Divorce: Crisis, Challenge, Or Relief? 9780814723647

Not since William Goode's Women in Divorce in the 1950's have we had such a comprehensive study of adjustment

161 67 16MB

English Pages [328] Year 1991

Report DMCA / Copyright

DOWNLOAD FILE

Polecaj historie

Divorce: Crisis, Challenge, Or Relief?
 9780814723647

Citation preview

DIVORCE

The early phase of this study was supported in part by the National Institute on Aging Program Project grant No. AG00002, entitled "Divorce: A Psychosocial Study of Adaptation/' Project Director: D. A. Chiriboga; Program Principal Investigator: M. Fiske. The follow-up investigation was supported in part by National Institute of Mental Health grant No. MH33713, "Mental Illness and Divorce: A Life Span Study/' Principal Investigator: D. A. Chiriboga.

DIVORCE CRISIS, CHALLENGE OR RELIEF?

David A. Chiriboga, Linda S. Catron and Associates

NEW YORK UNIVERSITY PRESS New York and London

Copyright © 1991 by New York University All rights reserved Manufactured in the United States of America

Library of Congress Cataloging-in-Publication Data Chiriboga, David Anthony. Divorce : crisis, challenge or relief? / David A. Chiriboga, Linda S. Catron and associates. p. cm. Includes bibliographical references (p. ) and index. ISBN 0-8147-1450-1 (alk.paper) 1. Divorce—California—Longitudinal studies. 2. Divorce— California—Psychological aspects—Longitudinal studies. I. Catron, Linda S., 1948. II. Title. HQ835.C2C45 1991 306.89- one person = 2," or

SOCIAL SUPPORTS

199

"yes, more than one person = 3." The average score of 2.62 indicated that a separated man or woman generally had somewhat more than one confidant. Only 8% reported they had no confidant, 21% reported one confidant and the majority, 71%, had more than one. With 92% having at least one confidant, it can be seen participants were by and large not lacking in having available someone they could count on for understanding and advice. However, although few were without a confidant, the literature (e.g., Lowenthal and Haven 1968) suggests that those few may constitute a high-risk group. It is important to add that men were less likely to have a confidant than women. Support Behavior: Support Seeking. This measure was used to assess the behavioral dimension of social support. Derived from an instrument created by Pearlin and Lieberman (1976), it is based on responses to the question "Did you try to get any advice or assistance or talk to anyone about the separation or divorce?" If respondents had sought assistance they were asked to check off which of the following 13 categories of help givers they consulted: spouse, parent, child, relative, friend, neighbor, co-worker, doctor, clergy, counselor, lawyer, self-help groups, other. Responses to these questions provide an indication of how much help with the separation and divorce the individual has sought. In comparison to the Active Network and Confidant measures which are indicators of potential social support, the Support Seeking measure is an indicator of actual social support-seeking activity. We developed a simple summary score of the number of help-seeking categories approached by respondents. For the total sample, the average number of help-giving categories people turned to was 2.84 (SD = 2.27). The mean Support Seeking score for men, at 2.56 (SD = 2.33) tended to be lower than the mean for women (3.04, SD = 2.21). In other words, on the average women tended to seek assistance from more categories of help-givers than did men. Perceived Support. This measure assessed the participants' evaluation of the quality of support available to them. Perceived Support was similar to the Support Seeking measure in that it assessed social support in terms of the individual's actual experience of support, rather than in terms of potential support. Information came from responses to two questions: 1. "How often are you without anyone to talk to about yourself?" (coded— never = 4, once in a while = 3, fairly often = 2, very often = 1).

2OO

CONTRIBUTING FACTORS

2. ''How often are you without anyone to share experiences and feelings with?" (coded—never = 4, once in a while = 3, fairly often = 2, very often - 1). We added the responses to these two questions in order to obtain a summary score, with higher responses reflecting a greater perception of support on the respondent's part. In the entire sample the mean score on perception of support was 6.45, with a standard deviation of 1.81. Consistent with findings for the measures of Confidant and Support Seeking, men perceived themselves to have significantly less support than did women. For men the average score was 6.05 (SD = .85), while for women it was 6.72 (SD= 1.73). Intel-correlations of the Four Measures of Social Support

As you may recall, our goal was to develop relatively distinct measures of support. Our findings indicate that in this we were successful: there were generally low intercorrelations among the support variables. Support Seeking in fact was virtually independent of all other support variables. The closest association was between Perceived Support and Confidant (r= .28, p= .000). While it makes sense that persons with more confidants would perceive greater support, the relationship still is a rather weak one. There was also a tendency for people with larger networks to have more confidants and to perceive that they were supported. While these associations make sense, more important perhaps is that the four measures of social support are essentially independent of each other, a finding that lends credence to the notion that they represent different dimensions of support. This in turn emphasizes the importance of a multidimensional approach to the measurement of support. The goal should not be to find the single best measure of social support but, rather, to capture social support in all its complexity by identifying the central facets of the various forms of support. While the intercorrelations remain low, some interesting differences emerge between men and women when they are looked at separately. For men but not for women, the larger the Active Network, the higher the level of Support Seeking. For women but not for men, the larger the Active Network, the higher the score on the Confidant measure. For both men and women, the Confidant measure was significantly related to Perceived Support. The cor-

SOCIAL SUPPORTS

201

relation was higher for women. These differences in the intercorrelations among measures of social support suggest that there may be differences in the salience of each dimension of social support for men and women. THE CONSISTENCY OF SOCIAL SUPPORTS OVER TIME

As we turn to the issue of consistency in social supports over time, keep in mind respondents were interviewed during two very distinct phases of an evolving stress process. At the first interview they all had recently separated from their spouses and were in what could be considered the "impact" phase of divorce. By the time of the second interview, most had in some way accommodated to the necessity of change; their lives were not marked by the chaos and indecision of the earlier period. Given the situations our participants faced at the two points, we would anticipate considerable variation in the availability, use and perceived adequacy of supports over the course of the crisis. In order to answer this question three statistical techniques were used. The first, the £-test, looks at stability of group means but is relatively insensitive to individual variation. The second, correlations, looks at the individual's consistency from one time period to the next but disregards variation in the actual level of scores. The third, Chi-square analysis, looks at general patterns of stability and change. Average Group Differences over Time

One question that arises when studying support in a stress context is whether groups of people remain stable over time. For both men and women, there were no differences in group averages between first and second interviews for any of the four measures of support. In other words, when looked at from the perspective of group means, there was apparent consistency in each of the dimensions of social support for both men and women from one time of measurement to the next. Correlation over Time Consistency over time in social supports can be evaluated in ways that place greater emphasis on how individuals fare at each point. Correlations assessed whether knowing social support scores at the first interview gave us any idea of where the individual would be in the long run. For men, the correlations

202

CONTRIBUTING FACTORS

of Active Network (r=.76y £=.00) and Support Seeking (r=.59, £=.00) suggest reasonably high stability in these two measures between the first and second interviews. The measures for Perceived Support (r= .29, p = .00) and Confidant (r. = .27, £ = .00), however, were much less stable. A very different pattern is revealed in the correlations of social supports for women. Neither Active Network (r=.36, £=.00), nor Support Seeking (r=.36, £ = .00), nor the Confidant measure (r=.33, £ = .00) was particularly stable over time. Perceived Support (r=.51, £=.00) demonstrated the highest stability for women and this stands in marked contrast to the results for men. Overall, then, while the group averages held relatively constant over the years of the study, the results of the correlations showed that for certain social support variables there was less stability than for others. We also found that there was generally less stability in the social supports of women compared with those of men, and men and women also differed in terms of which measures were most stable. For men the most stable measures of support tapped the structural and behavioral dimensions of support, while for the women the variable that displayed the highest consistency tapped the perceptual dimension of support. The Chi-Square Analyses

We used the Chi-square statistic to provide information on how many people had roughly the same or different levels of support when we reinterviewed them. When men's Active Networks were compared from first to second interview, considerable consistency was apparent: 61% of the men manifested the same level of network activity at both measurement points. What change was found tended to be in the direction of an expanded network: 28% of the men reported having a larger Active Network while 11% reported a smaller Active Network. In short, the Active Network of men was relatively stable over time; when change occurred, it was more than twice as likely to occur in the direction of adding to the Active Network. This suggests that the experience of divorce may have prompted these men to enlarge their social networks. More than half of the men remained constant in their access to a confidant, with approximately 22% increasing and another 22% decreasing such access. Essentially the same distributions were found in Support Seeking and Perceived Support. The only difference is that slightly more men (27%)

SOCIAL SUPPORTS

2O3

decreased in the number of people talked to about the divorce and slightly more (25%) increased in their perception of support. Both these latter results are expectable since by the second interview the level of negative events had decreased. In other words, there was less need of Support Seeking and, as we have reported later in this chapter, people with lower levels of stress are more satisfied with their levels of support. Among women, about 45% fell into the same categories of Active Network at both time periods; nearly 23% reported fewer resources and 32% reported an expanded network. As was the case for men, in other words, when change did occur, it tended to be towards a larger rather than a smaller Active Network. This growth in network, evident for both sexes, suggests that men and women were accomplishing the third transitional task; reintegration back into society. They had become less isolated by the second contact. The number of confidants available to participants was quite stable. We found that over 18% of women declined in number of confidants reported, 13% reported an increase, and an impressive 69% remained in the same category at both times. In other words, what well may be the linchpin of support, the presence of people with whom one can truly be free and open, is likely to remain anchored through time. Stability was less evident in Support Seeking, where we found that only 46% of the women reported talking to the same numbers of people, 31% talked to fewer people and a surprising 23% talked to more people. Thus, somewhat less than half the women talked to the same number of people at both times; of those women who changed, more reported talking to fewer people than talking to more people at follow-up. Comparing men and women, men again appear to be somewhat more stable than women on this measure, as they were on the Active Network measure. However, the direction of change was similar for those men and women who changed on this measure, both reporting talking to somewhat fewer people about the divorce at follow-up. Despite the reductions, it is noteworthy that substantial numbers of men and women were in fact still seeking advice and support nearly four years after their separation. Nearly 52% of the women reported the same level of Perceived Support at both interviews, but at the second interview 20.5% reported less Perceived Support and 27.6% reported higher Perceived Support. Women and men were very similar in the amount of change they displayed in this measure of support from baseline to follow-up.

2O4

CONTRIBUTING FACTORS

Summing Up: Consistency over Time in Support

Substantial levels of consistency in the social supports of men and women are evident as they pass through the transition of divorce. This consistency is particularly striking given that participants were interviewed at a point of major upset as well as some years later. To expand on this last point, it should be emphasized that as a major, nonnormative life event, divorce by its very nature can be very disruptive of social networks and systems of support (Thoits 1982; Weiss 1975). Hence we had expected that the period of time between the first and second interviews would be accompanied by substantial change, not consistency, in social support. There are undoubtedly many factors which cooperated to produce the levels of stability observed in this study. Timing of measurement, for example, can influence findings (Thoits 1982; House 1981; see also chapter 6). Since it may take a while for stress to exert its full effect, measurements taken earlier in the transition could miss its impact on social support; measurements taken late in the process, similarly, could miss change entirely since support could have ebbed and flowed in the intervening time period. Since we first interviewed people an average of six months into the period of separation, we can assume that the transition would generally have progressed sufficiently to have exerted some impact on participants. And despite the fact that three and a half years is a relatively long time, our data suggest that divorce for most people was not entirely a "done deal" at follow-up. More possible is a durability to supports, even during times of strain. Perhaps the much-publicized changes associated with divorce—new friends, disruption of old networks, new lifestyles—have been overemphasized. A majority of people appear on the whole to keep their basic or "core" social connections intact throughout crises like divorce. Aneshensel and Frerichs (1982) and Ferraro, Mutran and Barresi (1984) provide longitudinal results supporting such a consistency hypothesis. There is no question that some divorces result in major changes in living patterns and social relationships, but this may be true for fewer people than the media and popular opinion would lead us to believe. Furthermore, some elements of support systems would seem to be naturally resistant to change: for example, relatives and those close enough to be considered confidants. Is Consistency Adaptive? A related point is that consistency in social support may itself contribute to the reductions in symptomatology we found during

SOCIAL SUPPORTS

205

the course of the study. If supports are relatively consistent in availability over the months and years of divorce, could this very consistency have a major stabilizing effect and facilitate coping with the many problems attendant on the breakup of a marriage? To follow up on this question, we computed change scores for confidant availability. What we found emphasized what we already know: men and women differ markedly in the significance of supports. Our scoring system divided people into those who had fewer confidants at the second interview, the same number, or more. We first looked at psychological symptoms reported at the second interview and found a tendency for men with greater consistency in availability of confidants to report more symptoms than other men, while women who were consistent over time reported fewer symptoms than other women. These findings indicate that consistency is in fact linked to better functioning, but only for women. Associated with greater symptomatology, consistency among men may indicate a continuing need of support rather than an ongoing part of normal life. A significant interaction between gender and consistency was also found for our measure of overall happiness as measured at follow-up. Unlike the case for symptoms, where women averaged higher scores than men, the sexes were equal in happiness at follow-up. As would be expected on the basis of the results for symptoms, the least happy men were those who were consistent in confidants since the first interview; the happiest were those who had experienced a decrease. In contrast, the happiest women were consistent in confidant availability. All in all, these results strongly suggest that an ordinary part of life for well-adjusted women is to maintain consistency in an inner circle of friends that, incidentally, usually includes more than one person. Men, on the other hand, may expand their inner circle as a means of resolving problems, but in normal circumstances operate with a more restricted group of close friends. DIFFERENTIAL ACCESS TO SOCIAL SUPPORT

One of our goals was to see if the stress process model proposed in chapter 3 could help explain why people vary in their access to social support. To explore this issue we tried to predict each of the social support measures on the basis of three sets of variables: social background factors, divorce context factors, and exposure to negative events. These three sets were chosen on the basis of literature and theory. Both social background factors and negative

206

CONTRIBUTING FACTORS

events, for example, have been reported to be associated with level of social support (e.g., Veroff, Douvan and Kulka 1981; Liem and Liem 1978; Dean and Ensel 1983; McFarlane et al. 1983). As for the divorce context variables, some researchers (e.g., Goode 1956; Chiriboga and Cutler 1977) report such measures to predict adaptation to divorce, while others find no association (e.g., Brown et al. 1980). Whether or not context variables (in this case, length of marriage, time separated, control over divorce) affected adaptation through their impact on the supports was of interest. It turned out that the three sets of predictors were generally unrelated to the four dimensions of social support. One individual measure, however, did exert a modest but significant effect on the availability of support: Negative Events. For both men and women, higher levels of stress were associated with people feeling they had less adequate levels of support. McFarlane et al. (1983) and Dean and Ensel (1983) noted a similar relationship between perception of support and stress exposure. To a small but significant extent, high stress was also linked to increased Support Seeking, but only for men. What seems to happen is that while men generally are less likely to seek support, increased stress leads them to increase Support Seeking to a level where their support system approximates that of women. As suggested by our findings for consistency, then, we see that while males may characteristically rely less on supports than do women, they do respond under conditions of stress. For divorcing men, the relationships between Negative Events and Perceived Support, and between Negative Events and Support Seeking provide additional information on their use of social support dimensions. Under conditions of high stress, we found that men talk to greater numbers of people about the breakup of their marriage, yet perceive themselves to be less supported. There are a number of possible explanations for this phenomenon. First, there may be a time lag involved. Even though men sought more support at times of greater stress it does not mean that they instantly received its benefits and this delay creates dissatisfaction. Second, men staggering under the burden of many stressors may devalue their support because it is not a magical cure. Finally, since men generally maintain a more frugal support system, they may have to work harder to obtain any support, and what they obtain may not suffice. This last possibility might be labeled a "deficit" hypothesis, since the point is that the support systems of men may be less adequate in the face of crisis.

SOCIAL SUPPORTS

207

THE ROLE OF SOCIAL SUPPORT DURING THE IMPACT PHASE OF DIVORCE

As a means of assessing the contribution of social supports in understanding adaptation, we examined four conceptual domains encompassed by the stress process model of divorce: (1) predisposing factors, (2) social stressors, (3) psychological mediators, and (4) social mediators. Level of psychological symptoms was the outcome measure of interest. 1. Predisposing factors are seen here as durable personal and social characteristics as well as structural characteristics directly linked to the divorce that can lead to greater or lesser vulnerability to its impact, for example, age, income, number of years married, number of months separated (e.g., Rabkin and Streuning 1976). Two separate sets of variables were used to measure predisposing factors: a set which measures social background characteristics and a set which measures divorce context characteristics. 2. Social stressors are those life events which occurred within the same general time frame of the separation and divorce and which had the potential to further disturb life patterns. The summary measure of all negative events reported for our Life Events Questionnaire over the year prior to the interview became our measure of social stress. 3. Psychological mediators here include the person's appraisal of the stressfulness of the divorce, and his or her self-esteem. 4. Social mediators, like psychological mediators, concern the resources available to an individual which can reduce or modify the impact of stress. We considered social supports as the mediators of interest, and included the four measures of support that have been described in prior sections: Active Network, Support Seeking, Confidant, and Perceived Support. In our model of the stress process, both predisposing factors and exposure to stress are hypothesized to precede and affect stress appraisal. Stress appraisal is placed before self-esteem in the ordering since appraisal is hypothesized to have a direct effect on self-esteem, but the literature is unclear about the exact role of supports in the stress process. Since support can potentially enter into the stress model at several points, our first procedure was designed to clarify at which point support may be most helpful.

2O8

CONTRIBUTING FACTORS

Supports and Some Precursors to Maladaptation

As we have just mentioned, social support can affect the stress process at several different points. In a series of hierarchical multiple regression analyses, we actually explored the effect of placing social supports at different steps in the model. By adding in social support at various stages of the hypothesized stress process, it was hoped that more could be learned about the role of social support as a mediator in the processes antecedent to the development of psychological symptoms. Specifically, we looked at the utility of supports when trying to predict several personal characteristics evident during the impact phase of divorce: divorce-related distress, self-esteem and psychological symptoms. When the level of divorce-related distress during marital separation was the dependent variable, the set of support variables was entered between social background, divorce context and negative events and distress, to see whether support made a difference to stress appraisal; it did not. In a second analysis, in which self-esteem became the dependent variable, measures of social support were entered immediately after distress; support did not make a difference in the level of self-esteem. Supports as a Predictor of Symptoms During the Impact Phase. Our next step was to consider what happened when we entered support measures immediately after self-esteem, with psychological symptoms as the outcome measure. It was at this stage of the process that social support had its effect. Social support was associated with the development of symptoms in divorcing persons, with higher levels of support being associated with fewer psychological symptoms. This relationship existed independently of the person's exposure to negative events, appraisal of stress or self-evaluation. Some variations in the role of support, however, were noted for men and women. Men. For recently separated men, neither social background nor divorce context factors were associated with psychological symptoms. In contrast, the experience of Negative Events during the past year was related very significantly to symptoms; Negative Events explained a substantial 15% of the variance in symptoms. To a lesser extent, appraisal of the divorce as stressful was also associated with symptoms: the appraisal of stress explained 4% of symptoms (p