Soc. Sci. Med. Vol. 30, No. 4, pp. 469-477, 1990
027%9536/90 $3.00 + 0.00 Copyright © 1990 Pergamon Press plc
Printed in Great Britain. All rights reserved
G E N D E R D I F F E R E N C E S IN THE E F F E C T OF U N E M P L O Y M E N T O N P S Y C H O L O G I C A L DISTRESS MARGARET E. ENSMINGERand DAVID D. CELENTANO Division of Behavioral Sciences and Health Education, Department of Health Policy and Management, The Johns Hopkins School of Hygiene and Public Health, 624 North Broadway, Baltimore, MD 21205, U.S.A. Abstract--In this paper we examine whether unemployment has a differential impact on the expression of psychological distress among men and women. Based on the traditional centrality of the work role to men and the family role to women, we defined several key domains that might affect unemployed men and women differentially: family circumstances, concerns and worries about children and family; coping responses; social support and social integration; and the centrality of the work role. While the study population either were or hoped to be in the labor force and had dependent children, they varied in their marital status and whether they were the custodial parent. Using data collected in Baltimore from those who had been unemployed but had returned to work, those who had remained continuously unemployed for a year, and those who had been continuously employed, we compared the patterns of men's and women's reactions to unemployment. The important differences in psychological symptoms in this population were related to employment status, problems with parenting, financial difficulties, perceived lack of social support, hostility, and feelings about unemployment. By and large, the patterns of these relationships were similar for men and women. These findings suggest that when gender differences in psychological distress are found they may be due to differences in role configurations of men and women rather than intrinsic gender differences.
Key words--unemployment, psychological distress, gender differences, coping
INTRODUCTION
have personality disorders and drug and alcohol problems [17]. The unemployment literature often implies that among those who are unemployed, men will suffer more adverse psychological effects than women because of the centrality of the work role for men [18]. Studies of the impact of unemployment have focused mainly on married men because of their traditional economic role in the family. With the increasing proportion of women who are now in the labor force and who are heads of households, unemployment may be as detrimental for women as for men.
In this paper we examine whether unemployment, a major life stressor, has a differential impact on expressions of psychological distress among men and women. There is an extensive literature describing the impact of unemployment on psychological distress, mental health, and health status. In general, these studies indicate that the unemployed suffer more depression, anxiety, and psychosomatic symptoms than those who are employed [1-6]. While we expect a general decrement in psychological well-being due to unemployment for both men and women, they may respond differently to this major life event. Studies of psychological functioning in a variety of settings and populations have indicated that, overall, women have a higher prevalence of psychological distress than men [7, 8]. Several explanations have been proffered for these differences. A major focus has been that the nature and accumulation of women's social roles are more stress provoking and less fulfilling than those of men [8, 9]. Others have suggested that women are exposed to more stressful life events or other hardships [10, 11]; that women are more vulnerable to the impact of stressful life events [12]; that women's nurturing roles entail a cost of caring [13]; that husbands receive more social support from their wives than they reciprocate [14]; that women's socialization makes them more likely than men to respond to life difficulties with self blame or a sense of helplessness and hopelessness [15, 16]; or that men and women have distinct styles of expressing mental disorder such that women are more likely to internalize distress while men are more likely to
Unemployment and women Studies that have compared the psychological status of working women with those who are not employed are often confounded because it cannot be determined whether women who are not working have ever been in the labor force or want to be in the labor force. In a review of 38 studies of the impact of paid employment on women's psychological wellbeing, Warr and Parry [19] concluded that it is rare for investigators to distinguish between women who are unemployed and seeking work and those women who are not employed and out of the labor market. The importance of this distinction is indicated in a study that examined how employment and attitudes toward employment affected married women's health [20]. W o m e n with favorable attitudes toward employment who were working were less likely to report psychophysiological symptoms than those with unfavorable attitudes. In contrast, for women out of the labor force, those with favorable employment attitudes were more likely to report psychophysiological 469
470
MARGARETE. ENSMINGERand DAVIDD. CELENTANO
symptoms than those who favored not working. Thus, within each labor force status category, women whose attitudes were compatible with their labor force status had better health. There is some evidence that women suffer from unemployment, although there are few studies that address this issue. Hall and Johnson [21] compared women (n = 96) who had lost their jobs due to a plant closing with a control sample of women (n = 51) from the same Swedish rural community who were working in similar jobs. The unemployed women were three times more likely to be classified as depressed measured by the Beck Depression Inventory. Few studies have directly compared the impact of unemployment on psychological functioning in men and women. In a study that examined gender differences in depression, Aneshensel et al. [22] compared men's and women's depression scores (CES-D) across different work and family roles. Employment, particularly full-time, appeared to benefit both sexes, and unemployment had large negative effects for both men and women. Housewives were not included in this analysis since there was no comparable category for men. In general, women had higher depression scores than men, but there was no gender by employment interaction, suggesting that men and women did not differ in their response to unemployment. Gore and Mangione [23] reported that both employment and marriage had a favorable impact on depression for both men and women, based on a study of a random sample of adults in Boston. Controlling for employment and marital status reduced the effect of gender on depression to nonsignificance in their data. McLanahan and Glass [24] examined why gender differences in psychological distress as reported in surveys have declined over the past several decades. They concluded that both the increased employment of women and the increased unemployment of men have contributed to this trend. However, several studies suggest that unemployment may be less important psychologically to women than to men. Jahoda [18] concluded, based on reviews of several studies in countries other than the United States, that unemployment is not as psychologically distressing to women as it is to men because the traditional housewife role is available as an alternative role. Several plant closing studies conducted in communities in the U.S. have found greater psychological consequences for unemployed men than for unemployed women [25, 26]. In these studies unemployed men and women did not differ in psychological distress. However, employed men had lower levels of distress than unemployed men, while employed and unemployed women had similar levels of distress. In a study of unemployment in Michigan, an identified subgroup who appeared to experience few adverse health or mental health effects of unemployment largely included single mothers of young children and women married to men who were working [3]. Whether the same factors explain the impact of unemployment for men and women has rarely been examined. Thoits [27] compared the psychological distress of married and unmarried men and women. She found, using panel data including some 1000 adults from a representative sample of Chicago
households, that stressors were unique to the different groups studied. For example, among the married, wives were more vulnerable to uncontrollable (or fateful) events, such as family deaths, injury to a loved one or job layoffs, while husbands were more vulnerable to controllable events (those experiences in which individuals may have some role in causing them to occur) such as divorce or being fired. Among women, the unmarried were more upset by health events than the married, while wives were more upset by uncontrollable events than unmarried women. She concluded that in investigating gender differences in psychological distress, one needs to compare men and women at equal risk for specific events as well as combinations of role relationships to determine whether women are in fact more psychologically affected or more reactive to particular events than men. If men and women with similar role involvements have similar rates of distress, this suggests that differential role involvements by men and women account for the gender differences often found in studies of psychological distress. RESEARCH QUESTIONS
The literature suggests that, in general, women have more psychological distress than men, but for those who are unemployed, men may suffer more psychological consequences than do the women. In this paper, we examine two issues: first, does unemployment affect the psychological status of childrearing men and women differently; and second, do the same factors explain the impact of unemployment for men as they do for women. Several key domains suggested by the literature that might affect unemployed men and women differentially in their psychological response to unemployment are: their family circumstances, their concerns and worries about their children and family, coping responses, social support and social integration, the meaning of the work role, and how much unemployment has affected their lives. Family circumstances may influence responses to unemployment. One of the most important of these is marital status. Married men, in general, seem to have better mental health than unmarried men or married women [10, 28, 29]. However, because of the traditional economic role that marriage entails for men, unemployment may have a greater impact on them than on married women or unmarried men. Married women are more likely to have an employed spouse than are married men; unmarried fathers commonly do not live with their children and are less likely to have day-to-day family responsibilities than either married men or single women with children. For unmarried parents, unemployment may be more stressful for women than for men, since those women are more likely than men to be the parent that is primarily responsible for children, both financially and as caretaker. We therefore hypothesize that: I. Unemployed married fathers will report more signs of psychological distress than unemployed married mothers or unemployed single fathers. 2. Unemployed unmarried mothers will report more psychological distress than unemployed married mothers or unemployed single fathers.
Gender differences in psychological distress
471
Men and women also differ in the child-rearing responsibilities they have. Mothers, including those in the labor force, are likely to have more responsibility for children than fathers. In the case of divorce or separation, mothers are more likely to be the custodial parent. F o r these reasons, the concerns and worries regarding children and child-rearing may be more psychologically distressing to women than to men. F o r those who are unemployed, these parenting concerns may be especially distressing. We hypothesize that: 3. Child-rearing worries and burdens will affect the psychological distress of women but not men.
8. Those unemployed who perceive that unemployment had greatly changed how they were thought of and how they structured their day will have more psychological distress than those who perceive little change. Since men will be more likely to perceive more change as a result of unemployment than women, controlling on this will decrease the unemployment-psychological distress relationship among men but not among women.
Men and w o m e n may have different coping responses to deal with life stress because of their
9. Financial strain will have a detrimental psychological effect on both men and women. Financial strain will explain some of the effect of unemployment on psychological distress, but it will not differentially affect men and women.
socialization. These may impact on their response to stressful life circumstances. Dohrenwend and Dohrenwend [17] suggested that men may be more likely to respond with outward aggression or hostility while w o m e n may be more likely to become depressed. Men may be less likely to talk about their troubles with unemployment while women may be more expressive about problems. W o m e n may also be better able to ask for help from others while men feel more constrained to be independent. All of these coping styles may affect psychological distress. We hypothesize that: 4. Unemployed men will have higher levels of hostility than unemployed women. Since hostility and psychological distress are interrelated, hostility will account for more of unemployed men's psychological distress than unemployed women's. 5. Those unemployed who discuss their problems will have less psychological distress than those who do not. Since this coping style is more likely to be present in women than in men, controlling on this coping style will decrease the unemployment-psychological distress relationship more among women than among men. In general, men and women seem to have different patterns of social support and social integration [13]. Since social support and social integration may buffer the impact of unemployment on psychological distress, gender differences in these patterns may influence the psychological response to unemployment. We hypothesize that: 6. Those unemployed with more social support and more social integration will have less psychological distress than those with less. Since high levels of social support and social integration are more likely to be present in women than in men, controlling on social support and social integration will decrease the unemployment-psychological distress relationship among women but not among men. Because of the traditional centrality of the work role to men's self concept, unemployed men may feel especially distressed. We expect both the meaning of work and how much unemployment is perceived as a problem to differ for men and women, and this will affect how men and women respond to being unemployed. We hypothesize that: 7. Those unemployed who feel that work did not bring out the best in them will have less psychological distress than those who felt that their work did bring out their best. Since men will be more likely to report meaningful work than women, controlling on the meaning of work will decrease the unemployment-psychological distress relationship among men but not among women.
We expect financial difficulties to have an impact on the unemployment of both men and women. We know from previous analyses [1] that financial problems are highly related to unemployment and affect the psychological distress of the unemployed. We hypothesize that:
Comparing the psychological distress of men and women who all have dependent children under the age of 16 and who have become unemployed in the past year provides a unique opportunity to examine whether men and women with similar family and work role involvements respond differentially to the same major life stressor. A comparison group of continuously employed men and women with children and living in the same neighborhoods as the unemployed allow us to determine whether men and women who have not recently experienced unemployment differ from each other and from the unemployed in psychological distress.
DATA AND METHODS
Population We present analyses based on a 1984 cross-sectional survey of child-rearing men and women in Baltimore, Maryland who were receiving unemployment benefits during January-April, 1983 and a control population of men and women living in the same neighborhoods, but who had been continuously employed during the previous 16 months. All respondents had dependents under the age of 17 years and lived in a catchment area of one of three of the seven mental health centers in Baltimore. Their names were obtained from State lists of those receiving unemployment compensation (n = 383). A b o u t half of those who were receiving unemployment during 1983 were reemployed when interviewed in 1984 while about half remained unemployed. Interviews were completed with 269 of those on the list (70%); 13% refused to be interviewed and we could not locate 16%. Interviews lasted about one hour and were conducted over the telephone (34%) or in the respondent's home (66%). In order to obtain a comparison group we randomly selected 100 of the unemployed respondents and interviewed someone on their block who was their same sex, race, had dependent children, and had been continuously employed during the study period. We were able to complete interviews with 92 matched community controls. Three groups will be compared: those who have been unemployed and were still unemployed a year later (n = 133); those who had been unemployed in 1983 but had returned to work within a year
472
MARGARETE. ENSMINGERand DAVIDD. CELENTANO
(n = 135); and those living in the same neighborhood as the others but who had been continuouslyemployed during the time of the study in the same job (n = 92). Most often gender differences in psychological distress have been examined in samples chosen to represent the adult population of a country, a large metropolitan area, etc., and include respondents of varying stages of life, social status, marital status, as well as job status. The differential effect of unemployment on men and women is not easily studied in such samples because there are too few unemployed unless the sample is extremely large. (For example, in a panel sample of 1106 adult respondents, aged 18-65, originally chosen to represent the Chicago metropolitan area, only 59 respondents reported being laid off or fired in the 4 year interval before 1976 [30]; in a community survey of 1000 conducted in Los Angeles during 1979, 69 respondents responded they were unemployed [22].) In addition, the unemployed would differ from each other in terms of stage of life, parenting responsibilities, neighborhood, as well as gender. These factors need to be taken into account in the analyses. In contrast, the respondents for this study are all from similar neighborhoods, all are in the labor force, and all are parents of children from 0 to 16 years. Thus, the study design holds constant many of the social characteristics that might affect individuals who are unemployed. We are able to examine in greater detail how these child-rearing men and women are affected by unemployment than we would if studying a more diverse (even if more representative) population. The analyses examine whether unemployed men and women have similar psychological distress and whether the same factors relate to their functioning. Several domains reflecting the posited hypotheses are included. First, within the family domain, marital status, child-care arrangements, parenting concerns and worries, and family relationships may differentially influence the psychological distress of unemployed men and women. Second, men and women may have different coping responses. Third, social support and social integration for men and women are likely to differ, and these may affect the impact of unemployment. Fourth, factors related to the
*Hostility represents one type of coping style frequently aroused as a way of coping with stress. We had some concern that the measure of hostility might overlap with the GHQ, the psychologicaldistress measure. Uhlenhuth et aL [36] reported that 73% of persons scoring high on hostility also were considered to have a major depression, 56% were demonstrated to have a syndrome of general anxiety, and almost half were labeled as having other high distress. The correlations between the hostility and the GHQ items were examined in order to assess their independence. The GHQ items correlated significantly higher with the GHQ total (mean item-to-total correlation was 0.61) than with the measure of hostility (where an average correlation of 0.30) was found. A factor analysis confirmed this finding: all five of the hostility measures fell together on one factor while none of the GHQ items loaded above 0.40 on this factor (and only one was found with a loading above 0.30). On the basis of these findings, we conclude that the measures of hostility and psychiatric symptoms are empirically as well as conceptually distinct.
meaning of work and the meaning of unemployment are hypothesized to differ for men and women. The comparison sample of employed men and women demonstrate whether gender differences we find are general gender differences that are found among both employed and unemployed populations or whether unemployment affects psychological processes differentially for men and women. A description of these domains and their operationalization follows.
Measures The dependent variable is a measure of psychological distress derived from a subset of 20 questions from the General Health Questionnaire (GHQ). This instrument was designed to detect psychiatric disorders among respondents in community settings [31]. Several studies have shown the GHQ to be an effective measure of psychiatric morbidity in general populations [32-34]. In this study we use items which include aspects of the dimensions of depression, anxiety, and social impairment. A summed score of these 20 items is used. The independent variables include family circumstances, family concerns and worries, coping, social integration and social support, financial difficulties, the meaning of work, and perceived problems of unemployment. Family circumstances includes both family role responsibilities and parenting concerns. The role responsibilities examined include whether the respondent is married or unmarried; the number of children the respondent has; and whether the respondent is a primary caretaker of the children. Parenting concerns are indicated by two m e a s u r e s (l) whether they had more or fewer problems as a parent than other parents with similarly aged children and how much of the time problems of being a parent were on their mind; and (2) the parent's concern about their child's health. The measure of family coping consisted of five items selected from questions used by Pearlin and Schooler [35] asking respondents how they dealt with disagreements and conflict in the family. A high score indicates poor communication, i.e. respondents ignored disagreements or conflict within their family. This scale had an internal consistency coefficient of 0.64 (Cronbach's alpha). We also asked two questions concerning whether the respondent ignored problems or talked with others about them and whether the respondent held problems inside or showed them. A measure of hostility was selected from an instrument developed by Uhlenhuth et al. [36] modified after the Hopkins Symptom Checklist. Five items were included in the scale and asked how much in the past 7 days the respondent was distressed by being easily upset, irritated or annoyed; getting angry over things that aren't really too important; temper outbursts; wanting to hurt somebody or smash something; or feeling critical of others (Cronbach alpha = 0.78).* Hostility was included because of its potential as a means of coping with unemployment and the expectation that unemployed men would more often report hostility than unemployed women. Social integration was measured by the number and extent of participation in community organizations and by church membership and participation. Social
473
Gender differences in psychological distress Table 1. Mean ratings of psychological distress for marital status and gender, by employment status
ANALYSES AND RESULTS
Marital status
Employment group Marital status Married Male Female Unmarried Male Female
Continuously employed
Returned (n) to work (n)
Unemployed
(n)
35.2 35.7
(43) (20)
37.9 39.0
(53) (18)
41.4 41.6
(33) (22)
32.6 38.1
(6) (23)
37.4 37.1
(36) (24)
43.8 42.8
(37) (42)
Analysis of variance: F = 3.352; R 2 = 0.087; P < 0.001.
support was assessed by asking respondents if they loaned money to friends or relatives; if they borrowed money from friends/relatives; if they wished to know more people to get together with; if they wished to know more people to talk with about personal matters; and how often the respondent talked with someone about personal matters. In order to measure the meaning of working, we asked each respondent whether they expected to have pleasure from their job and whether they felt their current or most recent job brought out the best in them. For those respondents who have been unemployed or were currently unemployed we asked a series of eight questions about perceived problems of unemployment: had it affected the way they felt about themselves, the way those around them felt about them, how much it had been a major problem in their life, in the way they structured their time, their life at home, and their dependence on others, and how much they had to worry about getting a job. These items were combined in a scale that had an internal consistency reliability coefficient of 0.70. In the financial strain area, a scale of bill problems was selected from Pearlin and Schooler [35]. Respondents were asked how much difficulty they had in seven areas: paying for food, transportation, medical care, utilities and heating, telephone, mortgage/rent, and other bills. These were combined in a scale which had an internal consistency reliability coefficient of 0.86 (Cronbach's alpha). Jobs were also scored from 0 to 9 on their relative status according to the Bureau of the Census, with professional and technical jobs coded on the low end of the scale.
We first examine, by analysis of variance, the relationship between G H Q scores and marital status, gender, and employment status, the major role configurations for those in this study. Table 1 shows the G H Q means for these categories. The unemployed have much higher G H Q scores than either the continuously employed or those who had returned to work. There was no main effect for either gender or marital status on psychological distress. The only substantive gender difference seen in Table 1 is the comparison of the unmarried who were continuously employed. In this category, the difference between males and females in G H Q means was 5.5 (t = 2.20, P < 0.05), a little over a half standard deviation. For all other employment-marital groups, the G H Q means for males and females were very similar. In order to examine whether unemployment affects the psychological status of men and women differently, we performed multiple regressions including gender, employment status, and two interaction terms of gender and employment status, and variables from each domain. We analyzed each domain separately including all the variables from that domain. In none of these multiple regressions were the standardized regression coefficients for gender or gender by employment status interaction terms significantly related to G H Q scores. In order to examine whether unemployment affects men and women differently depending on the seven key domains outlined earlier, we analyzed multiple regression equations for males and females separately-regressing G H Q scores on the variables from each domain as well as employment status.
Family circumstances The multiple regression analyses included marital status, the number of dependent children, whether the respondent had the caretaking responsibility for the child as well as employment status. As shown in Table 2, none of these indicators of family status affect either men's or women's report of psychiatric symptoms (nor did they interact with employment status on the G H Q scores). After adjusting for family status, employment continues to have a strong impact on the G H Q scores for both men and women with the continuously employed and those who had returned
Table 2. Regression of G H Q scores by family status, and employment status for males and females
Males Independent variables Marital status 1 = Married 2 = Unmarried Child-care responsibility 1 = Not custodial parent 2 = Custodial parent Number of Children Reemployed Continuously employed R2= Adjusted R2= f P ~
027%9536/90 $3.00 + 0.00 Copyright © 1990 Pergamon Press plc
Printed in Great Britain. All rights reserved
G E N D E R D I F F E R E N C E S IN THE E F F E C T OF U N E M P L O Y M E N T O N P S Y C H O L O G I C A L DISTRESS MARGARET E. ENSMINGERand DAVID D. CELENTANO Division of Behavioral Sciences and Health Education, Department of Health Policy and Management, The Johns Hopkins School of Hygiene and Public Health, 624 North Broadway, Baltimore, MD 21205, U.S.A. Abstract--In this paper we examine whether unemployment has a differential impact on the expression of psychological distress among men and women. Based on the traditional centrality of the work role to men and the family role to women, we defined several key domains that might affect unemployed men and women differentially: family circumstances, concerns and worries about children and family; coping responses; social support and social integration; and the centrality of the work role. While the study population either were or hoped to be in the labor force and had dependent children, they varied in their marital status and whether they were the custodial parent. Using data collected in Baltimore from those who had been unemployed but had returned to work, those who had remained continuously unemployed for a year, and those who had been continuously employed, we compared the patterns of men's and women's reactions to unemployment. The important differences in psychological symptoms in this population were related to employment status, problems with parenting, financial difficulties, perceived lack of social support, hostility, and feelings about unemployment. By and large, the patterns of these relationships were similar for men and women. These findings suggest that when gender differences in psychological distress are found they may be due to differences in role configurations of men and women rather than intrinsic gender differences.
Key words--unemployment, psychological distress, gender differences, coping
INTRODUCTION
have personality disorders and drug and alcohol problems [17]. The unemployment literature often implies that among those who are unemployed, men will suffer more adverse psychological effects than women because of the centrality of the work role for men [18]. Studies of the impact of unemployment have focused mainly on married men because of their traditional economic role in the family. With the increasing proportion of women who are now in the labor force and who are heads of households, unemployment may be as detrimental for women as for men.
In this paper we examine whether unemployment, a major life stressor, has a differential impact on expressions of psychological distress among men and women. There is an extensive literature describing the impact of unemployment on psychological distress, mental health, and health status. In general, these studies indicate that the unemployed suffer more depression, anxiety, and psychosomatic symptoms than those who are employed [1-6]. While we expect a general decrement in psychological well-being due to unemployment for both men and women, they may respond differently to this major life event. Studies of psychological functioning in a variety of settings and populations have indicated that, overall, women have a higher prevalence of psychological distress than men [7, 8]. Several explanations have been proffered for these differences. A major focus has been that the nature and accumulation of women's social roles are more stress provoking and less fulfilling than those of men [8, 9]. Others have suggested that women are exposed to more stressful life events or other hardships [10, 11]; that women are more vulnerable to the impact of stressful life events [12]; that women's nurturing roles entail a cost of caring [13]; that husbands receive more social support from their wives than they reciprocate [14]; that women's socialization makes them more likely than men to respond to life difficulties with self blame or a sense of helplessness and hopelessness [15, 16]; or that men and women have distinct styles of expressing mental disorder such that women are more likely to internalize distress while men are more likely to
Unemployment and women Studies that have compared the psychological status of working women with those who are not employed are often confounded because it cannot be determined whether women who are not working have ever been in the labor force or want to be in the labor force. In a review of 38 studies of the impact of paid employment on women's psychological wellbeing, Warr and Parry [19] concluded that it is rare for investigators to distinguish between women who are unemployed and seeking work and those women who are not employed and out of the labor market. The importance of this distinction is indicated in a study that examined how employment and attitudes toward employment affected married women's health [20]. W o m e n with favorable attitudes toward employment who were working were less likely to report psychophysiological symptoms than those with unfavorable attitudes. In contrast, for women out of the labor force, those with favorable employment attitudes were more likely to report psychophysiological 469
470
MARGARETE. ENSMINGERand DAVIDD. CELENTANO
symptoms than those who favored not working. Thus, within each labor force status category, women whose attitudes were compatible with their labor force status had better health. There is some evidence that women suffer from unemployment, although there are few studies that address this issue. Hall and Johnson [21] compared women (n = 96) who had lost their jobs due to a plant closing with a control sample of women (n = 51) from the same Swedish rural community who were working in similar jobs. The unemployed women were three times more likely to be classified as depressed measured by the Beck Depression Inventory. Few studies have directly compared the impact of unemployment on psychological functioning in men and women. In a study that examined gender differences in depression, Aneshensel et al. [22] compared men's and women's depression scores (CES-D) across different work and family roles. Employment, particularly full-time, appeared to benefit both sexes, and unemployment had large negative effects for both men and women. Housewives were not included in this analysis since there was no comparable category for men. In general, women had higher depression scores than men, but there was no gender by employment interaction, suggesting that men and women did not differ in their response to unemployment. Gore and Mangione [23] reported that both employment and marriage had a favorable impact on depression for both men and women, based on a study of a random sample of adults in Boston. Controlling for employment and marital status reduced the effect of gender on depression to nonsignificance in their data. McLanahan and Glass [24] examined why gender differences in psychological distress as reported in surveys have declined over the past several decades. They concluded that both the increased employment of women and the increased unemployment of men have contributed to this trend. However, several studies suggest that unemployment may be less important psychologically to women than to men. Jahoda [18] concluded, based on reviews of several studies in countries other than the United States, that unemployment is not as psychologically distressing to women as it is to men because the traditional housewife role is available as an alternative role. Several plant closing studies conducted in communities in the U.S. have found greater psychological consequences for unemployed men than for unemployed women [25, 26]. In these studies unemployed men and women did not differ in psychological distress. However, employed men had lower levels of distress than unemployed men, while employed and unemployed women had similar levels of distress. In a study of unemployment in Michigan, an identified subgroup who appeared to experience few adverse health or mental health effects of unemployment largely included single mothers of young children and women married to men who were working [3]. Whether the same factors explain the impact of unemployment for men and women has rarely been examined. Thoits [27] compared the psychological distress of married and unmarried men and women. She found, using panel data including some 1000 adults from a representative sample of Chicago
households, that stressors were unique to the different groups studied. For example, among the married, wives were more vulnerable to uncontrollable (or fateful) events, such as family deaths, injury to a loved one or job layoffs, while husbands were more vulnerable to controllable events (those experiences in which individuals may have some role in causing them to occur) such as divorce or being fired. Among women, the unmarried were more upset by health events than the married, while wives were more upset by uncontrollable events than unmarried women. She concluded that in investigating gender differences in psychological distress, one needs to compare men and women at equal risk for specific events as well as combinations of role relationships to determine whether women are in fact more psychologically affected or more reactive to particular events than men. If men and women with similar role involvements have similar rates of distress, this suggests that differential role involvements by men and women account for the gender differences often found in studies of psychological distress. RESEARCH QUESTIONS
The literature suggests that, in general, women have more psychological distress than men, but for those who are unemployed, men may suffer more psychological consequences than do the women. In this paper, we examine two issues: first, does unemployment affect the psychological status of childrearing men and women differently; and second, do the same factors explain the impact of unemployment for men as they do for women. Several key domains suggested by the literature that might affect unemployed men and women differentially in their psychological response to unemployment are: their family circumstances, their concerns and worries about their children and family, coping responses, social support and social integration, the meaning of the work role, and how much unemployment has affected their lives. Family circumstances may influence responses to unemployment. One of the most important of these is marital status. Married men, in general, seem to have better mental health than unmarried men or married women [10, 28, 29]. However, because of the traditional economic role that marriage entails for men, unemployment may have a greater impact on them than on married women or unmarried men. Married women are more likely to have an employed spouse than are married men; unmarried fathers commonly do not live with their children and are less likely to have day-to-day family responsibilities than either married men or single women with children. For unmarried parents, unemployment may be more stressful for women than for men, since those women are more likely than men to be the parent that is primarily responsible for children, both financially and as caretaker. We therefore hypothesize that: I. Unemployed married fathers will report more signs of psychological distress than unemployed married mothers or unemployed single fathers. 2. Unemployed unmarried mothers will report more psychological distress than unemployed married mothers or unemployed single fathers.
Gender differences in psychological distress
471
Men and women also differ in the child-rearing responsibilities they have. Mothers, including those in the labor force, are likely to have more responsibility for children than fathers. In the case of divorce or separation, mothers are more likely to be the custodial parent. F o r these reasons, the concerns and worries regarding children and child-rearing may be more psychologically distressing to women than to men. F o r those who are unemployed, these parenting concerns may be especially distressing. We hypothesize that: 3. Child-rearing worries and burdens will affect the psychological distress of women but not men.
8. Those unemployed who perceive that unemployment had greatly changed how they were thought of and how they structured their day will have more psychological distress than those who perceive little change. Since men will be more likely to perceive more change as a result of unemployment than women, controlling on this will decrease the unemployment-psychological distress relationship among men but not among women.
Men and w o m e n may have different coping responses to deal with life stress because of their
9. Financial strain will have a detrimental psychological effect on both men and women. Financial strain will explain some of the effect of unemployment on psychological distress, but it will not differentially affect men and women.
socialization. These may impact on their response to stressful life circumstances. Dohrenwend and Dohrenwend [17] suggested that men may be more likely to respond with outward aggression or hostility while w o m e n may be more likely to become depressed. Men may be less likely to talk about their troubles with unemployment while women may be more expressive about problems. W o m e n may also be better able to ask for help from others while men feel more constrained to be independent. All of these coping styles may affect psychological distress. We hypothesize that: 4. Unemployed men will have higher levels of hostility than unemployed women. Since hostility and psychological distress are interrelated, hostility will account for more of unemployed men's psychological distress than unemployed women's. 5. Those unemployed who discuss their problems will have less psychological distress than those who do not. Since this coping style is more likely to be present in women than in men, controlling on this coping style will decrease the unemployment-psychological distress relationship more among women than among men. In general, men and women seem to have different patterns of social support and social integration [13]. Since social support and social integration may buffer the impact of unemployment on psychological distress, gender differences in these patterns may influence the psychological response to unemployment. We hypothesize that: 6. Those unemployed with more social support and more social integration will have less psychological distress than those with less. Since high levels of social support and social integration are more likely to be present in women than in men, controlling on social support and social integration will decrease the unemployment-psychological distress relationship among women but not among men. Because of the traditional centrality of the work role to men's self concept, unemployed men may feel especially distressed. We expect both the meaning of work and how much unemployment is perceived as a problem to differ for men and women, and this will affect how men and women respond to being unemployed. We hypothesize that: 7. Those unemployed who feel that work did not bring out the best in them will have less psychological distress than those who felt that their work did bring out their best. Since men will be more likely to report meaningful work than women, controlling on the meaning of work will decrease the unemployment-psychological distress relationship among men but not among women.
We expect financial difficulties to have an impact on the unemployment of both men and women. We know from previous analyses [1] that financial problems are highly related to unemployment and affect the psychological distress of the unemployed. We hypothesize that:
Comparing the psychological distress of men and women who all have dependent children under the age of 16 and who have become unemployed in the past year provides a unique opportunity to examine whether men and women with similar family and work role involvements respond differentially to the same major life stressor. A comparison group of continuously employed men and women with children and living in the same neighborhoods as the unemployed allow us to determine whether men and women who have not recently experienced unemployment differ from each other and from the unemployed in psychological distress.
DATA AND METHODS
Population We present analyses based on a 1984 cross-sectional survey of child-rearing men and women in Baltimore, Maryland who were receiving unemployment benefits during January-April, 1983 and a control population of men and women living in the same neighborhoods, but who had been continuously employed during the previous 16 months. All respondents had dependents under the age of 17 years and lived in a catchment area of one of three of the seven mental health centers in Baltimore. Their names were obtained from State lists of those receiving unemployment compensation (n = 383). A b o u t half of those who were receiving unemployment during 1983 were reemployed when interviewed in 1984 while about half remained unemployed. Interviews were completed with 269 of those on the list (70%); 13% refused to be interviewed and we could not locate 16%. Interviews lasted about one hour and were conducted over the telephone (34%) or in the respondent's home (66%). In order to obtain a comparison group we randomly selected 100 of the unemployed respondents and interviewed someone on their block who was their same sex, race, had dependent children, and had been continuously employed during the study period. We were able to complete interviews with 92 matched community controls. Three groups will be compared: those who have been unemployed and were still unemployed a year later (n = 133); those who had been unemployed in 1983 but had returned to work within a year
472
MARGARETE. ENSMINGERand DAVIDD. CELENTANO
(n = 135); and those living in the same neighborhood as the others but who had been continuouslyemployed during the time of the study in the same job (n = 92). Most often gender differences in psychological distress have been examined in samples chosen to represent the adult population of a country, a large metropolitan area, etc., and include respondents of varying stages of life, social status, marital status, as well as job status. The differential effect of unemployment on men and women is not easily studied in such samples because there are too few unemployed unless the sample is extremely large. (For example, in a panel sample of 1106 adult respondents, aged 18-65, originally chosen to represent the Chicago metropolitan area, only 59 respondents reported being laid off or fired in the 4 year interval before 1976 [30]; in a community survey of 1000 conducted in Los Angeles during 1979, 69 respondents responded they were unemployed [22].) In addition, the unemployed would differ from each other in terms of stage of life, parenting responsibilities, neighborhood, as well as gender. These factors need to be taken into account in the analyses. In contrast, the respondents for this study are all from similar neighborhoods, all are in the labor force, and all are parents of children from 0 to 16 years. Thus, the study design holds constant many of the social characteristics that might affect individuals who are unemployed. We are able to examine in greater detail how these child-rearing men and women are affected by unemployment than we would if studying a more diverse (even if more representative) population. The analyses examine whether unemployed men and women have similar psychological distress and whether the same factors relate to their functioning. Several domains reflecting the posited hypotheses are included. First, within the family domain, marital status, child-care arrangements, parenting concerns and worries, and family relationships may differentially influence the psychological distress of unemployed men and women. Second, men and women may have different coping responses. Third, social support and social integration for men and women are likely to differ, and these may affect the impact of unemployment. Fourth, factors related to the
*Hostility represents one type of coping style frequently aroused as a way of coping with stress. We had some concern that the measure of hostility might overlap with the GHQ, the psychologicaldistress measure. Uhlenhuth et aL [36] reported that 73% of persons scoring high on hostility also were considered to have a major depression, 56% were demonstrated to have a syndrome of general anxiety, and almost half were labeled as having other high distress. The correlations between the hostility and the GHQ items were examined in order to assess their independence. The GHQ items correlated significantly higher with the GHQ total (mean item-to-total correlation was 0.61) than with the measure of hostility (where an average correlation of 0.30) was found. A factor analysis confirmed this finding: all five of the hostility measures fell together on one factor while none of the GHQ items loaded above 0.40 on this factor (and only one was found with a loading above 0.30). On the basis of these findings, we conclude that the measures of hostility and psychiatric symptoms are empirically as well as conceptually distinct.
meaning of work and the meaning of unemployment are hypothesized to differ for men and women. The comparison sample of employed men and women demonstrate whether gender differences we find are general gender differences that are found among both employed and unemployed populations or whether unemployment affects psychological processes differentially for men and women. A description of these domains and their operationalization follows.
Measures The dependent variable is a measure of psychological distress derived from a subset of 20 questions from the General Health Questionnaire (GHQ). This instrument was designed to detect psychiatric disorders among respondents in community settings [31]. Several studies have shown the GHQ to be an effective measure of psychiatric morbidity in general populations [32-34]. In this study we use items which include aspects of the dimensions of depression, anxiety, and social impairment. A summed score of these 20 items is used. The independent variables include family circumstances, family concerns and worries, coping, social integration and social support, financial difficulties, the meaning of work, and perceived problems of unemployment. Family circumstances includes both family role responsibilities and parenting concerns. The role responsibilities examined include whether the respondent is married or unmarried; the number of children the respondent has; and whether the respondent is a primary caretaker of the children. Parenting concerns are indicated by two m e a s u r e s (l) whether they had more or fewer problems as a parent than other parents with similarly aged children and how much of the time problems of being a parent were on their mind; and (2) the parent's concern about their child's health. The measure of family coping consisted of five items selected from questions used by Pearlin and Schooler [35] asking respondents how they dealt with disagreements and conflict in the family. A high score indicates poor communication, i.e. respondents ignored disagreements or conflict within their family. This scale had an internal consistency coefficient of 0.64 (Cronbach's alpha). We also asked two questions concerning whether the respondent ignored problems or talked with others about them and whether the respondent held problems inside or showed them. A measure of hostility was selected from an instrument developed by Uhlenhuth et al. [36] modified after the Hopkins Symptom Checklist. Five items were included in the scale and asked how much in the past 7 days the respondent was distressed by being easily upset, irritated or annoyed; getting angry over things that aren't really too important; temper outbursts; wanting to hurt somebody or smash something; or feeling critical of others (Cronbach alpha = 0.78).* Hostility was included because of its potential as a means of coping with unemployment and the expectation that unemployed men would more often report hostility than unemployed women. Social integration was measured by the number and extent of participation in community organizations and by church membership and participation. Social
473
Gender differences in psychological distress Table 1. Mean ratings of psychological distress for marital status and gender, by employment status
ANALYSES AND RESULTS
Marital status
Employment group Marital status Married Male Female Unmarried Male Female
Continuously employed
Returned (n) to work (n)
Unemployed
(n)
35.2 35.7
(43) (20)
37.9 39.0
(53) (18)
41.4 41.6
(33) (22)
32.6 38.1
(6) (23)
37.4 37.1
(36) (24)
43.8 42.8
(37) (42)
Analysis of variance: F = 3.352; R 2 = 0.087; P < 0.001.
support was assessed by asking respondents if they loaned money to friends or relatives; if they borrowed money from friends/relatives; if they wished to know more people to get together with; if they wished to know more people to talk with about personal matters; and how often the respondent talked with someone about personal matters. In order to measure the meaning of working, we asked each respondent whether they expected to have pleasure from their job and whether they felt their current or most recent job brought out the best in them. For those respondents who have been unemployed or were currently unemployed we asked a series of eight questions about perceived problems of unemployment: had it affected the way they felt about themselves, the way those around them felt about them, how much it had been a major problem in their life, in the way they structured their time, their life at home, and their dependence on others, and how much they had to worry about getting a job. These items were combined in a scale that had an internal consistency reliability coefficient of 0.70. In the financial strain area, a scale of bill problems was selected from Pearlin and Schooler [35]. Respondents were asked how much difficulty they had in seven areas: paying for food, transportation, medical care, utilities and heating, telephone, mortgage/rent, and other bills. These were combined in a scale which had an internal consistency reliability coefficient of 0.86 (Cronbach's alpha). Jobs were also scored from 0 to 9 on their relative status according to the Bureau of the Census, with professional and technical jobs coded on the low end of the scale.
We first examine, by analysis of variance, the relationship between G H Q scores and marital status, gender, and employment status, the major role configurations for those in this study. Table 1 shows the G H Q means for these categories. The unemployed have much higher G H Q scores than either the continuously employed or those who had returned to work. There was no main effect for either gender or marital status on psychological distress. The only substantive gender difference seen in Table 1 is the comparison of the unmarried who were continuously employed. In this category, the difference between males and females in G H Q means was 5.5 (t = 2.20, P < 0.05), a little over a half standard deviation. For all other employment-marital groups, the G H Q means for males and females were very similar. In order to examine whether unemployment affects the psychological status of men and women differently, we performed multiple regressions including gender, employment status, and two interaction terms of gender and employment status, and variables from each domain. We analyzed each domain separately including all the variables from that domain. In none of these multiple regressions were the standardized regression coefficients for gender or gender by employment status interaction terms significantly related to G H Q scores. In order to examine whether unemployment affects men and women differently depending on the seven key domains outlined earlier, we analyzed multiple regression equations for males and females separately-regressing G H Q scores on the variables from each domain as well as employment status.
Family circumstances The multiple regression analyses included marital status, the number of dependent children, whether the respondent had the caretaking responsibility for the child as well as employment status. As shown in Table 2, none of these indicators of family status affect either men's or women's report of psychiatric symptoms (nor did they interact with employment status on the G H Q scores). After adjusting for family status, employment continues to have a strong impact on the G H Q scores for both men and women with the continuously employed and those who had returned
Table 2. Regression of G H Q scores by family status, and employment status for males and females
Males Independent variables Marital status 1 = Married 2 = Unmarried Child-care responsibility 1 = Not custodial parent 2 = Custodial parent Number of Children Reemployed Continuously employed R2= Adjusted R2= f P ~
