Survey says: General Social Survey tracks public opinion trends over decades

JHU's Stephen Morgan, who has worked on the GSS for 10 years, calls it a 'gold standard' for social science research

Even if you haven't heard of the General Social Survey, you're likely familiar with its findings. They make for some eye-catching headlines.

"Americans are becoming less happy, and there's research to prove it," the Los Angeles Times wrote last month. Perhaps that's because "the share of Americans not having sex has reached a record high," according to The Washington Post. Or maybe it's because the number of people who have no religion has risen 266 percent in three decades, according to The Daily Mail.

In politics, meanwhile, "trusting the media has become a partisan issue," declared The Economist. In fact, "Republicans and Democrats have never been more divided on confidence in the media," reported the Post, analyzing 46 years of trend data.

"[The General Social Survey] was born in an era when it was difficult to get your hands on data ... Now we announce the GSS data release by email and Twitter, downloads begin, and conclusions are available within hours."
Stephen L. Morgan
Bloomberg Distinguished Professor of Sociology and Education

But there's something a majority of Americans agree on: "A record 61 percent of Americans now want weed to be legalized across the U.S., with support rising in every age group," Business Insider reported.

These and a deluge of similar media reports about trends in public opinion are based on data from the 2018 General Social Survey, which was published last month and adds to decades of social science research collected by the survey since 1972.

Stephen L. Morgan, a Bloomberg Distinguished Professor of Sociology and Education at Johns Hopkins University, has worked on the GSS for 10 years and calls it a "gold standard" study. Since 2016, he has served as one of the principal investigators for the project, which is funded primarily by the National Science Foundation.

"The GSS is one of my favorite parts of my job," Morgan says. "It's fun to watch journalists jump on the data after we tweet out its release, and this year was particularly active. Data journalists write headline-worthy stories based on simple trend comparisons, sometimes within hours of the data release. Academic researchers are much slower, but we can go deeper."

The survey, conducted by NORC at the University of Chicago every other year, is based on approximately 2,500 face-to-face interviews with a nationally representative sample of English and Spanish speakers who reside in the United States. Since its debut more than 46 years ago, the GSS has been one of the nation's most influential studies in the social sciences, used to generate results that have been published in thousands of journal articles and books.

Interviews can last more than two hours and include hundreds of questions developed by pollsters and social scientists over the decades. These questions are designed to consistently measure trends over time and also to address emerging topics. Between survey years, principal investigators, including Morgan, conduct a thorough review of the questionnaire and survey methods to ensure that each question gets to the heart of a person's opinion on an issue and that the data collected enable scientists to analyze how opinions evolve.

Some questions match those offered on typical telephone polls—such as political party identification and level of completed education—but others are specific to the study. Since 1984, for example, respondents have been asked whether the Bible should be considered:

  • "the actual word of God and is to be taken literally, word for word"
  • "the inspired word of God but not everything should be taken literally, word for word"
  • "an ancient book of fables, legends, history, and moral precepts recorded by man"

According to Morgan, that question alone is a strong predictor of other attitudes, not just views on religion.

The survey delves deeply into the respondent's origins and identities, asking questions not only about their personal histories but also their spouses and parents. It asks about respondents' daily lives and experiences, including:

  • whether they've had back pain in the past year
  • how often they feel exhausted or "used up" after their workday
  • how they'd rate their general happiness
  • whether they attend church with members of a different race

Other questions—about drug use or sexual behavior, for example—are so sensitive that the respondent is asked to read the question on a laptop and key their response themselves, rather than offer the answer verbally to an interviewer they have just met.

Leading the project is subtle, painstaking work in part because of the survey's legacy of consistent, detailed coding for demographics and opinions.

"It's an authoritative source on a number of social trends, such as attitudes about civil liberties and free expression, gun ownership, and attitudes toward spending by the federal government," Morgan says. "We do not want to ruin our record of providing consistent measurement, even if an old question begins to feel stale."

In order to gauge a respondent's view of free speech, the survey asks whether "an admitted communist" should be allowed to give a speech in the respondent's community and, in a separate question, also be allowed to teach at a college or university. These questions are still part of the survey even though threats from communism and red-baiting are no longer at the forefront of the social consciousness as they were decades ago.

In 2008, just before Morgan began working on the GSS, a parallel set of questions was developed for a "Muslim clergyman who preaches hatred of the United States." Morgan says investigators have determined that "clergyman" is a "needlessly gendered reference," and that they plan to change the wording in the 2020 survey to "an Islamic religious leader," in the hopes that the question will stand the test of time and "isn't too much of a departure from what we have asked in the past six surveys," he says.

Lately the survey's depth of data collection has proved a boon to those analyzing political controversies.

"We're the only survey that I know of that's coded occupation consistently and in a detailed fashion for such a long time," Morgan says. "It makes us able to model what the white working class really is, what their racial attitudes are, and whether they support certain parties, candidates, and policies."

As recent media stories have noted, the data show that contrary to some concerns about racial polarization, opinions about government spending to benefit minority groups have shifted. A record number of GSS respondents say that they don't believe the federal government spends enough on improving conditions for African-Americans.

"This isn't a change that is confined to liberal Democrats," Morgan notes. "Even conservatives moved in the same direction on a lot of these indicators, which is consistent with broad-based recognition that a lot of work remains to be done to promote racial integration in the U.S."

He calls these results "encouraging for our civil society."

Overall, Morgan says, the GSS should be thought of as a piece of social science infrastructure.

"It's a great open-science data source that anyone can use," he says. "It was born in an era when it was difficult to get your hands on data—you had to wait for magnetic tapes to arrive in the mail. Now we announce the GSS data release by email and Twitter, downloads begin, and conclusions are available within hours."

He says he always looks forward to seeing how the data is used and he hopes to see more social scientists—including those here at Hopkins—diving into the data. For his part, he's using the survey in his freshman seminar course, Public Opinion and Democracy. It has a certain nostalgic quality for Morgan.

"I first analyzed GSS data as a college freshman myself in 1990," he says. "My students are not too far behind the journalists, and I expect some great final papers from them at the end of the semester."

Explore Key Trends in the GSS data online.