Article summary

Divorce, Family Conflict, and Adolescents' Well-Being Author(s): David Mechanic and Stephen Hansell Source: Journal of Health and Social Behavior, Vol. 30, No. 1 (Mar., 1989), pp. 105-116 Published by: American Sociological Association Stable URL: http://www.jstor.org/stable/2136916 Accessed: 20-03-2018 14:23 UTC

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide

range of content in a trusted digital archive. We use information technology and tools to increase productivity and

facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at

http://about.jstor.org/terms

American Sociological Association is collaborating with JSTOR to digitize, preserve and extend access to Journal of Health and Social Behavior

This content downloaded from 130.68.183.9 on Tue, 20 Mar 2018 14:23:49 UTC All use subject to http://about.jstor.org/terms

Divorce, Family Conflict, and Adolescents' Well-Being*

DAVID MECHANIC STEPHEN HANSELL

Rutgers University

Journal of Health and Social Behavior 1989, Vol. 30 (March):105-116

The relative effects of family conflict and divorce on the well-being of adolescents were examined in a longitudinal study of a heterogeneous community sample. Higher levels of family conflict were associated with increases in adolescents' depressed mood, anxiety, and physical symptoms over time. In contrast, neither recent divorce nor earlier divorce was associated with longitudinal changes in any health outcomes. Also, adolescents living in intact families with high conflict had significantly poorer well-being than those living in families of divorce with low conflict. Finally, the longitudinal effects of divorce and family conflict did not differ by age and sex, but Hispanic adolescents experienced more negative effects of family conflict than non-Hispanic whites, and the well-being of Asian adolescents was influenced more strongly by recent divorce.

The divorce of parents is a major stressful life event for children and adolescents, often

having negative effects on health and well-

being (Emery 1982; Guidubaldi and Perry

1985; Wallerstein and Kelly 1980). Increas- ing evidence suggests that the process of

divorce often involves conflict and family

disharmony which are more distressing and

damaging to children than the separation itself (Dancy and Handal 1984; Emery 1982; Enos and Handal 1986; Hetherington, Cox, and

Cox 1981; Peterson and Zill 1986; Shaw and Emery 1987; Slater and Haber 1984). Chil- dren often manifest an "acute distress syn-

drome" (Rutter 1979) immediately after

divorce, but the effects are usually time- limited (Hetherington 1979), whereas the effects of family conflict are often more

enduring (Hetherington, Cox, and Cox 1978).

In addition, many parents who are embroiled in marital discord never divorce, and yet

* Direct correspondence to: David Mechanic, Institute for Health, Health Care Policy, and Aging Research, Rutgers University, 30 College Avenue, New Brunswick, NJ 08903. This research was supported in part by grants from the NIMH (No. 39590) and the William T. Grant Foundation. Requests for reprints should be addressed to David Mechanic, Institute for Health, Health Care Policy, and Aging Research, Rutgers University, 30 College Ave., New Brunswick, NJ 08903.

children in such families often have serious problems (Hetherington et al. 1978). Finally, divorces that end painful family conflict actually may improve children's well-being (Emery 1982; McCord, McCord, and Thurbur 1962).

With a few notable exceptions (Fursten- berg, Peterson, Nord, and Zill 1983; Guidu- baldi and Perry 1985; Peterson and Zill 1986), most of the research on the relative effects of divorce and family conflict has been based on relatively small and homogeneous samples, often drawn from clinical popula- tions (Emery 1982). Existing community- based studies have not examined the relative effects of these predictors on longitudinal changes in adolescents' well-being. In addi- tion, few studies have used more than a single measure of well-being, although Emery (1982) notes that the effects of divorce and family conflict may vary according to the outcomes that are assessed.

Data that were collected as part of another study allowed us to explore the longitudinal effects of divorce and family conflict in a large, heterogeneous community sample of adolescents. Measures of four conceptually and empirically distinct dimensions of well- being were available: depressed mood, anxi- ety, self-esteem, and physical symptoms. These secondary data were not designed to

105

This content downloaded from 130.68.183.9 on Tue, 20 Mar 2018 14:23:49 UTC All use subject to http://about.jstor.org/terms

106 JOURNAL OF HEALTH AND SOCIAL BEHAVIOR

answer the specific questions addressed, and thus have shortcomings that will be described; nevertheless, this study provides an opportu- nity to begin to examine the longitudinal effects of divorce and family conflict on multiple dimensions of well-being among adolescents in the community.

Divorce is a complex process, and may involve a constellation of stressful life events that tax the child's coping capacities (Guidu- baldi and Perry 1985; Wallerstein 1985a). Examples of such stressors include residential and school changes, family economic difficul- ties, alienating custody disputes, and separa- tion from friends. Although the influences of stressors associated with divorce probably diminish over time as families reconstitute themselves, long-term effects persist for some children (Hetherington, Cox, and Cox 1985; Rickel and Langner 1985). Thus it remains unclear to what degree separation from a parent through divorce adversely affects the child's health and well-being independently of the stressful life events that often are associated with it.

Both divorce and family conflict probably erode family members' capacity to nurture children's well-being by diverting time and attention from the children and by undermin- ing the children's perceptions of parental interest (Hess and Camara 1979; Hethering- ton, Cox, and Cox 1979; Rutter 1971). In addition, parents engaged in marital discord may be more likely to neglect their children, to punish them as a result of their own frustration and distress, or to manipulate them to gain advantages in the conflict with their spouse. Consequently the negative effects of family conflict may be caused by decreases in the quality of family relationships.

The evidence about sex differences in the impact of divorce is mixed. Some studies suggest that divorce has a more negative impact on boys than on girls (Emery 1982; Emery and O'Leary 1982; Porter and O'Leary 1980; Wallerstein and Kelly 1980). Other studies, however, found no sex differences (Slater, Stewart, and Linn 1983), and non- clinical studies generally found fewer sex differences than studies based on clinical samples (Emery 1982). Still others suggest that the long-term effects of divorce may be worse for girls than for boys (Hetherington et al. 1985; Wallerstein 1985b).

Thus the existing literature suggests five hypotheses about the relative effects of

divorce and family conflict, which we examine in this secondary analysis. First, the deleterious effects of family conflict are expected to be more damaging than the effects of divorce or separation from a parent. Second, children of divorce who go on to live in a conflict-free setting are expected to experience less distress than those in intact families characterized by conflict. Third, we expect the negative effects of divorce to be mediated through the stressful life events that often accompany divorce. Fourth, we expect the negative effects of divorce and family conflict to be mediated through a reduction in the quality of relationships between parents and adolescents, as reflected in adolescents' perceptions of lower parental interest and greater punishment. Finally, recent divorce will affect the well-being of adolescent boys more than that of girls over a one-year period, while earlier divorce will influence girls' health more than boys'.

In addition, little evidence exists as to whether divorce and family conflict have stronger effects on younger or on older children, or on children with varying racial and ethnic backgrounds (Emery 1982). We did not form specific hypotheses about such effects, but the heterogeneous sample in this study provided an opportunity to explore them. Finally, divorce is only one of many causes of separation from a parent that may loosen crucial attachment bonds and may degrade adolescents' well-being. In our anal- yses we attempted to distinguish the effects of divorce from those of separations from parents or stepparents caused by parental death.

SAMPLE AND PROCEDURES

Longitudinal data were obtained from the first two waves of a three-wave, three-year study of adolescents' health and well-being. We recruited adolescents from the seventh, ninth, and eleventh grades of 19 public schools in five communities in New Jersey, which serve the inner city as well as middle-class suburbs. Involvement in the study required the cooperation of each school system, the personnel of each individual school, and the written consent of both students and their parents.

All adolescents who were present in health or gym classes in the fall of 1984 were asked

This content downloaded from 130.68.183.9 on Tue, 20 Mar 2018 14:23:49 UTC All use subject to http://about.jstor.org/terms

ADOLESCENT'S WELL-BEING 107

to volunteer; approximately 80 percent (3,215) agreed to participate. Permission forms were mailed at least twice to their parents with a letter describing the study. We received written permissions from 48 percent (1,528) of the parents; 81 percent (1,235) of these adolescents subsequently completed usable questionnaires in school under the supervision of our field staff. In the second wave, 97 percent (1,193) of these adolescents were resurveyed with the same measures used in Wave One. The analyses reported here are based on the 1,067 cases that provided complete data on all measures used in the analysis, approximately 33 percent of the adolescents present in the health or gym classes from which respondents were re- cruited originally.

In addition we interviewed one parent, usually the mother, of 88 percent (842) of a random sample of 957 of the adolescents who were studied in the first wave. These

interviews, conducted by telephone, lasted an average of 45 minutes; we asked the parents about their own health as well as about the health of their adolescent.

This sample contains a variety of selection biases, such as the omission of students who were not in school and of those who declined to participate. Most important, however, only about half of the parents returned signed consents; phone consent and passive consent were not acceptable to our human subjects reviewers. Comparisons with census data showed that the distributions of sex, race, and ethnicity in our sample were very similar to those in the population. Parents of our adolescents, however, had significantly higher levels of education than adults of similar age in the population. Thus the sample is broad and heterogeneous, but cannot be considered representative.

MEASURES

A major independent variable was the adolescents' report of the degree of family conflict in Wave One. A single item asked whether or not there was "a lot of quarreling or fighting where you live in the past year." Although this item did not refer specifically to parental conflict, and may have included quarreling among siblings or between parents and adolescents, we assumed that all quarrel- ing or fighting contributed to a noxious family

environment. Thirty-eight percent of the adolescents reported frequent quarreling or fighting at home.

A limitation of this measure is that when adolescents who lived with one parent reported family conflict, we did not know whether they were referring to continuing marital conflict between the parent at home and the parent who had left, or to conflict among other people who currently lived at home. Thus our measure of family conflict may have had somewhat different meaning for adolescents in single-parent households than for those in families with two parents. We cannot resolve this ambiguity with the data available in this secondary analysis, but we present some information that supports the convergent validity of our measure of family conflict. We also examined whether family conflict had different effects for adolescents who did and who did not live with two parents or stepparents.

We created two dummy variables for parental marital status in Wave One. One dummy variable was coded 1 for the seven percent of the adolescents whose parents or stepparents had been divorced or separated recently (in the past year), and 0 for others. The second dummy variable was coded 1 for the 19 percent of the adolescents who reported an earlier divorce or marital separa- tion (more than a year ago), and 0 for others. The remaining 74 percent of the adolescents lived in intact families.'

We derived a measure of current separation from parents based on who lived with the ad- olescent. Yet because no distinction was made between natural parents and stepparents in the available date, this measure did not capture all separations from natural parents. In particular, it did not reflect current separation from a nat- ural parent for the 12 percent of the adoles- cents who had experienced parental divorce or death and who currently lived with one natural parent and one stepparent or with two steppar- ents. The distribution of this measure showed that 81 percent of the adolescents lived with two parents or stepparents, 15 percent lived with a mother or stepmother only, two percent lived with a father or stepfather only, and two percent reported living with no parent or step- parent. We combined the last three categories to contrast current separation from at least one parent or stepparent with no separation.

Adolescents are separated from parents and

This content downloaded from 130.68.183.9 on Tue, 20 Mar 2018 14:23:49 UTC All use subject to http://about.jstor.org/terms

108 JOURNAL OF HEALTH AND SOCIAL BEHAVIOR

stepparents for reasons other than divorce. The data did not include specific information about the reasons for separations, so we could not disentangle the relationships among di- vorce and separation from parents unambigu- ously. We asked adolescents, however, whether a parent or stepparent had ever died; thus for the four percent of the sample who answered affirmatively, we could assess the effects of one specific type of separation from parents other than divorce.

Other independent variables assessed in Wave One were sex, grade level, race and ethnicity, parental education, and an eight- item scale of the stressful life events reported by adolescents in the past year other than divorce, death of a parent, separation from a parent, and family conflict. The low internal consistency of this scale (alpha = .50) reflected the relative independence of these life events. In the second wave, adolescents also assessed the undesirability of each life event. Items for this scale and for others that were created for this study are listed in the appendix.

Adolescents' perceptions of parental inter- est in Wave One were measured with nine-item scales of paternal interest (alpha = .91) and maternal interest (alpha = .88). These two scales were summed into a scale of overall perceived parental interest; ratings for single parents were doubled to obtain the same metric as the sum of scores from both parents. Two five-item scales assessed adoles- cents' perceptions of paternal punishment (alpha = .70) and maternal punishment (alpha = .71). These two scales also were summed into a scale of overall perceived parental punishment; ratings for single parents were doubled. We asked adolescents to rate parental interest and punishment only for parents or stepparents with whom they currently lived. Thus a limitation of these measures is that we had no information about adolescents' perceptions of the interest or punishment by parents or stepparents from whom they were separated currently.

We assessed four dependent measures of well-being in both waves of data collection. The Center for Epidemiological Studies Depression scale (CESD) (Radloff 1977) consisted of 20 items that measured relatively common symptoms of depressed mood (alpha = .90). Although the CESD was developed for adults, Schoenbach, Kaplan, Wagner, Grimson, and Miller (1983) found that it had

good internal consistency in a study of adolescents.

In addition, we created a four-item scale that measured anxiety (alpha = .79), and we used Rosenberg's (1965, 1979) scale of 10 items to assess self-esteem (alpha = .82). We also included a 12-item measure of common physical symptoms (Mechanic 1979, 1980) (alpha = .66). The moderate internal consis- tency of this scale reflected the fact that symptoms such as headaches and stomach- aches were reported frequently, but that others, such as constipation and vomiting, were reported relatively rarely.

Parents were not asked to rate family conflict, but we asked them whether they were very happy, somewhat happy, or not very happy with their marriages; responses to this question helped to illuminate the meaning of adolescents' reports of family conflict. Parents also reported how often their child exhibited behavioral problems on a four-item scale that included uncontrollable temper tantrums or outbursts, irritability, doing the opposite of what was asked of him or her, and disagreeing or quarreling with people for no apparent reason (alpha = .73). Parents also reported how much they disagreed with their adolescent about seven types of conventional behavior such as coming home on time, doing homework, and doing chores and jobs around the house (alpha = .79). Finally, a 19-item scale asked parents to rate their child's psychological distress in terms of sadness, loneliness, nervousness, having enough en- ergy, and sleep disturbances (alpha = .83).

RESULTS

Sample Characteristics

The distribution of adolescents by grade and sex in the final sample was not significantly different from the distributions in the sample of adolescents who volunteered or in the sample for whom we received parental permission. In addition, the percent- ages of adolescents who volunteered, and of parents who gave their permission, were not associated significantly with adolescents' well- being aggregated by school district. We speculate, however, that selection biases made it more likely that adolescents in our sample were more affluent, healthier, and happier than those in the sample of volunteers

This content downloaded from 130.68.183.9 on Tue, 20 Mar 2018 14:23:49 UTC All use subject to http://about.jstor.org/terms

ADOLESCENT'S WELL-BEING 109

or the total school population. Because adolescents from the poorest and most disrupted families were less likely to be surveyed, selection biases may have attenu- ated somewhat the effects of divorce and family conflict.

The sample included students with hetero- geneous racial and ethnic backgrounds and so- ciocultural characteristics. Of the 1,067 ado- lescents, 54 percent were female and 46 percent were male; 35 percent were in the seventh grade in Wave One, 37 percent were in the ninth grade, and 28 percent were in the elev- enth grade. Eight percent of the adolescents were black, four percent were Hispanic, four percent were Asian, 81 percent were non- Hispanic whites, and three percent were mem- bers of other races or ethnicities. Finally, five percent of the parents had less than a high school education, 23 percent had graduated from high school, 15 percent had some col- lege, 24 percent had graduated from college, and 33 percent had postgraduate training.

Correlates of Divorce and Family Conflict

Table 1 displays the joint distribution of the divorce of parents or stepparents, current separation from a parent or stepparent, and

TABLE 1. Mean Family Conflict by Divorce, Current Separation from Parents, and Parental Death

Adolescent Currently Separated from One or Two Parents or Stepparents

Parent or Stepparent Ever Died

No Yes

Parent or Stepparent No .30 .22

Ever Divorced (3%, 27) (2%,18) Yes .41 .29

(14%, 147) (1%, 7)

Adolescent Not Currently Separated from Two Parents or Stepparents

Parent or Stepparent Ever Died

No Yes

Parent or Stepparent No .37 .63 Ever Divorced (70%, 743) (1%, 8)

Yes .45 .50

(10%, 111) (1%, 6)

NOTE: N = 1,067. For each cell, the percentage of the total sample and the number of cases are given in parentheses.

the death of a parent or stepparent in relation to family conflict. Fully 70 percent of the adolescents reported no current separation

from parents and no parental divorce or death. The second most common pattern involved 14

percent of the adolescents, who were sepa- rated currently from at least one parent and

who reported parental divorce but no parental

death. Finally, 10 percent of the adolescents reported parental divorce but no current separation from parents or parental death. Other patterns were much less frequent.2

Mean levels of family conflict also are displayed in Table 1 for each combination of the divorce of parents or stepparents, current separation from parents or stepparents, and the death of a parent or stepparent. A three-way analysis of variance showed that adolescents whose parents had divorced reported significantly higher levels of family conflict than did those from intact families (F = 5.64, df = 1, 1057, p < .02). Parental death was not associated significantly with family conflict despite high levels of reported conflict; the absence of statistical significance may reflect the small number of adolescents who reported the death of a parent or stepparent. Finally, current separation from parents was not associated significantly with family conflict, and there were no significant interaction effects.

Thirty-six percent of the 795 adolescents from intact families reported family conflict, as did 40 percent of the 200 adolescents whose parents or stepparents had divorced earlier and 51 percent of the 72 adolescents whose parents or stepparents had divorced recently. These data suggested that although the process of divorce was associated signifi- cantly with higher levels of family conflict, the absolute differences in perceptions of family conflict between adolescents from intact families and from families of divorce were relatively small.

The Wave One interviews with parents provided some evidence about the meaning of adolescents' reports of family conflict. Par- ents' reports of greater satisfaction with their marriages were associated significantly with adolescents' reports of less family conflict (r = - .17) and with higher levels of perceived parental interest (r = .18). In contrast, parents' reports of disagreement with their adolescents were not associated significantly with adolescents' reports of family conflict, but were associated significantly with less

This content downloaded from 130.68.183.9 on Tue, 20 Mar 2018 14:23:49 UTC All use subject to http://about.jstor.org/terms

110 JOURNAL OF HEALTH AND SOCIAL BEHAVIOR

perceived parental interest (r = - .17) and with greater perceived parental punishment (r = .25). These correlations suggested that adolescents' ratings of family conflict re- flected parental marital conflict more strongly than conflict between adolescents and their parents.3

Divorce, Family Conflict, and Adolescents' Well-Being

The well-being of this sample was fairly stable across waves, as shown by the autocorrelations for depressed mood (r = .59), anxiety (r = .53), self-esteem (r = .64), and physical symptoms (r = .57). Adolescents' reports of family conflict, how- ever, were somewhat less stable over time (r

.38). The longitudinal influences of family

conflict, divorce, current separation from parents, and parental death on adolescents' well-being were tested in ordinary least squares (OLS) regressions that also included the adolescents' sex, grade level, race and ethnicity, and parental education as predic- tors. For each dependent variable measured in Wave Two, we included all Wave One predictors and the same dependent variable as we measured in Wave One. Thus the results of these regressions, which are shown in Table 2, indicated changes in well-being over

time that were associated with each predictor.4 The results supported the first hypothesis-

that family conflict would have more direct effects on longitudinal changes in well-being than would divorce, current separation from parents, or parental death. Specifically, higher levels of family conflict were associated significantly with longitudinal increases in depressed mood, anxiety, and physical symp- toms. In contrast, divorce, current separation from parents, and parental death were not associated significantly with changes in any health measure over time.

Other results showed that in comparison with males, females had significant longitudi- nal increases in depressed mood, anxiety, and physical symptoms, and significant decreases in self-esteem. Also, older adolescents evi- denced significant increases in anxiety over time, while blacks and Asians showed significant decreases in anxiety over time.5

Because of the ambiguities in our measure of family conflict, it is possible that this measure had different effects for adolescents who reported a current separation from at least one parent or stepparent than for those reporting no separation. In analyses that are not shown, we tested the interaction effect involving current separation and family con- flict for each dependent measure of well- being, but none were significant. Although it was not necessary statistically, we ran separate regressions for each group. In

TABLE 2. Regressions of Wave Two Adolescents' Well-Being on Wave One Predictors

Wave Two Wave Two Wave Two Wave Two Wave One Predictors CESD Anxiety Self-Esteem Physical Symptoms

Wave One Well-Being .54*** .49*** .63*** .55 Family Conflict .07** .07* -.02 .07 Recent Divorce .05 .04 -.04 .03 Earlier Divorce .01 .00 .02 .01 Current Separation from Parents .01 .02 .00 -.01 Parent Ever Died .02 .00 .00 .00 Sex - .08** - . 10*** .07** -.07 Grade .04 .12*** .02 .02 Black .03 - .06* .01 - .02 Hispanic .03 - .01 .03 - .05 Asian .01 -.05* -.01 -.02 Other Race -.01 .01 .00 .02 Parental Education -.03 .01 .03 .01 Adjusted R2 .36 .32 .42 .33

NOTE: N = 1,067. Family conflict, divorce, and parental death are coded 1 for yes, 0 for no; current separation from one or more parents or stepparents is coded 1, no current separation is coded 0; sex is coded 1 for male, 0 for female; dummy variables for race and ethnicity are coded 1 for the relevant group, 0 for others, with non-Hispanic whites as the reference category.

* Standardized beta is significant, p<.05. ** Standardized beta is significant, p<.01.

* Standardized beta is significant, p<.001.

This content downloaded from 130.68.183.9 on Tue, 20 Mar 2018 14:23:49 UTC All use subject to http://about.jstor.org/terms

ADOLESCENT'S WELL-BEING 111

analyses that are not presented here, the significant direct effects for each group were almost identical to those shown in Table 2 for the whole sample. The only difference was that family conflict did not affect depressed mood significantly for adolescents currently separated from at least one parent or steppar- ent. These data suggested that our measure of family conflict had similar meaning for adolescents who experienced varying degrees of current separation from parents or stepparents.6

Other data allowed us to examine the perceived undesirability of divorce and family conflict. In the second wave we asked adolescents to rate the undesirability of the life events they reported. In support of the first hypothesis, the results showed that family conflict was significantly more unpleas- ant than divorce (X2 = 34.8, df = 4, p < .001). Of the 398 adolescents who reported frequent quarreling or fighting at home in the past year and who rated its undesirability, fully 69 percent rated it very bad or somewhat bad, 25 percent rated it neither good nor bad, and only 6 percent rated it somewhat good or very good. In contrast, of the 88 adolescents who reported that their parents divorced or separated in the past year and who rated its undesirability, 42 percent rated it very bad or somewhat bad, 33 percent rated it neither good nor bad, and fully 25 percent rated it somewhat good or very good.

The means in Table 3 supported the second hypothesis. T-tests showed that for every measure of well-being in both waves, adoles- cents in intact families characterized by high family conflict had significantly poorer well- being than adolescents experiencing family divorce who reported low family conflict.

In addition, the 37 adolescents who rated divorce as bad or very bad in the second wave had significantly higher levels of depressed mood, anxiety, and physical symptoms and significantly lower self-esteem than the 617 adolescents in intact families with no family conflict. The well-being of these 37, how- ever, was not significantly different from that of the 348 adolescents in intact families with family conflict. Thus divorce perceived as undesirable had approximately the same cross-sectional association with well-being as did family conflict in intact families.

The findings for the 51 adolescents who rated divorce as very good, good, or neither good nor bad in Wave Two were less clear. These respondents did have significantly lower anxiety and higher self-esteem than the 348 adolescents in intact families with family conflict, but they also had significantly higher depressed mood than the 617 adolescents in intact families without conflict.

Mediating Effects of Stressful Life Events and Relationships with Parents

We examined the third hypothesis-that stressful life events would mediate the longitudinal effects of divorce-by adding our measure of life events to the second step of a hierarchical regression for each depen- dent variable. The results, which are not shown, did not support our hypothesis. Stressful life events did not have any significant direct or mediating effects on well-being over time.

Similarly, we examined the fourth hypoth- esis-that perceptions of the quality of relationships with parents would mediate the

TABLE 3. Mean Adolescents' Well-Being by Divorce and Family Conflict

No Parent or Stepparent Parent or Stepparent Ever Divorced Divorced

No Family Conflict Family Conflict No Family Conflict Family Conflict Well-Being Measure (N = 507) (N = 288) (N = 155) (N = 117)

Wave One CESD 12.93 17.50 14.65 18.52 Wave Two CESD 12.00 15.64 13.50 17.76 Wave One Anxiety 7.51 8.84 7.74 8.85 Wave Two Anxiety 7.26 8.23 7.35 8.63 Wave One Self-Esteem 30.80 29.07 30.36 28.94 Wave Two Self-Esteem 31.52 30.33 31.39 30.02 Wave One Physical Symptoms 21.08 22.01 20.77 22.79 Wave Two Physical Symptoms 20.42 21.47 20.19 22.25

NOTE: For every measure of well-being in both waves, t-tests showed that adolescents in intact families with high conflict had significantly poorer well-being than adolescents in families of divorce with low conflict.

This content downloaded from 130.68.183.9 on Tue, 20 Mar 2018 14:23:49 UTC All use subject to http://about.jstor.org/terms

112 JOURNAL OF HEALTH AND SOCIAL BEHAVIOR

longitudinal effects of family conflict. In analyses that are not shown, we added perceived parental interest and punishment to the second step of a hierarchical regression for each dependent variable. Perceived paren- tal interest was associated with longitudinal increases in self-esteem, but contrary to our fourth hypothesis, perceived relationships with parents did not mediate any of the longitudinal effects of divorce, family con- flict, or the other predictors.

We also examined regression models in which overall perceived parental interest and punishment were disaggregated into separate scales of perceived maternal and paternal interest and punishment. The results, which are not shown, showed that higher levels of perceived maternal punishment were associ- ated significantly with longitudinal increases in depressed mood, and that higher levels of perceived maternal interest were associated significantly with increases in self-esteem over time. Contrary to our hypotheses, however, adolescents' perceptions of relation- ships with mothers and fathers did not mediate any of the direct longitudinal effects of family conflict, divorce, or the other predictors.

Subgroup Differences

The fifth hypothesis-that the effects of recent divorce would be stronger for boys, while the effects of earlier divorce would be stronger for girls-was tested in separate regressions for each dependent variable, in which the appropriate interaction term was entered after all other predictors. We also tested interactions involving sex and family conflict. The results, which are not shown, indicated that one interaction was significant, but in view of the large number of interactions tested, this effect must be interpreted cau- tiously. Family conflict was associated with significantly higher levels of depressed mood in Wave One for girls than for boys.

In addition, we tested all of the interactions involving grade level with divorce and family conflict. One interaction effect involving grade level was significant, but this result must be interpreted cautiously. Adolescents in the seventh grade who reported recent divorce also reported significant increases in anxiety over time, whereas recent divorce was not

associated with increased anxiety for ninth graders or eleventh graders.

Exploratory analyses that are not shown revealed more consistent patterns of signifi- cant interactions involving race and ethnicity. Asian adolescents from recently divorced families reported significantly more depressed mood, anxiety, and physical symptoms over time than did adolescents with other racial and ethnic backgrounds. Also, Hispanic adolescents who reported family conflict evidenced significantly greater depressed mood, anxiety, and physical symptoms over time than did other adolescents.

DISCUSSION

The results supported the first hypothesis- that family conflict would have more negative effects on the well-being of adolescents than would divorce or separation. However, the measure of family conflict employed here did not specify which family members were involved, although other data suggested that it reflected parental marital conflict more than parent-child conflict. Relationships between parents in conflict, and between parents and their children, are far more complex than our measures revealed; these findings need to be replicated with more representative samples and with detailed measures of family conflict. The longitudinal results, however, showed that family conflict influenced more dimen- sions of the well-being of adolescents in this sample than did divorce.

We obtained these results controlling for current separation from a parent or steppar- ent, and for whether a parent or stepparent had ever died. Neither separation from parents nor parental death had any significant longitudinal effects on adolescents' well- being; therefore they could not mediate any of the effects of divorce. Of course, our relatively crude measure of current separation did not distinguish between natural parents and stepparents; it is possible that a more adequate measure of separation would pro- duce different results. In addition, we specu- late that the absence of any effect of parental death on adolescents' well-being is due to the small number of adolescents reporting this life event and to the varying lengths of time since it occurred.

The data also supported the second hypoth- esis-that children of divorce living in a

This content downloaded from 130.68.183.9 on Tue, 20 Mar 2018 14:23:49 UTC All use subject to http://about.jstor.org/terms

ADOLESCENT'S WELL-BEING 113

conflict-free setting would have significantly higher levels of well-being than those living in intact families characterized by high levels of conflict. There was also evidence that in some instances divorce may protect adoles- cents from harmful and damaging family interactions. For example, adolescents rated family conflict as significantly more undesir- able than divorce; it is potentially important that fully 25 percent of the children experienc- ing divorce defined it as a positive event. Taken together, these results suggest that the nature of family climates and conflicts may be more crucial than household structure per se, although these data also show that recent divorce had a direct negative effect on adolescents' depressed mood over time.

Hypotheses Three and Four were not supported. Life events and the perceived quality of relationships with parents did not have significant main effects on adolescents' well-being over time, and therefore could not mediate the longitudinal effects of divorce and family conflict. The absence of mediating effects may be due to the high stability of well-being in this sample over a one-year period, and also may reflect the chronic nature of much family conflict. This possibil- ity may help to explain why a substantial proportion of adolescents reported divorce to be a positive experience.

It could also mean, however, that different causal processes are operating. For example, exploratory analyses revealed that adolescents who were perceived by their parents as disagreeable and negativistic elicited more punishment from parents over time. In addition, significant cross-sectional associa- tions existed between punishment reported by parents and parental interest perceived by the adolescent, although parental punishment did not affect perceived interest significantly over time. Thus our longitudinal data provided clear evidence about the relative effects of family conflict and divorce on adolescents' well-being, but did not yield unambiguous findings regarding potential mediating pro- cesses. This is an important area for further research.

Like other large-scale community-based studies, our results did not support the fifth hypothesis-that divorce and family conflict would affect boys and girls differently. In addition, we found no evidence of different effects for various age groups. More consis- tent differences were found among racial and

cultural groups. Family conflict had signifi- cantly stronger negative effects for Hispanics on three measures of well-being than for members of other ethnic groups. Hispanics are more likely to have large extended families characterized by strong relationships (Keefe and Casas 1978). Such families can provide much social support, but conflict in a tightly knit extended family may be especially stressful for adolescents who have not yet developed independent support networks.

Other results showed that recent divorce had significantly stronger negative effects for Asian adolescents than for members of other subgroups on three health outcomes. The tendency for fewer Asian adolescents to report earlier parental divorce or separation may reflect the strong respect for the family typical of this group (Wilkinson 1969). A divorce or separation in such a family may be particularly devastating and may contribute to especially high levels of distress among adolescents. Although the special vulnerabili- ties of Hispanic and Asian adolescents to family conflict and divorce respectively are intriguing, possible sample biases lead us to urge that these results be regarded as no more than suggestive for further exploration.

In conclusion, these results help to illumi- nate the relative effects of divorce and family conflict on various dimensions of adolescents' well-being over time. In addition, the finding that divorce and family conflict have varying relationships with conceptually and empiri- cally different dimensions of well-being suggests that diverse causal processes may be operating, and that we may need to expand our models of divorce and family conflict as more evidence accumulates. The magnitude of the significant coefficients reported here was relatively modest, but we believe that they are important in a practical sense, because it may be possible to make parents more aware of the negative effects of family conflict on adolescents and to teach them better ways of resolving such discord.

NOTES

1. The overall proportion of adolescents who reported a divorce or a separation in this sample (26%) is very similar to national estimates. The National Center for Health Statistics (1987), for example, estimated that in 1985, 26 percent of all 16-year-olds had experienced the divorce of their parents, as had 24 percent of 14-year-olds

This content downloaded from 130.68.183.9 on Tue, 20 Mar 2018 14:23:49 UTC All use subject to http://about.jstor.org/terms

114 JOURNAL OF HEALTH AND SOCIAL BEHAVIOR

and 21 percent of 12-year-olds. Yet the proportion of adolescents in this sample report- ing divorce or separation in the past year (7%) far exceeds the yearly rate of involvement in divorce for any age cohort of adolescents. Our data do not provide an explanation of this anomaly, but we speculate that a reporting bias may have been operating. If divorce and separation are highly salient life events, adoles- cents may underestimate the time interval since the event and may report earlier divorces as occurring in the past year.

2. A log-linear analysis showed that all three partial associations among separation from a parent or stepparent and the death or divorce of a parent or stepparent were significant. When parental death was controlled, adolescents whose parents had divorced were significantly more likely to be separated currently from at least one parent than were those from intact

families (X2 = 320.41, df = 1, p < .001). When divorce was controlled, adolescents who reported the death of a parent were significantly more likely to be separated currently from at

least one parent (X2 = 52.70, df = 1, p < .001). Finally, when current separation from parents was controlled, those who reported the death of a parent were significantly less likely to report parental divorce (X2 = 12.15, df = 1, p < .001).

3. Correlations that are not presented in the text are available from the authors.

4. This study examined four conceptually and empirically distinct dimensions of adolescents' well-being often studied in epidemiological research, although few studies have examined more than one of these dimensions simulta- neously. Unfortunately we had available only one scale each of anxiety, physical symptoms, and self-esteem, and two scales of depressed mood. With two waves of data, we could not estimate measurement models for each scale using procedures such as LISREL or two-stage least squares; therefore we used OLS regression to estimate longitudinal effect coefficients. Consequently it is possible that autocorrelated measurement errors may bias the OLS longitu- dinal effect coefficients presented here. We considered the alternative strategy of develop- ing measurement models based on the individ-

ual items in each scale, but preliminary results showed that the factor structures of all four dependent measures were complex and would require detailed discussion beyond the scope of this paper. We also considered using the four scales to construct a model of global well-being and examining the extent to which adolescents are high on all four component dimensions. This approach, however, would address a different research question.

5. Forty-five adolescents reported that their par-

ents divorced between Waves One and Two. Divorce between waves was not associated significantly with any dimension of well-being measured in Wave Two. Family conflict measured in Wave Two, however, was associ- ated significantly with well-being measured in Wave Two (depressed mood, r = .27; anxiety, r = .25; self-esteem, r = -.17; physical symptoms, r = .20). Also, Wave Two family conflict was associated significantly with di- vorce between waves (r = .11), but, surpris- ingly, Wave One family conflict was not associated significantly with divorce between waves.

6. We also examined mean differences in well- being in both waves for adolescents separated from at least one parent or stepparent in comparison with those who were not separated from a parent or stepparent. Out of eight t-tests between groups, only one was significant. Adolescents living with fewer than two parents or stepparents had significantly higher Wave Two CESD scores than those living with two parents or stepparents. These tests provided further evidence that few differences existed between groups.

APPENDIX Items of Scales Created

for This Study

1. Adolescent Anxiety (a little, some, much, very much) In the past three months, how much have you

a. Felt nervous? b. Felt frightened? c. Felt tense or keyed up? d. Felt restless or fidgety?

2. Adolescent Stressful Life Events in the Past Year (yes, no) a. Have there been big money worries where

you live, like not having enough money for food, rent, clothes, heat, electricity, or the phone?

b. Has anyone ever hurt you or threatened to hurt you in a way that left you afraid for some time?

c. Did any of your parents or stepparents lose their job?

d. Did your family receive food stamps? e. Did your family receive welfare? f. Did your family move to a new home or

apartment?

g. Did you have a close friend move away? h. Did you change schools?

3. Adolescent Perceptions of Parental Interest in the Past Year (never, not very much, some, a great deal; asked separately for each parent) How much did your father (or stepfather) . . .

This content downloaded from 130.68.183.9 on Tue, 20 Mar 2018 14:23:49 UTC All use subject to http://about.jstor.org/terms

ADOLESCENT'S WELL-BEING 115

a. Please you? b. Show you he likes you? c. Show interest in you as a person? d. Talk about schoolwork with you? e. Tell you when you did something well? f. Listen to your ideas and opinions? g. Talk to you at mealtime? h. Love you? i. Spend time with you when you needed him?

4. Adolescent Perceptions of Parental Punishment in the Past Year (never, not very much, some, great deal; asked separately for each parent or stepparent) How much did your father (or stepfather) ... a. Stop talking to you when he was angry with

you? b. Yell or shout at you? c. Make you feel guilty when you did some-

thing wrong?

d. Threaten you when you did something wrong?

e. Hit you when you did something wrong? 5. Parental Perceptions of Disagreement with

Adolescent (never, rarely, sometimes, often, very often) During the last 6 months, how often have you and (child) disagreed about... a. Doing homework? b. How well he/she is doing with schoolwork? c. Doing chores and jobs around the house? d. Giving you enough respect? e. Letting you know where he/she is when

he/she is out late? f. Coming home on time? g. Getting along with other family members?

6. Parental Perceptions of Adolescent Behavior Problems (never, rarely, sometimes, often, very often) How often . . . a. Does your child show an uncontrollable

temper, for example, behavior like scream- ing, swearing, loss of control, angry out- bursts, throwing things, or breaking things?

b. Is your child irritable? c. Does your child do the opposite of what is

asked of him/her? d. Does your child disagree or quarrel with

people for no apparent reason? 7. Parental Perceptions of Adolescent Distress

(never, rarely, sometimes, often, very often) During the last three months how often has he/she . a. Shown signs of loneliness? b. Shown signs of nervousness? c. Seemed depressed or blue? d. Been cheerful? e. Seemed to have guilt feelings? f. Had crying spells? g. Spent a lot of time alone? h. Had a poor appetite? i. Been enthusiastic?

j. Had trouble sleeping? k. Had the energy he/she needed? 1. Been so restless that he/she couldn't sit in a

chair very long? m. Seemed to enjoy life? n. Acted as though he/she just couldn't get

going? o. Had aches and pains? p. Disliked school? q. Spent time with friends? r. Done his/her homework? s. Gotten into fights with people outside the

family?

REFERENCES

Dancy, Barbara L., and Paul J. Handal. 1984. "Perceived Family Climate, Psychological Ad- justment, and Peer Relationships of Black Adolescents: A Function of Parental Marital Status or Perceived Family Conflict?" Journal of Community Psychology 12:222-29.

Emery, Robert E. 1982. "Interparental Conflict and the Children of Discord and Divorce." Psychological Bulletin 92:310-30.

Emery, Robert E., and K. Daniel O'Leary. 1982. "Children's Perceptions of Marital Discord and Behavior Problems of Boys and Girls." Journal of Abnormal Child Psychology 10: 11-24.

Enos, Diane M., and Paul J. Handal. 1986. "The Relationship of Parental Marital Status and Perceived Family Conflict to Adjustment in White Adolescents." Journal of Consulting and Clinical Psychology 54:820-24.

Furstenberg, Frank F., James L. Peterson, Christine W. Nord, and Nicholas Zill. 1983. "The Life Course of Children of Divorce: Martial Disrup- tion and Parental Contact." American Sociolog- ical Review 48:656-67.

Guidubaldi, John, and Joseph D. Perry. 1985. "Divorce and Mental Health Sequelae for Children: A Two-Year Follow-Up of a Nation- wide Sample." Journal of the American Acad- emy of Child Psychiatry 24:531-37.

Hess, Robert D., and Kathleen A. Camara. 1979. "Post-Divorce Family Relationships as Mediat- ing Factors in the Consequences of Divorce for Children." Journal of Social Issues 35:79-96.

Hetherington, E. Mavis. 1979. "Divorce: A Child's Perspective." American Psychologist 34:851-58.

Hetherington, E. Mavis, Martha Cox, and Roger Cox. 1978. "The Aftermath of Divorce." Pp. 149-76 in Mother-Child, Father-Child Relation- ships, edited by M. Matthews. Washington, DC: National Association for the Education of Young Children.

. 1979. "Family Interaction and the Social, Emotional and Cognitive Development of Chil- dren Following Divorce. " Pp. 71-87 in The Family: Setting Priorities, edited by V. Vaugh

This content downloaded from 130.68.183.9 on Tue, 20 Mar 2018 14:23:49 UTC All use subject to http://about.jstor.org/terms

116 JOURNAL OF HEALTH AND SOCIAL BEHAVIOR

and T. Brazelton. New York: Science and Medicine.

. 1981. "Effects of Divorce on Parents and Children." Pp. 233-88 in Nontraditional Fami- lies, edited by M. Lamb. Hillsdale, NJ: Erlbaum.

. 1985. "Long-Term Effects of Divorce and Remarriage on the Adjustment of Children." Journal of the American Academy of Child Psychiatry 24:518-30.

Keefe, Susan E., and Jesus M. Casas. 1978. "Family and Mental Health among Mexican Americans: Some Consideration for Mental Health Services." Pp. 1-24 in Family and Mental Health in the Mexican American Com- munity, edited by Jesus M. Casas and Susan E. Keefe. Los Angeles: Spanish-Speaking Mental Health Research Center, University of Califor- nia.

McCord, Joan, William McCord, and Emily Thurbur. 1962. "Some Effects of Paternal Absence on Male Children." Journal of Abnor- mal Psychology 64:361-69.

Mechanic, David. 1979. "Development of Psycho- logical Distress among Young Adults." Ar- chives of General Psychiatry 36:1233-39.

. 1980. "The Experience and Reporting of Common Physical Complaints." Journal of Health and Social Behavior 21:146-55.

National Center for Health Statistics. 1987. Monthly Vital Statistics Report 36, 8, Supple- ment. Hyattsville, MD: Public Health Service.

Peterson, James L., and Nicholas Zill. 1986. "Marital Disruption, Parent-Child Relation- ships, and Behavior Problems in Children." Journal of Marriage and.ane Family 48:295-307.

Porter, Beatrice, and K. Daniel O'Leary. 1980. "Marital Discord and Childhood Behavior Prob- lems." Journal of Abnormal Child Psychology 8:287-95.

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

Rickel, Annette U., and Thomas S. Langner. 1985. "Short- and Long-Term Effects of Marital Disruption on Children." American Journal of Community Psychology 13:599-611.

Rosenberg, Morris. 1965. Society and the Adoles- cent Self-Image. Princeton: Princeton University Press.

. 1979. Conceiving the Self. New York: Basic Books.

Rutter, Michael. 1971. "Parent-Child Separation: Psychological Effects on the Children." Journal of Child Psychology and Psychiatry and Allied Disciplines 12:233-60.

. 1979. "Maternal Deprivation, 1972-1978: New Findings, New Concepts, New Ap- proaches." Child Development 50:283-305.

Schoenbach, Victor J., Berton H. Kaplan, Edward H. Wagner, Roger C. Grimson, and Francis T. Miller. 1983. "Prevalence of Self-Reported Depressive Symptoms in Young Adolescents." American Journal of Public Health 73:1281-87.

Shaw, Daniel S., and Robert E. Emery. 1987. "Parental Conflict and Other Correlates of the Adjustment of School-Age Children Whose Parents Have Separated." Journal of Abnormal Child Psychology 15:269-81.

Slater, Elisa J., and Joel D. Haber. 1984. "Adolescent Adjustment Following Divorce as a Function of Familial Conflict." Journal of Consulting and Clinical Psychology 52:920-21.

Slater, Elisa J., Krista J. Stewart, and Margaret W. Linn. 1983. "The Effects of Family Disruption on Adolescent Males and Females." Adoles- cence 8:931-42.

Wallerstein, Judith S. 1985a. "Children of Di- vorce: Preliminary Report of a Ten-Year Follow- Up of Older Children and Adolescents." Journal of the American Academy of Child Psychiatry 24:545-53.

. 1985b. "Children of Divorce: Recent Research-Introduction." Journal of the Ameri- can Academy of Child Psychiatry Special Issue: 515-17.

Wallerstein, Judith S., and Joan B. Kelly. 1980. Surviving the Breakup: How Children and Parents Cope with Divorce. New York: Basic Books.

Wilkinson, Thomas 0. 1969. "Family Structure and Industrialization in Japan." Pp. 98-101 in Marriage and the Family: A Comprehensive Reader, edited by Jeffrey K. Hadden and Marie L. Borgatta. Itasca, IL: Peacock.

DAVID MECHANIC, is University Professor, the Rene' Dubos Professor of Behavioral Sciences, and the Director of the Institute for Health, Health Care Policy, and Aging Research at Rutgers University.

STEPHEN HANSELL, Associate Professor in the Department of Sociology and the Institute for Health, Health Care Policy, and Aging Research at Rutgers University, is interested in the social psychology of illness behavior. His recent studies involve adolescents, elderly members of an HMO, and persons with AIDS.

This content downloaded from 130.68.183.9 on Tue, 20 Mar 2018 14:23:49 UTC All use subject to http://about.jstor.org/terms