reports, family structure, cohabiting stepfamily, mother reports, cohabiting, report, stepfamilies, Journal of Marriage and the Family, mother, the adolescent, school problems, Journal of Marriage and Family, American Journal of Sociology, stepfamily, adolescent, Demography, boundary changes, University of North Carolina at Chapel Hill, Princeton University, single parent families, Bumpass, Population Studies, Cherlin, Center for Family and Demographic Research Working Paper, Population Research, Social Forces, References Bearman, biological siblings, biological children, family form, complex family, Harvard University Press, National Longitudinal Study of Adolescent Health, parents and children, Carolina Population Center
Phone: (419) 372-7279[email protected]Bowling Green State University Working Paper
Family Structure Measurement: Reconciling Adolescents and Mother Reports of Cohabitating Stepfamilies Susan L. Brown Wendy D. Manning Department of Sociology & Center for Family and Demographic Research Bowling Green State University Bowling Green, Ohio 43403
1 FAMILY STRUCTURE MEASUREMENT: RECONCILING ADOLESCENT AND MOTHER REPORTS OF COHABITING STEPFAMILIES* Susan L. Brown and Wendy D. Manning Department of Sociology and Center for Family and Demographic Research Bowling Green State University Bowling Green, OH 43403 Ph: 419-372-9521 Fax: 419-372-8306 [email protected]
*The research for this paper is supported by a grant to the first author from the National Institute of Child Health and human development
(K01-HD42478-02) and by the Center for Family and Demographic Research, Bowling Green State University, which has core funding from the National Institute for Child Health and Human Development (R21- HD042831-01). This research uses data from the Add Health project, a program project designed by J. Richard Udry (PI) and Peter Bearman, and funded by grant P01-HD31921 from the National Institute of Child Health and Human Development to the Carolina Population Center, University of North Carolina at Chapel Hill, with cooperative funding from 17 other agencies. Persons interested in obtaining data files from The National Longitudinal Study of Adolescent Health should contact Add Health, Carolina Population Center, 123 West Franklin Street, Chapel Hill, NC 27516-2524 (http//www.cpc.unc.edu/addhealth).
2 The authors thank I-Fen Lin for helpful comments on an earlier version of this manuscript.
3 FAMILY STRUCTURE MEASUREMENT: RECONCILING ADOLESCENT AND MOTHER REPORTS OF COHABITING STEPFAMILIES Abstract We used data from the first wave of the National Longitudinal Study of Adolescent Health to examine family boundary ambiguity in adolescent and mother reports of residing in cohabiting stepfamilies. Among the 532 mothers who reported living with a cohabiting partner, only one-third of their teenage children also reported residing in a cohabiting stepfamily. Moreover, the relationship between family structure and adolescent well-being differed depending on whether we used adolescent or mother reports of family structure. This dramatic discrepancy in reporting presents an important measurement problem for family scholars. Key words: adolescence, cohabitation, family structure, measurement, well-being
4 FAMILY STRUCTURE MEASUREMENT: RECONCILING ADOLESCENT AND MOTHER REPORTS OF COHABITING STEPFAMILIES Cohabitation is now a common experience among American adults and children. A majority of persons in their twenties and thirties have cohabited, and the modal path of entry into marriage is cohabitation (Bumpass & Lu, 2000; Bumpass & Sweet, 1989; Bumpass, Sweet, & Cherlin, 1991). Cohabitation is a family status that includes children; almost half of cohabiting unions have children present. Bumpass and Lu (2000) estimate that 40 percent of children will spend some time in a cohabiting family before age 16. These figures demonstrate the importance of obtaining reliable and valid measures of cohabitation in our research on family structure and living arrangements. Few scholars have critiqued the measures of cohabitation used in social science
research (for an exception, see Manning & Smock, 2003). This omission may owe in part to the fact that measures of cohabitation on national surveys are relatively new, appearing for the first time in a national sample in the 1982 National Survey of Family Growth (NSFG). In this paper, we argue that cohabitation poses significant measurement challenges and contributes to the ambiguity of family boundaries. Using data from the first wave of the National Longitudinal Study of Adolescent Health (Add Health), we compare adolescent and mother reports of current cohabitation and document the extent to which discrepancies exist. We also identify predictors of discrepant reporting. We then investigate whether and how these discrepancies are related to adolescent well-being. We conclude by outlining approaches to enhance our measurement of children's family structure. Rationale The complexity of children's living arrangements is increasingly apparent. As a declining share of
5 children resides with two biological married parents, a growing share lives in an array of other arrangements, including married stepfamilies, cohabiting stepfamilies, single-parent families, and without biological parents (Bumpass & Lu, 2000; Seltzer, 2000; Sun, 2003). Cohabitation is an important part of many children's family experiences; two-fifths of children are expected to spend some time living with cohabiting parents (Bumpass & Lu, 2000). Consequently, ignoring cohabitation misrepresents children's living arrangements. About 20 percent of single-mother families include a cohabiting partner and nearly one-third of single-father families also contain a cohabiting partner (London, 1998; Manning & Smock, 1997). Most stepfamilies are formed through marriage, but some are maintained informally through cohabitation. Among adolescents living in stepparent families, one-third are living with cohabiting parents and two-thirds are living with married parents (Manning & Lamb, 2003). Thus, the evidence indicates that research on children's living arrangements must include cohabitation as a family type. Cohabitation may be a particularly important family structure to study because of the high rates of instability. Children experience considerable instability in their family lives, and cohabiting families are arguably more unstable than any other (Graefe & Lichter, 1999; Manning, Majumdar, & Smock, 2004). A recent study by Raley and Wildsmith (2004) indicates that once we account for spells of parental cohabitation, levels of family instability increase by 30 percent among White children and more than double among Black children. Moreover, children living in a cohabiting family experience more family instability than those residing with a non-cohabiting single mother (Raley & Wildsmith). Recent data collection
efforts have accounted for this growth in cohabitation as measures of family structure now typically include cohabitation as a family form. Nonetheless, our ability to reliably
6 and validly measure cohabitation as a living arrangement for children is less clear. Our goal in this paper is to reconcile adolescent and parent reports of cohabitation to assess the effectiveness of current measures of family structure. Measuring Cohabitation Early estimates of cohabitation were derived through indirect measurement strategies. The Census Bureau
measured cohabitation by determining the number of POSSLQ households, that is, partners of opposite sex sharing living quarters. POSSLQs are defined as those households containing only two persons of the opposite sex who are unrelated and at least age 15. This definition excludes cohabitors with resident children and those living in complex households. It also mistakenly identifies persons living as roommates for cohabiting partners. Casper and Cohen (2000) introduced an adjusted POSSLQ measure that captures many of those cohabitors with children and yields more generous estimates of the cohabiting population from the 1970s to the 1990s than the original measure. A comparison of the adjusted POSSLQ measure with direct measures of cohabitation available in data sets such as the 1987-88 National Survey of Families and Households (NSFH) and 1995 NSFG reveals that these direct measures produce even larger estimates of the cohabiting population. Direct questions about cohabitation (referred to as unmarried partners in household rosters) were first included in the 1990 decennial Census and the 1995 Current Population Survey (CPS). Direct measures of cohabitation from the 1995-1997 CPS surveys actually yield lower estimates than the adjusted or unadjusted POSSLQ or other surveys (e.g., NSFH and NSFG). This pattern of findings led Casper and Cohen (2000) to caution researchers to be sensitive to the ways in which cohabitation is conceptualized and measured, particularly when making comparisons across surveys.
7 Today, most national data collections include direct measures of cohabitation. Some surveys include questions about current and prior times the respondent has lived together with someone of the opposite sex. Another strategy is to ask respondents to report on their relationships to other household members by completing a household roster. The most common method to identify cohabitors is to include relationship types on these rosters such as "partner" or "unmarried partner." New research has focused on measurement issues and cohabitation, suggesting that current measurement strategies may be less than ideal. Based on in-depth interviews with 115 cohabitors, Manning and Smock (2003) conclude that many cohabitors do not understand the term "unmarried partner" and would not use the term to describe their cohabiting relationship. This may lead to underestimation of cohabitation. In her analysis of data from new parents, Knab (2004) uncovers considerable complexity in the measurement of cohabitation. Her study reveals that reports of cohabitation may vary according to the number of nights the couple spends together. Moreover, mothers and fathers with newborns do not always consistently report whether they were cohabiting when the child was born and their reports of cohabitation sometimes change during follow-up interviews (Teitler, Reichman, & Koball, 2004). theoretical framework
The measurement challenges posed by emerging family forms are not new. There is extensive research on the ambiguities surrounding married stepfamilies, which Cherlin (1978) characterized as "incomplete institutions" because the norms and expectations involved in this family type are not clearly defined. Stepfamilies require individual members to create kinship ties and establish among themselves the contours of their responsibilities and obligations to one another. Doing the work of kinship is difficult
8 for many stepfamilies and contributes to their instability (Cherlin & Furstenberg, 1994). The ambiguity surrounding stepfamily members
' roles is evidenced in Furstenberg's (1987) study showing that many individuals do not report stepfamily members when asked to list the people in their family. For example, 15% of parents did not report stepchildren who resided in the household (versus only 1% of parents who neglected to mention biological children). And, whereas about 7% of children failed to mention a biological mother or father, 31% of children did not include a residential stepparent in their family list. Children were also more likely to omit residential stepsiblings than biological siblings (41% compared to 19%). Similarly, White (1998) found that children's reports of siblings are unreliable, particularly when step- and half-siblings are involved. Using data from the two waves of the NSFH, she calculated that about 16% of respondents overreported and another 15% underreported their siblings. These discrepancies are largely attributable to the classification difficulties posed by complex family forms, including stepfamilies that involve the presence of step- and halfsiblings. The incomplete institutionalization of new family forms is linked to the measurement challenges involved with complex family structures (White, 1998). Without shared understandings of the norms and roles involved in these "nontraditional" families, family boundary ambiguity leads to inconsistencies in reports of who is in and who is out of the family (Ganong & Coleman, 1994; Stewart, forthcoming). Therefore, it is not surprising that there are discrepancies in reports of membership in complex family structures. Stated differently, individuals define their families and consequently the reliability of our measures may be compromised. The more complex the family form, the greater the family boundary ambiguity (Boss, 1980; Stewart, forthcoming). White (1998, p. 732) argued that "family structure has
9 a larger subjective component than we have accorded it...incongruity is not error." Family structure reports, particularly for complex families, are likely to depend in part on who is doing the reporting. Discrepancies may occur between siblings, partners, or the parent and child. In their study of adolescents following parental divorce, Bunchanan, Maccoby, and Dornbusch (1996) encountered discrepancies in terms of the presence of new partners, the remarriage status (i.e., cohabiting versus married) of a parent, and the duration of the new relationship. Cohabiting stepfamilies are arguably even less institutionalized than married stepfamilies, which are formed through a tie that is legally binding. Although increasingly common, cohabiting stepfamilies are predicated on informal ties between two adults and their partner's children. The growing propensity to substitute cohabitation for remarriage suggests that remarriage is "becoming less obligatory and socially regulated. It follows that informal unions are generally less stable and secure arrangements" (Cherlin & Furstenberg, 1994, p.362). Indeed, Stewart (forthcoming) found that family boundary ambiguity, operationalized as a discrepancy in stepparents' reports of their (and their partner's) children, was greater among cohabiting stepfamilies than married families (29 versus 11%). The Present Study Family life today is diverse and complex. The instability and incomplete institutionalization of cohabiting families likely contributes to family boundary ambiguity, which in turn calls into question the efficacy of our family structure measurement strategies. A decade ago, family scholars were concerned with the boundary ambiguity created by married stepfamilies (Cherlin & Furstenberg, 1994). Now, we extend this line of inquiry by focusing on cohabiting stepfamilies. The National Longitudinal Study of Adolescent Health (Add Health) data provide a unique
10 opportunity to compare family structure reports of parents and adolescents. No other national data include family structure reports from both children and parents. Still, most prior studies using the Add Health rely on the child's perspective of family structure (e.g. Bearman & Brььckner, 2001; Demuth & Brown, 2004; Harris, Duncan, & Boisjoly, 2002; Meier, 2003; Videon, 2002). We begin by documenting whether and how mothers and adolescents (dis)agree about being in a cohabiting stepfamily. We examine all pairs in which either (or both) the adolescent or the mother reports the current family structure as a cohabiting stepfamily. Then, we describe the extent to which mothers and adolescents report differing family structure. In particular, the discrepancy centers around either the nature of the mother's relationship to the partner (i.e., married versus cohabiting stepfamily) or the presence of the partner (i.e., single-mother family versus cohabiting stepfamily). Next, we examine factors associated with discrepant family structure reports to determine which adolescent-mother pairs are more (or less) likely to provide conflicting information about their current living arrangements. Prior work on family boundary ambiguity suggests that several demographic characteristics
(e.g., older age of child, nonwhite, and less education) as well as union duration and prior marital experience may heighten the odds that the adolescent and mother do not report the same family structure (Madden-Derdich, Leonard, & Christopher, 1999; Stewart, forthcoming). Finally, we investigate the importance of discrepant family structure reports for adolescent wellbeing. Discrepancies between mother and adolescent reports may represent part of the reason why prior work on the association between parental cohabitation and adolescent well-being has not yielded consistent findings (cf. Manning, 2002). We estimate a series of models using (1) the mother's family
11 structure report, (2) the adolescent's family structure report, and (3) a combined mother and adolescent family structure report. This approach allows us to evaluate how various family structure measurement strategies are related to adolescent well-being to determine whether the source of information (i.e., mother or adolescent) or discrepancy between sources is linked to adolescent outcomes. Method Data We use the first wave of the National Longitudinal Study of Adolescent Health (Add Health) collected in 1995. The Add Health includes both an in-home parent interview and an in-home adolescent interview. The respondents were students in grades 7 through 12 from a sample of 80 high schools and 52 middle school
s in the UNITED STATES
. The analytic sample for our paper is based on adolescents who have a biological or adoptive mother who responds to the parent interview (N=10,488). The Add Health are appropriate for our analyses for several reasons. The primary advantage of the Add Health is that the data include questions about family structure directed to both the adolescent and the parent. Other national data sources rely on the parent's report of family structure (e.g., CPS, NLSY, NSAF, NSFG, NSFH, PSID, and SIPP). For a summary of the question wording and reporting source for cohabitation in various national surveys, including the Add Health, see the Appendix. Other benefits of using the Add Health include the large sample that ensures a sufficient number of parents who are cohabiting and questions that tap several dimensions of adolescent wellbeing. Family Structure
12 Adolescents are asked to fill out a household roster that includes mother's or father's cohabiting partners as a family category. For these analyses respondents who report they are living with their biological or adoptive mother and their "mother's partner" are coded as living in cohabiting stepparent families. The family categories include: two biological parents, married stepparent, cohabiting stepparent, and single mother family. The parent interview includes several questions that are used to establish family structure. Mothers are coded as living in a cohabiting stepfamily if they report they are currently living in a "marriage-like" relationship and they are not living with the biological father of their child. The question about the type of relationship was prefaced with a series of questions that started with "The next questions are about your marriages and marriage-like relationships." Respondents reported on the number of relationships and then were asked "Thinking about your present or most recent such relationship. During what years were you married or living with this person." The parent then replies whether or not they were married or living with someone in each year and whether the relationship was a "marriage or marriage-like relationship." Finally the parent is asked "Is this relationship still going on?" We categorize mothers into the same four family categories as adolescents: two biological parent, married stepparent, cohabiting stepparent, and single-mother family. Adolescent Outcomes We focus on two adolescent outcomes: delinquency and school problems. The former taps externalizing behaviors and the latter is an indicator of academic adjustment. Both of these measures are reported by the adolescent. Delinquency is a behavioral outcome that summarizes the adolescent's participation in the following activities during the past year: deliberately damage property; shoplift;
13 joyride; steal something worth more than $50; go into a house or building to steal something; use or threaten to use a weapon to get something; steal something worth less than $50; engage in a gang fight; behave in a loud, unruly manner in public. Respondents' reports of engagement in these behaviors, ranging from (0) never to (3) five or more times in the past year, were summed to create a count variable of the frequency with which the respondent engages in various delinquent behaviors. Values range from 0 to 27, with higher values indicating more frequent participation in delinquent activities. School problems measures the respondent's difficulty in the school context. The four items comprising the scale indicate the degree to which, since the start of the school year, the adolescent respondent has had problems getting along with teachers, paying attention in school, getting homework done, and getting along with other students. Responses for each item range from (0) never to (4) every day. Responses were summed to create the problems scale, which ranges from 0 to 16, with higher values indicating more school problems. This measure has a Cronbach alpha reliability of 0.85. Other Covariates We include control variables
in our models that measure factors that are related to family boundary ambiguity or adolescent well-being. We use three measures of the child's demographic characteristics: age (coded in years), gender (male is coded one, female zero), and race-ethnicity (White (reference), African-American, Latino, and Other). Mothers report on the family's socioeconomic status
, including maternal education, marital history, and parental income. Education is coded into four categories: less than high school, high school (reference), some college, and college graduate. Marital history is a dichotomous variable indicating whether the mother has ever been married. We log the value of Family income
. Missing cases are imputed to the mean and a dummy variable flags the imputation. We
14 include two measures of the adolescent's appraisal of the mother's parenting. Mother-adolescent relationship quality is comprised of the following four items: how close you feel to your mother, your mother is warm and loving toward you, you are satisfied with the way your mother and you communicate with each other, you are satisfied with your relationship with your mother. Values for each item range from 1 to 5, with higher values indicating better relationship quality. The Cronbach's alpha for the scale is 0.85. Maternal supervision is a count variable that sums the frequency with which the mother is home when the adolescent leaves for school, returns from school, and goes to bed. Responses range from (1) always to (5) never and are reverse-coded such that higher values indicate more supervision. We also include indicators of the length and quality of the mother's cohabiting relationship. Relationship duration measures the number of years the mother and her partner have lived together. Relationship happiness gauges the mother's report of her happiness with her current relationship ranging from (1) completely unhappy to (10) completely happy. Analytic Strategy First, we document the extent of discrepancy in family structure reports. We tabulate the percentage of mothers whose reports agree with their adolescent's report of family structure as well as the percentage of adolescents whose reports agree with their mother's report of family structure, focusing on discrepancies in the cohabiting stepfamily category (N=10,488). We then examine the types of discrepancies between adolescents and mothers for all pairs in which either the mother or the adolescent reported living in a cohabiting stepfamily (N=617). Second, we investigate predictors of discrepant family structure reports. We begin with the assumption that mothers are more accurate reporters of their intimate relationship
s, particularly since
15 532 mothers report cohabiting whereas just 190 adolescents report living in a cohabiting stepfamily. Among the pairs in which mothers report living in a cohabiting stepfamily, multinomial logistic regression
is used to establish the odds that the adolescent reports a different family structure than the mother. We contrast the odds that the adolescent reports living in a married stepfamily versus a cohabiting stepfamily. We then compare the odds that the adolescent claims to live in a single-mother family rather than a cohabiting stepfamily. Third, we estimate multivariate models to evaluate the linkages between various measures of family structure and our two indicators of adolescent well-being. We use negative binomial regression to model delinquency because it is an over-dispersed count variable (i.e., its standard deviation is greater than its mean) and ordinary least squares regression for our analyses of school problems. We estimate two sets of models. The first set uses adolescent reports of family structure and the second uses mother reports. We estimate a zero-order or bivariate model that includes only the basic family structure variables (i.e., two biological, married stepparent, cohabiting stepparent, and single mother). The second model we present adds the control variables. We then recode family structure to account for parent and teen discrepant reports of cohabiting stepfamilies. The family categories include two biological parent, married stepparent, single mother, mother and adolescent agree cohabiting stepparent (reference), mother cohabiting stepparent-adolescent single mother, and mother cohabiting stepparentadolescent married stepparent. The reference category indicates concordance between adolescent and mother reports of living in a cohabiting stepfamily. To ensure that the data are nationally representative of adolescents in the United States design effects must be taken into account (Bearman, Jones & Udry, 1997). All of our analyses are conducted using STATA survey estimation procedures
16 to obtain correct standard errors (Chantala & Tabor, 1999). Results Adolescent and Mother Reports of Family Structure Overall, there is a high level of congruence between mother and adolescent reports of family structure. We find that 86% of mothers and adolescents report living in the same family structure (result not shown). Table 1 shows that the match between adolescent and mother reports of family structure varies considerably by family type. The first column is based on the adolescent report of family structure. Typically, adolescents who report living with two biological parents also have mothers who report this same family structure (93%). A slightly lower percentage (82%) of adolescents who claim to live with single mothers have mothers who also state they are a single mother. Similarly, 86% of adolescents who report living with married stepparents have mothers who also report living in married stepfamilies. The greatest level of incongruence occurs among adolescents who report living in cohabiting stepfamilies. We find that just two-thirds of teenagers who state they are living in a cohabiting stepfamily have mothers who also report living in a cohabiting stepfamily. Among teens who report living in a cohabiting stepfamily, one-fifth of their mothers report being single mothers and another 13% claim to be in a married stepfamily (results not shown). [TABLE 1 ABOUT HERE] The second column of Table 1 focuses on the mother's report of family structure. We find nearly perfect congruence (99%) between mother and adolescent reports of living with two biological parents. We also find that when mothers report being single mothers, 89% of adolescents' reports agree. In contrast, only 69% of mothers who state they live in married stepfamilies have an adolescent
17 who also reports living in a married stepfamily. In fact, one-quarter of mothers who report living in a married stepfamily have a teen who reports living with two biological parents (results not shown). The family category with the highest level of incongruence is cohabiting stepfamily. Only one-third of mothers who report living in this family type have an adolescent who also claims to be living with a cohabiting stepparent. Most often the mismatch occurs because teens report they live with single mothers (45%) and close to one-fifth state they live with married stepparents (results not shown). This pattern of findings is consistent with family boundary ambiguity as the more complex the family form, the greater the discrepancy in reporting. Table 2 focuses on the type of mismatch in reports that occurs among those adolescent-mother pairs in which either (or both) the adolescent or the mother reports living in a cohabiting stepfamily (N=617). There is a very high level of discord in reports about cohabiting stepfamilies. We find that when either the mother or adolescent reports living in a cohabiting stepfamily there is agreement for only 31% of adolescents and mothers. The most common type of disagreement (38%) is when a mother reports living in a cohabiting stepfamily and the adolescent reports residing in a single-mother family. In this situation the teenager does not appear to recognize their mother's cohabiting partner. Another type of disagreement exists for 18% of mothers and adolescents in which the mother reports living with a cohabiting partner and the adolescent states they are living with a married mother and stepfather. This may occur because the teen is embarrassed to report that the mother is cohabiting and not married or alternatively the mother has told the child she is married when she is not. It is less common for the adolescent to claim they are living with a cohabiting stepparent and the mother reports living alone or being married (8% and 6%, respectively).
18 [TABLE 2 ABOUT HERE] Predicting Discrepancies Between Adolescent and Mother Reports Next we examine the characteristics of adolescents and mothers who agree versus disagree about living in a cohabiting stepfamily. Our analyses are limited to the 532 mothers who report living a cohabiting stepfamily. We focus on this subsample because we believe that mothers may be better able to provide information about their union status. Table 3 presents the multinomial logistic regression results. The reference category in these models is agreement about living in a cohabiting stepparent family. The first contrast category is the adolescent reports the mother is married and the second contrast category is the adolescent reports the mother is single. [TABLE 3 ABOUT HERE] The results in Table 3 indicate that teens' reports of their mother's union status vary according to race-ethnicity, income, mother's marital status, mother-adolescent relationship quality, and quality of the mother's cohabiting relationship. The first column shows the effects of the coefficients predicting whether the teen reports the mother is married rather than cohabiting. Both Hispanic and Black teens are more likely than Whites to report their mother is married than cohabiting. Teenagers from higher income families have higher odds of reporting their mother is married than cohabiting. The mother's report of the quality of her cohabiting relationship is related to the adolescent's classification of the family. Teenagers who have mothers who are happier have greater odds of reporting their mother is married versus cohabiting. The second column shows the estimates of the odds that teenagers report their mother is single versus cohabiting. Black teens are more likely to report their mother is single than cohabiting.
19 Respondents with mothers who have ever been married have lower odds of reporting their mother is single than cohabiting. Finally, teenagers who have better relationships with their mothers are more likely to report their mother is single than cohabiting. Family Structure Reports and Adolescent Well-Being Our final task is to examine whether and how the discrepancy between adolescent and mother reports of family structure is related to adolescent well-being. We present each measure of well-being (delinquency and school problems) separately. The first two columns of Table 4 focus on adolescent reports of family structure. When we rely on the adolescent's report of family structure, we find that living in a cohabiting stepfamily is associated with higher levels of delinquency than living in a two biological parent, married step, or single-mother family (Model 1a). In the multivariate model, the effect of living in a single-mother family versus a cohabiting stepfamily is reduced to nonsignificance (Model 1b). [TABLE 4 ABOUT HERE] The next two columns in Table 4 show the relationships between family structure and delinquency when we rely on the mother's reports of living arrangements. Here, teens living in cohabiting stepfamilies have higher levels of delinquency, on average, than teens living in two biological, married stepfamilies, or single-mother families (Model 2a). The second model includes the control variables and shows that teens living in married stepfamilies and two biological parent families have lower levels of delinquency than their counterparts living in cohabiting stepfamilies, but the difference between single-mother families and cohabiting stepfamilies is no longer statistically significant (Model 2b). In other words, whether we use the adolescent's or mother's report of family structure, the
20 pattern of findings is the same. We do note that the magnitude of the coefficients is consistently larger using the adolescent versus mother reports of family structure, which is perhaps not surprising when we consider that delinquency is reported by the adolescent. The final two columns present the effects of discrepancies or agreement in mother and adolescent reports of living in cohabiting stepfamilies. We break apart the reference category of cohabiting stepfamily used in Models 2a and 2b to distinguish among adolescents who report living in a married stepfamily, a single-mother family, and a cohabiting stepfamily. The reference category is agreement between the mother and adolescent who both report living in a cohabiting stepfamily. The coefficients indicate that adolescents who agree their mother lives in a cohabiting stepfamily have similar levels of delinquency as teens who disagree. These results hold in both the bivariate and multivariate models (Models 3a and 3b). Thus, whether the adolescent agrees or disagrees with the mother about being in a cohabiting stepfamily is not consequential for delinquent behavior. The next series of models focus on school problems, as shown in Table 5. When we rely on the adolescent's report of family structure, we find that living in a cohabiting stepfamily is associated with higher levels of school problems than living in a two biological parent family, married stepfamily, or single mother family (Model 1a). The multivariate model (Model 1b) yields similar findings, except the difference in delinquency for teens in cohabiting stepfamilies versus single-mother families is only marginally significant (p=.08). [TABLE 5 ABOUT HERE] When we rely on mother reports of family structure, the relationship between family structure and school problems changes. We still find that living in a cohabiting stepfamily is associated with more
21 school problems than living with two biological parents. But, adolescents living in cohabiting stepfamilies now exhibit similar levels of school problems as teens living in single-mother families or married stepfamilies (Model 2a). These associations persist in the multivariate models (Model 2b). Thus, the relationship between family structure and school problems depends on whether we rely on adolescent or mother reports of family structure. Moreover, the magnitude of the coefficients is considerably larger for adolescent than mother reports of family structure. The final two models show how the discrepancy or agreement between adolescent and mother reports of family structure are related to school problems. The results indicate that a specific type of mismatch between mother and teen reports of living in a cohabiting stepfamily is associated with fewer school problems. Teens who claim their mother is single when the mother reports cohabiting have fewer school problems than teens who agree with their mother about living in a cohabiting stepfamily (Model 3a). This finding is consistent with that shown in Table 3 that teens who feel closer to their mother are more likely to report living in a single-mother family when their mother reports living in a cohabiting stepfamily. Indeed, in the multivariate model (Model 3b) that includes a control for motheradolescent relationship quality, this association between family structure discrepancy (i.e., adolescent reports single-mother family and mother reports cohabiting stepfamily) is only marginally significant (p=.11). Teens who report their mother is married rather than cohabiting have similar levels of school problems as teens who agree with their mother about living with a cohabiting stepparent. Discussion We used data from the wave one adolescent and parent in-home questionnaires of the Add Health to
22 examine the extent to which adolescents and mothers provide discrepant reports of family structure. Guided by family boundary ambiguity theory, which states that the greater the family complexity, the more likely is inconsistency in reporting who is in and out of the family, we anticipated that the greatest discrepancy in reporting would occur among those living in cohabiting stepfamilies. Indeed, whereas two-thirds of adolescents agreed with their mothers' reports of living in a married stepfamily, just onethird of adolescents whose mothers said they live in a cohabiting stepfamily reported the same family type. Nearly 90 percent of adolescents concurred with mothers who reported being single and over 99 percent agreed with mothers who reported being part of a two biological parent family. Consistent with prior work on family boundary ambiguity in married stepfamilies, our analyses identified several of the same sociodemographic predictors of discrepancies in adolescent and mother reports of residing in a cohabiting stepfamily. Adolescents who are Black or Hispanic, have higher family incomes, and mothers who are happier in their intimate relationship are more likely to report living in a married stepfamily than agree with their mother who reports living in a cohabiting stepfamily. Adolescents who are Black, feel very close to their mothers, and whose mothers have been married previously are more likely to report living in a single-mother family. The mother's report of the duration of the cohabiting stepfamily is not related to discrepancy in reporting, which reinforces Furstenberg's (1987) finding of no effect of duration on ambiguity in married stepfamilies. Certainly, our documentation of considerable ambiguity in the reporting of cohabiting stepfamilies between adolescents and their mothers is important for how we measure family structure. As cohabitation continues to increase in popularity and more children are exposed to this family type, the questionable validity of relying on a single reporter of family structure becomes more consequential.
23 A school-based survey, the Add Health has a higher response rate
for adolescents than parents. Yet it seems that parents provide more accurate reports of their living arrangements. Using adolescent reports of family structure yields only half as many cohabiting families as tallied from mother reports in the Add Health. This discrepancy will affect our estimates of children in cohabiting families. From a demographic perspective, misclassification of children's living arrangements will yield an inaccurate picture of the distribution of children across various family structures. In addition to the demographic consequences of boundary ambiguity in cohabiting stepfamilies, our work reveals its significance for research on the association between family structure and adolescent well-being. Our examination of two domains of well-beingexternalizing behavior and academic performance
shows that whose report of family structure we use matters. While we obtained a similar pattern of findings regardless of whether we used the adolescent or mother reports of family structure for delinquency, the same is not true for school problems. From the models using adolescent family structure reports, we concluded that being in a cohabiting stepfamily is related to more school problems than being in a two biological parent family, a married stepfamily, or a singlemother family. In contrast, from the models using mother family structure reports, school problems among adolescents in cohabiting stepfamilies are only higher than those in two biological parent families. Adolescents who report living in a single-mother family but have mothers who say they are cohabiting actually exhibit fewer school problems than adolescents who agree with their mother's report of living in a cohabiting stepfamily. Adolescents who have a good relationship with their mother are particularly likely to report being in a single-mother family, which may explain why this group is better adjusted academically. More generally, the size of the family structure coefficients was consistently larger for
24 adolescent versus mother reports for both delinquency and school problems. Essentially, the association between family structure and adolescent outcomes attenuates when we use mother reports. We are not able to gauge the extent to which the high level of discrepancy in adolescent and mother reports of a cohabiting stepfamily is due to measurement error. The Add Health ascertains adolescent reports of family structure through a household roster, whereas mother reports are achieved through a detailed series of questions about living with someone in a marriage or marriage-like relationship. We might have achieved greater consistency in reporting had adolescents and mothers been asked the same series of questions about their current living arrangements. Still, the pattern we obtained is consistent with that which we expected according to family boundary ambiguity theory. The more complex the family form, the greater the discrepancy in adolescent and mother reports. Particularly since cohabitation is an incomplete institution that is predicated on informal ties, greater attention should be paid to how we measure cohabitation in national surveys (Casper & Cohen, 2002; Manning and Smock 2003). Inconsistencies in question wording (see Appendix) may account for some of the reporting discrepancies we documented in this paper. Our study demonstrates that family boundary ambiguity is not uncommon, especially in complex family forms, namely, cohabiting stepfamilies. As family complexity intensifies, the validity of our measures of family structure may be undermined. We provide evidence that adolescents and mothers are more likely to disagree than agree about living in a cohabiting stepfamily. Whose report of family structure we use in our analyses of adolescent well-being affects our conclusions about how adolescents fare in various family forms. Lynn White (1998, p. 732) was prescient when she suggested researchers refer to "`perceived family structure'" since respondents actively construct their
25 families. Future data collection efforts should obtain family structure information from multiple sources, including PARENTS AND CHILDREN
, to permit additional research on family boundary ambiguity as discrepancies yield meaningful insights not only about the social construction of family membership, but also its influence on individual well-being.
26 References Bearman, P. S., Jones, J., & Udry, J. R. (1997). The National Longitudinal Study of Adolescent Health: research design
. http://www.cpc.unc.edu/projects/addhealth/resdesign/index/htm Bearman, P. S., & Brььckner, H. (2001). Promising the future: Abstinence pledges and the transition to first intercourse. American Journal of Sociology, 106, 859-912. Boss, P. (1980). Normative family stress: Family boundary changes across the life span. Family Relations, 29, 445-450. Buchanan, C. M., Maccoby, E. E., & Dornbusch, S. M. (1996). Adolescents after divorce. Cambridge, MA: Harvard University
Press. Bumpass, L., & Lu, H. (2000). Trends in cohabitation and implications for children's family contexts in the United States. Population Studies, 54, 29-41. Bumpass, L., & Sweet, J. (1989). National estimates of cohabitation: Cohort levels and union stability. Demography, 26, 615-625. Bumpass, L. L., Sweet, J. A., & Cherlin, A. J. (1991). The role of cohabitation in declining rates of marriage. Journal of marriage and the family
, 53, 913-927. Casper, L. M., & Cohen, P. N. (2000). How does POSSLQ measure up? Historical estimates of cohabitation. Demography, 37, 237-245. Chantala, K., & Tabor, J. (1999). Strategies to perform a design-based analysis using the Add Health data. Carolina Population Center, University of North Carolina at Chapel Hill, Chapel Hill, NC. Cherlin, A. J. (1978). Remarriage as an incomplete institution. American Journal of Sociology, 84,
27 634-650. Cherlin, A. J., & Furstenberg, F. F. (1994). Stepfamilies in the United States: A reconsideration. Annual Review of Sociology, 20, 359-381. Demuth, S., & Brown, S. L. (2004). Family structure, family processes, and adolescent delinquency: The significance of parental absence versus parental gender. Journal of Research in Crime and Delinquency, 41, 58-81. Furstenberg, F. F. (1987). The new extended family. In K. Pasley & M. Ihinger-Tallman, Remarriage and Stepparenting: Current Research and Theory (pp. 42-64). New York: Guilford Press. Harris, K. M., Duncan, G. J., & Boisjoly, J. (2002). Evaluating the role of 'nothing to lose' attitudes on risky behavior in adolescence. Social Forces, 80, 1005-1039. Ganong, L. H., & Coleman, M. (1994). Remarried family relationships. Thousand Oaks, CA: Sage. Graefe, D. R., & Lichter, D. T. (1999). Life course transitions of American children: Parental cohabitation, marriage, and single motherhood. Demography, 36, 205-217. Knab, J. (2004). Who's in for how much: The impact of definitional changes on the prevalence and outcomes of cohabitation. Center for Research on Child Well-Being Working Paper 200405. Princeton University
. London, R. A. (1998). Trends in single mothers' living arrangements from 1970 to 1995: Correcting the Current Population Survey. Demography, 35, 125-131. Madden-Derdich, D. A., Leonard, S. A., & Christopher, F. S. (1999). Boundary ambiguity and coparental conflict after divorce: An empirical test of the family systems model of the divorce process. Journal of Marriage and the Family, 61, 588-598.
28 Manning, W. D. (2002). The implications of cohabitation for children's well-being. In A. Booth & A. C. Crouter (Eds.), Just living together: Implications of cohabitation for children, families, and social policy
(pp. 121-152). Mahwah, NJ: Lawrence Erlbaum Associates
. Manning, W. D., & Smock, P. J. (1997). Children's living arrangements in unmarried mother families. Journal of Family Issues, 18, 526-544. Manning, W. D., & Smock, P. J. (2003). Measuring and modeling cohabitation: New perspectives from qualitative data
. Center for Family and Demographic Research Working Paper, 2003- 10. Bowling Green State University. Manning, W. D., Smock, P. J., & Majumdar, D. (2004). The relative stability of cohabiting and marital unions for children. Population Research and Policy Review, 23, 135-159. Manning, W. D., & Lamb, K. A. (2003). Adolescent well-being in cohabiting, married, and single parent families. Journal of Marriage and Family, 65, 876-893. Meier, A. (2003). Adolescents' transition to intercourse, religiosity and attitudes about sex. Social Forces, 81, 1031-1052. Raley, R. K., & Wildsmith, E. (2004). Cohabitation and children's family instability. Journal of Marriage and Family, 66, 210-219. Seltzer, J. A. (2000). Families formed outside of marriage. Journal of Marriage and the Family, 62, 1247-1268. Stewart, S. D. (in press). Boundary ambiguity in stepfamilies. Journal of Family Issues. Sun, Y. (2003). The well-being of adolescents in households with no biological parents. Journal of Marriage and Family, 65, 894-909. Teitler, J., Reichman, N., & Koball, H. (2004). Bias in retrospective reports of cohabitation among new parents. Center for Research on Child Well-Being Working Paper, 2004-07. Princeton
29 University. Videon, T. M. (2002). The effects of parent-adolescent relationships and parental separation on adolescent well-being. Journal of Marriage and Family, 64, 489-503. White, L. K. (1998). Who's counting? Quasi-facts and stepfamilies in reports of number of siblings. Journal of Marriage and the Family, 60, 725-733.
SL Brown, WD Manning