Paper

usband and Wife Differences in Response to Undesirable Life Events*

RAND D. CONGER FREDERICK O. LORENZ Iowa State University

GLEN H. ELDER, JR. University of North Carolina, Chapel Hill

RONALD L. SIMONS XIAOJIA GE Iowa State University

Journal of Health and Social Behavior 1993, Vol. 34 (March):71-88

In the present study of 451 married couples living in the rural midwest, gender differences were examined in reports of exposure and vulnerability to specific types of^ undesirable life events. Consistent with expectations derived from either a social structural or identity perspective, the results demonstrated that men are more likely than women to report exposure to and to be distressed by work and financial events. Women, on the other hand, are more strongly influenced by exposure to negative events within the family but not within their network of friends. Outcomes vary according to the type of emotional distress. Financial stress, for example, increases hostility among men more than among women, but wives are more likely than husbands to report somatic complaints in response to the same stressor. The findings demonstrate the need for future research that more directly investigates the intraindividual and social mechanisms which account for gender differences in a broad range of emotional and behavioral responses to varying types of significant life changes.

The present investigation addresses a cen- emotional distress suggests that women are at tral issue in social epidemiology: the role of greater risk than men for some types of gender in the stress process (Mirowsky and psychological problems such as depression Ross 1989). Research on sex differences in and anxiety. Men, on the other hand, are

more likely to demonstrate other symptoms of distress such as alcohol or drug use (Cleary

^ , . , ,, ... . ,u 1 .u 1987; Pearlin 1989). In addition to gender * We would like to thank the anonymous ,... ' . r ,. . ,•

reviewers whose comments considerably improved differences m types of distress, a second line the manuscript. This paper is based on collabora- of inquiry has focused on the differential tive research involving the Iowa Youth and reactivity of men and women to varying types Families Project at Iowa State University, Ames, of life crises with the expectation that and the Social Change Project at the University of gender-related patterns of socialization, self- North Carolina, Chapel Hill. The combined research effort is currently supported by National Institute of Mental Health grant no. MH43270, no. MH00567, and the Iowa Agriculture and National Institute on Drug Abuse grant no. Home Economics Experiment Station. For re- DA05347, the John D. and Catherine T. Mac- prints, contact Rand D. Conger, Department of Arthur Foundation Program for Successful Adoles- Sociology, Iowa State University, Ames, IA cent Development among Youth in High-Risk 50011. Journal Paper no. 14942 of the Iowa Settings, Bureau of Maternal and Child Health Agriculture and Home Economics Experiment grant no. MCJ-109572, Research Scientist Award Station, Ames, Iowa, project no. 2931.

71

72 JOURNAL OF HEALTH AND SOCIAL BEHAVIOR

conceptions, or responsibilities within proxi- mal social networks will affect both exposure and response to undesirable life events (Aneshensel, Rutter, and Lachenbruch 1991; Kessler and McLeod 1984; Lennon 1987; Thoits 1987, 1991; Wethington, McLeod, and Kessler 1987).

GENDER, DISTRESS, AND LIFE EVENTS

A number of studies have used diverse measures (e.g., symptom checklists and psychiatric diagnoses), research procedures (e.g., questionnaire self-reports and intensive clinical interviews), and study populations (e.g., urban versus rural) in demonstrating that women report more symptoms of anxiety, depression, and psychophysiological com- plaints (somatization) than men (Cleary 1987; Pearlin 1989), although some divergent find- ings have been reported (Aneshensel et al. 1991). The symptoms of distress are typically those employed in studies of gender differ- ences in vulnerability to undesirable life changes (Bolger et al. 1989; Kessler and McLeod 1984; Thoits 1987; Wheaton 1990). There is growing recognition, however, that men are at higher risk of demonstrating other symptoms of distress such as overt aggres- sion, criminal misconduct, or substance abuse (Aneshensel et al. 1991; Cleary 1987; Lennon 1987; Pearlin 1989). Thus, both contradictory findings regarding women's greater vulnera- bility to distress (e.g., Aneshensel et al. 1991), as well as evidence for higher rates of some emotional or behavioral problems for men, suggest the need for additional research on gender differences for specific symptoms or disorders. An important element in such research is the identification of factors that differentially infiuence emotional distress for men and women.

Explaining Sex Differences in Distress

Several approaches have been taken in efforts to explain why women are more likely than men to demonstrate some symptoms of psychological distress. One might assume that women are disadvantaged by the lower social status they often experience and by role overload or the multiple role demands made of them, such as worker and homemaker.

This social status approach to gender differ- ences in vulnerability suggests that fewer social resources increase the risk of distress, in contrast to life event models which hypothesize that greater exposure to specific life crises accounts for gender differences in psychological functioning. Several recent reviews of the empirical evidence (e.g., Mirowsky and Ross 1989) report modest empirical support for social status explana- tions for different levels of psychological distress among men and women; however, having fewer roles also has been found to operate to women's disadvantage (Thoits 1986). A second perspective on women's greater psychological distress proposes that women lack adequate personal or social resources for coping with the stresses and strains of life (Kessler and McLeod 1984; Thoits 1987). Kessler and McLeod (1984) note, however, that women may fare better than men in their response to many stressful events such as divorce or widowhood. They suggest that if women do in fact have fewer personal or coping resources, they should not demonstrate better adjustment than men to certain types of life crises. Moreover, in a careful test of the coping resource approach, Thoits (1987) found that differences in feelings of mastery and control by men and women did not account for gender differences in distress (see also Ross and Mirowsky 1989).

Some investigators have suggested that lower status groups (including women) generally suffer greater emotional problems because they are exposed to a larger number of undesirable life events (Dohrenwend 1970; Thoits 1987). Although there is evidence that for some types of events and for some symptoms of emotional distress differences in event exposure may account for gender effects on distress (e.g., Thoits 1987), most studies do not find that exposure alone accounts for such sex differences. In sum, variations in social status, multiple role demands, coping re- sources, or event exposure do not appear to account for the greater risk of certain types of psychological distress for women com- pared to men. Moreover, no systematic study has been done within the stress and coping literature specifically designed to identify factors that might explain the higher levels of certain types of distress for men. Although in addition to the stress process

RESPONSES TO UNDESIRABLE LIFE EVENTS

many other factors likely contribute to gender differences in psychological well- being (e.g., gender-related differences in lifestyle), several investigators now suggest that the study of differential response to particular types of life stresses may provide a better understanding of sex differences in emotional distress.

Differential Vulnerability to Undesirable Life Events

Although gender-related research during the past decade has been concerned primarily with explaining women's higher rates of psycholog- ical difficulties, the growing recognition that men are at greater risk than women for many problems other than anxiety and depression sug- gests the need for a new avenue of investigation focused on exposure to events that differen- tially influence the vulnerability of men and women to specific types of psychopathology. Kessler and McLeod (1984), for example, sug- gested that investigators had for too long as- sumed that women are generally more vulner- able than men to stressful life events. They proposed instead that women are particularly reactive only to certain types of stressors, es- pecially undesirable events that occur within their network of family and friends. Here and elsewhere (e.g., Wethington et al. 1987), Kessler and his colleagues suggest that wom- en's roles often involve a greater commitment to nurturing and caring for those around them. They note that women are more likely to bear the emotional costs of caring when family or friends have acute or chronic difficulties.

The findings from Kessler and McLeod's (1984) research were supportive of their per- spective. A combined analysis of data from five independent studies showed that women were more psychologically distressed than men but were not more likely to report a greater overall level of stress (negative life events). However, women did indicate a larger number of crises in their network of close relationships. Men actu- ally reported greater exposure to physical health and financial problems than women, and also experienced similar numbers of marital disrup- tions and love losses. Most important, men and women responded almost equally to income loss, divorce or separation, and other love losses in terms of symptoms of depression and psycho- physiological distress. Working men, however, were more adversely affected by income loss

73

than women who were homemakers. As ex- pected, women were more negatively influ- enced by deaths, ill health, and other adverse events within their social networks. Women's differential response to network events primar- ily accounted for gender differences in psycho- logical well-being in these analyses.

In an important extension of the above study, Thoits (1987) found that women were more likely than men to experience increased anxiety or depression as a result of uncontrol- lable network or personal events. Men were more reactive to negative controllable events such as financial problems. Thus, her results provided support for the earlier study by Kessler and McLeod (1984). Contrary to the earlier study, though, women's differential response to poor health had the greatest impact on their emotional functioning. Net- work events played a secondary role in women's greater vulnerability to specific stressors. Also contrary to Kessler and McLeod's results, women were no more emotionally vulnerable than men to deaths among family and friends. Given the combi- nation of consistent and inconsistent findings across these two studies, further analyses clearly are needed of possible gender differ- ences in reaction to negative life events.

The results from the above two studies suggest that men may be more vulnerable than women to economic difficulties and women more reactive to crises in their social networks. More recent research adds to the credibility of this hypothesis. For example, Bolger and his colleagues (1989) examined the impact of daily stressors on negative mood. Over a six-week period, married couples were asked to record their levels of stress and negative feelings each day in a diary kept with them. For present purposes, the important findings were that women were more adversely affected than men by prob- lems in relationships such as arguments with spouse and others within their social net- works. Men, on the other hand, were more distressed by financial problems. Similar results are reported by Wheaton (1990) in a three-wave panel study of Canadian adults. He found that men's emotional well-being (depression and anxiety) was more strongly infiuenced by work-related stressors while women were more responsive to stresses involving social relationships. Using a psy- chiatric measure of affective or anxiety disorder, an outcome most comparable to that

74 JOURNAL OF HEALTH AND SOCIAL BEHAVIOR

employed by the other studies discussed here, Aneshensel et al. (1991) report contrary fmdings that show greater male reactivity to network events and greater female vulnerabil- ity to financial strains. Thus, although not entirely consistent, the weight of the evidence from this series of studies is highly suggestive that men and women are differentially vulnerable to varying types of undesirable life events.

Attempts to explain these sex differences to reaction to specific negative events have led to two general approaches to the issue, both related to social structural conditions typical to the lives of men and women. The first perspective suggests that gender-related so- cialization practices lead to daily roles, responsibilities, and divisions of labor that uniquely affect men and women (Aneshensel and Pearlin 1987; Kessler and McLeod 1984). Thus, men typically are taught to emphasize occupational success and to acquire the skills necessary to support themselves and a family. Women, on the other hand, by observation or instruction, learn to take greater responsibility than men for providing nurturance in relation- ships, raising children, and caring for a home.

Thoits (1987, 1991) proposes an extension of this social structural perspective by suggest- ing that, as a result of their experiences, men and women typically have divergent self- concepts (see also Stryker and Serpe 1982). She hypothesizes that, on average, they de- velop different role identities which determine expected rights, responsibilities, and priorities in multiple life domains such as work and fam- ily. Stressors that threaten these identities are most likely to create distress for either men or women. If men, for example, are most likely to identify with the worker/family provider role, they will be most distressed by problems such as economic reversals. Conversely, if women most identify with a family/caregiverrole, they will be most vulnerable to negative events in that area of life. The important point here is that either an identity or social structural approach to understanding differential response to nega- tive life events links reactivity to the primary roles and responsibilities in an individual's life.'

RESEARCH OBJECTIVES

The primary purpose of the present study is to investigate further the relationship between gender and distress using a new population of

respondents. As noted, although certain themes are present in earlier research, inconsistencies remain and new data are needed to clarify the relationship between gender and stress pro- cesses. Moreover, the above review demon- strates that overall rates of distress are similar for both sexes; thus, the question today relates not to a female disadvantage but to the identi- fication of specific stressors that differentially influence particular dimensions of distress for men and women. Especially important, men are more likely than women to engage in overt antisocial behavior toward others, such as ag- gressive acts, in response to stressful life con- ditions (Cleary 1987; Pearlin 1989). For exam- ple, unemployed men are more likely to become violent at home than unemployed women. For that reason, in the present study we employ several measures of psychological symptoms, including hostile feelings and behaviors, as well as several indicators of theoretically relevant undesirable events. In addition, to address the influence of previously discussed social param- eters that describe important structural condi- tions or identity-relevant responsibilities (e.g., marital status, residential location, family roles) on the stress process, we employ a sample of relatively traditional husbands and wives who have fairly clearly demarcated roles in family life.

Drawing on the above review, we hypothe- sized that men in the present sample of husbands and wives would be more likely than women to demonstrate symptoms of distress indicative of overt antisocial behav- ior, operationalized in these analyses as hostile behaviors or feelings. Women, on the other hand, were expected to report higher levels of depression, anxiety, or somatic complaints. Because married couples, com- pared to men and women in general, are more likely to share similar life stresses, socioeco- nomic circumstances, personality characteris- tics, and emotional problems (Caspi and Herbener 1990; Thoits 1987; Weissman 1987), this design holds constant several conditions that might confound the relation- ship between gender and distress and thus is a conservative approach to contrasting the effects of some social factors influencing gender differences in psychopathology.

Based on the social structural and identity perspectives, we also hypothesized that hus- bands would report more undesirable events related to financial circumstances, and women would report more crises regarding interpersonal relationships. The sample used

RESPONSES TO UNDESIRABLE LIFE EVENTS

for analysis is particularly well-suited to addressing this proposition in that the couples come from traditional, two-parent, rural families in which the husbands are the primary breadwinners and the wives typically work no more than part-time and have the primary responsibility for home and family. In Thoits' terms (1991, p. 106), these may be thought of as "configural identities" and their concomitant role responsibilities suggest that these men should be more involved in and aware of the economic side of family life and the wives more involved in the maintenance of family relationships and more knowledge- able about crises which arise in that domain.

Also based on the social structural and identity perspectives, we proposed that women would be most vulnerable to crises in their interpersonal networks and men most reactive to financial or work-related difficulties. The strongest support for this hypothesis would be provided by two types of complementary findings: (1) within-group (i.e., within-gender) comparisons showing that, relative to other stressors, work or financial problems are the best predictors of distress for men and network crises the best predictors for women, and (2) between- group comparisons demonstrating significant differences in husband and wife reactions to financial and network events. Importantly, within- and between-group findings in such analyses do not always produce these consistencies (e.g., Aneshensel et al. 1991; Kessler and McLeod 1984); an important contribution of the present study, then, will be to determine whether they might be present in a sample with clearly differenti- ated social roles.

To address this last proposition, we first examined patterns of reactivity within each group of respondents. Within-group regres- sion equations predicting emotional distress for women, for example, were expected to produce larger standardized coefficients for network than financial events. The same analyses were expected to demonstrate rela- tively larger effects for financial or work events for men. Between-group comparisons also were conducted to determine whether network events, as proposed, would have a significantly greater impact on wives' than husbands' distress, and financial or work events would have a larger association with husbands' than wives' distress. As a corollary of this hypothesis, we expected that men's

75

differential reactions to work and financial stressors would be most evident for hostility (predicted to be a more male-oriented form of distress), and women's differential reactions to network difficulties would be most evident for symptoms of depression, anxiety, and somatization.

Finally, we hypothesized that negative family events would be more strongly related to all forms of psychological distress than negative friend events for both husbands and wives. This hypothesis is tested in the within-group analyses. Most previous re- search on stress and gender has aggregated friend and family crises (e.g., Aneshensel et al. 1991; Kessler and McLeod 1984; Thoits 1987), thus obscuring possible differences in their relative effects. We extend earlier findings by disaggregating friend and family events and propose that, given their heavy investments in and mutual responsibilities for family life, these respondents will be more reactive to crises involving family than friends across all measures of psychological distress.

METHODS

Sample and Procedures

A total of 451 White, married couples were recruited for the study. On average, they had been married 18 years when interviewed in 1989. The couples lived in one of eight adjacent, predominantly rural counties in a midwestem state. Consistent with the rural location of the study, none of the couples lived in a town with a population greater than 6,500. The couples are participants in a research project on socioeconomic change in the rural midwest, the focus of which is the investigation of individual characteristics and family pro- cesses that promote successful adaptation by adults and children during stressful times. Because the goal of the research is to study adaptive processes within families, the poten- tially confounding effects of family struc- ture, race, and geographic location were held constant.

The sample includes families that had a child in the seventh grade in a public or private school in one of these eight rural counties during the 1988-89 academic year. The counties were selected because of their

76 JOURNAL OF HEALTH AND SOCIAL BEHAVIOR

rural character and heavy dependence on the agricultural economy, attributes that created enormous social and economic upheaval during the 1980s (Murdock and Leistritz 1988). In addition to the seventh grader, at least one sibling was within four years of age in each household. Only families containing both biological parents of the seventh grader were eligible for the study. Thus, these couples were involved in relatively stable marriages that included significant invest- ments in childrearing. Of those families eligible for participation, 78 percent agreed to be interviewed. To be included, all four family members (husband, wife, seventh grader, sibling) had to agree to participate. Because no data are available for nonrespon- dents, we cannot determine how they differed from participants. We are assured, however, given the wide variability in socioeconomic Status next described, that we interviewed a broad cross-section of eligible couples.

For the total sample, 34 percent of the families lived on farms, 12 percent in the countryside but not on a farm, and the remainder (54%) in small towns. Median family income from all sources (e.g., earn- ings, net farm income, interest, etc.) for the past year was $33,700 and ranged from a net loss to over $100,000; 11 percent of the families fell below the poverty line. These families were slightly poorer than married couples in the United States as a whole, a group that had a median income of $38,164 in 1988 (U.S. Bureau of the Census 1991). Consistent with our interest in relatively traditional couples with divergent responsibil- ities for men and women, the fathers were the primary breadwinners in most of these families as indicated by median earnings of $20,600, compared to $6,000 for wives. The median level of education for both husbands and wives was 13 years, quite comparable to the median of 13.1 years for Whites between 35 and 44 years of age in the U.S. in 1989 (U.S. Bureau of the Census 1991), and ranged from eighth grade to completion of the doctoral degree. Their median ages were 39 (husbands) and 37 (wives) years, ages characteristic of 80.7 percent of parents in married-couple families with children in 1989 (U.S. Bureau of the Census 1991). Because of the study's selection criteria, total family size was above average (median = 5.00). The national median of 4.13 for married- couple families (U.S. Bureau of the Census

1991) suggests that couples in this sample had one more child on average than their counterparts in the nation as a whole. Thus, these couples were similar in terms of age and education to married couples with children in general; however, they were somewhat poorer and had larger families than would be expected from national norms. Given the nature of the study questions, only data for the married couples were considered in these analyses.

The names and addresses of possible participants were obtained from the 34 schools with seventh-grade students in the eight counties. All public and private schools in these counties cooperated. Families were sent a letter explaining the project and then were contacted by telephone and asked to participate. A personal visit was made to those without a telephone. After agreeing to be interviewed, each family was visited twice at home.

During the first visit, the project inter- viewer explained the purpose of the study, and obtained informed consent and demo- graphic information (e.g., ages of all family members, number of people living in the home, etc.). Then each of the four family members separately completed a set of questionnaires that asked about topics such as recent life changes, family economic circum- stances, the quality of family relationships, styles of family interaction, psychological distress, and other issues relevant to the study. In the present paper, we are concerned primarily with the data regarding life changes and emotional distress. This first visit took an average of two hours. During a second visit that occurred about two weeks after the first, the family participated in a series of video- taped interviews to be used for qualitative analyses of family process. Those data were not relevant to questions addressed here. At the conclusion of the second interview, each participating family member was compen- sated at a rate of approximately $10 per hour for his/her time and effort.

Measures

The study involved four basic domains of measurement: gender (husband or wife), undesirable or negative life events, psycho- logical distress, and a set of sociodemo- graphic control variables that might relate

RESPONSES TO UNDESIRABLE LIFE EVENTS 77

both to distress and life changes (Mirowsky and Ross 1989). The sociodemographic control variables included (a) family per capita income, (b) education in years, (c) household size, (d) age in years, and (e) employment status (1= employed at least part-time, 0 = unemployed). Because finer discriminations regarding employment status (e.g., part-time, full-time, unemployed) did not affect the analyses, a simple dichotomy was used. The employment variable did not necessarily indicate low or high stress. For example, some individuals in the study chose not to be employed so that they could stay home with children, pursue an education, take early retirement, etc. Thus, current employment status may or may not have been a result of adverse work changes during the past year.

Undesirable Life Events. Life events were assessed using items primarily from the PERI measure of objective life-event ratings (Dohrenwend et al. 1978). The use of negative life events was consistent with the central role they have been shown to play in the relationship between life changes and emotional functioning (Lin and Ensel 1989; Mirowsky and Ross 1989; Thoits 1983). In separate questionnaires, husbands and wives were asked to indicate independently whether any of 35 different events had happened to them, to a member of their family, or to a friend during the past 12 months. The events were clustered a priori into eight sets of items that were theoretically meaningful within the context of this study. The total score for each cluster is the sum of affirmative responses (1 =yes, 0 = no) for those events.

The events selected for study emphasize our interest in differentiating between crises hypothesized to be particularly salient for these men, who are the primary providers for their families, versus those especially perti- nent for the women in the study who have the major responsibilities for childrearing and household maintenance. These theoretical interests necessitated a sample of negative events related to work and finances as well as crises related to important social networks (e.g., friends, nuclear family, extended fam- ily). We used the research by Kessler and McLeod (1984) as a guide in developing negative event categories.

The first four event types were concerned with negative changes that directly involved the respondents. The first of these involved

job disruptions (four items) during the past year such as being fired, laid off, or demoted. Difficulties with work were considered sepa- rately from financial problems (eight items) which included losses in income or debt problems during the past year. It was particularly important to consider these two areas of change individually as many of the respondents were self-employed and experi- enced financial problems unrelated to loss of employment. These two categories of events include those items indicative of income loss in the Kessler and McLeod (1984) study.

For completeness and consistency with earlier studies (e.g., Kessler and McLeod 1984; Thoits 1987), we also determined whether the respondent had experienced a serious illness or injury during the past year (one item). We included two items about legal problems, e.g., being involved in a court case of some type. Many of these couples had financial difficulties (e.g., loss of a farm) that included legal action. Thus, we expected that these events would most influence the lives of men.

The last four dimensions of change in- volved negative experiences within important social networks: the nuclear family, extended family, and friends. Here we extend earlier work by discriminating between family and friend events. Because of the nature of the sample, two categories of events from the Kessler and McLeod study were not pertinent: separation/divorce and other love loss (except for losses of friends included in our friend events measure). Friend events (five items) included changes during the past year such as health or marital problems among close friends. Deaths among friends were included with the items. Such events were rare for respondents at midlife and the results of the analyses were similar whether friends' deaths were considered separately or as part of the overall category.

Problems within the nuclear family (seven items) included experiences such as a marital separation or increased conflict with the spouse or children. Extended family events (four items) involved questions related to issues such as financial or personal problems among relatives outside the nuclear family. Finally, we considered family health changes (four items) for both nuclear and extended family members within a single cluster of items, e.g., serious illness or accidents experienced by spouse, children, in-laws, or

78 JOURNAL OF HEALTH AND SOCIAL BEHAVIOR

other close relatives. Health events were combined for nuclear and extended families because almost all negative health events involved grandparents. The findings reported here would not change significantly if health events were disaggregated to discriminate between the nuclear and extended family. Moreover, all deaths in the past year occurred in the extended family and they influenced distress in the same fashion as other family health events; thus, they were included with the other items in this category.

Psychological Distress. The SCL-90-R was used to measure symptoms of emotional problems (Derogatis 1983). This instrument has demonstrated reliability and validity as a measure of psychological distress. Although earlier stress research has been concerned primarily with anxiety and depression (Mirowsky and Ross 1989), four SCL-90 subscales were of particular interest in this study: depression, anxiety, somatization, and hostility. The hostility subscale was included to determine whether men are more likely than women to respond to stress in an angry manner, as might be suggested by epidemio- logical studies of psychiatric disorder (Cleary 1987). The other subscales are those typically used in earlier studies of gender differences in distress (e.g., Kessler and McLeod 1984; Thoits 1987).

Respondents were asked to report on a five-point scale how much discomfort they had experienced during the past week related to each symptom of somatization, depression, anxiety, and hostility. Responses ranged from no discomfort to extreme discomfort. Scores were summed to determine the level of symptomology. The higher the score on a measure, the greater the feelings of distress. The somatization subscale (twelve items) includes bodily complaints such as headaches or nausea that are known to be correlated with anxiety and depression (Mirowsky and Ross 1989). Thirteen symptoms (e.g., "feeling

blue") are included in the depression mea- sure, ten in the anxiety measure (e.g., "feeling fearful"), and six in the hostility subscale (e.g., "having urges to break or smash things"). Internal consistency coeffi- cients for this sample ranged from .67 (wives' hostility) to .87 (husbands' and wives' depression). The median coefficient of .79 across spouses and subscales suggests reason- ably good reliability for these measures.

RESULTS

Appendix A provides the correlations, means, and standard deviations for all study variables, reported separately for husbands and wives. The measures of emotional distress were moderately to highly correlated for both spouses, ranging from r = .43 (somatization and hostility for both husband and wife) to r = .79 (depression and anxiety for husbands). The highest correlation be- tween an event category and distress for husbands was .29 between negative financial events and depression, consistent with study hypotheses. For women, the highest event/ distress correlation was .26 between nuclear family crises and depression, also as ex- pected.

Gender Differences in Level of Distress and Number of" Undesirable Events

Our first hypothesis was that men would report more symptoms of hostility, while women would report greater levels of depres- sion, anxiety, and somatization. Table 1 provides the mean scores for distress by gender. The multivariate F-test [F(4447) = 8.02, /?<.OO1] for repeated measures across correlated groups indicates a statistically significant difference in the means, the wives averaging significantly higher scores than

TABLE 1. Mean Scores for Emotional Distress for Husbands and Wives

SCL-90 Subscales

Somatization Depression Anxiety Hostility

Husbands

16.10 18.87 12.68 7.90

Wives

16.63 20.82 13.22 8.33

Univariate F(, 450,

3.39+ 25.46*** 4.59* 8.14**

Note: Multivariate ^,4447, = 8.02, p<.001. Multivariate and univariate tests are for repeated measures across dependent groups. Mean scores reflect the total summed score for each subscale.

V < 1 0 ; * p < . 0 5 ; * * p < . 0 1 ; ***p<.001.

RESPONSES TO UNDESIRABLE LIFE EVENTS 79

husbands for all forms of distress: somatiza- tion, depression, anxiety, and hostility. The greatest gender difference was for depression, the least for somatization. For this sample, men did not report higher levels of hostile feelings or behaviors, contrary to our expec- tations.

Consistent with our second hypothesis, men reported an average of .84 adverse financial events compared to .61 (p<.OOl) for women (Table 2). There were no significant gender differences for job disrup- tions, illness or injury, or legal problems. The results for network events were consistent with expectations for family relationships but not for friends (Table 2). Women reported more negative events than men for all three areas of family change. These spouses reported very similar numbers of friend events, however. Taken together, the results for gender differences in reports of various domains of undesirable life changes are consonant with the study hypothesis.

Within-Group Differences in Vulnerability to Stress

To address our third hypothesis, regression equations for each dependent variable were computed (Tables 3a and 3b). The standard- ized coefficients in parentheses indicate the relative effects of the independent variables within each equation. According to our third hypothesis, men should be most vulnerable to financial or work crises compared to the other negative events they experience and women should be most distressed by network events.

For men, the within-group findings are reasonably consistent with the hypothesis.

The largest standardized coefficients for husbands involve negative financial events for each dimension of emotional distress except somatization, which was most strongly re- lated to a control variable (employment status) followed by negative friend events (b* = .l2, p<.Ol, Table 3a). Men's anxiety was most strongly associated with financial problems {b* = .lS, p<.OOl, Table 3a), as was depression {b*=.24, p<.OOl, Table 3b) and hostility (Z>* = .18, p < . 0 0 1 . Table 3b). Moreover, in three out of four regressions, the findings for husbands were consistent with our fourth hypothesis that family, compared to friend, crises would be more strongly associated with emotional distress. Only in the case of somatization did the findings suggest that negative friend events were more distressing for men than family events.

For women, the findings also were gener- ally consistent with the within-group predic- tion that they would be most affected by network events. The largest standardized coefficients for anxiety (nuclear family events), depression (extended family events), and hostility (nuclear family events) all relate to network changes. Also important, and consistent with our final hypothesis, was the finding that crises within the family were far more important for these women than friend events, which were not statistically significant for any outcome. Wives' symptoms of somatization, however, were most strongly associated with their own health problems, which may refiect some confounding between the two measures. The results for men, however, did not reflect such a confound. Thus, for both husbands and wives, the within-group analyses were supportive of

TABLE 2. Mean Scores for Undesirabte Life Changes for Husbands and Wives

Life Events Husbands Wives Univariate F,,

Personal Job disruptions Financial problems Illness or injury Legal problems

Network Friend events Nuclear family Extended family Family health

.24

.84

.05

.09

1.07 .37 .76 .48

.20

.61

.08

.07

.98

.54

.92

.58

1.35 16.91** 2.09 2.28

2.31 18.62** 9.57* 8.20*

Note: Multivariate dependent groups.

* p<.Ol; ** p<.0Ol.

437) = 8.51, p<.0Ol. Multivariate and univariate tests are for repeated measures across

80 JOURNAL OF HEALTH AND SOCIAL BEHAVIOR

- Q

o t/5

I I I I

8

I I

8S I I I

(S O — OS

I I

q oioo—^ o q S q q I w - ^ ^ ^ I ^ I ^ ^

8

o — o o I I

I I

§ e . | g a 2

5 « o 6 LS u III

1 ^ \ii^

i •? "ĝ.S

^ E z l l l " ' " " ^ - S •£ ? I

•a o

2 » on *

g o

T3

•o a.

RESPONSES TO UNDESIRABLE LIFE EVENTS 81

00 — O —

I I I I

S S S : S — o —(N — O I I

I I

IT) VO O — r>1 O

8

S I

• " s i l l

82 JOURNAL OF HEALTH AND SOCIAL BEHAVIOR

study hypotheses except for those related to somatization. We consider these divergent results for this measure of distress in our later discussion of the findings.

Between-Group Analyses

Tables 3a and 3b also provide the unstand- ardized regression coefficients for between- gender comparisons. We hypothesized that wives, compared to their husbands, would be more vulnerable to network events, while men would be more reactive to financial and work problems. Because husbands and wives are paired groups, we employed generalized least squares estimators for seemingly unre- lated regression equations rather than ordinary least squares (Johnston 1984; Zellner 1962). The findings were only partially consistent with the third hypothesis.

The results for hostility provided the best fit with expectations for the differential vulnerability of men. The unstandardized coefficients predicting hostility (Table 3b) were significantly larger for men than women for financial problems {b=.'i5, husbands; fc=.O8, wives; p < . 0 5 for the difference in slopes) and for legal problems (Z)=1.18, husbands; fc=.16, wives; p < . 0 5 for the difference in slopes). Also consistent with the third hypothesis, men were more likely than women to report symptoms of anxiety (p<.10) or depression (p<.10) as a result of legal problems.

Although wives were somewhat more likely than their husbands to be vulnerable to depression (Table 3b) as a function of crises in the extended family {b=.19, husbands; b= 1.40, wives; p<.\0 for the difference in slopes), the findings were generally inconsis- tent with the thesis that women would be more vulnerable than men to network events. Indeed, the association between friend events and somatization was a bit greater for men than women (/j<.10). Also contrary to expectations, the relationship between finan- cial problems and somatization (Table 3a) was stronger for wives {b—.59) than for husbands (fo=.O6, p<.05 for the difference in slopes).

DISCUSSION

Four hypotheses guided these analyses. The results did not support our first proposi-

tion that husbands would report more symp- toms of hostility than their wives, although they responded to some events in a more hostile fashion than their spouses, a point we will return to later in the discussion. These analyses did replicate the usual finding that women, compared to men, report higher levels of depression, anxiety, and somatic complaints (Pearlin 1989). We suspect that our measure of problem behavior, self- reported hostile feelings, and actions during the past week may not correspond as well as expected with more general antisocial tenden- cies, including illegal acts, that characterize men more than women (Cleary 1987). Indeed, low levels of irritable and hostile behavior are often associated with depressive symptoms (Downey and Coyne 1990) and we suspect that the hostility measure employed here is more consonant with aversive acts of this type than with the more dramatic forms of antisocial behavior found to be more prevalent among men in earlier research. Future research on these issues should include measures that emphasize overt antisocial tendencies to a greater degree than the indicator of hostility used here.

The data did support our second hypothesis that women would report more undesirable network events and men more negative financial changes during the past year. The present results are consistent with Kessler and McLeod's central findings that reports of income loss are more typical of men while the prevalence of network crises is greater for women. Although the events are not exactly the same as in their study, the discrimination between economic and interpersonal domains is the same. In addition, these results demonstrate that, at least for relatively traditional married women, negative family events are more common than problems among friends. In contrast to these findings and the Kessler and McLeod results, Thoits (1987) reported that married women in her study did not experience more negative network events than men nor did men report more financial problems. Moreover, she found that women reported more negative health events than men while Kessler and McLeod found the reverse. In this study we find no gender differences for negative health changes during the past year, a result that may reflect the relatively young age of the respondents. Clearly, much additional work is needed on the varying dimensions of social

RESPONSES TO UNDESIRABLE LIFE EVENTS 83

context and life stage that determine risk for exposure to different life stressors for both men and women. For these relatively tradi- tional couples at mid-life, however, gender- based predictions for differential exposure to negative events based on Kessler and McLeod's research received considerable empirical support.

The findings related to our third hypothesis demonstrate the importance of considering variation in psychological outcomes and stressors within the same set of analyses (see also Aneshensel et al. 1991; Lennon 1987). We proposed that both within- and between- group analyses would demonstrate greater vulnerability of wives to negative network events and of husbands to financial and work events. For men, three out of four of the within-group analyses were consistent with the hypothesis. In the case of somatization, though, crises among friends were the events most strongly related to psychological symp- toms for men. For the between-group analy- ses, men were significantly more vulnerable than women to negative financial or legal events only for hostility, a response we predicted would be more typical of husbands. For symptoms of depression and anxiet̂ y, men were marginally {p<.lO) more vulnerable than women to legal problems. For somatiza- tion, women were actually more vulnerable to financial problems than men.

As noted, the greatest vulnerability of men to financial and legal problems occurred when hostility was the measure of distress. Thus, even though men were not more hostile than their wives in general, they reacted in a more hostile fashion in response to sex-typed negative events. In this sense, the results for hostility were as predicted. In addition, the findings suggest that the mean level of distress represents only one of several impor- tant dimensions of gender effects in the stress process. The complex interaction found in this study among gender, event type, and psychological outcomes is equally important.

For women, the within-group analyses were consistent with the hypothesis of greater vulnerability to network events except for somatic complaints, which were most strongly related to illness or injury during the past year. This finding concerning wives' somatization may result, at least in part, from confounds in measurement in that somatic complaints at the time of the interview may reflect continuing symptoms from prior ill-

ness or injury. Contrary to this thesis, however, illness or injury reported by men was only marginally (p<.10) related to somatization, which was better predicted by network events for husbands.

Although the strong association between somatization and men's employment status suggests that unemployment leads to somatic symptoms, it also suggests another possible confound in the somatization measure in that ill or injured husbands may be less able to work. This possibility cannot be tested with these data because we would need to have a measure of somatic complaints prior to changes in employment status. Future re- search will be needed to clarify the degree to which these within-gender findings for soma- tization result from measurement confounds or model misspecification or whether they refiect real differences in the stress process for this measure of distress. Until findings are available from research directed toward these measurement and theoretical issues, we will not know whether the within-group results for somatization, which are so inconsistent with the results for the other measures of psycho- logical status, place in doubt its value as an outcome variable in stress reseju'ch.

The between-group analyses did not sup- port our thesis that wives would be more vulnerable than husbands to negative network events. Only for crises in the extended family did wives demonstrate greater vulnerability than husbands, and this finding held only for depression and was only marginally signifi- cant (p<.lO). Moreover, negative friend events were marginally more predictive of men's than women's somatic symptoms.

Particularly unexpected was the finding that women were more reactive than men to negative financial changes when the criterion measure was somatization. This finding again demonstrates the importance of considering varying relationships between distress and stress processes for men and women. Future research will be needed to determine whether the present finding of greater hostility by men and greater somatization by women in response to financial stress is generalizable beyond the present study population. How- ever, as indicated earlier, the findings for somatization are somewhat suspect and should be interpreted cautiously. In any case, these findings again sensitize us to the fact that sex differences in stress processes may involve complex interactions among gender.

84 JOURNAL OF HEALTH AND SOCIAL BEHAVIOR

types of distress, and nature of the stressor. Indeed, our findings are consistent with the conclusion reached by Aneshensel et al. (1991) that gender differences in vulnerability to stress may vary depending on the types of disorder and stress being considered. Despite the small number of significant between- group differences in reactivity to network events, the results clearly show that women experienced more undesirable family events than men and were more distressed by them than by other crises in their lives. Their enmeshment in a greater number of family problems than men increased the likelihood that these role-related difficulties would exacerbate their psychological distress. Simi- larly, men were more distressed by financial than other problematic changes in their lives and were more likely than their wives to experience such events.

In retrospect, the failure to find pervasive and significant between-group differences for relationship events may not be surprising. These couples were involved in marriages that averaged almost two decades in length and they had a greater than average (mean = 2.95) number of children. From their behav- ior we can infer that they had an above average commitment to stable family relation- ships and perhaps greater than normal inter- personal skills that may be necessary to maintain continuity in family life. They also lived in a rural environment known to be characterized by close ties within extended family systems (Marotz-Baden, Hennon, and Brubaker 1988). Given their life histories and location in family-oriented social systems, the husbands in this sample may well be as sensitive as their wives to negative family events, as these data suggest. Moreover, the reactivity of men to friends' crises, contrary to findings from earlier research, may reflect the generally closer social ties in rural environments in contrast to the urban settings of most prior studies. The between-group findings tell us that for this very family- invested group, men are almost as reactive as wives to family crises. Although not entirely consistent with study hypotheses, this result is similar to one reported by Wethington et al. (1987), who found that men were about as vulnerable as women to family crises and were less reactive only to negative events involving more distal social relations. Future research will need to clarify whether this finding is generalizable to similar samples or

to the population of men and women as a whole.

Regarding our fourth hypothesis, with a single exception (i.e., husband's somatic symptoms), both spouses experienced greater emotional distress as a function of stress in family compared to friend social networks. This finding is consonant with the above

. discussion regarding the heavy investment in family life by this sample of respondents. Moreover, it is consistent with either a social structural or identity perspective. If our thesis about the greater importance of family ties in rural compared to urban settings is correct, we would not expect the exact same pattern of findings to hold in urban settings in which friendships may rival family relations in their importance. Rather than simply comparing urban and rural respondents on stressors and outcomes, however, future studies should examine these processes while also investigat- ing the value placed by urban or rural respondents on family ties. For urban and rural people with equivalent values about family life, the results of such research may be quite the same regardless of geographic location.

Taken together, these findings suggest that social roles and responsibilities, and perhaps self-conceptions concomitant to such statuses, play an important role in determining the salience and infiuence of negative life events. Overall, the analyses are more consonant with Kessler and McLeod's (1984) findings than with Thoits' (1987). In particular, as did Kessler and McLeod, we found that women are more exposed than men to social network crises, especially in the family, and that men experience a larger number of undesirable financial events. Contrary to the results reported by Thoits (1987) but consistent with the Kessler and McLeod (1984) study, we found that relationship events played a greater role in women's vulnerability than did health problems. We will be more confident about these findings, however, when these results are replicated with longitudinal data. It is likely that distress leads to a more problem- atic social environment, a possibility noted earlier for the relationship between employ- ment status and somatization. An understand- ing of such reciprocal connections will require prospective, longitudinal research.

There is also evidence, however, that contextual factors may influence ties to traditional notions of gender roles and respon-

RESPONSES TO UNDESIRABLE LIFE EVENTS 85

sibilities. Samples with quite different demo- graphic characteristics than the couples in this study may produce very different results (Aneshensel et al. 1991). Thus, the next stage in this research must go much further in elaborating the mechanisms at work. Rather than inferring role responsibilities or self- conceptions, for example, these constructs must eventually be measured directly and entered into our analyses as others have suggested (Brown, Bifulco, and Harris 1987; Thoits 1991). In addition, more information should be generated about specific commit- ments and activities in major life domains such as work and family that might determine resilience or vulnerability to particular stres- sors, especially when men and women have similar role responsibilities (Lennon 1987). As Lennon (1987) also has suggested, progress in research on gender and psycho- logical distress will require a multiplicity of theoretically relevant measures related to work and family roles and emotional and behavioral outcomes.

Also important in future research will be greater attention to specific life stages, life circumstances, and residential situation. For example, we have suggested that the hus- bands in this study may have been more reactive than expected to network events because of the norms and priorities in a rural environment. In addition, the wives studied here may have been less responsive to financial and work-related crises because of their generally secondary responsibilities in these areas in a more traditional, relatively conservative family context. It also may be the case that family problems were especially distressing to both men and women in these families because of their concern for adoles- cents soon to be on their own and for their own aging parents, many of whom are experiencing significant life transitions such as retirement, chronic health problems, or loss of a spouse. These cross-cutting dimen- sions of social life should be investigated

carefully in future research on gender differ- ences in distress and on how these differences may be mediated or moderated by everyday social contingencies that are located within and vary by larger social contexts and stage in life. Conducting such intensive research may require an emphasis on relatively homoge- neous samples that can be subdivided by variables of theoretical interest (e.g., role identities) while being held constant on other significant factors (e.g., marital status). Such a strategy, compared to large-scale surveys of broad cross-sections of the adult population, may provide a cost-effective means for identifying the values, beliefs, and social processes that may help to account for gender differences in psychological distress.

NOTES

1. In the present analyses we are most concerned with the relation between specific types of life events and gender-related emotional outcomes. For these associations, either a social structural or identity perspective leads to similar predic- tions. The intervening processes that account for such relationships, however, may be quite different. In identity theory, the role incumbent must feel that some important dimension of self has been threatened by an undesirable event for psychological problems to occur. From a structural perspective, this cognitive mediator is not a necessary part of the process and the important linking variables between events and emotional distress may have more to do with increased role demands, the imposition of barriers that produce failure in meeting norma- tive expectations, or other such structural features of the proximal environment. From a structural perspective, these environmental con- tingencies should influence psychological dis- tress or well-being even in the absence of self-perceptions that describe a high degree of identification with a specific role. Thus, with appropriate data, one could test empirically whether the identity mediator or moderator is necessary or whether structural demands alone account for variations in psychological status.

86 JOURNAL OF HEALTH AND SOCIAL BEHAVIOR

iv e s]

.o

s

o

1

a o

5

.3

I

cs

2

3

I55

o

s

Pi

.3 8

.3 2

.3 8

!o

.2 0

0 0

.6 1

.0 8

.0 7

s —

.9 8

0̂ S

*n ON .5 8

»n

07

§?̂

.4 0

^ — •*

.7 0

.7 8

0000 — 0000000 — 0 — 0 I I I I I I I I I

m in in c^ 0*1 p*̂ in •̂" ^0 ^^ 0*1 00 ^^ 00 CN I G*^ P ^ ^^ 0 0 0 0 0 0 0 0 0 0 — O ( N — O l O - J O N

I I I I I I I

O O O O O — O — O — — O t N O I I

0 0 0 0 I I I

— — o o o r j o o o o o o i f N t N — 0 0 — I I I I I I I I I I I I M IP;

I — o o o o o r - 0 0 I I I • •

C N t N t N — O — O — — I —

~~ —" O O O O d O I O fS —" O O O O O O ~™ I I I - : -.:

S u-1 — v o o o O N f s l o v o n o o - — w n ^ ( N O \ —o o o o o — I o — O O O O O O O O r ^ I I I I • •

I I I I I • •

0 ^ O ^ * ^ T t • 0 ^ I C S 0 \ O O O » O ^ f * ^ 0 0 — - O N d ^ I O O < N — ( N O — O O O O O O —

I I I I • - :

8 5 S S I 2 3 5 3 5 2 2 3 2 3 5 2 2 ; I I I I

r ^ f ^ o o | i o O N ' » t ' n i ^ O N \ o o o i n o c N > r ) i y - i ' ^ u - ) • / ^ T T V ^ ' 0 0 — 0 ^ 0 — 0 0 . — 0 0 < N c * ^ r o

I I I I - • •

I - . V O i v o o n o — — — ( N o o o o o — ( N c n I I I I I - : '

in 1 i l O r ^ O O O O r ^ - ^ O v f N ^ O ^ C N O O I O C N O O O C N O — O ^ — O O O O O O t ^ f * ^

I I I I - : '

o o v r ^ ^ O N - o o \ — o » n o ^ - ^ — a \ » n f ^ r - r - u - > O ( N — — O c s f s o — O O O O ' * - *

I I I - : •

i.jtilillliii^ 1, *-J HC '̂ ̂ ̂ UH S HJ UH Z U UM Qd UJ tL ̂ UJ

"" (N CO ̂ <n ̂ p^ 00 ON O "~ CN CO ̂ in ̂ o p».

1

il

g'l Al S

.1

RESPONSES TO UNDESIRABLE LIFE EVENTS 87

REFERENCES

Aneshensel, Carol S. and Leonard I. Pearlin. 1987. "Structural Contexts of Sex Differences in Stress." Pp. 75-95 in Gender and Stress, edited by R. C. Bamett, L. Biener, and G. K. Baruch. New York: Free Press.

, Carolyn M. Rutter, and Peter A. Lachen- bruch. 1991. "Social Structure, Stress, and Mental Health: Competing Conceptual and Analytic Models." American Sociological Re- view 56: 166-78.

Bolger, Niall, Anita DeLongis, Ronald C. Kessler, and Elizabeth A. Schilling. 1989. "Effects of Daily Stress on Negative Mood." Journal of Personality and Social Psychology 57: 808-18.

Brown, George W., Andrew Bifulco, and Tirril O. Harris. 1987. "Life Events, Vulnerability and Onset of Depression: Some Refinements." British Journal of Psychiatry 150:30-42.

Caspi, Avshalom and Ellen S. Herbener. 1990. "Continuity and Change: Assortative Marriage and the Consistency of Personality in Adult- hood." Journal of Personality and Social Psychology 58:250-58.

Cleary, Paul D. 1987. "Gender Differences in Stress-Related Disorders." Pp. 39-72 in Gender and Stress, edited by R. C. Bamett, L. Biener, and G. K. Tsanich. New York: Free Press.

Derogatis, Leonard R. 1983. SCL-90-R Manual-II. Towson, MD: Clinical Psychometric Research.

Dohrenwend, Barbara S. 1970. "Social Class and Stressful Events." Pp. 313-19 in Psychiatric Epidemiology, edited by E. H. Hare and J. K. Wing. New York: Oxford University Press.

, Larry Krasnoff, Alexander R. Askenasy, and Bruce P. Dohrenwend. 1978. "Exemplifica- tion of a Method for Scaling Life Events: The PERI Life Events Scale." Journal of Health and Social Behavior 19:205-29.

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

Johnston, John. 1984. Econometric Methods, 3rd edition. New York: McGraw-Hill.

Kessler, Ronald C. and Jane D. McLeod. 1984. "Sex Differences in Vulnerability to Undesir- able Life Events." American Sociological Re- view 49:620-21.

Lennon, Mary C. 1987. "Sex Differences in Distress: The Impact of Gender and Work Roles." Journal of Health and Social Behavior 28:290-305.

Lin, Nan and Walter M. Ensel. 1989. "Life Stress and Health: Stressors and Resources." American Sociological Review 54:382-99.

Marotz-Baden, Ramona, Charles B. Hennon, and Timothy H. Brubaker. 1988. Families in Rural America: Stress, Adaptation and Revitalization. St. Paul, MN: National Council on Family Relations.

Mirowsky, John and Catherine E. Ross. 1989.

Social Causes of Psychological Distress. New York: Aldine.

Murdock, Steve H. and F. Larry Leistritz (Eds.). The Farm Financial Crisis: Socioeconomic Dimensions and Implications for Producers and Rural Areas. Boulder, CO: Westview Press.

Pearlin, Leonard 1. 1989. "The Sociological Study of Stress." Journal of Health and Social Behavior 30:241-56.

Ross, Catherine E. and John Mirowsky. 1989. "Explaining the Social Patterns of Depression: Control and Problem Solving—or Support and Talking?" Journal of Health and Social Behav- ior 30:206-19.

Stryker, Sheldon and Richard T. Serpe. 1982. "Commitment, Identity, Salience, and Role Behavior." Pp. 199-218 in Personality, Roles, and Social Behavior, edited by W. Ickes and E. S. Knowles. New York: Springer-Verlag.

Thoits, Peggy A. 1983. "Dimensions of Life Events that Influence Psychological Distress: An Evaluation and Synthesis of the Literature." Pp. 33-103 in Psychosocial Stress: Trends in Theory and Research, edited by H. B. Kaplan. New York: Academic Press.

1986. "Multiple Identities: Examining Gender and Marital Status Differences in Distress." American Sociological Review 51: 259-72.

1987. "Gender and Marital Status Differ- ences in Control and Distress: Common Stress versus Unique Stress Explanations." Journal of Health and Social Behavior 28:7-22.

. . 1991. "On Merging Identity Theory and Stress Research." Social Psychology Quarterly 54:101-12.

U.S. Bureau of the Census. 1991. Statistical Abstract of the United States. Washington, DC: U.S. Department of Commerce.

Weissman, Myma M. 1987. "Advances in Psychi- atric Epidemiology: Rates and Risks for Depres- sion." American Journal of Public Health 77:445-51.

Wethington, Elaine, Jane D. McLeod, and Ronald C. Kessler. 1987. "The Importance of Life Events for Explaining Sex Differences in Psychological Distress." Pp. 144-56 in Gender and Stress, edited by R. C. Bamett, L. Biener, and G. K. Baruch. New York: Free Press.

Wheaton, Blair. 1990. "Life Transitions, Role Histories, and Mental Health." American Socio- logical Review 55:209-23.

Zellner, Amold. 1962. "An Efficient Method of Estimating Seemingly Unrelated Regressions and Tests for Aggregation Bias." Journal of the American Statistical Association 57:348-68.

88 JOURNAL OF HEALTH AND SOCIAL BEHAVIOR

Rand D. Conger is professor of sociology, Iowa State University, and Director of the NIMH-funded Center for Family Research in Rural Mental Health. Some of his research interests include the transition from adolescence to adulthood, family research methods such as the direct observation of family interactions, and the influence of family dynamics and acute or chronic stressors on physical and mental health.

Frederick Lorenz is associate professor in the Departments of Statistics and Sociology, Iowa State University. His research interests include the effects of measurement error in modeling family processes and individual adaptation to psychosocial stressors.

Glen H. Elder, Jr. is Howard W. Odum Distinguished Professor of Sociology and Research Professor of Psychology, University of North Carolina, Chapel Hill, where he directs the Social Change project on life course and family studies. His research activities include studies of families and children under economic pressure in the Iowa project and research on the life course effects of World War II.

Ronald L. Simmons is professor of sociology and principal investigator for the Single Parent Project at Iowa State University. His research interests include parenting practices and adolescent developmental outcomes, family processes and individual development in single-parent households, and homeless adolescents and adults.

Xiaojia Ge is research associate. Center for Family Research in Rural Mental Health, Iowa State University. His research interests include family research methods, adolescent development, and family stress processes.