Family Structure Pathways and Academic Disadvantage among Adolescents in Stepfamilies

Authors


Abstract

Using National Longitudinal Study of Adolescent Health (Add Health) data, this research documents the prevalence of the different stepfamily forms in which American adolescents live, examines the family structure pathways through which adolescents traveled to arrive at their current family form, and explores the effects of these pathways on grades, school-related behavior, and college expectations (N = 13,988). Compared to those who have always lived with both biological parents, youth in pathways including divorce/separation or a nonunion birth experience significantly lower academic outcomes, while those whose pathways include parental death do not. Specific effects vary, however, according to the outcome examined. For example, the combination of divorce/separation and movement into the least common of family forms is associated with particularly poor GPA outcomes. Divorce/separation is also more detrimental than nonunion birth for college expectations, particularly when coupled with a transition into a stepfamily based on cohabitation. Divorce/separation and nonunion birth have similar, negative effects on school behavior problems. Overall, results indicate that living in a stepfamily does not benefit youth, and can in some ways disadvantage them, even compared to their peers in single-mother families. This is especially the case if youth transition into a stepfamily following a combination of stressful family experiences. These findings underscore the importance of examining family effects from a longitudinal perspective.

Stepfamilies now make up the fastest growing family type in the United States (Hetherington and Stanley-Hagan 2000), particularly if two-parent families formed via cohabitation (in which one adult is not biologically related to the child/ren) are included in that group.1 To date, however, most stepfamily research has not adequately accounted for the structural complexity and diversity of contemporary American living arrangements (Booth and Dunn 1994; Coleman, Ganong, and Fine 2000; Stewart 2001). We also lack a rigorous longitudinal examination of family structure over the life course of children living in stepfamilies (Cooksey 1997; Martinson and Wu 1992; Thomson, Hanson, and McLanahan 1994). Stepfamily formation can follow a divorce/separation of biological parents, nonmarital childbearing, or the death of one biological parent. These three possibilities represent very different kinds of family changes and experiences which may lead to differing effects on adolescent well-being.

The purpose of this article is to examine critically the differential effects of distinct family structure pathways on the academic outcomes of adolescents in stepfamilies. Throughout their lives, individuals experience a constellation of family changes that make up family structure pathways, beginning with family structure at birth and including each specific family structure transition that occurs thereafter. Examining the combination of the current and past family structure experiences youth have faced will be beneficial for several reasons. First, the results will enhance our knowledge of the various trajectories of family structure experiences that contemporary American youth encounter. Second, the results will allow us to determine whether the specific kinds of family structure experiences a child has endured condition the effects of their current family structure. Finally, the results will help us to better understand which family contexts place adolescents at greater risk for poor academic outcomes. In doing so, we will also be better able to evaluate the potential consequences of public policies, such as recent welfare reform and government-sponsored marriage initiatives, that provide incentives for single parents to marry.

Using the National Longitudinal Study of Adolescent Health (Add Health), this research documents the prevalence of differing stepfamily forms and the pathways by which contemporary American adolescents move into them. It also explores the effects of family structure pathways on several adolescent academic outcomes, including grades, school-related behavior, and college expectations. It addresses two specific research questions: (1) Do adolescents who have experienced different family structure pathways show different academic outcomes? (2) Do adolescents in different stepfamily forms experience poorer academic outcomes than their counterparts in two biological parent families, regardless of family structure pathways? That is, do some particular family structure pathways place adolescents in stepfamilies at risk for poor outcomes, while others do not?

Academic Outcomes of Youth

Adolescent academic outcomes are significantly related to long-term employment, occupational, and economic well-being outcomes (Haveman and Wolfe 1994; Huurre et al. 2006; Rindfuss, Cooksey, and Sutterlin 1999; Wiesner et al. 2003). It is during adolescence that academic attitudes and performance begin to have real effects upon later educational attainment and, ultimately, upon adult socioeconomic achievement. An understanding of how family environments influence academic outcomes during adolescence is, therefore, an important step toward understanding how families of origin continue to influence socioeconomic success during adulthood, and is necessary for the development of useful interventions that could help target young people at risk for lowered socioeconomic attainment.

Much research has shown that family structure is associated with the academic outcomes of children and adolescents. In general, adolescents in stepfamilies fare moderately worse than those living with both biological parents in terms of grades, achievement scores, high school completion, school attendance (Astone and McLanahan 1991, 1994; McLanahan and Sandefur 1994; Pong 1997; Pong and Ju 2000; Zill 1996), and behavior (Day 1992; Hoffmann and Johnson 1998). Controlling for demographic and socioeconomic characteristics, youth in stepfamilies often exhibit lower levels of academic achievement than do their peers in single-mother families (Zill 1996). Studies have also shown that, holding marital status constant, living in a stepmother family is generally associated with more psychological and social problems than is living in a stepfather family (Cherlin 1992; Fine and Kurdek 1995; Zill 1996). These results suggest that youth living in stepmother families may also face a risk of poor academic outcomes. Although some recent research indicates that adolescents in married and cohabiting stepfamilies may not vary significantly from one another (Brown 2004), other work shows that youth in cohabiting stepfamilies experience lower levels of academic achievement and higher levels of school-related behavior problems than do youth in married stepfamilies (Manning and Lamb 2003; Thomson, Hanson, and McLanahan 1994).

A number of important factors are associated with both academic outcomes and current family structure. For example, the academic outcomes of children and youth are directly and indirectly compromised by economic deprivation, which is strongly associated with stepparent and single-parent family structures. Youth in these family forms, particularly in cohabiting stepfamilies and single-mother families, tend to experience lower incomes, lower levels of parental education, and higher levels of poverty (Amato 1993; McLanahan and Sandefur 1994; Pong 1997; Pong and Ju 2000; Thomson 1994; Thomson, Hanson, and McLanahan 1994). Being African American or Hispanic, a boy, or an older adolescent, having a larger number of siblings, attending public school, and having low verbal ability are also factors associated with both poorer academic outcomes and living in a stepparent or single-parent family (Downey 1995; Entwisle, Alexander, and Olson 1997). Being an immigrant, on the other hand, is often associated with both intact family structures and better academic outcomes (Tillman, Guo, and Harris 2006; Portes and Rumbaut 1996). Stepfamilies in general, and cohabiting stepfamilies in particular, tend to have less well-established family roles and relationships and produce lower levels of parental support and involvement than do other family forms (Menaghan, Kowaleski-Jones, and Mott 1997). In addition to being a less common family form, stepmother family arrangements are also selective of older boys, children who have preexisting behavioral problems, and children whose mothers are incarcerated, engaged in substance abuse, or have a history of psychological problems (Cherlin 1992).

These factors have all been associated with poorer child outcomes, and appear to explain a portion of the academic disadvantage experienced by youth living in stepfamilies (McLanahan and Sandefur 1994; Pong 1997; Thomson 1994; Thomson, Hanson, and McLanahan 1994), as well as some of the differences in outcomes by stepfamily type. However, significant differences between stepfamily types and inconsistencies in the stepfamily literature remain. This may be, in part, because specific types of past family structure experience are not fully accounted for in previous research. Reliance upon single time-point measures of family structure limits our ability to understand why various outcomes may exist across family types. Life changes experienced throughout childhood inevitably alter long-term social trajectories (Elder 1985, 1998). When family structure experiences are not measured across the life course, crucial differences in the family contexts of children may be masked (Cooksey 1997; Martinson and Wu 1992; Thomson, Hanson, and McLanahan 1994).

Family Structure and Theory

Researchers have explored a variety of theoretical explanations for the negative effects of stepfamily living. Those that best address research questions pertaining to the significance of specific family structure pathways, above and beyond the importance of current family structure, can be drawn from variations of stress theory and theories regarding parental investment.

Stress theory centers on the idea that stressful life events produce role strain, reduce self-esteem and mastery and, ultimately, lead to more negative social and psychological outcomes (Amato 2000; Pearlin et al. 1981; Plunkett et al. 1997; Thoits 1995). In particular, the multiple negative events generated by alterations in parental union status produce detrimental levels of stress for both parents and children. Parental union formations and dissolutions are directly associated with role transitions for children (Harris 1997) and with many other stressful changes in family lifestyle and resources (McLanahan and Sandefur 1994; Menaghan, Kowaleski, and Mott 1997). For example, family structure change is often accompanied by changes in economic status and maternal employment patterns, residential moves, and the loss of contact with a nonresident parent and other members of a social support system (Amato, Lomis, and Booth 1995; Cherlin 1992; McLanahan and Sandefur 1994). Stressed parents may respond to family changes with a diminished quality of parenting, supervision, and interaction (Hoffmann and Johnson 1998).

Stepfamily formation, in particular, also leads to the introduction of new household members and family relationships, a change in social relations and networks (Wallerstein, Corbin, and Lewis 1988), and usually the need to adapt to new household routines and activities (Menaghan, Kowaleski, and Mott 1997). Our society does not have a well-established set of norms dictating the appropriate roles for stepfamily members, especially when families are formed through cohabitation or when children from multiple families are blended into one household. Nor does our society have a network of institutionalized support to help children and parents adjust to stepfamily life (Cherlin 1978). The stressors associated with stepfamily formation may be amplified for youth who live with half- and/or step-siblings (Ganong and Coleman 1994). Some recent research has indicated that, net of socioeconomic and background characteristics, living in a “blended” family is associated with significantly more negative outcomes for youth (Ginther and Pollak 2004).

In addition to increasing exposure to stress, stepfamily formation also leads to the introduction of a new parent figure, who may be less willing to invest time, emotion, and money toward a child's development and academic success. For example, stepparent–child relationships, on average, are more conflict-ridden than relationships with biological parents, and stepparents tend to offer children less parental support, closeness, and supervision (Buchannan, Maccoby, and Dornbusch 1996; Cherlin 1992; Coleman, Ganong, and Fine 2000). The introduction of a stepparent also generally leads to a decline in the amount of undivided attention and supervision that children receive from their resident biological parent (Pong 1997). Furthermore, stepparents report feeling significantly lower levels of obligation to provide financial support for stepchildren's postsecondary education (Aquilino 2005). Both biological parents and stepparents report providing less financial support for their children's education when they are living in a stepfamily (Zvoch 1999). Lowered social and financial investments may signal to children a lack of parental interest and lowered expectations for academic achievement and college attendance. In turn, children in stepfamilies may be less likely to seek and obtain academic assistance when needed and may not be as motivated to reach for academic success or to pursue higher education.

An accumulation of multiple life changes and deficits in parental investment can threaten a young person's sense of security (Seltzer 1994), reduce access to material and emotional resources, and disrupt behavior and academic performance (Capaldi and Patterson 1991; Menaghan, Kowaleski, and Mott 1997). As a result, the number of family structure changes experienced during childhood and the length of exposure to stepparent and single-parent family forms are known to be associated with negative youth outcomes (Ganong and Coleman 1994; Hetherington and Stanley-Hagan 2000; Kurdek and Fine 1993; Wojtkiewicz 1993a, 1993b; Wu 1996; Wu and Martinson 1993). These findings indicate the importance of considering family structure and family changes over time; that is, by taking a longitudinal perspective when examining family effects.

Simply measuring current family structure, the number of family structure transitions or the length of time spent in a given family form may obscure important differences in the effects of specific family structure pathways on the well-being of youth. The formation of a stepfamily can occur through either marriage or cohabitation. It can also take place following a divorce/separation of biological parents, a nonmarital/non-cohabiting birth (henceforth I will refer to this as a “nonunion” birth), or the death of a biological parent. Different family structure pathways are associated with different experiences for children. For example, although they may experience the same number of family structure transitions and the same length of exposure to their current family, adolescents who transition into a stepfamily following the divorce of biological parents may experience different emotional reactions and lifestyle adjustments than adolescents who transition into a stepfamily following the death of a biological parent. The effects of that transition may, furthermore, depend upon whether the stepfamily is formed through marriage or cohabitation. Varying kinds and combinations of family change may produce different levels of stress for parents and children and be associated with different levels of adult-to-child investment. As such, some pathways may lead to a greater accumulation of negative family-related experiences than others. Thus, the outcomes of individuals with the same current family structure, but different family structure pathways, may vary significantly.

Family Structure Pathways and Differential Academic Outcomes

According to stress theory and theories of parental investment, pathways associated with more stressful experiences or more limited investments will be more likely to lower emotional well-being, lower access to social and material resources, and, ultimately, disrupt the behavior and academic performance of adolescents (Capaldi and Patterson 1991; Menaghan, Kowaleski, and Mott 1997; Pong and Ju 2000). Disentangling these effects is difficult, undoubtedly, and little research has assessed the effects of specific family structure pathways on youth well-being.

The majority of stepfamilies are formed following a divorce, and a growing proportion is formed following a nonunion birth or the separation of cohabiting parents (Cherlin and Furstenberg 1994). Although youth who have never lived with both biological parents may have accumulated more exposure to deficits in economic and parental resources, they have likely experienced fewer stressful family changes than have the children of divorced or separated families. Divorce and separation are often preceded by high levels of family conflict and tension that negatively affect parent–child interaction and child well-being (Furstenberg and Cherlin 1991). The processes of separation and divorce can also engender additional conflict, which may escalate when a child's resident parent forms new romantic relationships (Furstenberg and Cherlin 1991). Furthermore, compared to youth who have never lived with both biological parents, those who have experienced divorce/separation are more likely to have regular contact with their nonresident parent (Seltzer and Bianchi 1988). This contact is associated with higher levels of conflict between children and their resident parents (Cherlin 1992; Furstenberg and Cherlin 1991) and between biological parents. Contact with nonresident parents also increases the likelihood that youth will have to manage relationships with multiple father figures or multiple mother figures. Conflict, hostility, role confusion, and parental dissatisfaction with visitation are all detrimental to youth outcomes (Acock and Demo 1999; King and Heard 1999; Plunkett and Henry 1999), and have been shown to have negative effects on self-esteem and educational attainment (Amato 1993; Biblarz and Gottainer 2000). Thus, youth whose biological parents have divorced/separated may face a greater accumulation of stressful experiences than those who have never lived with both biological parents.

Despite the grief associated with a parent's death, the loss of a parent in this manner might be more easily understood and accepted by a child than either the loss of a parent as the result of divorce/separation or the lack of contact with a parent who has always been nonresidential. The ongoing conflict and negative experiences associated with divorce and separation often lead children to develop hostile feelings toward their nonresidential parent (Parish and Kappes 1980; Rozondal 1983). The death of a parent, on the other hand, is more often associated with the development of positive memories of and feelings toward the missing parent (Silverman, Nickman, and Worden 1992). Furthermore, although the death of a parent during childhood is no longer as common an event in the United States as it once was, society-wide norms regarding the appropriate way to interact with and support grieving families are fairly well established. As a result, children who have lost a parent to death are more likely to maintain a relationship with and receive financial support from grandparents or other adults outside of their households than are children of divorce or nonunion childbearing (Drew and Smith 1999; Johnson 1992).

Families that experience a death also tend to experience less of a decline in their standard of living than do those that experience divorce or separation, in part because the U.S. government provides guaranteed support to both widowed spouses and children in the form of social security survivor's benefits (Biblarz and Gottainer 2000; Holden and Smock 1991). These continuing social and financial supports may help youth in widowed families better manage the stress of their loss. Overall, children of widowed parents report significantly less life stress, family conflict, antisocial behavior, anxiety and depression, and more supportive parenting than do children of divorced parents. In fact, children of widowed parents do not significantly differ on these measures from children of continuously married parents (Short 2002). So, while family structure pathways that include the death of a parent may be associated with a great deal of emotional upheaval for youth and their surviving parent, the greater access to social supports and economic resources may lead to the accumulation of fewer negative changes, overall, than pathways that involve divorce or a nonunion birth.

Youth who live in stepfamilies are also affected by their current living arrangements. Adolescents in the less common types of stepfamilies may face more stressful and confusing family relationships and roles as a result of low institutional and normative support for them (Cherlin 1992; Menaghan, Kowaleski, and Mott 1997; Thomson 1994). For example, youth may see married and cohabiting stepparents as filling very different roles in their lives. A married stepparent may be viewed as a more permanent and committed member of the family, thereby deserving of a closer and more prominent role in the child's life. A cohabiting stepparent, on the other hand, may be viewed as the biological parent's temporary romantic partner, who competes with the child for the parent's time and attention (Buchannan, Maccoby, and Dornbusch 1996). The stress associated with the family relationships and roles in less common stepfamily forms may be amplified by the fact that these youth, particularly those living with stepmothers, are more likely to have contact with nonresident parents than youth in married stepfather families (Brown 2002; Raley and Wildsmith 2004).

For reasons such as these, the less common stepfamily forms may be associated with a greater accumulation of family-related stress. In fact, stepmothers often report their family situations as more stressful than do stepfathers (Fine and Schwebel 1992; Hobart 1991; Whitsett and Land 1992) and child–stepmother relationships are more conflict-ridden than child–stepfather relationships (Cherlin 1992; Clingempeel and Segal 1986; Ganong and Coleman 1994; Kurdek and Fine 1993). Cohabiting stepparents tend to offer stepchildren less social support, closeness, and supervision than do married stepparents (Coleman, Ganong, and Fine 2000), and children are more likely to be engaged in conflict with a cohabiting stepparent than with a married one (Buchannan, Maccoby, and Dornbusch 1996). In general, youth in cohabiting families report more emotional and behavioral problems and lower levels of school engagement than do youth in married families (Brown 2004; Manning and Lamb 2003).

It is the combination of the past and the current family structure, however, that is of most interest here. Guided by theory and the past research on divorce, widowhood, nonunion birth and current family structure, I expect certain kinds and combinations of family structure experiences to produce a greater accumulation of family-related stress and investment deficits than others, and, as a result, to be associated with more negative individual outcomes for youth.

In particular, I expect the following: (1) all family structure pathways that include stepfamily or single-parent family living will be associated with more disadvantaged outcomes than those that include only two biological parent living, regardless of the specific pathways experienced; (2) family structure pathways that include divorce/separation will be associated with more disadvantaged outcomes than other pathways involving change; (3) family structure pathways that include the death of a biological parent will be associated with less disadvantaged outcomes than other pathways involving change; (4) family structure pathways that include a combination of divorce/separation and a transition into a less common stepfamily form (e.g., cohabiting stepfather, married stepmother) will be associated with the greatest academic disadvantage; and (5) the number of prior family structure transitions and the length of exposure to stepparent and/or single-parent families that an adolescent experiences will be negatively associated with academic outcomes. However, these factors will not fully explain the divergent outcomes associated with the various family structure pathways.

Data and Methods

Data come from the National Longitudinal Study of Adolescent Health (Add Health), a nationally representative study of adolescents in grades 7 through 12 in the United States. The study entailed in-depth interviews with adolescents and their parents (conducted separately) and provides detailed information regarding child outcomes, family composition, and family structure experiences. Add Health used a multistage, stratified, school-based, cluster sampling design, including students from 80 public and private high schools, as well as students from one junior high or middle school feeding into each high school (Harris et al. 2003).

Add Health involves multiple waves of data collection and several data collection components. This research utilizes data collected from In-Home interviews during Wave I (1994–1995) and selected data from the Wave I Parental Questionnaire and School Administrator Questionnaire. The parental questionnaire provides retrospective accounts of parent relationship histories, including information on the three most recent “marriages or marriage-like relationships” having occurred over the past 18 years (98 percent of parents indicate having had three or fewer such relationships). These histories pinpoint the year in which all marriages or cohabitations began and ended, and indicate whether past relationships ended through divorce, separation, or death. The adolescent in-home interview provides information regarding the length of time that the respondents had lived with their resident parent(s), and whether and for how long they had lived with any nonresident parents. I combined the parental histories and the adolescent reports to capture the family structure pathways that adolescents have lived in throughout their lives.

The sample includes 13,988 adolescents who had a completed Parental Questionnaire, and who lived with at least one biological parent. Adolescents not living with a biological parent (about 5 percent of all Add Health respondents) were excluded because their parental questionnaires may fail to provide accurate family structure histories. Because immigrants are more likely than nonimmigrants to have incomplete parental questionnaires, a measure of immigrant generation status is also included in all analytical models to address potential bias in linking the Add Health parent data to the adolescent data (Harker 2000).

Although this framework assumes that social causation underlies the relationship between family structure pathways and adolescent academic outcomes, social selection likely plays a role. For example, stepfamilies that result from either family disruption or nonunion birth may also be characterized by preexisting negative parental and familial characteristics that precipitated those family events and continue to exert a negative effect upon youth outcomes (Capaldi and Patterson 1991; Cherlin et al. 1991; McLanahan and Sandefur 1994). In contrast, entering a stepfamily following the death of a parent may represent a more random process. The characteristics of widowed families, therefore, may more closely resemble those of people living in two biological parent families. Although previous research has shown that family structure effects on educational outcomes remain statistically significant even when controlling for unmeasured family influences (Sandefur and Wells 1999), causal inferences should be drawn from the findings of this study with caution. This work improves upon much previous stepfamily research by empirically modeling the pathways that lead to current living arrangements, yet the data used do not allow for extensive examination of parental and child characteristics that pre-date data collection.

Measures

Academic Outcomes.  The three dependent variables are grade point average (GPA), school-related behavior problems, and college expectations. GPA is a self-reported, continuous variable ranging from 1 (D/F) to 4 (A), which measures grades for the past academic year.2 School-related behavior problems are measured as an index (α = .69) representing the mean item score across four self-reported, five-category ordinal items (having trouble getting along with students, getting along with teachers, paying attention in school, and getting homework done) with responses ranging from 0 to 4 (Never to Every day). College expectations are assessed through the adolescents’ responses to a question regarding the likelihood of attending college. These expectations were originally measured on an ordinal scale, ranging from 1 (low) to 5 (high). Because the distribution is very highly skewed toward the upper end of the distribution (the majority of the sample responded with a “5”), responses were dichotomized to indicate holding expectations of college attendance that are below average for the sample (responses of 1–4) versus expectations of college attendance that reflect the sample's average response (responses of 5).

Current Family Structure.  Measures of current family structure are created for use in the descriptive analyses, which provide a picture of the current living arrangements of America's youth. Current family structure captures both the biological and legal relationships between the adolescent and all co-resident parent figures, as reported by the adolescent. Adolescents are classified as living in two biological parent families (N = 8189), married stepfather families (N = 1631), married stepmother families (N = 357), cohabiting stepfather families (N = 335), cohabiting stepmother families (N = 35), single-mother families (N = 3091), or single-father families (N = 343).3

Family Structure Pathways.  Family structure pathways, the main independent variable in the multivariate analyses, is based on a combination of current family structure status (as described above) and information on the past marital/relationship experiences of the respondent's resident biological parent. Past marital/relationship experiences are coded according to whether the resident biological parent had been divorced or separated from the adolescent's other biological parent (with whom they had lived during marriage or cohabitation); had never been married to and never cohabited with the other biological parent; had been widowed by the other biological parent (while living together); or had always lived with the other biological parent.4

Each respondent is classified as having experienced one of 19 possible combinations of current family structure and past marital/relationship history. Low frequencies among some of the family structure pathways necessitated that I drop nine of the less frequent ones when conducting the multivariate analyses. Thus, the multivariate models include a total of 10 family structure pathways. For descriptive purposes, however, it is important to explicate all 19 of the possible family structure pathways. I show 10 generalized family structure pathways in Figure 1. With the exception of Pathway 1, each is further divided in two when used in the analyses (leading to a total of 19 pathways), depending upon whether the resident biological parent is a mother or a father.

Figure 1.

Possible Family Structure Pathways.
Note: The final step of each pathway, with the exception of Pathway 1, can be further divided into two categories, indicating the gender of the resident biological parent.

The possible pathways indicated in Figure 1 include living with two biological parents who have lived with one another since the respondent's birth (Path 1; N = 8189); living with a single mother following a divorce/separation of biological parents (Path 2; N = 1637), a nonunion birth (Path 10; N = 1257), or the death of one biological parent (Path 7; N = 197); living with a single father following a divorce/separation (Path 2; N = 225), a nonunion birth (Path 10; N = 83), or a death (Path 7; N = 35); living with a married stepfather following a divorce/separation (Path 3; N = 935), a nonunion birth (Path 9; N = 629), or a death (Path 6; N = 67); living with a married stepmother following a divorce/separation (Path 3; N = 243), a nonunion birth (Path 9; N = 82), or a death (Path 6; N = 32); living with a cohabiting stepfather following a divorce/separation (Path 4; N = 216), a nonunion birth (Path 8; N = 99), or a death (Path 5; N = 20); living with a cohabiting stepmother following a divorce/separation (Path 4; N = 20), a nonunion birth (Path 5; N = 10), or a death (Path 5; N = 5).

Due to low frequencies, I could not separately examine all of the family structure pathways in the multivariate analyses and nine pathways were ultimately dropped.5 These pathways included youth living in all widowed families (N = 159), with the exception of widowed single-mother families. Also included were youth in the remaining cohabiting stepmother families (N = 30), and youth in married stepmother and single-father families formed following a nonunion birth (N = 165). As a result, the multivariate models presented in this text include a total of 10 family structure pathways. The primary reference category consists of adolescents who have always lived with both biological parents.

Family Structure Transitions and Family Structure Exposure.  Simple measures of family structure transitions and exposure to various family structures are used primarily in the descriptive analyses, but are also briefly used in the multivariate analyses in order to establish the usefulness of the more complex measures of family structure pathways. Family structure transitions are measured as the cumulative number of times an adolescent had experienced a change in his/her resident family structure (i.e., the formation or dissolution of either a marriage or a cohabiting relationship). Exposure to stepparent and/or single-parent family structures is measured as the proportion of each adolescent's entire life (measured in years) spent living in a single-parent or stepparent family. Exposure to current family structure is measured as the proportion of each adolescent's entire life (measured in years) spent living in his/her current family structure.

Control Variables.  Controls for economic deprivation are measured as follows: total family income in 1994 (under $16,000; $16,000–$34,999; $35,000–$59,999; $60,000 or more; missing), the highest level of education obtained by a resident parent (less than high school education; high school diploma or GED; greater than high school; missing), and resident mother's working status (employed full-time/not employed full-time). I also control for the respondent's race (white, black/African American, Hispanic/Latino, or Asian), gender, age in years, number of resident siblings, school sector (public versus private), Revised Peabody Picture Vocabulary Test scores (AH-PVT), and immigrant generation status (1st, 2nd, and 3rd + generation youth). First generation immigrants are individuals who were born in a foreign country (and not as a U.S. citizen) to foreign-born parents. Second generation youth were born in the United States to at least one foreign-born parent. Third-plus generation youth were born in the United States to native-born parents. Finally, given the interrelated nature of grades, academic expectations and behavior (Zimmerman, Caldwell, and Bernat 2002), the analysis examining GPA includes controls for the respondent's college expectations and school behavior. The analysis examining school-related behavior includes controls for college expectations and GPA, and the analysis of college expectations controls for GPA.

Analysis Strategy

To provide an accurate depiction of the family living arrangements in which contemporary American children are being raised, the descriptive analyses document the prevalence of each current family structure and then turn to an examination of adolescents’ family structure pathways. The multivariate analyses examine the effects of specific family structure pathways on academic outcomes. I use Ordinary Least Square (OLS) regression analyses to study the effects of family structure pathways on GPA and school-related behavior problems, and logistic regression analyses to study the effects on college expectations. To adjust for the multistage, stratified, school-based, cluster sampling design, the models are estimated by using the robust estimator of variance (a Huber or White estimator of variance) in STATA. Differential sampling probabilities among individuals are controlled by using the Add Health grand sample weights in all estimation procedures (Chantala and Tabor 1999).

Descriptive Results

Table 1 presents a view of the current family structure distribution of 13,988 adolescents in grades 7–12. Of these adolescents, 58.5 percent lived in a two biological parent family, 16.9 percent lived in a stepfamily, and 24.5 percent lived in a single-parent family. Of those in stepfamilies, 69.2 percent lived with married stepfathers, 15.1 percent with married stepmothers, 14.2 percent with cohabiting stepfathers, and 1.5 percent with cohabiting stepmothers (see Table 1).

Table 1.  Weighted Means of Academic Outcomes, by Current Family Structure
 2-bio parents N = 8189 (58.5%)All step-families N = 2358 (16.9%)All single parents N = 3434 (24.5%)Full sample N = 13,988 (100%)StepfamiliesSingle-parent families
Married stepfather N = 1631 (11.7%)Married stepmother N = 357 (2.6%)Cohabiting stepfather N = 335 (2.4%)Cohabiting stepmother N = 35 (0.3%)Single mother N = 3091 (22.1%)Single father N = 343 (2.5%)
  1. Note: Means in the same row that do not share subscripts differ at P < .05 in Chi-square tests.

Self-reported GPA (1–4)2.93a2.68b2.64b2.822.71b2.76b2.46c2.48b2.65b2.50c
School-related behavior problems (0–4) .99a1.15b1.14b1.061.13b1.18b1.26c1.09a1.13b1.22c
Lower than average college expectations (0–1) .39a .51b .54b .45 .48b .51b .64c .65a .53b .62c

Table 1 also indicates that respondents reported fairly positive academic outcomes. The average self-reported GPA was 2.8 (approximately a B/B-), the average youth reported “rarely” experiencing school-related behavior problems, and only 45 percent of youth reported college expectations that were below the highest response possible. Chi-square tests indicate, however, that academic outcomes varied according to measures of family structure. Compared to adolescents in almost all other family forms, those living in two biological parent families reported significantly higher GPAs, reported fewer school-related behavior problems, and were less likely to hold lower than average college expectations.

Chi-square tests also indicate that significant differences in academic outcomes exist between adolescents in the various types of stepparent and single-parent families. For example, youth in married stepfather families are more likely than their peers in two biological parent families to express lower than average expectations for college attendance (48 percent of the former group as opposed to 39 percent of the latter), but they are less likely to do so than youth in cohabiting stepfather, single-father, and cohabiting stepmother families. Between 62 to 65 percent of youth in these family forms reported lower than average college expectations. Across the three academic outcomes, youth in cohabiting stepfather and single-father families tend to fare significantly worse than youth in other family forms (see Table 1). Youth in cohabiting stepmother families also appear to fare more poorly in terms of college expectations and grades (chi-square tests do not indicate statistical significance here, most probably because of the very small number of respondents in this family type).

Table 2 presents a distribution of the 10 complex family structure pathways that are included in the multivariate analyses. As can be determined by the number of youth found in each family structure pathway, a greater proportion of those living in a stepfamily (as opposed to a two biological parent family or a single-parent family) have pathways that include a divorce/separation of their biological parents (see Table 2). Additional descriptive results (shown in Appendix A) indicate that youth living in stepfamilies have also experienced a greater average number of family structure transitions, have spent more of their lives in either stepparent or single-parent families, and have spent less of their lives in their current family situation than other youth. Overall then, youth who currently live in stepfamilies are more likely to have experienced the kinds of stressful family structure events and pathways that may negatively affect their ability to do well academically.

Table 2.  Weighted Means of Demographic and Family Background Characteristics, by Family Structure Pathway
 2-bio parents-always together (N = 8189)StepfamiliesSingle-parent families
Married stepfather-divorced/separated (N = 935)Married stepfather-never married/cohabited (N = 629)Married stepmother-divorced/separated (N = 243)Cohabiting stepfather-divorced/separated (N = 216)Cohabiting stepfather- never married/cohabited (N = 99)Single mother-widowed (N = 197)Single mother- divorced/separated (N = 1637)Single mother-never married/cohabited (N = 1257)Single father-divorced/ separated (N = 225)
Gender (male = 1).52.49.50.57.53.64.45.48.46.70
Age in years15.7415.7915.7115.9515.5615.5315.8115.6915.8016.24
White/Caucasian.77.81.60.80.65.55.58.60.38.75
Black/African American.08.08.21.06.15.22.24.24.46.12
Hispanic/Latino.11.09.16.10.17.19.13.12.13.09
Asian.04.01.02.04.01.01.04.03.02.04
1st generation immigrant.04.03.04.04.03.02.06.04.05.04
2nd generation youth.11.08.09.07.14.08.07.10.06.11
3rd generation youth.85.89.87.89.83.89.88.86.89.85
Number of resident siblings1.491.411.711.931.321.231.441.261.30.89
Revised Peabody Picture Vocabulary Scores103.70103.1098.10102.1099.7196.8199.30100.7095.20100.70
Public school attendance.91.93.95.95.97.98.92.94.97.91
Resident mother works full-time.47.58.51.42.55.46.48.65.53N/A
Family income ≤$15,999.06.09.14.09.27.30.23.34.44.15
Family income $16,000–$34,999.18.22.29.19.37.36.47.34.26.35
Family income $35,000–$59,999.33.33.28.30.18.10.16.18.09.32
Family income ≥$60,000.30.26.19.30.07.11.05.06.05.11
Family income missing.13.10.10.12.11.13.09.08.16.08
Parents < high school education.08.08.13.06.19.17.23.14.24.14
Parents have high school education.28.31.34.33.36.36.28.35.38.34
Parents > high school education.62.59.49.57.37.30.49.47.33.49
Parents’ education missing.02.02.04.04.08.17.06.04.05.03

Table 2 also presents a weighted mean distribution of demographic and family background characteristics by family structure pathway. Compared to other youth, those who have experienced family structure pathways that include cohabitation, single parenthood or nonunion births are more likely to report demographic and family background characteristics that are associated with poorer adolescent academic outcomes. For example, those in married stepfather families formed following a nonunion birth, cohabiting stepfather families formed following either a divorce/separation or a nonunion birth, and single-mother families (particularly those formed following a nonunion birth), are more likely to be African American, more likely to have lower levels of family income and parental education, and more likely to have lower AH-PVT scores. Youth in single-father families formed after a divorce/separation are also more likely to be male, older, have lower levels of family income, and have parents with lower education levels.

Multivariate Results

Measures of family structure pathways were used in the multivariate analyses, where a set of models was estimated for each of the three dependent variables: GPA, school-related behavior problems, and lower than average college expectations. In each set of models, the first model establishes outcome differentials by family structure pathways, with the “two biological parents, always living together” pathway used as the primary reference category. The second model controls for demographic and socioeconomic characteristics that may explain some of the pathway differentials. The third model adds controls to determine whether any remaining pathway differentials may be explained by other related academic outcomes.

An additional set of models was estimated for each of the outcome variables to determine whether controlling for the number of prior family structure transitions and the length of exposure to stepparent and/or single-parent families can mediate the direct effects of the various family structure pathways. As these measures did not mediate the significant pathway differences for any of the outcome variables and did not significantly improve the fit of the analytical models, the results are not shown in the tables (results are available upon request).

Grade Point Average.  Although family structure pathways by themselves appear to account for only a small amount of the variance in GPA outcomes (approximately 4 percent), Model 1 in Table 3 indicates that all but one of the family structure pathways are significantly associated with GPA. Compared to youth who have always lived with both biological parents, youth who have experienced any pathway involving the divorce/separation of biological parents or a nonunion birth (a birth that occurs to a woman who is neither married to nor cohabiting with the child's father) report significantly, although moderately, lower GPAs. Living in a single-mother family following the death of a biological parent, however, is not associated with a significantly different GPA than is continuously living with both biological parents.

Table 3.  Summary of Ordinary Least Square (OLS) Regression Analysis for Family Structure Pathways and Other Variables Predicting GPA (N = 13,627)
VariableModel 1Model 2Model 3
BSE BBSE BBSE B
  • *

    P < .05;

  • **

    P < .01;

  • **

    P < .001.

  • Notes: †indicates significant difference (P < .05) from the following family structure pathways: married stepfather family formed following divorce/separation and married stepfather family formed following a nonunion birth (see footnote 6); #indicates significant difference (P < .05) from the following family structure pathways: single-mother family formed following divorce/separation, single-mother family formed following a nonunion birth, single-mother family formed following a death (see footnote 7).

2-bio parents—always lived together (reference category)
Married stepfather—bio parents divorced/separated−.187***.036−.163***.032−.104***.030
Married stepfather—bio parents never married/cohabiting−.284***.049−.157***.044−.099**.038
Cohabiting stepfather—bio parents divorced/separated−.452***.077−.281***, , #.072−.161**, #.065
Cohabiting stepfather—bio parents never married/cohabiting−.452***.70−.189**.069−.115.066
Married stepmother—bio parents divorced/separated−.173*.070−.122.067−.037.049
Single mother—bio parent widowed−.124.092 .023.083 .021.070
Single mother—bio parents divorced/separated−.222***.027−.098***.025−.034.021
Single mother—bio parents never married/cohabiting−.369***.032−.136***.032−.080**.031
Single father—bio parents divorced/separated−.443***.062−.281***.060−.174**.054
Gender (male = 1, female = 0)  −.236***.020−.135***.018
Years of age  −.030***.008−.031***.007
Black/African American  −.024.033−.078**.029
Hispanic/Latino  −.084*.042−.079*.038
Asian   .081.048 .076.049
1st generation immigrant   .306***.050 .183***.053
2nd generation youth   .054.037 .024.033
Number of resident siblings   .011.008 .013.007
Revised Peabody Picture Vocabulary Scores   .013***.001 .011***.001
Public school  −.093.055−.081.045
Resident mother works  −.006.016−.011.015
Family income ≤$15,999  −.152***.032−.089**.029
Family income $16,000–$34,999  −.157***.031−.105***.028
Family income $35,000–$59,999  −.075***.023−.047*.021
Family income missing  −.082**.032−.047.028
Parents < high school education  −.226***.036−.143***.034
Parents have high school education  −.172***.019−.119***.018
Parents’ education missing  −.290***.047−.199***.045
Lower than average college expectations    −.334***.019
School-related behavior problems    −.307***.014
R-square.035 .169*** .313***

Controls for demographic and socioeconomic characteristics (Model 2) reduce the significance of the differences found between the GPAs of youth who have always lived with two biological parents and those of youth whose pathways include divorce/separation or nonunion birth. This indicates that the lower GPAs associated with pathways that include divorce/separation and nonunion birth may be partially explained by lower socioeconomic resources and a greater likelihood of being a racial/ethnic minority. Still, most of the GPA differences that were statistically significant in the previous model remain so.

Overall, compared to peers whose family structure pathways include continuously living with both biological parents, youth whose pathways include living in any stepfather family (either married or cohabiting) following either a divorce/separation or a nonunion birth face a significant GPA disadvantage. Contrary to expectations, however, the magnitude of GPA disadvantage does not differ significantly between youth living in a married stepfather family following a divorce/separation and those living in a married stepfather family following a nonunion birth.6 Similarly, there is no significant difference between the magnitude of disadvantage experienced by youth living in cohabiting stepfather families following divorce/separation and those in cohabiting stepfather families following a nonunion birth. Youth in cohabiting stepfather families formed following divorce/separation, however, do report a significantly larger GPA disadvantage than youth who have experienced any of the family structure pathways that include living in a married stepfather family (significance indicated on Table 3).

Additional regression analyses also indicate that there are no significant differences in the grades of youth whose family structure pathways include living in a married stepfather family following either a divorce/separation or a nonunion birth and those whose pathways include living in a single-mother family following the same kinds of family structure experiences.7 Thus, holding constant past family structure experiences, youth in married stepfather families do not fare significantly better or worse than youth in single-mother families. Youth living in cohabiting stepfather families formed following divorce/separation, however, do report significantly lower grades than youth living in single-mother families that were formed following a divorce/separation (significance indicated on Table 3).

Thus, with the exception of youth living in a single-mother family following the death of a parent, youth currently living in stepparent and single-parent families report significantly lower GPAs than youth who have continuously lived with both biological parents, regardless of their past family structure experiences. At the same time, youth who live in cohabiting stepfather families formed following a divorce/separation tend to experience a greater magnitude of GPA disadvantage than youth who have ended up living in married stepfather families or single-mother families formed after either a divorce/separation or a nonmarital birth. As a result, the findings provide some support for the expectation that youth who have experienced family structure pathways that include a combination of divorce/separation and a transition into a less common family form tend to be among the most disadvantaged in terms of grades.

The final model of Table 3 adds controls for college expectations and school-related behavior problems. Results indicate that both of these measures are highly significant predictors of GPA. Including these controls also further reduces the significant differences in GPA by family structure pathway. For example, the significant difference between the two biological parent/always together pathway and the pathway including cohabiting stepfather families formed following a nonunion birth are completely mediated. Thus, the lower college expectations and poorer school-related behavior of youth who have experienced the latter pathway help to explain their poorer grades. Even after these controls are added to the models, however, youth in cohabiting stepfather families formed after divorce/separation continue to face a significant GPA disadvantage. Not only do they earn significantly lower grades than youth who have continuously lived with both biological parents, but they also continue to earn significantly lower grades than youth in all family structure pathways that include living in a single-mother family (significance indicated on Table 3). As a result, there remain significant pathway differences in GPA that cannot be fully explained by demographic and family background characteristics, academic expectations, or school-related behavior problems.8

In sum, compared to youth who have always lived with both biological parents, youth who have experienced any family structure pathway that includes either a married or cohabiting stepparent family or a nonwidowed single-parent family report significantly lower GPAs. While the magnitude of GPA disadvantage is similar across most of these family structure pathways, it is greater for children who end up living in a cohabiting stepfather family following a divorce/separation. On the other hand, youth living in single-mother families following the death of their biological father do not report significantly lower GPAs than do those who have always lived with both biological parents. These findings illustrate the greater specificity of results obtained when the complete family structure pathways of youth are examined. Although research using cross-sectional measures of current family structure has shown that living in a cohabiting stepfamily or a single-mother family places children at higher risk for poor outcomes (see Appendix B, Model 1 for cross-sectional results), the results here indicate that the negative outcomes associated with these family structures are conditioned by the past family structure experiences of youth. Furthermore, the results indicate that the mechanisms that underlie GPA disadvantage may vary by specific family structure pathway (e.g., the factors that mediate the GPA disadvantage associated with living in a cohabiting stepfather family following a nonunion birth do not fully mediate the disadvantage associated with living in a cohabiting stepfather family following a divorce/separation).

School-Related Behavior Problems.  The amount of variance that is explained by family structure pathways is more modest for school-related behavior problems than for GPA. However, consistent with the previous analyses, the findings indicate that youth who have experienced any pathway involving divorce/separation of biological parents or a nonunion birth report significantly higher levels of school-related behavior problems than do youth who have continuously lived with two biological parents. Furthermore, living in a single-mother family following the death of a biological parent is not associated with significantly higher levels of school-related behavior problems (see Table 4).

Table 4.  Summary of Ordinary Least Square (OLS) Regression Analysis for Family Structure Pathways and Other Variables Predicting School-Related Behavior Problems (N = 13,627)
VariableModel 1Model 2Model 3
BSE BBSE BBSE B
  • *

    P < .05;

  • **

    P < .01;

  • ***

    P < .001.

2-bio parents—always lived together (reference category)
Married stepfather—bio parents divorced/separated.139***.031.137***.031.075**.028
Married stepfather—bio parents never married/cohabiting.134**.048.136**.047.075.041
Cohabiting stepfather—bio parents divorced/separated.241***.060.226***.056.110.057
Cohabiting stepfather—bio parents never married/cohabiting.264**.087.216*.088.149.090
Married stepmother—bio parents divorced/separated.212**.073.191**.074.139*.058
Single mother—bio parent widowed−.051.066−.042.070−.039.055
Single mother—bio parents divorced/separated.156***.029.167***.031.129***.027
Single mother—bio parents never married/cohabiting.136***.038.159***.034.109***.031
Single father—bio parents divorced/separated.230***.061.194**.061.079.058
Gender (male = 1, female = 0)  .194***.019 .098***.018
Years of age  .005.007−.004.006
Black/African American  −.111**.035−.111***.031
Hispanic/Latino  −.031.039−.065.036
Asian  .017.054.048.054
1st generation immigrant  −.289***.052−.171***.052
2nd generation youth  −.066.037−.044.033
Number of resident siblings  .003.008.007.007
Revised Peabody Picture Vocabulary Scores  −.002*.001 .003***.001
Public school  −.019.041−.057.033
Resident mother works  .018.021.020.020
Family income ≤$15,999  −.008.032−.084**.031
Family income $16,000–$34,999  −.006.029−.081**.027
Family income $35,000–$59,999  −.003.022−.040.022
Family income missing  .011.034−.029.030
Parents < high school education  .084*.037−.015.036
Parents have high school education  .004.019−.074***.018
Parents’ education missing  .139**.054.022.052
Lower than average college expectations     .343***.014
GPA    −.126***.021
R-square.012.039***.165***

Although several demographic and socioeconomic characteristics are significant predictors of school-related behavior problems (gender, age, race, immigrant status, test scores, parents’ education), controlling for background factors in Model 2 does not explain the significant differences between the various family structure pathways. Similar to the findings for GPA, the magnitude of the increase in school-related behavior problems does not differ significantly between youth living in a married stepfather family following a divorce/separation and those living in a married stepfather family following a nonunion birth. There is also no significant difference in the magnitude of increase in behavioral problems between youth living in cohabiting stepfather families following divorce/separation and those in cohabiting stepfather families following a nonunion birth. Moreover, additional regression analyses indicate no significant differences between the behavioral outcomes of youth whose family structure pathways include living in a married stepfather family formed following either a divorce/separation or a nonunion birth and those whose pathways include living in a single-mother family following the same kinds of family structure experiences. Finally, in contrast to the GPA findings, the analyses indicate little difference between the behavioral outcomes of youth in cohabiting stepfather families formed following divorce/separation and the behavioral outcomes of youth whose pathways include living in married stepfather or single-mother families. So, regardless of past family structure experiences, youth currently living in nonwidowed stepparent and single-parent families are more likely to report school-related behavior problems than are youth who have always lived with both biological parents. Furthermore, the magnitude of increase in behavioral problems is similar across the various family structure pathways.

Controlling for college expectations and GPA in the final model fully mediates the association between school behavior problems and the following four family structure pathways: living in a married stepfather family formed following a nonunion birth; a cohabiting stepfather family formed following a divorce/separation; a cohabiting stepfather family formed following a non-union birth; and a single-father family following a divorce/separation. Thus, significant differences between the behavioral outcomes of youth in these family structure pathways and youth in the two biological parent/always together pathway may largely be explained by the fact that youth in the former pathways tend to hold lowered college expectations and poorer grades. Controlling for expectations and grades does not fully explain, however, the heightened behavior problems of youth in family structure pathways that include living in a married stepfamily formed following divorce/separation (either a stepfather or stepmother) or living in a single-mother family formed following either divorce/separation or a nonunion birth.9

In sum, with the exception of youth in single-mother families formed following a parental death, youth in all of the family structure pathways examined here report significantly higher levels of school-related behavior problems than do youth who have always lived with both biological parents. All of these family structure pathways appear to be associated with a similar magnitude of increase in behavior problems. Yet, Model 3 shows that the factors that explain the increased behavioral problems of youth who have not lived continuously with both biological parents may differ by family structure pathway.

College Expectations.  Logistic regression analyses indicate that family structure pathways account for a small but significant amount of variance in adolescents’ college expectations (see Table 5). The findings in Model 1 indicate that youth in all of the family structure pathways that include living in either a stepparent or a single-parent family are significantly more likely to report lower than average college expectations than are youth who have always lived with both biological parents.

Table 5.  Summary of Logistic Regression Analysis for Family Structure Pathways and Other Variables Predicting Lower than Average College Expectations (N = 13,627)
VariableModel 1Model 2Model 3
BSE BeB (Odds ratio)BSE BeB (Odds ratio)BSE BeB (Odds ratio)
  • *

    P < .05;

  • **

    P < .01;

  • ***

    P < .001.

  • Notes: †indicates significant difference (P < .05) from the following family structure pathways: married stepfather family formed following divorce/separation and married stepfather family formed following a nonunion birth (see footnote 6);‡indicates significant difference (P < .05) from the following family structure pathway: cohabiting stepfather family formed following a nonunion birth (see footnote 6);#indicates significant difference (P < .05) from the following family structure pathways: single-mother family formed following divorce/separation, single-mother family formed following a nonunion birth, and single-mother family formed following a death (see footnote 7).

2-bio parents—always lived together (reference category)
Married stepfather—bio parents divorced/separated .265.0891.304** .240.0981.271* .116.1061.123
Married stepfather—bio parents never married/cohabiting .491.1251.635*** .240.1381.271 .120.1431.127
Cohabiting stepfather—bio parents divorced/separated1.124.2143.077*** .713.2332.039**, †, ‡, # .525.2311.691*
Cohabiting stepfather—bio parents never married/cohabiting .726.2612.067** .109.2851.115−.061.294 .941
Married stepmother—bio parents divorced/separated .456.2011.578* .369.2321.446 .305.2261.357
Single mother—bio parent widowed .537.2081.711* .162.1951.175 .200.2171.222
Single mother—bio parents divorced/separated .494.0871.638*** .173.0891.189* .098.0891.103
Single mother—bio parents never married/cohabiting .613.0871.847*** .104.1071.110−.016.113 .984
Single father—bio parents divorced/separated1.015.1732.761*** .657.1791.930*** .473.1791.605**
Gender (male = 1, female = 0)    .557.0521.745*** .391.0551.479***
Years of age   −.035.018 .966−.065.019 .937
Black/African American   −.293.094 .746**−.334.099 .716***
Hispanic/Latino    .193.1111.213 .134.1171.144
Asian   −.189.162 .828−.111.171 .895
1st generation immigrant   −.427.182 .652*−.186.201 .830
2nd generation youth   −.144.105 .866−.103.109 .902
Number of resident siblings    .022.0231.022 .030.0241.030
Revised Peabody Picture Vocabulary Scores   −.021.002 .980***−.011.003 .990***
Public school    .241.1821.272 .157.1801.170
Resident mother works   −.138.048 .871**−.163.052 .850***
Family income ≤$15,999    .900.1082.460*** .840.1082.316***
Family income $16,000–$34,999    .755.08342.127*** .675.0871.964***
Family income $35,000–$59,999    .425.0861.530*** .389.0871.475***
Family income missing    .479.0981.615*** .444.1031.559***
Parents < high school education    .776.1012.172*** .645.1051.907***
Parents have high school education    .676.0681.966*** .584.0721.793***
Parents’ education missing    .608.1331.837*** .417.1441.517**
GPA      −.867.047 .420***
Constant−.430 1.117 3.290 
Chi-square–9234.5–8433.7***–7922.8***

Controls for demographic and socioeconomic characteristics in Model 2 reduce to nonsignificance the difference in expectations between youth who have continuously lived with both biological parents and youth in several of the other family structure pathways, including those living in married stepfather, cohabiting stepfather, or single-mother families formed following a nonunion birth; married stepmother families formed following a divorce/separation; and single-mother families formed following the death of a parent. Lower than average socioeconomic status, in particular, appears to explain the greater likelihood of lower than average college expectations among youth with these family structure pathways. Yet, the college expectations of youth living in both married and cohabiting stepfather families formed following divorce/separation and in both single-mother and single-father families formed following a divorce/separation continue to be lower than their peers of similar background characteristics who have always lived with both biological parents.

The extent to which college expectations are reduced is significantly greater for youth in cohabiting stepfather families who have experienced divorce/separation than for youth in cohabiting stepfather families formed following a nonunion birth (significance indicated on Table 5). Among youth living in married stepfather and single-mother families, however, the extent to which college expectations are reduced is fairly similar for those whose pathways include a divorce/separation and those whose pathways include a nonunion birth. So for youth in the most common of stepfamily and single-parent forms (i.e., married stepfather and single-mother families), there is little distinction between the effects of pathways that include divorce/separation and those that include a nonunion birth. However, there is a salient distinction between these pathways for youth in the quickest growing of family types, the cohabiting stepfather family.

In terms of across-group comparisons, additional regression analyses suggest that there are no significant differences between the college expectations of youth whose family structure pathways include living in a married stepfather family following either a divorce/separation or a nonunion birth and those whose pathways include living in a single-mother family following the same kinds of family structure experiences. Youth living in cohabiting stepfather families formed following divorce/separation, however, do report significantly lower expectations than youth living in single-mother families that were formed following a divorce/separation and youth in all pathways that include living with a married stepfather (significance indicated on Table 5). So, as with GPA, youth who have experienced family structure pathways that include a combination of divorce/separation and a transition into a less common family form tend to be among the most disadvantaged in terms of college expectations.

The addition of self-reported grades to Model 3 mediates to nonsignificance the remaining differences between youth in pathways that include continuously living with two biological parents and youth in pathways that include living in a married stepfather or single-mother family following a divorce/separation. Even after all controls are added to the analyses, however, adolescents in pathways that include living in a cohabiting stepfather family formed after a divorce/separation remain over 60 percent more likely to hold lower than average college expectations than youth who have always lived with both biological parents.10

In sum, these findings indicate that college expectations are affected by the specific family structure pathways experienced by youth, even after measures of demographic and family background characteristics and self-reported grades are taken into account. On average, youth who have always lived with both biological parents are the least likely to have lower than average college expectations, while those in pathways that include a divorce/separation are among the most likely to report lowered expectations. In particular, living in a cohabiting stepfather family following a divorce/separation appears to be detrimental to the college expectations of youth. These youth are significantly more likely to report lowered expectations than are youth in other pathways, including youth who live in the same current family structure, but have had different past family structure experiences, and youth who live in a different current family structure, but have had similar past family structure experiences. Again, the findings illustrate the greater specificity of results obtained when the complete family structure pathways of youth are examined (see Appendix B, Model 3 for cross-sectional results), and suggest that the mechanisms that underlie educational differences may vary according to the specific kinds of past family structure experiences youth have had.

Discussion

This study documents the prevalence of the different stepfamily forms found in America today, examines the family structure pathways through which adolescents in stepfamilies have traveled to arrive at their current family form, and explores the effects of these pathways on adolescent academic outcomes. Developing a longitudinal picture of children's lives is a useful step in understanding how modern-day family structures affect the long-term academic and socioeconomic outcomes of youth. The results confirm the theoretical argument that it is important to distinguish between adolescents in stepfamilies according to their family structure pathways. Simply examining current family structure and basic measures of change or length of exposure to nonintact family forms is not a sufficient method for capturing the effect of the family contexts in which young people live and develop.

On average, adolescents who live in stepfamilies are at risk for poorer academic outcomes than are adolescents who live with both biological parents, regardless of the family structure pathways they have followed. Yet, not all stepfamily experiences are the same. For example, controlling for sociodemographic background factors, youth in married stepfather families report grades and college expectations that are higher than those of youth in cohabiting stepfather or single-father families, but similar to those of youth in single-mother families. Some particular family structure pathways also appear to place adolescents at a moderately greater risk for poor outcomes. Although small sample sizes preclude the inclusion of youth from previously widowed stepfamilies in the regression models, analyses indicate that youth in single-mother families who have experienced the death of a biological parent are not significantly disadvantaged in terms of their academic achievement, expectations (once SES is controlled), or behavior. Pathways that include divorce/separation and nonunion births, however, are associated with significantly poorer GPAs and college expectations and higher levels of school-related behavior problems among youth in all stepfamily and single-parent family types. For youth in married stepfamilies, I find little difference in the magnitude of the negative effects associated with pathways that include divorce/separation and pathways that include a nonunion birth. For youth in cohabiting stepfamilies, however, pathways that include a divorce/separation are associated with a significantly greater reduction in college expectations than are those including a nonunion birth. Across the academic outcomes, the disadvantages of divorce/separation are particularly pronounced for youth who have also transitioned into a cohabiting stepfamily or a single-father family.

These findings provide some support for arguments that family structure pathways that include more stressful combinations of family changes and adjustments tend to have a more powerful impact on adolescent well-being. All pathways of family structure change, with the exception of change brought about as the result of parental death, are associated with more negative academic outcomes during adolescence. Yet, youth in the least common of family forms may be more vulnerable to the stresses associated with past family disruption than other youth, especially when it comes to their expectations for the future. Thus, it is youth who have had this combination of experiences, rather than all youth living in stepfamilies, who may be at greatest risk for poorer outcomes during their adolescent years and well into their adult lives. Although family structure pathways are not the most crucial determinant of a young person's outcome, they do have significant and measurable consequences.

While these findings are fairly congruent with the hypotheses, the study does have some limitations. Social selection likely plays a role in the relationship between family structure and adolescent academic outcomes. There are preexisting negative parental and familial characteristics that may precipitate a family change or a particular family structure pathway and exert continuing negative effects upon adolescent academic outcomes, such as poor emotional or physical health, negative or difficult personality characteristics, substance use, and serious behavioral problems (Brown 2004; Capaldi and Patterson 1991; Cherlin et al. 1991; McLanahan and Sandefur 1994). These kinds of preexisting factors could not be examined due to data limitations. The lack of negative effects for youth living in previously widowed families, however, may support the contention that selection plays some role in the poorer outcomes of children who have experienced family disruption or have never lived with both biological parents. As a result, causal inferences should be drawn from the findings of this article with caution. This study does, however, address the problem of selection in a way that much previous research on family structure has not, by explicitly examining the kinds and combinations of past family structure experiences that youth have endured. Another limitation of the data is that the measure of college expectations had to be a dichotomized, indicating either average or lower than average expectations for college attendance. Thus, the analyses cannot explore the variability among respondents who were categorized together in the “lower than average” response category. Future research should seek to clarify the association between lower than average expectations and actual college attendance, as well as the substantive differences between respondents within this diverse response category.

Future research on family structure and adolescent outcomes should also take a longitudinal approach, continuing to account for the family structure pathways of young people. Given that the distinction between the different kinds of past family structure change appear to be more salient for youth in cohabiting stepfather families than in the more common stepfamily forms, this approach will become increasingly important as the number of children growing up within cohabiting unions continues to rise at a rapid rate. Future data collection should also focus on gathering more measures of preexisting parental and familial characteristics that may help to further reassure us that our findings are not due to social selection, and to more directly examine the specific stress mechanisms that may make some family structure experiences more difficult than others.

In conclusion, this research indicates that the way to improve academic outcomes among adolescents in stepparent and single-parent families is much more complicated than merely striving to ensure that all children grow up with two parents. In general, living with both biological parents in a stable family structure environment does seem to be most conducive to academic success. But, many children in America today do not have this opportunity. Often we have assumed that the “best” alternative to a two biological parent family is a two-parent stepfamily. Yet, the findings suggest that experiencing family structure pathways that include a combination of stressful family changes, such as divorce/separation and a later transition into a stepfamily (particularly if that stepfamily is based upon cohabitation rather than marriage), is generally not associated with better academic outcomes than is growing up in a single-mother family following a divorce/separation or a non-marital birth.

Public policies such as recent welfare reforms and government sponsored marriage initiatives that provide incentives for single parents to marry may need to be reexamined given these results. The effects of such policies on the well-being of youth will depend upon the stability of the resulting marriages and the kinds of family structure pathways that the young people involved experience. If government policies encourage the marriage of biological parents, and provide continuing support to help maintain these marriages, they may be useful tools for improving the lives of many children. The same may also be true of policies that encourage cohabiting families to transition to marriage (and to maintain those marriages). However, if these policies encourage the formation of unions that will ultimately fail or the creation of stepfamilies that include children who have already experienced a family breakup, they may have the unintended consequence of further jeopardizing child well-being. Before instituting public policy that affects the marital behavior of parents, policy makers should carefully consider whom they wish to target with their programs and what kinds of continuing support services they can offer newly married couples.

ENDNOTES

  • *

    Please direct all correspondence to Kathryn Harker Tillman, Department of Sociology, 526 Bellamy Building, Florida State University, Tallahassee, FL 32306-2270. Phone: 859-644-1669; Email: ktillman@fsu.edu. This article is based on data from the Add Health project, a project designed by J. Richard Udry, Peter S. Bearman, and Kathleen Mullan Harris, and funded by a grant P01 HD31921 from the National Institute of Child Health and Human Development, with cooperative funding from 17 other agencies. Acknowledgment is due Ronald R. Rindfuss and Barbara Entwisle for assistance in the original design of the Add Health. Persons interested in obtaining data files from Add Health should contact Add Health, Carolina Population Center, 123 W. Franklin Street, Chapel Hill, NC 27516-2524 <http://www.cpc.unc.edu/addhealth/contract.html>. This research was partially funded by support from the National Institute of Child Health and Human Development through grant 1 U01 HD37558-01 as part of the NICHD Family and Child Well-being Research Network. Special thanks is due Kathleen Mullan Harris, Karin L. Brewster, and Patricia Yancey Martin for their helpful comments and suggestions.

  • 1

    There is debate within the literature on whether or not an adult who cohabits with unrelated children should be classified as a “stepparent.” Yet, cohabiting living arrangements are becoming increasingly common for children in America today and recent research has shown that, similar to men in married stepfamilies, cohabiting men often make substantial financial and emotional investments in the children in their household and take on many “parent-like” roles and responsibilities (Stewart 2001). As such, the use of terms such as “cohabiting stepfather” and “cohabiting stepfamily” is becoming more widely accepted in the academic community (and will be used throughout this article).

  • 2

    The use of self-reported GPA, as opposed to data gathered from school records, is not ideal. Research has indicated that self-reported GPAs tend to be somewhat over-inflated, particularly by adolescents who have low actual GPAs. A recent meta-analysis, however, indicates that self-reported GPAs are more reliable when respondents are asked to report separately on each specific subject area and the time frame of recall is relatively short (Kuncel, Credé, and Thomas 2005). The Add Health collects GPA data in this manner.

  • 3

    Adolescents are classified as living in a “cohabiting stepfather” or “cohabiting stepmother” family if they live with one biological parent who is cohabiting with an unrelated adult. Unfortunately, the data limit the ability to determine the marital status of biological parents who are living together. Thus, adolescents are classified as living in a “two biological parent family” if they live with both biological parents, regardless of the marital status of those parents.

  • 4

    Because the data limit the ability to determine the marital status of biological parents who are living together, the family structure pathways do not capture transitions between cohabitation and marriage among biological parents. For example, if a child's biological parents were cohabiting when the child was born and they later married, the child would simply be classified as living with two biological parents who had lived together since the child's birth (the same as if the child had been born to a married couple). Similarly, if a child's biological parents were cohabiting when the child was born and they later separated, the child would be classified as living with a single mother (or father) following a divorce/separation (again, the same as if the child had been born to a married couple).

  • 5

    Multivariate models including an “other family structure pathway” measure, composed of the nine pathways with small frequencies, were also analyzed. The results of these analyses did not differ significantly from the results of the more parsimonious models presented in the text.

  • 6

    To determine whether the magnitude of the coefficients differed significantly, I utilized the “test” command in STATA, which provides hypothesis testing after estimation. This statistical testing was conducted during the analysis of all three of the outcome variables. Significant findings are indicated in Tables 3–5.

  • 7

    Additional regression analyses, in which the reference category for the family structure pathways variable was altered, were conducted to determine whether there were significant differences in the outcomes of youth in the various family structure pathways. This type of additional modeling was conducted during the analysis of all three of the outcome variables. Significant findings are indicated in Tables 3–5.

  • 8

    Separate analyses show that the number of prior family structure transitions, the length of prior exposure to stepparent and/or single-parent family forms, and the length of exposure to current family structure are all significantly associated with GPA. To determine whether the number of prior transitions and length of exposure can explain the significant pathway differences, I estimated models that included measures of these factors in addition to the various family structure pathways. I also estimated models that included interactions between family structure pathways and each of the transition and exposure measures. The addition of these measures (and their interactions) does not mediate the significant relationships between family structure pathways and GPA, nor does it significantly improve the fit of the analytical model.

  • 9

    While the number of prior family structure transitions and length of family structure exposure experienced are significantly associated with school-related behavior problems, no further insights or improved model fit were obtained when these measures are added to the multivariate models.

  • 10

    While the number of prior family structure transitions and length of family structure exposure experienced are significantly associated with college expectations, no further insights or improved model fit were obtained when these measures are added to the multivariate models.

Table Appendix A.  Weighted Means of Past Family Structure Experience Variables, by Current Family Structure
 2-bio parents N = 8189All stepfamilies N = 2358All single parents N = 3434Full sample N = 13,988StepfamiliesSingle-parent families
Married stepfather N = 1631Married stepmother N = 357Cohabiting stepfather N = 335Cohabiting stepmotherN = 35Single mother N = 3091Single father N = 343
Past family structure experiences 
Number of past family structure transitions .001.631.24.581.551.881.771.791.201.51
Proportion of life lived in a step- or single-parent family .00 .71 .67.28 .74 .66 .68 .59 .68 .58
Proportion of life lived in current family1.00 .39 .47.77 .45 .28 .26 .12 .49 .31
Table Appendix B.  Summary of Regression Analyses for Current Family Structure and Other Variables Predicting GPA, School-Related Behavior Problems, and Lower than Average College Expectations (N = 13,627)
VariableGPASchool-related behavior problemsLower than average college expectations
BSE BBSE BeB (Odds ratio)SE B
  • *

    P < .05;

  • **

    P < .01;

  • ***

    P < .001.

2-bio parents (reference category)
Married stepfather−.104***.025.075**.0261.118.096
Cohabiting stepfather−.208***.063.142**.0531.688*.365
Married stepmother−.039.048.143**.0571.349.304
Single mother−.057**.018.118**.0231.066.072
Single father.195***.056.087.0581.605**.287
Gender (male = 1, female = 0)−.139***.018.100***.0181.486***.082
Years of age−.023***.006−.007.006 .935***.018
Black/African American−.089**.028−.112***.031 .707***.067
Hispanic/Latino−.089*.038−.058.0371.140.132
Asian.072.046.052.055 .895.155
1st generation immigrant.162**.053−.169***.052 .822.170
2nd generation youth.019.033−.040.034 .909.100
Number of resident siblings.011.007.006.0071.033.024
Revised Peabody Picture Vocabulary Scores.010***.001.003***.001 .989***.003
Public school−.072.044−.061.0321.172.214
Resident mother works−.012.015.021.020 .848***.044
Family income ≤$15,999−.103***.030−.078**.0312.321***.257
Family income $16,000–$34,999−.111***.028−.070**.0281.976***.173
Family income $35,000–$59,999−.060**.022−.032.0231.489***.130
Family income missing−.054.028−.026.0301.570***.159
Parents < high school education−.144***.037−.030.0371.878***.199
Parents have high school education−.124***.018−.078***.0171.802***.131
Parents’ education missing−.199***.046−.003.0571.557**.223
Lower than average college expectations−.372***.025    
GPA  −.346***.014 .421***.020
School-related behavior−.313***.014−.141***.027  
R-square.309    

Ancillary