More Heat Than Light: A Critical Assessment of the Same-Sex Parenting Literature, 1995–2013

Abstract

Same-sex marriage is one of the great policy issues of our time, and part of this debate hinges on the ability of same-sex couples to parent. Most gay parenting studies conclude that children raised by gay parents perform as well, if not better, than their counterparts in heterosexual families. This conclusion, which may or may not be true, is not scientifically warranted because of various limitations: Some results are misreported; most of the literature is exploratory and made up of small qualitative samples, biased data, and other research design failures; the studies concentrate on lesbian families; and outcome measures have been limited. Although these problems prevent scientific generalizations, social scientists have treated the preliminary, nonconclusive research as authoritative. Quite naturally, those within public policy circles have adopted this unwarranted position. Regardless of what science ultimately demonstrates about same-sex family structure, it is important to safeguard the research process from political pressures: either anti-gay marriage or pro-gay rights.

KEYWORDS:

• same sex parenting
• child outcomes

Notes

A note on nomenclature. The nonheterosexual world is varied, loosely defined, fluid, and evolving in terms of labels for various sexual orientations. This makes it cumbersome to find a word(s) to describe this community as a class of people. The word “homosexual” might seem to fit as an umbrella term, but it is seldom used within the literature and no doubt carries a pejorative tone. Some moniker's are long and awkward and often over-/under-inclusive for a specific context. Here I use the term “same-sex” to describe the entire class of individuals who fall outside the heterosexual norm. When speaking of a specific class I use “lesbian”, “bisexual”, and so on.

Andersson et al. (2006, p. 81) noted: “The lack of representative samples is the most fundamental problem in quantitative studies on gays and lesbians, which commonly rely on self-recruited samples from an unknown population.”Sweet's (2009) survey finds that (1) less than 20 studies met her scientific selection criteria, (2) in all cases the tests had limited power and tended to accept the null of no difference between same-sex and opposite-sex parenting, (3) most studies were not done in the United States, and (4) all studies had very small sample sizes, especially when broken down by gender. Stacey and Biblarz (2001) are more pessimistic and note that few studies grapple with these issues (pp. 164–166).

Two others used samples drawn from a population: Rothblum et al. (2008) and Andersson et al. (2008).

For example, Balsam et al. (2008) and Golombok et al. (2003).

This indirect source of bias is a problem in many large data sets, including the U.S. Census. In these cases, same-sex couples are usually identified by tagging respondents who claim they are married or common law and living with someone of the same sex. On the one hand, this is likely to oversample same-sex households because it picks up many nonsexual same-sex relations (e.g., brothers, cousins, roommates, etc.) and because it relies on respondents to accurately answer a combination of questions. The Williams Institute (2011) completed a study of the 2010 census and noted that the bias works both ways: same-sex couples get coded as opposite-sex couples, and opposite-sex couples get coded as same-sex couples:[T]he total national error rate is approximately 0.25% ((0.2%*0.74)+(0.4%*0.26)) …. This measurement problem means that Census tabulations of same-sex couples may be biased too high, yielding an over-count. However, there is also reason to believe that Census procedures can undercount same-sex couples. Gates (2010) estimates that at least 15% of same-sex couples are not counted in Census Bureau tabulations either because they identified themselves as something other than “husband/wife” or “unmarried partner” or neither partner was Person 1 in the household. Some same-sex couples are unwilling to identify themselves as such on the Census due to concerns about confidentiality. Same-sex couples may experience stigma and discrimination and consider it too risky to identify as spouses or unmarried partners on a government survey like the Census. Instead, they may choose to call themselves roommates or unrelated adults. Couples where neither partner is Person 1 cannot be identified on the Census since identification relies upon knowing the relationship between Person 1 and others in the household …. O'Connell and Gooding 2007 found evidence that the over-count and undercount may effectively offset each other. If this is true, then Census same-sex couple tabulations may be a fairly accurate. However, the presence of the miscoded different-sex couples within the identified same-sex couples could seriously bias the reported rates of male versus female couples, child-rearing, and the overall geographic distribution of same-sex couples across a state.

Rivers et al. (2008) uses a random British survey similar to that used by Wainright and Patterson and ends up with a sample of 18 lesbian households.

The fraction of lesbians and gays within the population is a number in some dispute, and a number that depends on how lesbians and gays are defined. Allen and Lu (2013) found that lesbians make up only one-third of 1% of the Canadian population. This estimate comes from the Canadian Community Health Survey, the only large, random, nationally represented data set, I know of, that directly identifies sexual orientation (including bisexuality) for both singles and couples. Reliable estimates of lesbian numbers in the United Staes are typically less than 2% of the population (Black et al., 2007, p. 54).

The Canada Census does not separate married or common-law same-sex couples. When children of same-sex couples are compared with common-law opposite-sex couples, they perform slightly worse. In Canada, common-law relationships have the same legal status as marriage, unlike in most states of America.

Of the 60 studies examined here, only 22 dealt with gay male parents (15 of these since 2008). This is another source of bias that warrants caution in drawing any conclusions about nonlesbian families.

According to Nock (2001, p. 37), to properly test any hypothesis regarding same-sex parenting, a sample size of 800 is required. Most would reject Nock's claim, but given the complexity of same-sex family structure and formation, sample sizes greater than 300 are likely necessary to generate any power.

Sweden adopted same-sex marriage in 2009, and so Andersson et al. actually are comparing divorce hazards between different types of unions.

Nor is it not impressive in Canada where same-sex marriage has been legal since 2005. According to Allen and Lu (2013) only 12.2% of lesbians and 4.9% of gays are married.

There is one other study that is particularly well done. Sarantakos (1996) does not use a random sample but rather draws on a longitudinal study and uses objective, verifiable, hard measures of performance that are not self-reported. He finds a “difference” as well, with children from gay households doing significantly worse in mathematics, language, and other school-related matters. Interestingly, this study is not mentioned in most literature surveys. However, Schumm (2012) examined all of Sarantakos work on nontraditional families and questions why it has been dismissed and ignored.

Goldberg (2010, p. 110) cites the 1993 Virginia case of Bottoms v. Bottoms as a “catalyst” for the literature. In the case, the court awarded custody of Sharon Bottom's son to her mother, Kay Bottoms, on the grounds that her lesbian sexual orientation made her an unfit mother.

Stacey and Biblarz (2001, pp. 159–160) noted as follows:

• [S]ocial science research on lesbigay [sic] family issues has become a rapid growth industry that incites passionate divisions. For the consequences of such research are by no means “academic,” but bear on marriage and family policies that encode Western culture's most profoundly held convictions about gender, sexuality, and parenthood.

• As does Goldberg (2010, p. 110):

  Legal decisions such as this [Bottoms vs. Bottoms 1993] served as the catalyst for a steady wave of research studies that compared lesbian and gay parents with heterosexual parents to determine whether parental sexual orientation has implications for parent functioning.

Ironically, some writers within this literature point out the problem of researcher bias in the literature that opposes same-sex families (Stacey & Biblarz, 2001, p. 162):

• The deeply rooted hetero-normative convictions about what constitutes healthy and moral gender identity, sexual orientation, and family composition held by contributors to this literature hinders their ability to conduct or interpret research with reason, nuance, or care.

Conducting a standard difference of means, or any other type of standard statistical test, only makes sense when the conditions of a probability sample are met. This condition is almost never the case.

The same sentiments are driven home in Goldberg's book, in which she states in lesbian families there is “… greater sense of openness and communication within the family … greater tolerance for diversity, a greater sensitivity to discrimination, and growing up in a loving environment …. freedom from traditional models … less gender- stereotyped ideas” (2010, p. 97).

In a similar vein, Wainright et al. (2004) claim that “a warm accepting style of parenting is related to optimal outcomes for adolescents ….” (p. 1887). The latest study by Bos et al. (2013) also claims that children of lesbian homes are more civic minded and respectful of democratic institutions.

Gartrell and Bos (2010), concluding on the performance of the National Lesbian Longitudinal Family Study (NLLFS) children at age 17, stated (p. 7) the following:

• The lower levels of externalizing problem behavior among the NLLFS adolescents may be explained by the disciplinary styles used on lesbian mother households.

  The NLLFS mothers reported using verbal limit-setting more often with their children. Other studies have found that lesbian mothers use less corporal punishment and less power assertion than heterosexual fathers.

See Fedewa and Clark (2009, p. 317) for another example of the despite there being “no studies … it is generally accepted that … ” reasoning.

An interesting tension arises in the context of lesbian families and feminist ideology: the raising of sons. Chrisp (2001, p. 198) provides an almost emotional account of the dilemma faced by a feminist lesbian mother of a son:

• Few mothers, I believe, would not want their sons to make their own decisions, to be their own people, and to have the world full of opportunity for them. And yet, we need them [the sons] to reject the patriarchal status quo that has inhibited our own opportunities and will continue to do so for the women in their lives.

• These sentiments are repeated in Biblarz and Savei's 2010 survey (p. 482):

  Lesbian mothers raising sons may face unique tensions in wanting social and socioeconomic success for their sons when that may mean colluding with cultural ideas of hegemonic masculinity that encourage male achievement but involve the subordination of women.

• Ironically, because lesbians using some type of sperm donor insemination want to maximize the probability of conception, they often inseminate on the day of ovulation, which increases the chance of having a son (Chrisp, 2001, p. 198).

One element of feminist theory that crops up within this literature is the irrelevance of biology—a major theoretical competitor to feminist theories of the family. The importance of gender over sex has already been mentioned. Males can make good parents if they parent like a female. Another area is in the role of “social mother,” the mother not biologically related to the child. Goldberg et al. (2008) claim that by the time a child is 3.5 years old, children are indifferent between the biological and social mother. Based on a biased sample of 30 couples and some very soft questions they conclude the following: “These women demonstrate the power of ‘social motherhood’ in creating maternal connections that transcend biological relatedness over time” (p. 432). Finally, an almost humorous example is found in Goldberg (2010). After noting that birth mothers tend to specialize in the household, she quickly retorts, “This is not to say that lesbian mothers are reproducing gender relations along the lines of biology; rather it suggests that they are shaped by (and also shape) broader social patterns and various structural and symbolic forces” (p. 99). Everything is a social construct.

As noted throughout, this cannot be concluded for lesbian studies and certainly cannot be concluded for same-sex parents, for whom there is almost nothing known.