Sexual Morality & Social Values

Homosexuality: the largest attitude shift in GSS history

The topline series on whether homosexual relations are "always wrong" is the central exhibit of this chapter — and one of the central exhibits of any contemporary book on American attitude change.

Two features of Figure 1 are worth slowing down on.

The first is the direction of the early-1990s movement. The standard narrative on this question — "Americans steadily became more accepting of homosexuality over fifty years" — turns out to understate what happened. The series actually rose through the 1980s, peaking in 1991 at 78%. That peak coincides with the late peak of the HIV/AIDS public-health crisis and the cultural moment when American moral conservatism on this topic was at its most assertive. The post-1991 fall is what dominates the chart visually, but the period the fall is relative to is not a placid 1970s baseline — it is a sharp early-1990s spike. The liberalization is real, large, and post-1991.

The second is the 2018 → 2021 dip. The 2021 wave produced the lowest reading in the series at 27%, partly attributable to the GSS's shift to web interviewing (sexual-morality items are sensitive to the mode of administration), and partly reflecting actual continued attitude movement. The 2022 and 2024 readings have bounced back somewhat, settling at 33%. Whether the post-2021 stabilization holds, or whether the long downward trend resumes, is an open question on present evidence.

The partisan structure underneath this collapse is the chapter's central finding.

The widening of the partisan gap on this question is not what most readers expect from "polarization" framings. Both parties moved. Republicans, in 1973, said homosex was always wrong at 76%; in 2024, 51%. That is a 25-point Republican liberalization — a real change, on a question conservatives have historically owned. The asymmetry is that the Democratic line moved 54 points over the same window. Two parties liberalizing at different speeds produce a partisan gap that grows even as both ends of the gap recede from the original consensus position.

The age structure: replacement plus within-cohort change

The age view of homosex is one of the cleanest cohort-replacement plots in the GSS.

Three observations on the age structure.

First, the 1973 65+ cohort (people born 1880s through 1908) said homosex was always wrong at 93%. That is a near-unanimous answer for an entire age group on any moral question in the GSS. The 2024 65+ cohort (people born 1955–1959) — the same age bracket, fifty years later — says it at 43%. The cohort that occupies the 65+ slot has been almost entirely replaced, and the replacement cohort is the one that came of age during the gay-rights movement of the 1970s and 1980s. The 50-point drop in the 65+ cell is the clearest example of generational replacement in this book.

Second, within-cohort movement also matters. The 1973 18–34 cohort (born 1939–1955) is roughly the 2024 65+ cohort — same people, fifty years older. They started at 62% in 1973 and they're at 43% in 2024 — a 19-point liberalization across their own lives. So the 30-point drop on the 65+ line is not purely replacement: it is some replacement plus some within-cohort attitude change. People do change their minds on this question.

Third, the 2024 18–34 cohort sits at 22% — the lowest reading on any age band in any year of the series. If the cohort-replacement projection from §3.4 applies here, contemporary 18-to-34-year-olds will age into the 65+ cohort over the next thirty years and produce a 65+ reading in the low 20s by 2055. The question's long downward trajectory has not finished.

Church attendance: still the dominant axis

The proposal that became this book described sexual morality as a domain where "religion matters more than party." The GSS data confirms that — though with one important amendment.

The amendment is that the attendance axis is not just more important than the party axis — it has grown more important over fifty years. The 1973 attendance gap (25 points) was already larger than the partisan gap (3 points). The 2024 attendance gap (49 points) is larger than the 2024 partisan gap (33 points) by a factor of nearly 50%. Religious attendance and partisan affiliation are correlated, but on sexual-morality questions specifically, attendance is the master variable. Weekly church attenders of either party are more conservative on this question than secular voters of either party.

This pattern recurs across the chapter's other items, sometimes more sharply.

Premarital sex: where attendance dominates almost completely

The premarital-sex question — premarsx_rec, "what about a man and a woman having sex relations before marriage" — produces a topline collapse from 37% saying always wrong in 1972 to 17% in 2024. The partisan gap on this item is real but smaller than on homosex, and the attendance gap is enormous.

Now look at the attendance breakdown for the same item — and the magnitude of what attendance explains:

The party-only and attendance-only views describe different populations. A Republican who never attends church is, on this item, much closer to a Democrat who never attends church than either is to a weekly-attending member of either coalition. Premarital sex is where the proposal's "religion matters more than party" reading is most clearly true.

Extramarital sex: stable consensus, growing partisan gap

The extramarital sex item — xmarsex_rec, infidelity — produces a topline that has been remarkably stable across the entire fifty-year history of the GSS. About three-quarters of Americans, in nearly every wave, say cheating is always wrong.

Stable topline, but the partisan structure has changed in a way that is worth flagging.

The extramarital-sex chart is one of the more unexpected findings in this chapter. The topline gives almost no signal that anything is happening — the country has been at roughly 70-80% saying always wrong for fifty years. Underneath the topline, the parties have separated: Republicans have become more strict on infidelity (75% → 84%, a 9-point intensification), while Democrats have become less strict (67% → 56%, an 11-point softening). The 2024 partisan gap of 29 points is one of the larger ones in this chapter, on an item whose topline shows essentially no movement.

The explanation that the data alone cannot test, but that is consistent with the pattern, is that the Republican coalition has consolidated around a moral-conservative position on personal-sexual-conduct questions even as the party's overall direction on social issues has been described as drifting. The Democratic coalition's slow softening on this question is consistent with a separation between sexual-morality framings and political identity for Democratic voters.

Teen sex: the unexpected liberalization

The last item in this chapter is the strangest. teensex_rec asks about a 14-to-16-year-old having sex with another teenager — a question one would expect to produce stable, near-universal bipartisan opposition. It doesn't.

The teensex chart is a genuine puzzle. The proposal that became this book described it as "bipartisan consensus protecting minors" at 65-70% saying always wrong. That description was approximately right when the proposal was written but is no longer accurate. Since 2000 the question has moved 25 points downward, and as of 2024 fewer than half of Americans say teen sex is always wrong.

Some of this is almost certainly a response framing effect rather than a moral attitude change. The question wording ("a man and a woman before they are married") and the underlying interpretation ("two 15-year-olds in a high-school relationship" versus "an adult man and a 15-year-old girl") may have shifted in the public's reading of the question over time. Some of it may be a generational effect: respondents whose own teenage years involved more permissive cultural norms may answer this question differently than respondents whose teenage years were spent under stricter ones. The data cannot adjudicate. The data say only that this question, expected to be the chapter's stable-consensus exhibit, has moved as much as any other item in the chapter — in the same downward direction.

What this chapter changes

Three patterns recur across the four items.

First, the country has liberalized on every sexual-morality question the GSS tracks. The homosex decline (45 points from peak) is the largest. The premarsx decline (20 points) is the second. Teen sex has liberalized 19 points since the 2000 peak. Even extramarital sex, the most-stable topline in the chapter, has moved enough on the partisan version to register as polarized. The aggregate American attitude on sex-and-morality questions in 2024 is substantially less restrictive than it was in 1973.

Second, the partisan gap has widened on every item — but for different reasons. On homosex, both coalitions liberalized and Democrats liberalized faster. On premarital sex, both coalitions liberalized at roughly the same speed and the partisan gap stayed roughly stable. On extramarital sex, the parties moved in opposite directions — Republicans intensified opposition while Democrats softened. On teen sex, both coalitions softened in roughly the same direction. The mechanism behind each widening is different, even though the direction (widening) is the same.

Third, church attendance dominates partisan identification on these questions. On homosex, the 2024 attendance gap (49 points) is 50% larger than the 2024 partisan gap (33 points). On premarital sex, the attendance gap (39 points) is roughly twice the partisan gap (20 points). Religious attendance is the master variable on sexual-morality questions in a way that party identification is not. A book that frames cultural-political polarization purely in terms of Democrats and Republicans misses what the GSS has been showing for half a century: the deepest divide on these questions is between regular religious practice and its absence.

The next chapter turns to a different set of cultural questions — race and explanations for inequality — where the partisan structure is similar in direction but very different in shape, and where the demographic-axis-that-dominates is education rather than religion.