Anti Social Effect

Sexuality Research and Social Policy
https://doi.org/10.1007/s13178-020-00509-y
The Anti-Social Effects of Legalizing Same-Sex Marriage: Fact

or Fiction?
Laura Langbein 1 & Brandon Ranallo-Benavidez 2 & Jane E. Palmer 3
Accepted: 27 October 2020

# Springer Science+Business Media, LLC, part of Springer Nature 2020
Abstract
Introduction Previous research examines the effects of same-sex marriage on many child and family outcomes, but only a small
subset examines the effects of laws on those outcomes. We evaluate the effects of same-sex marriage legalization in the USA on
four socio-familial outcomes.
Methods We use currently available public data from the U.S. Census and CDC to analyze changes in state-level legalization of
same-sex marriage on rates of child poverty, divorce, marriage, and children living in single-parent households within each state
from 2011 to 2016. The estimators use traditional cross-sectional time-series methodologies, along with adjusting for high-
dimensional fixed-effects (HDFE) clustering to account for both spatial and temporal dependence of state-time observations.
Results We find no evidence to validate claims of negative ramifications from same-sex marriage legalization on these outcomes.
Discussion With respect to the arguments articulated in Supreme Court amici briefs, we show that assertions of negative social
effects of legalized same-sex marriage are largely unsupported.
Conclusion In addition to illustrating the gains from HDFE estimators, we conclude that warnings of likely negative effects from
same-sex marriage, such as disallowing adoption by same-sex couples, are not credible.
Keywords Same-sex marriage . LGBT . Socio-familial outcomes . High-dimensional fixed-effects . Multi-way clustering
Enacting policies that support a stable family and propitious Previous research examines the effects of same-sex mar-
childhood undergird both proponents’ and opponents’ argu- riage on many child and family outcomes, but only a small
ments concerning same-sex marriage legalization. This paper subset examines the effects of laws on those outcomes. This
directly responds to key claims filed before the Supreme Court study examines whether empirical evidence supports claims
that same-sex marriage federalization would deleteriously af- of negative consequences of same-sex marriage laws. Many
fect children and families. In so doing, this study is motivated state legislatures are currently deliberating or have recently
by a desire to link public policy research with assertions made enacted new policies regarded as unfavorable to LGBTQ peo-
to policymaking bodies such as the Supreme Court. Coupling ple (e.g., Family Equality Council [FEC], 2018) such as
such research directly with arguments proffered to curb the “bathroom bills,” religious exemption bills, or bills that allow
expansion of the right to marry is vital—both to ensure re- child welfare agencies to deny same-sex couples for foster
search is responsive to political actors’ claims and to bolster placement or adoption. The idea that same-sex couples or
the utility of such research in the creation of future policies. LGBTQ people should not be parents is not new despite sub-
stantial evidence that children of same-sex couples fare just as
well, or better, than children of opposite-sex couples (Adams
* Laura Langbein & Light, 2015; Biblarz & Stacey, 2010; Herek, 2006; Frank,
langbei@american.edu Weeden, & Baker, 2018; Manning, Feltro, & Lamidi, 2014;
Meezan & Rauch, 2005; Potter, 2012).1 The issue remains
1
Department of Public Administration & Policy, American controversial around the world: while it is legal in most
University, 4400 Massachusetts Avenue NW, Kerwin Hall,
Washington, DC 20016, USA
1
2
Department of Political Science, Winthrop University, Rock There are extensive discussions that critique (e.g., Marks, 2012) and defend
Hill, SC, USA the validity (e.g., Amato, 2012) of these studies. One major study that found
negative impacts of same-sex parents on child outcomes (Regnerus, 2012) was
3
Department of Justice, Law & Criminology, American University, criticized when a replication did not come to the same conclusion (Cheng &
Washington, DC, USA Powell, 2015)
Sex Res Soc Policy
democracies, including the Americas, W. Europe, and Though these briefs assert numerous testable claims,
Australia and New Zealand, it still remains controversial in we limit our empirical analysis to four core socio-
many of those places, and in most of Africa and Asia as well familial outcomes: childhood poverty, divorce rates, mar-
(Felter & Renwick, 2019; Masci & DeSilver, 2019; Masci, riage rates, and single-parenting, each of which several
Sciupac, & Lipka, 2019). This study assesses some of the oppositional briefs associate with same-sex marriage le-
opponents’ concerns that legalized same-sex-marriage would galization (e.g., 47 Scholars, 2015; 100 Scholars, 2015;
adversely affect the stability of families and the welfare of Alabama, 2015; Anderson, 2013; American College of
children. Pediatricians, 2015; Family Research Council, 2015;
Otter, 2015; Kentucky General Assembly, 2015;
Louisiana et al., 2015; Republican National Convention
Committee, 2015). We investigate state-level variation in
Consequences of Legalizing Same-Sex these outcomes before and after legalization using fixed-
Marriage effects, cluster-robust standard errors, and other panel da-
ta techniques (e.g., Cameron & Trivedi, 2010; Correia,
According to Posner (1973), a “law is supposed to change 2017b; Wooldridge, 2010). We find no support for the
human conduct by altering the returns that individuals get empirical claim of adverse consequences resulting from
from different kinds of behavior” (Romaniuc, 2016, p. 1). same-sex marriage laws on childhood poverty, divorce
The corresponding question in this study is whether rates, marriage rates, and single-parenting.
changing a law by legalizing same-sex marriage affects Following a review of the pertinent literature and the hy-
“the returns” of avoiding costly forms of some deviant potheses from the oppositional briefs, we describe the design
behaviors related to marriage and family. Specifically, and the data in this research. Next, we present results from
opponents of same-sex marriage legalization fear that if three different model specifications—(1) high-dimensional
“traditional marriage” is no longer codified in the law or fixed-effects (HDFE) multi-way clustering (Cameron,
expected socially, people will perceive that as also open- Gelbach, & Miller, 2011; Correia, 2017a, 2017b); (2) cross-
ing the door to anti-family values such as divorce or sin- sectional time-series with state-fixed effects and year indicator
gle-parenting. regressors with panel-corrected standard errors (Cameron &
Decrying the ills of legalizing same-sex marriage is not Trivedi, 2010; Wooldridge, 2010), similar to Ludwig and
new (see Family Research Council [FRC], 2015). Indeed, fol- Cook’s (2000) identification strategy in their analysis of the
lowing the 2004 United States Presidential election, some po- Brady Handgun law; and (3) seemingly unrelated regression
litical commentators concluded that George W. Bush’s trium- with state and year indicator regressors. We find no statistical-
phant re-election hinged in some part on his opposition to ly significant differences in family outcomes within states that
same-sex marriage (Chinni, 2004; Dao, 2004; Lewis, 2005). legalized same-sex marriage after the Supreme Court’s
Moreover, as noted above, even though public sentiments Windsor ruling (2013) (the treated group of the reluctant com-
have changed since 2004, claims about potential negative con- pliers) as compared to states that had already legalized same-
sequences of legalized same-sex marriage remain (Anderson, sex marriage (the control group of earlier compliers). We con-
2013; Gallup, 2019; PRRI, n.d.). sider the importance for policy analysis of our null findings in
Many of these claims were presented to the Supreme Court the context of policymakers’ framing of this issue. We also
in its landmark Windsor (2013) and Obergefell (2015) cases. point to our use of multi-dimensional fixed effects (Correia,
Assertions of negative social impacts range from concerns that 2017b) as an important addition to policy analysis when pol-
some people might be less likely to marry due to changing icy adoptions by states (or other political units) are not inde-
social norms related to marriage (e.g., 100 Scholars, 2015) and pendent events. We end by noting limitations of this study,
fears of extensive cultural ruination (Young & Nathanson, needed next steps, and conclusions.
2007). For example, a brief jointly filed by the state govern-
ments of fifteen states,2 warned the Supreme Court: “…the
amici States […] highlight the negative consequences that will Prior Research
flow from a decision that the Fourteenth Amendment compels
recognition and adoption of same-sex marriage. Those conse- In theory, legalizing same-sex marriage can have positive,
quences would be severe, unavoidable, and irreversible” negative, or no different societal effects. Consistent with
(Louisiana et al., 2015, p. 13). the language of social losses articulated by the opposition
amici, we focus only on the negative effects.
2 Conceptually, if empirical evidence verifies these claims,
Louisiana, Utah, Texas, Alaska, Arizona, Arkansas, Georgia, Idaho, Kansas,
Montana, Nebraska, North Dakota, Oklahoma, South Dakota, and West it would be evidence of negative externalities from legal-
Virginia (listed here in same order as listed on the legal brief). izing same-sex marriage. Laws forbidding certain types of
Sex Res Soc Policy
marriage serve as barriers to entry.3 Such barriers might 100 Scholars, 2015; Allen & Price, 2015; Hawkins &
make economic sense if the contract adversely affects Carroll, 2015). These challenges provide evidence of negative
other parties outside the marriage, especially children: social consequences from legalization of LGBTQ rights to
these are negative externalities. However, if laws marry with respect to outcomes such as economic stability
restricting certain types of otherwise voluntary marriage (Allen, 2006; Cohen & Wright, 2011), high school graduation
contracts either reduce gains to the individuals involved rates (Allen, 2013), juvenile delinquency (Wardle, 2007), the
or gains to society, then those laws may exacerbate social spread of disease (Texas Values, 2015), state-level welfare
costs, rather than reduce them. That is our focus in this funding solvency (100 Scholars, 2015; Wardle, 2001), mar-
paper, as well as the focus of limited previous research riage rates (Cere & Farrow, 2004; Cherlin, 2004, 2009),
(see Langbein & Yost Jr., 2009; Trandafir, 2014a, 2015).4 fatherlessness (Hawkins & Carroll, 2015; Wax, 2007), di-
One of the earliest quantitative analyses of the impact of vorce (Marquardt, 2006; Wax, 2009), family formation
same-sex marriage laws in the USA was Langbein and Yost (Murray, 2013; Wardle, 2007), and even the foundational no-
Jr. (2009) finding of no negative externalities imposed by its tion of intimacy (Giddens, 1992), in addition to other negative
legalization. Critiqued for its limited time frame, questionable impacts.
data replicability, and low statistical power (Allen & Price, It is common to warn that same-sex marriage unravels the
2015), the Langbein and Yost study leverages empirical caus- connections that bring “responsible procreation” into a famil-
al identification strategies to scrutinize the hypothesized neg- ial setting. Supporting this argument, some studies suggest
ative effects of same-sex marriage. The fact that Langbein and that same sex marriage laws will mean that heterosexual
Yost Jr. (2009) found no such causation has been affirmed in men will feel less pressure to engage in stable family forma-
succeeding scholarship using a longer time frame (e.g., tion (Cere & Farrow, 2004; 100 Scholars, 2015; Hawkins &
Dillender, 2014; Trandafir, 2014a, 2014b, 2015). For exam- Carroll, 2015; Marquardt, Glenn, & Clark, 2010; Marquardt,
ple, using difference-in-difference strategies, Dillender (2014) 2010; Wardle, 2001, 2007; Wax, 2007, 2009; Young &
also found no support for claims that same-sex marriage re- Nathanson, 2007). Further, citing studies of the negative ef-
duces opposite-sex marriage rates in the USA. Additionally, fects of fatherlessness on children (Allen, 2010; Ellis et al.,
in three different studies, Trandafir uses synthetic control 2003; Flouri, 2007; Flouri & Buchanan, 2003; Harper &
methodologies (Trandafir, 2014a, 2014b), difference-in- McLanahan, 2004), the Obergefell (2015) brief submitted to
difference models (Trandafir, 2015), and country-by-country the Supreme Court by the 100 Scholars of Marriage (2015)
case studies of OECD nations and the Netherlands, conclud- argued that men who procreate are particularly susceptible to
ing that there is no evidence of adverse effects of the introduc- cultural arguments that dissolve the link between marriage and
tion of same-sex marriage on family formation. biological parenthood (Allen, 2010; Cherlin, 2004, 2009;
More recently, several studies have analyzed the impact of Cohen & Wright, 2011; Hawkins & Carroll, 2015; Manning
laws supporting LGBTQ rights in the USA and globally on a & Lamb, 2003; Marquardt, 2010; Murray, 2013).
variety of economic outcomes. For example, Badgett, Though there are myriad postulated effects opponents
Nezhad, Waaldijk, and van der Meulen (2014) analyzed data have presented in other research (as recounted above), this
from 132 countries from 1966 to 2011. They found that in- study intentionally focuses on testable claims made in amici
creasing legal rights for LGB persons is positively associated briefs submitted before the Supreme Court beseeching that
with an increase in real gross domestic product per capita. body to refuse to constitutionalize a right to marriage regard-
Nevertheless, many of these studies (Langbein & Yost Jr., less of sex or gender. Framing opposition to marriage equal-
2009; Dillender, 2014; Trandafir, 2014a), all of which find no ity before the judiciary may require a secular, legal, consti-
evidence of negative effects of same-sex marriage, have been tutional framing to arguments in a way that is not necessarily
rebuked, either for poor methodology (e.g., Allen & Price, required in other research or writing. Nevertheless, specifi-
2015; Dinno, 2014), or for erroneous results entirely (e.g., cally testing empirically based concrete hypothesized nega-
tive ramifications of the legalization of marriage equality,
3
stated in amici briefs themselves, speaks directly to the pol-
In economics, non-price entry barriers (such as refusing to hire based on race
or gender) are inefficient: they impose costs not only on the excluded group
icy concerns articulated before the High Court—a meaning-
but also on consumers or employers, while conferring smaller benefits (higher ful and necessary contribution. The law should not ignore
wages) on the privileged group. There are no social benefits. reality. While this study cannot speak directly to all of claims
4
Considerable research shows that marriage in general benefits both the two
made before the Court, it can evaluate an empirically falsifi-
parents and the children, partly due to scale economies along with more in-
come. In addition to within family benefits, external benefits of marriage able assumption in which petitioners couch their briefs.
include reduced demand for many social services (e.g, welfare and other in- While this paper cannot address all potential effects of legal-
come supports; police) that might otherwise be required if there were more izing same-sex marriage in the USA, it sheds light on an
single-parent families. See Ribar (2015); Sawhill and Thomas (2005); Thomas
and Sawhill (2002); Sawhill and Haskins (2016); McLanahan and Sawhill empirical claim that, if it were true, might justify legal ques-
(2015); Zissimopoulos, Karney, and Rauer (2015). tions about same-sex marriage.
Sex Res Soc Policy
These alleged negative externalities connect most of Hypothesis 1: Same-Sex Marriage Legalization Will
the hypotheses offered by the briefs in opposition to Increase Child Poverty
same-sex marriage. Our research examines the empirical
veracity of some of these claims and improves on previ- A brief filed by the Republican National Convention
ous scholarship on same-sex marriage in the USA in Committee (RNCC) states: “The traditional family guards
many ways. We also examine outcomes (child poverty, against poverty and the need for the growth of the welfare
divorce rates, marriage rates, and the share of children state” (RNCC, 2015). Similarly, Idaho’s Republican
raised in single-parent households) that, if they occur, Governor submits in his brief: “…children do better emotion-
should be apparent within a few years of the passage of ally, socially, intellectually and economically when raised by
a law legalizing same-sex marriage, if they happen at all. their biological mothers and fathers” (Otter, 2015, p. 3,
Specifically, compared to previous research, this study emphasis added). The mechanism for this increase in poverty
collectively adds (1) more recent years of analyses, (2) is often understood to be the lack of social pressure for het-
state and year fixed effects that also account for clustering erosexual copulation to take place in wedlock, resulting in
between states, (3) improved comparison groups between fewer fathers forming committed bonds with their children
“treatment” and “control” states, and (4) significant im- and wives (Allen, 2006; Cere & Farrow, 2004; Cherlin,
provements to the econometric analytics of prior research, 2004, 2009; Hawkins & Carroll, 2015; Manning & Lamb,
including at least 6 different estimators of the 4 focal 2003; Wardle, 2001, 2007; Wax, 2007; Young &
parameters. Importantly, this study allows for large Nathanson, 2007). Following from these expectations, we test
groups of reasonably comparable states to be included in the null hypothesis against the alternative hypothesis that
both “treatment” and “control” designations. Instead of same-sex marriage legalization will increase childhood
looking at only one or two “early-adopting” states (e.g., poverty.
Massachusetts or Vermont) compared to all the others, we
compare early, pre-Windsor adopters to the reluctant, but Hypothesis 2: Same-Sex Marriage Legalization Will
nonetheless compliant post-Windsor adopters. We omit Increase Divorce Rates
the most reluctant post-Obergefell adopters. We also rec-
ognize that states are not independent actors by using The second hypothesis follows from the idea that
HDFE estimators that allow for cross-sectional correlation “redefining” marriage destabilizes it. For example, in a
between state estimators. In other words, we account for brief submitted to the Supreme Court by the government
the fact that Oklahoma residents with—or in search of— of the state of Alabama, the amici curiae write, “The
same-sex partners may change their place of residence to States’ actual interest—promoting child-rearing by both
states with conforming laws. If this kind of omitted cross- biological parents…” (Alabama, 2015, p. 3) implies that
section correlation exists, using HDFE estimators reduces the Court should maintain opposite-sex-only definitions
the variance of estimators, raising t-statistics, and making of marriage to promote biological, nuclear family durabil-
it harder to find insignificant estimates, which is a higher ity. This admonition is nearly universal in the opposition-
threshold for this study as it rests on a claim of no al briefs (passim 100 Scholars, 2015; Texas Values,
(harmful) effect. 2015). Many of the briefs claim that same-sex marriage
undermines this goal by encouraging parents to dissolve
marriages via divorce (e.g., 47 Scholars, 2015; 100
Scholars, 2015; ACP, 2015; FRC, 2015). Fearful of a
Hypotheses decline in “domestic stability” (Kentucky General
Assembly, 2015, p.14), the briefs frequently warn the
This study tests null hypotheses that same sex marriage Supreme Court Justices that opposite-sex couples will dis-
laws have no effect, compared to one-tail hypotheses solve their unions more readily if same-sex marriage is
asserting that a change in the law will have deleterious allowed, because re-writing marriage law to make it
effects by increasing child poverty, divorce rates, and the gender-neutral fundamentally erodes popular understand-
share of children raised in single-parent households, and ing of the significance of the marital bond (e.g., Louisiana
by decreasing marriage rates. These claims are particular- et al., 2015). Connecting same-sex marriage directly to
ly important, because they are foundational to the claims opposite-sex divorce, Idaho Governor Otter wrote, “For
of other adverse (and longer-term) outcomes of same-sex children, when these [opposite-sex marriage] norms are
marriage, including economic instability (Allen, 2006; vibrant in families and society, the consequence is fewer
Cohen & Wright, 2011), lower high school graduation fatherless (and in some cases motherless) children and
rates (Allen, 2013), juvenile delinquency (Wardle, less divorce, neglect and abuse” (Alabama, 2015, p. 3,
2007), and the spread of disease (Texas Values, 2015). emphasis added).
Sex Res Soc Policy
Hypothesis 3: Same-Sex Marriage Legalization Will the necessity of “encouraging responsible [i.e., within
Decrease Marriage Rates w e d l o c k ] p r o c r ea t i o n by o p po s i t e- s ex c o up l es ”
(Alabama, 2015, p. 8) as a primary rationale for the
Part of the argument for the deleterious effects of same-sex state’s continued reliance on opposite-sex-only definitions
marriage on subsequent marriage rates is that heterosexuals, of marriage. The brief by 47 Scholars (2015) provides a
particularly men, who previously had been only marginally particularly illustrative articulation of the hypothesized
interested in marrying, induced into marriage by societal ex- mechanism by which same-sex marriage legalization
pectations, no longer would abide the bonds of matrimony would undermine family stability and increase single-
(e.g., 100 Scholars, 2015; Hawkins & Carroll, 2015; Young parenthood:
& Nathanson, 2007). Citing the work of Wax (2007, 2009),
the brief of the 100 Scholars of marriage (2015) explains the Moreover, recognizing same-sex bonds as marriages
rationale behind the expectation of an increase in single-parent would send the message—which other public institu-
families if marriage is not restricted to heterosexual couples: tions would reinforce—that it matters not, even as a rule,
whether children are reared by their biological kin, or by
[I]t’s not a father’s ‘see[ing] a child being raised by two a parent of each sex at all; indeed, that it is bigoted to
[married] women’ that will reduce heterosexual males’ think otherwise. As this message is internalized—as
enthusiasm for marriage. It’s the fact that, before they mothers and fathers each come to seem optional—it will
become fathers, marriage will have already been be harder to send the message that fathers, say, are es-
redefined in a way that signals their involvement is less sential. Men are likely to feel less urgently any respon-
important. Although not all heterosexual fathers or po- sibility to stick with their wives and children to offer any
tential fathers will have less interest in marriage, distinctive benefits of paternity, and men and women are
some—especially those at the margins—undoubtedly less likely to feel motivation or social pressure to com-
will (100 Scholars, 2015, p. 29, emphasis in original). mit to each other in marriage before having children in
the first place (2015, p. 17).
The same brief warns the High Court that, in the long
run, there will be decreases in marriage rates. The RNCC This paragraph summarizes not only the hypothesized neg-
reiterates these concerns as well: “Same-sex marriage has ative effect of same-sex marriage legalization on dual parent-
not promoted marriage, either, as marriage rates continue hood but also the mechanism by which it is expected to
to decline in most countries that recognize it” (RNCC, function.
2015, p. 6). Idaho’s Governor expounds: “redefining mar-
riage in genderless terms—which is necessary to the ac-
commodation of same-sex marriage—will undermine the Data and Methods
norms of man-woman marriage, resulting in fewer mar-
riages” (Otter, 2015, p. 4). As these assertions make clear, To test the hypotheses presented above, we utilize public-
many oppositional public-policymakers fear a decline in ly available data from multiple US government sources. It
marriages, implying fear that genderless marriage would would be possible to use data from these sources that
result in a negative externality. includes all states and DC for most years between 1990
and 2016. However, rather than compare non-adopters
Hypothesis 4: Same-Sex Marriage Legalization Will (pre-2011) to all adopters (both early, later, and forced),
Increase Single-Parenting we believe that a better comparison is between earlier and
later adopters. Using this selected sample of relatively
Same-sex marriage opponents also argue that, by comparable states, we reduce the likelihood that omitted
diminishing the normative connection between reproduc- variables, including pre-existing economic and political
tion and [opposite-sex] marriage, the legalization of same- differences, account for our results. By omitting the latest,
sex marriage will produce an increase in unwed mothers “forced,” post-Obergefell, adopters, we examine whether
(100 Scholars, 2015; Wax, 2009; Young & Nathanson, adoption of SSM, independent of other differences, af-
2007). As Idaho’s Governor asserts before the Court, fects family outcomes. The treatment states are the later
“The man-woman definition […] conveys [the norm that] (but not the latest) adopters, while the “control” (early)
men and women should postpone procreation until they states had legalized same-sex marriage before 2014. This
are in a committed, long-term relationship…i.e., hetero- also gives us a longer time frame to examine whether the
sexuals’ sexual behavior should be ‘channeled’ into such laws could affect family outcomes. In other words, in-
stable relationships” (Otter, 2015, pp. 11–12). Indeed, the cluding the latest adopters in our analysis would be likely
Alabama state government’s oppositional brief refers to to bias the sample in favor of a “no impact” result.
Sex Res Soc Policy
Independent Variable same-sex couples to wed. These are the “control” states; they
are early-adopter states with no policy change in our time
Windsor (2013) was the change that affected the treatment period. We code the “treatment” states as 1 after 2014.
states in our analysis. That ruling only re-legalized same-sex These are states that only started offering same-sex couples
marriage in one state—California—but it laid the precedent marriage certificates in 2014. This places 34 states in our
many Circuit Courts later used to invalidate bans on same-sex primary model specification.6
marriage state by state. Thus, our treatment group is the flurry There is another reason that we limit our data to 2011–
of states affected by rapid caselaw changes between 2014 and 2016. If same-sex marriage does have the negative social out-
2015, the “later treatment” states that legalized same-sex mar- comes hypothesized by the oppositional briefs, we would ex-
riage only after the Windsor (2013) decision. Even later, the pect that, compared to states that already allowed same-sex
Supreme Court constitutionalized same-sex marriage rights marriage, states that did not but were then required7 in 2014 or
nationwide in the caselaw of Obergefell (2015). The post- later to do so would see a marked post-policy effect in the
Windsor states are likely to be politically more comparable short-term outcome measures that we examine. These pre/
than selecting a larger group that includes both the forced, post comparative analyses require control states and treatment
post-Obergefell, adopters and the earlier, pre- and post- states to have parallel prior trends (and constant differences) in
Windsor adopters (which is all the other states). We use data each of the dependent variables (e.g., divorce rate) before the
from years and states that had legalized same-sex marriage policy intervention (Correia, 2017b; Ludwig & Cook, 2000;
before Obergefell (2015) could have any impact, which means Wooldridge, 2010). For three of the four dependent variables
the data in our analyses span the years 2011–2016. We com- (child poverty, divorce rate, and single-parent households),
pare early adopters5 to late adopters, omitting the most recent the prior trends assumption clearly holds in both model spec-
(and reluctant) adopters. ifications (with and without 2013 same-sex marriage legaliz-
While same-sex marriage laws rapidly expanded in the ing states). However, while marriage rates in control and treat-
mid-2010s, most states legalized same-sex marriage only ment states do not follow similar patterns prior to the year
post-Windsor, decided in 2013 (see Fn. 5.). Because there 2014, they became similar, and unchanging, after 2011 (see
are not yet enough post-2015 years of data available for Figs. 1, 2, 3, and 4).
robust empirical investigations of the impact of the
Obergefell decision, and because these states are clearly
the most reluctant adopters, our analyses drop states that Dependent Variables
only legalized same-sex marriage in 2015 (see Fn. 5). In
addition to their non-comparability, we also drop the very Child Poverty
late adopting states because of the lack of time for the
anticipated negative consequences to occur: If states adopt To measure child poverty, we use the American
a law in 2015 legalizing same-sex marriage, it is not like- Community Survey data from the U.S. Census on the
ly to affect the dependent variables of divorce, childhood percent of children below the federal poverty level by
poverty, single-parenting, and aggregate marriage rates state and year, 2011–2016. The time trends of child pov-
until long after 2016. We compare states that already erty rates in the treatment and control states follow re-
had legalized same-sex marriage prior to 2014 to states markably similar trends prior to and after the 2014 insti-
that legalized same-sex marriage in 2014, when Circuit tution of same-sex marriage. Additionally, for nearly all
Courts across the nation began invalidating bans on the states in our analysis, the common trends in the de-
LGBTQ marriage rights, utilizing the Windsor precedent. pendent variables in the treatment and control states were
Using the methodological frame of Ludwig and Cook’s observable even before the first same-sex marriage in
(2000) test of the effect of the Brady Handgun law, we code America in 2004 and continued to follow similar relative
the main independent or policy variable (same-sex marriage patterns across both the Great Recession of the late 2000s
legalization in 2014) as 0 for all observations prior to 2014 and and during the recovery since.
continuing as 0 after 2014 for states that had already allowed
5
The early adopter control states include the 2013 legalizers (CA, DE, HI,
MD, MN, NJ, NM and RI) and the earliest adopters (MA, 2004; CT, 2008; IO,
6
V, 2009; NH, DC, 2010; NY, 2011; ME, WA, 2012). The treated states, forced However, as a robustness check, we also provide a secondary model speci-
to adopt by Windsor (except IL, which voluntarily adopted after Windsor) fication that drops all states from our “control” group which did not have same-
include OR, ID, MT, WY, NV, UT, AZ, CO, OK, WI, IL, IN, PA, WV, sex marriage for multiple years prior to 2014. The results in Tables 1, 2, 3, and
VA, NC, and SC. Omitted states (the late legalizers, forced to legalize by 4 do not change.
7
Roe in 2015) include ND, SD, NE, KS, TX, MO, AR, LA, MI, OH, KY, Note that Illinois was the only state post-Windsor to voluntarily legalize
TN, MS, AL, GA, FL) same-sex marriage.
Sex Res Soc Policy
Fig. 1 Time trends of children

living in poverty. The blue line
illustrates the annual child
poverty rate in states that were
early adopters of same-sex mar-
riage. The red line illustrates the
annual child poverty rate in states
that were late adopters of same-
sex marriage. The green vertical
line illustrates the policy inter-
vention and the green years mark
the timeframe of the study.
Throughout the entire window,
treatment states had a higher share
of children in poverty (measured
as living under the federal poverty
level) than control states.
Treatment states range from
20.7% of children in poverty in
2011 to 17.5% in 2016. Control
states range from 18.4% of chil-
dren in poverty in 2011 to 16.0%
in 2016
Divorce Rates divorce rate had been declining for nearly all the past
two decades, except for maybe a small uptick during
To measure divorce rates, we use information from the the Great Recession. While the levels of divorce differ,
U.S. Centers for Disease Control and Prevention (CDC) the trends in treatment and control states are parallel, no
on the number of divorces per 1000 residents per state matter whether 2013 legalizers are included or excluded
per year. For both treatment and control states, the in the analyses.
Fig. 2 Time trends of divorce

rates. The blue line illustrates the
annual divorce rate in states that
were early adopters of same-sex
marriage. The red line illustrates
the annual divorce rate in states
treatment states had a higher di-
vorce rate (measured as the num-
ber of divorces per 1000 popula-
tion) than control states. In the
analytic sample of 2011 through
2016, treatment states’ average
divorce rates range from 4.1 per
1000 population in 2011 to 3.4 in
2016. Control states’ average di-
vorce rates range from 3.3 per
1000 population in 2011 to 2.8 in
2016
Sex Res Soc Policy
Fig. 3 Time trends of marriage

rates. The blue line illustrates the
annual marriage rate in states that
were early adopters of same-sex
marriage. The red line illustrates
the annual marriage rate in states
treatment states had a higher
marriage rate (measured as the
number of marriages per 1000
population) than control states. In
the analytic sample of 2011
through 2016, treatment states’
average divorce rates range from
8.5 per 1000 population in 2011
to 8.0 in 2016. Control states’
average marriage rates range from
7.6 per 1000 population in 2014
to 7.0 in 2015
Marriage Rates far back as 2000, the marriage rate had been significantly
lower in control states (the earliest adopters of same-sex
Using data from the CDC on the number of marriages per marriage), but the magnitude of the difference drops be-
1000 residents per state per year to measure marriage rate, tween 2000 and 2016. For example, in 2014, the differ-
we see that, unlike the other three dependent variables, ence between the average of treatment states’ and control
marriage rates may not meet the parallel trends assump- states’ marriage rates is 0.7 marriages per 1000, compared
tion between control and treatment states before 2014. As to a difference of 3.8 marriages per 1000 in 2000.
Fig. 4 Time trends of share of

children in single-parent house-
holds. The blue line illustrates the
annual percent of children in
single-parent households in states
that were early adopters of same-
sex marriage. The red line illus-
trates the annual percent of chil-
dren in single-parent households
in states that were late adopters of
same-sex marriage. The green
vertical line illustrates the policy
intervention and the green years
mark the timeframe of the study.
control states had a higher percent
of children in single-parent than
treatment states. In the analytic
sample of 2011 through 2016,
treatment states’ average ranges
from 31.8% in 2011 to 30.7% in
2016. Control states’ average
ranges from 33.6% in 2011 to
32.7% in 2015
Sex Res Soc Policy
However, our comparison starts in 2011, when the prior Estimation Models
trends and differences in marriage rates had become
similar. In the examination of the effects of same-sex marriage on
2014 legalizing states, we use the following model for each
Single-Parent Households of the four dependent8 variables:
Y i;t ¼ β0 þ βk X i;t;k þ βT i;t þ d i;t þ g t−1
The final dependent variable is the percent of children living
in single-parent households as measured in the American þ ei;t ð1Þ
Community Survey. In both treatment and control states, the
In this equation, Yi, t represents a dependent variable affected
average single-parenthood numbers climbed since 2000 but by same-sex marriage legalization (say, divorce rate) for state i in
leveled off generally after the recession. In short, the trends
year t, and Xi, t, k represents the set of k control variables described
and differences are steady over time, regardless of whether we
above. Ti, t represents the treatment effect, or the policy variable,
drop 2013 legalizing states from the control group or not. for one of the outcomes. Ti, t equals 0 for the legalizing states in
the sample prior to 2014, and 1 in any state for years 2014, 2015,
Control Variables and 2016 that legalized same-sex marriage after the Supreme
Court’s Windsor decision. Therefore, in our primary model
Finally, each of the models contains the same ten control
specification, β is the focal policy coefficient for the estimated
variables. Using data from the Bureau of Labor Statistics,
treatment effect of being a state that was compelled to legalize
we control for the percentage of the adult population who same-sex marriage in 2014 compared to those that had already
are unemployed in each state by year. Parental unemploy-
legalized. The model includes separate dichotomous indicator
ment tends to exacerbate childhood poverty (Lichter &
variables for each state, di, t, to capture unmeasured state-
Crowley, 2004). From the Census’ American specific fixed effects that both correlate with policy adoption
Community Survey, we include median household income
and cause the level of the dependent variable to differ across
(adjusted to 2016 real dollars and measured in thousands)
states. More specifically, the fixed effects indicators capture both
as a covariate. The impact of household income on family unmeasured state characteristics that induce (or reduce) the pro-
structure and child poverty is profound (Lerman, 1996).
pensity to adopt laws to protect SSM and that may cause the four
Considering research showing a relationship between spa-
family outcomes that we examine in this study. The model also
tial factors and child poverty rates (Voss, Long, Hammer, includes a set of (lagged) year-indicator variables, gt − 1. We use a
& Friedman, 2006), we control for population density.
lagged indicator of year to capture the effects of unmeasured prior
From their research on gifted children from disadvantaged
events (such as major electoral and policy changes) on the out-
backgrounds, Goings and Ford (2017) find not only that comes we measure conditional on the observed covariates.
race is a key factor in poverty and its intergenerational
We use three different estimates of the focal parameter in Eq.
transmission but also that the racial composition of age
(1). Following Correia (2017a, 2017b), we implement a high-
cohorts is crucial. Consequently, we control for each dimensional fixed-effects (HDFE) model with multi-variable
state’s childhood and overall racial composition by year.
state-year clustering as our primary specification model. The
For the racial composition of each state, we measure the
Correia methodology uses graph theory (Cameron et al.,
percentage of the population (both childhood and overall) 2011) to make joint clustering on multiple variables (e.g.,
who are White alone as well as the percentage who are
state-year clusters) feasible. In this case, clustering by state-
Latino (alone or Latino and at least one other racial/ethnic
year pairs accounts not only for likely homogeneity between
background); we omit the category Black alone as a ref- years but also between similar states: Montana, Wyoming, and
erence. Because Chen and Corak (2008) found limited
Colorado are not only later adopters, they are also near each
evidence that demographic age distribution also influ-
other; the same is true for Wisconsin, Illinois and Indiana. The
ences child poverty, we control for the overall age distri-
same is true for the cluster of control states (the earliest
bution by state each year, measuring the percentage of the
adopters) in New England; they also share ideological similar-
population who are children (age 0–17) and the percent-
ities with California. Thus, the use of bi-variate clustering by
age who are age 18 through 64, leaving the percentage
who are 65 or older as the omitted reference category. 8
For ease of interpretation, we use state-year levels in the dependent variables.
Lastly, Casper, McLanahan, and Garfinkel (1994) explore
However, because we use state fixed effects, the focal coefficient B represents
a “gender-poverty gap,” because women and men experi- the average within state difference in the dependent variable between treatment
ence different rates of poverty in America. Thus, we con- and control state (Ludwig & Cook, 2000); it is not a between state difference.
trol for the percentage of the state’s population each year As a robustness check, when we use difference-in-difference estimates of the
relation between state-year changes in the policy adoption variable to state-
who are female, with the percent male being the omitted year changes in dependent variables, along with the control variables and state
category of reference. fixed effects, the results do not change appreciably.
Sex Res Soc Policy
both time and state means that both the temporal and spatial net effect is to favor statistically significant estimates (in this
homogeneity between (some) state-year pairs is accounted for, case, of negative effects), if they are present. Yet, none of the
reducing estimates of standard errors. While it is common to estimates are significant at even nearly conventional levels.
expect temporal clustering, spatial clustering is not commonly
expected, but it is not uncommon. Spatial (cross-sectional) clus-
tering is mostly regional, but pairwise spatial clustering also Results
accounts for common patterns of mobility by, say, gay and
lesbian young people migrating from a non-SSM state to a state Child Poverty
where the possibility of same sex marriage is fully legal.
Finally, if we re-estimate the same models reported in columns As shown by Table 1, in all six model specifications for this
1–2 of Tables 1, 2, 3, and 4 below, which include traditional dependent variable, the policy variable is statistically insignif-
state and year fixed effects, but without the HDFE adjustment, icant. None of the six models find evidence to support the
the results are the same, but with larger standard errors. The claim that same-sex marriage increases child poverty rates
HDFE method is available in Stata (Correia, 2017b). (hypothesis 1). Figure 1 provides a time trend graph illustrat-
Another model specification uses simple cross-sectional ing the year-to-year aggregate change in childhood poverty in
time-series analyses with the xtreg command in Stata, adjusting early-adopting states compared to late-adopting states.
errors in panels for within-state and time homoscedasticity and Compared to states that had already legalized same-sex mar-
auto-correlation between years. Finally, we recognize that marriage, states that legalized same-sex marriage in 2014 saw no
riage, divorce, childhood poverty, and single-parent families are significant impact on the percent of children living in poverty
clearly related. For example, while marriage rates are not the in the three following years. Even the substantive implication
same as single-parenting, because married people may decide of the insignificant coefficients under a one-tail test implies
not to have children, most marriages do result in children, so the that SSM might even reduce childhood poverty: all the esti-
two are clearly related. Consequently, we cannot assume that mates are negative, with a range from 0.01 to 0.2% fewer
the estimates for each of our four equations are independent of children in poverty.
each other. While the equations include the same independent Overall, these results corroborate prior research demon-
variables, the omitted variables included in the error terms of strating a lack of evidence of negative externalities of same-
each of the four equations may well include the same (omitted) sex marriage (Langbein & Yost Jr., 2009; Dillender, 2014;
variables. Thus, while the equations “seem” unrelated, they Trandafir, 2014a). As expected, the percent of the adult pop-
really are not: the stochastic terms in the four regressions are ulation who are unemployed, holding other variables constant,
unlikely to be independent of each other, because omitted fac- is statistically significant and practically important in all six
tors that affect marriage also affect single-parenting, divorce models—signifying that a 5% increase in the unemployment
rates, and childhood poverty. Consequently, we estimate a rate coincides with increases of about 2–2.5 in the percent of
seemingly unrelated regression with state and year fixed ef- children living in poverty.
fects.9 We also estimate each of the models with 2013 legaliz-
ing states included and excluded from the control group.10 Divorce Rates
Finally, we note that all three adjustments (HDFE, xtreg, and
SUR) to reduce correlation of error terms within and between Again, in all six model specifications, the same-sex marriage
each estimating equation are designed to reduce variance in the legalization in 2014 variable is insignificant, as shown in
focal parameter estimate of the effects of legalizing SSM. The Table 2. Figure 2 provides a time trend graph comparing
early-adopting states’ divorce rates to late-adopting states’ di-
9
The SUR estimates show that the residuals are correlated: vorce rates. Compared to states that already had legalized
Correlation matrix of residuals:
childpov divorce marriage sglpar same-sex marriage, this signifies that states forced to do so
childpov 1.0000 in 2014 saw no increases in divorces per 1000 residents (hy-
divorce 0.0259 1.0000 pothesis 2). While none of the coefficients are significant, they
marriage − 0.1606 0.5399 1.0000
sglpar 0.3548 0.0124 0.0768 1.0000
all have a positive sign, implying under a one-tail hypothesis
Breusch-Pagan test of independence: chi2(6) = 246.067, Pr = 0.0000. test that full legalization of marriage might have induced a few
We also estimated an xtreg with state-level fixed effects (Cameron & divorces among opposite-sex couples. This could possibly
Trivedi, 2009; Wooldridge, 2010). Similar to Ludwig and Cook’s (2000)
assessment of the Brady bill, we also performed a specification of ordinary
account for some of the marriages we note in the following
least squares analysis with controls, heteroskedasticity-robust standard errors, table. In any case, the number of divorces is small, ranging
and state and year fixed effect regressors. We do not show the results, but they from 0.008 = 8 divorces per thousand to 0.06 = 60 divorces
do not differ from the results that we present.
10 per thousand.
Excluding 2013 legalizing states is a robustness check to provide a model
where all the “control” states had same-sex marriage legalization for more than Like the previous model, the only statistically significant
just 1 year. The results do not change. variables are some of the controls. In all six models, the
Sex Res Soc Policy
Table 1 Percent of children in poverty and same-sex marriage legalization (2011–2016)
(1) (2) (3) (4) (5) (6)

HDFE HDFE XTREG with XTREG with SUR SUR
AR(1) AR(1)
Regressors 2013 States 2013 States 2013 States 2013 States 2013 included 2013 excluded
included excluded included excluded
2014 SSM Legalizers − 0.170 (0.304) − 0.143 − 0.189 (0.440) − 0.010 (0.500) − 0.135 − 0.176 (0.407)
(0.426) (0.338)
% Unemployed 0.489*** 0.385* (0.176) 0.576*** (0.176) 0.602*** (0.188) 0.473*** 0.450**
(0.120) (0.162) (0.193)
Median Inc. (1000s) − 0.0464 − 0.0209 − 0.157*** − 0.173*** − 0.0378 − 0.0223
(2016 real dollars) (0.0461) (0.0498) (0.029) (0.032) (0.0306) (0.0342)
Population Density 0.00217 − 0.00139 0.00101*** 0.00079*** − 0.00039 − 0.00163
(0.00281) (0.00456) (0.000332) (0.00024) (0.0033) (0.00399)
% Children White as only race − 0.581 (0.299) − 0.143 (0.530) − 0.020 (0.058) − 0.126 (0.079) − 0.270 − 0.0666
(0.398) (0.481)
% Children Latino (in combination or 0.543 (0.316) 0.735 (0.451) − 0.133 (0.214) − 0.028 (0.232) 1.069** 0.784 (0.595)
alone) (0.493)
% Total Population White as only race − 0.00352 0.0909 (0.143) − 0.0049 (0.0547) − 0.037 (0.085) 0.0719 0.0819
(0.0918) (0.0657) (0.0737)
% Total Population Latino (in − 0.568 (0.400) 0.190 (0.884) 0.305 (0.265) − 0.046 (0.299) − 0.528 0.368 (1.13)
combination or alone) (0.879)
% Children (age 0–17) of total − 0.831 (0.594) − 0.578 (0.597) − 0.280 (0.238) − 0.459** − 0.817 − 0.722 (0.667)
population (0.207) (0.627)
% age 18–64 of total population − 0.422 (0.219) − 0.498 (0.420) − 0.693** (0.309) − 1.166*** − 0.562 − 0.577 (0.464)
(0.305) (0.424)
% Female of total population − 0.249 (1.556) − 0.538 (1.305) 0.713 (0.522) 0.604 (0.446) − 0.0516 − 0.302 (1.71)
(1.608)
Observations 216 168 216 168 191^ 162^
State FE Yes Yes Yes Yes Yes Yes
Year FE Yes Yes Yes Yes Yes Yes
R2 0.969 0.967 0.970 0.967
Adj-R2 0.959 0.955
Within R2 0.205 0.177
Within Adj-R2 0.166 0.123
Number of states 36 28 36 28 36 28
Number of years 6 6 6 6 6 6
Robust standard errors in parentheses

***p < 0.01; **p < 0.05; *p < 0.1
unemployment rate and the percent of children who are Latino already allowed for same-sex marriage (Table 3). In fact,
are associated respectively with a slight or .4 increase in the while insignificant, the estimates all have (small) positive
divorce rate. Similarly, for each percent of the statewide pop- signs, ranging from 0.06 to 0.42 per thousand, or from 60 to
ulation who are children (as compared to the reference cate- 420 more marriages. While there may be a few more divorces
gory of the percent aged 65 and over), divorce rates in all six from legalizing SSM, there are more marriages too, consistent
models increase by about 0.5. Again, there is no evidence of a with the possibility of fewer children in poverty from SSM.
negative effect of same-sex marriage on divorce rates. Figure 3 provides a time trend graph of the changes in the
marriage rate for the two groups of states. The marriage rates
Marriage Rates were not on parallel trajectories in early- and late-adopting
states prior to the Supreme Court’s ruling, but during the pe-
Examining the six different model specifications, we again see riod of our analysis the trends are parallel; the corresponding
that same-sex marriage legalization in 2014 in the treatment table shows that the differences in marriage rates are not sig-
states does not have a predicted statistically significant effect nificant. The full results also show important state and time
on marriage per 1000 population, compared to the states that specific differences, but none of the results imply that same-
Sex Res Soc Policy
Table 2 Divorces per 1000 and same-sex marriage legalization (2011–2016)
(1) (2) (3) (4) (5) (6)

AR(1) AR(1)
2014 SSM legalizers 0.00807 0.0560 0.0258 (0.0704) 0.0708 (0.0828)
0.00807 0.0560
(0.0612) (0.0707) (0.0561) (0.0673)
% Unemployed 0.113** 0.0972* − 0.00389 − 0.0180 0.113*** 0.0972***
(0.0404) (0.0456) (0.0306) (0.0357) (0.0268) (0.0318)
Median Inc. (1000s) (2016 real dollars) − 0.00485 − 0.00648 − 0.00498 − 0.00665 − 0.00485 − 0.00648
(0.00850) (0.00807) (0.00485) 0.00573 (0.00508) (0.00565)
Population density 1.33e-05 0.000456 0.0000394 0.0000572 0.0000133 0.000457
(0.000462) (0.000683) (0.0000823) (0.0000913) (0.000543) (0.000659)
% Children White as only race 0.162* 0.0780 − 0.00104 0.00258 (0.0171) 0.162** 0.0780
(0.0695) (0.0699) (0.0148) (0.0661) (0.0796)
% Children Latino (in combination or 0.401** 0.470** 0.0881 (0.0539) 0.105 (0.0807) 0.401*** 0.470***
alone) (0.111) (0.120) (0.0818) (0.0984)
% Total population White as only race − 0.00860 − 0.0102 0.00355 (0.0123) 0.00125 (0.0141) − 0.00860 − 0.0102
(0.0172) (0.0186) (0.0109) (0.0122)
% Total population Latino (in 0.00601 − 0.299 (0.159) − 0.111 (0.0687) − 0.128 (0.113) 0.00601 (0.146) −0.299 (0.187)
combination or alone) (0.245)
% Children (age 0–17) of total − 0.548** − 0.527** − 0.0305 − 0.0415 − 0.548*** − 0.527***
population (0.148) (0.151) (0.0581) (0.0640) (0.104) (0.110)
% age 18–64 of total population 0.0795 (0.121) 0.101 (0.117) − 0.0232 − 0.0194 0.0795 (0.0704) 0.101 (0.0768)
(0.0659) (0.0726)
% Female of total population 0.504 (0.328) 0.385 (0.331) − 0.439*** − 0.485*** 0.504* (0.267) 0.385 (0.282)
(0.137) (0.154)
Observations 191 162 191 162 191 162
Year FE Yes Yes Yes Tes Yes Yes
R2 0.969 0.972 0.969 0.972
Adj-R2 0.959 0.962
With R2 0.360 0.348
Number of statesa 32 27 32 27 32 27

***p < 0.01; ** p < 0.05; *p < 0.1
a
It is not mandatory that states report divorces to the CDC—the treatment states of California, Hawaii, and Minnesota, and the control state of Indiana did
not report. This explains the difference in number of observations for this dependent variable compared to the N for the three other tables
sex marriage laws are responsible for any differences in mar- adopting states. When we compare states that already
riage rates. allowed same-sex marriage to those states forced to do
so in 2014, the hypothesis that single-parenting would
Single-Parent Households increase following nationalization of same-sex marriage
is not supported. The range of the insignificant estimates
Finally, regarding the percent of children living in single- is from − .27 to + .29 per thousand, or 270 fewer kids to
parent households, Table 4 illustrates that none of the six 280 more kids in single-parent households. While not
models find a statistically significant relationship with consistently negative, the estimates are small and not sta-
same-sex marriage legalization. Figure 4 provides a graph tistically significant. Recall further that the results in
of time trends of the percent of children living in single- Table 1 suggest either no increase in children in poverty
parent households for treatment states compared to early- or a slight reduction, if we use a 1-tail test. Finally, none
Sex Res Soc Policy
Table 3 Marriages per 1000 and same-sex marriage legalization (2011–2016)
(1) (2) (3) (4) (5) (6)

AR(1) AR(1)
2014 SSM legalizers 0.0552 (0.329) 0.294 (0.396) 0.262 (0.223) 0.416 (0.261) 0.0857 (0.208) 0.283 (0.253)
% Unemployed 0.266 (0.187) 0.314 (0.181) 0.139 (0.0986) 0.177 (0.115) 0.253**
0.334***
(0.0996)
(0.120)
Median Inc. (1000s) (2016 real dollars) − 0.0114 − 0.00510 − 0.0127 (0.0156) − 0.0214 − 0.00990
− 0.00575
(0.0136) (0.0154) (0.0183) (0.0189)
(0.0213)
Population density 0.000216 0.000824 − 0.000321 − 7.04e-5 − 0.00150
0.000739
(0.00217) (0.00358) (0.000477) (0.000549) (0.00202)
(0.00248)
% Children White as only race − 0.0267 − 0.0943 − 0.0597 0.0570 − 0.0483 − 0.0684
0.206 (0.246)
(0.264) (0.473) (0.0866) (0.300)
% Children Latino (in combination or 1.343* (0.653) 1.677* (0.751) 0.846*** 0.291 1.135*** 1.722*** 1.69***
alone) (0.407) (0.304) (0.371)
% Total population White as only race − 0.0319 0.0167 (0.0407) − 0.00848 0.0442 (0.0458) 0.0401 0.0141
(0.0262) (0.0285) (0.0405) (0.0459)
% Total population Latino (in − 2.96* (1.205) − 3.88** − 1.09*** (0.373) − 1.408** − 2.82*** − 3.82***
combination or alone) (1.449) (0.591) (0.542) (0.703)
% Children (age 0–17) of total − 0.588 (0.451) − 0.686 (0.500) − 0.119 (0.288) − 0.189 (0.318) − 0.703* − 0.731*
population (0.387) (0.416)
% age 18–64 of total population 0.952* (0.451) 0.796 (0.402) 0.378 (0.248) 0.359 (0.269) 0.756*** 0.773***
(0.262) (0.289)
% Female of total population − 1.484 (1.399) − 1.787 (1.661) − 1.54** (0.660) − 1.62** − 1.85* (0.991) − 1.72 (1.06)
(0.774)
Observations 216 168 216 168 191^ 162^
R2 0.988 0.990 0.988 0.990
Adj-R2 0.985 0.986
With R2 0.292 0.372

***p < 0.01; **p < 0.05; *p < 0.1
of the control variables appears clearly associated with sex marriage on society, and on children in particular, at least
state level differences in single-parenting; the differences with currently available data. When comparing states that
that we see in the data appear to be state specific, and were forced to legalize same-sex marriage in 2014 to those
unrelated to the nationalization of same-sex marriage. that already had it, the effect of the policy change is not sig-
nificant.11 This study should temper concerns and assuage
some of the strongest fears of the deleterious effects of
Discussion “redefining marriage” (Hawkins & Carroll, 2015; Louisiana
et al., 2015, p. 13).
These results should help to assuage the concerns expressed As the summary of one-sided t tests against the null hy-
by opponents of the legalization of a right to marriage irre- potheses (Table 5) illustrates, no statistically significant evi-
spective of sex or gender in amici briefs submitted to the dence substantiates the oppositional briefers’ claims of dele-
Supreme Court. Furthermore, this study is in line with and
reaffirms prior research (e.g., Dillender, 2014; Langbein & 11
These findings are robust to replacing rates in the dependent variables with
Yost Jr., 2009; Trandafir, 2014a) that empirical analysis does log of rates. Neither the statistical nor substantive results changed. Data avail-
not support claims about the negative ramifications of same- able upon request.
Sex Res Soc Policy
Table 4 Percent of children in single-parent households and same-sex marriage legalization (2011–2016)
(1) (2) (3) (4) (5) (6)

AR(1) AR(1)
2014 SSM Legalizers − 0.0682 − 0.162 (0.577) 0.289 (0.514) 0.284 (0.583) 0.0695 (0.442) − 0.271 (0.542)
(0.361)
% Unemployed − 0.0306 0.132 (0.140) 0.314 (0.198) 0.329 (0.209) 0.0536 (0.211) 0.240 (0.256)
(0.132)
Median Inc. (1000s) (2016 real dollars) − 0.0403 − 0.0677 − 0.130*** − 0.139*** − 0.0474 − 0.0645
(0.0534) (0.0724) (0.0328) (0.0360) (0.0400) (0.0455)
Population density − 0.00537 − 0.00655 0.000951*** 0.000920*** − 0.00397 − 0.00740
(0.00774) (0.00858) (0.000321) (0.000235) (0.00428) (0.00531)
% Children White as only race − 0.144 0.127 − 0.199*** − 0.323*** −0.230 0.304
(0.447) (1.131) (0.0644) (0.0895) (0.521) (0.640)
% Children Latino (in combination or 0.481 (0.524) 0.610 (0.619) − 0.360* (0.205) − 0.528** (0.231) 0.479 (0.645) 0.721 (0.792)
alone)
% Total population White as only race 0.0265 0.00242 0.0968 (0.0632) 0.0918 (0.0996) 0.0480 − 0.0128
(0.0559) (0.131) (0.0859) (0.0982)
% Total population Latino (in − 0.284 (0.878) 0.661 (1.393) 0.383 (0.251) 0.449 (0.288) − 0.107 (1.15) 1.06 (1.50)
combination or alone)
% Children (age 0–17) of total − 0.310 (0.346) − 0.695 (0.440) − 0.738*** − 1.210*** − 0.645 (0.821) − 0.926 (0.888)
population (0.232) (0.210)
% age 18–64 of total population − 0.00670 − 0.292 (0.535) 0.157 (0.316) − 0.537* (0.322) − 0.0903 − 0.475 (0.618)
(0.496) (0.555)
% Female of total population 2.540 (2.659) 3.637 (3.272) 1.304*** (0.501) 0.125 (0.445) 3.77* (2.10) 4.18* (2.27)
Observations 216 168 216 168 191^ 162^
R2 0.966 0.964 0.965 0.965
Adj-R2 0.955 0.952
2
With R 0.144 0.168

***p < 0.01; **p < 0.05; *p < 0.1
terious effects12 from legalizing same-sex marriage across the by omitting the most reluctant of adopters; (2) incorporating a
American states. Using multiple specifications to estimate the multitude of econometrically rigorous falsification tests, iden-
effect of states that were forced to legalize same-sex marriage tification strategies, and robustness checks (e.g., Correia’s
laws to earlier adopters, we find no differences in family wel- high-dimensional fixed-effects to account for spatio-
fare between these states. Specifically, the late adopters did temporal inter-dependence; Ludwig and Cook’s (2000) iden-
not look any different than the early adopters with respect to tification strategy for analyzing policy impact; and SUR esti-
child poverty, divorce rates, marriages, or the percent of chil- mates to account for dependent variables that only appear to
dren in single-parent households. be unrelated); and, finally, (3) directly addressing arguments
This study contributes to the literature by (1) utilizing a by opponents by focusing on the mechanisms and hypotheses
fuller and more recent panel of “treatment” and “control” that they submitted to the Supreme Court in their own words.
states than was used in prior research (e.g., Langbein & Yost Another significant contribution sheds data-based evidence on
Jr., 2009) and using “comparable” treatment and control states a politically-charged topic such as LGBTQ-rights. To avoid
caricaturing or misrepresenting the arguments of opponents to
12
Opponents prefigured adverse effects of legalized same-sex marriage; there- same-sex marriage, we deliberately highlight key aspects of
fore, we used directionally one-sided tests. their empirically testable assertions. We hope this approach to
Sex Res Soc Policy
Table 5 Summary of one-sided t test P values for effect of legalizing same sex-marriage on each DV
Model specification Tested alternative hypotheses (Ha)
Child poverty will Divorce rate will Marriage rate will Share of children in single-parent households
increase increase decrease will increase
P values on same-sex marriage coefficient from one-sided t test
(1) HDFE—2013 states 0.70 0.45 0.56 0.57

included
(2) HDFE—2013 states 0.62 0.23 0.75 0.61
excluded
(3) XTREG—2013 states 0.70 0.45 0.59 059
included
(4) XTREG—2013 states 0.64 0.24 0.89 0.61
excluded
(5) SUR—2013 states 0.66 0.44 0.66 0.44
included
(6) SUR—2013 states 0.67 0.20 0.87 0.69
excluded
For each regression, the control variables are state-year values for unemployment rate, median income in 2016 real dollars, population density, percent of
children with White as only race, percent of children who are Latino either alone or in combination with other ethnicities and races, percent of the
population who are under age 18, percent of the population age 18 to 64, and percent of the population who are female. Therefore, for each, race referent
group is Black alone, age referent group is percent over age 65, and sex referent group is percent male
***p < 0.01; **p < 0.05; *p < 0.1
defining research by using multiple measures and methods statistical assessments (100 Scholars, 2015; Allen, 2010;
will militate against concerns of both “liberal bias” (Yancey, Dillender, 2014; Langbein & Yost Jr., 2009; Trandafir,
2011) or “political correctness” (Pujol, 2016). 2014a).
While this study contributes to an ongoing, sometimes The second limitation of this research concerns the lack of
fraught and other times tepid, deliberation in American society qualitative data. While the preponderance of research on this
about the appropriate role of sexuality, intimacy, gender, and topic relies solely on quantitative, publicly accessible data,
family in our lives, it does not “end the debate”—nor can any comparative case studies using place-specific qualitative data
study do so. The value of this research is that it provides are crucial for assessing why same-sex marriage legalization
evidence-based scholarship that uses many empirical methods has or does not have expected effects. Many of the arguments
with consistent findings of “no association” to inform the de- about the likely effects of same-sex marriage depend on our
bate. It does not silence the debate: like all others, this study empirical beliefs about how redefining marriage in law would
does have limitations. change mindsets, cultural philosophies, and moral ideas about
The first limitation of the current study is the absence of procreation, family connectedness, and child-rearing (e.g.,
longer-term post-test national data. This study is the first na- 100 Scholars, 2015; Allen & Price, 2015; Marquardt, 2010).
tionwide assessment directly drawing hypotheses from oppo- Rigorous qualitative studies can provide useful information
sitional public-policymakers’ own words to underpin the eval- about what social beliefs and behaviors are responsible for
uation of broad-scale judicial intervention legalizing same-sex shifts in family formation and child-care, and whether differ-
marriage at the state-level in the USA. However, while we ences in these social norms and practices are associated with
focus on outcomes that are likely to be observed shortly after changes in same-sex marriage laws. We note that the use of
adoption of marriage laws, given the relative recency of some state fixed effects controls for cultural and normative differ-
these laws, future research should return to these questions ences between states, but it does not specifically isolate and
with a longer term and broader set of outcome measures. examine their importance.
These could include psychological outcomes (Riggle, Lastly, our data are limited to the US context. While studies
Rostosky, & Horne, 2010) as well as more manifest outcomes from Canada (e.g., Cere & Farrow, 2004), the Netherlands
like economic stability, high school graduation rates, the (Trandafir, 2014a; Trandafir, 2014b), and OECD countries
spread of disease, juvenile delinquency, along with other var- broadly (Trandafir, 2015) are becoming more numerous, stud-
iables mentioned in the introduction. Updating the analyses as ies in non-OECD nations are scarce at best. Legalization of
new data become available is critical. Opponents, advocates, same-sex marriage in Mexico (starting in Mexico City in
and researchers of any stance alike have reiterated the need for 2010), Argentina (2010), Brazil (2013), Colombia (2016),
both continuing and high-quality data acquisition and sound Costa Rica (forthcoming), Israel (if married outside the
Sex Res Soc Policy
country), Taiwan (forthcoming), and Uruguay (2013), or Research Involving Human Data This is an observational study that uses
publicly available aggregate data on humans. Specifically, as noted in the
same-sex registered unions in Aruba (2016), Chile (2015),
abstract of the manuscript, “we use currently available public data from
and Ecuador (2019) offer prime research arenas outside of the U.S. Census and CDC to analyze changes in state-level legalization of
Canada and the USA. Extending studies of the effects of same-sex marriage on rates of child poverty, divorce, marriage, and chil-
same-sex marriage to countries that have different cultures dren living in single-parent households within each state from 2011 to
2016.”
and different trends in marriage rates and child-rearing is es-
sential. Study of the effect of normative and behavioral laws
on normative and behavioral practice regarding families in
countries where different socio-political forces act on the fam-
References
ily as an institution and on the individual as a (potential)
Adams, J., & Light, R. (2015). Scientific consensus, the law, and same
member of the family is vital, since the findings from this sex parenting outcomes. Social Science Research, 53, 300–310.
research may suffer from aggregation bias (Goodfriend, Alabama (2015) http://sblog.s3.amazonaws.com/wp-content/uploads/
1992; Langbein & Lichtman, 1976) and may not be valid 2015/04/14-556_State_of_Alabama.pdf.
Allen, D. W. (2006). An economic assessment of same-sex marriage
for other countries.
laws. Harvard Journal of Law and Public Policy, 29(3), 949–980.
Allen, D. W. (2010). Who should be allowed into the marriage franchise?
Drake Law Review, 58, 1043–1079.
Allen, D. W. (2013). High school graduation rates among children of
same-sex households. Review of Economics of the Household, 11,
Conclusion 635–658.
Allen, D. W., & Price, J. (2015). Same-sex marriage and negative exter-
Contributing to the ongoing debate of issues related to nalities: A critique, replication, and correction of Langbein and
LGBTQ rights on the state and local level in American pol- Yost. Econ Journal Watch, 12(2), 142–160.
Amato, P. (2012). The well-being of children with gay and lesbian par-
itics, using multiple estimation models, this study demon-
ents. Social Science Research, 41(4), 771–774.
strates that assertions that legalizing same-sex marriage American College of Pediatricians, Family Watch International, Marks,
would (1) increase child poverty, (2) increase divorce rates, L. D., Regnerus, M. D., Sullins, D. P. (2015). Amici curiae brief
(3) lower marriage rates, and (4) increase the percent of chil- supporting respondents, Obergefell v. Hodges, 135 S. Ct. 2584.
dren living in single-parent households cannot be supported. Anderson, R. (2013). Marriage: What it is, why it matters, and the con-
sequences of redefining it. Washington DC: The Heritage
We compare states that had legalized same sex marriage Foundation.
before the Supreme Court made it national in 2014 to states Badgett, M., Nezhad, S., Waaldijk, C., & van der Meulen, R. (2014). The
that had to change because of the ruling. Utilizing Correia’s 962 relationship between LGBT inclusion and economic develop-
(2017b) approach to high-dimensional, fixed effects multi- ment: 963 An analysis of emerging economies. Washington DC: US
Agency 964 for International Development & The Williams
way clustering, as well as more traditional panel data analy- Institute.
sis methods (see Cameron & Trivedi, 2010; Wooldridge, Biblarz, T., & Stacey, J. (2010). How does the gender of parents matter?
2010), as well as other estimation procedures, including Journal of Marriage and Family, 72(1), 3–22.
SUR regression, we conclude that, at least with the currently Cameron, A. C., Gelbach, J. B., & Miller, D. L. (2011). Robust inference
with multiway clustering. Journal of Business & Economic
available data, many of the worst fears about the “severe, Statistics, 29(2), 238–249.
unavoidable, and irreversible” (Louisiana et al., 2015, p. Cameron, A. C., & Trivedi, P. K. (2009). Microeconometrics using stata.
13) negative consequences of legalizing same-sex College Station: STATA Press.
marriage—or the idea that it reduces returns for individuals Cameron, A. C., & Trivedi, P. K. (2010). Microeconomics using Stata.
College Station, TX: Stata Press.
to participate in the market of marriage—have been un- Casper, L. M., McLanahan, S. S., & Garfinkel, I. (1994). The gender-
founded. Regardless of one’s stance on whether that was poverty gap: What we can learn from other countries. American
the ethically or morally correct path for the Supreme Court Sociological Review, 59(4), 594–605.
to take American public policy, the absence of statistically Cere, D., & Farrow, D. (2004). Divorcing marriage: Unveiling the dan-
gers in Canada’s new social experiment. Montreal, Ontario: The
significant harmful ramifications of same-sex marriage le- Institute for the Study of Marriage, Law and Culture.
galization should be encouraging. Chen, W. H., & Corak, M. (2008). Child poverty and changes in child
poverty. Demography, 45(3), 537–553.
Compliance with Ethical Standards Cheng, S., & Powell, B. (2015). Measurement, methods, and divergent
patterns: Reassessing the effects of same-sex parents. Social Science
Research, 52, 615–626.
Conflict of Interest The authors have no relevant financial or non- Cherlin, A. J. (2004). The deinstitutionalization of American marriage.
financial interests to disclose. The authors have no conflicts of interest
Journal of Marriage and Family, 66(4), 848–861.
to declare that are relevant to the content of this article. All authors certify
Cherlin, A. J. (2009). The marriage go-round. New York, NY: Alfred A.
that they have no affiliations with or involvement in any organization or
Knopf.
entity with any financial interest or non-financial interest in the subject
Chinni, D. (2004). Is the post-election red tinge a mandate? Don’t bet on
matter or materials discussed in this manuscript. The authors have no
it. Christian Science Monitor. https://www.csmonitor.com/2004/
financial or proprietary interests in any material discussed in this article.
1104/p09s01-codc.html.
Sex Res Soc Policy
Cohen, L. R., & Wright, J. D. (2011). Research handbook on the eco- Langbein, L., & Yost Jr., M. A. (2009). Same-sex marriage and negative
nomics of family law. Northampton, MA: Edward Elgar. externalities. Social Science Quarterly, 90(2), 292–308.
Correia, S. (2017a). REGHDFE: Stata module for linear and instrumen- Lerman, R. I. (1996). The impact of the changing U.S. family structure on
tal-variable/GMM regression absorbing multiple levels of fixed ef- child poverty and income inequality. Economica, 63(250), S119–
fects. Boston, MA: Statistical software components, Boston College S139.
Department of Economics. https://ideas.repec.org/c/boc/bocode/ Lewis, G. B. (2005). Same-sex marriage and the 2004 presidential elec-
s457874.html. tion. PS: Political Science and Politics, 38(2), 195–199.
Correia, S. (2017b). A feasible estimator for linear models with multi-way Lichter, D. T., & Crowley, M. L. (2004). Welfare reform and child pov-
fixed effects. (working paper). http://scorreia.com/research/hdfe. erty: Effects of maternal employment, marriage, and cohabitation.
pdf. Social Science Research, 33(3), 385–408.
Dao, J. (2004, November). 4. New York Times: Same-sex marriage issue Louisiana, Utah, Texas, Alaska, Arizona, Arkansas, Georgia, Idaho,
key to some GOP races https://www.nytimes.com/2004/11/04/ Kansas, Montana, Nebraska, North Dakota, Oklahoma, South
politics/campaign/samesex-marriage-issue-key-to-some-gop-races. Dakota, and West Virginia. (2015). Amici curiae brief supporting
html. respondents, Obergefell v. Hodges, 135 S. Ct. 2584.
Dillender, M. (2014). The death of marriage? The effects of new forms of Ludwig, J., & Cook, P. J. (2000). Homicide and suicide rates associated
legal recognition on marriage rates in the United States. with implementation of the Brady handgun violence prevention act.
Demography, 51, 563–585. Journal of the American Medical Association, 284(5), 585–591.
Dinno, A. (2014). Comment on “The effect of same-sex marriage laws on Manning, W. D., Feltro, M. N., & Lamidi, E. (2014). Child well-being in
different-sex marriage: Evidence from the Netherlands.”. same-sex parent families: Review of research prepared for American
Demography, 51, 2343–2347. Sociological Association amicus brief. Population Research &
Ellis, B. J., Bates, J. E., Dodge, K. A., Fergusson, D. M., Horwood, L. J., Policy Review, 33(4), 485–502.
Pettit, G. S., & Woodward, L. (2003). Does father absence place Manning, W. D., & Lamb, K. A. (2003). Adolescent well-being in co-
daughters at special risk for early sexual activity and teenage preg- habiting, married, and single-parent families. Journal of Marriage
nancy? Child Development, 74(3), 801–821. and Family, 65, 876–893.
Family Equality Council. (2018). Joint Adoption Law. https://www. Marks, L. (2012). Same-sex parenting and children’s outcomes: A closer
familyequality.org/get_informed /resources/equality_maps/joint_ examination of the American Psychological Association’s brief on
adoption_laws/. lesbian and gay parenting. Social Science Research, 41(4), 735–751.
Family Research Council. (2015). Amicus curiae brief supporting respon- Marquardt, E. (2006). Between two worlds: The inner lives of children of
dents and affirmance, Obergefell v. Hodges, 135 S. Ct. 2584. divorce. New York, NY: Three Rivers Press.
Felter, C., & Renwick, D. (2019). Same-sex marriage: Global compari- Marquardt, E. (2010). The kids are not all right. Human Life Review,
sons. Council on Foreign Relations. https://www.cfr.org/ 36(3), 113–115.
backgrounder/same-sex-marriage-global-comparisons. Marquardt, E., Glenn, N. D., & Clark, K. (2010). My daddy’s name is
Flouri, E. (2007). Fathering and adolescents’ psychological adjustment: donor: A new study of young adults conceived through sperm do-
The role of fathers’ involvement, residence and biology status. nation. New York, NY: Institute for American Values. http://
Child: Care, Health and Development, 34(2), 152–161. americanvalues.org/catalog/pdfs/Donor_FINAL.pdf
Flouri, E., & Buchanan, A. (2003). The role of father involvement in Masci, D., & DeSilver, D. (2019). A global snapshot of same-sex mar-
children’s later mental health. Journal of Adolescence, 26, 63–78. riage. Pew Research Center. https://www.pewresearch.org/fact-
Frank, N., Weeden, K., & Baker, K. (2018). What does the scholarly tank/2019/06/21/global-snapshot-same-sex-marriage/.
research say about the well-being of children with gay or lesbian Masci, D., Sciupac, E., & Lipka, M. (2019). Gay marriage around the
parents? Ithaca, NY: Center for Study of Inequality. https:// world. Pew Research Center: Religion and Public Life. https://www.
whatweknow.inequality.cornell.edu/topics/lgbt-equality/what-does- pewforum.org/fact-sheet/gay-marriage-around-the-world/.
the-scholarly-research-say-about-the-wellbeing-of-children-with- McLanahan, S., & Sawhill, I. (2015). Marriage and child wellbeing
gay-or-lesbian-parents/. revisited: Introducing the issue. The Future of Children, 25(2), 3–9.
Gallup (2019). Support for gay marriage stable. https://news.gallup.com/ Meezan, W., & Rauch, J. (2005). Gay marriage, same-sex parenting, and
poll/257705/suupport-gay-marriage-stable.aspx.?version=print. America's children. The Future of Children, 15(2), 97–115.
Giddens, A. (1992). The transformation of intimacy: Sexuality, love, and Members of the Kentucky General Assembly. (2015). Amici curiae brief
eroticism in modern societies. Stanford, CA: Stanford University supporting respondents, Obergefell v. Hodges, 135 S. Ct. 2584.
Press. Murray, C. (2013). Coming apart: The state of white America, 1960–
Goings, R. B., & Ford, D. Y. (2017). Investigating the intersection of 2010. New York, NY: Crown Forum.
poverty and race in gifted education journals: A 15-year analysis. Obergefell v. Hodges, 576 (US Supreme Court, 2015). https://www.
Gifted Child Quarterly, 62(1), 25–36. supremecourt.gov/opinions/14pdf/14-556_3204.pdf.
Goodfriend, M. (1992). Information-aggregation bias. The American Otter, C. L. (2015). Amici curiae brief, Obergefell v. Hodges, 135 S. Ct.
Economic Review, 82(3), 508–519. 2584.
Harper, C. C., & McLanahan, S. S. (2004). Father absence and youth Posner, R. A. (1973). Economic analysis of law. New York, NY: Little,
incarceration. Journal of Research on Adolescence, 14(3), 369–397. Brown and Co..
Hawkins, A. J., & Carroll, J. S. (2015). Beyond the expansion framework: Potter, D. (2012). Same-sex parent families and children’s academic
How same-sex marriage changes the institutional meaning of mar- achievement. Journal of Marriage and Family, 74, 556–571.
riage and heterosexual men’s conception of marriage. Ave Maria Public Religion Research Institute. (n.d.) The American Values Atlas.
Law Review, 13(2), 219–235. http://ava.prri.org/home#lgbt/2017/states/lgbt_ssm.
Herek, G. (2006). Legal recognition of same-sex relationships in the Pujol, J. (2016). The United States safe space campus controversy and the
United States: A social science perspective. American paradox of freedom of speech. Church, Communication and
Psychologist, 61(6), 607–621. Culture, 1(1), 240–254.
Langbein, L., & Lichtman, A. (1976). Across the great divide: Inferring Regnerus, M. (2012). How different are the adult children of parents who
individual level behavior from aggregate data. Political have same-sex relationships? Findings from the new family struc-
Methodology, 3(4), 411–439. tures study. Social Science Research, 41(4), 752–770.
Sex Res Soc Policy
Republican National Convention Committee. (2015). Amicus curiae brief Trandafir, M. (2015). Legal recognition of same-sex couples and family
supporting respondents, Obergefell v. Hodges, 135 S. Ct. 2584. formation. Demography, 52, 113–151.
Ribar, D. (2015). Why marriage matters for child wellbeing. The Future United States v. Windsor, 570 US 744 (2013).
of Children, 25(2), 11–27. Voss, P. R., Long, D. D., Hammer, R. B., & Friedman, S. (2006). County
Riggle, E., Rostosky, S., & Horne, S. (2010). Psychological distress, child poverty rates in the US: A spatial regression approach.
well-being, and legal recognition in same-sex couple relationships. Population Research and Policy Review, 25(4), 369–391.
Journal of Family Psychology, 24(1), 82–86. Wardle, L. D. (2001). Multiply and replenish: Considering same-sex
Romaniuc, R. (2016). What makes law change behavior? An experimen- marriage in light of state interests in marital procreation. Harvard
tal study. Review of Law & Economics, 12(2), 447–475. Journal of Law and Public Policy, 24(3), 771–814.
Sawhill, I., & Haskins, R. (2016). The decline of the American family: Wardle, L. D. (2007). The fall of marital family stability and the rise of
Can anything be done to stop the damage? The Annals of the juvenile delinquency. Journal of Law & Family Studies, 10(2), 83–
American Academy of Political and Social Science, 667(1), 8–34. 110.
Sawhill, I., & Thomas, A. (2005). For love and money? The impact of Wax, A. L. (2007). Engines of inequality: Class, race, and family struc-
family structure on family income. The Future of Children, 15(2), ture. Family Law Quarterly, 41(3), 567–599.
57–74. Wax, A. L. (2009). Review: The family law doctrine of equivalence.
Scholars. (2015). Amici curiae brief supporting respondents and affir- Michigan Law Review, 107(6), 999–1017.
mance, Obergefell v. Hodges, 135 S. Ct. 2584. Wooldridge, J. M. (2010). Econometric analysis of cross-section and
Scholars of Marriage. (2015). Amici curiae brief supporting respondents, panel data. Boston, MA: MIT Press.
Obergefell v. Hodges, 135 S. Ct. 2584. Yancey, G. (2011). Compromising scholarship: Religious and political
Texas Values. (2015). Amicus curiae supporting respondents, Obergefell bias in American higher education. Waco, TX: Baylor University
v. Hodges, 135 S. Ct. 2584. Press.
Thomas, A., & Sawhill, I. (2002). For richer or for poorer: Marriage as an Young, K. K., & Nathanson, P. (2007). Redefining marriage or
anti-poverty strategy. Journal for Policy Analysis and Management, deconstructing society: A Canadian case study. Journal of Family
21(4), 587–599. Studies, 13, 133–178.
Trandafir, M. (2014a). The effect of same-sex marriage laws on different- Zissimopoulos, J. M., Karney, B. R., & Rauer, A. J. (2015). Marriage and
sex marriage: Evidence from the Netherlands. Demography, 51, economic well-being at older ages. Review of the Economics of the
317–340. Household, 13, 1–35.
Trandafir, M. (2014b). Reply to comment on “The Effect of Same-Sex
Marriage Laws on Different-Sex Marriage: Evidence from the Publisher’s Note Springer Nature remains neutral with regard to jurisdic-
Netherlands”. Demography, 51, 2349–2350. tional claims in published maps and institutional affiliations.

Anti Social Effect

Uploaded by

Copyright:

Available Formats

Anti Social Effect

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Anti Social Effect

Uploaded by

Copyright:

Available Formats

Sexuality Research and Social Policy

The Anti-Social Effects of Legalizing Same-Sex Marriage: Fact

Accepted: 27 October 2020

Fig. 1 Time trends of children

Fig. 2 Time trends of divorce

Fig. 3 Time trends of marriage

Fig. 4 Time trends of share of

Table 1 Percent of children in poverty and same-sex marriage legalization (2011–2016)

(1) (2) (3) (4) (5) (6)

Robust standard errors in parentheses

Table 2 Divorces per 1000 and same-sex marriage legalization (2011–2016)

(1) (2) (3) (4) (5) (6)

Robust standard errors in parentheses

Table 3 Marriages per 1000 and same-sex marriage legalization (2011–2016)

(1) (2) (3) (4) (5) (6)

Robust standard errors in parentheses

(1) (2) (3) (4) (5) (6)

Robust standard errors in parentheses

Model specification Tested alternative hypotheses (Ha)

(1) HDFE—2013 states 0.70 0.45 0.56 0.57

You might also like

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.