Many observers have been struck by how quickly public opinion has shifted on homosexuality in the United States. A quarter-century ago, about 12 percent of Americans agreed that “homosexual couples should have the right to marry one another.” And only a decade ago, Americans opposed gay marriage by healthy 20-25 point margin. Now, most Americans support it. Politically, what was once an easy winning issue for the GOP is increasingly becoming a drag on the party’s candidates.
The pattern of change on the wider question of homosexuality has also been striking. In the mid-1970s, about 70 percent of Americans told pollsters that “sexual relations between two adults of the same sex” were “always wrong.” In the 2010s only 46 percent did. Note this, however: Americans’ views of homosexuality changed little for the first half of those years; indeed the percent who damned gay relations grew a bit. Then, in the 1990s, expressions of tolerance skyrocketed.
We see roughly a similar pattern of change in public opinion about other major issues: In most cases, a clear consensus holds for a long time. When opinions start to change, the change takes up increasing speed toward a much more even division. That is when the topic becomes socially and politically divisive. A majority forms around a new consensus and the pace of change slows again as the most committed supporters of the old view reluctantly come around; some never do. Researchers call this pattern the S-curve or, more properly, the sigmoid. Some readers will recognize this as the standard description for the diffusion of innovations. In this post, I discuss a few examples of the process and the implications it has for understanding social change. (This post draws from Century of Difference, Ch. 9, and here).
The figure below shows the public opinion trends on gay issues. In red is the percentage of American respondents to the GSS (General Social Survey) who answered “not at all wrong” to a question asking whether they thought that “sexual relations between two adults of the same sex” were always, almost always, sometimes, or not at all wrong. The purple diamonds show support for gay marriage. What we see – or I see, at least – is that beginning swing up the sigmoid curve. The GSS does not have data before 1972, but it is hard to image that voiced tolerance of homosexuality was ever much higher than 10 percent. What we cannot tell is where support for gay issues will level off. Given that much of the change entails “generational succession” – the recently born, who are more accepting, replacing the older, who are less, as they die – some upward movement is almost guaranteed. One point to observe is that the heated controversy over gay marriage coincides with the period when the acceptance of homosexuality reached about 30 to 40 percent.
Race and Religion
We are in the midst of rapid change on homosexuality. We can look back historically and see other opinion changes that are more or less complete. The Gallup Poll and later the GSS asked Americans whether they would vote for a qualified candidate whom their own party nominated if he was Negro or black (the term changed over the years). The main graph below shows the trend. Note that Gallup did not bother to ask the question before 1958 (when 37 percent said yes), so unlikely seemed the idea, thus we cannot see the lead-up to 1958. Surveys continued to ask the question even as “yes” answers went into the 90-percent range, thus we can see the topping out of yeses at nearly 100 percent. (By the way, the post-Obama-election percentages are just 2 points higher than the pre-Obama ones.)
The inset figure shows the trend for voting for a Jew. Gallup started asking that question in 1937, when only 46 percent already said that they would vote for a Jew, and stopped asking that question in 1987, when 89 percent said they would. The issue presumably was no longer sufficiently controversial. As with black candidates, we don’t see the low-percentage tail at the start, but unlike the question about blacks, we also don’t see the topping off.
In 1938, Gallup asked Americans, “Do you approve of a married woman earning money in business or industry if she has a husband capable of supporting her?” Only 22 percent said yes and Gallup did not bother to ask again. Starting in 1972, the GSS asked the question and, as shown below, we see much of the sigmoid which describes the large change in how Americans viewed women’s roles. We also see that approval seems to have topped off, not at 100 percent, but at about 80 percent. This may be a ceiling for Americans’ approval of married women working.
Of course, many potential changes in public opinion do not – or have not yet – happened. For example, even though views on premarital sex have shifted to the liberal side since the 1950s in roughly this S-curve way, Americans’ views of extramarital sex have changed little (and, indeed, have become more conservative – see this earlier post).
On some issues, the resistance to change became so fierce that what seemed to be a shift in opinion stopped. From 1960 to the mid-1970s, it appeared that Americans were moving to the “choice” position on abortion. That move stalled and even partly reversed afterward, leaving the public division on the issue at pretty much at the same place for about 30 years now.
And then, we can see cycles in pubic opinion. One concerns capital punishment. The graph below shows the percentage of Americans who told the Gallup Poll that they were against the death penalty for murder. From 1953 to 1966, one would have thought that American opinion was shifting rapidly against the death penalty. Then the trend reversed sharply, probably because of the wave of violent crime that hit the U.S. starting in the mid-1960s. By 1995, only 13 percent were opposed to capital punishment. By then, the violent crime wave was receding and once again opinion moved against the death penalty, reaching 35 percent in late 2013, the same level as in 1960.
Despite such alternate cases, the sigmoid diffusion-of-innovation pattern is common. Among its implications are these: Mass opinion change can come very rapidly, even after decades of consensus, creating the feeling of a cultural earthquake – but it will usually taper off. We typically see the greatest social conflict on a topic near the center of the sigmoid curve – between the 30- or 35-yard lines, so to speak. And, our data to describe the changes will often be incomplete, because survey-takers often don’t bother polling on issues where there is a widespread consensus – within the 20-yard lines, so to speak. Sometimes they, and we, are quite surprised.
 General Social Survey (GSS) data through 2012; Pew Report, 2013; Gallup, 2013.
 GSS, 2010-2012; Gallup, 2012, has 42 percent.
 One indicator of attention to the topic is the frequency of the phrase “gay marriage” in American books, as seen in Google’s ngram, with the take-off starting in the 1990s:
(Cross-posted on the Boston Review BR Blog on January 9, 2014.)