The Effect of Social Relationships on Psychological Well-Being: Are Men and Women Really So Different?

by Debra Umberson, Ellen Slaten, Kristine Hopkins, James S. House, Meichu D. Chen
Citation
Title:
The Effect of Social Relationships on Psychological Well-Being: Are Men and Women Really So Different?
Author:
Debra Umberson, Ellen Slaten, Kristine Hopkins, James S. House, Meichu D. Chen
Year: 
1996
Publication: 
American Sociological Review
Volume: 
61
Issue: 
5
Start Page: 
837
End Page: 
857
Publisher: 
Language: 
English
URL: 
Select license: 
Select License
DOI: 
PMID: 
ISSN: 
Abstract:

THE EFFECT OF SOCIAL
RELATIONSHIPS ON PSYCHOLOGICAL WELL-BEING:
ARE MEN AND WOMEN REALLY SO DIFFERENT?*
Debra Umberson Meichu D. Chen

University of Texas University of Texas
James S. House Kristine Hopkins Ellen Slaten

University of Michigan University of Texas University of Texas

We assess evidence for gender differences across a range of relationships and consider whether the form and quality of these relationships affect the psy- chological functioning of men and women differently. Data from a national panel survey provide consistent evidence that men's and women's relation- ships differ However; we find little evidence for the theoretical argument that women are more psychologically reactive than men to the quality of their relationships: Supportive relationships are associated with low levels of psy- chological distress, while strained relationships are associated with high lev- els of distress for women and for men. However; if women did not have higher levels of social involvement than men, they would exhibit even higher levels of distress relative to men than they currently do. We find little evidence for the assertion that men and women react to strained relationships in gender- specific ways-for example, with alcohol consumption versus depression.

ocial theorists provide compelling rea-

sons to expect men and women to expe- rience relationships differently (Chodorow 1978), and many empirical studies identify gender differences in specific characteristics of social relationships. Bestsellers describe the difficulties and misunderstandings that follow from these purported gender differ- ences (Tannen 1991). However, much of the empirical evidence for these differences is based on nonrepresentative samples or se- lected aspects of men's and women's rela- tionships. Are men's and women's relation- ships really different? Do relationships affect the well-being of men and women in differ- ent ways?

'~irect correspondence to Debra Umberson (Umberson@jeeves.la.utexas.edu). We thank Judith Howard, Margaret M. Marini, Arthur Sakamoto, Jordan Steiker, Yu Xie and, especially, Christine L. Williams for their comments on ear- lier versions of this paper. This research was sup- ported in part by a FIRST Award to Debra Umberson (NIA #AG05440) and NIA #AGO5562 (James S. House, Principal Investigator).

Our main purpose is to assess links con- necting gender, relationships, and psycho- logical functioning. We first analyze data from a national sample to assess gender dif- ferences in the quality and quantity of a range of kin and nonkin relationships. We examine the frequency with which men and women socialize informally, the number of people to whom they feel close, and the de- gree to which their relationships are charac- terized by strain and support. We also con- sider whether relationships affect the psycho- logical functioning of men and women dif- ferently. We examine whether the quality of a relationship is associated with psychologi- cal distress for one gender more than for the other. We use measures of psychological functioning that recognize that men and women may express psychological distress in different ways (Aneshensel, Rutter, and Lachenbruch 199 1 ).
THEORETICAL EXPLANATIONS

Social theorists, working from diverse per- spectives, have long argued that women are

American Sociological Review, 1996, Vol. 61 (October:837-857)

more nurturant and relationship-oriented than are men (Chodorow 1978; Parsons 1955). Explanations for these gender differ- ences typically highlight the importance of early childhood experiences. Socialization theorists argue that, through the process of socialization, boys are encouraged to express traditionally male behaviors, such as aggres- siveness and competitiveness, which are bar- riers to intimacy (Marini 1988). Girls are en- couraged to restrict competition and aggres- sion aid to exhibit emotions and nurturance, factors that may foster intimacy and the for- mation of relationships. These gender differ- ences are reinforced by peer group activities, educational processes, and the mass media (Marini 1988). Gilligan (1982) argues that gender-role socialization leads girls to de- velop a stronger moral sense of caring for others.

Chodorow (1978) argues that because adult women tend to be the primary caretak- ers of infants, most children begin life with a feminine identification. Whereas boys are compelled to break with this identification to establish a masculine identity, girls are en- couraged to maintain this primary attachment to their mothers. Consequently, girls, and later, adult women, often experience prob- lems developing a strong psychological sense of separation from others. This may lead to women's "loss of self in overwhelm- ing responsibility for and connection to oth- ers" (Chodorow 1978:59). Although Cho- dorow does not discuss psychological dis- tress as an outcome of "loss of self," other writers have drawn this conclusion (Jack 1993). Supporting this point of view, Kessler and McLeod (1984) report that women's psy- chologlcal well-being is more adversely af- fected than is men's when friends and rela- tives encounter problems (but see Aneshensel et al. 1991).

Structuralists argue that the different struc-

-

tural positions of adult men and women in society are associated with different oppor- tunities, constraints, and demands that influ- ence the types of relationships they have with others (Epstein 1988; Fischer and Oliker 1983; Kanter 1977). For example, the de- mands and rewards of paid work and family roles typically differ for men and women. Women are more likely to occupy subordi- nate positions in the paid workplace (Marini
AMERICAN SOCIOLOGICAL REVIEW

1988; Reskin and Hartman 1986) and in mar- riage (Bernard 1972), to have jobs in which the provision of care and concern is a pri- mary feature (e.g., nursing, social work) (Baines, Evans, and Neysmith 1991), and to shoulder primary responsibility for child- care, even if they are also employed outside the home (Nock and Kingston 1988). These structural contingencies are associated with particular types of interpersonal behavior. Subordinate positions are associated with ex- pression of feelings and self-disclosure (Kol- lock, Blumstein, and Schwartz 1985). Child- care is associated with nurturant behavior (Risman 1987), and the demands associated with child-rearing may preclude opportuni- ties to form and maintain relationships with adults (Fischer and Oliker 1983). Jobs re- quiring emotional labor and provision of care elicit nurturant behavior and require indi- viduals to make their personal needs second- ary to those of others (Baines et al. 1991; Hochschild 1983).
PREVIOUS EMPIRICAL RESEARCH

Gender Differences in Relationships

Many studies have considered how relation- ships differ for men and women. However, the research findings are often inconsistent or limited in generalizability. Some studies have concluded that women have larger so- cial networks than men (Veroff, Kulka, and Douvan 1981). However, recent research suggests that men's and women's social net- works are similar in size (Moore 1990). Other studies report that women are more likely than men to have a confidant-a person with whom they have an emotionally in- timate relationship-while men have more ties to formal associations (Booth 1972). Women report that they receive more soc~al support from coworkers, relatives, friends (Turner and Marino 1994), and adult children (Umberson 1992b) than do men. Wellman and Wortley (1990) find that women are more likely than men to provide social sup- port to both kin and nonkin. Results are in- consistent regarding social support within marriage-some suggest that women provide more social support to their spouses than do men (Vanfossen 1981), while others suggest that men and women receive similar levels of social support from their spouses (Turner and Marino 1994). Women are more likely than men to provide care to sick and elderly relatives (Gerstel and Gallagher 1993).
Relationships and Psychological Well-Being

Two separate literatures examine the effects of relationships on the psychological func- tioning of men and women: the literature on social support and integration, which empha- sizes the positive consequences of relation- ship involvement, and the literature on role strains, which emphasizes the costs.

Social support/integration. Social support (the emotionally sustaining content of rela- tionships) and social integration (the exist- ence of certain key relationships) both have been linked to mental health outcomes. The positive association of social support with well-being is supported by hundreds of stud- ies (House, Umberson, and Landis 1988). Al- though most studies include a control vari- able for sex of respondent, the possibility of gender differences in the effect of social sup- port on well-being typically is not explored (House et al. 1988). However, Turner and Marino (1994) find that social support from several relationships has similar effects on men's and women's levels of depression. Re- search on social integration and well-being has a longer tradition in sociology, but yields inconsistent results regarding gender differ- ences. Although social integration may in- here in various types of social involvements, most research focuses on marriage as a key source of social integration. Research in the 1970s suggests that the married were in bet- ter mental health than the unmarried and that this marital benefit was greater for men than for women (Gove and Tudor 1973). How- ever, recent studies question whether there is a mental health advantage associated with marriage for either men or women, at least when the comparison group is the never-mar- ried (Horwitz and White 1991; Fox 1980). The married do exhibit better mental health than the divorced or widowed, presumably because the latter groups have lost a key source of social integration (Gerstel, Reissman, and Rosenfield 1985; Umberson, Wortman, and Kessler 1992). Two longitudi- nal surveys provide inconsistent results re-

garding gender differences and divorce: Aseltine and Kessler (1993) report that women's mental health is more adversely af- fected by divorce than men's, while Booth and Amato (1991) find no gender differences in the mental health effects of divorce. Umberson et al. (1992) report that widow- hood has stronger adverse effects on depres- sion for men than for women.

Role strains. Many studies attempt to ex- plain women's higher rates of psychological distress by examining gender differences in family and job roles. Such studies often point to relational aspects of men's and women's roles as contributing to distress. Women's roles and relationships are often seen as more demanding and less rewarding than men's, therefore more conducive to depression. For example, marital relationships (Gove and Tu- dor 1973), relationships with children (Simon 1992), and relationships with im- paired family members (Gerstel and Galla- gher 1993) may be characterized by more demands and obligations for women than for men (Aneshensel and Pearlin 1987; Lennon and Rosenfield 1992). Studies often consider how role strains contribute to the gender dif- ferences in depression. Few studies consider whether equivalent levels of role strain affect men and women differently and those stud- ies provide mixed results. Pearlin and Lieber- man (1979) conclude that marital strains have stronger effects on women than on men while Barnett (1994), who restricted her analysis to dual-earner couples, finds no gen- der difference in the impact of marital role strain on distress. Young and Kahana (1989) report that women experience more distress than do men as a result of caregiving, while Gerstel and Gallagher (1993) find no gender difference. Simon (1992) concludes that pa- rental role strain affects men and women similarly.

In sum, previous findings on gender differ- ences in relationships and psychological functioning are inconsistent. One problem with summarizing these findings is that each study, even those using representative com- munity samples, focuses on a narrow range of relationship types and dimensions. As a result, generalizations about overall patterns of gender differences in relationships cannot be drawn. Furthermore, gender differences in the effects of relationships on psychological functioning may depend on how we measure psychological distress.
Gendered Expressions of Psychological Distress

Most community studies of psychological distress have emphasized measures of de- pression. These studies consistently find that women are more depressed than men (Miro- wsky and Ross 1989). However, several re- cent studies provide persuasive evidence that men and women simply express psychologi- cal distress in different ways: "Depression and some anxiety disorders [are] more com- mon among women than among men, and antisocial personality and alcohol abuse-de- pendence [are] more common among men than women" (Aneshensel et al. 1991: 168). Recent research shows that stressful life events and financial strain (Aneshensel et al. 1991), marital conflict (Horwitz and Davies 1994), and negative work conditions (Lennon 1987) are more strongly associated with sub- stance abuse for men and psychological dis- tress for women. Examining only one gen- der-typed expression of distress may bias re- sults, making it appear that one sex is more psychologically reactive than the other to their relationships. The literature on gend- ered expressions of distress emphasizes the need to include male-prevalent and female- prevalent expressions of distress when study- ing gender differences in psychological reac- tions to stress and relationship quality (Ane- shensel et al. 1991; Horwitz and Davies 1 994).
RESEARCH DESIGN

Previous research on gender differences in relationships is limited because many stud- ies rely on nonrepresentative samples or ana- lyze only one type of relationship (e.g., mari- tal relationships or friendships) or only one dimension of relationships (e.g., supportive aspects, but not strained aspects). Further- more, studies typically do not examine gen- der differences in the impact of relationships on psychological well-being, and when they do they consider only one gender-prevalent dimension of psychological functioning. We address each of these issues and extend our theoretical understanding of gendered social
AMERICAN SOCIOLOGICAL REVIEW

experiences (as expressed in the quantity and quality of relationships) and gendered ex- pressions of psychological distress (as ex- pressed in depression and alcohol consump- tion). We use a nationally representative sample that is uniquely suited to address the issue of gender differences in relationships and the impact of relationships on well-be- ing. This sample includes information on nu- merous types and aspects of social relation- ships and the respondents' psychological functioning. The few studies that have con- sidered gender differences in the effects of relationships have tended to rely on cross- sectional data. Results based on cross-sec- tional data cannot tell us whether the asso- ciation between relationships and psycho- logical distress is ephemeral or long-lasting. We examine this issue by using a two-wave panel survey to analyze gender differences in the impact of relationships on change in psy- chological functioning over time.

We address four questions raised by previ- ous theoretical and empirical work on gen- der and relationships: (1) Are there consis- tent gender differences across measures of social integration, social support, relation- ship strain, and provision of care? (2) Are women more psychologically reactive than men to the form and content of their relation- ships, and do gender differences in the psy- chological effects of relationships depend on gender-typed expressions of psychological distress? (3) Do gender differences in rela- tionships explain some of the gender differ- ences in alcohol consumption and depres- sion? (4) Does the quality of relationships have long-term effects on psychological functioning, and do such effects differ ac- cording to gender?
DATA

The data are from a national two-wave panel survey ("Americans' Changing Lives") conducted in the contiguous United States (House 1986). Face-to-face interviews were conducted with 3,617 persons ages 24 and older in 1986. In 1989, 2,867 of the original respondents were reinterviewed.' The survey

I Attrition between surveys was 21 percent (N = 750).Twenty-two percent of the attrition was due to mortality; the rest was due to nonresponse.

was designed to evaluate stress, social rela- tionships, and health over the life course. In- terviews averaged about 90 minutes in length. Our basic analyses concerning gen- der differences in relationships are based on the 1986 survey of 3,617 individuals. Analy- ses of gender differences in the effects of re- lationships on psychological functioning are based on the 1986 and 1989 surveys.
Measures

"Social relationships" is a general term com- prising three theoretically distinct dimen- sions: social integration, social networks, and relational content (House et al. 1988). Social integration refers to the existence, quantity, or frequency of specific relation- ships. Relational content refers to the quality of relationships. Social networks refer to the structures existing among a set of relation- ships (e.g., homogeneity). We focus prima- rily on measures of social integration and re- lational content in several different types of relationships.

Social integration. The measures of for- mal and informal social integration are de- rived from work by Veroff et al. (1981). Formal social integration is measured with a scale based on responses to two questions:

(1)
    "How often do you attend meetings or programs of groups, clubs, or organizations that you belong to?" and (2) "How often do you usually attend religious service^?"^ Informal social integration is measured with a scale developed from two survey questions:
(1)
    "How often do you get together with friends, neighbors, or relatives and do things like go out together or visit in each other's homes?" and (2) "In a typical week, about how many times do you talk on the telephone with friends, neighbors, or relatives?" Re-sponse categories for each social integration question include: "more than once a week," "once a week," "two or three times a month," "about once a month," "less than once a month," and "never." Advisoq integration refers to the number of individuals available for advice or help and was measured by ask- ing respondents: "About how many friends or other relatives (i.e., other than spouse, chil-

*All scales were constructed by summing item responses and converting scores to z-scores.

dren, parents) do you have whom you could call on for advice or help if you needed it?"

Relational content. Social support refers to the positive, emotionally sustaining quali- ties of relationships, whereas relationship strain refers to the negative or demanding aspects of relationships. These positive and negative qualities of relationships can have independent effects on well-being (House et al. 1988). Specific measures, based on work by House and his colleagues (House and Kahn 1985), are obtained for relationships with one's spouse, children over the age bf 16, mothers, fathers, and friends and rela- tives (other than spouse, child, or parent). The social support scales are based on two questions: "How much does your (type of re- lationship, e.g., spouse) make you feel loved and cared for?" and "How much is (helshe) willing to listen when you need to talk about your worries or problems?" The relationship strain scales are based on responses to two items: (1) "How much do you feel (helshe) makes too many demands on you?" and (2) "How much is (helshe) critical of you or what you do?" Response categories for the support and strain measures include "a great deal," "quite a bit," "some," "a little," and "not at all." Supportive ties are also evalu- ated with a confidant measure: "Is there any- one in your life with whom you can really share your very private thoughts and feel- ings" (0 = no, 1 = yes). In addition, respon- dents were asked how many confidants they have. The health regulation measure is based on research suggesting that an important function of relationships is to affect the health and health behaviors of significant others (House et al. 1988). Respondents were asked "How often does anyone tell or remind you to do anything to protect your health (of- ten, sometimes, rarely, or never)?" Scores range from 1 to 4 with high scores indicating more frequent regulation.

The provision of care to those who cannot fully care for themselves may be an impor- tant source of relationship strain (Gerstel and Gallagher 1993). We include four measures of care provision. The first measure deter- mines whether respondents provide care to others because of "physical or mental illness, disability, or for some other reason." Respon- dents were asked "Are you currently involved in helping someone like this by caring for them directly or arranging for their care by others?" (0 = no, 1 = yes). Those respondents who reported a care-provider role were asked how many hours during the previous year they spent providing care for acute and chronic health conditions. Care-provider stress was measured with a single item: "How stressful is it for you to care for (him/ her) or to arrange for (histher) care?" Re- sponse categories were "very," "quite," "somewhat," "not too," or "not at all stress- ful." The potential confounding of role strains with psychological outcomes is re- duced somewhat by excluding any language of distress in the wording of the strain ques- tion (Pearlin 1989). An exploratory analysis (not shown) indicated that inclusion of care- provider hours in the analysis reduces the es- timated effect of provider strain on distress in the present sample. This also suggests that strain reflects the nature of care provision.

Sociodemographic variables. The primary independent variable is gender (0 = male, 1 = female). Sociodemographic variables in- clude employment status (0 = not employed, 1 = employed full-time or part-time), marital status (married, divorced, widowed, and never-married-married is the excluded cat- egory), parental status (childless, children under 18 in the house, children age 18 or older in the house, children of any age living in a separate residence-childless is the ex- cluded category), income (in thousands of dollars), age (in years), education (in years), and race (1 = African American, 0 = other).

Psychological well-being. We analyze two measures of psychological functioning that have been identified as more prevalent for one gender or the other-depression and al- cohol consumption (Aneshensel et al. 1991). Depression is measured with an 1 1 -item ver- sion of the Center for Epidemiologic Studies Depression (CES-D) Scale. Respondents were asked how often they experienced each of the following feelings during the previous week: "I felt depressed," "I felt lonely," "people were unfriendly," "I enjoyed life," "I did not feel like eating, my appetite was poor," "I felt sad," "I felt that people disliked me," "I could not get going," "I felt that ev- erything I did was an effort," "my sleep was restless," and "I was happy." Response op- tions for each question were "hardly ever," "some of the time," and "most of the time."

Items were coded so that high scores indi- cate high levels of depression. Very high scores may indicate clinical depression, but as measured here, scores represent a con- tinuum of distress (Mirowsky and Ross 1989). The CES-D Scale has consistently demonstrated reliability and validity in com- munity surveys (Radloff 1977).

Alcohol consumption is typically measured by considering both frequency of drin- king and amount consumed (Berkman and Breslow 1983). However, there may be prob- lems with using total alcohol consumption as a measure of distress because longitudinal health and mortality data indicate that mod- erate drinking is associated with better health outcomes than is heavy drinking or absti- nence (Berkman and Breslow 1983). We code alcohol use as a trichotomy: nondrink- ers, moderate drinkers (fewer than 90 drinks per month), and heavy drinkers (90 drinks or more per month). These amounts are based on previous research on alcohol and health outcomes (Berkman and Breslow 1983). We calculated alcohol consumption per month by multiplying the number of days in the past month on which the respondent reported drinking by the number of drinks the respon- dent reported typically drinking on those days.

Means and standard deviations of all vari- ables are presented in Table 1, along with the t-values for the difference between the means for men and women and alpha reliability co- efficients for multi-item measures. The mean scores support much previous research indi- cating that women exhibit higher levels of depression than do men while men exhibit higher levels of alcohol consumption than do women.
RESULTS
Are There Gender Differences in Relationships?

The mean scores on relationship variables, presented separately for men and women in Table 1, provide striking evidence for gender differences in most relationship characteris- tics. Concerning general relationships, women report greater formal and informal social integration, more support from friends, and say they are more likely to have a confi-

Table 1. Means and Standard Deviations for Variables Used in the Analysis: U.S. Adults Ages 24 to 96 in 1986

T-valueMen Women

for Gender Variable Mean S.D. N Mean S.D. N Difference Alpha

General Relationships, 1986

Informal social integration -.25 1.06
Formal social integration -.06 .99
Advisory integration 9.15 8.98
Friendtrelative social support -. 19 1.07
Friendtrelative strain -.03 1.01
Has confidant .84 .37
Number of confidants 2.21 1.89
Health regulation 2.01 1.14

Family Relationships, 1986

Spouse social support
Spouse strain
Child social support
Child strain
Father social support
Father strain
Mother social support
Mother strain

Care-Giving Relationships, 1986

Care provider .15 .35 1,358 .17 .38 2,259 -1.96' NA Care-provider strain 2.29 1.22 199 2.69 1.21 386 -3.77"' N A Hours of acute care

provision 16.88 40.39 199 30.62 61.56 386 -3.23*'* N A Hours of chronic care provision 100.75 76.55 199 120.60 78.31 386 -2.95** NA

Psychological Functioning

Depression, 1986 -.03 .99 1,358 .19 1.08 2,259 -6.32*** .89 Depression, 1989 -. 10 .97 1,037 .09 1.04 1,830 -5.02"" .82 Alcohol consumption, 1986 24.95 48.96 1,358 6.39 18.56 2,259 13.41"' N A Alcohol consumption, 1989 16.77 35.32 1,037 5.15 17.94 1,830 9.90"' NA

*P < .05 **p< .01 111 p < ,001 (two-tailed tests)

dant. Men, however, report greater advisory more social support from their spouses. integration and health regulation from others. There are no gender differences in respon- The measures on which women score higher dents' reports of support from their mothers than men indicate that women are more in- and fathers. Although women's family roles volved in intimate relationships. are often portrayed in the literature as more

Concerning family relationships, gender demanding than men's, women report less differences depend on the particular relation- strain than do men in relationships with their ship examined. Compared to men, women spouses, mothers, and fathers. report more social support from their adult For cure-giving relationships, women children. Compared to women, men report score higher than men on all measures.

Women are more likely to provide care to others. Compared to men who provide care, women also spend more hours providing care and report more strain associated with their care-provider role.

Although there are statistically significant differences on most relationship measures, what does this mean in a practical sense? Some of the differences are quite small. For example, 87 percent of women and 84 per- cent of men in our sample identify someone as a confidant. Seventeen percent of women and 15 percent of men provide or arrange care for an impaired person. However, other gender differences are more striking. For ex- ample, while most men and women have a confidant, 49 percent of men, but only 20 percent of women identify their spouse as their primary confidant (p < ,002). Also, among those who provide or arrange care for others, women average 3 1 hours per year car- ing for individuals with acute care needs, while men average 17 hours per year. Fur- thermore, the care-provision measure refers to arranging or providing care to others. Among those respondents who are care pro- viders (N = 199), 83 percent of the female care providers provide direct care (or provide and arrange care), while only 71 percent of the male care providers are involved in di- rect care provision (p < .OO1).

Next we assess evidence for the structural- ist hypothesis that gender differences in re- lationships derive from the contingencies faced by men and women in their paid work and family roles. For example, regarding paid work, the parenting role may be more constraining for women than it is for men; and being a parent may interfere with women's potential for informal social integra- tion, but have little impact on men. We as- sessed this issue by regressing each of the relationship variables (one model for each relationship measure) on gender, the socio- demographic variables, and interaction terms between gender and each of the role-status variables (i.e., employment status, parental status, and marital status). Significant inter- action terms suggest that the estimated effect of gender on relationships depends on men's and women's social roles. The models re- ported in Tables 2 through 4 are the final models; they do not include interaction terms when, in the base models, there were no sig-
AMERICAN SOCIOLOGICAL REVIEW

nificant interactions @ < .05), or in the final models, interactions were not statistically significant. The coefficient for gender from the base model (without interactions) is also provided for those models that included in- teraction terms.

General relationships. Paralleling the gender differences reported in Table 1, Table 2 shows that women score higher than men on informal social integration, formal social integration, and social support from friends and relatives; women score lower than men on advisory integration. Gender differences in health regulation and having a confidant depend on men's and women's family roles. For example, having an adult child in the home or a child living outside the home significantly increases men's probability of having a confidant while the effect on women is not ~ignificant.~

Perhaps men face fewer of the social constraints as- sociated with parenting, and are less likely than women to have confidants outside of family, leading them to rely more on adult children for their close relationships. The estimated effect of gender on health regula- tion depends on marital status: Divorced and never-married men are significantly less likely than their female counterparts to re- port that others attempt to regulate their health. Among the married, however, men are much more likely than women to report that others attempt to regulate their health. It is not surprising that married men experi- ence more health regulation than do married women (or other men) because traditional gender roles typically encourage women to monitor the health of husbands (Umberson 1992a).

Family relationships. Table 3 presents co- efficients predicting family relationships.

We calculated the effects of parental status separately for men and women. The effects of pa- rental status for men are the main effects of pa- rental status in Table 2 (-,035, .796*, and ,478' for having a minor in the home, having an adult child in the home, having a child outside the home). The effects of parental status for women are equal to the sum of the main effect and the corresponding interaction (-,432, -.373, ,136). The standard errors for these effects are also cal- culated (.232, ,225, ,202). The detailed calcula- tion for these and other effects involved in inter- actions is available from the authors.
Selected Independent Variables: U. S. Adults Ages 24 to 96 in 1986

General Relationships
Independent Variable     Informal Social Integration     Formal Social Integration     Wend/ Relative Social Support     Friend1 Relative Strain     Number Has of Confidant Confidants     Advisory Integration     Health Regulation
Gender (Female)' Base model (without interactions)     ,318"'     ,202"'     ,385"'     -.020     ,042"     ,070     -.684*     -. 139"'
Final model (with interactions)     -    -    -    -    ,791"'     -    -    -.280***
Sociodemographic Variables Marital Statusb Divorcedlseparated Widowed     ,098' ,296"'     -.278**' -.098* -.083 ,021     ,253"'. -.034     -.679***-. 130 -1.297" ,540"' -.I66 -.500     -.294** -.I95
Never-married     -.083     -.273*** -.078     ,156'     -1.1 19"'     -.267*     -1.231'     -.406***
Parental StatusC
Child < 18at home
                       
Child 218 at home
                       
Child away, any age
African American
                       
Employed
Family income
Years of education                         
Age                     
Interactions Child < 18 x female Child 2 18 x female     --    --    --    --    -.397 -1.169"     --    --    --
Child away x female Divorced x female Widowed x female Never-married x female     ----    ----    ----    ----    -.342 ---    ----    ----    -,387"' ,222 ,468"'
Constant Rzllog-likelihood     -.272* ,100     -1.273"' ,071     -.670*** ,786"' ,049     .lo6     2.035"' -1,417     1.875"' ,024     8.206"' ,031     2.715"' ,018

Note: OLS regression was used for continuous dependent variables and RZ is reported; logistic regression was used for dichotomous dependent variables and log-likelihood is reported. Standard errors are available from the authors. N = 3,617.

a "Gender" coefficients for base and final models are provided for comparison if the base model is not the final model.

"Married" is the omitted category.

"Childless" is the omitted category.

** ***

'p S .05 p 5.01 p S ,001(two-tailed tests)

Compared to men, women report lower lev- children aged 16 and older occurs primarily els of strain with fathers and spouses and because being divorced lowers the receipt of lower levels of social support from spouses. social support from older children much An interaction between gender and being di- more for men than for women. This may oc- vorced when predicting social support from cur because divorce is more likely to result

Table 3. Unstandardized Coefficients from the Regression of Family Relationships on Gender and Selected Independent Variables: U. S. Adults Ages 24 to 96 in 1986

Family Relationships

Father Mother Spouse Child Independent Social Father Social Mother Social Spouse Social Child Variable Support Strain Support Strain Support Strain Support Strain

Ger~der (Female)"

Base model
(withoutinteractions) -.010 -.195*** ,021 -.075 -.424"* -.130** ,323"' ,004
Final model
(with interactions) ------,246"' -

Sociodemograplzic Variables

Marital Statush
Divorcedlseparated -.226" ,092 -. 118 ,199" -.074 ,093 -.726*** ,177"
Widowed ,069 ,056 ,059 ,051 ,323 ,264 ,126 -.004
Never-married -.021 ,075 -.lo4 ,156' -.021 ,031 -.598 -.080

Parental StatusL
Child < 18 at home
Ch~ld2 18 at home
Child away, any age
African American
Employed
Family ~ncome
Years of educat~on

Age

Itlternctior~s

---.558*** -

Divorced x female --

Widowed x female ------. 150

Never-married x female ----,539

Constant -.I 17 ,716" ,148 ,776"' ,719"' -. 187 -.814"* ,587"' R2 ,019 ,068 ,016 ,080 ,054 ,044 ,094 ,132 Number of cases 1,231 1,231 1,885 1,885 2,075 2,075 2,337 2,337

Note: OLS regression was used for continuous dependent variables. Logistic regression was used for dichoto- mous dependent variables. Standard errors are available from the authors.

"'Gender" coefficients for base and final models are provided for comparison if the base model is not the final model.

"Married" is the omitted category.

"Childless" is the omitted category.

I* ***

*p 5 .05 p 5.01 p < ,001 (two-tailed tests)

in less contact and lower-quality relation- care-provider role, regardless of family and
ships with children for fathers than for moth- work roles.
ers (Aauilino 1994; Cooney 1994).

,

.
Care-provider relationships. Consider-Do Relationships Affect the Psychological

ation of sociodemographic variables does not
Functioning of Men and Women

alter the basic results for care-provider rela-
Dqferently ?

tionships reported in Table 1. Table 4 shows that women are more likely than men to be We test whether relationships affect the psy- care providers, to spend more hours provid- chological functioning of men and women in ing care, and to experience more strain in the different ways or to different degrees. We

Table 4. Unstandardized Coefficients from the Regression of Care-Giving Relationships on Gender and Selected Independent Variables: U. S. Adults Ages 24 to 96 in 1986

Independent Variable

Gender (Female)

Base model (without interactions)

Sociodemographic Variables

Marital Status" Divorcedtseparated Widowed

Parental Statusc Child < 18 at home Child 2 18 at home Child away, any age African American Employed Family income Years of education Age

Constant R2/log-likelihood Number of cases
Care-Provider Relationships
Care     Care-    Hours Providing     Hours Providing
Provider     Provider Strain     Chronic Care     Acute Care
,235"     ,443"'     6.80"'     8.615"'
-.312*     -. 107     -4.974     7.246"
-.766**'     -.236     -13.069"'     1.546
-.055     ,011     -6.990'     2.069
,276     -.234     3.303     3.233
,354'     ,092     3.796     4.947
,212'     -. 184     5.100"     -3.174
-.017     -.026     -.953     -.999
-.006*     ,001     -.043     ,075
,037'     ,024         
,000     ,002         
2.125"'     1.934"'     11.651     4.756
-1,573     ,057     .0 17     ,026
3,617     585     3,617     3,617

Note: OLS regression was used for continuous dependent variables and R2 is reported; logistic regression was used for dichotomous dependent variables and log-likelihood is reported. Standard errors are available from the authors.

"Married" is the omitted category. "Childless" is the omitted category.

** ***

'p 5 .05 p < .01 p < ,001 (two-tailed tests)

begin by estimating the effects of relation- ships on depression and alcohol use using the 1986 cross-sectional survey data. This analy- sis examines the association between current relationship involvement and current psycho- logical functioning. We tested a series of in- termediate models to derive the final model for each of the dependent variables, depres- sion and alcohol use. We began by estimat- ing separate models for each possible inter- action between gender and a relationship measure when predicting depression and al- cohol use. We regressed depression scores and alcohol use on gender, the sociodemo- graphic variables, all relationship variables, and an interaction term for gender with one of the relationship variables. If more than one interaction was significant when predict- ing a dependent variable, we estimated addi- tional models that included all statistically significant interaction terms to determine if a particular interaction was a predominant source of the effect. Variables indicating missing data for were also included. Table 5 presents two models for depression, non- drinker, and heavy drinker. In each case, Model 1 includes the sociodemographic vari- ables, gender, and the relationship measures; Model 2 is a restricted final model, which in- cludes gender, all significant sociodemo- graphic variables, all significant relationship measures, significant interaction terms, and the corresponding lower-order relationship measures involved in interaction^.^

General relationships. For depression, the restricted model (Model 2 in Table 5) shows that six of the eight general relationship mea- sures are significantly associated with depres- sion: informal social integration, formal so- cial integration, social support from friends and relatives, and advisory integration are associated with lower levels of depression, while health regulation and friendtrelative relationship strain are associated with el- evated levels of depression. The integration, support, and strain measures are associated with depression in the expected direction. Perhaps the health regulation measure is posi- tively associated with depression because ef- forts from others to regulate health behavior are perceived as critical and demanding or because poor health may be associated with depression and elicit regulation efforts from others (Umberson 1992b). None of the inter- actions between gender and the general rela- tionship measures are statistically significant, suggesting that these general aspects of rela- tionships affect men's and women's levels of depression in similar ways.

Few of the general relationship measures are associated with alcohol use. Formal so- cial integration increases the probability of being a nondrinker (compared to being a moderate drinker) and decreases the prob- ability of being a heavy drinker (compared to being a moderate drinker). This may oc- cur because the measure of formal social in- tegration includes participation in religious organizations, which often precludes alco- hol use. Strain with friends and relatives de- creases the probability of being a nondrinker (as opposed to being a moderate drinker), but does not significantly increase the probability of being a heavy drinker. An interaction between gender and informal so- cial integration indicates that informal social integration significantly decreases the probability of being a nondrinker (compared to being a moderate drinker) for women but not for men.

Family relationships. Social support from one's spouse and adult children is inversely associated with depression, while strain in

"esults for the intermediate models are avail- able from the authors.

relationships with one's spouse, mother, and father is positively associated with depres- sion (Table 5, Model 2 for depression). These family relationship measures have similar ef- fects on depression for men and women. An interaction between gender and child rela-

-

tionship strain remains in the final model for depression: Relationship strain with children over age 16 significantly contributes to de- pression for women but not for men.

For alcohol use, two significant interac- tions between gender and spouse relationship measures suggest that spouse relationship quality is associated with alcohol use in dif- ferent ways for men and women. Social sup- port from one's spouse significantly de- creases women's, probability of being heavy drinkers (compared to being a moderate drinker), but has no significant effect for men. Spouse relationship strain and gender significantly interact when predicting being a nondrinker but the effect of spouse rela- tionship strain on alcohol use is not statisti- cally significant for either gender.

Previous research has suggested that mari- tal status per se may be an important indica- tor of social integration that affects men and women differently (Gove and Tudor 1973). To explore this possibility, we estimated models (not shown) that included an interac- tion between gender and marital status when

-

predicting psychological functioning. The re- sults indicated that the effect of marital status on depression and the probability of being a nondrinker is similar for men and women. However, the analysis revealed a significant interaction between marital status and gender

-

when predicting being a heavy drinker: Being divorced increases the probability of being a heavy drinker for men and women, but this effect is significant only for women. Divorce and heavy drinking may be more strongly as- sociated for women because heavy drinking may be more likely to lead to divorce for women than for men (Mulford 1977).

Care-provider relationships. Hours spent providing care for acute conditions and care- provider strain are positively associated with depression, and these estimated effects are similar for men and women. None of the care-provider measures or interactions be- tween gender and these measures attained

-

statistical significance when predicting alco- hol use (see Table 5).

Table 5. Unstandardized Coefficients from the Regression of Depression, Nondrinker, and Heavy Drinker on Selected Independent Variables: U.S. Adults Ages 24 to 96 in 1986
Depression
Independent Variable Model 1     Model 2
Gender (female)         
Sociodetnographic Variables         
Marital Status"
       
DivorcedISeparated
       
Widowed
       
Never-married
       
Parental Statusb
       
Child < 18 at home
       
Child 2 18 at home
       
Child away, any age
       
African American
       
Employed
       
Family income
       
Years of education
       
Age
       
General Re(at~onships         
Informal social integration         
Formal social integration         
Friendlrelative support         
Friendlrelative strain         
Has confidant         
Number of confidants         
Number of advisors         
Advisory integration         
Health regulation         
Do Gender Differences in Relationships Explain Gender Differences in Psychological Functioning?

Table 5 shows that, even with controls for all of the relationship measures, there are sig- nificant gender differences in depression and alcohol consumption. This suggests that tak- ing into account the nature of relationships does not explain the higher levels of depres- sion among women or the higher levels of alcohol consumption among men. We com- pared the results in Table 5 to results from our basic models (not shown), in which we regressed the dependent variables only on the sociodemographic variables. A comparison of the coefficient for gender in the basic models and the final models in Table 5 suggests how

Nondrinker Heavy Drinker
Model I Model 2 Model 1 Model 2

(Table 5 contlt~uedon next page)

inclusion of the relationship measures affects the coefficient for gender when predicting depression and alcohol consumption. The co- efficient for gender actually increases by 33 percent (from ,123 to ,183) when predicting depre~sion.~

This suggests that women's greater involvement in positive relationships may protect them from depression and that if we were able to equalize relationships for men and women, the gender gap in depres- sion would actually increase (also see Turner and Marino 1994). The coefficients for gen-

The coefficient of ,183 is from the model that includes the gender x child strain interaction; therefore the coefficient is valid when child strain equals 0. Because the overall mean for child strain in the sample is 0, the coefficient is reason- ably applied to the sample as a whole.

(Table 5 continued from previous page)
Depression     Nondrinker     Heavy Drinker
Independent Variable     Model 1     Model 2     Model 1     Model 2     Model 1     Model 2
Family Relationships             
Spouse social support     -.135***     -.139***     ,060     .061     -.009     ,144
Spouse strain     ,061"     ,065"     ,002     -.I39     ,171     ,134
Child social support     -.075**'     -.082***     -.047     -    -. 184     -
Child strain     ,042     -.035     -.021     -    -. 106     -
Father social support     ,028     -    ,091     -    -.032     -
Father strain     ,079"     .072**     ,106     -    .05 1     -
Mother social support     -.029     -    -.060     -    ,180     -
Mother strain     ,055'     ,067"     -.085     -    ,104     -
Care-Giving Relationships             
Care provider     -.I27     -.I74     -.I51     -    ,501     -
Provider strain     ,097"     ,042"     ,003     -    ,117     -
Hours of acute care     ,001'     ,001     -.001     -    -.001     -
Hours of chronic care     -.000     -    ,002     -    -.004     -
Interactions             
Child strain x female     -    .I 12"     -    -    -    -
Spouse support x female     -    -    -    -.020     -    -.703*
Spouse strain x female     -    -    -    ,228'     -    -.010
Informal social integration x female     -    -    -    -. 172'     -    -.358
Data Presence IndicatorsC             
Relationships with spouse     ,339"'     ,300"'     -.030     ,004     -.611     -.409
Relationships with children     -.010     ,014     -.029     -    -1.3 18"'     -
Relationships with mother     ,037     ,037     ,103     -    ,327     -
Relationships with father     -.074     -.063     -. 144     -    -. 144     -
Constant     ,604"'     ,599"'     .577     ,377     1.297     -.891
~~Ilog-likelihood     ,245     ,244     -2,467     -2,467     -2,479     -2,479

Arore: OLS regression was used to assess the depression variable and R2 is reported. Multinomial logit models (LIMDEP) were used to assess the trichtomous alcohol use outcome and log-likelihood is reported. Number of cases for each alcohol category is: 1837 for nondrinker, 130 for heavy drinker, and 1650 for moderate drinker. Standard errors are available from the authors. N = 3,617.

"'Married" is the omitted category.
"Childless" is the omitted category.
"Nonmissing" is the omitted category.

*p< .05 **p5 .Ol ***p < .001 (two-tailed tests)

der when predicting alcohol use are largely psychological functioning. However, the unchanged by inclusion of the relationship quality of relationships may have more last- measures, suggesting that involvement in re- ing effects on psychological functioning. For lationships does not influence the observed example, spouse relationship strain in one gender difference in alcohol use. year may be associated with a decline in psy- chological well-being over the next several years, and this effect may differ for men and

Relationship Change and Psychological

women. We assess the long-term effects of
Functioning

relationships on psychological functioning by To this point, we have considered how ongo- regressing the 1989values for depression and ing relationships are associated with current alcohol consumption on gender, the 1989values of the sociodemographic variables, mea- sure of change in relationships (1989 value minus the 1986 value), the 1989 values for the relationship measures, and the 1986 value for the de~endent variable. We include the 1989 value's for the relationship measures to control for the effects of current relationship status on psychological functioning. The re- lationship change measures are included to assess the lasting impact of 1986 relation- ships on psychological functioning in 1989. Alcohol consumption is measured as a con- tinuous variable for the analysis reported in Table 6 because we are predicting the impact of relationships on a subsequent increase or decrease in alcohol consumption. We fol- lowed the same estimation procedures de- scribed for the analysis presented in Table 5 to derive final models that also include sig- nificant interactions between gender and the relationship variables. Results are presented in Table 6. Again, two models are presented for each dependent variable: Model 1 shows the results without interaction effects; Model 2 shows only significant effects from Model 1, including significant interaction effect^.^

Table 6 shows that increases in formal so- cial integration and advisory integration are associated with reduced depression over time. Increased friendlrelative support is associated with reduced alcohol consumption over time. These beneficial long-term effects of rela- tionships are similar for men and women. Hours of acute care provision are positively associated with alcohol consum~tion over time. None of the other relationship change measures is associated with psychological functioning in 1989 for either men or women.

The cross-sectional associations between 1989 relationships and psychological func- tioning in 1989 largely parallel the 1986 cross-sectional results. Two significant inter- actions between 1989 relationships and gen- der occur in the final models of Table 6. The gender x child social support interaction in- dicates that social support from children is associated with lower levels of depression for women, but not for men. Having a confidant

Nonsignificant variables may be presented in Table 6 because a nonsignificant variable is in- volved in a significant interaction term or, in the case of models predicting heavy drinker or non- drinker, a variable was significant in predicting the other drinking category.

is associated with significantly higher levels of alcohol consumption for men but has little effect on women's alcohol consumption. However, high levels of alcohol consumption do not necessarily indicate heavy drinking, merely higher consumption levels than for those men who do not have a confidant.

Causal order. Longitudinal studies provide persuasive evidence that social relationships have a causal effect on subsequent well-be- ing (House et al. 1988). However, serious depression and alcohol problems undoubt- edly affect the availability and quality of re- lationships.

To sort out the nature of this reciprocity is beyond the scope of this study. We begin to address issues of causality by examining the impact of relationships on subsequent levels of depression and alcohol consumption. However, causal order cannot be established by longitudinal data over a three-year period. Relationships change over time and are prob- ably reciprocally influenced by psychologi- cal state-a complex association that requires a much longer longitudinal profile to untangle.
DISCUSSION AND CONCLUSION
Are There Gender Differences in Relationships ?

Clearly, men and women have different rela- tionships. Women score higher than men on measures indicating more intimate and infor- mal social ties, such as having a confidant, receiving social support from friends and family, and visiting with friends. Women also are more likely to provide care to impaired individuals. Although men report less social support from most relationships (except in marriage), they report having more people to call on for advice or help, and they are more likely to have someone attempt to regulate their health. We conclude that men receive more of the instrumental aspects of relation- ships (e.g., practical, tangible aspects) while women receive more of the intimate, interac- tive aspects of relationships. Women also provide instrumental aspects of relationships as suggested by their higher levels of care- giving. Most research on relationships has focused on the emotional functions of rela- tionships, such as social support (e.g., feel-

Table 6. Unstandardized Coefficients from the Regression of Depression and Alcohol Consumption in 1989 on Selected Independent Variables: U.S. Adults Ages 24 to 96 in 1986

Depression, 1989 Alcohol Consumption, 1989

Independent Variable Model 1

Gender (female) ,029

Sociodemographic Variables, 1989

Marital Status"
Divorcedlseparated -.043
Widowed ,038
Never-married .03 1

Parental statusb
Child < 18 at home ,068
Child 2 18 at home ,068

Child away, any age ,067 Employed -. 182"' African American ,069 Family income -.003*** Years of education -.O 19"'

-.003

Age

Relationship Chunge, 1986-1 989

Formal social integration Advisory integration -
Friendlrelative support Hours of acute care provision

General Relationships, 1989

Informal social integration -.047**
Formal social lntegratlon -,058***
Friendlrelative social support -.05 1
Friendlrelative strain .I 15***
Has confidant ,004
Number of confidants -.012
Instumental integration -.003
Health regulation, 1986 ,055"'

ing cared for and listened to). Future re- search on gender and relationships should in- clude measures of the receipt and provision of instrumental and emotional aspects of re- lationships. Finally, marital, parental, and employment status usually have similar ef- fects on men's and women's relationships.
Are Women More Psychologically Reactive than Men to Their Relationships?

Our second major finding is that there are few gender differences in the effects of rela- tionships on psychological functioning. Con-
Model 2     Model I     Model 2
,028     -4.749"'     .I 12
-
       
-
       
-
       
-
       
-
       
-
       
-. 145"'
       
-
       
-.003***
       
-.018***
       
-
       
-.062**
       
-.006
       
-
       
-
       

(Tuble 6 tontlnued on next page)

trary to much sociological theory, women do not seem to be more psychologically sensi- tive than men to the circumstances and qual- ity of their relationships. When men and women are involved in similar relationships, those relationships affect levels of depression and alcohol use of men and women in strik- ingly similar ways. This finding supports the structuralist position that similar social con- ditions elicit similar psychological reactions from individuals.

Previous studies of the effects of social re- lationships have focused almost exclusively on symptoms of depression. However, be-

(Table 6 c.onrinueii from previous pujie)

Depression, 1989 Alcohol Consumption, 1989

Independent Variable
Faniily relnrionshi[~r, 1989

Spouse soc~al support
Spouse \train
Child social support
Child strain
Father social support
Father strain
Mother social support
Mother \train
Care-Giving Relaliot~shi[~s.

1989

Care provider
Care-provider strain

Hours of acute care provision Hours of chronic care provision
/nferactions

Female x child social
support, 1989
Female x confidant, 1989
Vcilue of depenilenr vnrictble, 1986

Data presence indicators, 1989' Spouse relationship Child relationship Father relationship Mother relationship

Constant R2 Number of cases

Model I Model 2 Model I Model 2

-. 152 ,094"' ,000 ,000

-

-

,346"'

,201* -.O 17 -.O 13 ,023

.3 19 ,375 2,867

-.202** 094"'

-

-

-. 1 17"

-

,360"' ,215"' -.001

-

-

,151' ,377 2,867

2.862 -.737 -.003 .021 * -.008

--

--6.1 15'

,404"' ,407"'

.-,435 -.33 1 .421

    5.589'

,286 ,287 2,867 2,867

h'ore: Standard errors are ava~lable from the authors.
"%larr~cd" is the omitted category.
"Childless" is the omitted category.
'"Nonmissing" is the omitted category.

***

*P < .05 **[I 5 .ol [J 5 ,001 (two-tailed tests)

cause depression is a female-prevalent mea- sure of psychological distress, we empha- sized the need to consider a male-prevalent expression of distress-alcohol use. Previous research suggests that women are more emo- tionally reactive to relationships when emo- tional reactivity is measured in terms of de- pression (Aneshensel et al. 1991; Horwitz and Davies 1994). We find evidence for this only in relationships with children over age 16.The quality of relationships with children in terms of support and strain has more im- pact on depression among women than it does among men. This supports the hypothesis of women's greater sensitivity to strained rela- tionships and corresponds to previous research suggesting that the parental role is more salient to the psychological well-being of women than of men (Scott and Alwin 1989; but see Simon 1992). However, we emphasize that men's and women's depres- sion levels respond similarly to most of the relationship measures considered in this study.

Although previous research on gender dif- ferences in psychological reactivity suggests that men are more sensitive to relationships when alcohol consumption is the indicator of psychological distress, this appeared to be the case for only two relationship measures: Informal social integration decreases men's probability of being nondrinkers (but not women's), and having a confidant is posi- tively associated with alcohol consumption for men (but not women). Perhaps social contexts are more likely to facilitate drink- ing for men than for women. In several cases, relationship status affected women's alcohol use more strongly than men's: Social support from one's spouse and simply being married seemed to have more impact on women's al- cohol use than it did on men's. In sum, we find little evidence that men's alcohol con- sumption levels are more responsive to rela- tionship quality or that women's levels of depression are more responsive to relation- ship quality.
Do Relationships Explain the Gender Differences in Depression and Alcohol Consumption?

We emphasize the lack of gender differences in the consequences of relationships for de- pression and alcohol consumption: All things being equal, supportive relationships seem to alleviate distress, and strained relationships seem to increase distress for women and men. But all things are not equal: We find clear gender differences in the quantity and quality of relationships and these differences influence men's and women's levels of psy- chological distress. For example, even if men and women are equally affected by support- ive relationships with friends, women benefit more psychologically because the sheer amount of social support women receive from friends is higher than the amount that men receive from friends. Women's support- ive relationships mitigate their distress, but women begin with much higher rates of de- pression than men. If women did not have the higher levels of support and integration that they have relative to men, they would exhibit even higher levels of depression relative to men than they currently do. Thus, women's greater involvement in certain types of rela- tionships-especially the positive aspects of those relationships-protects them psycho- logically (also see Turner and Marino 1994). Mirowsky and Ross (1989) emphasize the social causes of psychological distress when explaining the high rates of depression among women. Relationships represent a social alleviation of psychological distress that affects the gender difference in depression in the opposite direction. In contrast, relation- ships have less influence on alcohol consumption than on depression, and relation- ships do not explain any of the gender differ- ence in alcohol consumption.
Are There Gender Differences in the Long-Term Effects of Relationships?

The impact of ongoing relationships on cur- rent levels of psychological distress seems to be much stronger than the effects of relation- ships at one point in time on distress levels three years later. This suggests that ongoing relationships are usually more important to current psychological functioning than are past relationships, and that most relationship measures do not have long-term effects on psychological functioning. Of course, these results do not demonstrate that relationships have no long-term psychological consequences--only that the particular dimensions of relationships measured here are not sig- nificantly associated with subsequent psy- chological functioning. Although we control for current relationships when predicting the impact of past relationships on current psy- chological functioning, current relationships, in part, probably reflect past relationships.

When there are long-term effects, we find that two of the social integration items are associated with subsequent depression levels. Furthermore, when relationship quality has long-term effects, it is for alcohol consump- tion rather than for depression. Most previ- ous research has focused on relationship quality (especially social support) rather than on social integration in predicting depres- sion. Future research should consider both social integration and relationship quality as important aspects of relationships that may affect psychological distress. Also, alcohol consumption should be considered in addi- tion to depression as an indicator of distress. Finally, we find no gender differences in the long-term consequences of relationships.

Theory on Gender, Relationships, and
Psychological Functioning

Sociological theory provides reasons to ex- pect women to be more emotionally respon- sive than men to social relationships (Cho- dorow 1978; Gilligan 1982). We find little evidence that women are more emotionally sensitive than men to the quality of their re- lationships. However, we find that women's greater involvement in positive relationships relative to men's is partly responsible for keeping women's depression levels as low as they are. That is, without these positive rela- tionships, women would be more depressed relative to men than they currently are. Re- sults strongly suggest that social structural position (as indicated by gender and other sociodemographic variables) influences the form and content of relationships. However, social structural position as defined by gen- der does not appear to make women more

. .

sensitive than men to their relationships: In similar relationship conditions, men and women are affected similarly by those rela- tionships. Future theoretical work on gender, relationships, and psychological functioning must reconcile gender difference in the quan- tity and quality of relationships with gender similarity in psychological responses to the quantity and quality of relationships.

The quest to identify gender differences may sometimes impair our ability to accu- rately understand social phenomena: "The extent to which people attribute qualities and capacities to the two sexes is an example of how the concepts "male" and "female" cause the sorting and skewing of perceptions of re- ality by a focus on differences rather than similarities" (Epstein 1988:12). We find that men and women are similar in their psycho- logical reactions to the nature and quality of their relationships, a finding that contradicts much sociological theory and suggests that much previous research on gender and rela- tionships may have overemphasized gender differences. However, we find substantial gender differences in the form and content of relationships, which suggests that men and women have different relationship experi- ences. It is important to recognize gender differences arid gender similarities in order to integrate theory and research into a coher- ent picture of "gendered" reality.

Debra Umberson is Associate Professor of Soci- ology at the University of Texas, Austin. Her re- search focuses on structural determinants of psy- chological and physical health, genrler and rela- tionships, and bereavertlent anrl the family. She is now writing a book on the impact of a pnrent's death on adult children andfamilies based on re- search supported by a FlRSTAwarrlfrotn the Nci- tional lnslitute on Aging. She is also conducting research on genrler and domestic violence.

Meichu D. Chen is Research Affiliate wllh the Population Research Center of the University of Texas, Austin. Her interests include social strati- fication, demography, and quantitative methods.

James S. House is Director of the Survey Re- search Center anrl Professor of Sociology at the University of Michigan, Ann Arbor. His current research focuses on the role ofpsychosocialfac- tors, especially socioeconoinic status, in the maintenance of health and effective functioning over the life course, including the reciprocal re- lationship between health and paid work or other productive activity.

Kristine Hopkins is a Ph. D. candidate in Sociol- ogy at the University of Texas, Austin. Her dis- sertation research focuses on reproductive deci- sion-making in Brazil and explores wonlen's atti- tudes and experiences with surgical birth and contraception.

Ellen Slaten received her Ph. D. fronz the Univer- sity of Texas, Austin in 4'996 and is now Assistant Professor of Sociology at the University of Texas Health Science Center in San Antonio. Her research focuses on AIDS caregivers.
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. Barnett, L. Biener, and G. K. Baruch. New York: Free Press.

Aneshensel, Carol S., Carolyn M. Rutter, and Pe- ter A. Lachenbruch. 1991. "Social Structure, Stress, and Mental Health." American Socio- logical Review 56: 166-78.

Aquilino, William S. 1994. "Impact of Childhood Family Disruption on Young Adults' Relation- ships with Parents." Journal of Marriage and the Family 56:295-3 13.

Aseltine, Robert H. and Ronald C. Kessler. 1993. "Marital Disruption and Depression in a Com- munity Sample." Journal of Health and Social Behavior 34:237-5 I.

Baines, Carol, Patricia Evans, and Sheila Neysmith, eds. 1991. Wornen's Caring: Femi- nist Perspectives on Social Welfare. Toronto, Ontario: McClelland and Steward.

Barnett, Rosalind. 1994. "Home-to-Work Spillover Revisited: A Study of Full-Time Em- ployed Women in Dual-Earner Couples." Journal of Marriage and the Farnily 56:647-56.

Berkman, Lisa F. and Lester Breslow. 1983.

Health and Ways of Living: The Alanzeda County Study. New York: Oxford University Press.

Bernard, Jessie. 1972. The Future of Marriage. New Haven, CT: Yale University Press. Booth, Alan. 1972. "Sex and Social Participa- tion." Arnerican Sociological Review 37: 183-

92.

Booth, Alan and Paul R. Amato. 1991. "Divorce and Psychological Stress." Journal of Health and Social Behavior 32:396-407.

Cooney, Teresa M. 1994. "Young Adults' Rela- tions with Parents: The Influence of Recent Pa- rental Divorce." Journal of Marriage and tlie Farnily 56:45-56.

Chodorow, Nancy. 1978. The Reproduction of Mothering. Berkeley, CA: University of Cali- fornia Press.

Epstein, Cynthia Fuchs. 1988. Deceptive Distinc- tions: Sex, Gender, and the Social Order. New Haven, CT: Yale University Press and New York: Russell Sage Foundation.

Fischer, Claude S. and Stacey J. Oliker. 1983. "A Research Note on Friendship, Gender, and the Life Cycle." Social Forces 62(1): 124-33.

Fox, John. 1980. "Gove's Specific Sex-Role Theory of Mental Illness: A Research Note." Journal of Health and Social Behavior 21 :260-

67.

Gerstel, Naomi and Sally K. Gallagher. 1993. "Kinkeeping and Distress: Recipients of Care, and Work-Family Conflict." Journal of Mar- riage and the Family 55:598-607.

Gerstel, Naomi, Catherine Kohler Reissman, and Sarah Rosenfield. 1985. "Explaining the Symp- tomatology of Separated and Divorced Women and Men: The Role of Material Conditions and Social Networks." Social Forces 64:84-101.

Gilligan, Carol. 1982. In a Different Voice. Cambridge, MA: Harvard University Press.

Gove, Walter R. and Jeannette F. Tudor. 1973. "Adult Sex Roles and Mental Illness." Arnerican Journal of Sociology 78: 1-24. Hochschild, Arlie. 1983. The Managed Heart. Berkeley, CA: University of California Press.

Horwitz, Allan V. and Lorraine Davies. 1994. "Are Emotional Distress and Alcohol Problems Differential Outcomes to Stress? An Explor- atory Test." Social Science Quarterly 75:607

21.

Horwitz, Allan V, and Helene R. White. 1991. "Becoming Married, Depression, and Alcohol Problems among Young Adults." Journal of Health and Social Behavior 32:221-37.

House, James S. 1986. "Americans' Changing
AMERICAN SOCIOLOGICAL REVIEW

Lives: Wave I" [MRDF]. Ann Arbor, MI: Sur- vey Research Center [producer]. Ann Arbor, MI: Inter-University Consortium for Political and Social Research [distributor].

House, James S. and Robert L. Kahn. 1985. "Measures and Concepts of Social Support." Pp. 83-108 in Social Support and Health, edited by S. Cohen and S. L. Syme. New York: Academic Press.

House, James S., Debra Umberson, and Karl Landis. 1988. "Structures and Processes of So- cial Support." Annual Review of Sociology 14:293-318.

Jack, Dana Crowley. 1993. Silencing tlie Self: Wornen and Depression. New York: Harper Collins.

Kanter, Rosabeth M. 1977. Men and Women of the Corporation. New York: Basic Books.

Kessler, Ronald C. and Jane C. McLeod. 1984. "Sex Differences in Vulnerability to Undesir- able Life Events." American Sociological Re- view 49:620-3 1.

Kollock, Peter, Philip Blumstein, and Pepper Schwartz. 1985. "Sex and Power in Interaction: Conversational Privileges and Duties." American Sociological Review 50:34-46.

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

Lennon, Mary Clare and Sarah Rosenfield. 1992. "Women and Mental Health: The Interaction of Job and Family Conditions." Journal of Health and Social Behavior 33:3 16-27.

Marini, Margaret Mooney. 1988. "Sociology of Gender." Pp. 374-93 in The Future of Sociol- ogy, edited by E. F. Borgatta and K. S. Cook. Beverly Hills, CA: Sage.

Mirowsky, John and Catherine E. Ross. 1989. Social Causes of Psyclzological Distress. New York: Aldine de Gruyter.

Moore, Gwen. 1990. "Structural Determinants of Men's and Women's Personal Networks." American Sociological Review 55:726-35.

Mulford, Harold A. 1977. "Women and Men Problem Drinkers: Sex Differences in Patients Served by Iowa's Community Alcoholism Cen- ters." Joiirnal of Studies on Alcoliol 38:1624

39.

Nock, Steven L. and Paul W. Kingston. 1988. "Time with Children: The Impact of Couples' Work-Time Commitments." Social Forces 6759-85.

Parsons, Talcott. 1955. "Family Structures and the Socialization of the Child." Pp. 35-131 in Family, Socialization and Interaction Process, edited by T. Parsons and R. Bales. Glencoe, IL: Free Press.

Pearlin, Leonard I. 1989. "The Sociological Study of Stress." Journal of Health and Social Behav-

ior 30:241-56.

Pearlin, Leonard I. and Morton A. Lieberman. 1979. "Social Sources of Emotional Distress." Pp. 217-48 in Research in Community and Mental Health, edited by R. Simmons. Green- wich, CT: JAI Press.

Radloff, Lenore S. 1977. "The CES-D Scale: A Self-Report Depression Scale for Research in the General Population." Applied Psychologi- cal Measurenient 1 :385-401.

Reskin, Barbara F. and Heidi I. Hartman, eds. 1986. Women's Work, Men's Work: Gender Segregation on the Job. Washington, DC: Na- tional Academy Press.

Risman, Barbara J. 1987. "Intimate Relationships from a Microstructural Perspective: Men Who Mother." Gender and Society 1:6-32.

Scott, Jacqueline and Duane Alwin. 1989. "Gen- der Differences in Parental Strain: Parental Role or Gender Role?" Journal of Farnily Is- sues 10:482-503.

Simon, Robin W. 1992. "Parental Role Strains, Salience of Parental Identity, and Gender Dif- ferences in Psychological Distress." Journal of Health and Social Behavior 33:25-35.

Tannen, Deborah. 1991. You Just Don't Under- stand: Women and Men in Conversation. New York: Valentine Books.

Turner, R. Jay and Franco Marino. 1994. "Social Support and Social Structure: A Descriptive

Epidemiology." Journal of Health and Social Behavior 35: 193-21 2.

Umberson, Debra. 1992a. "Gender, Marital Sta- tus, and the Social Control of Health Behav- ior." Social Science and Medicine 34:907-17.

. 1992b. "Relationships between Adult Children and Their Parents: Psychological Consequences for Both Generations." Journal of Marriage and the Family 54:664-74.

Umberson, Debra, Camille B. Wortman, and Ronald C. Kessler. 1992. "Widowhood and Depression: Explaining Gender Differences in Vulnerability to Depression." Journal of Health and Social Behavior 33: 10-24.

Vanfossen. Beth E. 1981. "Sex Differences in the Mental Health Effects of Spouse Support and Equity." Joiirnal of Health and Social Behav- ior 22: 130-43.

Veroff, Joseph, Richard A. Kulka, and Elizabeth Douvan. 1981. Mental Health in American: Patterns of Help-Seeking from 1957 to 1976. New York: Basic Books.

Wellman, Barry and Scot Wortley. 1990. "Differ- ent Strokes from Different Folks: Community Ties and Social Support." American Journal of Sociology 96:558-88.

Young, Rosalie F. and Eva Kahana. 1989. "Speci- fying Care-Giving Outcomes: Gender Relation- ship Aspects of Caregiver Strain." The Geron- tologist 29:660-66.

Comments
  • Recommend Us