What Effects Would Redefining Marriage Likely Have On Children Of Heterosexuals? Testing Competing Hypotheses With European Data re: Marriage, CBOW & Abortions

Social scientists and commentators have offered opposing hypotheses about the likely effects on children of redefining marriage in genderless terms.  Some advocates of the redefinition have postulated that, by “expanding” marriage to include same-sex couples, such a redefinition would strengthen the overall marriage culture and thereby increase the percentage of children raised by two parents in stable households.  This hypothesis is an important component of what is sometimes called the “conservative case” for same-sex marriage.[1]


How might these competing hypotheses be tested against real-world data?  One problem, as Justice Alito noted in his opinion in United States v. Windsor, is that genderless marriage is still too new – especially in the United States – to permit robust testing using rigorous social-science methods.[5]  However, in Europe, six nations have had genderless marriage regimes long enough to permit at least preliminary testing and evaluation:  Norway, Sweden, Iceland, and the Netherlands, which adopted same-sex civil union regimes that were the practical equivalent of marriage in 1993, 1995, 1996, and 1998, respectively, and subsequently adopted genderless marriage expressly; and Belgium and Spain, which expressly adopted genderless marriage regimes in 2003 and 2005, respectively.[6]  In three separate respects, data from these nations undermine the “Conservative Case” hypothesis and support the Allen-Hawkins-Carroll hypothesis.


Marriage rates.  The most obvious implication of the Conservative Case hypothesis is an expected increase in marriage rates generally—not only among homosexuals, but also among progressive heterosexuals, for whom the stigma of marital inequality would be lifted.  On the other hand, the most obvious implication of the Allen-Hawkins-Carroll hypothesis is an expected decline in marriage rates among heterosexuals.  Unfortunately, these six nations generally do not report marriage rates separately for man-woman and same-sex couples.[7]  But given that heterosexuals make up the vast majority of the population, a decline in marriage rates among heterosexuals should—in equilibrium and over time—show up as a decline in overall marriage rates.


The data strongly suggest that marriage rates among heterosexuals have indeed declined in most of these six nations since their adoption of genderless marriage (or equivalent) regimes.  The following table shows the changes in marriage rates in each of these nations from their adoption of genderless marriage until 2010, the last year for which data are consistently available, and compares those changes with the changes in European nations generally over similar periods:


Table 1:  Marriage Rates In European Nations

Adopting Same-Sex Marriage (Or Practical Equivalents) Before 2006[8]


Nation Year adopted SSM or equivalent Marriage rate in prior year[9] Marriage rate, 2010 Percent change
All EU-27 nations   5.1 (1998) 4.4 -13.7%
    4.9 (2004)   -10.2%
 Norway 1993 (2009)[10] 4.3 4.8 11.6%
 Sweden 1995 (2009) 4.6 5.3 15.2%
 Iceland 1996 (2010) 5.6 4.9 -12.5%
 Netherlands 1998 (2001) 5.5 4.5 -18.1%
 Belgium 2003 3.9 3.9 0
 Spain 2005 5.1 3.6 -29.4%


As the table shows, marriage rates have generally declined in Europe during this period—by over 13 percent between 1998 and 2010, and over 10 percent from 2004 to 2010.  But the decline in one of these nations–Spain at 29.4% over the period from 2004 to 2010–was significantly in excess of the overall European decline of 10.2 percent over that same period.  And that is obviously consistent with the Allen-Hawkins-Carroll prediction that the adoption of genderless marriage leads some percentage of the heterosexual male population to lose interest in marriage altogether.


The data for two other countries—Iceland, which saw significant but slightly smaller declines in marriage rates than the European average over a similar period, and Belgium, which saw no change—are also consistent with this prediction.  That is because, all else being equal, the advent of officially sanctioned same-sex marriage could be expected to cause a small but temporary increase in overall marriage rates because of pent-up demand for marriage by same-sex couples.  If that expectation is correct—as same-sex marriage advocates claim[11]—then it appears that marriages involving heterosexual men were also probably declining in Iceland at a faster clip than the overall decline in European marriage rates.  That could also be true in Belgium—depending on the relative percentage of same-sex couples reflected in the marriage statistics—although the data do not conclusively show that.


Thus, by comparison to the rest of Europe, marriage rates among heterosexuals appear to have declined more rapidly than one would expect in at least three of the six European nations that were “early movers” in enacting same-sex marriage or its functional equivalent.[12]  That correlation tends to support the Allen-Hawkins-Carroll hypothesis.  It also contradicts the Conservative Case hypothesis, which would predict increased overall marriage rates—something that clearly did not occur in four of the six nations.


By contrast, the Swedish and Norway experience during this period appears at first blush to support the Conservative Case hypothesis.  But the fact that the experience of four of these six nations is inconsistent with that hypothesis suggests that some other exogenous—and possible transitory—factor explains the increase in Sweden’s and Norway’s marriage rates.


In sum, though it is far from conclusive, the European marriage data provide some support for the Allen-Hawkins-Carroll hypothesis, while generally undermining the Conservative Case Hypothesis.


Out-of-wedlock births.  Both hypotheses predict additional changes in measures related to procreation.  Specifically, with a decrease in males’ interest in marriage (but no decrease in their interest in sex), the Allen-Hawkins-Carroll hypothesis would predict an increase in the percentage of out-of-wedlock births, an increase in the percentage of pregnancies that end in abortion, or some combination of the two.  That is because unmarried women who become pregnant as a result of sexual relations with men who lack an interest in marriage would increasingly face the choice between completing the pregnancy and bringing the baby into an unmarried household, or obtaining an abortion.  By contrast, the Conservative Case hypothesis would predict a reduction in both out-of-wedlock births and abortions, as the “expansion” of marriage (under that hypothesis) would both legitimize the institution and reinforce its pro-child norms.  Here again, data for the six European nations that first adopted genderless marriage generally confirm the Allen-Hawkins-Carroll hypothesis and decisively refute the Conservative Case hypothesis.


The chart below, for example, shows the percentage of live births to unmarried mothers:


Table 2:  Out-of-Wedlock Births In European Nations

Adopting Same-Sex Marriage (Or Practical Equivalents) Before 2006[13]


Nation Year adopted SSM or equivalent Out-of-wedlock birth % in prior year Out-of-wedlock birth %, 2010 Percent change
All EU-27 nations   27.3 (2000) 38.3 40.2%
    29.1 (2003)   31.61%
 Norway 1993 49.0[14] 54.8 32.2%
 Sweden 1995 54.7 54.2 -1.0%
 Iceland 1996 64.0 64.3 0.5%
 Netherlands 1998 20.8 44.3 89.0%
 Belgium 2003 28.0[15] 46.2 65.0%
 Spain 2005 25.1 35.5 41.4%


As the chart shows, four of the six nations saw substantial increases in the percentage of children born to unmarried mothers.  In one case—Norway—the increase (32.2 percent) was less than the overall percentage increase in all of Europe during a similar period.  But that is offset by the fact that at 49 percent, Norway’s initial unwed birth percentage was already about twicethe overall European percentage—indicating that Norway had far less room for an increase than other nations.  And clearly, Norway’s adoption of same-sex marriage did nothing to reverse its high percentage of out-of-wedlock births—contrary to the “Conservative Case” hypothesis.


More troubling, three of the six nations saw increases in out-of-wedlock births substantially in excess of the European average:  The Netherlands saw an increase in unwed births of 89 percent, more than double the European average increase during the same period of 40 percent.  Belgium also saw an increase of 65 percent.  And in just five years, Spain saw a jump of more than 41 percent, as compared with a Europe-wide increase of about 32 percent during the period from 2003 to 2010.  All of these nations started out at levels comparable to or below the overall European average.  These nations’ experience therefore strongly confirms the Allen-Hawkins-Carroll hypothesis and contradicts the “conservative case” hypothesis.


In the two remaining nations—Sweden and Iceland—the percentage of unwed births remained essentially unchanged.  But as we will see, during these years both nations also saw substantial increases in their abortion rates, indicating that many expectant mothers chose abortion rather than bringing a baby into an unmarried household.  As with Norway, moreover, the initial unwed birth percentage in both nations was already substantially in excess of the European average—more than twice as high in Iceland’s case.  If the Conservative Case hypothesis were correct, we would expect to see at least some significant decline in the percentage of unwed births in these nations—especially in light of the substantial period during which same-sex marriage has been available.  But there has been no such decline, and that fact once again refutes the Conservative Case hypothesis.


In short, European statistics on unwed births strongly refute the Conservative Case hypothesis, while supporting (albeit less strongly) the Allen-Hawkins-Carroll hypothesis.


  1.   That conclusion is confirmed by statistics on abortions in the six “early mover” European nations.  Given that abortion is an obvious practical alternative to an unwed birth, it is important to examine both sets of statistics together.  Moreover, if the “Conservative Case” hypothesis is correct, we would expect to see a reduction in abortions (which have generally been decreasing worldwide in any event).  That is because, under that hypothesis, allowing same-sex couples to marry will reinforce the value of marriage and the pro-child norms that it embraces—thereby leading more unmarried heterosexual couples to choose marriage rather than abortion when faced with an unexpected pregnancy.


By contrast, if the Allen-Hawkins-Carroll hypothesis is correct, we would expect to see an increase in abortions as a result of the alienation of heterosexual men from marriage.  That alienation would naturally lead to a relative increase in the number of unmarried but pregnant women.  And because unmarried pregnant women are much more likely than married pregnant women to obtain abortions,[16] a relative increase in the former will naturally lead to higher abortion rates.  That tendency would also be reinforced if, as suggested by Allen-Hawkins-Carroll and others, the adoption of genderless marriage necessarily changes the public meaning or perception of marriage from an institution principally concerned with procreation and children to one that is principally concerned with the well-being of adults.[17]  In most societies, marriage is the only social institution largely dedicated to children, and its high status stands as a constant reminder to society that the interests of children should take precedence over the interests of adults.  But a society that redefines marriage to accommodate the romantic interests of a small subset of the adult population necessarily conveys to its members that adult interests can appropriately trump the interests of children.  That message will tend to legitimize decisions by non-married and married citizens to place their own interests above the interests of their children – including their unborn children.  And that, in turn, will tend to increase the abortion rate.


Consistent with the Allen-Hawkins-Carroll hypothesis, moreover, among all six nations there appears to have been a substantial increase in abortion, as measured by the percentage of all pregnancies:


Table 3:  Abortion Percentages In European Nations

Adopting Same-Sex Marriage (Or Practical Equivalents) Before 2006[18]


Nation Year adopted SSM or equivalent Abortion % in prior year Abortion %


Percent change
All EU-27 Nations   27.6 (1998) 19.1 -30.1%
    23.2 (2004)   -17.7%
 Norway 1993 20.1 20.3 1.0%
 Sweden 1995 22.4 25.2 12.5%
 Iceland 1996 15.9 17.8 11.9%
 Netherlands 1998 10.6 13.4 26.4%
 Belgium 2003 12.4 13.4 8.0%
 Spain 2005 15.8 18.8 19.0%

As the chart shows, since 1998, abortion percentages in Europe generally have declined dramatically—by some 30 percent from 1998-2011 and nearly 18 percent from 2004-2011.   These results are confirmed by a 2012 joint study by the Guttmacher Institute and the World Health Organization showing that overall abortion rates (the number of abortions per 1000 women of child-bearing age) in the developed world have consistently declined since 1995 (up to 2008, the last year analyzed by the study).[19]  Specifically, in developed countries other than Eastern Europe (where abortion rates have been higher), between 1995 and 2008 the average abortion rate declined by about 15 percent.


But at the same time, all but one of the six “early mover” nations saw a substantial increase in the abortion percentage (as well as abortion rates).  Spain’s progression is especially remarkable:  Over the 2004-2011 period, it saw an increase of 19 percent.  Over a somewhat longer period (1997-2011), the Netherlands saw an even larger increase of more than 26 percent.  The average change in the abortion percentage for the entire group was 11.3 percent.


For Sweden and Iceland, moreover, the sharp increase in abortions explains why the percentage of live births to unmarried mothers did not increase significantly:  More pregnant women simply chose abortion rather than bringing a new baby into an unmarried household.  And for Norway, which saw only a modest increase in its abortion percentage, it appears that a higher percentage of unmarried pregnant women chose the latter course rather than choosing an abortion.


Especially in light of the Europe-wide trend, the fact that abortion rates in all six of the “early mover” European nations increased rather than declined contravenes the “Conservative Case” hypothesis and supports the Allen-Hawkins-Carroll hypothesis.


Conclusion.  All of these results—for marriage rates, out-of-wedlock births and abortions—are summarized in the following table:


Table 4:  Do the European Data As A Whole Support The “Conservative Case”

Hypothesis (“CC”) Or The Allen-Hawkins-Carroll Hypothesis (“AHC”)?


Nation Marriage rate Out-of-Wedlock Birth Percentage Abortion Percentage Overall conclusion
 Norway CC Neutral AHC Neutral
 Netherlands AHC AHC AHC AHC
 Belgium Neutral AHC AHC AHC
Overall conclusion AHC AHC AHC AHC


As the table shows, some of the individual data points on marriage rates and out-of wedlock births either support the “Conservative Case” (“CC”) hypothesis (Sweden’s marriage rates) or are neutral as between the two hypotheses (e.g., Belgium’s marriage rates).  But when taken as a whole, each of those data series—as well as all the data on abortions—supports the Allen-Hawkins-Carroll (“AHC”) hypothesis.


Except for Norway, the same is true when one examines the three data items for each nation.  Even the data for Sweden, whose marriage rate history alone seems to support the “Conservative Case” hypothesis, on the whole support the Allen-Hawkins-Carroll hypothesis because the data on out-of-wedlock births and abortions are consistent with that hypothesis and not the Conservative Case hypothesis.[20]


In short, the early European data as a whole appear to contradict the “Conservative Case” hypothesis and support the conclusion by Allen, Hawkins and Carroll that the adoption of a genderless marriage regime tends to alienate substantial numbers of heterosexuals from marriage—much to the detriment of their children.

-Posted by Lynn D. Wardle, 27 June 2014


[1] See, e.g., Jonathan Rauch, “Conservative Case for Gay Marriage,” USA Today, May 16, 2013, available at USTToday.com; A. Barton Hinkle, “A Conservative Case for Gay Marriage,” Reason, October 7, 2013, available at reason.com; Theodore B. Olson, “The Conservative Case for Gay Marriage,” Newsweek, January 8, 2011, available at Newsweek.com; James Kirchick, “The Conservative Case for Gay Marriage,” New York Daily News, March 5, 2013; [others?]

[2] E.g., Douglas Allen, “An Economic Assessment of Same-Sex Marriage” 29 Harvard Journal of Law and Public Policy 949-980 (2006); Douglas Allen, “Who Should Be Allowed Into the Marriage Franchise?” 58 Drake Law Review 1043-75 (2010).

[3] See, e.g., [E.g., Hawkins-Carroll amicus brief in Latta v. Otter (9th Circuit, June 26, 2009), or Allen-Hawkins-Carroll article]

[4] See, e.g., Girgis, Anderson, & George, What is Marriage? Man and Woman: A Defense, at 23-28 (2012).

[5] United States v. Windsor, 133 S.Ct. 2675, 2715-16 (Alito, J., dissenting).

[6] Denmark adopted a registered partnership arrangement for same-sex couples in 1989.  But as to adoption and other significant matters, and unlike the arrangements in Norway and Sweden, Denmark’s registered partnership arrangement did not give same-sex couples the same rights as married couples.  That did not occur until Denmark expressly redefined marriage in 2012.

[7] The Netherlands is the only apparent exception.

[8] Sources:  Statistics on marriage rates for 2010 are available on the Eurostat website under “marriage and divorce statistics; other marriage rates available from Eurostat at epp.eurostat.ec.europa.eu/tgm.

[9] The earliest year for which data are consistently available is 1998, so that year is used for nations that adopted same-sex-marriage-equivalent regimes before then.

[10] For Norway, Sweden and Iceland, the year in parentheses is the year in which marriage was formally redefined in genderless terms, after having been effectively redefined previously because of a marriage-equivalent civil union or registered partnership regime.

[11] See, e.g., Chris Kirk & Hanna Rosin, “Does Gay Marriage Destroy Marriage?, Slate, May 23, 2012 (“Start with Massachusetts, which endorsed gay marriage in May 2004. That year, the state saw a 16 percent increase in marriage. The reason is, obviously, that gay couples who had been waiting for years to get married were finally able to tie the knot. In the years that followed, the marriage rate normalized but remained higher than it was in the years preceding the legalization. So all in all, there’s no reason to worry that gay marriage is destroying  marriage in Massachusetts.”).

[12] Mircea Trandafir finds that the adoption of a same-sex marriage equivalent regime in the Netherlands had no statistically significant effect on overall marriage rates in that nation.  Trandafir, “The Effects of Same-Sex Marriage Laws on Different-Sex Marriage:  Evidence from the Netherlands,” 15 Demography 317-341 (2014).  However, he finds that the adoption of that regime—and the subsequent express adoption of genderless marriage in 2001—did have a statistically significant impact on two major Dutch population groups:  (a) those living in large urban areas and (b) those without strong religious affiliations.  Id. at 336-37.  This conclusion likewise supports the Allen-Hawkins-Carroll hypothesis, given that men in these two groups, who would typically have weaker ties to groups that would likely reaffirm the importance of marriage, would likely be more susceptible to the social messages created by the redefinition of marriage.

[13] Source:  Eurostat

[14] The earliest numbers for Norway, Sweden and Iceland were from 1998, so that is what is used here.

[15] Data are unavailable for Belgium in 2001 and 2002, so the number for 2000 was used instead.

[16] See, e.g., National Center for Health Statistics, Data Brief No. 136 (December 2013), available at www.cdc.gov/nchs/data/databriefs/db136.pdf (in the U.S., the abortion rate for unmarried women is “almost five times higher than for married women”).

[17] See, e.g., Girgis, Anderson, & George, What is Marriage? Man and Woman: A Defense, at 23-28 (2012).

[18] Source:  Wm. Robert Johnston, Abortion Statistics and Other Data, last updated 14 April 2014, www.johnstonsarchive.net.

[19] See Guttmacher Institute, “Facts on Induced Abortion Worldwide,” January 2012, available at www.guttmacher.org/pubs/fb_IAW.html.

[20] The same is true for the Netherlands even under Trandafir’s conclusion that the advent of a same-sex marriage-equivalent regime there had no statistically significant effect on overall marriage rates.  Even if that is true, the Netherlands saw such remarkable increases in both unwed births and abortions that the data for that nation overall support the Allen-Hawkins-Carroll hypothesis.  The same is true of the data on marriage rates overall:  Even if the Netherlands data were treated as neutral as between the Allen-Hawkins-Carroll and the Conservative Case hypotheses, the data overall would still favor the former.  And of course, as noted previously, Trandafir also concluded that the Netherlands’ same-sex marriage regime caused a decline in marriage rates among the more urban and less religious portions of the nation’s population.

Lynn Wardle