar surveys. The first was West Germany in 1980, followed by Great Britain and Australia. In 1984 Smith helped organize a collaboration among those four—a series of shared questions on topics like class differences, equality, and the welfare state—whose results would allow researchers to study social trends across national borders. Today the International Social Survey Programme (ISSP) includes some 57 countries.

Each ISSP questionnaire focuses on a single, internationally relevant theme, and those themes repeat once a decade. In 2013 the theme was national identity; in 2014 it will be citizenship. Cross-national surveys offer a useful, and sometimes surprising, look at which ideas or values or habits are shared, and which are singular. The United States has been the runaway leader or tied for first place for national pride each year that the issue has been investigated. Digging deeper, however, the data reveals a more universal trend. “In 30-some countries, when I analyzed the data, national pride was highest among the oldest adults and lowest among the younger adults,” Smith says. The difference, he believes, is related not to a change in feeling as people age, but to generational turnover; young people now are less nationalistic than their elders were when they were young. “So this really is a global pattern, and something that I think represents a historic change, growing cosmopolitanism, the impact of globalization.”

Playing on an old joke (“How’s your wife?” “Compared to what?”), Smith articulates the significance of these comparative studies. “You grow up in a society, and whatever the society is, whether it’s a conservative Muslim society or a fairly secular, prosperous Scandinavian society, you tend to grow up and say, ‘Yeah, that’s what people are like; that’s what society is like,’” he says. “The fact is, well, that’s what your society is, that you’ve grown up in at this point in time. That’s not what human society is.”

Most GSS questions have been a part of the survey’s “replicating core” for decades; there are questions about religion, family, work, and politics that have remained the same, word for word, since the early 1970s. Some questions, about racial integration and gender roles, for instance, go back even further. “There’s a little saying,” Smith says: “‘The way to measure change is not to change the measure.’” Before the GSS launched, grad student Kathleen Schwartzman, PhD’85, now a sociologist at the University of Arizona, spent nearly a year combing through old surveys, culling questions to include. And so from its earliest moment, the GSS offered a perspective on the past; it was already tracking trends.

Sometimes, though, new questions do get added to the survey. Some belong to “topical modules” that often appear only once and bore deeply into a particular theme: multiculturalism, social networks, medical care, work organization. In 2004, Smith developed a module on spiritual transformations, studying people’s experiences with conversion or rebirth or the collapse of faith, and the demographics and triggering events.

Once in a while, new questions are added to the core. It’s rare, though, Smith says, because for every new question, an old one must be taken out to preserve the survey’s length. Plus, writing questions that will hold up year after year, decade after decade, is hard. The terminology gets outdated, or the concepts cease to be relevant. Sometimes it’s hard to make questions narrow and precise enough. In the 1990s, Smith and his staff wanted to get at the issue of comparable worth by looking at pay rates for men and women. But the problem was, so many occupations were largely segregated—construction workers, secretaries, nurses—that it became difficult to isolate gender as a variable. In the end, Smith gave up. “We came up with some questions, we tested them, and the majority of them failed completely.”

When new questions concern technology, writing them is even harder. “The questions we asked in the early 2000s about web and Internet use we’ve had to drop because they’re already too dated in terms of the terminology,” Smith says. “There was one we had, and the only example we could give to convey to people what we meant was ‘Blackberry.’ That was basically the device that defined this emerging market. And now there’s a 50-50 chance that Blackberry won’t even be around by the end of the next few years.”

He’s wary of latching onto the newest term: smartphone. “Maybe ‘smartphone’ is going to last, and maybe it won’t,” he says. “I mean, we could ask something like, ‘Do you have a device which could contact the Internet?’ And that would be pretty stable. The problem is, no one thinks in that kind of terminology. They think, smartphone, Wi-Fi-connected laptop, tablet.” In cases like this, he says, it can take several survey questions to nail down the answer to a single idea. “And you’re still less certain that you’re measuring the same thing over time.”

Social media has Smith completely stumped so far. The GSS doesn’t measure Americans’ social media behavior and attitudes because Smith and his staff haven’t yet found a clear and durable set of terms. For one thing, he says, “social media,” which entered the lexicon only a few years ago, doesn’t mean the same thing from person to person. For another, “if we had jumped on the first major example of social media in the early 2000s, MySpace, and asked questions about that, think where we’d be today. MySpace barely exists and has morphed into something it wasn’t originally.”

The GSS could ask respondents if they have Facebook accounts, but as time goes on, that information may not signify much about the overall trend in social media engagement. “It’s only true if Facebook is maintaining its market share of the social media universe,” Smith says. “Which it’s unlikely to do.”

Besides the vast and expanding list of social media sites—each with its own slightly different concept and purpose—there stands the deepening thicket of online and text messages. Does a Facebook message count the same as an e-mail? What about a direct message on Twitter? How to categorize Snapchats and Instagrams? “And somehow,” Smith says, “you want to know both the total of all of that and the different components of all that. And it’s very difficult, because the platforms are so different, and because they keep changing.” At some point, he would like to add social media questions to the GSS core. But at the moment, “I just don’t think there’s a reliable way to do that. It’s the exception, something that I just don’t think we can measure.”

Administering the GSS is harder than it used to be. More expensive too. Mostly, says Smith, that’s because it takes more work to get every interview. It’s harder to find people at home, even on Saturdays and in the evenings, and when you do find them at home, it’s harder to get them to talk. They’re busy, or they’re skeptical, or they can’t be bothered. “Men are less cooperative,” Smith says. Even when they’re home, they’re more likely to say, “I’m busy.” Or, “The ball game’s on,” or, “I’m not interested.” Often it takes several visits for interviewers—all of them trained professionals for whom the GSS is part-time work—to talk their way inside.

Once they’re in, the survey itself takes about 90 minutes to complete and covers about 300 questions. Most are multiple choice, some yes or no, and a few open-ended. Sometimes the survey takes longer than planned. The phone rings; company comes over; children need to be fed. Smith tells one story about a GSS respondent 20 years ago who took three hours to complete the survey. “He actually gave his answers fairly quickly,” but for almost every quesiton he quoted biblical chapter and verse to support his position, Smith says. “Like, Luke 12:16, or Leviticus, or whatever.” The interviewer diligently wrote down each one.

The GSS used to be conducted annually—with the exception of three missed years, when the money fell short—but in 1994, the National Science Foundation, which underwrites the survey, reduced its funding, forcing NORC to conduct it every other year instead. To keep collecting as much data, the GSS began targeting 3,000 respondents instead of 1,500. Though cheaper than mounting annual surveys, it’s an imperfect solution, Smith says, but one that preserves the survey’s overall robustness and reliability. “If I had my druthers,” he notes, “I’d go back to our original design,” since gaps in the data can limit the ability to do some types of analyses that involve pairing GSS findings with other, more frequently updated sources.

Another innovation: in 2006, the GSS began returning to some respondents for reinterviews in two subsequent GSS surveys—launching a “rolling panel” of people interviewed a total of three times in four years, allowing for the study of not only aggregate but also individual change over time.

The survey’s respondents represent a cross section of the country. Dividing cities and counties into 400 sample sections, measured in blocks, NORC researchers randomly draw addresses and a target respondent, 18 or older, from each household. “And there’s no substitution allowed,” Smith says. If the targeted respondent declines the survey—as sometimes they do—“you can’t go to another household. You can’t do the wife instead of the husband; you can’t do the unemployed son instead of the 60-hour-a-week working single mom. Because then you’re creating bias in the sample. If everyone in the target population—that is, adults living in households in the United States—has an equal probability of selection, then you will get a representative sample.” After that, the interviewers just have to get them to open the door.