Legalizing Marriage for Same-Sex Couples Did Not Harm Family Formation in the United States

Illustration by Good Studio/Adobe Stock

Two decades ago, the United States was divided by heated debates over the merits of legalizing marriage for same-sex couples. Those who were in favor argued that granting same-sex couples access to marriage would strengthen commitment for same-sex couples, extend the financial benefits of marriage to same-sex households, and improve outcomes for children being raised by same-sex parents. Those who were opposed argued that granting legal status to marriages between same-sex partners would alter the foundation of marriage and thereby diminish its value for different-sex couples, ultimately harming children by making them less likely to be raised in stable, two-parent families. Twenty years have passed since the first marriage licenses were issued to same-sex couples in Massachusetts in May 2004, making it possible to evaluate the evidence supporting or refuting these arguments.

Against this backdrop, RAND researchers conducted a review of evidence and new analyses to identify the effects of extending legal status and legal marriage to same-sex couples in the United States. This research brief highlights findings on the effects of granting that legal status on the broader population: specifically, individuals who are not in formalized same-sex couples. A companion research brief highlights findings on effects specific to lesbian, gay, bisexual, and transgender individuals.

Allowing same-sex couples to marry led to...

• an approximately 2-percent increase in marriage among different-sex couples
• a 10-percent increase in marriage overall (all couples)
• a 0- to 10-percent decrease in cohabitation among different-sex couples
• no consistent change in divorce among different-sex couples

Identifying and Analyzing Evidence

Opponents of legalizing marriage for same-sex couples contended that treating the marriages of same-sex and different-sex couples equally would have negative implications for different-sex marriages and households. Central to this perspective is the observation that marriage rates in the United States have been decreasing for several decades while alternative family forms (e.g., living alone, cohabitation, single-parent households) have been increasing. Opponents contended that granting legal status to the marriages of same-sex couples would further accelerate the "deinstitutionalization of marriage" by lowering the perceived value of marriage for different-sex couples. Such arguments suggest that legalizing marriage for same-sex couples should lead to fewer different-sex couples seeking marriage, as well as possible increases in divorce.

To evaluate how legalizing marriage for same-sex couples affected these outcomes, the RAND researchers undertook two approaches. The first approach was to search four independent research databases (Web of Science, Scopus, Social Sciences Abstracts, and Policy File Index) for studies that analyzed data to determine the effects of granting legal status to same-sex couples, via civil unions, domestic partnerships, or marriages, in the United States. From their original search and from identifying additional articles from research reviews and citations, the researchers identified 96 articles that met their criteria.

The studies described in these articles found that legalizing marriage for same-sex couples had neutral or positive effects on the general population, including positive effects on public opinion, public health, and economic outcomes. Yet the research addressing specific predictions about how legalization would affect trends in family formation and attitudes toward marriage was limited in scope and had not taken advantage of recent methodological advances. To address the limitations of those studies, the RAND researchers' second research approach was to analyze data from five large data sources—the American Community Survey (ACS), the Current Population Survey (CPS), National Vital Statistics Data from the Centers for Disease Control and Prevention (CDC), the Panel Study of Income Dynamics (PSID), and the Monitoring the Future Main Study (MTF)—to better assess whether recent changes in household formation rates and attitudes toward marriage might be attributable to changes in same-sex couples' ability to legally marry.

No Evidence of a Retreat from Marriage—If Anything, an Increase

In their literature review, the researchers found seven studies that examined the implications of legalizing marriage for same-sex couples on marriage rates among different-sex couples. None of these studies found any evidence of negative effects on marriage rates for different-sex couples. Similarly, three studies examined whether explicitly banning marriage for same-sex couples protected different-sex couples but found no evidence of that effect either.

These prior studies, while suggestive, do have several limitations. First, some of the methods used by these studies have become outdated, and more-rigorous methods are available for analyses of current data. Second, findings based on states that were early to grant legal status to the marriages of same-sex couples may not be applicable to states that granted legal status later. Third, previous research was not able to distinguish between no effect and the possibility of multiple effects that obscure each other, such as an increase in marriage among same-sex couples and a retreat from marriage among different-sex couples.

To address these limitations, the RAND researchers turned to ACS, CPS, PSID, and CDC data, applying multiple estimation techniques to these data to identify how marriage rates may have changed following legal recognition of marriage for same-sex couples. The researchers considered changes in both the share of the adult population that is married (marriage stock) and the rate of new marriages (marriage flow). With the ACS and CPS data, they were able to limit their analyses to different-sex couples; analyses using the CDC and PSID data included all couples. The researchers used the latest advances in difference-in-differences methods to estimate the effects of enacting state-level policies legalizing marriage for same-sex couples on total marriage rates. They specified analyses by two levels of analysis (state and individual), two estimation methods (two-way fixed-effects and an interaction-weighted estimator), and three controls specifications (no controls, controls for state-specific economic and marriage-market conditions,^[1] and controls plus a cohort-specific linear time trend). Their analyses compared changes in family formation in states that changed their marriage policies between 2000 and 2014 with changes in states that did not change their policies until 2015, the year of the Supreme Court's Obergefell v. Hodges decision, which required all states to issue marriage licenses to same-sex couples. Figure 1 shows the results of these analyses across data sources.

A detailed description of the figure is in the text directly below the figure note.

SOURCE: Features data from the ACS, CDC, CPS, and PSID.

NOTE: Stat. sig. refers to statistical significance at the ɑ < 0.05 level. The analytic period is 2000 to 2014. The figure summarizes findings from difference-in-differences regression models of each outcome. Bars indicate the proportion of the total number of post-legalization coefficients from all models in each significance category. Bars for the PSID and CDC data have half the number of coefficients summarized because PSID analysis is conducted only at the individual level and CDC analysis is conducted only at the state level. For details, see Benjamin R. Karney, Melanie A. Zaber, Molly G. Smith, Samuel J. Mann, Marwa AlFakhri, Jessie Coe, Jamie L. Ryan, Catria Gadwah-Meaden, Christy Mallory, Brad Sears, and Chandra Garber, 20 Years of Legal Marriage for Same-Sex Couples in the United States: Evidence Review and New Analyses, RAND Corporation, RR-A2912-1, 2024.

^aAll marriages are included, regardless of partner sex.

The two panels in this figure show the distribution of estimates for the effect of legalizing marriage for same-sex couples on two measures of marriage: the share of the population that is currently married and the rate of new marriages. Each row represents dozens of effect estimates (e.g., one year post-legalization, two years post-legalization, and so forth; analysis at the state level, analysis at the individual level; variation in underlying estimator and specification) with a given dataset. Because these estimates range in magnitude and in precision, we report whether they are detectably different from zero ("statistically significant") and, if statistically significant, whether the effect represents an increase or a decrease.

The data sources that allowed analysis of marriage stock at the state level showed no significant change in the stock of individuals reporting that they were currently married after states legalized marriage for same-sex couples. This finding held regardless of state-specific economic and marriage-market conditions and cohort-specific linear time trends. Similarly, the data sources that allowed analysis of marriage stock at the individual level showed no change in marriage stock resulting from legalization of marriage for same-sex couples, and some analyses suggested that the married share of the population increased.

In the ACS data, which allowed the researchers to distinguish the gender composition of couples, the results suggest that marriage entry increased among different-sex couples following legalization of marriage for same-sex couples. The CDC data offer still more evidence: Although the researchers could document only total marriages (of both same-sex and different-sex couples) with the CDC data, and not influences on different-sex marriages specifically, there is overwhelming evidence of an increase in overall marriage. Across analyses, estimated effects are generally either not statistically significant or indicative of a significant increase in marriage; the few models with significant decreases are of smaller magnitude.

Using the ACS and CPS data, the researchers further investigated whether legalization of marriage for same-sex couples would lead to more nonmarital cohabitation for different-sex couples. They found no evidence of this. Instead, they found some indications that nonmarital cohabitation decreased among different-sex couples after marriage for same-sex couples was legalized, suggesting a movement toward marriage (consistent with the increase documented above) or living apart while unmarried. The analyses consistently indicate that there has been no increase in cohabitation. The estimated magnitude of the potential decrease is sensitive to model and specifications.

No Consistent Evidence of Impacts on Divorce

Opponents of legalizing marriage for same-sex couples expressed concerns about divorce rates, predicting that allowing same-sex couples to marry would diminish the value of marriage for different-sex couples, leading to increased rates of divorce. In analyses that mirrored their analyses of effects on marriage rates, the researchers used ACS, CPS, and PSID data to assess whether legalizing marriage for same-sex couples led to changes in the share of the population currently divorced (divorce stock) or the rate of new divorces (divorce flow). As with the above analyses using the ACS and CPS, they limited their analyses to individuals not in a formalized (cohabiting or married) same-sex couple. Their estimates pertain to the effects on divorce in states that changed their marriage policies between 2000 and 2014. Figure 2 summarizes what these analyses found across specifications.

A detailed description of the figure is in the text directly below the figure note.

SOURCE: Features data from the ACS, CPS, and PSID.

NOTE: Stat. sig. refers to statistical significance at the ɑ < 0.05 level. The analytic period is 2000 to 2014. The figure summarizes findings from difference-in-differences regression models of each outcome. Bars indicate the proportion of the total number of post-legalization coefficients from all models in each significance category. The PSID bar has half the number of coefficients summarized because PSID analysis is conducted only at the individual level. See Karney et al., 2024, for details.

^aAll marriages are included, regardless of partner sex.

Taken together, analyses of these three datasets yield mixed findings of the effect of changes in marriage policy on divorce. Most estimates are not statistically significant, and the majority of those that are statistically significant point to a decline in divorce; however, the researchers found both positive and negative statistically significant estimates in the ACS, depending on model specification.

When the researchers looked at divorce stock, they found that analyses of ACS data on individuals showed a significant increase, a significant decrease, or no significant change, depending on the specification. CPS analyses found both no statistically significant change and a statistically significant decrease. PSID analyses of individuals suggest that there was no statistically significant change in the share divorced.

Results for the effect of changes in marriage policy on divorce flow generally suggest that there was no change in the flow of new divorces, and some analyses suggest that there was a statistically significant decrease. ACS estimates suggest that there was a decrease in divorce flow five years after marriage policy change. PSID statistics suggest that there was no statistically significant change in divorce. (The CPS does not separately assess recent divorces, so the researchers could not analyze divorce flow.) Altogether, these analyses cannot rule out an increase, a decrease, or no change in divorce rates resulting from changes in marriage policy.

Changes in Marriage Attitudes

Using MTF data, the researchers explored potential changes in marriage attitudes among a population generally not yet embarking on family formation: high school seniors. The researchers viewed the marriage attitudes of these students as a potential indicator for future family formation decisions. Figure 3 summarizes changes in marriage attitudes among high school seniors after their states legalized marriage for same-sex couples. The researchers found no consistent evidence of a negative shift in attitudes toward the institution of marriage, and they found some evidence of an improvement.

A detailed description of the figure is in the text directly below the figure note.

SOURCE: Features data from MTF.

NOTE: Stat. sig. refers to statistical significance at the ɑ < 0.05 level. The analytic period is 2000 to 2014. The full survey questions are as follows: "Most people will have fuller and happier lives if they choose legal marriage rather than staying single, or just living with someone" (agree or mostly agree); "Having a good marriage and family life" (quite or extremely important); "One sees so few good or happy marriages that one questions it as a way of life" (disagree or mostly disagree); and "It is usually a good idea for a couple to live together before getting married in order to find out whether they really get along" (disagree or mostly disagree). The figure summarizes findings from difference-in-differences regression models of each outcome. See Karney et al., 2024, for details.

^aThis outcome measures a negative sentiment toward marriage and/or family formation, so it was reverse-coded (assessing disagreement) to be consistent in interpretation and direction with other outcomes.

Three of the survey items in Figure 3 specifically address marriage. After states legalized marriage for same-sex couples, the RAND researchers observed no change in the perception that marriage leads to fuller lives, although most of the imprecise estimates are positive. The researchers also found no consistent change in the perceived importance of a good marriage. They found some increases in disagreement with the statement, "One sees so few good or happy marriages that one questions it as a way of life." They also observed some modest increases in disagreement with the statement regarding cohabitation. In a small number of specifications (across outcomes), they documented small decreases, though the vast majority of their models suggested that there was either no change or an improvement in attitudes toward marriage.

Conclusions

Overall, the analyses found no evidence of a retreat from marriage resulting from the legalization of marriage for same-sex couples. In fact, the only evidence of change that the researchers found is of a possible increase in marriage resulting from legalization of marriage between same-sex partners—an increase that is not just the result of newly marrying same-sex couples. Explanations for such an increase might include (1) renewed salience of the value of marriage created by the prolonged national discussion of marriage for same-sex couples, (2) heterosexual individuals perceiving themselves as allies waiting to marry until same-sex couples could, or (3) bisexual individuals waiting for states to license marriages between same-sex couples before forming different-sex marriages in an acknowledgment of their underlying identity. In comparing results for states that legalized marriage for same-sex couples earlier with results for states that did so later, the researchers found no difference in the estimated effects, suggesting that these effects may be generalized to states that legalized after the Obergefell v. Hodges decision.

The researchers' analyses suggest that the arguments available to opponents of legal marriage for same-sex couples today are narrower than they once were. Where once it was possible to speculate about the consequences of extending legal status to same-sex couples in the United States, now the consequences are known. There is no empirical basis for concerns that allowing same-sex couples to marry will negatively affect the intentions and choices of different-sex couples and families. The only empirical evidence of effects that the researchers found suggests that policies allowing marriage for same-sex couples lead to a renewed salience of marriage among the broader public.

Note

• [1] The term marriage-market conditions refers to the characteristics of potential spouses (i.e., the number of men and women in a given age and/or education grouping).

Available for Download

Download PDF

Topics

• Children and Families Legislation
• Courts
• Government Legislation
• LGBTQ+ Populations
• Marriage and Divorce
• United States