Comparing the Role of Selection in Early Adolescent Substance Use Disparities Related to Single-Mother Family Structures Across Three Affluent Countries.

Comparing the Role of Selection in Early Adolescent Substance Use Disparities Related to Single Mother Family Structures across Three Affluent Countries - PMC Skip to main content An official website of the United States government Here's how you know Here's how you know Search Log in Primary site navigation Search Log in Search PMC Full-Text Archive Search in PMC Journal List User Guide As a library, NLM provides access to scientific literature. Inclusion in an NLM database does not imply endorsement of, or agreement with, the contents by NLM or the National Institutes of Health. Learn more: PMC Disclaimer | PMC Copyright Notice Demography . Author manuscript; available in PMC: 2026 Jun 4. Published in final edited form as: Demography. 2025 Oct 1;62(5):1661–1687. doi: 10.1215/00703370-12253547 Comparing the Role of Selection in Early Adolescent Substance Use Disparities Related to Single Mother Family Structures across Three Affluent Countries Jared D Thorpe 1 , Robert Crosnoe 2, * Author information Copyright and License information 1 University of North Carolina at Chapel Hill 2 University of Texas at Austin * Correspondence concerning this article should be addressed to the first author at Population Research Center, University of Texas at Austin, Austin, TX 78712 ( crosnoe@austin.utexas.edu ). PMC Copyright notice PMCID: PMC13232638  NIHMSID: NIHMS2180623  PMID: 41065549 The publisher's version of this article is available at Demography Abstract This study investigated the association between family structure and onset of substance use by early adolescence (e.g., prior to age 14) in Australia and the U.K. and U.S. with a focus on the role of selection in this association. Leveraging nationally representative surveys, logistic regression models estimated this association while iteratively controlling for three sets of selection mechanisms and testing for differential robustness in the face of unobserved traits. Results revealed higher rates of substance use among early adolescents living with single mothers compared to those living with married mothers and early adolescents living with cohabiting mothers falling in between. In general, mothers’ own family formation histories and especially their socioemotional adjustment emerged as key confounds that attenuated these associations, while their socioeconomic histories more often acted like suppressors. Unobserved confounds also appeared to be at work. Such patterns were fairly consistent across countries but with some evidence that single motherhood mattered more to early adolescent substance use (particularly alcohol) and was less selective in terms of both observed and unobserved confounds in Australia, the country with the most policy buffers for families and youth facing hardship. Keywords: family structure, substance use, adolescence, selection, socialization Introduction In single mother families, young people tend to experience more social and economic instability and insecurity over time, which can undermine their trajectories into adulthood. Consequently, rising numbers of children in single mother families in many affluent Western countries have fueled often intense debate about the role of family structure in the intergenerational transmission of inequality and the divergence in children’s developmental ecologies by family structure ( Heuveline et al. 2003 ; McLanahan 2004 ). The higher rates of substance use among adolescents in single mother families—especially during the critical early years of adolescence—is one family structure disparity engendering such debate ( Brown and Rinelli 2010 ; Cavanagh 2008 ; Hemovich and Crano 2009 ; Hoffmann 2017 ; Ohannessian et al. 2015 ; Richmond-Rakerd et al. 2017 ). Recent research has contextualized this particular disparity by showing that it can be both similar and different in the U.S. and the U.K. ( Stritzel et al. 2022 ). Notably, selection into single mother families and socialization within them both drive the association between this family structure and substance use during early adolescence, a period that is marked by the pubertal transition and other rapid developmental changes, is conventionally defined as ages 10–14, and is characterized by strong normative controls on drinking and drug use ( Baum et al. 2014 ; Schulenberg et al. 2015). Typically, research has statistically controlled for selection, focusing on the “leftover” as evidence of socialization within and across contexts. That approach sheds light on the contextualized role of family life in youth development, pointing to the need to intervene in or better support single mother families as a means of reducing early adolescent substance use ( Cavanagh 2008 ; Hoffmann 2017 ). A complementary approach is to explicitly unpack the mechanisms of selection underlying the association between single mother families and early adolescent substance use and compare them across contexts. Doing so can reveal social conditions that affect both mothers and adolescents, suggesting the need to target downstream factors leading to single motherhood ( Cross et al. 2022 ; McLanahan 2004 ). This study, therefore, starts with the association between single mother family structures and substance use emerging around age 14 and then directly tests the role of four different mechanisms of selection into single mother families. These mechanisms concern demographic, economic, and psychosocial processes that lead a woman into single motherhood and also have implications for their children’s behaviors, including observable mechanisms like family formation, socioeconomic attainment, and adjustment and those more difficult to observe. We wanted to identify mechanisms that might lead to the appearance that living with a single mother causes early adolescents to engage in substance use even if it actually does not. This study also examined whether these selection processes varied across countries with both similarities (e.g., economic status, welfare regime) and differences (e.g., norms of marriage and fertility, public investments in children) as a means of determining where selection and its implications for theory and policy are most pronounced. Thus, we modeled young adolescents’ alcohol and marijuana use as a function of living with a mother who was single, married, or cohabiting in three countries while iteratively controlling for multiple sets of observable confounds and testing the robustness of results to unobservable confounds. This modeling drew on the harmonized pooling of three nationally representative studies of birth cohorts in Australia (Longitudinal Study of Australian Children, LSAC), the U.K. (Millennium Cohort Study, MCS), and the U.S. (National Longitudinal Survey of Youth—Child and Young Adult study, NLSY79-CYA). Our focus on mothers was motivated conceptually and practically. Mothers head the large majority of single parent families in Western countries and maintain greater responsibility for parenting on average (see Heuveline et al. 2003 ), and our adolescent-focused population data sets tracked mothers’ union formation and dissolution more comprehensively than fathers’. This mother-focused study can then set up a broader research agenda on the role of selection in single parent families (and other family structures) more generally. Linking Single Mother Families to Early Adolescent Substance Use The heightened behavioral problems of young people living in single mother homes illustrate life course theory’s principle of linked lives ( Elder et al. 2003 ), which asserts that the social, institutional, and personal experiences of parents are connected to the developmental trajectories of their children. Here, family structure reflects mothers’ histories of partnering and uncoupling, which can introduce stressors into the developmental ecologies of their children to disrupt adjustment and functioning over time ( Amato 2005 ; Breivik and Olweus 2006; Cavanagh 2008 ; Flewelling and Bauman 1990 ; Gaydosh and Harris 2018; McLanahan and Sandefur 1994 ; Turney and Halpern-Meekin 2020). Although receiving less attention than other developmental domains, a solid base of evidence indicates that this particular linked lives phenomenon applies to adolescents’ substance use. This link to substance use includes substance use in or before early adolescence ( Cavanagh 2008 ; Crawford and Novak 2002 ; Hoffmann and Johnson 1998 ; Krohn et al. 2009 ; Thomas et al. 1996 ; Wagner et al. 2010 ). We focus on early adolescence because, in general, it is a critical stage of dramatic physical, emotional, social, and neurological change that translates early experiences into future trajectories and, more specifically, represents an age group for whom substance use is uncommon, strongly sanctioned, and signals greater long-term risk ( Baum et al. 2014 ; Ellickson et al. 2003 ; Guttmannova et al. 2011 ; Mundt 2011 ; Peleg-Oren et al. 2009 ; Schulenberg et al. 2015; Stritzel et al. 2022 ). As a form of linked lives, the association between single mother families and adolescent substance use (including and especially early) should be contextualized within sociohistorical context ( Elder et al. 2003 ). As a layer of sociohistorical contextualization, this association likely varies across countries. Many country-level phenomena could moderate the link between mothers’ partnership statuses and their children’s outcomes, including more social psychological dynamics like relative comparison, but differences in policy contexts for families across countries have garnered more attention ( Heuveline et al. 2003 ; Kamerman and Kahn 1997 ). The concept of contingent protection offers a lens to consider those policy contexts by focusing on the generosity of the social safety net in a society and how much it can stand in for family or community support. If some marginalizing, disadvantaging, or stigmatizing status or circumstance places a young person at greater risk of poorer outcomes, that risk will decrease in magnitude as the strength of the social safety net in a context increases. That contingent protection occurs when the state provides buffers (e.g., income assistance, child care subsidies, parental leave) against the hardships of adolescents and their parents ( Cavanagh and Fomby 2019 ; Green et al. 2017). One recent study specifically looking at single mothers and early adolescent substance use, however, revealed unexpected consistency in this association across the U.S. and U.K., casting some doubt on contingent protection ( Stritzel et al. 2022 ). As part of this line of inquiry, there has long been concern with differentiating selection and socialization in the observed correlations between single parent family structures and adolescent outcomes within and across countries. The former captures the likelihood that the reasons women become single mothers may also affect their children regardless of how they themselves are affected ( Duncan et al. 2004 ). For example, women’s social histories (e.g., educational attainment) prior to family formation and any genetic predispositions towards poorer adjustment could lead to non-marital fertility and post-fertility union dissolution while also directly affecting their children’s wellbeing ( Carneiro et al. 2013 ; Hoffmann and Cerbone 2002 ; Lynskey et al. 2010 ; Smock and Greenland 2010 ). The latter captures mothers’ role in creating and managing children’s family, school, and community ecology. For example, single mothers face more challenges than partnered mothers accumulating resources—practical (e.g., money to create positive opportunities), parenting-related (e.g., monitoring adolescent behavior and peer activities), and climate-related (e.g., protection from strain and conflict)—to support positive development ( Conger et al. 1997 ; McLanahan and Sandefur 1994 ). Amidst such resource disparities, adolescents may engage in substance use as self-medication or because they do not have as many controls on their propensity for risk-taking ( Cavanagh 2008 ; Crawford and Novak 2002 ; Flewelling and Bauman 1990 ; Hoffmann 2002 ; Turner et al. 1991 ; Wagner et al. 2010 ). For the most part, research in this area has dealt with selection in order to focus on socialization. In this tradition, studies treat selection as a statistical nuisance to be assessed or addressed in order to isolate socialization and better understand why single parent family structures matter and how social stratification operates ( Carlson 2006 ; Cross et al. 2022 ; Frisco et al. 2007 ). Less common has been the explicit study of selection, treating it as theoretically meaningful in its own right and requiring careful elucidation and cross-context comparisons that might lead to different but important policy implications ( Cavanagh and Fomby 2019 ; Lee and McLanahan 2015 ). Past research already has documented the dynamic association between single mother family structures and early adolescent substance use that points to potential socializing effects while also demonstrating the similarity of this association across two similar countries ( Stritzel et al. 2022 ). A complement to that literature is research that documents selection mechanisms underlying this association and explores how they vary across countries. Elucidating and Comparing Mechanisms of Selection Into Single Mother Families Following this charge, the first goal of this study is to assess four mechanisms of selection beyond basic sociodemographic factors (e.g., gender, race/ethnicity). All four focus on women’s life course experiences that may influence their entry into single motherhood while also having meaning for how their early adolescents behave. First, maternal family formation refers to the broader pathways of partnership and fertility that women construct over time, particularly such potentially destabilizing experiences as younger age at first birth, non-marital fertility, and union dissolution ( Doherty and Needle 1991 ; Ermisch and Pevalin 2005 ; Lichter and Graefe 2001 ; McLanahan and Sandefur 1994 ; Poortman 2007 ). Second, maternal socioeconomic attainment refers to a constellation of truncated educational attainment, low income, and unemployment that situates women in disadvantaged positions within a system of stratification ( Cross et al. 2022 ; Hillman et al. 1993 ; Miech and Chilcoat 2005 ; Musick et al. 2012 ; Oppenheimer 1988 ). Third, maternal adjustment and functioning refers to the basic level of women’s wellbeing, such as poorer mental health and their own substance use, that may seep into other parts of their lives ( Bahr et al. 2005 ; Cortes et al. 2009 ; Menasco and Blair 2014 ). Fourth, there is a set of factors that are harder to discern in observational data, such as personality traits (e.g., neuroticism), genetic predispositions (e.g., volatility), and biosocial processes (e.g., stress responsiveness), that can undermine how women relate to others and their environments, and these unobserved traits need to be assessed even if only collectively ( Deak and Johnson 2021 ; Jockin et al. 1996 ; Mcgue and Lykken 1992 ). Across these mechanisms, women’s experiences can influence their likelihood of single motherhood. These experiences also can create family stressors and reduce family resources in ways that might encourage early adolescents’ coping through risky behavior or weaken social controls on such behavior. As such, early adolescents with single mothers may engage in substance use even if their family structure is not the sole or partial reason for this engagement during this critical developmental period. After elucidating these selection mechanisms, our second goal is to compare them across three selected countries: Australia, the U.K., and the U.S. Although the U.S. is more racially diverse than the other countries, all three have majority White and English-speaking populations. These countries are all wealthy and have “liberal” welfare regimes that are more dependent than other affluent countries on informal (versus state) protections and more tolerant of inequality ( Bradbury et al. 2015 ; Kamerman and Kahn 1997 ). Within these similarities, however, the three countries have key differences. For example, the U.S. spends the lowest proportion of gross domestic product (GDP) on child and family policies (e.g., parental leave, child care and preschool services, health care, tax breaks) of the three, particularly relative to Australia (OECD 2022). Following contingent protection, therefore, single mothers and their children would have fewer policy buffers against the challenges associated with this family structure in the first country than the other two. How might this country-level process of contingent protection moderate the role of selection in associations between single mother family structures and early adolescent substance use? The first two pathways of influence on single motherhood and early adolescent substance—women’s family formation and socioeconomic histories—should make a bigger difference in the U.S. than in Australia or the U.K. That is because the U.S. offers families fewer state supports, protections, and investments to help them avoid being trapped in broad systems of social and economic stratification ( Adema et al. 2016 ; Boddy et al. 2011 ; Cowan and Cowan 2008 ; Woods 2012 ). For example, while all mothers in the U.K. and Australia receive some duration of paid parental leave, most mothers in the U.S. are unpaid, if they even qualify for leave at all. As other examples, child care is less broadly subsidized by the state in the U.S., which is the only one of the three to lack universal health care ( Adema et al. 2016 ; Klerman et al. 2012 ). Consequently, potential disadvantages in American mothers’ pasts—such as teen fertility and poverty—have more latitude to undermine their partnerships while posing more unchecked risks to their children as they leave childhood. At the same time, the second two mechanisms of selection—women’s adjustment and unobserved personal traits—could make a bigger difference in Australia or the U.K. than in the U.S. Because the somewhat stronger safety net in the former two countries has more power to break the translation of the aforementioned systemic disadvantages into single motherhood and early adolescent problems ( Daly and Ferragina 2018 ; Glass et al. 2016 ; Thévenon 2011 ), personal attributes and experiences less widely targeted by large-scale social policy become relatively more important reasons why women enter single motherhood and their children grow up in conditions conducive to early substance use. Study Aims and Research Questions In the conceptual model for this study ( Figure 1 ), substance use in the critical etiological period of early adolescence among young people living in single mother families reflects, at least in part, four sets of observable and unobservable mechanisms of selection into this family structure, but the power of these four selection mechanisms varies according to differences in the policy context of three countries. Based on this conceptual model, the first aim of this study is to identify which selection mechanisms account for most of the apparent association between single mother family structure and early adolescent substance use, and the second is to test whether the family formation and socioeconomic history mechanisms matter more in the U.S. while the maternal adjustment and unobserved traits mechanisms matter more in the U.K. and Australia. Figure 1. Open in a new tab Conceptual model of cross-country selection mechanisms in links between single mother families and early adolescent substance use In pursuing these aims, we account for two additional complications. First, ample research documents that not all maternal partnerships play the same role in youth development. For example, mothers in cohabiting relationships have a second adult figure in the home to share finances and help with child care, housework, and other family responsibilities, but, at the same time, mothers’ cohabiting partners often are less involved with young people and provide less support to mothers in actively parenting children than married partners. Thus, young people raised in families with mothers and unmarried partners often benefit from some aspects of a two-parent family structure while less fully realizing the advantages of such two-parent families, suggesting an “in-between position” when comparing developmental outcomes by family structure ( Carlson and Berger 2013 ; Cavanagh and Fomby 2019 ; Brown and Rinelli 2010 ). Second, different forms of substance use have their own etiologies that might condition the impact of an ecological factor. For example, highly social activities like alcohol use often are more reactive to parental control than other forms of substance use that tend to be less associated with adolescent social life and “partying” ( Schulenberg et al. 2014 ; Trucco and Hartmann 2021 ). Consequently, these aims will involve comparisons of maternal singlehood, cohabitation, and partnership across early adolescent alcohol and marijuana use. Methods Data All three data sets regularly followed representative samples of children from birth, with mothers initially reporting on children but adolescents eventually reporting on their own. For the Australian data set, the LSAC had a birth cohort panel ( N = 5,107) consisting of approximately 20 children born per postcode in Australia sampled from Australia’s Medicare enrollment database. Their initial survey was in 2004 when the focal children were 0–1 years old, and they have been followed up biennially since that time with the most recent wave released in 2020–2021 when they were approximately age 16. For the U.K. data set, MCS children were born between September 2000 and January 2002 ( N = 18,980). They were sampled through Child Benefit Records and have been followed up in 7 waves (ages 3, 5, 7, 11, 14, 17, and 22). The U.S. data set, NLSY79-CYA, follows children born to mothers in the original NLSY79 cohort—a nationally representative survey of adolescents in the U.S. conducted by the U.S. Bureau of Labor Statistics and sampled through a multi-stage stratified area probability sample of dwelling units and group quarter units, with oversamples of Hispanic and Black populations ( N = 11,521). The Child and Young Adult cohort began in 1986 with ongoing biennial surveys (most recently released in 2022). Information about adolescent substance use was not collected prior to 1994, which, coupled with the focus on early adolescence, meant studying youth born between 1980 and 2000. For each country, analytical samples included youth who had a maternal survey at birth. Although somewhat arbitrary, early adolescence is typically defined as ending around age 14 (and, indeed, that is the cutoff used by the Journal of Early Adolescence ). Thus, we further narrowed the analytical samples to youth followed up at the wave nearest to age 14 (sometimes slightly younger or slightly older) and were still living with their biological mothers at that wave, leaving 3,033 (Australia), 11,009 (U.K.), and 5,639 (U.S.) youth. Measurement Early adolescent substance use. To harmonize measurement across the three studies, we dichotomized adolescent reports at the final wave of the two forms of substance use, differentiating any use (1) from no use (0) in the previous year for alcohol, but, given cell size limitations, differentiating ever used (1) from never used (0) for marijuana. We acknowledge the perceived limitations of such blunt indicators but also the emphasize the aforementioned heightened risk of any substance use up to this age. As such, although our harmonization limited the level of detail we could include about adolescent substance use, these measures still capture an important phenomenon of early adolescence that warrants attention. Current family structure. Based on parent-reported household roster data in each data set, the analytical sample from each country was restricted to households with the biological mother of the focal child. The current partnership status of the focal child’s biological mother at the final wave designated family structure (dummy variables for single, cohabiting, and married) in early adolescence. The biological relations of a male partner with the child could be determined for most youth in some but not all data sets. Maternal family formation history. Comparing these family structure measures across all available waves since birth allowed the measurement of duration of exposure to the current family structure; more specifically, the proportion of waves since birth that the child had been in the same family structure that they were in at the early adolescence wave. We also measured age at first birth and the partnership status of the mother at the time of the study child’s birth. Maternal socioeconomic history. Household income (within-country quintiles), educational attainment (dummy variables for less than high school, high school, college degree, advanced degree), and employment status (dummy variables for out of the labor force, less than 30 hours/week, more than 30 hours/week) captured mothers’ socioeconomic circumstances when the study child was born. This measurement at the child’s birth creates a long temporal distance relative to the focal predictor and outcomes when children were adolescents. This temporal distance was necessary to make certain the proper time-ordering among the selection mechanisms, family structures, and adolescent outcomes and to increase the likelihood that the associations we captured with these measures were more likely to represent confounding factors rather than mediational pathways. As a sensitivity check, we also estimated models with alternate versions of each of these socioeconomic factors that were measured concurrently to adolescent family structure and substance use. These models generally produced slightly stronger but otherwise similar results, with the only exception being that the contemporaneous measure of maternal employment in the U.K.—but not the one measured at the child’s birth—significantly predicted adolescent alcohol use. Maternal adjustment and functioning. All three data sets included measures of maternal mental wellbeing, but they were not directly comparable. LSAC and MCS mothers responded to a 6-item inventory on depressive symptoms experienced in the past month, with responses to all items available. The NLSY79-CYA mothers responded to the 7-item Center for Epidemiologic Studies-Depression (CES-D) inventory, with only the aggregate score available. The measures, independent of country, were all assessed during the first wave of data collection after the child’s birth. To harmonize these measures, responses to each item in the LSAC and MCS were summed into aggregate scores, and then mothers’ scores on all three measures were standardized within-country. Thus, the final measure of maternal mental wellbeing gauged relative standing to other mothers within the same country. We also measured mothers’ cigarette and alcohol use in the year prior to their children’s birth (dichotomous variables indicating engagement in each behavior versus no such engagement). Again, to probe the impact of the temporal distance between the selection mechanisms and the focal predictors and outcomes, we estimated alternate models with more contemporaneous measures of maternal mental health (there were no contemporaneous measures of substance use). These models revealed similar but stronger results with the more contemporaneous measures. Sociodemographic controls. To account for the diversity of the samples, we measured race/ethnicity (which could only be dichotomized as White versus Indigenous, non-White/immigrant, or non-White/non-immigrant given the varying composition and diversity of the three populations) and gender of the child. Pooled models also controlled for country, and the U.S.-specific models controlled for birth year, given that the U.S. sample represented children born to a cohort of women and not a birth cohort in its own rights (see the results section for more on this issue). Plan of Analyses The first set of analyses involved the estimation of basic associations between the adolescent family structure variable and the early adolescent substance use outcomes in the pooled sample across countries. Logistic regression models predicted the odds of each binary measure of substance use at the final (early adolescence) wave by the dummy variables for current family structure. Subsequent models added sociodemographic controls and then iteratively introduced each of the theoretically motivated subsets of variables tapping three of the four selection mechanisms, assessing any percentage decline in the family structure coefficients associated with each mechanism. Additionally, we calculated the Bayesian Information Criterion (BIC) for each of these stacked models. The BIC was calculated as k ln n - 2 ln L ^ , where L ^ was the maximum likelihood, k was the number of parameters in the model, and n was the sample size. Smaller BIC values indicated a better fit to the data. As such, if the addition of a block of variables better described the data than the baseline model, we should see the BIC drop between the two models, with larger changes indicating greater relative value of the additions. To gauge the fourth selection mechanism (unobserved traits), we calculated the Impact Threshold for Confounding Variables (ITCV) for any significant family structure coefficient. The ITCV quantified how much an unobserved confound would have to be correlated with both family structure and early adolescent substance use to reduce the coefficient for the former on the latter in the regression to non-significance (Frank et al. 2008). Put another way, it gauged whether unobserved variables not in the model could nullify the estimates of family structure “effects” if included. The ITCV equation was: r xy – r # xy / 1 – r # xy , where r # xy = t / SQRT[( n – q – 1) + t 2 ], t was the critical t -value, n was the sample size, and q was the number of model parameters. When covariates were included in the model, the equation became: ITCV no covariates x [SQRT (1 – R 2 xg )(1- R 2 yg )], where g was the set of covariates, R 2 xg was the R 2 value from a regression predicting the focal independent variable by the covariates, and R 2 yg was the R 2 value from a regression predicting the outcome by the covariates. The second set of analyses replicated the basic modeling steps with the sample for each country. We tested for significant differences in the coefficients for all selection variables across countries by utilizing the methods outlined by Clogg, Petkova, and Haritou (1995) , which gauged for the equality of coefficients across samples. They were essentially Wald tests for b ^ x 1 = b ^ x 2 where the null hypothesis was rejected when [ b ^ x 1 - b ^ x 2 ] / SQRT [ ( σ ^ 2 b ^ x 1 ) + ( σ ^ 2 b ^ x 2 ) ] was larger than the appropriate X 1 2 threshold. These calculations were performed using the seemingly unrelated estimation package in Stata 18. We also compared the ITCV calculations for any significant coefficients across the countries. For both sets of analyses, multiple imputation via chained equations in Stata 18 addressed all missing data. Sample weights helped to address the biases of differential attrition across waves within each data set. Results Table 1 describes the study variables for the three country-level data sets. In all countries, drinking alcohol was a more common form of substance use than smoking marijuana, even accounting for the different time scales of the variables (any use in past year for the former, any lifetime use for the latter). Drinking alcohol was significantly less common in the U.S. (19% of early adolescents having drunk in the last year) than in Australia (50%) or the U.K. (48%). Smoking marijuana was slightly more common in Australia (6% of early adolescents having ever done so) than in the U.K. (5%) and U.S. (5%). Most country-level differences were significant. Table 1. Weighted descriptive statistics of study variables by country Australia ( n = 3,033) U.K. ( n = 11,009) U.S. ( n = 5,639) Mean/Prop SD Mean/Prop SD Mean/Prop SD Early adolescent substance use Consumed alcohol in last 12 months 0.497 0.483 a 0.187 a b Ever used marijuana 0.064 0.051 a 0.052 b Mother’s current partner status Single 0.198 0.279 a 0.257 a b Cohabiting 0.143 0.192 a 0.055 a b Married 0.659 0.529 a 0.688 a Sociodemographic controls Immigrant/Indigenous/Non-White 0.036 0.192 a 0.216 a b Child gender (female) 0.488 0.485 0.494 Birth year (years after 1980) 8.345 5.385 Maternal family formation history Proportion waves in current partner status 0.717 0.403 0.805 a 0.273 0.806 a b 0.284 Age at first birth 28.062 5.558 25.112 a 5.680 24.278 a b 5.143 Partner status at child’s birth Single 0.091 0.185 a 0.167 a b Cohabiting 0.162 0.271 a 0.063 a b Married 0.748 0.545 a 0.770 b Maternal socioeconomic history Highest education Less than high school 0.043 0.159 a 0.235 a b High school diploma 0.526 0.544 a 0.462 b College degree 0.319 0.169 a 0.187 a b Advanced degree 0.112 0.128 a 0.116 a b Income quintile 1–20% 0.194 0.239 a 0.141 a b 21–40% 0.203 0.202 a 0.140b 41–60% 0.202 0.185 a 0.176 a 61–80% 0.203 0.179 a 0.256 b 81–100% 0.198 0.194 a 0.287 b Mother’s employment status Unemployed 0.579 0.701 a 0.246 a b Employed, less than 30 hours/week 0.248 0.148 a 0.197 a b Employed, more than 30 hours/week 0.173 0.151 a 0.556 a b Maternal adjustment Mental health (standardized) −0.011 1.010 0.062 a 1.029 −0.073 a b 0.925 Cigarette use during pregnancy 0.206 0.371 a 0.316 a Alcohol consumed during pregnancy 0.800 0.811 a 0.458 a b Open in a new tab Notes: Significant differences determined by one-way ANOVA. a significantly different from Australia ( p < .05). b significantly different from U.K. ( p < .05). Independent of country, living with a married mother and her partner around age 14 was the most common family structure that adolescents experienced, but this family structure was significantly more common in the U.S. and Australia (69% and 66% respectively) than in the U.K. (53%). In the U.S., 26% of early adolescents lived with a single mother while 6% lived in homes with a mother and her coresidential but unmarried partner. The corresponding percentages were 20% and 14% in Australia and 28% and 19% in the U.K. (with all country-level differences, except the proportion with married mothers between the U.S. and Australia, reaching statistical significance). Thus, single motherhood and maternal cohabitation were most common in U.K., with the latter accounting for a larger share of the adolescents not living with a married mother in that country than in other countries. The difference of this U.K.-U.S. comparison from current population data (which shows higher rates of single parent families in the U.S.) likely reflects the older cohort structure of the U.S. data set. Recall that the focus of this study was on mechanisms of selection into single mother families that might also influence early adolescent substance use. The multivariate models that we present below highlight how selection factors were associated with early adolescent substance and how these associations accounted for associations between family structure and early adolescent substance use. Yet, the selection argument also rests on the degree to which the selection factors also predict family structure. Thus, before getting to those focal multivariate models of early adolescent substance use, we also estimated logistic models predicting the likelihood of living in a single mother family around age 14 by the selection factors in general (also controlling for country) and within each country. These results informed how we identified variables as key selection factors in addition to whether, as described below, they predicted the substance use outcomes and how their inclusion in the substance use models attenuated coefficients for family structure in those models. In the pooled sample, the odds of living in a single mother family at this age significantly increased with shorter duration in the current family structure, younger maternal age at birth, non-marital birth, lower maternal incomes and educational attainment, and more maternal depression at birth and substance use during pregnancy. These patterns largely carried over to the models with the U.K. and U.S. The most notable cross-country difference was the lack of socioeconomic differentiation in single mother families in Australia, and controlling for race/ethnicity and socioeconomic factors also revealed greater odds of being in a single mother family for youth in the U.S. than in the other countries. Patterns of Early Adolescent Alcohol Use by Family Structure Table 2 presents the results of logistic regression models for alcohol use around age 14 in the pooled sample. The baseline model (Model 1) shows that early adolescents with married mothers had 34.4% lower odds of having drunk in the past year than their peers with a single mother ( p < .001). In contrast, early adolescents with cohabiting mothers had odds that were not statistically significantly different than their peers with a single mother (or relative to their peers with married mothers, as revealed by alternate models rotating the reference category). Model 2 added basic demographic variables, which actually increased the association between having a cohabiting mother versus a single mother by 14.5%. The remaining models in Table 2 iteratively and then collectively tested the focal selection mechanisms. Table 2. Odds ratios from logistic regressions predicting alcohol use the previous year in early adolescence by current family structure across countries Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 OR SE OR SE OR SE OR SE OR SE OR SE Mother’s current partner status Cohabiting 1.107 (0.057) 0.987 (0.053) 1.018 (0.056) 1.073 (0.057) 1.056 (0.056) 0.929 (0.053) Married 0.656 *** (0.024) 0.606 *** (0.023) 0.785 *** (0.036) 0.604 *** (0.023) 0.729 *** (0.028) 0.783 *** (0.037) Country (vs. U.S.) Australia 4.462 *** (0.229) 3.085 *** (0.165) 4.464 *** (0.242) 4.514 *** (0.257) 3.449 *** (0.186) 3.071 *** (0.191) U.K. 3.466 *** (0.138) 2.750 *** (0.114) 3.471 *** (0.144) 3.679 *** (0.175) 2.636 *** (0.111) 2.565 *** (0.131) Demographic controls Immigrant/Indigenous/Non-White 0.418 *** (0.018) 0.562 *** (0.026) Child gender (female) 1.006 (0.032) 1.022 (0.033) Maternal family formation history Proportion of waves in current partner status 0.716 *** (0.045) 0.741 *** (0.048) Age at first birth 0.996 (0.003) 0.986 *** (0.003) Partner status at birth (vs. single) Cohabiting 1.218 ** (0.070) 0.997 (0.061) Married 0.927 (0.048) 0.840 *** (0.047) Maternal socioeconomic history Highest education (vs. high school) Less than high school 1.051 (0.067) 1.059 (0.072) Bachelor's degree 0.941 (0.043) 1.007 (0.048) Advanced degree 0.839 ** (0.045) 0.916 (0.052) Employment status (vs. unemployed) Employed, less than 30 hrs/week 1.129 * (0.062) 1.059 (0.059) Employed, more than 30 hrs/week 0.973 (0.048) 0.968 (0.049) Income quintile (vs. median quintile) 1–20% 0.668 *** (0.035) 0.760 *** (0.042) 21–40% 0.951 (0.050) 0.984 (0.053) 61–80% 1.189 ** (0.062) 1.173 ** (0.062) 81–100% 1.163 ** (0.060) 1.150 ** (0.061) Maternal adjustment Mental health (standardized) 1.064 *** (0.018) 1.063 *** (0.018) Substance use during pregnancy Cigarettes 1.417 *** (0.051) 1.358 *** (0.052) Alcohol 2.484 *** (0.100) 2.046 *** (0.087) Constant 0.290 *** (0.011) 0.460 *** (0.023) 0.267 *** (0.014) 0.309 *** (0.021) 0.154 *** (0.008) 0.230 *** (0.021) Percentage change in effect size Cohabiting −112.150 *** −83.178 *** −31.776 *** −47.664 *** −166.355 *** Married 14.535 *** −37.500 *** 15.116 *** −21.221 *** −36.919 *** BIC 24589.290 24147.290 24550.300 24491.070 23869.930 23679.720 Change in BIC −442.000 −38.990 −98.220 −719.360 −909.570 Open in a new tab Notes: The outcome measured any use in the prior year. n = 19,682 * p < 0.05; ** p < 0.01; *** p < 0.001 Beginning with the maternal family formation mechanism, Model 3 revealed significantly greater odds of early adolescents drinking when their parents had cohabited at their births and significantly lower odds when they had been in the current family structure for longer. These additions attenuated the previously observed association between single mother family structure and early adolescent alcohol use reported for Model 1 by 38%, and the decline in the BIC suggested that these additions improved model fit. Model 4 added the variables for maternal socioeconomic history, with significantly greater odds of early adolescent alcohol use in the context of part-time maternal employment and higher family income and significantly weaker odds in the context of advanced maternal education. In total, these additions actually strengthened the previously observed association between single mother family structure and early adolescent alcohol use from Model 1 by 15%, with a larger decline in the BIC indicating meaningful improvement to model fit. Model 5 tested the maternal adjustment mechanism, revealing higher odds of alcohol use for early adolescents with mothers who reported more depression and substance use around the time the adolescents were born. These additions attenuated the previously observed association between single mother family structure and early adolescent alcohol use from Model 1 by 21%, and the decline in the BIC suggested that these additions improved model fit. Not surprisingly, the comprehensive model (Model 6) also revealed attenuation of the maternal marriage coefficient (37%) and a large decline in the BIC. To explore the fourth selection mechanism (unobserved characteristics), we calculated the ITCV for the maternal marriage coefficient in the comprehensive model. This coefficient had an ITCV of .051, which would indicate that some confound or set of confounds not included in the models would have had to be correlated with maternal marital status and early adolescent alcohol use at a level of .227 to nullify the finding. That is a strong level of robustness, although it leaves room for unobserved confounds to be at play. Table 3 provides an overview of results from the same modeling series estimated within each country. In Australia, the married versus single mother contrast was statistically significant and larger than in the pooled sample, and it was attenuated by 13% between the baseline model (1) and comprehensive model (6). Maternal adjustment appeared to be the only important selection mechanism in terms of the attenuation of the baseline maternal marriage coefficient (10%) and the decline in BIC. The ITCV for this coefficient in Model 6 was stronger in Australia (.08, or .28 x .28 correlations) than in the pooled sample, indicating less room for unobserved confounds. Australia also differed from the pooled sample in early adolescents’ significantly lower odds of drinking in homes with a cohabiting (versus single) mother, a difference that also appeared to be more closely related to maternal adjustment than the other selection mechanisms. Table 3. Odds ratios from logistic regressions predicting alcohol use in early adolescence by current family structure in three countries OR (SE) Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Australia Mother’s current partner status Cohabiting 0.621 ** 0.620 ** 0.604 ** 0.612 ** 0.617 ** 0.597 ** (0.103) (0.103) (0.114) (0.105) (0.104) (0.117) Married 0.534 *** 0.535 *** 0.538 * 0.526 *** 0.579 *** 0.594 * (0.065) (0.065) (0.130) (0.066) (0.072) (0.144) Percentage change in effect size Cohabiting 0.264 4.485 2.375 1.055 6.332 Married −0.215 −0.858 1.717 −9.657 ** −12.876 BIC 4200.723 4213.319 4230.144 4263.637 4172.683 4279.017 Change in BIC 12.596 29.421 62.917 −28.040 78.294 U.K. Mother’s current partner status Cohabiting 1.154 1.001 1.095 1.103 1.052 0.962 (0.097) (0.086) (0.098) (0.093) (0.092) (0.089) Married 0.708 *** 0.683 *** 0.879 0.629 *** 0.791 *** 0.846 * (0.044) (0.043) (0.066) (0.041) (0.052) (0.068) Percentage change in effect size Cohabiting −99.351 *** −38.312 −33.117 *** −66.234 *** −124.675 *** Married 8.562 −58.562 *** 27.055 *** −28.425 *** −47.260 *** BIC 14950.100 14329.940 14885.060 14762.500 14080.660 13895.230 Change in BIC −620.160 −65.040 −187.600 −869.440 −1054.870 U.S. Mother’s current partner status Cohabiting 0.839 0.818 0.712 0.814 0.774 0.672 * (0.148) (0.145) (0.130) (0.146) (0.141) (0.126) Married 0.630 *** 0.596 *** 0.842 0.693 *** 0.699 *** 0.804 * (0.055) (0.056) (0.087) (0.067) (0.063) (0.086) Percentage change in effect size Cohabiting 13.043 78.882 *** 15.528 40.373 ** 103.727 *** Married 9.189 −57.297 *** −17.027 −18.649 *** −47.027 *** BIC 5443.662 5446.360 5423.451 5503.469 5432.535 5495.493 Change in BIC 2.698 −20.211 59.807 −11.127 51.831 Open in a new tab Note: * p < 0.05; ** p < 0.01; *** p < 0.001 n = 3,033 (Australia), 11,008 (U.K.), 5,640 (U.S.) Model 2 added the sociodemographic controls to Model 1. Model 3 added the maternal family formation variables to Model 1. Model 4 added the maternal socioeconomic history variables to Model 1. Model 5 added the maternal adjustment variables to Model 1. Model 6 included all possible predictors. The U.K. results mostly replicated the results from the pooled model, which is not surprising given that the U.K. sample was the largest. Among early adolescents in this country, drinking was significantly higher in single parent families compared to married families, with 47% attenuation in these odds when all other factors were taken into account. The additions of maternal adjustment and family formation further attenuated these associations (see the declining BIC), and the addition of maternal socioeconomic history appeared to have a suppressor effect that also led to a declining BIC. The ITCV for the maternal marriage coefficient was smaller (.04, or .20 x .20 correlations) in the U.K. than in the pooled sample, indicating more room for unobserved confounds. The difference in alcohol use for early adolescents in single mother versus married homes was larger in the U.S. than in the U.K. but had similar levels attenuation between Models 1 and 6. This attenuation again reflected maternal adjustment and family formation (according to significant attenuation and declining BIC), with no evidence of the socioeconomic suppressor effect, and the ITCV of .05 was similar to the pooled sample. By Model 6, a significant contrast between early adolescents with cohabiting versus single mothers had emerged, but this model had poorer model fit than Model 1. Across all three countries, therefore, early adolescents in single mother families drank more alcohol than their peers with married mothers, with those living with cohabiting mothers falling in between (except for in Australia). Also in all three countries, this family structure difference was attenuated by accounting for key mechanisms of observable selection (although perhaps less so in Australia), and maternal adjustment appeared to be the most important of these selection mechanisms. In the U.S. and U.K., maternal family formation also seemed to matter, but there was also an apparent suppressor effect of maternal socioeconomic history in these countries (likely driven by more use associated with higher family income while single mother families tended to have lower incomes). Unobserved confounds did seem to be at least somewhat implicated in this family structure difference, although perhaps less so in Australia. Patterns of Early Adolescent Marijuana Use by Family Structure Table 4 present the results of logistic regression models for marijuana use around age 14 in the pooled sample. According to the baseline model (Model 1), having a married mother was associated with reduced odds of early adolescent substance use (61.0%, p < .001), while having a cohabiting mother was not associated with significantly different odds. This family structure difference was attenuated by 26% between the baseline model (1) and the comprehensive model (6). This attenuation appeared to be driven by maternal adjustment (specifically, maternal depression and substance use) and maternal family formation history (specifically, less duration in current family structure, younger maternal age at birth, and maternal non-marital fertility), the addition of which significantly attenuated the previously observed association between maternal marriage and early adolescent marijuana use in Model 1 (by 20% and 12% respectively) and decreased the BIC. The ITCV for maternal marriage in Model 6 was .058, indicating that an unobserved confound would have to be simultaneously correlated with both maternal marriage and early adolescent marijuana use at .24 or higher for its inclusion in the model to reduce this coefficient to non-significance. Table 4. Odds ratios from logistic regressions predicting marijuana use the previous year in early adolescence by current family structure across countries Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 OR SE OR SE OR SE OR SE OR SE OR SE Mother’s current partner status Cohabiting 0.896 (0.087) 0.888 (0.087) 0.799 * (0.081) 0.900 (0.088) 0.883 (0.087) 0.812 * (0.083) Married 0.390 *** (0.029) 0.387 *** (0.029) 0.514 *** (0.046) 0.405 *** (0.032) 0.461 *** (0.035) 0.548 *** (0.050) Country (vs. U.S.) Australia 1.223 * (0.122) 1.179 (0.128) 1.387 ** (0.151) 1.223 (0.139) 1.213 (0.128) 1.243 (0.158) U.K. 0.805 ** (0.063) 0.786 ** (0.066) 0.863 (0.071) 0.786 ** (0.073) 0.732 *** (0.060) 0.740 ** (0.076) Demographic controls Immigrant/Indigenous/Non-White 0.934 (0.080) 1.038 (0.096) Child gender (female) 0.884 (0.060) 0.889 (0.060) Maternal family formation history Proportion of waves in current partner status 0.712 ** (0.085) 0.717 ** (0.086) Age at first birth 0.971 *** (0.006) 0.975 ** (0.007) Partner status at birth (vs. single) Cohabiting 1.129 (0.120) 1.110 (0.121) Married 0.813 * (0.080) 0.882 (0.092) Maternal socioeconomic history Highest education (vs. high school) Less than high school 1.055 (0.121) 0.920 (0.110) Bachelor's degree 0.887 (0.094) 1.044 (0.114) Advanced degree 0.936 (0.116) 1.153 (0.151) Employment status (vs. unemployed) Employed, less than 30 hrs/week 1.052 (0.111) 1.053 (0.112) Employed, more than 30 hrs/week 0.872 (0.092) 0.881 (0.094) Income quintile (vs median quintile) 1–20% 1.149 (0.131) 1.078 (0.126) 21–40% 1.231 (0.140) 1.199 (0.138) 61–80% 1.129 (0.133) 1.189 (0.140) 81–100% 1.163 (0.141) 1.305 * (0.160) Maternal adjustment Mental health (standardized) 1.129 *** (0.037) 1.124 *** (0.037) Substance use during pregnancy Cigarettes 1.808 *** (0.130) 1.716 *** (0.131) Alcohol 1.247 ** (0.105) 1.231 * (0.107) Constant 0.093 *** (0.006) 0.103 *** (0.010) 0.083 *** (0.008) 0.085 *** (0.012) 0.061 *** (0.005) 0.054 *** (0.010) Percentage change in effect size Cohabiting 7.692 93.269 *** −3.846 12.500 80.769 ** Married 0.492 −20.328 *** −2.459 −11.639 *** −25.902 *** BIC 7537.871 7553.542 7527.908 7613.953 7461.079 7568.272 Change in BIC 15.671 −9.963 76.082 −76.792 30.401 Open in a new tab Notes: The outcome measured any lifetime use. n = 19,682 * p < 0.05; ** p < 0.01; *** p < 0.001 The within-country models in Table 5 indicate a consistent contrast between single mother and married mother family structures in early adolescent marijuana use, perhaps slightly stronger in Australia. The attenuation of this family structure difference by the selection mechanisms was larger in the U.K. (31%) than in the other two countries (21%), and the only case in which this significant attenuation decreased the BIC was in the U.K. (for maternal adjustment). The ITCV for maternal marriage in Model 6 was smaller in the U.K. (.047) and larger in the U.S. (.065). There was no significant maternal marriage coefficient in Model 6 in Australia, indicating no more room for unobserved confounds (and the corresponding ITCV values for this coefficient in the other modeling steps was consistently stronger). Table 5. Odds ratios from logistic regressions predicting marijuana use in early adolescence by current family structure in three countries OR (SE) Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Australia Mother’s current partner status Cohabiting 0.708 0.707 0.541 0.724 0.708 0.569 (0.196) (0.196) (0.182) (0.201) (0.200) (0.194) Married 0.334 *** 0.342 *** 0.446 * 0.357 *** 0.391 *** 0.475 (0.072) (0.072) (0.178) (0.073) (0.086) (0.193) Percentage change in effect size Cohabiting 0.342 57.192 ** −5.479 0.000 47.603 Married −1.201 −16.817 −3.453 −8.559 *** −21.171 BIC 1357.777 1370.910 1377.541 1421.831 1362.822 1463.628 Change in BIC 13.133 19.764 64.054 5.045 105.851 U.K. Mother’s current partner status Cohabiting 0.989 0.978 0.888 0.999 0.988 0.920 (0.166) (0.168) (0.153) (0.168) (0.169) (0.164) Married 0.375 *** 0.374 *** 0.529 *** 0.386 *** 0.492 *** 0.567 *** (0.052) (0.052) (0.083) (0.056) (0.072) (0.092) Percentage change in effect size Cohabiting 100.000 918.182 −90.909 9.091 627.273 Married 0.160 −24.640 *** −1.760 −18.720 *** −30.720 *** BIC 3862.581 3880.142 3868.900 3941.255 3810.044 3924.180 Change in BIC 17.561 6.319 78.674 −52.537 61.599 U.S. Mother’s current partner status Cohabiting 0.750 0.736 0.648 0.724 0.701 0.618 (0.207) (0.204) (0.189) (0.202) (0.197) (0.183) Married 0.371 *** 0.356 *** 0.505 *** 0.411 *** 0.431 *** 0.500 ** (0.054) (0.055) (0.098) (0.067) (0.067) (0.101) Percentage change in effect size Cohabiting 5.600 40.800 ** 10.400 19.600 52.800 ** Married 2.385 −21.304 ** −6.359 −9.539 *** −20.509 ** BIC 2345.770 2361.119 2360.354 2410.528 2350.690 2457.168 Change in BIC 15.349 14.584 64.758 4.920 111.398 Open in a new tab Note: * p < 0.05; ** p < 0.01; *** p < 0.001 n = 3,033 (Australia), 11,008 (U.K.), 5,640 (U.S.) Model 2 added the sociodemographic controls to Model 1. Model 3 added the maternal family formation variables to Model 1. Model 4 added the maternal socioeconomic history variables to Model 1. Model 5 added the maternal adjustment variables to Model 1. Model 6 included all possible predictors. Across all three countries, therefore, early adolescents in single mother families were more likely to have used marijuana than their peers with married mothers, with those living with cohabiting mothers falling in between no matter what the country. Again, the selection mechanisms accounted for some portion of this family structure difference in all three countries, less so in Australia, and maternal adjustment appeared to be the most consistent factor in selection. There was room for unobserved confounds to play a role also in the U.S. and U.K. but not Australia. Ancillary Analyses Given the nature and distribution of the data across the countries, we estimated several additional models to assess the sensitivity of the results presented above. First, because of differences in sample sizes across the three panel studies, U.K. youth accounted for more than half of the pooled sample. To assess the potential for this distribution to bias the results, we re-estimated models with a random sample of youth from the U.K. drawn to match the number of early adolescents from the U.S. sample. Results from these models saw a slight attenuation of effect sizes relative to the models with the larger pooled sample including more U.K. youth. In the pooled models, for example, the association for having a married mother changed from a 34.4% reduction in the odds of drinking to a 33.7% reduction, but there were no changes in direction or significance. Second, the U.S. panel study did not follow a true birth cohort of children over time. Instead, it followed children born at different times across a 44-year period to women from a nationally representative sample of adolescents and young adults. This sampling means that some of the early adolescents in the U.S. turned 14 as early as 1994, while others did not turn 14 until as late as 2014. This uniqueness of the U.S. sample is why the models presented above controlled for birth year. To go further, we re-estimated the models while restricting the U.S. sample to only youth early adolescents born within 10 years of the birth of the cohorts from the samples of the other two countries. The results from this set of models were similar to the results from the models presented above in both direction and magnitude—for example, the odds ratio for married versus single mother in the U.S. for early adolescent marijuana use changed from 0.371 to 0.393—but saw slight decreases in the levels of significance while still remaining statistically significant. Conclusions Demographers have long emphasized that observed family structure disparities often reflect the strong selectivity of family structure itself, with social and personal circumstances that lead to certain family structures also implicated in the developmental trajectories of the people in them ( Cross et al. 2022 ). This emphasis on selection is not solely important for isolating family structure effects and understanding why family structure matters. It also is important for grounding family structure disparities within the social, economic, political, and interpersonal systems that shape and stratify the opportunities and challenges of parents and children. Such grounding can advance policy intervention by using family structure as a proxy for refining services and supports for young people (and their parents) in need. This study, therefore, tested conceptually defined mechanisms of selection into single mother families that might also matter to early adolescent behavior but with the twist that the role of these mechanisms might vary across countries. We did so with a focus on the critical period of early adolescence in which any emergence of substance use occurs on top of the developmental upheaval of puberty, is misaligned with normative maturation, is viewed as problematic by adults and institutions and sanctioned accordingly, and signals distress more than social engagement ( Ellickson et al. 2003 ; Guttmannova et al. 2011 ; Schulenberg et al. 2015) Our first aim was to identify which selection mechanism led to correlations between family structure and early adolescent substance use independent of the causal impact of the former on the latter. The key distinction for early adolescent substance was between living with a single mother and a married mother. Although this family structure difference appeared to be rooted in all three sets of observable selection factors that we studied, it was most consistently and robustly confounded by mothers’ own mental health and substance use when they had children (maternal adjustment) followed by their life course patterns of marriage and fertility (maternal family formation). Aspects of their socioeconomic histories also were generally key confounds, at least for their alcohol use, but taking into account these histories tended to reveal stronger (rather than weaker) associations between family structure and early adolescent drinking. That likely was because of the unique nature of alcohol use as a domain of maladjustment during adolescence, when apparent problems are often grounded in socioeconomic advantages (like higher family incomes) rather than disadvantages as is seen with other domains of maladjustment like poorer mental health or delinquent behavior ( Humensky 2010 ; Luthar et al. 2018 ). Notably, unobservable confounds also seemed to matter to the associations between family structure and early adolescent substance use, although these associations appeared to be fairly robust in the face of such confounds. Our second was to test hypotheses about the countries in which selection mechanisms would be most and least important. We had expected that maternal family formation and socioeconomic histories would matter more in the U.S., given its weaker policy protections against systematic inequality, than in Australia or the U.K. but that maternal adjustment and unobserved factors would matter less in the U.S., given the greater opportunity for various selection factors to have competing effects in this less supportive policy context. Instead, we found more similarities than differences in the patterns of selection, family structure, and early adolescent substance use across these three countries. When differences emerged, Australia often stood out. Disparities between single mother and married mother families appeared somewhat stronger in this country, where a disparity between cohabiting mother and single mother families also emerged for alcohol use. Moreover, these disparities appeared to be somewhat less rooted in observable and unobservable selection factors in Australia. We had argued that, of the three, this country had the strongest policy buffers for parents and early adolescents facing hardships, and perhaps such protections created a context in which developmental problems related to family structure were more likely to reflect family structure itself rather than the other inequalities that might shape family structure. The anticipated distinctions between selection rooted in systematic stratification and selection reflecting personal processes may have been diluted in such a context. One takeaway is that explicitly exploring selection reinforced the appearance that maternal marriage—more than a mother simply having a partner—is an apparent protective factor against early adolescent substance use. This pattern may reflect a generally lower level of parental investment from cohabiting partners that creates a middle ground between having a partner in parenting and not ( Berger et al. 2008 ; Carlson 2006 ; Carlson and Berger 2013 ). Indeed, recent evidence from the same U.S. sample indicates that cohabiting partners invest less in children than do married partners in ways that matter to key life course outcomes ( Murray and Montelpare 2022 ). Yes, these patterns reflect selection mechanisms that need to be taken into account, but they tend to be fairly robust overall in ways that point to a role for differential socialization processes within single, cohabiting, and married mother families. Another takeaway is that selection mechanisms seemed to be more important in linking family structure to early adolescent alcohol use than marijuana use, no matter the country. Perhaps that is because, although both early adolescent alcohol and marijuana use are legally prohibited for early adolescents in all countries, the social sanctioning of the former tends to be weaker in these countries than the latter (and Australia and the U.K. allow older adolescents to drink). There is the possibility that such differences reflect statistical problems associated with common versus rarer behaviors, but the differential social control of different forms of early adolescent substance use could mean that family structure disparities in early drinking reflect many kinds of influence while the corresponding disparities in early marijuana use isolate something specifically going on in the family. These takeaways are relevant to ongoing policy debates about family structure in three ways. First, the strong role of selection in disparities in early adolescent substance use by single mother family structure lends support to the philosophy behind two-generation policies: investing in the stability and security of mothers might reduce developmental disadvantages and create developmental opportunities for their children. In contrast to most other developmental domains, though, early adolescent substance use might be more reactive to investments in the social and emotional (versus economic) wellbeing of mothers. Second, the general pattern in which selection did not appear to completely account for these disparities left open room for family-focused intervention. At least at the broad population pattern (versus within specific racial/ethnic groups), there may be some potential for policies and programs that directly target family circumstances and processes after children’s births to reduce early substance use. Third, the weaker-than-expected role of selection in Australia could suggest that policy contexts reducing single motherhood might change the nature of this family structure experience when it does occur. In other words, the single mother families that emerged in Australia might have had internal dynamics and circumstances not seen in the other countries where such families were more common, and they also might have drawn more stigmatization (from self and other). Of course, any policy relevance of this kind of research requires acknowledgement of its limitations and a call for future research to address them. For example, isolating causal impact is notoriously difficult with observational data, and our examination of selection did not solve that problem. More advanced statistical techniques are needed to attributing family structure disparities to selection and socialization ( Morgan and Winship 2014 ). Yet, recall that our goal was not to control for selection but instead to unpack it, and many such methods take care of selection effects collectively rather than iteratively comparing them. In this sense, this study treating selection as a conceptual (and policy) issue to be explored can be seen as a complement to studies that have treated selection as a statistical problem to be solved and accordingly employed those advanced techniques (Stritzel et al. 20220). Also notoriously difficult is harmonizing data cross data sets for cross-country comparisons, often leading to simpler measurement (e.g., dichotomizing substance use) or other compromises (e.g., using country-relative scores for maternal mental health) while accepting differences in sampling strategies and timeframes. We certainly dealt with those challenges here, but we think that what we have done lays a foundation on which future studies can build. As a final example, we had good reasons to focus on single mother families, but that focus does not minimize the experiences of single fathers and their children or preclude us from advocating for future research to apply what we have done here to this important segment of the family population. Improving and extending this line of research is important, as family structure is a meaningful component of the developmental ecology that also is intricately entangled with other ecological systems both more macro-level (e.g., economic stratification) and micro-level (e.g., personality traits). As a result, determining how young people’s trajectories reflect processes and condition that unfold as a result of what their families look like and how these processes and conditions occur as a result of other phenomena that make their families look like they do is difficult to assess. Advancing that determination, however, is important because it identifies not just who is at risk but why, helping us figure out where resources and supports need to go. Acknowledgments The authors acknowledge the support of grants from the National Institute on Drug Abuse (1R03DA046046-01A1, PI: Robert Crosnoe), the Eunice Kennedy Shriver National Institute of Child Health and Human Development (P2CHD042849, PI: Debra Umberson; T32HD007081, PI: Shannon Cavanagh and Bridget Goosby), and the National Science Foundation (1519686; PIs: Elizabeth Gershoff and Robert Crosnoe) to the Population Research Center at the University of Texas at Austin. Opinions reflect those of the authors and not necessarily the opinions of the granting agency. Contributor Information Jared D. Thorpe, University of North Carolina at Chapel Hill Robert Crosnoe, University of Texas at Austin. References Adema W, Clarke C, & Frey V (2016). Paid parental leave and other supports for parents with young children: The United States in international comparison. International Social Security Review, 69(2), 29–51. 10.1111/issr.12100 [ DOI ] [ Google Scholar ] Amato PR (2005). The impact of family formation change on the cognitive, social, and emotional well-being of the next generation. Future of Children, 15(2), 75–96. 10.1353/foc.2005.0012 [ DOI ] [ PubMed ] [ Google Scholar ] Bahr SJ, Hoffmann JP, & Yang X (2005). Parental and peer influences on the risk of adolescent drug use. Journal of Primary Prevention, 26(6), 529–551. 10.1007/s10935-005-0014-8 [ DOI ] [ PubMed ] [ Google Scholar ] Blum RW, Astone NM, Decker MR, & Mouli VC (2014). A conceptual framework for early adolescence: A platform for research. International Journal of Adolescent Medicine and Health, 26(3), 321–331. 10.1515/ijamh-2013-0327 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Berger LM, Carlson MJ, Bzostek SH, & Osborne C (2008). Parenting practices of resident fathers: The role of marital and biological ties. Journal of Marriage and Family, 70(3), 625–639. 10.1111/j.1741-3737.2008.00510.x [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Boddy J, Smith M, & Statham J (2011). Understandings of efficacy: Cross-national perspectives on “what works” in supporting parents and families. Ethics and Education, 6(2), 181–196. 10.1080/17449642.2011.622992 [ DOI ] [ Google Scholar ] Bradbury B, Corak M, Waldfogel J, & Washbrook E (2015). Too many children left behind: The U.S. achievement gap in comparative perspective. New York: Russell Sage. [ Google Scholar ] Brown SL, & Rinelli LN (2010). Family structure, family processes, and adolescent smoking and drinking. Journal of Research on Adolescence, 20(2), 259–273. 10.1111/j.1532-7795.2010.00636.x [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Carlson MJ (2006). Family structure, father involvement, and adolescent behavioral outcomes. Journal of Marriage and Family, 68(1), 137–154. 10.1111/j.1741-3737.2006.00239.x [ DOI ] [ Google Scholar ] Carlson MJ, & Berger LM (2013). What kids get from parents: Packages of parental involvement across complex family forms. Social Service Review, 87(2), 213–249. 10.1086/671015 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Carneiro P, Meghir C, & Parey M (2013). Maternal education, home environments, and the development of children and adolescents. Journal of the European Economic Association, 11, 123–161. 10.1111/j.1542-4774.2012.01096.x [ DOI ] [ Google Scholar ] Cavanagh SE (2008). Family structure history and adolescent adjustment. Journal of Family Issues, 29(7), 944–980. 10.1177/0192513X07311232 [ DOI ] [ Google Scholar ] Cavanagh SE, & Fomby P (2019). Family instability in the lives of American children. Annual Review of Sociology, 45(1), 493–513. 10.1146/annurev-soc-073018-022633 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Clogg CC, Petkova E, & Haritou A (1995). Statistical methods for comparing regression coefficients between models. American Journal of Sociology, 100(5), 1261–1293. 10.1086/230638 [ DOI ] [ Google Scholar ] Conger RD, Conger KJ, & Elder GH (1997). Family economic hardship and adolescent adjustment: mediating and moderating processes. In Duncan GJ & Brooks-Gunn J (Eds.), Consequences of growing up poor (pp. 288–310). New York: Russell Sage Foundation. [ Google Scholar ] Cortes RC, Fleming CB, Mason WA, & Catalano RF (2009). Risk factors linking maternal depressed mood to growth in adolescent substance use. Journal of Emotional and Behavioral Disorders, 17(1), 49–64. 10.1177/1063426608321690 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Cowan P, & Cowan C (2008). Diverging family policies to promote children’s well-being in the UK and US: Some relevant data from family research and intervention studies. Journal of Children’s Services, 3(4), 4–16. 10.1108/17466660200800022 [ DOI ] [ Google Scholar ] Crawford LA, & Novak KB (2002). Parental and peer influences on adolescent drinking: The relative impact of attachment and opportunity. Journal of Child and Adolescent Substance Abuse, 12(1), 1–26. 10.1300/J029v12n01_01 [ DOI ] [ Google Scholar ] Cross CJ, Fomby P, & Letiecq B (2022). Interlinking structural racism and heteropatriarchy: Rethinking family structure’s effects on child outcomes in a racialized, unequal society. Journal of Family Theory & Review, 14(3), 482–501. 10.1111/JFTR.12458 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Daly M, & Ferragina E (2018). Family policy in high-income countries: Five decades of development. Journal of European Social Policy, 28(3), 255–270. 10.1177/0958928717735060 [ DOI ] [ Google Scholar ] Deak JD, & Johnson EC (2021). Genetics of substance use disorders: A review. Psychological Medicine. 10.1017/S0033291721000969 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Doherty WJ, & Needle RH (1991). Psychological adjustment and substance use among adolescents before and after a parental divorce. Child Development, 62(2), 328–337. 10.1111/j.1467-8624.1991.tb01534.x [ DOI ] [ PubMed ] [ Google Scholar ] Duncan G, Magnuson K, & Ludwig J (2004). The endogeneity problem in developmental studies. Research in Human Development, 1(1), 59–80. 10.1207/s15427617rhd0101&2_5 [ DOI ] [ Google Scholar ] Elder GH, Johnson MK, & Crosnoe R (2003). The emergence and development of life course theory. In Mortimer J and Shanahan M (Eds.), Handbook of the life course (pp. 3–19). New York: Kluwer Academic/Plenum. 10.1007/978-0-306-48247-2_1 [ DOI ] [ Google Scholar ] Ellickson PL, Tucker JS, & Klein DJ (2003). Ten-year prospective study of public health problems associated with early drinking. Pediatrics, 111(5), 949–955. 10.1542/peds.111.5.949 [ DOI ] [ PubMed ] [ Google Scholar ] Ermisch J, & Pevalin DJ (2005). Early motherhood and later partnerships. Journal of Population Economics, 18, 469–489. 10.1007/s00148-004-0216-z [ DOI ] [ Google Scholar ] Flewelling RL, & Bauman KE (1990). Family structure as a predictor of initial substance use and sexual intercourse in early adolescence. Journal of Marriage and the Family, 52(1), 171–181. 10.2307/352848 [ DOI ] [ Google Scholar ] Frisco ML, Muller C, & Frank K (2007). Parents’ union dissolution and adolescents’ school performance: Comparing methodological approaches. Journal of Marriage and Family, 69(3), 721–741. 10.1111/j.1741-3737.2007.00402.x [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Glass J, Simon RW, & Andersson MA (2016). Parenthood and happiness: Effects of work-family reconciliation policies in 22 OECD countries. American Journal of Sociology, 122(3), 886–929. 10.1086/688892 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Guttmannova K, Bailey JA, Hill KG, Lee JO, Hawkins JD, Woods ML, & Catalano RF (2011). Sensitive periods for adolescent alcohol use initiation: Predicting the Lifetime occurrence and chronicity of alcohol problems in adulthood. Journal of Studies on Alcohol and Drugs, 72(2), 221–231. 10.15288/jsad.2011.72.221 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Hemovich V, & Crano WD (2009). Family structure and adolescent drug use: An exploration of single-parent families. Substance Use and Misuse, 44(14), 2099–2113. 10.3109/10826080902858375 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Heuveline P, Timberlake JM, & Furstenberg FF (2003). Shifting childrearing to single mothers: Results from 17 western countries. Population and Development Review, 29(1), 47–71. 10.1111/j.1728-4457.2003.00047.x [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Hillman SB, Sawilowsky SS, & Becker MJ (1993). Effects of maternal employment patterns on adolescents’ substance use and other risk-taking behaviors. Journal of Child and Family Studies, 2(3), 203–219. 10.1007/BF01321331 [ DOI ] [ Google Scholar ] Hoffmann JP (2002). The community context of family structure and adolescent drug use. Journal of Marriage and Family, 64(2), 314–330. 10.1111/j.1741-3737.2002.00314.x [ DOI ] [ Google Scholar ] Hoffmann JP (2017). Family structure and adolescent substance use: An international perspective. Substance Use and Misuse, 52(13), 1667–1683. 10.1080/10826084.2017.1305413 [ DOI ] [ PubMed ] [ Google Scholar ] Hoffmann JP, & Cerbone FG (2002). Parental substance use disorder and the risk of adolescent drug abuse: An event history analysis. Drug and Alcohol Dependence, 66(3), 255–264. 10.1016/S0376-8716(02)00005-4 [ DOI ] [ PubMed ] [ Google Scholar ] Hoffmann JP, & Johnson RA (1998). A national portrait of family structure and adolescent drug use. Journal of Marriage and the Family, 60(3), 633–645. 10.2307/353534 [ DOI ] [ Google Scholar ] Humensky JL (2010). Are adolescents with high socioeconomic status more likely to engage in alcohol and illicit drug use in early adulthood? Substance Abuse: Treatment, Prevention, and Policy, 5(1), 1–10. 10.1186/1747-597X-5-19 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Jockin V, McGue M, & Lykken DT (1996). Personality and divorce: A genetic analysis. Journal of Personality and Social Psychology, 71(2), 288–299. 10.1037/0022-3514.71.2.288 [ DOI ] [ PubMed ] [ Google Scholar ] Kamerman SB, & Kahn AJ (Eds.). (1997). Family change and family policies in Great Britain, Canada, New Zealand and the United States. New York: Oxford University Press. https://www.jstor.org/stable/3551532?origin=crossref [ Google Scholar ] Klerman JA, Daley K, & Pozniak A (2012). Family and medical leave in 2012: Executive summary. Cambridge, MA: Abt Associates. [ Google Scholar ] Krohn MD, Hall GP, & Lizotte AJ (2009). Family transitions and later delinquency and drug use. Journal of Youth and Adolescence, 38(3), 466–480. 10.1007/s10964-008-9366-8 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Lee D, & McLanahan S (2015). Family structure transitions and child development: Instability, selection, and population heterogeneity. American Sociological Review, 80(4), 738–763. 10.1177/0003122415592129 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Lichter DT, & Graefe DR (2001). Finding a mate? The marital and cohabitation histories of unwed mothers. In Wu L & Wade B (Eds.), Out of wedlock: Causes and consequences of nonmarital fertility. New York: Russell Sage. [ Google Scholar ] Luthar SS, Small PJ, & Ciciolla L (2018). Adolescents from upper middle class communities: Substance misuse and addiction across early adulthood. Development and Psychopathology, 30(1), 315–335. 10.1017/S0954579417000645 [ DOI ] [ PubMed ] [ Google Scholar ] Lynskey MT, Agrawal A, & Heath AC (2010). Genetically informative research on adolescent substance use: Methods, findings, and challenges. Journal of the American Academy of Child and Adolescent Psychiatry, 49(12), 1202–1214. 10.1016/j.jaac.2010.09.004 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Mcgue M, & Lykken DT (1992). Genetic influence on risk of divorce. Psychological Science, 3(6), 368–373. 10.1111/j.1467-9280.1992.tb00049.x [ DOI ] [ Google Scholar ] McLanahan S (2004). Diverging destinies: How children are faring under the second demographic transition. Demography, 41(4), 607–627. 10.1353/dem.2004.0033 [ DOI ] [ PubMed ] [ Google Scholar ] McLanahan S, & Sandefur G (1994). Growing up with a single parent: What hurts, what helps. Cambridge, MA: Harvard University Press. [ Google Scholar ] Menasco MA, & Blair SL (2014). Adolescent Substance Use and Marital Status in Adulthood. Journal of Divorce and Remarriage, 55(3), 216–238. 10.1080/10502556.2014.887382 [ DOI ] [ Google Scholar ] Miech R, & Chilcoat H (2005). Maternal education and adolescent drug use: A longitudinal analysis of causation and selection over a generation. Social Science and Medicine, 60(4), 725–735. 10.1016/j.socscimed.2004.06.025 [ DOI ] [ PubMed ] [ Google Scholar ] Morgan SL, & Winship C (2014). Counterfactuals and causal inference: Methods and principles for social research. New York: Cambridge University Press; 10.1017/CBO9781107587991 [ DOI ] [ Google Scholar ] Mundt MP (2011). The impact of peer social networks on adolescent alcohol use initiation. Academic Pediatrics, 11(5), 414–421. 10.1016/j.acap.2011.05.005 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Murray N, & Montelpare J (2022). The direct and indirect effects of cohabitation on children’s human capital development. Western Undergraduate Economic Review, 21, 1–26 [ Google Scholar ] Musick K, Brand JE, & Davis D (2012). Variation in the relationship between education and marriage: Marriage market mismatch? Journal of Marriage and Family, 74(1), 53–69. 10.1111/j.1741-3737.2011.00879.x [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Ohannessian CMC, Finan LJ, Schulz J, & Hesselbrock V (2015). A long-term longitudinal examination of the effect of early onset of alcohol and drug use on later alcohol abuse. Substance Abuse, 36(4),440–444. 10.1080/08897077.2014.989353 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Oppenheimer VK (1988). A theory of marriage timing. American Journal of Sociology, 94(3), 563–591. 10.1086/229030 [ DOI ] [ Google Scholar ] Peleg-Oren N, Saint-Jean G, Cardenas GA, Tammara H, & Pierre C (2009). Drinking alcohol before age 13 and negative outcomes in late adolescence. Alcoholism: Clinical and Experimental Research, 33(11), 1966–1972. 10.1111/j.1530-0277.2009.01035.x [ DOI ] [ PubMed ] [ Google Scholar ] Poortman AR (2007). The first cut is the deepest the role of the relationship career for union formation. European Sociological Review, 23(5), 585–598. 10.1093/esr/jcm024 [ DOI ] [ Google Scholar ] Richmond-Rakerd LS, Slutske WS, & Wood PK (2017). Age of initiation and substance use progression: A multivariate latent growth analysis. Psychology of Addictive Behaviors, 31(6), 664–675. 10.1037/adb0000304 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Schulenberg J, Patrick ME, Maslowsky J, & Maggs JL (2014). The epidemiology and etiology of adolescent substance use in developmental perspective. In Lewis M & Rudolph K Handbook of Developmental Psychopathology (pp. 601–620). Boston, MA: Springer US. 10.1007/978-1-4614-9608-3_30 [ DOI ] [ Google Scholar ] Smock PJ, & Greenland FR (2010). Diversity in pathways to parenthood: Patterns, implications, and emerging research directions. Journal of Marriage and Family, 72(3), 576–593 10.1111/j.1741-3737.2010.00719.x [ DOI ] [ Google Scholar ] Stritzel H, Green M, & Crosnoe R (2022). A cross-national comparison of the linkages between family structure histories and early adolescent substance use. Social Science and Medicine, 315. 10.1016/j.socscimed.2022.115540 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Thévenon O (2011). Family policies in OECD countries: A comparative analysis. Population and Development Review, 37(1), 57–87. 10.1111/j.1728-4457.2011.00390.x [ DOI ] [ PubMed ] [ Google Scholar ] Thomas G, Farrell MP, & Barnes GM (1996). The effects of single-mother families and nonresident fathers on delinquency and substance abuse in Black and White adolescents. Journal of Marriage and the Family, 58(4), 884–894. 10.2307/353977 [ DOI ] [ Google Scholar ] Trucco EM, & Hartmann SA (2021). Understanding the etiology of adolescent substance use through developmental perspectives. Child Development Perspectives, 15(4), 257–264. 10.1111/CDEP.12426 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Turner RA, Irwin CE, & Millstein SG (1991). Family structure, family processes, and experimenting with substances during adolescence. Journal of Research on Adolescence, 1(1), 93–106. [ Google Scholar ] Wagner KD, Ritt-Olson A, Chou CP, Pokhrel P, Duan L, Baezconde-Garbanati L, et al. (2010). Associations between family structure, family functioning, and substance use among hispanic/latino adolescents. Psychology of Addictive Behaviors, 24(1), 98–108. 10.1037/a0018497 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Woods DR (2012). Family policy in transformation: US and UK policies. New York: Palgrave/Macmillan. 10.1057/9780230348844 [ DOI ] [ Google Scholar ] ACTIONS View on publisher site PDF (436.2 KB) Cite Collections Permalink RESOURCES Similar articles Cited by other articles Links to NCBI Databases Follow NCBI NCBI on X (formerly known as Twitter) NCBI on Facebook NCBI on LinkedIn NCBI on GitHub NCBI RSS feed Connect with NLM NLM on X (formerly known as Twitter) NLM on Facebook NLM on YouTube National Library of Medicine 8600 Rockville Pike Bethesda, MD 20894 Web Policies FOIA HHS Vulnerability Disclosure Help Accessibility Careers NLM NIH HHS USA.gov Back to Top

Related documents

• Cross-system collaboration between the child welfare system and substance use treatment providers: a qualitative analysis of strategies, facilitators, and barriers.#645536
• Multisystemic Therapy - Building Stronger Families (MST-BSF): Substance misuse, child neglect, and parenting outcomes from an 18-month randomized effectiveness trial.#209626
• Young adult polysubstance use trajectories and later non-prescribed opioid use and drug use disorder symptoms: A propensity score-matched analysis#1133336
• Placental Pathophysiology in Maternal Psychoactive Substance Use: Biological, Clinical, and Forensic Perspectives.#315390
• Changing Patterns of Recent Substance Use in Emerging Adulthood and Their Associations With Cognitive Functions and Mental Health.#823440