Abstract
Is the system of norms comprising traditional, natural marriage—featuring formally enacted, irrevocable, exclusive man/woman sexual union preceded by chastity—essential for children’s development and well-being, as Catholic teaching asserts? Review of an extensive body of diverse research finds that, compared to children continuously living with two parents, married parents, or their own biological parents, children in other family arrangements consistently experience lower emotional well-being, physical health, and academic achievement. Competing research has variously attributed this difference to a lack of married parents, two parents, complementary man/woman parents, or family stability, but these possibilities have not previously been studied in combination. To address this question, family structure differences and determinants of child well-being (reverse coded to show child distress) were examined using the 2008–2018 National Health Interview Surveys (n = 82,635). Adjusted odds ratios (AOR) for child emotional problems were higher with less than two parents (AOR = 1.42, 95% CI 1.27–1.56), unmarried parents (1.46, 95% CI 1.31–1.61), unstable parents (1.55, 95% CI 1.27–1.76), or less than two biological parents (AOR = 1.70, 95% CI 1.55–2.87 for one biological parent; 4.77, 95% CI 3.95–5.77 for no biological parents). When combined in the same model, only the lack of joint biological parentage accounted for higher distress, with outcomes significantly worse without the biological father than without the biological mother (interaction AOR = 1.33, 95% CI 1.04–1.71). This evidence strongly supports the claim that maximum child development occurs only in the persistent care of both of the child’s own biological parents. Marriage benefits children primarily by ensuring such care. Implications are discussed.
Summary:
Children raised apart from the care of both natural parents consistently experience lower developmental outcomes. Traditional, religious marriage norms—a lifelong, exclusive sexual union between man and woman—benefit children by establishing strong conditions that promote such care. More than any other family arrangement, marriage assures to children the care of their own mom and dad.
Keywords
One of the most far-reaching social changes in recent Western society has been the dramatic decline in the proportion of children who grow up in in the care of their own married father and mother. A century ago (1920) almost all children (above 90 percent) grew to age eighteen with intact married biological parents; by 2020, less than a third (29 percent) could expect to do so (Parker, Horowitz, and Rohal 2015). Today less than 60 percent of children even begin life with married parents (Ventura 2009; Martin et al. 2019), and half of those who do will be denied the residential care of one or both parents before age eighteen (Amato 2000).
This transition has accompanied the gradual shift from a rural, agricultural economy to an urban, industrial one, which motivated smaller, less extended families. Most of the change, however, has occurred since the 1950s, following the rising workforce participation of women, which made marriage economically unnecessary for many women, and the sexual revolution allied with the birth control pill, which undermined marriage as the gatekeeper for sexual experience. During this time, the “nuclear family,” defined by the Centers for Disease Control and Prevention (CDC) as a married father and mother with their own biological or adoptive children (The precise wording is “one or more children living with two parents who are married to one another and are each biological or adoptive parents to all children in the family”; Blackwell 2010, 2), rapidly gave ground to other family arrangements. First came divorce, the rate of which more than doubled, rising from one in five of marriages in the 1950s to one in two of 1970s marriages. Widespread divorce left in its wake a spate of unintended single parent families, 85 percent of the time a single mother. A generation later increasing numbers of divorce-averse young adults began to preempt marriage altogether, choosing to cohabit in informal marriage-like partnerships or to intentionally bear children without a partner.
Is the Nuclear Family Essential for Children?
Western culture has long privileged conjugal marriage and the nuclear family (also termed the “traditional” or “natural” family) in law and custom. As alternative family forms have proliferated, the primacy of the nuclear family has met with challenge and opposition from almost every corner of Western culture. Feminists have opposed it for perpetuating male dominance, gays and lesbians for imposing heteronormativity, and women’s advocates for locking women into abusive relationships. Individualists have criticized it for stigmatizing singleness, communitarians for thwarting collectivity. The first and most enduring criticism has been that since the nuclear family requires a single lifelong partner, it unnecessarily restricts sexual freedom. Today its grounding in man/woman sexual relations is widely challenged by left-leaning social scientists as “essentialist”, that is, assuming that cultural understandings of sexual difference are biologically real. As alternative family forms have become the majority, even many right-leaning theorists argue that the defense of the nuclear family, while correct in principle, is unworkable in practice. On this view, the nuclear family is now obsolete, unlikely to return, and it would be better to adapt moral principles and social policy to manage the alternatives that are rapidly displacing it.
The justification for the primacy of the nuclear family is both religious and practical. In Genesis, a text sacred to all the Abrahamic faiths (Judaism, Christianity and Islam) which have influenced Western culture, procreative marriage between man and woman is authoritatively presented as a pattern established by God in creation (Genesis 2:18-25). As part of the created order, it can be understood by the natural law, that is, the set of obligations and rights that derive from the nature of man and are discernible by reason, without any special revelation (Pius XI 1930). Classic Catholic theology appeals to reason, not revelation, to establish that marriage constitutes a “fixed, definite and settled arrangement, which will enable man and woman not only to procreate, but also to protect the offspring until they are capable of looking after themselves” (Brown 1955, 84). Modern Catholic theology makes the same kind of appeal by emphasizing that marriage between man and woman is consonant with the physiology of males and females (John Paul II 2006).
The Catholic Church, in accord with almost all variants of the Abrahamic faiths, asserts that the nuclear family is superior to other kinship arrangements for the optimal benefit of the man and woman partners, their children, and society at large. Recent magisterial thought has tended to restate parental natural law obligations from the perspective of the child in the more modern language of rights. Pope John Paul II taught in 1987: “The child has the right to be conceived, carried in the womb, brought into the world and brought up within [conjugal heterosexual] marriage: it is through the secure and recognized relationship to his own parents that the child can discover his own identity and achieve his own proper human development” (Congregation for the Doctrine of the Faith 1987, sec. II.A.1). Cardinal Ratzinger, the future Pope Benedict XVI, speaking for Pope John Paul II in 2003 of homosexual unions, invoked a specific set of juridical rights to affirm the necessity of man/woman parents for child development: As experience has shown, the absence of sexual complementarity in these unions creates obstacles in the normal development of children who would be placed in the care of such persons. They would be deprived of the experience of either fatherhood or motherhood…[which is] an environment that is not conducive to their full human development. This is gravely immoral and in open contradiction to the principle, recognized also in the United Nations Convention on the Rights of the Child, that the best interests of the child, as the weaker and more vulnerable party, are to be the paramount consideration in every case. (Ratzinger and Amato 2003, sec. 7)
Consistent with its basis in natural law, Catholic teaching authority has grounded this claim in arguments and evidence that appeal to any rational mind (John Paul II 1993). A critical question in the current debate, therefore, is whether the available evidence supports the primacy of the nuclear family for child well-being. Is it true that both man/woman natural parents are essential for optimal child development? The experience of children in alternatives to the nuclear family, which have as their common element the removal from the child of the joint care of both natural parents, offers a kind of natural experiment to test this hypothesis. To that end, the present study reviews prior research and examines population data on child well-being in various family structures in order to determine whether, as the Church claims, children in the care of their own natural parents have consistently better well-being or not.
Research Review
The debate over the nuclear family has prompted a spate of studies on the effect of changing family structures and transitions on the health and well-being of children. This diverse, extensive literature has been comprehensively reviewed elsewhere (McLanahan, Donahue, and Haskins 2005; Ginther and Pollak 2004; Manning 2015; Amato 2014; Ribar 2015). In the review that follows, I will focus on the question of interest in this study, that is, the current relative well-being and development of children in nuclear families compared to other family arrangements. Since much of our information on this topic comes from fertility surveys that only include women, for the sake of simplicity, I will not report parental measures for men unless there are important differences by parent sex, as for example, with single parenting or stepparents. Even when not explicitly identified as such, I will adopt the plausible inference that parents who have been in a continuous married or cohabiting relationship since the birth of a child are the biological parents of the child.
Divorce: Gateway to Dysfunction for Children
A review of the effect on children of family forms that defect from the nuclear family must begin by considering the effects of divorce. Portions of this section are adapted from Sullins (2017). Divorce is not a family form, of course, but is a gateway event that is implicated in most life trajectories that lead either by default or design to almost all nonnuclear family forms with children. Most single parents, remarried stepfamilies, and same-sex partnerships and marriages initiated those arrangements following divorce. Cohabitation and intentional single parenting are arrangements that often attempt to preempt the disruption of future divorce. The consequences of divorce for children, therefore, are expressed and compounded by the stress and relative dysfunction of the alternative family arrangements into which divorce propels them.
Today more children experience divorced parents than continuously married ones. As already noted, about half of all children born into married families will experience the divorce of their parents before age eighteen (Amato 2000). Half of these children of divorce, in turn, will go on to experience the remarriage of one or both of their parents (Furstenberg et al. 1983). A fifth of them, or 10 percent of all children born into married households, will witness the divorce of their parents two or more times (Bramlett and Mosher 2002). Since the 1980s, the rate of divorce has dropped for first marriages (from 50 percent to 40 percent) but risen for second and subsequent marriages, amid an increasing churn of marriage and marriage-like pairings over the life course (Cherlin 2010).
Research on these changes over the past five decades has found that parental divorce results in widespread negative physical, social, economic, and psychological consequences for the children affected. In a 2014 review, Paul Amato of Pennsylvania State University summarized the findings of decades of divorce research as follows: “In general, the accumulated research shows that children with divorced parents, compared with children with continuously married parents, exhibit more conduct problems, have more emotional problems, obtain lower academic test scores and school grades, and have more problems with social relationships. Divorce also is associated with weaker emotional ties with parents—especially fathers. These disadvantages appear to persist into adulthood” (Amato 2014, 12–13; see also 2000). Amato found that, on average over all studies and measures in both Europe and America, divorce reduced child well-being by about a third of a standard deviation compared to children with intact married parents.
Early Loss of a Father
Much of the reduction in child well-being following divorce stems from the effective loss of the child’s biological father early in life. This happens for two reasons. First, in about 85 percent of US divorces, custody of the children is awarded to the mother, thereby making the father, at best, a noncustodial parent. Paternal parenting is impaired, at minimum, and more often lost entirely. On the 1996 National Longitudinal Survey of Adolescent to Adult Health only one child in seventeen (6.3 percent) whose parents had divorced rated their father as “warm, loving and cared for them,” compared to almost one in two (43 percent) children whose parents were in a first marriage.
Furstenberg et al. (1983, 656) found that frequent contact with the outside parent occurred in only 17 percent of disrupted families, as fathers gradually withdrew from the former relationship and often moved away. Other studies confirm that “only about half of children with a nonresident father receive any child support or see their fathers more than a few times a year” (Stewart 2010). Even among the closest nonresident fathers, there is still a substantial loss of relationship imposed by the occasional, limited nature of their postdivorce interactions with their children (Kalter 1987, 595). Second, about 80 percent of couples that will ever divorce do so in the first ten years of marriage, which means that children are typically very young when their parents divorce. For most outcomes, the harm of divorce is greater for younger children than for older children and adolescents. This is particularly true for emotional and psychological trauma, as the divorce devastates children’s primary relational system just at the time when they are entering some of the most important and complex tasks necessary for their proper emotional and psychic development.
Main Outcomes
Research has focused on three main outcomes that are negative for children outside of nuclear families: poverty, emotional and behavioral development, and educational attainment.
Poverty
Financial resources and support for the children are greatly reduced, largely as a result of the loss of the father’s earnings (Hogendoorn, Leopold, and Bol 2020). Lower family income, especially if it is below poverty, has been found to be associated with deprivation and reduced well-being for children in a number of ways, including poorer overall health, increased mortality rates, greater risk of abuse or neglect, lower developmental scores at any age, reduced educational attainment, and lower earnings as adults (Hogendoorn, Leopold, and Bol 2020). Divorce strongly increases the risk of poverty for custodial mothers and children but not for men (Mortelmans 2020; Lundberg, Pollak, and Stearns 2016; Aber et al. 1997). According to 2012 Census data, female single parent households, at 38.8 percent, were over four times as likely to be in poverty as were married households, at 8.8 percent (Vespa, Lewis, and Kreider 2013, table 5). This may account, in part, for findings that children do better with single fathers than with single mothers although as we shall see other factors may be at work. The negative effects of deprivation on child well-being are nonlinear, becoming increasingly adverse as one descends from middle-income affluence, and especially harmful for families below the poverty line (Hogendoorn, Leopold, and Bol 2020, 1091). Partly for this reason, while public aid programs can lessen the effect of economic limitations (Furstenberg 2005, 80), equalizing resources does not fully bridge the gap in well-being between children of divorce and children with intact married parents (Amato 2010).
Apart from comparison with the nuclear family, variation in financial resources among the alternative family structures do not align well with differences in child well-being. Intentional single mothers have even lower education and income than postdivorce single mothers (Pew Research Center 2013) but do not experience consistently lower child outcomes. Remarriage after postdivorce single parenting often brings increased financial security but not increased child well-being. Intact cohabiting parents also have lower income but greater child well-being, on average, than do married stepparent families (Manning and Brown 2006). On the other hand, gay and lesbian parents tend to have higher education and are less likely to be in poverty than comparable heterosexual parents (Prokos and Keene 2010, 945), but their children have lower well-being (Sullins 2015; Allen 2013). In sum, while economic resources have a strong effect, other elements of family structure are probably more definitive for child well-being.
Education
One of the most widely acknowledged consequences of parental divorce is its disabling effect on academic achievement. Research in many countries has shown that compared to children with intact married parents, the children of divorce earn lower grades and experience more problems in school (Arkes 2015), are less likely to complete secondary education (McLanahan, Tach, and Schneider 2013), and less likely go on to college or university (Bernardi and Radl 2014). These deficits are related to reduced economic resources (Havermans, Botterman, and Matthijs 2014), which are in turn related to lower paternal investment. Judith Wallerstein’s twenty-five-year longitudinal study of children of divorced parents found that the children in her US sample were less likely to have completed college or attained a graduate degree than were their parents (Buttenheim 2001, 10). This downward mobility was “a direct result of the fathers’ failure to contribute to their children’s college expenses: only 29 percent of the divorced children received full or consistent partial support from their parents for college, compared to 88 percent of the children from intact families” (Buttenheim 2001, 10). Most of the fathers could have afforded to contribute but felt no need to exceed their legal obligations to their children, which ended when they turned eighteen. Affirming what the Catholic/natural view of the family would predict, Wallerstein concluded that “at least for some fathers, the tie to their biological children diminishes outside the original marital relationship” (Buttenheim 2001, 12).
Research that includes other postdivorce family forms has confirmed the importance of biological parents for educational success. Turunen, examining academic achievement scores by family form from full-population Swedish register data (N = 874,812), found that “boys and girls living with both biological parents…had the highest mean grade-point scores”; “children with [single parents] are the second most successful group, and the children [in blended families] score somewhat lower.…The children scoring lowest are those with the most complex family background…” (Turunen 2014, 578).
Behavioral/emotional problems
Like the other outcomes discussed so far, almost every study that has examined the question has found that children in alternative family arrangements are subject to a wide range of emotional and behavioral problems at higher rates than children of intact marriages or with two biological parents. On most measures, the children in nonnuclear families are about twice as likely to experience these problems as are children of an intact marriage or with two biological parents.
Research on behavioral and emotional outcomes has been more likely to move beyond binary comparisons such as married/divorced or intact/nonintact and to include the presence/absence of biological parents, specifying a wider array of postdivorce parent/child relationships. Dawson’s study of the 1988 NHIS was one of the first nationally representative studies to report that children living with two biological parents were less likely to experience behavioral or emotional problems than children living in other family types (Dawson 1991, 579). McLanahan and Sandefur’s (1994) analysis of four nationally representative data sets reported that “adolescents who have lived apart from one of their parents during some period of childhood are twice as likely to drop out of high school, twice as likely to have a child by age twenty, and one and a half times more likely to be ‘idle’—out of school and out of work—in their late teens and early twenties” (McLanahan and Sandefur 1994, 2). More recently, Bramlett and Blumberg, using the 2003 National Survey of Children’s Health, reported that children living with their mother (but not their father) in single or stepfamilies after divorce experienced twice the rate of both moderate and severe emotional problems as those living with two biological parents, a difference which persisted in the presence of sociodemographic and economic controls (Bramlett and Blumberg 2007, 553, Exhibit 3). Gorman and Braverman (2008) found that all measures of health care utilization were lower for children with stepparents or single parents compared to two married parents.
In 2010, Blackwell and a team of demographers from the CDC’s National Center for Health Statistics reported findings from the 2001–2007 National Health Information Surveys (NHIS) comparing children in nuclear (intact married) families with those with postdivorce single parents, remarried stepparents (blended), and unmarried and cohabiting parents (among others) on a wide range of indicators of physical and emotional health.(Blackwell 2010) In the pattern which is by now familiar, on almost every indicator examined children being raised in single parent, stepparent (blended) or cohabiting parent families exhibited poorer health than those in nuclear families. Figures 1 –4 reproduce selected findings, showing that the children in nuclear families were significantly less likely to be in good, fair, or poor health versus very good or excellent health (Figure 1); to have a basic action disability (Figure 2); to be generally not well behaved or disobedient to adults (Figure 3); or to have definite or severe emotional or behavioral difficulties (Figure 4). Their conclusion: “Children living in blended (i.e., stepparent), cohabiting, unmarried biological or adoptive, extended, and other families were generally disadvantaged relative to children in nuclear families,…” (Blackwell 2010, 27).
Percentages of children under age eighteen in good, fair, or poor health by family structure: United States, 2001–2007. Source: Blackwell (2010, 11). N = 12,604.
Percentages of children under aged four to seventeen who had a basic action disability by family structure: United States, 2001–2007. Source: Blackwell (2010, 13). N = 12,604.
Percentages of children under aged four to seventeen who were generally not well behaved or did not usually do who adults requested in the past six months by family structure: United States, 2001–2007. Source: Blackwell (2010, 24). N = 12,604.
Percentages of children under aged four to seventeen who had definite or severe emotional or behavioral difficulties, by family structure: United States, 2001–2007. Source: Blackwell (2010, 26). N = 12,604.
A close look at Blackwell et al.’s results confirms the well-attested finding that remarriage after divorce does not improve child well-being. Like most research in this area, they do not pursue such differences beyond noting that the two types of family are comparably situated with respect to nuclear families (Blackwell 2010, 35). A small group of studies, however, has recently begun to examine differences among the diverse nonnuclear family forms more closely in order to better specify the determinants of child well-being in any family arrangement.
Parsing Marriage and Biology
Many studies of family structure and child outcomes conflate marriage and biological parenting. Survey data in this area have almost always measured parent marital status but often leave children’s biological relationship to parents unmeasured. Children who have resided continuously since birth with the same set of married man/woman parents can be reasonably inferred to be living with their biological parents, but many surveys do not include all the information to make this inference. In addition, almost no studies, including the present one, are able to eliminate the inaccuracy in measures of biological parenting due to the small proportion of nonbiological children included due to joint infant adoption or assisted reproductive techniques. Recent studies of both types of children suggest that outcomes are little different than those of similarly placed biological children, though these findings are tentative (Sehmi et al. 2020; Golombok 2020). Studies of cohabiting parents also seldom distinguish biological from nonbiological parents although those that do tend to find that, as Manning summarizes, “Generally, young children living in two biological parent cohabiting families fare as well as children residing in two biological parent married families,…” (Manning 2015, 7). Blackwell et al. found that this was true for some, but not all, indicators, noting: “Interestingly, children living in unmarried biological families share some of the health characteristics of both nuclear and cohabiting families” (Blackwell 2010, 27). They called for additional research into this family form which separated the effects of marriage and biology.
Conway and Li compared nuclear families with nine alternative family forms distinguished by marital status and biological parent gender on measures of physical health, school engagement, and behavioral problems. Their results confirm the nearly universal finding that “nontraditional families are associated with lesser child outcomes, even after controlling for a large list of economic resources and inputs.” Among the nontraditional families, they found that “two-parent cohabiting families fall between [nuclear] families and other [non-nuclear] families,” suggesting that “living with both biological parents is the more empirically important distinction for child well-being than if the household is headed by a married couple or not” (Conway and Li 2012, 371).
They also found that children living without their biological mother but with their biological father, either singly or with a stepmother, fared better than those living in comparable arrangements without their biological father but with their biological mother (Conway and Li 2012). Studies that make the distinction have also typically confirmed that children do better with a single father than a single mother (Blackwell 2010; Dawson 1991; Ziol-Guest and Dunifon 2014). Ziol-Gruest and Dunifon (2014) conclude that living with a biological father has a “key role” in assuring children’s health, which is not reducible to income or health insurance status. In sum, this stream of emerging research that defined parent/child relationships more precisely sharpened previous findings to suggest that, in addition to marriage, biological relationship and parent gender may also be consequential for child well-being.
Method
Analytic Strategy
As indicated above, Catholic teaching references at least four qualities of the nuclear family that are necessary for healthy child development: the complementarity of two man/woman parents, the child’s own biological parents, married parents, and security or stability. These qualities are conceptually distinct but not mutually exclusive in practice. Each of the family forms that currently defect from the nuclear family lacks one or more of them. Table 1 illustrates. The seven alternative family forms shown offer successively fewer features of the nuclear family as one moves from unmarried cohabitation, which differs only in that it lacks married parents; to remarried stepfamilies, which lack both stability and both biological parents; to cohabiting stepfamilies, intentional single parents, and same-sex married parents, who each retain only one element of the nuclear family; to postdivorce single parents and same-sex cohabiting parents, who offer none of the four nuclear family features. It is not known which of these qualities or features, or combination of them, account for observed reductions in child well-being outside of nuclear families when they are not present. As the reviewed literature above demonstrates, each of them has its advocates and supporting research; however, they have not heretofore all been studied in combination. The present study attempts to amend that research gap.
Nuclear Family Compared to Alternative Family Forms.
The analysis proceeded in two stages. First, I examined representative population data to test children’s need for natural, that is, their own joint biological, parents comparing the family forms presented in Table 1. This procedure reproduces that found in almost all prior research in this area. The broadness of the claim requires that child well-being be better in nuclear families when compared to all or almost all alternate arrangements, not just a subset of the alternatives. To further identify what factors may account, more or less, for differences in child well-being across family structures, in a second stage, I isolated the features which are collected in the nuclear family but dispersed or missing in alternate arrangements: marriage, two complementary parents, stability, and joint biological parents. Reflecting extensive research on the special harm of father absence, I also isolated, for families with a missing biological parent, the gender of the missing parent. Statistical models that compared the combined effect of these features were analyzed to determine their relative importance in explaining reduced child well-being when each is absent and therefore what specific feature or features of the nuclear family were essential for optimal child development.
Data
Data for this analysis were derived from the National Health Interview Survey (NHIS). The NHIS is the principle source of public health information about the US population. Since 1957, the US CDC’s National Center for Health Statistics has annually interviewed between 35,000 and 40,000 households, collecting extensive health and demographic data on 75,000 to 100,000 individuals comprising a nationally representative sample of the civilian noninstitutionalized population of the United States. For each family that includes children under age eighteen, detailed supplemental health information from a parent or other knowledgeable informant is collected for one child chosen at random (the “sample child”). Year-specific information on sample design and questionnaires is available at http://www.cdc.gov/nchs/nhis/nhis_questionnaires.htm.
The NHIS collects an extensive family roster that defines the relationships among all members of the household, much like the US census. It also employs an array of well-validated measures to assess the physical and mental health of the population. These features make it ideal for the present study. The 2010 CDC study of family structure and child health by Blackwell et al. reviewed above used the combined 2001–2007 NHIS data. In many ways, the present study employs a similar analysis using subsequent NHIS data from 2008 to 2018. A more complete description of the NHIS data can be found in Blackwell et al.’s study. An earlier study by Sullins, from which the previous paragraph is drawn, also describes the NHIS data more extensively as well as the composition of any variables used in the present study which are not described below. The interested reader is referred to these sources.
In the present analysis, the definition of family forms and child outcomes follow those of Blackwell et al. and Sullins. All child health data are based on parent reports. “Poor child health” compares the categories “Good, Fair, and Poor” with “Excellent and Very Good” from a five-category rating of the child’s general physical health. The definition of the nuclear family qualities is straightforward. “Two parents” and “Married parents” are bivariate indicators coded 1 when those conditions are present for a child and 0 when they are not. “Stable parents” similarly indicates children who have been continuously with the same set of parents (1) since birth versus those who have not (0). Since the absence of these conditions is what is of interest, these variables are reverse-coded for analysis. “Absence of bio father (mother)” is coded 1 for children in parenting arrangements where his or her biological father (mother) is not present but his/her biological mother (father) is present and 0 otherwise. “Biological parents” indicates the number of the child’s own biological parents who are present in the family and thus has responses of 2, 1, or 0. These values are reverse-coded in order to express increasing departure from the composition of nuclear families, which by definition includes joint biological parents.
Child outcomes are expressed as dichotomous measures, which are analyzed in logistic regression models resulting in exponentiated coefficients expressed as odds ratios (ORs). ORs express the probability of being in one outcome category versus another, conditional upon being in one treatment category versus another. ORs can be substantively interpreted in terms of risk, risk ratio, or relative risk although these statistics are technically slightly different. All analysis models include the following potential confounders: parent education (BA degree or higher), parent income (expressed as poverty status), child race, age, and sex. Models also include the sex of the family informant on the NHIS interview to counter gender bias in parent reports and the survey year to equalize any time trend. All models also employ statistical weights and adjust the variance for complex survey sample design so as to present adjusted odds ratios (AOR), confidence intervals, and significance tests that are statistically representative of the US population. Analyses were performed using SPSS 25 and Stata 13 software. This research was performed subject to the oversight of the Institutional Review Board of the Catholic University of America, which determined that, as a secondary analysis of deidentified publicly available data, it was exempt from additional human subject protocol review.
Results
Table 2 reports the means or proportions for the covariates, and Table 3 for the outcome measures, by family structure and cell size in the analysis. Three family types—cohabiting step, same-sex married, and same-sex cohabiting—are represented by a small number of cases, resulting in less robust statistical tests of significance. With this complicating exception, it can be seen in Table 2 that parent education and income, and child race and age, are significantly different from those of nuclear families in all other alternative family forms. Although they will not be reported in the analysis that follows, as they are not of interest in this study, all of these variables have a significant effect on child outcomes. Except for same-sex married couples, parent education is highest and poverty is lowest in nuclear families. A smaller proportion of nonwhite children live in nuclear families than in any other family form. Average child age is lowest in intact cohabiting families, reflecting their relatively higher rate of dissolution in proximity to childbirth. Average child age is higher in stepfamilies, postdivorce single parent families, and same-sex parent families, which bring in children from the dissolution of a prior relationship, but lower among intentional single parents than it is in nuclear families.
Unadjusted Population-Weighted Descriptive Statistics for Covariates in the Analysis, by Family Structure: NHIS 2008–2018.
Difference from nuclear family, by t-test:
†p < .10.
*p < .05.
**p < .01.
***p < .001.
Unadjusted Prevalence of Child Outcome Measures, by Family Structure (Population-Weighted): NHIS 2008–2018.
Note: ADD = attention-deficit disorder.
Difference from nuclear family, by t-test:
*p < .05.
**p < .01.
***p < .001.
Table 3 presents the proportion of children experiencing a range of negative outcomes by family structure. Mirroring Blackwell’s findings for many of these measures in a previous period of the same data, the prevalence of these emotional and developmental difficulties is observed to be significantly higher in almost every nonnuclear family form. The differences are strongest and most consistent for the psychological metrics presented in the first two rows: serious emotional problems as assessed by the Harvard-validated Strengths and Difficulties Questionnaire, and having received a diagnosis of attention-deficit disorder (ADD). These conditions are significantly higher for children in every other family structure compared to nuclear families, despite the small sample size for some categories. Physical health ratings are poorer, compared to nuclear families, for children in every family form except with same-sex married parents. Up to three times as many children have been retained a grade in school in postdivorce families, whether remarried or single parent, or with same-sex parents.
Four more outcomes are presented at the bottom of Table 3 that expand or illustrate the dimensions shown in the top four measures but were not included in further analysis. “Saw doctor for mental health condition” reports the proportion of children who had seen a doctor for a mental health problem, and “Medication for emotional problem” reports the proportion of children who had been prescribed psychotropic medication, in the past year. As with emotional problems and ADD, these variables report related activities that were significantly higher in all alternative family structures (with one exception; children with intact cohabiting parents were not more likely to see a doctor for a mental health condition). Having a learning disability and being in special education in school present two other conditions related to academic achievement. Again, with one exception (special education with intact cohabiting parents), both of these conditions that reflect challenges for child development are much more common for children in every family structure except the nuclear family.
The unadjusted differences among family forms shown in Table 3 may have been partly or wholly due to the effects of the covariates shown in Table 2. To address this possibility, the top four outcomes of Table 3 were estimated from logistic regression models controlling for the covariates in Table 2, as well as the sex of the respondent and the survey year. The results are shown in Table 4. The table presents AORs comparing the well-being of children in nuclear families (the reference) with that in the seven alternative family forms presented in Tables 1 –3. The four outcomes examined—emotional problems, physical health, academic achievement, and a diagnosis of ADD—generally represent the main spheres of harm analyzed in prior research. The measures are scored in terms of declining well-being or increasing distress, so that an OR greater than one indicates that child distress with the index family form is greater than for children in nuclear families.
Adjusted Odds Ratios for Child Outcomes by Family Structure: NHIS 2008–2018.
Note: Odds ratios are predicted from logistic regression models adjusted for parent education and income, child race, sex and age, sex of respondent, and survey year. All covariates were significant; results are suppressed in the table for simplicity.
Difference from nuclear family, by t-test:
*p < .05.
**p < .01.
***p < .001.
Consistent with virtually all prior research, every AOR shown in the table is greater than one, indicating that child distress is higher in every alternative family form compared with the nuclear family. The lack of significance for cohabiting stepparent families and same-sex married parents, despite point estimates above unity, probably reflects the lack of statistical power due to the small number of cases in these categories (376 cohabiting stepparent and 133 same-sex married families; see Table 3).
All adjusting covariates had a statistically significant effect on the outcome in every predictive model examined. Comparing the AORs presented in Table 4 with the corresponding prevalences reported in Table 3 suggests that the covariates had a mixed effect on the unadjusted differences. The AORs for poor physical health and school retention in Table 4 tend to be moderated from the corresponding unadjusted comparisons shown in Table 3; for ADD, they tend to be larger; and for emotional problems, the two are generally similar.
The nonnuclear family forms do not appear to follow a pattern of increasing distress as fewer qualities of the nuclear family are retained, as suggested by Table 1 and the related discussion. Well-being with single parents, for example, was better than with remarried stepparents for all outcomes. For the two psychological measures, well-being was highest (the odds on distress was lowest) with intact cohabiting parents, suggesting that joint biological parents are important for these outcomes. For three of the four outcomes examined in Table 4, well-being is better with cohabiting than with remarried stepparents and same-sex parents, suggesting that parental marriage may not overcome the harm of an absence of joint biological parents and/or instability. And child emotional problems and school retention are higher with married than with cohabiting same-sex parents. Overall, these distinctions suggest, at least, that the retained qualities of the nuclear family interact in complex ways, and more likely that some of them are more important than others, in affecting child well-being in the nonnuclear family forms.
The models presented in Table 5 explore these possibilities further, presenting AORs for child emotional problems related to the particular nuclear family qualities lacking in various alternative arrangements, specifically: joint biological parents, the biological mother (but not the biological father), the biological father (but not the biological mother), two complementary parents, married parents, and stability. The first four test the qualities identified in Catholic teaching presented in Table 1; the latter two test the thesis, found in the literature, that children fare worse with a missing father than with a missing mother. Except for biological parents, which has three design categories (both, one, or none), all of these effects were coded as bivariate measures indicating the presence (0) or absence (1) of each quality.
Adjusted Odds Ratios (AORs) for Parent Relationship Features Predicting Child Emotional Problems: NHIS 2008–2018.
Note: Odds ratios are predicted from logistic regression models adjusted for parent education and income, child race, sex and age, sex of respondent, and survey year. All covariates were significant; results are suppressed in the table for simplicity.
Difference from reference category, by t-test:
*p < .05.
**p < .01.
***p < .001.
All six of these defections are individually associated with increased child emotional problems, as the “Zero Order” reports. For the four qualities identified in Catholic teaching, the bivariate AORs show significantly increased risk ranging from 42 percent to 70 percent (for one bio parent compared to none). Consistent with prior research, the bivariate AOR for emotional problems with the absence of the biological mother, at 1.07, was much lower than with the absence of the biological father, at 1.49. The difference between these two coefficients, however, is not statistically significant (not shown). 1
The most powerful bivariate effect on child well-being is biological parentage. Compared to children with joint biological parents, those with only one biological parent experience 1.70 times higher risk of emotional problems, and almost five times (4.78) the risk for children residing with no biological parents. The second column of Table 5 (“With bio parents”) presents models that combine each of the other qualities with biological parentage. In these models, the effect of biological parentage is generally increased, while each of the other qualities of the nuclear family is either rendered null or reversed. In effect, the apparent zero order association of child emotional problems with these measures is accounted for by the presence or absence of biological parents.
The variable “Absence of bio mother” captures the effect of arrangements where the biological father is present but not the biological mother. These include single fathers and remarried or cohabiting fathers with stepmothers. At zero order (column 1), child emotional problems are half again higher with the absence of the biological father but unaffected by the absence of the biological mother. When adjusted for biology (column 2), the associations with the sex of the missing parent is reduced, but their relative magnitude remains the same.
In the final model (column 3), child well-being is accounted for by just two effects: biological parentage and father absence. This model fits the three categories of biological parent/child relationship and the interaction of father absence with mother absence, which is significant. The Hosmer–Lemeshow statistic of 0.56 indicates it has a good fit to the data. Child emotional problems strongly increase with the loss of any biological parent and are a third higher (AOR 1.33) when the missing biological parent is the father rather than the mother.
Discussion
As the above analysis has shown, the evidence that children flourish best under the uninterrupted care of their natural mother and father is among the strongest we have for any proposition in the social sciences. If the proliferation of alternative family forms over the past fifty years were a social experiment in parenting arrangements, there would now be more than enough evidence to conclude that it was a failure. Single parenting, stepparenting, and cohabitation may or may not have improved the lives of adults, but they have demonstrably degraded the lives of children.
Just as modernity is characterized by the differentiation, specialization, and rationalization of formerly undifferentiated roles and functions in other spheres of life, many have interpreted the rise of diverse family forms in terms of the differentiation of functions formerly collected in the nuclear family. Legitimate romantic intimacy, sexual expression, a stable public relationship, economic cooperation, and childbearing, once relegated exclusively to the nuclear family centered in marriage, have now become severed from marriage and one another, each to occupy a distinct set of relationships and arrangements. On this view, given proper social support or due to disappearing social support for marriage, the new functional arrangements may soon replace marriage.
But the functions of the nuclear family that matter for children—two complementary partners, marriage, stability—do not function apart from biological parenting. In both the specific findings of this study (Table 4) and the wider literature on child well-being, parental marriage with suboptimal biological parentage did not improve child well-being. Marriage after divorce, which cuts across biology by introducing a nonbiological parent figure, increasing the disaffection of the nonresidential biological parent and dividing the affection of the residential biological parent, did not improve and reduced estimated child well-being compared to postdivorce single parenthood. Marriage between same-sex partners also reduced estimated child well-being compared to cohabiting same-sex parents. In these cases, the introduction of marriage that did not reflect increased biological propinquity with the children actually harmed child well-being. Intact cohabiting parents, being joint biological parents who differ definitionally from nuclear families only in that they lack marriage, experienced estimated child outcomes that were better than those in any married family form without joint biological parents.
Parental marriage did improve child well-being in one biological condition: when the parents were the joint biological parents of all the children in the household. Children with intact married (i.e., nuclear family) parents fared significantly better on all measures than those with intact cohabiting parents (see Table 4). Whether this difference is due to unobserved functional differences or an independent marriage effect cannot be resolved by the evidence in view in this study. In either case, the power of marriage to benefit children is in biology. Marriage or its correlates have a positive effect on child well-being only when biology is optimal; failing that condition, marriage has either no effect or a negative effect. And if the functions found in the nuclear family are carried by biology, they cannot operate independently of biological relationships found in the nuclear family itself. There is no substitute for the procreative relationship of man, woman, and their own child.
Although an almost universal religious ideal, the greatest benefit of sacred marriage for children may rest in its conformity to biological nature. In Catholic thought, such conformity is explicit. Formal vows of marriage are incomplete without a biological action, that is, sexual intercourse between the partners that consummates the marriage. Without the biological ability to commit such an act, one is not qualified to enact a Catholic marriage. On this view, marriage is inseparable from biological sex relations that are likely to result in children. This does not reduce marriage to biology; it would be more accurate to say that marriage is elevated to biology because it is by this biological action that sacramental marriage is consummated and rendered indissoluble.
In the Catholic faith, every one of the seven sacraments is expressed in matter. For marriage, the matter expressed consists, in prospect and in part, of the bodily sexual union of the persons receiving the sacrament which they promise to one another at the wedding and enact with one another subsequently, in a total gift of bodily self. A priest or deacon does not “marry” persons; he only witnesses their vows to marry one another. The ministers of marriage are the couple themselves. In the mutual ministration of their bodies to and with one another, “the sacrament and its conferring of grace take place in virtue of the mutual consent previously expressed and still continuing” (Lemkuhl 1910) . In the words of the New Testament, their “one flesh” union expresses the “great sacrament” of the relation of Christ and the Church (Douay-Rheims Bible 1582, v. Ephesians 5:31-32). Just as Christ is focused not on himself but on the sons and daughters he came to save, so this view of marriage orients the focus of the partners to the sons and daughters they are likely to create. Their sexual union is not just mutually fulfilling but also generative and sacrificial. The love of man and woman who sacrifice themselves for those they have created mirrors the love of God who has sacrificed himself for his creation. Such sacrifice brings new life and allows children to mature most in accord with their nature.
On this view, marriage can be understood as a social arrangement that ensures to children the undistracted and uninterrupted care of their own mother and father. Consistent with this, the Catholic faith does not recognize the possibility of dissolving a consummated marriage, the validity of second marriages after divorce, and proscribes all sexual relationships outside of marriage. These boundary conditions—premarital chastity, marital fidelity, and unrepeatable lifelong commitment—ensure that children born to the partners will both be the biological offspring of both partners and will be their only biological offspring. Marriage so defined thus assures the biological conditions for the maximum developmental benefit of each child. A relational arrangement that is not in line with these biological conditions may bear the name of marriage, but it does not have the same power to serve the best interests of any children involved. When “marriage” is used to institutionalize arrangements that thwart biological parentage, as with remarriage or same-sex marriage by partners who already have children by another partner, it can even degrade the best interests of the child.
Policies aiming to serve the best interests of the child often consider such factors as parental income, stability, and quality of relationship with the child. These findings suggest that, in adjudicating family breakdown, judicial remedies and social policy should place greater emphasis on preserving and enabling the bonds of the natural parents and child. This would first entail erecting higher barriers in law and custom to the dissolution of a nuclear family than currently exist. Divorce by mutual consent should be legally curtailed. When divorce occurs, financial provision for the children and the involvement of the nonresidential parent should be protected by stronger sanctions than at present. In all family types and configurations, the rights of the child to the care of the natural parents and, of the natural parents to care for the child, should be the foremost consideration. Marriage by one biological parent to a person unrelated to the child should not include a presumption of parenthood or a parenting role on the part of the unrelated person. Adoption, custody, guardianship, or the erection of any similar right with respect to the child by someone who is not his or her biological parent should also be resisted and permitted only when both biological parents agree, and the parent who would be displaced has specifically waived her/his corresponding rights.
Conclusion
The evidence presented in this study demonstrates without qualification that children experience lower developmental outcomes in the absence of the coresidential care of their own natural parents. Such care, therefore, is necessary (though may not be sufficient) for children’s maximum development. Traditional, religious marriage norms—lifelong, a sexual union, between man and woman, observing chastity outside marriage and fidelity inside it—benefit children by establishing strong conditions for such care, and may add other benefits. All things being equal, when children can only have the residential care of one natural parent, well-being is better when they retain the father.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research and/or authorship of this article: This study received funding from Ruth Institute.
