The Big Shift: Changing Views on Marriage for Gay Couples

Support for allowing gay and lesbian couples to marry has risen dramatically in the past decade. One often-cited reason for the change is that younger voters with more accepting views are replacing older voters in the population. The more important reason, though, is that Americans in every demographic, political, and religious group across the country are changing their minds on this issue. Although marriage advocates have not yet been able to turn that edge into victory at the ballot box, the numbers indicate that this year could be different.

Using data from 98 national surveys conducted between 2004 and 2011 with more than 128,000 responses, we dug underneath the topline numbers to gain insights about how different groups are evolving on this issue. Specifically, we find that:

1. Support for marriage has risen 16 points since 2004, with major shifts across every demographic group.
2. 75% of the growth has come from people changing their minds.
3. In at least 13 states, marriage support has surpassed the majority mark, including in two that will see votes this November.

Demographic Shifts

Support for marriage has risen 16 points since 2004, and the shift is occurring across every demographic group, with even those groups most resistant to marriage warming at least 8 points.

The number of Americans who favor allowing gay and lesbian couples to marry is rising dramatically, increasing 16 points between 2004 and 2011. Since 2004 was a low point both in public support for marriage and in the political atmosphere around the issue, we used that year as a starting point. We found that acceptance of marriage for gay couples rose at a rate of over 2 points a year through 2011, and recent polls indicate that support appears to have risen another 2 points since 2011.*

Given the data available, we focused our analysis on the change between 2004 and 2011, both overall and for each group. Our best guess is that, since 2011, each group has continued to increase at the same rate they had been moving, meaning the absolute numbers are higher than those reported here.

Although different demographic groups have warmed at different speeds, every single group has moved in a positive direction since 2004. Even among evangelical Protestants, one of the groups most likely to oppose marriage for gay and lesbian couples, support grew by 8 points from 2004 to 2011. This works out to more than one point of positive movement per year—a striking pace of progress for any issue. Moderates moved more quickly than any other group we sampled, gaining 21 points in support for marriage between 2004 and 2011. In fact, they have moved nearly 10 points just since the last Presidential election. And no group’s views have grown more negative on the issue over the seven-year time frame we studied.

Compared to the national average, many of the groups who moved more quickly are fairly predictable. Support for marriage among college graduates rose 17 points since 2004, and the same was true among white Americans, those who do not attend church weekly, and younger people who were born in 1980 or later. Democrats, mainline Protestants, and those living in the Northeast gained nearly 19 points of support for marriage during the seven-year time span we examined, as did those with some level of college education. But other groups with higher than average increases in support for marriage defied common wisdom: Catholics, for example, gained more than 17 points from 2004 to 2011, while support among Asian Americans and people living in the South grew 18 points.

Political Party

In 2011, Democrats were 32 points more likely than Republicans to favor marriage for gay and lesbian couples, as well as nearly 19 points more likely to favor marriage than Democrats had been just seven years earlier. Between 2004 and 2011, support for marriage also grew by 15 points among Independents and by nearly 11 points among Republicans. In 2004, only 16% of Republicans favored allowing gay and lesbian couples to marry. By 2011, that support had risen to 26%, with a large portion of the gain occurring in the years since 2008.* Just over one-quarter of self-described Republicans counted themselves as marriage supporters in 2011, and that fraction is clearly growing rapidly. In fact, more Republicans now support marriage for gay couples than call themselves “pro-choice.”**

All data for 2004, 2008, and 2011 is available in Appendix C.

* According to Gallup, 22% of Republicans consider themselves pro-choice. Poll, Gallup, May 3-6, 2012, Accessed September 28, 2012. Available at: http://www.gallup.com/poll/154838/Pro-Choice-Americans-Record-Low.aspx.

Ideology

Not surprisingly, liberals support marriage at higher rates than people of other ideological persuasions. In 2004, liberals, Jewish voters, and those who said they ascribe to no religion were the only groups in which a majority favored marriage. At that time, liberals were 23 points more likely than moderates and 43 points more likely than conservatives to favor allowing gay and lesbian couples to marry. Support among liberals grew by 13 points between 2004 and 2008 but has been stable at about 7 in 10 since the last Presidential election.

Support for marriage among self-described moderates, however, has grown steadily and quickly since 2004. In 2004, only 33% of moderates favored marriage. By 2011, 54% held that position, an astounding 21-point gain in just 7 years. This is the fastest rate of increase of any demographic group we examined. The rapid transformation of views among moderates illustrates just how quickly the politics are shifting on the issue of marriage, and how different the dynamics will be in 2012 compared to 2004 and even 2008.

Today, nearly 70% of liberals and a solid majority of moderates favor allowing gay and lesbian couples to marry. Support remains a minority position among conservatives, only one-quarter of whom favor marriage for gay couples. However, that proportion is growing quickly. The number of self-described conservatives who favor allowing gay couples to marry doubled in the seven years from 2004 to 2011, and it rose an above-average 9 points just since the last Presidential election (3 points a year). Among conservatives born in 1980 or later, 45% favored marriage for gay couples in 2011, basically the same rate as the general population.

Religion

Most people who describe themselves as nonreligious supported marriage for gay couples throughout the period we studied, boasting some of the highest levels of support among any group in 2004 (58%). Support among Jewish voters outstripped even that number, clocking in at 65% in 2004 and 74% by 2011. Catholics and mainline Protestants started lower but have seen exponential gains, and majorities of both groups now favor marriage. In the seven years we studied, support among Catholics rose from 35% to 52%—a gain of 17 points or 2.5 points a year. For mainline Protestants, the change was even bigger: 19 points from 2004 to 2011. By last year, both groups boasted majority support for marriage. Half of Catholics and Protestants born in the 1950s and 1960s, and sizeable majorities of those born since, favored allowing gay couples to marry.

Evangelical Protestants had the lowest level of marriage support of any group we studied. In 2004, only 12% of them supported allowing gay and lesbian couples to marry. However, even in this more resistant group, there have been significant changes since that low water mark. Support had nearly doubled by 2011 to 20%, and the pace of change sped up between 2008 and 2011. In addition, young Evangelicals (those born in 1980 and afterwards) are twice as likely to support allowing gay couples to marry as those born in the 1950s or earlier.

Religious Attendance

Religiosity is as important as religious identification in explaining support for marriage. Those who say they attend church weekly are significantly less likely to support marriage for gay couples than those who are less observant. Among those who do not go to church weekly, support for marriage neared 40% in 2004 and gained 17 points before 2011, culminating at 57%. Weekly churchgoers were a little less than half as likely to favor marriage in each year, but their support has also risen significantly, gaining 12 points since 2004, with nearly two-thirds of that shift happening just between 2008 and 2011. However, even among younger weekly churchgoers (those born in 1980 and afterwards), support for marriage was only at 39% in 2011—7 points below the national average.

Education

Support for marriage is higher among more-educated Americans. Folks with graduate degrees have led the way, with 59% supporting marriage by 2011. But majorities of those with a college degree or some college now support marriage, and their support has grown at faster-than-average rates since 2004—by 17 points and 19 points, respectively.

High school graduates also changed their attitudes over this time period, with an additional 15 points supporting marriage by 2011. Younger high school graduates (those born since 1980) favor marriage at above-average levels, as do those with some college who were born since 1960. Among college graduates and Americans with a graduate degree, only those born before 1940 support marriage at below-average rates.

Region

Folks in the Northeast and on the Pacific Coast support marriage for gay couples at higher rates than those in the rest of the country, but support is rising at very similar rates in all regions. In 2011, majorities of Americans living in the Northeast, on the Pacific Coast, and in the Midwest supported marriage. Less than 30% of those in the South and the Mountain West favored marriage in 2004, but support rose by 18 and 19 points, respectively, in the two regions by 2011, and even in those regions, support is approaching 50%.

Race

By 2011, about half of Asian American, white, and Latino voters favored allowing gay and lesbian couples to marry. Support was rising fastest among Asians and whites. Only about one-third of African Americans supported marriage by 2011, but their support has risen nearly as rapidly as other groups’ and appears to have risen faster between 2008 and 2011, even before President Obama endorsed marriage for gay couples. In addition, the 2011 data shows that younger African Americans and Latinos support marriage at above-average rates (47.8% and 57% respectively for those born in 1980 or after).

Gender

Women have supported marriage more than men in every year, and the gender gap has increased over time. In 2004, nearly a third of American women favored allowing gay and lesbian couples to marry, while men lagged about 5 points behind. By 2011, women had crossed the majority mark with 51% support, gaining 18 percentage points over the seven year period. Support among men increased as well, albeit more slowly, creating a 10-point gender gap on the issue by 2011.

Generational Shifts

Only one-quarter of the increase in support is due to younger voters replacing older voters in the population; 75% of the growth has come from people changing their minds.

Younger people are much more likely than older people to favor allowing gay and lesbian couples to marry. In 2011, for instance, support was 10 points higher among Americans born in the 1940s than among those born earlier and 19 points higher for those born in the 1980s than for those born in the 1950s. In fact, our data shows that the probability of supporting marriage goes up 0.8 percentage point by each year of birth (except for a small plateau between 1953 and 1961)—meaning a person born in 1981 is 16 points more likely to favor marriage than someone born in 1961. The general pattern of younger people being more likely than their elders to support marriage holds in every demographic group we examined.

In theory, generational replacement could explain much of the national warming. As people born in the 1980s and 1990s replace those born in the 1920s and 1930s in the voting age population, support for marriage could increase in the electorate even without anyone changing his or her mind.

In fact, however, most of the increase in support is due to Americans changing their minds in favor of marriage for gay couples. Since 2004, support has risen in every single birth cohort. Support rose 12 points among those born in the 1960s, for instance, even though basically the same people made up that group in 2004 and 2011.

On average, the increase has been about 12 points for those born in each decade. Those born in the 1970s are the only group changing their minds substantially slower than that average. Regression analysis shows that Americans were 12 points more likely to support marriage in 2011 than in 2004 among people born in the same year. Although immigration or death rates might account for a tiny part of that increase, the far more likely explanation is genuine changes in attitudes.

Between 2004 and 2011, support for marriage across the country went up 16 points: from 30% to 46%. But among people born within the same year, support rose 12 points during the same time period. This means that a quarter of the total change was due to replacement of older voters by a younger generation, but a full 75% of the shift was due to Americans changing their mind on the issue of marriage.

State Shifts

In at least 13 states, marriage support has surpassed the majority mark, including in two that will see votes this year, and the other two with 2012 votes are on the cusp of a majority.

Marriage opponents often point out that marriage bans have always won at the ballot box in order to combat the perception that they are losing ground.* However, a look at the timing and location of those ballot initiative votes quickly explains why that pattern has prevailed thus far—and why this year will almost certainly mark the end of marriage’s losing streak.

This record is only true of votes that banned marriage for gay couples without limiting other forms of legal recognition. An Arizona ballot initiative that attempted to ban both marriage and civil unions failed in 2006, but a pure marriage ban was then passed in 2008.

Of the 32 statewide votes where marriage lost, 27 took place in or before 2006, which was an eon ago on this issue. In 2006, nationwide support for marriage was at about 32%, and Massachusetts was the only state where a majority may have supported marriage (we estimate 49% favored marriage in the state that year, and Rhode Island was the next closest with 46%).* And in the case of the five more recent ballot votes, we estimate that support for marriage was still below majority levels: 37% in Florida in 2008, 41% in Arizona in 2008, 47% in California in 2008, 46% in Maine in 2009, and 38% in North Carolina in 2012.

State-by-state data is available in Appendix B.

But by 2011, the most recent year for which we have data, 13 states plus the District of Columbia have estimated marriage support of 52% or higher. Five of those states banned marriage through statewide vote back when majorities opposed it: California, Alaska, Colorado, Maine, and Oregon. Two of those states—Washington and Maine—will have ballot initiative votes this year, marking the first time there has been a marriage vote in a state where the estimated support is above 50%. In Washington, 2011 support was estimated at 55%, and in Maine the number was 52%, but support is probably about 2 points higher in 2012 given the pace of warming on this issue. The two additional states with ballot initiative votes on marriage in 2012 both sit on the tipping point: Maryland with 48% estimated support and Minnesota with 47% in 2011. If those numbers have risen 2 to 3 points in the past year as predicted by national and state trends, they may be over the majority mark as well.

Five other states also lie on the cusp of majority support: Nevada (50% estimated support in 2011), Hawaii (50%), Wisconsin (49%), Arizona (49%), and Illinois (48%). When you add those states to Minnesota, Washington, and the 13 states with solid majorities, that means there are now 20 states where the data indicates that marriage could very possibly win at the ballot box. By contrast, in 23 of the states where marriage bans have already been passed by voters, support is still significantly below the majority threshold, and voters would probably reaffirm the bans if they were on the ballot this year. But based on the data and the line-up for state votes in 2012, it is likely this year will be the turning point when voters accept marriage at the ballot for the first time in a statewide ballot initiative.

Conclusion

Americans’ views on marriage for gay and lesbian couples are changing rapidly. Support for allowing gay couples to marry has risen markedly since 2004 in every demographic group we examined. On average, the rise has been 16 percentage points in seven years, but support is rising even among those who are most likely to oppose marriage (e.g., support among evangelical Protestants has risen 8 points since 2004). Younger Americans in every demographic group are much more likely than their elders to favor marriage for gay couples, and support will continue to rise simply from younger, more accepting voters replacing older ones in the population. The bulk of the increase in support, however, has come from people within each age group in each demographic category reconsidering the issue and changing their minds. This evolution in public opinion is likely to manifest itself in a groundbreaking way this year: by making 2012 the first year that a state passes marriage by popular vote. Given the pace of change on this issue, that milestone is likely to be simply the first in a long line of victories for marriage at the ballot box in the future.

Appendix A

Methodology

We run most of our analyses on data for 128,032 respondents to 98 national surveys conducted by 12 polling firms between January 2004 and December 2011. We obtained most data sets from the Roper Center for Public Opinion Research. The surveys used 21 different question wordings. Our dependent variable is coded 1 for those who favored allowing gay and lesbian couples to marry and 0 for those who gave any other response (including “Don’t Know,” as well as “Civil Unions” for those asked whether they favored marriage, civil unions, or no legal recognition). This is a conservative measure of support for marriage, as those who are undecided are counted as not supporting marriage rather than being dropped. To estimate changes over time, we ran logit models with dummy variables representing each survey year.

Question wording matters, and some questions lead to higher estimates of support for marriage than others. To estimate support as if all respondents had been asked the same question, we added 20 dummy variables for question asked to the logit model, using the most commonly asked question (“Do you think marriages between same-sex couples should or should not be recognized by the law as valid, with the same rights as traditional marriages?”) as the reference group. This technique makes the simplifying assumption that differences in question wording have the same impact on the probability of answering “yes” across time and groups. Using Stata, we first ran our logit model on the full sample, then used the predict command to estimate the probability of answering that “marriages between same-sex couples should … be recognized by the law as valid, with the same rights as traditional marriages” in each year. We then repeated the logit models on different sub-groups (e.g., Democrats, evangelical Protestants, those born between 1940 and 1949) and estimated the probability of supporting marriage for gay couples for each group for each year. Some surveys did not ask some demographic questions, so not all surveys could be included in all analyses.

To estimate support in the states, we added data for 25,111 respondents to 21 additional surveys conducted from 1992 through 2003. We also eliminated all the respondents where the survey did not reveal the state in which the respondent lived. This lowered the total sample slightly, to 126,918. We added 50 dummy variables for the states and 9 extra dummy variables for survey year. Because the number of surveys for the first dozen years was fairly limited, we assumed the same time trends for all states during that period, but we added 51 linear time variables, one for each state, for the period since 2004, allowing support to grow faster in some states than others. We estimated support in each state in each year and report those estimates for 2011 and for years in which a state held a popular vote on a ban on marriage for gay couples.

Appendix B

Marriage Bans and Support

Appendix C

Polling Data