Family structure and children's socioeconomic attainment: a Canadian sample.

Seabrook, Jamie A. ; Avison, William R.

WITH THE PROLIFERATION of different family forms in many western countries over the last few decades, research investigating the influence of family structure on children's socioeconomic attainment has expanded dramatically, particularly in the United States. Unfortunately, longitudinal research on this topic has been very limited in Canada. One exception is a study by Strohschein, Roos, and and Brownel (2009) that assessed the relationship between family structure histories and high school completion using a 1984 birth cohort in the province of Manitoba. This study found that, compared to children from stable two-parent families, children who experienced any change in family structure were less likely to graduate from high school. A key limitation of this study, however, was that there were no data available on household income. Many studies have shown that family income is associated with children's high school graduation, college attendance, and years of education, even after controlling for family structure (Brooks-Gunn, Guo, and Furstenberg 1993; Haveman and Wolfe 1994; Haveman, Wolfe, and Spaulding 1991). Consequently, it is possible that household income may have a greater relative effect on predicting children's high school graduation than family structure per se.

A recent synthesis of Canadian research on the relationship between income and children's developmental and psychosocial outcomes has found that income's effect on child outcomes is greater for children living in poverty, and that children's outcomes worsen the longer they live in poverty (McEwan and Stewart 2014). Although this appears to support a policy agenda aimed at improving household income, the authors note that federal and provincial benefits that only marginally improve income have minimal benefits on child outcomes. Thus, to improve child outcomes, policies that provide larger income benefits to children residing in poor families than children from socioeconomically advantaged families are necessary.

It is well known that single-parent families are more likely than two-parent families to live in poverty (Fields 2003; Williams et al. 2013). In 2005, 26 percent of single-parent families and 6.8 percent of two-parent families were living in poverty in Canada (Taylor 2007). Children whose parents lack financial resources are less likely to receive high-quality child care, education, health care, material, and social goods (Becker 1964; Coleman 1988; McLanahan and Percheski 2008), and these differential resources can accumulate across the life course, leaving children from socio-economically disadvantaged families with less opportunity to build their human capital.

It is now a widely held view that children raised in single-parent families are more disadvantaged than children from two-parent families with respect to their educational attainment and subsequent life chances, and that much of this disadvantage is attributable to the limited economic resources and parental engagement in single-parent families (Biblarz and Raftery 1999; McLanahan and Sandefur 1994; Sun and Li 2001). However, most research on the association between family structure and children's socioeconomic status has been conducted in the United States, where other related sources of disadvantage may have more important effects on children's outcomes than single parenthood (Thomson and McLanahan 2012). For example, approximately one-half of all African American families live in single-parent households (U.S. Census Bureau 2010), and studies that do not account for race/ethnicity (e.g., Pong, Dronkers, and Hampden-Thompson 2003) may overestimate the relative importance of family structure on children's status attainment.

It is also conceivable that poor school quality and greater neighborhood disadvantage have a larger relative influence on children's attainment outcomes in different regions of the United States than in Canada. Compared to the poorest neighborhoods in Canada, there is significantly more low-income segregation, racial/ethnic segregation, and violent crime in disadvantaged neighborhoods in the United States (Oreopoulos 2008). Fryer and Levitt (2004) sampled children from 1,000 U.S. schools and found that in 35 percent of these schools, there was not a single African American student, and that among schools attended by African American students, the average proportion African American was 59 percent. Interestingly, the authors found that family structure was not related to children's academic test scores from kindergarten to second grade, but that parental education, poor school quality, and neighborhood disadvantage all contributed to widening test score gaps in the first two years of school between African American and white students. Similarly, in the Chicago Longitudinal Study of over 1,500 African American children who grew up in high-poverty neighborhoods, Ou and Reynolds (2008) found that high school quality and maternal education were two of the most important predictors of children's educational attainment by age 20, and that single-parent status at the child's birth was unrelated to subsequent educational success.

A consistent problem with research assessing the relationship between family structure and children's attainment outcomes has been the reliance on cross-sectional, short-term, and retrospective data. This has resulted in studies investigating the influence of family structure on children's academic achievement while they are still in school (e.g., Duncan and Magnuson 2005; Fryer and Levitt 2004; Ram and Hou 2003; Sun and Li 2011; Wagmiller et al. 2010), without following these children into their adult years to determine whether occupational or income attainment is affected by growing up in a single-parent family.

In addition to the greater economic hardship experienced by single-parent families, research from Canada suggests that single mothers experience higher prevalence rates of major depressive illness and higher levels of psychological distress than married mothers (Avison, Ali, and Walters 2007; Lipman, Offord, and Boyle 1997; Seabrook and Avison 2012), and that many of the consequences of family structure on women's mental health result from single mothers' greater exposure to economic hardship, caregiving stress, and limited social support relative to married mothers (Avison et al. 2007). These mental health differences between single mothers and married mothers contribute to differences in parental socialization, as two meta-analyses have shown maternal depression to be related to high levels of harsh parenting and low levels of parental engagement and nurturance (Downey and Coyne 1990; Lovejoy et al. 2000).

Despite some of the adverse social consequences that children from single-parent families experience, not all social scientists agree that family structure has a direct and independent effect on children's socioeconomic attainment. For example, Song, Benin, and Glick (2012) report that differences in high school completion rates between children from two-parent families and single-mother families are greatly reduced when both types of families have comparable economic resources. Similarly, once parental occupational status and employment status are controlled, men from single-parent families have comparable occupational attainment to those from two-parent families (Biblarz, Raftery, and Bucur 1997). If, in fact, the relationship between family structure and children's socioeconomic attainment is as large as many social scientists suspect, we would expect much of the relationship between single parenthood and children's lower socioeconomic attainment to persist once parental socioeconomic status is controlled.

FAMILY STRUCTURE AND CHILDREN'S SOCIOECONOMIC STATUS ACROSS THE LIFE COURSE

The life course perspective (Elder 1998) provides a conceptual framework that allows for the modeling of stability and change in family structure over time, both of which can influence children's status attainment as adults. Whereas cross-sectional research conceptualizes family structure as a static phenomenon, a life course perspective emphasizes the timing and duration of family structure for children's lives.

The defining characteristic of the life course perspective is the need to assess temporality over extended periods of time (Elder, Johnson, and Crosnoe 2003; George 2007; Seabrook and Avison 2012). When a child's family structure is assessed over many years, patterns of stability or change can be modeled into trajectories, and these trajectories allow researchers to study the life course consequences of family structure on children's social outcomes.

Substantial gains have been made in our understanding of how trajectories of family structure can influence children's socioeconomic status. Two meta-analyses have shown that children with divorced parents are more likely to drop out of high school and to have poorer cognitive skills and psychosocial well-being than children raised with continuously married biological parents (Amato 2001; Amato and Keith 1991). Moreover, the greater mental health problems among children whose parents have experienced divorce are largely explained by hostile parental conflict and parental neglect during their childhood (O'Connor 1999). This is consistent with the idea of cumulative disadvantage, which posits that early childhood adversity, such as family instability, economic hardship, and poor family mental health can lead to a protracted chain of life course "insults" across the life course (O'Rand and Hamil-Luker 2005; Seabrook and Avison 2012).

In addition to family structure histories, it is important to consider the types of transitions that children experience when examining the relationship between family structure and socioeconomic attainment. For example, children who experience many years of stability in family structure after a parental divorce are probably a different group than children who have experienced multiple changes in family structure over the course of their childhood. Research has shown that children who experience two parental dissolutions have lower high school completion rates than children who experience just one or none (Cavanagh, Schiller, and Riegle-Crumb 2006; Pong and Ju 2000; Strohschein et al. 2009).

With multiple determinants of children's status attainment, including family structure, parental socioeconomic status, parental and childhood mental health, and family stability, it is essential for researchers and policymakers to ascertain which factors have the largest relative influence on children's attainment. This will allow for a policy agenda that best fits the well-being of children who grow up in different family types. We analyzed data from a 14-year prospective study of single-parent and two-parent families living in London, Ontario, Canada, that allowed us to address these important issues.

METHOD

Study Design

Our data come from a case-comparison, three-wave panel study of single-parent and two-parent families living in London, Ontario, Canada, with interviews conducted in 1993 (wave 1), 1994 (wave 2), and between 2005 and 2008 (wave 3).

The Sample

The Single-Parent Family Study was a survey designed to compare the experiences of single and married mothers. Single-parent families were defined as families headed by mothers who were separated, divorced, widowed, or never-married and who had at least one child under the age of 17 in 1993. Single parents included women who may have been legally married but had subsequently separated and were not cohabiting, as well as single mothers who lived in extended families. Two-parent families were defined as families in which the mother lived with her husband or cohabiting partner and had at least one child less than 17 years of age living at home. The reason why there was no differentiation between cohabiting and married couples in two-parent families is that only 5 percent of Ontario couples were cohabiting at wave 1, leaving too small of a comparison group of cohabiting partners given the sampling target of 500 two-parent families. Married partners and cohabiting partners were therefore categorized as two-parent families when coding the data on family structure.

The 1989 London Municipal Assessment File, an enumeration of every household in the municipality, was used to create the sampling frame. A list of 4,078 households headed by single women with at least one child under age 17, and 1,341 households involving single-parent families living with extended families was generated from the Municipal Assessment File. To ensure adequate representation of single-parent families across socioeconomic groups, these lists of single-parent families were stratified across 13 geographic areas to account for differences in household income. To generate a comparison sample of two-parent families, a two-stage sampling strategy was used to obtain a sampling pool of families that matched for the age and sex of the oldest child under 17 years from the sampled single-parent families. In the first stage, all two-parent families were randomly sampled from the Municipal Assessment File and stratified by the same 13 geographic areas as the single-parent families. These families were subsequently stratified by the age and gender of the oldest child in the second stage of sampling. Thus, each single-parent family had a randomly selected comparison two-parent family from the applicable age-gender stratum.

From the entire sampling pool of single-parent families, a stratified random sample of 723 single mothers were invited to participate in the wave 1 interviews, and 518 (71.7%) agreed to participate; of the 766 partnered mothers who were contacted, 502 (65.5%) participated in the wave 1 interviews. Almost 75 percent of single mothers were separated/divorced and 22 percent were never married. Mothers from single-parent families had a mean of 1.7 children (SD = 0.8) at wave 1, compared to 2.0 children (SD = 0.9) from two-parent families. The mean age of the children at wave 1 was 10.6 years (SD = 4.0), and by wave 3, the children were 15-33 years of age [mean: 24.5; SD = 4.3]. At wave 3, we completed follow-up interviews with 349/518 (67.4%) of the original sample of single parents. Sixteen (3.1%) single mothers had died or were too ill to be interviewed; 35 (6.8%) refused to be interviewed; and 118 (22.8%) were lost to follow-up. Among the two-parent families, we successfully reinterviewed 430/502 (85.7%) partnered mothers; 3 (0.6%) had died or were too ill to be interviewed; 24 (4.7%) refused; and 45 (14.5%) were lost to follow up.

Measures

Two-hour structured interviews were administered to mothers at waves 1 and 2 in their homes. At wave 3, mothers and children independently participated in structured interviews and completed a life history calendar that documented important life experiences that occurred between waves 2 and 3.

Life history calendars were used to collect data on household composition, sources of income, the timing, and sequencing of events corresponding to residential moves, and employment history. The calendar design was based on the work of Freedman et al. (1988), and Turner, Sorenson, and Turner (2000). Data collected on household composition were used to create variables to measure trajectories of family structure. Participants were asked to indicate the month of entry and exit into a partnered relationship, which provided information on the number of times that a woman lived with a male partner, the total number of months that each mother lived in a partnered relationship and the total number of partners.

Mothers were asked to rate their level of depressive symptoms, as well as their child's behavioral and emotional problems at wave 1. Maternal depressive symptoms were measured with the Center for Epidemiological Studies Depression Scale (CES-D), a 20-item questionnaire measuring depressive symptoms in the general adult population over the past week (Radloff 1977). Final scores on the CES-D range from 0 to 60, with scores [greater than or equal to] 16 indicating clinically relevant depressive symptomatology.

Children's behavioral and emotional problems were rated using the child behavior checklist (CBCL). The CBCL is a parent-report questionnaire in which parents rate their child's internalizing (i.e., anxious, depressive) and externalizing (i.e., aggressive, hyperactive, noncompliant) behaviors over the past six months (Achenbach 1991). The CBCL consists of 113 questions, and higher scores indicate greater emotional and behavioral problems. Following Achenbach's recommendation, we dichotomized the CBCL T scores, such that CBCL scores > 63 indicated clinical importance in children's mental health problems. Given the strong correlation between children's CBCL internalizing and externalizing scores, we only provide results for children's externalizing scores in regression tables.

Children's socioeconomic attainment at wave 3 was estimated by college/university graduation, occupational status, and personal income. To ensure that all children had the opportunity to complete their education, we used a cutoff of age 24 years and older to assess whether each child had received a college diploma/university degree. Using the age cutoff of 24 years, we had data on 401 children for our analysis.

Children were also asked about their work history. Information about their longest jobs held was derived from children's life history chart data. The responses to these occupational outcomes were then coded using the Hollingshead Index of Social Position (Hollingshead 1975). Categories to these variables included: (1) higher executives, proprietors of large concerns; (2) business managers, proprietors of medium-sized businesses; (3) administrative personnel, small independent businesses; (4) clerical and sales workers, as well as technicians; (5) skilled manual employees; (6) machine operators and semi-skilled employees; and (7) unskilled employees. We reverse coded the occupational categories for ease of interpretation so that higher scores were reflective of higher socioeconomic status jobs. In total, 520 children were included in the analysis once we excluded children who were still in school. The reason for the discrepancy in sample size between the occupational and educational attainment measures is due to the older age cutoff for education.

The current personal income of the children was also asked. We had personal incomes recorded for 587 children; once children who were still in school were excluded from the analysis, we were left with 513 children. Eighteen income categories, ranging from less than $5,000 per year to $90,000 or more, indexed income. The midpoints of the categories were used to calculate a continuous variable for the regression analysis.

Trajectories of Family Structure

Family structure (single/partnered) at baseline, wave 2 (12 months later), wave 3 (12 years after wave 2), the number of times a mother was partnered, and the length of time a mother was partnered over the previous 12 years were entered into a latent class cluster analysis in order to determine distinct groups of mothers who shared similar trajectories of family structure. Considering what we know about the importance of family stability for better long-term social outcomes for children (Cavanagh et al. 2006; Fomby and Cherlin 2007), the decision to incorporate these variables was both theoretically informed based on the life course perspective, and based empirically on research on family structure and children's socioeconomic attainment.

Latent class cluster models were generated using Latent Gold Software (Vermunt and Magidson 2000). Model fit was assessed by comparing the likelihood ratio chi-square statistics, the log likelihood, the Bayes information criteria, the classification error, models with the fewest number of parameters, and bivariate residuals. Fit statistics indicated that a fourcluster model provided the best fit for the data:

(1) Temporally stable two-parent families (N = 338; 45.2% of all families in the study).

(2) Temporally stable single-parent families (N = 154; 20.6%).

(3) Single-parent families that transitioned to two-parent families over the past 12 years (N = 150; 20.1%).

(4) Two-parent families that transitioned to single-parent families over the past 12 years (N = 105; 14.1%).

Attrition Analysis

Our estimates of the influence of single parenthood on children's socioeconomic outcomes are vulnerable to two potential sources of attrition bias. First, some children were lost to follow-up because we could not relocate the families to reinterview the mothers and their children (n = 207). Second, another subsample was lost to follow-up because we could not locate and interview the children, even though their mothers had participated in the wave 3 interviews (n = 125).

To evaluate attrition bias, we used a methodological approach as suggested by Miller and Hollist (2007), in which a logistic regression analysis was computed with a dichotomous dependent variable (1 = stayers, 0 = dropouts) and the following wave 1 independent variables: family structure, child's age, mother's age, mother's education, mother's household income, mother's CES-D, mother's employment status, and child mental health problems. There were no statistically significant coefficients for any of the independent variables between those who remained in the study and those who dropped out, indicating no attrition bias. These results suggest that, despite the higher overall rates of attrition from single-parent families, patterns of attrition bias were virtually identical by family structure, indicating that the most disadvantaged children were lost to follow-up in both two-parent and single-parent families. In other words, differences in the socioeconomic outcomes of children at wave 3 are not the result of family structure differences in student academic performance or parental socioeconomic status at wave 1.

RESULTS

Concurrent Effects of Family Structure on Children's Status Attainment

The association of family structure at wave 1 with children's socioeconomic attainment at wave 3 is shown in Table 1. For all socioeconomic attainment outcomes, we found no significant differences between children who were living in single-parent compared to two-parent families at wave 1. These results therefore suggest no disadvantage in socioeconomic status for children who spent at least some of their childhood in a single-parent family compared to a two-parent family.

The Relationship between Trajectories of Family Structure and Children's Status Attainment

In Table 2, we compare trajectories of family structure on children's status attainment at wave 3. We found no significant differences in college/university graduation or personal income between the clusters. Unexpectedly, the mean occupational ranking for children from temporally stable single-parent families was 3.4 (SD = 1.8), which was significantly higher than the ranking of children from temporally stable two-parent families [mean: 2.7 (SD = 1.6), p = .004] and children whose families transitioned from two-parent status to single-parent status [mean: 2.7 (SD = 1.7), p = .026] between waves 2 and 3. Like the cross-sectional results examining the influence of family structure at wave 1 on children's socioeconomic status at wave 3, these findings provide no evidence that children raised in stable two-parent families have more advantageous attainment outcomes than children from other family types.

Regression Analyses

Children's educational attainment. Table 3 present the results of logistic regression models estimating the relative effects of family structure, parental resources, and family mental health on the likelihood of children obtaining a college diploma/university degree at wave 3. Models 1-3 display the effects of family structure at wave 1, whereas Models 4-6 examine the influence of trajectories of family structure.

In Model 1, we found that family structure was not associated with children's educational attainment, but males were about 54 percent less likely to graduate from college or university than females. We added in parental resources in Model 2, and found that mothers' age and education were positively associated with their children's education. In Model 3, we see that children with externalizing scores above the cut-point were 61.7 percent less likely to acquire a university degree/college diploma than those who scored below the cut-point.

In Models 4-6, we found that children from all four family structure clusters were just as likely to graduate with a college diploma or university degree. Female gender, mothers' age, and mothers' years of education at wave 1 all increased the probability of children graduating from college or university, whereas children who had clinically significant mental health problems at wave 1 were less likely to graduate.

Children's longest job. Table 4 presents estimates from regression models for children's longest job held. Family structure and family mental health at wave 1 had no significant influence on children's occupational attainment (Models 1-4). The only variables that influenced children's occupational status were children's age, mothers' age, and children's educational attainment, all of which were positively related to their occupational attainment.

In Model 5, both children's age and the temporally stable single-parent family cluster are positively related to children's job attainment at wave 3. When parental resources are added in Model 6, not only do children from temporally stable single-parent families continue to have a higher occupational ranking than children from temporally stable two-parent families, but so do children whose families transitioned from a single-parent family to a two-parent family between waves 2 and 3 (p < .01). In Model 7, we see that children's age, temporally stable single-parent families, and the single-parent to two-parent family cluster were all positively related to children's longest job held, as was children's years of education in Model 8.

Children's personal income. Multiple regression equations were computed to determine the effects of family structure, family mental health and parental resources at wave 1 on children's personal income at wave 3 in Models 1-4 of Table 5. In Model 1, family structure did not predict children's income. When mothers' age and socioeconomic status at wave 1 were included in Model 2, however, the measure of family structure became significant; children who were living in single-parent families at wave 1 had higher personal incomes than children who were living in two-parent families. This suppressor effect of mothers' income suggests that, if single mothers have the same household income as their married counterparts, their children would have even higher personal income attainment than children from two-parent families. Model 4 shows that children's education is strongly related to their personal income.

In Models 5-8 we see that children's age and male gender are positively related to personal income, but that trajectories of family structure have no significant relationship with income attainment. Children's total years of education was again positively related to their income attainment (Model 8).

DISCUSSION

Although many studies from the United States contend that the economic disadvantage of single-parent families relative to two-parent families is the key reason why children from the former have lower socioeconomic attainment (e.g., Astone and McLanahan 1991; McLanahan and Sandefur 1994), our results suggests that substantial differences in household income by family structure do not disadvantage children over time. Our findings are relatively consistent with a new synthesis of Canadian evidence that shows that, although a correlation exists between income and several developmental and psychosocial child outcomes, the relationship is in fact small in magnitude once education, employment, and neighborhood is controlled (McEwan and Stewart 2014). Instead of household income, we found that, regardless of family structure, higher levels of maternal education increase the probability that children will graduate from college/university, and that children raised in temporally stable single-mother families, and those whose families transitioned from a single-parent family to a two-parent families have higher occupational attainment than children from temporally stable two-parent families for their longest job held.

The fact that children raised in stable two-parent families did not have better socioeconomic outcomes than children from other family types was an unexpected finding. It may be that more people today are getting at least some postsecondary education, regardless of the type of family they are raised in. London, Ontario is also unique in that it has two highly accredited postsecondary institutions, Western University and Fanshawe College. This makes it easier for adolescents already living in London, Ontario, because they do not have to move to another city to pursue college or university. Moreover, with more people recognizing the value of a postsecondary education in order to find secure employment, the lack of differences in educational attainment by family structure may not be that surprising, considering that there were little differences in maternal education by family structure.

We are also unaware of the level of conflict and tension that may have been present in temporally stable two-parent families. Indeed, many marriages endure despite high levels of conflict. Recent research by Musick and Meier (2010) found that differences in academic achievement are negligible between children raised in single-mother families and those living in a home with constant fighting between parents.

It might also be the case that some children from disadvantaged homes acquire a strong level of resiliency. For example, with many single mothers having to work long hours due to family economic insecurity, their children may have to take on more responsibility in terms of decision making within the household, thereby enhancing their self-efficacy.

Reconciling Results from the Single-Parent Family Study with Other Studies

Although our findings on family structure and children's socioeconomic status seem counterintuitive, there are several ways to reconcile our results with those from other research. In the Single-Parent Family Study, mothers from two-parent families had only one year more of completed education than mothers from single-parent families, and the mean difference was only 0.4 years between mothers from temporally stable two-parent families and temporally stable single-parent families (data not shown). Moreover, when contrasting educational disparities by family structure between Canada and the United States, evidence suggests that a much larger gap in academic performance can be found in the United States. Hampden-Thompson (2009) found mean literacy scores to be 6.6 percent lower for children from single-mother families compared to children from two-parent families in the United States, whereas the difference was only 2.4 percent for children from Canada. This implies that underachievement in educational attainment may be more prevalent in single-mother families in the United States than is the case for Canada.

The fact that children from temporally stable single-parent families and those whose families transitioned from a single-parent family to a two-parent family had higher status jobs may be explained from the finding that mothers from these two clusters had the highest Hollingshead job rank of all four trajectories of family structure. This high occupational standing for mothers from single-parent families may be important because parents in two-parent families usually have higher status jobs than single parents (Astone and McLanahan 1991; Biblarz and Raftery 1999). Hampden-Thompson (2009) found that parents in two-parent families in Canada had a mean occupational ranking 1.9 times higher than parents in single mother families, whereas the difference was 4.1 times greater between two-parent families and single mother families in the United States. Biblarz and Raftery (1999) found no effect of growing up in single-mother families for children's educational and occupational attainment, and the authors argue that parents' labor force attachment and occupational standing is the key to understanding the relationship between family structure and children's socioeconomic status. In terms of policy implications, this suggests that encouraging family-friendly work environments that recognize that single mothers are no less valuable employees than partnered mothers can have significant and positive consequences with respect to their children's future status attainment.

Another possible explanation for the similar status attainment outcomes may be because single parenthood in London, Ontario, is less connected to neighborhood, school, and ethnic/racial disadvantage than in other parts of Canada and the United States. In a review of 25 studies mostly from the United States, 23 studies showed at least a moderate, independent neighborhood effect on health status controlling for socioeconomic status (Picket and Pearl 2001). Although Canadian research on the effects of disadvantaged neighborhoods is lacking, one study from Montreal, Quebec, found that only 3 percent of the variation in health status was attributable to neighborhoods, despite Montreal being Canada's most segregated and unequal major city with respect to income (Ross, Tremblay, and Graham 2004).

At the time of our study, London, Ontario, also had a relatively small percentage of visible minorities. Whereas 13 percent of the population of London in 2006 was comprised of visible minorities, 21 percent were visible minorities in the province of Ontario, and the average for Census metropolitan areas was 25 percent. In 2001, visible minorities made up 9 percent of the London population and 19 percent of the population in Ontario (Statistics Canada 2007). In our sample, there were only five black participants, all of whom were living in a single-parent family at wave 1. By contrast, the U.S. Census Bureau (2010) has recently revealed that 49 percent of African American children live in single-mother families. It is therefore possible that issues pertaining to race and ethnicity are confounding the relationship between family structure and children's attainment in the United States.

It may also be that moderate size cities, such as London, Ontario, have lower socioeconomic gradients that may partially account for the similar outcomes in status attainment by family structure. At wave 3, the mean household income of two-parent families was $63,400 compared to $34,600 for single-parent families. In Canada, the average family income in 2006 was $80,100 for married couples with children and $38,800 for single-mother families (Statistics Canada 2011). These household incomes represent a ratio of 1.8:1 favoring mothers in two-parent families in our study, and 2.1:1 favoring married couples with children over single-mother families in Canada. When the incomes in the Single-Parent Family Study are contrasted with those from the United States, the differences are even more noticeable. According to 2006 U.S. Census data, the average family income for single-mother families was only $28,865, compared to $89,096 for married couples with children, indicating a 3.1:1 ratio favoring married parents with children (U.S. House of Representatives 2008). Thus the socioeconomic gradient between single-parent and two-parent families appears higher for both Canada and the United States when compared to the families in this study.

Study Limitations

This study is not without limitations. Over the 14-year follow-up period, 24 percent (241/1,020) of mothers who were interviewed at wave 1 were lost to follow-up by wave 3. Of the 241 mothers who did not participate in the wave 3 interviews, 169 were single mothers at wave 1 and 72 were partnered mothers. Despite the higher overall attrition rates among single-parent families, however, we found no evidence for differential attrition bias by family structure. In others words, attrition could not account for the absence of socioeconomic attainment differences by family structure.

Another challenge we faced when investigating children's socioeconomic status outcomes had to do with the ages of the children in this study. With some evidence, for example, suggesting that family structure in very early childhood is more important for children's achievement outcomes than family structure in later childhood or adolescence (Ermisch and Francesconi 2001), subanalyses of children who were very young (e.g., five years of age or under) at wave 1 do not allow for adequate comparisons by family structure in socioeconomic outcomes because these children would be a maximum of 20 years of age at wave 3, not giving them enough time to complete college or university. Furthermore, if these children have not had the time to complete their postsecondary educational endeavors, it is unlikely that they will have already entered the labor force, let alone have had time to acquire a higher status job. On the other hand, those who have entered the workforce are likely to have been in low status occupations, given their lack of educational credentials.

The young ages of these children at wave 3 also limit our ability to adequately test other important outcomes. For instance, children from single-parent families have a greater likelihood of becoming divorced than do children from two-parent families. Children from divorced families have a 50 percent greater probability of divorcing when they are adults, are more apt to marry someone who has also experienced parental divorce, and are more likely to marry at a young age (Wolfinger 2005). Although the small subsample of children who had separated/divorced by wave 3 limits generalizability, we found, however, that 12 percent (6/51) of children who were raised in temporally stable two-parent families and who had married by wave 3 had separated, compared to just 3 percent (1/29) of children from temporally stable single-parent families. The corresponding percentages of children who had separated were 15 percent (4/26) and 11 percent (2/18) for children whose mothers transitioned from a single-parent family to a two-parent family and from a two-parent family to a single-parent family.

We were also unable to differentiate between children raised in families whose parents were married and those raised in cohabiting families. With only 5 percent of Ontario couples cohabiting at wave 1 (1993) of the study, married partners and cohabiting partners were both classified as being in a two-parent family when the first wave of interviews were conducted. As of 2011, however, common-law families now represent almost 17 percent of Canadian families, and have just surpassed the number of single-parent families (Statistics Canada 2012). Thus research that compares differences in children's socioeconomic outcomes between those living with cohabiting parents, married parents and single parents is an important line of research in the years to come.

CONCLUSION

Notwithstanding these limitations, this paper demonstrates the importance of evaluating the effect of family structure across childhood when assessing its influence on children's socioeconomic attainment. Findings from this study suggest that children from single-parent families have socioeconomic status attainment that is comparable to children from two-parent families. In our view, it seems reasonable to conclude that in circumstances where social disparities associated with race/ethnicity and neighborhood are not pronounced, and where educational and socioeconomic differences are only modestly associated with family structure, the impact of growing up in a single-parent household on children's socioeconomic attainment in early adulthood is negligible. For those concerned with social policy, this implies that greater attention ought to be paid to addressing disparities in education, occupation, and family income than to concerns with the kinds of families in which children grow up.

References

Achenbach, T.M. 1991. Manual for the Child Behavior Checklist 14-18 and 1991 Profile. Burlington, VT: University of Vermont, Department of Psychiatry.

Amato, P. 2005. "The Impact of Family Formation Change on the Cognitive, Social, and Emotional Well-Being of the Next Generation." Future of Children 15:75-96.

Amato, P.R. 2001. "Children of Divorce in the 1990s: An Update of the Amato and Keith (1991) Meta-Analysis." Journal of Family Psychology 15:355-70.

Amato, P.R. and B. Keith. 1991. "Parental Divorce and Adult Well-Being: A Meta-Analysis." Journal of Marriage and Family 53:43-58.

Astone, N.M. and S.S. McLanahan. 1991. "Family Structure and High School Completion." American Sociological Review 56:309-20.

Avison, W.R., J. Ali and D. Walters. 2007. "Family Structure, Stress, and Psychological Distress: A Demonstration of the Impact of Differential Exposure." Journal of Health and Social Behavior 48:301-17.

Becker, G.S. 1964. Human Capital: A Theoretical and Empirical Analysis. New York: National Bureau of Economic Research.

Biblarz, T.J. and A.E. Raftery. 1999. "Family Structure, Educational Attainment, and Socioeconomic Success: Rethinking the 'Pathology of Matriarchy.'" American Journal of Sociology 105:321-65.

Biblarz, T.J., A.E. Raftery and A. Bucur. 1997. "Family Structure and Social Mobility." Social Forces 75:1319-41.

Brooks-Gunn, J., G. Guo and F.F. Furstenberg Jr. 1993. "Who Drops Out of and Who Continues beyond High School? A 20-Year Follow-Up of Black Urban Youth." Journal of Research on Adolescence 3:271-94.

Cavanagh, S.E., K.S. Schiller and C. Riegle-Crumb. 2006. "Marital Transitions, Parenting, and Schooling: Exploring the Link between Family-Structure History and Adolescents' Academic Success." Sociology of Education 79:329-54.

Coleman, J.S. 1988. "Social Capital in the Creation of Human Capital." American Journal of Sociology 94:S95-120.

Downey, G. and J. Coyne. 1990. "Children of Depressed Parents: An Integrative Review." Psychological Bulletin 108:50-76.

Duncan, G.J. and K.A. Magnuson. 2005. "Can Family Socioeconomic Resources Account for Racial and Ethnic Test Score Gaps?" The Future of Children 15:35-54.

Elder, G.H., Jr. 1998. "The Life Course and Human Development." Pp. 939-91 in Handbook of Child Psychology: Volume 1: Theoretical Models of Human Development, edited by W. Damon and R.M. Lemer. New York: Wiley.

Elder, G.H. Jr., M.K. Johnson and R. Crosnoe. 2003. "The Emergence and Development of Life Course Theory." Pp. 3-19 in Handbook of the Life Course, edited by J.T. Mortimer and M.J. Shanahan. New York: Springer Science.

Ermisch, J.F. and M. Francesconi. 2001. "Family Structure and Children's Achievements." Journal of Population Economics 14:249-70.

Fields, J. 2003. "Children's Living Arrangements and Characteristics: March 2002." Pp. 20-547 in Current Population Reports. Washington, DC: U.S. Census Bureau.

Fomby, P. and A.J. Cherlin. 2007. "Family Instability and Child Well-Being." American Sociological Review 72:181-204.

Freedman, D., A. Thorton, D. Cambum, D. Alwin and L. Young DeMarco. 1988. "The Life History Calendar: A Technique for Collecting Retrospective Data." Pp. 37-68 in Sociological Methodology, edited by C. Clogg. San Francisco: Jossey-Bass.

Fryer, R.G. Jr. and S.D. Levitt. 2004. "Understanding the Black-White Test Score Gap in the First Two Years of School." Review of Economics and Statistics 86:447-64.

George, L.K. 2007. "Life Course Perspectives on Social Factors and Mental Illness." Pp. 191-218 in Mental Health, Social Mirror, edited by W.R. Avison, J.D. McLeod and B.A. Pescosolido. New York: Springer.

Hampden-Thompson, G. 2009. "Are Two Better Than One? A Comparative Study of Achievement Gaps and Family Structure." Compare: A Journal of Comparative and International Education 39:517-34.

Haveman, R., B. Wolfe and J. Spaulding. 1991. "Childhood Events and Circumstances Influencing High School Completion." Demography 28:133-57.

Haveman, R.H. and B. Wolfe. 1994. Succeeding Generations: On the Effects of Investments in Children. New York: Russell Sage Foundation.

Lipman, E.L., D.R. Offord and M.H. Boyle. 1997. "Single Mothers in Ontario: Sociodemographic, Physical and Mental Health Characteristics." Canadian Medical Association Journal 156:639-45.

Lovejoy, M.C., P.A. Graczyk, E. O'Hare and G. Neuman. 2000. "Maternal Depression and Parenting Behavior: A Meta-Analytic Review." Clinical Psychology Review 20:561-92.

McEwen, A. and J.M. Stewart. 2014. "The Relationship between Income and Children's Outcomes: A Synthesis of Canadian Evidence." Canadian Public Policy 40:99-109.

McLanahan, S. and C. Percheski. 2008. "Family Structure and the Reproduction of Inequalities." Annual Review of Sociology 34:257-76.

McLanahan, S. and G. Sandefur. 1994. Growing Up with a Single Parent. Cambridge: Harvard University Press.

Miller, R.B. and C.S. Hollist. 2007. "Attrition Bias." Pp. 57-60 in Encyclopedia of Measurement and Statistics: Volume 1, edited by N. Salkind. Thousand Oaks: Sage Publications.

Musick, K. and A. Meier. 2010. "Are Both Parents Always Better Than One? Parental Conflict and Young Adult Well-Being." Social Science Research 39:814-30.

O'Connor, T.G., K. Thorpe, J. Dunn and J. Golding. 1999. "Parental Divorce and Adjustment in Adulthood: Findings from a Community Sample." Journal of Child Psychology and Psychiatry 40:777-89.

O'Rand, A.M. and J. Hamil-Luker. 2005. "Processes of Cumulative Adversity: Childhood Disadvantage and Increased Risk of Heart Attack across the Life Course." The Journals of Gerontology: Series 60B: 117-24.

Oreopoulos, P. 2008. "Neighbourhood Effects in Canada: A Critique." Canadian Public Policy 34:237-58.

Ou, S.R. and A.J. Reynolds. 2008. "Predictors of Educational Attainment in the Chicago Longitudinal Study." School Psychology Quarterly 23:199-229.

Picket, K.E. and M. Pearl. 2001. "Multilevel Analyses of Neighbourhood Socioeconomic Context and Health Outcomes: A Critical Review." Journal of Epidemiology and Community Health 55:111-22.

Pong, S.L., J. Dronkers and G. Hampden-Thompson. 2003. "Family Policies and Children's School Achievement in Single-Versus Two-Parent Families." Journal of Marriage and Family 65:681-99.

Pong, S.L. and D.B. Ju. 2000. "The Effects of Change in Family Structure and Income on Dropping Out of Middle and High School." Journal of Family Issues 21:147-69.

Radloff, L. 1977. "The CES-D Scale: A Self-Report Depression Scale for Research in the General Population." Applied Psychological Measurement 1:385-401.

Ram, B. and F. Hou. 2003. "Changes in Family Structure and Child Outcomes: Roles of Economic and Familial Resources." The Policies Studies Journal 31:309-30.

Ross, N.A., S. Tremblay and. K. Graham. 2004. "Neighbourhood Influences on Health in Montreal, Canada." Social Science and Medicine 59:1485-94.

Seabrook, J.A and W.R. Avison. 2012. "Socioeconomic Status and Cumulative Disadvantage Processes across the Life Course: Implications for Health Outcomes." Canadian Review of Sociology I Revue Canadienne de Sociologie 49:50-68.

Song, C., M. Benin and J. Glick. 2012. "Dropping Out of High School: The Effects of Family Structure and Family Transitions." Journal of Divorce and Remarriage 53:18-33.

Statistics Canada. 2007. Catalogue no: 92-591-XWE. Ottawa.

Statistics Canada. 2011. Average Income after Tax by Economic Family Types. Summary Tables. Last updated June 15, 2011. Retrieved October 22, 2012 (http://www40.statcan.ca/1010cst01-famil21a-eng.htm).

Statistics Canada. 2012. Catalogue no: 98-312-X2011001. Ottawa.

Strohschein, L., N. Roos and M. Brownell. 2009. "Family Structure Histories and High School Completion: Evidence from a Population-based Registry." Canadian Journal of Sociology 34:83-103.

Sun, Y. and Y. Li. 2001. "Marital Disruption, Parental Investment and Children's Academic Achievement." Journal of Family Issues 22:27-62.

Sun, Y. and Y. Li. 2011. "Effects of Family Structure Type and Stability on Children's Academic Performance Trajectories." Journal of Marriage and Family 73:541-56.

Taylor, P.S. 2007. Family Poverty in Canada: Raising Incomes and Strengthening Families. Canadian Family Views, 14-18.

Thomson, E. and S.S. McLanahan. 2012. "Reflections on 'Family Structure and Child WellBeing: Economic Resources vs. Parental Socialization.'" Social Forces 91:45-53.

Turner, R.J., A.M. Sorenson and J.B. Turner. 2000. "Social Contingencies in Mental Health: A Seven Year Follow-Up Study of Teenage Mothers." Journal of Marriage and the Family 62:777-91.

U.S. Census Bureau. 2010. "One in Two African-American Children Lives in a Single-Parent Home." Current Population Survey: 1-20.

Vermunt, J.K. and J. Magidson. 2000. Latent Gold User's Guide. Belmont, MA: Statistical Innovations.

Wagmiller, R.L., E. Gershoff, P. Veliz and M. Clements. 2010. "Does Children's Academic Achievement Improve When Single Mothers Marry?" Sociology of Education 83:201-26.

Williams, K., S. Sassler, A. Freeh, F. Addo and E. Cooksey. 2013. "Policy Brief." Journal of Health and Social Behavior 54:277-77.

Wolfinger, N.H. 2005. Understanding the Divorce Cycle: The Children of Divorce in Their Own Marriages. Cambridge: Cambridge University Press.

JAMIE A. SEABROOK

Brescia University College at Western University

WILLIAM R. AVISON

Western University

Jamie A. Seabrook, Division of Food and Nutritional Sciences, Brescia University College at Western University, Children's Health Research Institute, 1285 Western Road, London, ON N6G 1H2, Canada. E-mail: jseabro2@uwo.ca Table 1 The Influence of Family Structure at Wave 1 (1993) on Children's Socioeconomic Attainment at Wave 3 (2005-2008) Single-parent Two-parent family family Variables (re = 518) (re = 502) p-Value % College diploma/University 58.8 60.4 0.74 degree (.n = 194) (n = 207) Job rank on Hollingshead Index 3.0 (1.7) 2.8 (1.6) 0.28 (n = 242) (n = 265) Personal income ($ thousands) 27.0 (12.2) 25.8(10.1) 0.42 (n = 246) (re = 267) Notes: The chi-square test compared educational differences between groups; the independent sample E-test compared mean differences in job rank and personal income. There were no significant differences between children from single-parent families and children from two-parent families. Table 2 The Influence of Trajectories of Family Structure on Children's Socioeconomic Attainment at Wave 3 (2005-2008) Stable two- Stable single- parent family parent family Variables (n = 338) (n = 154) % College diploma/ 64.3 61.6 University degree (n = 154) (n = 86) Job rank on 2.7 (1.6) 3.4 (1.8) Hollingshead index (n = 200) (n = 96) Personal income 26.2 (9.8) 29.4(13.0) ($ thousands) (n = 203) (n = 99) Single parent Two parent to to two parent single parent p-Value Variables (n = 150) (n = 105) % College diploma/ 58.7 50.9 0.35 University degree (n = 63) (n = 57) Job rank on 3.0 (1.7) 2.7 (1.7) 0.005 Hollingshead index (n = 94) (n = 73) Personal income 24.6 (9.5) 24.6 (9.5) 0.27 ($ thousands) (n = 93) (n = 72) A one-way analysis of variance was used to compare mean differences in job rank and personal income between the four trajectories of family structure. The chi-square test was used for differences in educational attainment. Table 3 Logistic Regression Assessing the Relative Effects of Family Structure, Mothers' Socioeconomic Status, and Family Mental Health at Wave 1 on Children Obtaining a College Diploma or University Degree at Wave 3 Model 1 Model 2 b OR b OR SPF wave l (a) -.138 .871 -.007 .993 Temporally stable SPFs (b) SPF transitioned to TPF (b) TPF transitioned to SPF (b) Child's age .042 1.043 -.022 .978 Male -.778 *** .459 -.891 *** .410 Mother's age .072 ** 1.075 Mother's education .120 ** 1.127 Mother's employed .099 1.104 FT or PT Mother's household .006 1.006 income Mother's CES-D Child's externalizing score > 63 Constant -1.120 -4.146 -2.738 -.285 Nagelkerke [R.sup.2] .049 .134 .173 .054 Model 3 Model 4 b OR b OR SPF wave l (a) -.021 .980 Temporally stable -.164 .849 SPFs (b) SPF transitioned -.324 .723 to TPF (b) TPF transitioned -.539 .583 to SPF (b) Child's age -.049 .952 .019 1.019 Male -.953 *** .386 -.743 ** .476 Mother's age .071 ** 1.073 Mother's education .100 ** 1.105 Mother's employed .135 1.144 FT or PT Mother's household -.011 .989 income Mother's CES-D -.006 .994 Child's -.959 ** .383 externalizing score > 63 Constant -3.575 -2.216 Nagelkerke [R.sup.2] .144 .183 Model 5 Model 6 b OR b OR SPF wave l (a) -.253 .777 -.277 .758 Temporally stable SPFs (b) -.284 .753 -.342 .710 SPF transitioned to TPF (b) -.379 .685 -.234 .792 TPF transitioned to SPF (b) -.046 .955 -.071 .932 Child's age -.869 *** .419 -.927 *** .396 Male .079 ** 1.082 .077 ** 1.080 Mother's age .139 ** 1.149 .125 ** 1.133 Mother's education .093 1.097 .166 1.181 Mother's employed FT or PT -.042 .959 Mother's household income -.014 .986 Mother's CES-D -.870 ** .419 Child's externalizing score > 63 Constant Nagelkerke [R.sup.2] Notes: * p < .05; ** p < .01; *** p < .001. CES-D, Center for Epidemiological Studies Depression Scale; SPF, single-parent family; TPF, two-parent family. (a) Two-parent family at wave 1 = reference group; (b) Temporally stable two-parent family = reference group. Table 4 Multiple Regressions Assessing the Relative Effects of Family Structure, Mothers' Socioeconomic Status and Family Mental Health at Wave 1 on Children's Longest Job Held at Wave 3 Model 1 Model 2 B Beta B Beta SPF wave l (a) .048 .014 .319 .095 Temporally stable SPFs (b) SPF transitioned to TPF (b) TPF transitioned to SPF (b) Child's age .177 *** .392 .131 *** .292 Male -.083 -.025 -.095 -.028 Mother's age .037 ** .140 Mother's .025 .039 education Mother's -.006 -.002 employed FT or PT Mother's .034 .090 household income Mother's CES-D Child's externalizing score > 63 Children's years of education Constant -1.645 -2.712 Adjusted [R.sup.2] .151 .172 Model 3 Model 4 B Beta B Beta SPF wave l (a) .285 .086 .298 .090 Temporally stable SPFs (b) SPF transitioned to TPF (b) TPF transitioned to SPF (b) Child's age .126 *** .277 .100 *** .220 Male -.094 -.028 .024 .007 Mother's age .037 ** .137 .030 ** .112 Mother's .014 .022 -.019 -.030 education Mother's .078 .020 .107 .028 employed FT or PT Mother's .027 .072 .017 .047 household income Mother's .006 .039 .008 .048 CES-D Child's -.353 -.084 -.180 -.043 externalizing score > 63 Children's .147 *** .223 years of education Constant -2.384 -3.180 Adjusted [R.sup.2] .161 .193 Model 5 Model 6 B Beta B Beta SPF wave l (a) Temporally .530 ** .127 768 ** .186 stable SPFs (b) SPF .310 .074 .583 ** .140 transitioned to TPF (b) TPF -.009 -.002 .044 .009 transitioned to SPF (b) Child's age .182 *** .402 .143 *** .317 Male -.087 -.026 -.098 -.029 Mother's age .032 ** .117 Mother's .004 .006 education Mother's -.056 -.014 employed FT or PT Mother's .047 ** .125 household income Mother's CES-D Child's externalizing score > 63 Children's years of education Constant -.1884 -2.694 Adjusted [R.sup.2] .180 .199 Model 7 Model 8 B Beta B Beta SPF wave l (a) Temporally .732 ** .179 .753 ** .184 stable SPFs (b) SPF .526 ** .127 .587 ** .142 transitioned to TPF (b) TPF .070 .015 .159 .035 transitioned to SPF (b) Child's age .138 *** .302 .111 *** .244 Male -.099 -.030 .027 .008 Mother's age .030 .111 .024 .087 Mother's -.006 -.009 -.038 -.059 education Mother's .042 .010 .070 .018 employed FT or PT Mother's .040 .107 .032 .086 household income Mother's .009 .054 .011 .066 CES-D Child's -.370 -.087 -.226 -.053 externalizing score > 63 Children's .141 ** .214 years of education Constant -2.389 -3.155 Adjusted [R.sup.2] .185 .213 * p < .05; ** p < .01; *** p < .001. CES-D, Center for Epidemiological Studies Depression Scale; SPF, single-parent family; TPF, two-parent family. (a) Two-parent family at wave 1 = reference group; (b) Temporally stable two-parent family = reference group. Table 5 Multiple Regressions Assessing the Relative Effects of Family Structure, Mothers' Socioeconomic Status and Family Mental Health at Wave 1 On Children's Personal Income at Wave 3 Model 1 Model 2 B Beta B Beta SPF wave l (a) .053 .006 1.174 ** .130 Temporally stable SPFs (b) SPF transitioned to TPF (b) TPF transitioned to SPF (b) Child's age .668 *** .550 .577 *** .476 Male 1.726 *** .191 1.737 *** .192 Mother's age .055 .076 Mother's -.022 -.013 education Mother's .125 .012 employed FT or PT Mother's .171 * .170 household income Mother's CES-D Child's externalizing score > 63 Children's years of education Constant -10.229 -12.066 Adjusted R.sup.2] .328 .343 Model 3 Model 4 B Beta B Beta SPF wave l (a) 1.182 ** .131 1.184 ** .131 Temporally stable SPFs (b) SPF transitioned to TPF (b) TPF transitioned to SPF (b) Child's age .576 *** .465 .495 *** .400 Male 1.776 *** .196 2.154 *** .238 Mother's age .054 .074 .031 .043 Mother's -.045 -.026 -.155 ** -.089 education Mother's .155 .015 .215 .021 employed FT or PT Mother's .146 ** .145 .118 ** .117 household income Mother's -.009 -.020 -.004 -.010 CES-D Child's -.720 -.065 -.121 -.011 externalizing score > 63 Children's .470 *** .263 years of education Constant -11.260 -13.714 Adjusted R.sup.2] .330 .375 Model 5 Model 6 B Beta B Beta SPF wave l (a) Temporally .200 .018 .885 .079 stable SPFs (b) SPF -.007 -.001 .594 .052 transitioned to TPF (b) TPF -.434 -.034 -.330 -.026 transitioned to SPF (b) Child's age .698 *** .570 .636 *** .519 Male 1.789 *** .196 1.764 *** .193. Mother's age .039 .054 Mother's -.011 -.006 education Mother's .383 .035 employed FT or PT Mother's .110 ** .108 household income Mother's CES-D Child's externalizing score > 63 Children's years of education Constant -10.936 -12.381 Adjusted R.sup.2] .353 .361 Model 7 Model 8 B Beta B Beta SPF wave l (a) Temporally .958 .086 1.027 .092 stable SPFs (b) SPF .588 .052 .762 .067 transitioned to TPF (b) TPF -.223 -.018 .092 .007 transitioned to SPF (b) Child's age .639 *** .510 .547 *** .436 Male 1.815 *** .199. 2.261 *** .247 Mother's age .036 .049 .014 .020 Mother's -.032 -.018 -.147 -.082 education Mother's .440 .041 .479 .044 employed FT or PT Mother's .086 .084 .064 .062 household income Mother's -.004 -.010 .003 .006 CES-D Child's -.740 -.065 -.186 -.016 externalizing score > 63 Children's .483 *** .268 years of education Constant -11.724 -14.285 Adjusted R.sup.2] .347 .392 * p < .05;** p < .0l; *** p < .001. CES-D, Center for Epidemiological Studies Depression Scale; SPF, single-parent family; TPF, two-parent family. (a) Two-parent family at wave 1 = reference group; (b) Temporally stable two-parent family = reference group.