Appraised Changeability of a Stressor as a Modifier of ...

1 downloads 0 Views 1MB Size Report
Rabkin & Struen- ing, 1976 .... 1982) and coping (Coyne, Aldwin, & Lazarus, 1981; Mitchell,. Cronkite ..... were tested simultaneously using Hotelling's T 2 test.
Journal of Personality and Social Psychology 1990, Vol. 59, No. 3, 582-592

Copyright 1990 by the American Psychological Association, Inc. 0022-3514/90/$00.75

Appraised Changeability of a Stressor as a Modifier of the Relationship Between Coping and Depression: A Test of the Hypothesis of Fit Peter P. Vitaliano, Deborah J. DeWolfe, Roland D. Maiuro, Joan Russo, and Wayne Katon Department of Psychiatry and Behavioral Sciences University of Washington The goodness of fitamong the appraised changeabilityof a stressor,coping, and depression in people with psychiatric,physicalhealth,work, and familyproblems was examined (N= 746).Itwas expected that problem-focused coping (as opposed to emotion-focused coping) would be used more and be more adaptive in situationsappraised as changeable as compared with situations appraised as not changeable. Although few relationshipsexistedbetween appraisaland coping, testsof fitbetween coping and depressed mood (maladaption)were much stronger.In people with nonpsychiatricconditions,problem-focused coping and depressed mood were negativelyrelated when a stressorwas appraised as changeable but were unrelatedwhen a stressorwas appraised as not changeable.Emotion-focused coping was positivelyrelatedto depressionwhen a stressorwas appraised as changeable. No general relationswere observed in the people with psychiatricconditions.

The cognitive-phenomenological model of stress provides a theoretical framework from which to test hypotheses and to conceptualize stress phenomena (Lazarus & Folkman, 1984). The model depicts distress as a function of a potentially stressful event/experience (stressor), the person's appraisal of that event, coping responses to appraisal, and reciprocal relationships among appraisal, coping, and distress. The stress response is predicted, in part, by the meaning that the responder assigns to the stressor (primary appraisal) and the judgments made about available coping options (secondary appraisal; Folkman, 1984). Both situational factors (cf. Rabkin & Struening, 1976; Thoits, 1983) and person factors (cf. Wrubel, Benner, & Lazarus, 1981) affect appraisal and coping. Two major categories of coping include problem-focused efforts directed toward altering the source of stress and emotion-focused efforts that regulate the person's emotional response to the stressor (Lazarus & Folkman, 1984). A third major coping strategy involves the seeking of social support for information, material assistance, or emotional support (Schaefer, Coyne, & Lazarus, 1981). The goodness of fit between appraisal and coping in response to a specified stressor is intrinsic to the cognitive-phenomenological model of stress. The term goodness of fit refers

to the appropriateness of a coping response, given a particular appraisal, and to the affeetive consequences of a match between appraisal and coping. For example, one would expect that the appraisal of a situation as one that could be changed (Folkman & Lazarus, 1980) would be linked to coping strategies that emphasize a problem-focused approach as opposed to emotion regulating or expressive approaches. The model would predict further that people who match their coping with their appraisal would be less likely to be distressed than those who mismatch such thoughts and behaviors. The major goal of this article is to examine the goodness of fit between the appraisal of a stressor (changeable versus not changeable), coping, and distress across six diverse samples. The samples were drawn from populations whose principal stressors were defined as being psychiatric, physical, work, or family problems. First, we review research that has investigated the appraisal of changeability, coping, and distress. For parsimony, we exclude three types of studies: (a) laboratory studies, because their findings may not be readily generalized to stress and coping in naturalistic settings, (b) studies that have not examined the appraisal of change, coping, and distress simultaneously, and (c) studies with total sample sizes of less than 40 people, because of their potentially unreliable conclusions.

Preparation of this article was supported by National Institute of Aging Grant AGO 6770-62 and the Harborview Anger Management Program. We thank Gay Armsden for her comments, Steven Dahlberg and Gordon Hall for the inclusion of their subjects, and Roslyn Siegel and Joanna Plattner for their assistance. Correspondence concerning this article should be addressed to Peter E Vitaliano, Stress and Coping Project, BB1538University Hospital, Psychiatry and Behavioral Sciences, RP- 10, University of Washington, Seattle, Washington 98195.

Literature Review Some conceptual overlap exists in the appraisals of a stressor as changeable and controllable. Although they are not synonymous, control can be viewed as a subcategory or condition o f the larger, less specific meaning associated with the word changeable. The appraisal of control has been defined as "the belief that one has at one's disposal a response that can influence the aversiveness of an event" (Thompson, 1981, p. 89). 582

APPRAISED CHANGEABILITY OF A STRESSOR Such an appraisal can reflect an assessment of objective reality or perceptions or beliefs about personal capability. In contrast, the term changeable is somewhat less associated with a specific locus of control. It can be a characteristic of the stressor or the individual's perception of his or her ability to modify it. In the current study, endorsement of the appraisal statement, "Is this situation one that you could change or do something about?" reflects an intention of control consistent with Thompson's (198 l) definition. Therefore, studies examining appraisals of control and changeable are both reviewed under the three guidelines established earlier. Relationships between appraisal, coping, and distress have been examined in a number of populations drawn from different age groups. In a study of130 children and adolescents 10-14 years of age, Compas, Malcarne, and Fondacaro (1988) investigated the perception of a situation as controllable. They found that when appraisal of a situation as controllable and coping with that situation were mismatched, distress-related behavioral problems were greatest. Specifically, the highest levels of behavior problems were found in combinations of low appraised control and high problem-focused coping and high appraised control and low problem-focused coping. Coping responses that involved the seeking of social supports were coded as either problem-focused or emotion-focused and were not analyzed separately. In a detailed investigation of the stress and coping processes of a sample of undergraduate students who were taking a midterm examination, Folkman and Lazarus (1985) used four emotions as the criteria in a series of regression analyses. Threat- and challenge-related emotions were experienced most frequently prior to the examination; harm- and benefit-related emotions were experienced most frequently following the posting of examination grades. The amount of control that subjects felt about the exam was positively correlated with problem-focused coping and an emphasis upon the positive and was negatively correlated with wishful thinking, distancing, and selfblame. Approximately 44% of the variance in challenge-related emotions was explained by four variables: feeling in control, the stakes associated with the exam, problem-focused coping, and tension reduction. Prior to the examination, subjects sought social support in the form of information. Immediately following the examination, students reported a significant decrease in informational support and a significant increase in emotional support obtained from social networks. Seeking social support contributed significantly to the variance in threat- and benefitrelated emotions. In another study using a sample of undergraduate students, Forsythe and Compas 0987) compared relationships between the appraisal of control, coping, and distress when the primary stressor was a major life event versus a daily hassle. They examined the goodness of fit between appraised control, problem-focused coping, and symptoms of distress. A significant interaction between appraisal and coping was found. Lower symptom scores were associated with relatively more problem-focused coping in events that were perceived as controllable and with relatively more emotion-focused coping in events perceived as less controllable. This significant interaction effect was explained only when the stressor was a major life event (e.g., "death of a family member; "entering college"). Thus, the he-

583

cessity of goodness of fit for lowered distress was not supported for daily events or hassles (e.g, "doing poorly on an exam or paper"). In a community sample of 85 married couples, significant variance in psychological distress was explained by primary appraisal (what was at stake in the situation), coping, and selected person factors such as mastery and interpersonal trust (Folkman, Lazarus, Gruen, & DeLongis, 1986). Secondary appraisal, including viewing the situation as changeable, had low correlations with psychological distress and was not significantly predictive of variance in distress when entered with other variables. Folkman and Lazarus 0986) compared these same people on high versus low levels of depressive symptomatology. They found that those high in depressive symptoms appraised themselves as having more at stake and as needing to hold themselves back from doing what they wanted to do. There were no overall differences between the groups on the appraisal of stressors as changeable. Rationale and Constructs The goodness of fit between the appraisal of a stressor as changeable or controllable and adaptive problem-focused coping strategies has received minimal attention from researchers, especially in psychologically challenged populations. Moreover, although some empirical support exists for the hypothesis that a mismatch of coping and the appraisal of change or control is related to greater distress in nonpsychiatric samples (Compas et al, 1988; Forsythe & Compas, 1987), little is known about the fit hypothesis in psychiatric samples. The present study investigates the goodness-of-fit hypothesis in both nonpsychiatric and psychiatric samples. We examine a criterion variable, depression, as it relates to the match between the appraisal of a stressor as changeable or not changeable and coping that is problem-focused, emotion-focused, or involves the seeking of social support. Coping efforts have generally been conceptualized as intrapsychic or personality-based processes (Byrne, 196 l; Haan, 1977; Miller, 1987) or cognitive and behavioral efforts to change or manage a stressor (Lazarus, & Folkman, 1984). The present study examines the cognitive-phenomenological model of stress, in which coping has been divided into problem-focused and emotion-focused strategies (Folkman & Lazarus, 1980). The mediating role of social support in the stressor-distress relationship has also received considerable attention in the stress literature (cf. Cohen & Wills, 1985; Leavy, 1983; Thoits, 1986). We specifically investigated the seeking of emotional and informational support in response to a particular stressor, as opposed to the experience of such support. The appraisal of a stressor as one that could be changed reflects an individual's cognitive set in response to a particular situation. The appraisal of a stressor as changeable was one of four secondary appraisal items included in Folkman and Lazarus's (1980) original investigation of stress and coping in a sample of community residents. Problem-focused coping strategies were most frequently used in situations that were appraised as changeable. Although the appraisal of change has been examined widely, it has not received as much interest as the appraisal of control. In an analysis of the construct of personal control,

584

VITALIANO, DEWOLFE, MAIURO, RUSSO, KATON

Folkman (1984) distinguished between situational control and personal control (generalized beliefs o f control or mastery). Situational control originates in the demand characteristics o f the stressor. When situations are highly ambiguous and do not offer clear indicators for action, person factors are more influential in the perception of control (Sehank & Abelson, 1977; Shalit, 1977). The appraisal of a situation as changeable was examined in the current study because it has a demonstrated relationship with patterns of coping (Coyne, Aldwin, & Lazarus, 1981; Folkman & Lazarus, 1980) and it is related to the theoretically rich construct of control. For example, an extensive literature exists on the perception of control over life events (Rothbaum, Weisz, & Snyder, 1982; Rotter, 1966), self-efficacy and outcome expectancy, (Bandura, 1977), and learned helplessness (Abramson, Seligman, & Teasdale, 1978). In the current study, depressed mood was selected as the criterion (i.e, indicator of distress or maladaptation) because of the extensive literature linking it with cognitive processes (Abramson, Seligman, & Teasdale, 1978; Beck, 1976; Beck & Epstein, 1982) and coping (Coyne, Aldwin, & Lazarus, 1981; Mitchell, Cronkite, & Moos, 1983; Mitchell & Moos, 1984; Billings & Moos, 1984; Folkman & Lazarus, 1986). Unlike a psychiatric disorder of depression, a diagnosable condition that would be primarily identifiable in psychiatric populations, some range of depressed mood would be expected across a variety o f samples. Folkman (in press) has suggested that an outcome of depressive disorder may be too distal and that a measure of more proximal indicators--such as affect, mood, or feelingwis more likely to relate to the specific effects o f coping. Hypotheses Life events appraised as uncontrollable have been more strongly associated with reports o f depression than have events appraised as controllable (cf. Thoits, 1983). The perception of control has been shown to be critical. In fact, in some instances, the perception of control has been shown to be more influential in its relationship to depression than the actual controllability of a situation (Taylor & Brown, 1988). On the basis of the concept of fit, our expectation was that problem-focused coping would be used proportionately more in situations that were appraised as changeable, as compared to situations that were appraised as unchangeable. The converse would be true for the use o f emotion-focused coping in situations appraised as changeable as compared to unchangeable. We also expected that problem-focused coping would be a more adaptive strategy to use than emotion-focused coping in situations appraised as changeable. We hypothesized that such a match or fit would be reflected in the following relationships: (a) a negative correlation between problem-focused coping and depression that would be greater when the situation was appraised as changeable as opposed to unchangeable and (b) a positive correlation between emotion-focused coping and depression that would be greater when the situation was appraised as changeable as opposed to unchangeable. Existing research on the appraisal of change, coping, and distress provides scant significant findings regarding the role of the seeking of social supports. As such, we anticipated either

nonspecific or insignificant relationships between appraisal, seeking of social supports, and depression. The appraisal o f change may be less proscriptive of the seeking o f social supports, when c o m p a r e d with the stronger association o f appraisal of change and problem-focused coping.

Method

Participants Several procedural steps were taken to test our hypotheses. A multicohort design (N = 746) was used in which samples were examined representing four problem categories. This design allowed us to test the generalizability of our findings over different life predicaments. Six groups of people were included who have been identified in the stress literature as experiencing levels of stress that require adjustment. Problems were examined in four of the five major areas considered in the life stress literature (Rahe, 1977): personal (here, psychiatric), physical health, work, and family. Financial problems were not examined in the current study. As such, this investigation follows recent recommendations of Barnett and Gotlib (1988). They suggested that the elucidation of the relationship between coping strategies and depression requires inclusion of the stressful event, thus providing the context in which the coping occurs. The people in each of the samples had experienced a common stressful event (e.g.,physical symptoms) or common predicament (e.g.,attempting suicide, referred for treatment because of anger/ dyscontrol problems, caring for an Alzheimer's patient).

Psychiatric Problems Three groups of psychiatric patients who had serious adjustment problems were examined. These included patients with anger/dyscontrol and assault problems, sexual offenders, and suicide ideators and attempters. These groups represent definitive examples of maladaptive coping in that their emotions and behaviors are ofconcern to both the patients and to others. Moreover, the samples examined were all cognitively intact and nonpsychotic, thus enhancing the reliability of their reports and their suitability for this type of study. Anger/dyscontrol. This sample included 371 psychiatric outpatients who were referred to a specialized treatment program for anger and dyscontrol problems at Harborview Community Mental Health Center, Seattle, Washington. Approximately 92% of the sample had independently documented histories of coping problems as evidenced by incidents of interpersonal conflict and assault (Maiuro, Cahn, & Vitaliano, 1986; Maiuro, Cahn, Vitaliano, Wagner, & Zegree, 1988). All patients met Diagnostic and Statistical Manual ofMental Disorders (DSM-IIL"American Psychiatric Association, 1980) criteria for adjustment reactions, impulse control disorder, or personality disorder. Demographic data for this sample are provided in Table 1. Suicidal. This sample included 42 suicide attempters and 18 suicide ideators who were under supervision at Harborview Hospital's inpatient psychiatric ward. All patients had seriously considered or attempted suicide within the past week (Maiuro, O'Sullivan, Michael, & Vitaliano, 1989). In line with previous studies (e.g, Dressier, Prusoff, Mark, & Shapiro, 1975; Paykcl & Rassaby, 1978; Weissman, Fox, & Klerman, 1973), a suicide attempt was defined as any clearly documented, intentionally self-inflicted, life-threatening injury (including by ingestion). Demographic variables for this sample are given in Table 1. Sex offenders. This sample consisted of 106 male patients at Western State Hospital, Fort Steillacoom, Washington. All had documented histories of sexual assault and were under treatment in a locked inpatient facility. All patients met DSM-III criteria for either

585

APPRAISED CHANGEABILITY OF A STRESSOR Table 1

Demographic Variables for the Six Samples Education Gender Groups Psychiatric problems Anger/dyscontrol Sex offenders Suicide patients Physical health problems Pelvic pain/tinnitus patients Work problems Camp counselors Family problems Spouse caregivers

Age

Marital status

n

Male

Female

M

SD

Single

Married

More than high school

High school or less

371 106 60

89.7 100 51.7

10.3 0 48.3

31.92 31.50 33.73

8.57 8.25 12.14

78 72.6 83.3

22 27.4 16.7

33.4 27.4 33.4

66.6 72.6 66.6

65

0

100

43.16

16.51

17.6

82.4

17.6

82.4

43

100

0

16.79

1.42

100

0

0

63.7

66.24

8.68

0

100

101

36.3

36.3

100 63.7

paraphilia, antisocial personality or dependent personality disorders (Hall, Maiuro, Vitaliano, & Proctor, 1986). Psychiatric patients in the anger/dyscontrol, suicide, and sex offender samples were examined prior to treatment. These groups were also screened to exclude patients with medical problems. Demographic variables for these samples are presented in Table I.

years and group leaders for a minimum of 3 years. They reported distress from a variety of work sources (e.g~ daily responsibility for 30 scouts, stage fright, and minimal personal time; Dahlberg, 1987). Demographic data for this sample are provided in Table I.

Physical Health Problems

In the present study, spouse caregivers of patients with Alzheimer's disease (AD) were used to represent family stress. As AD progresses, patients are less and less able to care for themselves, thus becoming more and more dependent on their caregivers. Researchers have documented both deleterious psychosocial effects (Deimling & Bass, 1986; Morycz & Blumenthal, 1983; Rabins, Mace, & Lucas, 1982; Zarit, Reever, & Bach-Peterson, 1980) and physiological effects (KiecoltGlaser et al, 1987) of AD patient demands on caregivers. Caregivers. The sample of caregivers consisted of 101 spouses of patients with AD who were participants in studies on stress vulnerability in AD Patients' families (Coppel, Burton, Becker, & Fiore, 1985) or AD Patient-spouse interactions (Vitaliano, M a i m , Ochs, & Russo, 1989). Research on caregivers has focused on several stressors as a result of the AD patient's illness. These include responsibility for housework, feeding/grooming patient, paying bills, and responding to patient's agitation (Vitaliano, Maiuro, Ochs, & Russo, 1989). Table i contains demographic information for this sample.

Two groups of patients with physical health problems were examined: people with tinnitus (ringing in the ears) and pelvic pain. At the time of assessment, these patients were all experiencing considerable pain or discomfort or both and were unaware o f the cause and prognosis of their condition. Pelvic pain/tinnitus. This sample included 29 patients with chronic pelvic pain for at least 3 months' duration and 36 patients with reported tinnitus, a problem often associated with high-frequency hearing loss because of noise exposure. All pelvic pain patients were examined in the University Hospital Gynecology Clinic in Seattle (Walker, Katon, Harrop-Grifliths, Holm, Russo, & Hickok, 1988). These women were awaiting an uncomfortable laparoscopic exam in which a flexible tube is passed through the umbilicus into the pelvic region. The tinnitus sample was recruited at the University Hospital Chronic Tinnitus Clinic (Harrop-Grifliths, Katon, Dobie, Sakai, & Russo, 1987). Several of the patients were quite distressed by their tinnitus, claiming it caused disability in their social, vocational, and family systems. Table 1 contains demographic data for this sample. Psychological distress has been reported in both of the physical health conditions examined. Patients with tinnitus have been shown to have high rates of depression (Stephens & Hallam, 1985; Sullivan et al, 1988; Tyler & Baker, 1983) and pelvic pain patients have been described as having high levels of both anxiety and depression (Castelnuevo-Tedesco & Krout, 1973; Magni, Salmi, DeLeo, & Ceola, 1984; Renaer, Vertommen, & Hijs, 1979; Walker et al, 1988). However, these samples differ from the non-medically ill psychiatric patients in this study because they presented with medical problems and have more variable rates of distress. Campcounselors. A sample of 43 people with work stress was examined. This sample consisted of Boy Scout camp counselors, a group with a high level of responsibility and repercussions associated with a failure to perform. They had considerable performance demands, high workload, and minimal control over the amount and pace of their work. The subjects were interviewed at their work sites in three summer camps in western Washington. They all had been Boy Scouts for 10

Family Stress

Measures Standardized measures of appraisal, coping, and depressed mood were administered to all samples. The appraisal of changeability was examined, using a one-item dichotomous measure, as used in the Folkman and Lazarus (1980) study. Coping was assessed, using a revised version of the Ways of Coping Checklist (WCCL; Vitaliano, Russo, Carr, Maiuro, & Becket, 1985), which was originally developed by Folkman and Lazarus (1980). Following the paradigm used by Folkman and Lazarus (1980), subjects were asked to respond to the coping items with respect to the major serious stressor in their life. In 4 of the 6 samples, all subjects in the sample were asked how they had coped (or were coping) with the problem identified in that sample (i.e, being a camp counselor, having pelvic pain/chest pain/tinnitus, committing sex offenses, attempting suicide). However, subjects in the spouse caregiver and anger/dyscontrol samples additionally were asked to list their particular problem. For example, the patients with anger/dyscontrol problems listed their particular interpersonal stressor, and the caregivers of AD patients listed their major stressor. This provided a basis

586

VITALIANO, DEWOLFE, MAIURO, RUSSO, KATON

for examining the role of specific problems within these two samples. In the revised version of the WCCL, four psychometric properties were examined. These included the reproducibility of the factor structure of the original scales, the internal consistency reliabilities and intercorrelations of the scales, and the construct and concurrent validity of the scales) Although there are five scales on the revised WCCL, analyses for the present study were not performed separately on each of these scales. To reduce the number of analyses, the Problem-Focused and Seeking Social Supports scales were left intact, but the three emotion-focused scales (Wishful Thinking, Self-Blame, and Avoidance) were grouped into one scale. These scales were also combined because a literature supports hypotheses about emotion-focused coping (Folkman & Lazarus, 1980; Pearlin & Schooler, 1978) and little data are available to support hypotheses about the separate scales. The raw score for a particular coping scale is computed by summing the appropriate item scores on that scale. Vitaliano, Maiuro, Russo, and Becker 0987) have shown that such scores are biased by the total number of efforts endorsed on all scales and therefore poorly reflect one's emphasis on specific strategies. However, if one corrects raw scores for the total number of efforts endorsed on all strategies, one is able to observe more clearly coping differences across pathological and nonpathological samples. Such scores, labeled relative scores, represent the percentage of efforts made on specific strategies. To compute the relative scores of the three scales, one first obtains the mean item scores for each scale (ie., raw score divided by number of items). Mean item (MI) scores are not biased by the number o fitems constituting a particular scale. A specific relative score is then computed by dividing the mean item score in question by the sum of the three mean item scores for each scale. For example, the relative score for problemfocused (PF) coping is PF% =

MI Sum of MI Scores

As noted earlier, only problem-focused coping, emotion-focused coping, and seeking social support were examined here because of the theoretical literature associated with these strategies and our attempt to make the analyses parsimonious. Psychological distress was operationalized as depression because it is a hallmark of mental disorder (Woodruff, Goodwin, & Guze, 1984). In this article, a continuous measure of depressed mood was chosen in order to tap a full range of symptoms and experiences. This construct was assessed using the Symptom Checklist (SCL-90) Depression Scale (Derogatis, 1977) and the brief version of the Beck Depression Inventory (BDI; Beck & Beck, 1972). These measures were chosen for theoretical as well as practical reasons. Respectable psychometric properties have been reported for both the SCL-90 (Derogatis & Cleary, 1977; Derogatis, Rickels, & Rocks, 1976) and the brief BDI (Beck & Beamesdorfer, 1974; Beck & Beck, 1972; Gallagher, Nies, & Thompson, 1982). These measures are appropriate for both nonpatient and patient samples, and they are easily administered. The BDI was administered to five of the samples, whereas the SCL-90 Depression Scale was administered to one of the samples.

Statistical Analyses Frequency distributions were obtained for all variables to examine them for subsequent inferential analyses. The internal consistency coefficients of the measures were calculated to determine the degree to which measurement error might suppress the power of the inferential analyses. The coping and depression scales were shown to have adequate reliabilities across all samples (alphas > .7). Frequency distributions were obtained for the coping scales for the two levels of appraisal (changeable and not changeable) across the six

samples. We also obtained the percentages for these two appraisals for the six samples. The distributions were checked for skewness and discontinuity to mitigate subsequent spurious correlations. Descriptive statistics were obtained for the 12 subsamples (six samples by two appraisals). For each subsample, the distributions were generated for the following four variables: problem-focused coping, emotion-focused coping, seeking social supports, and depressed mood. The breakdown of appraisal in each sample determined the sample sizes of the 12 subsamples. Tests of the hypotheses consisted of two series of analyses. The first statistical analysis consisted of comparisons of coping efforts across the two appraisal levels in each of the six samples. These preliminary tests examined the degree to which goodness of fit existed between coping and appraisal. In this manner, it was determined whether the mean coping score was significantly higher under matched (problem focused/appraised as changeable; emotion focused/appraised as not changeable) versus mismatched (problem focused/appraised as not changeable; emotion-focused/appraised as changeable) conditions. To minimize the number of tests performed, three criterion variables (problem-focused coping, emotion-focused coping and depression) were tested simultaneously using Hotelling's T 2 test. These tests were performed with depressed mood as a criterion variable because it has been argued that differences in depression between the groups could affect assessments of coping differences (Vitaliano, Maiuro, Russo, Katon, DeWolfe, & Hall, 1990). 2 Once we performed the tests of differences in the mean coping scores, we provided another test of the goodness-of-fit hypothesis. Within each sample, the correlations of depression with coping were obtained and tested for differences across the appraisal groups of change versus no change. The tests involved Fisher's two-sample Z test (Hays, 1970). Because the hypotheses were directional, one-tailed tests were performed. Analyses were performed first within samples to limit heterogeneity of stressors across individuals. However, to increase power and to maintain at least 40 cases per group (a requirement noted in the Introduction), the two samples with medical problems (pelvic pain and tinnitus) were combined. Where possible, we also examined the degree to which the type of stressor, within the samples of anger/dyscontrol patients and caregivers, modified the results for goodness of fit. These two samples were the only samples in which the stressor varied within groups.

1 The revised scales were developed using a combination of factoranalytic and rational approaches. Because the medical-student sample was the largest (n = 425) in relation to the number of items (r = 68), it was used for this analysis. The other two samples were then used to determine the degree to which the resulting scales, in relation to the original scales, were internally consistent and intercorrelated. The Blamed Self Scale was reproduced entirely. Three other factors resulted in three scales, labeled Problem Focused, Wishful Thinking, and Seeks Social Supports. These factors were so named because they contained high loading items that were on the original scales or were, on face, representative of these respective strategies. The final scale, labeled Avoidance, contained items from the original Mixed, Wishful Thinking, and Minimize scales. The revised scales were consistently shown to be more reliable (.83 vs..76 in the two test samples) and to share substantially less variance (mean shared variance = 12% vs. 20% in Alzheimer's disease spouses and 28% vs. 42% in psychiatric outpatients) than the original scales. 2 Because the three coping measures are ipsative (i.e, they sum to 100), their covariance matrix could not be inverted. Therefore, one of the scales (Seeks Social Supports) was analyzed separately for each sample.

APPRAISED CHANGEABILITY OF A STRESSOR Results

Relationships Between Stressors and Appraisal, Coping, and Depression Within Samples The frequency distributions o f the specific identified stressors obtained from the samples o f anger/dyscontrol patients and spouse caregivers were as follows. Among the psychiatric patients, 60% listed a person within their family, 23% listed a person at work, and 17% listed both people in the family and people at work to be their major source o f stress. Among the spouse caregivers, there were also three general sources o f stress: 71% o f the subjects listed an event/experience that was directly attributed to their spouse (e.g, patient cannot care for himself), 27% listed an event not directly attributed to their spouse but that resulted from the disease (e.g, filling out applications to secure medicaid for nursing home fees), and 2% listed problems unrelated to the patient's disease (e.g., breakup o f child's marriage). Analyses of variance (ANOVAS)indicated that no differences in coping, appraisal, or depression existed as a function o f these identified stressors within the spouse caregiver sample or patients with anger/dyscontrol problems sample.

Relationships Between Appraisal and Coping The sample sizes for the appraisal groups o f changeable and not changeable within each of the six samples are shown in Table 2. In addition, the means and standard deviations for coping and depression are provided for the two appraisal levels. Of the six T 2 tests performed, only three were significant: those for the suicidal patients, F(3, 56) = 5.02, p < .01; the caregivers, F(3, 97) = 3.00, p < .03; and the camp counselors, F(3, 39) = 3.45, p < .03. Within the group o f suicidal patients, people who appraised their stressor as changeable had higher problem-focused scores, F(1, 58) = 5.02, p < .004, but lower emotion-focused scores, F(I, 58) = 15.4, p < .001, than those who appraised their stressor as not changeable. Within the caregiver group, people who appraised their stressor as changeable were more depressed than those who appraised their stressor as not changeable, F(1, 99) = 9.17, p < .003. 3 Among the camp counselors, people who appraised their stressor as changeable had significantly lower emotion-focused coping scores F(I, 41) = 7.28, p < .01, and depression, F(I, 4 l) = 5.61, p < .02, than those who appraised their stressor as not changeable. These results suggest that the hypothesis o f fit between coping and appraisal was not generally supported in these analyses. We did not hypothesize specific relationships between appraisal and seeking social supports. In fact, the univariate tests revealed that the Seeks Social Supports scores were not significantly different for the appraisal groups in the anger/dyscontrol, sex offender, physical health problems, and spouse caregiver groups. Higher seeking social support scores were observed, however, for the condition of change versus no change in the sample o f suicidal patients, F(I, 58) = 13.56, p < .001, and the camp counselors, F(I, 41) = 7.08, p < .01.

Differences in Correlations of Coping and Depression Across the Appraisal Conditions Problem-focusedcoping. Table 3 contains the correlations of depression with coping for the two appraisal levels within each

587

sample. When the stressor was appraised as changeable, the correlations o f depression with problem-focused coping were negative and significantly different from zero in the three nonpsychiatric distressed samples: patients with physical health problems, spouse caregivers o f patients with AD, and camp counselors. In contrast, the correlations in the not changeable condition were either much smaller or insignificant in these three samples. The negative correlation of problem-focused coping with depression in the changeable condition (as stated in the research hypothesis) was significantly greater than the same correlation in the not changeable condition in these samples: camp counselors (Z = 1.73, p < .05), spouses o f patients with A D (Z = 2.25, p < .01), and patients with physical health problems (Z -- 1.88, p < .05). In contrast, the pattern o f results achieved in the psychiatric samples was less consistent. Although negative correlations between problem-focused coping and depression were consistently observed in the three groups, they were not systematically or significantly higher in the changeable conditions. In fact, in one case (the anger dyscontrol sample), there was a trend for the relationship to be stronger in the not changeable condition (Z = - 1.31, p < . 10). These results provided support for the hypothesis of fit between the appraisal o f change, problem-focused coping and depression, but only for the nonpsychiatric distressed samples. Emotion-focused coping. Support for the hypothesis o f fit was not as strong for emotion-focused coping as it was for problem-focused coping. In all samples, the positive correlation of emotion-focused coping and depression, under the condition o f change, was not significantly greater than the positive correlation o f emotion-focused coping and depression, under the condition of no change. However, supportive trends did occur. In the changeable condition, marginally significant positive correlations of emotion-focused coping and depression were obtained in the three nonpsychiatric distressed samples: patients with physical problems ( p < .09), spouses of patients with A D ( p < .07), and the camp counselors ( p < .10). In the not changeable condition, no relationships between emotion-focused coping and depression were obtained in two of the three nonpsychiatric samples: the camp counselors and spouses o f AD patients. Given the same trend for emotion-focusedcoping in all three nonpsychiatric distressed samples, we used Rosenthars (1984) procedure to collapse these samples to obtain better estimates of the correlations within the changeable and not changeable conditions. This was done to determine if the lack of significance was the result o f weak power in each sample. The rs in each appraisal by sample combination were converted to Fisher's Zs, the Zs were summed and averaged, and the average for each appraisal combination was used in the two-sample Z 3 Within the specific stressors in the caregiver sample, differences in coping did not exist between the appraisal groups. This could have resulted because of the small number of carcgivers with each type of stressor (72 caregivers listed a stressor directly attributed to their spouse, 27 caregivers listed a stressor that resulted from the disease itself, and 2 caregivers listed strcssors unrelated to the patient or to the disease), which had to be further broken down by the appraisal of change versus no change.

588

VITALIANO, DEWOLFE, MAIURO, RUSSO, KATON test. The mean Zs for changeable and not changeable were .58 (n = 87) and .23 (n = 122), respectively. These Fisher's Zs (correlations) were, in fact, significantly different (Z = 2.50, p < .01), supporting the goodness-of-fit hypothesis. As with problem-focused coping, no significant differences in emotion-focused coping occurred between the appraisal conditions in the three psychiatric samples. In fact, in the one instance in which a trend was observed (sex offenders), the direction of differences was opposite to the hypothesis in that the correlation was positive and high between emotion-focused coping and depression in the not changeable condition but not in the changeable condition.4 Seeking social supports. We did not hypothesize differences in the correlations of seeking social supports and depression under the conditions of change versus no change. No significant differences were observed in the exploratory comparisons of these correlations.

.o

Discussion

d

_ ~ ~

t~

o

O

o

- ~ ~a

¢q

~

.

~

O

Goodness of fit has been described on two levels: as the fit between the actual demands associated with a stressor and its appraisal and as the fit between that appraisal and subsequent coping (Folkman, in press). To determine fit, Forsythe and Compas (1987) incorporated adaptational outcomes with appraisal and coping. They used distress level as an indicator of the degree of match or mismatch between appraisal and coping. We investigated goodness of fit using two related but not identical methods. These involved (a) comparing the mean coping scores under conditions in which the problem was appraised as changeable or not changeable and (b) examining differences in the correlations of coping and depression across the appraisal groups. We found that support for the goodness of fit of coping and appraisal was dependent 9 n which components of the stress model were examined, the method used, and the sample or type of problem studied. The current findings suggest that goodness of fit is best demonstrated when appraisal, coping, and distress are considered simultaneously using correlational methods. When we simply compared coping means across appraisal groups, without considering their relationships with depression, fit was not observed. The cognitive-phenomenologicmodel of stress involves emotional factors, as well as cognitive and behavioral factors. An examination of the latter without the former would appear 4 Where possible, the influence of stressor type was examined with respect to the relationships of coping and depression in each appraisal group. In the psychiatric patients, coping and depression were not differentially related between appraisal groups in those patients who listed a person within their family as the source of their anger/stress (n = 223), a person at work (n = 85), or both people at work and people in their family (n = 63). In the caregivers, only one of the three stressor groups was large enough to examine with respect to stressor type: These were the people who listed an event/experience directly attributable to their spouse's illness (n = 72). The correlations between problem-focused coping and depression in the two appraisal conditions had patterns similar to the overall sample of caregivers (change r = -.56, n = 23 vs. no change r = -.22, n = 49); however, these were not significantly different because of the smaller sample size in these groups versus the overall sample size.

589

APPRAISED CHANGEABILITY OF A STRESSOR Table 3

Correlations of Coping With Depression in Groups With Appraisals of Changeable VersusNot Changeable Problem-Focused Not

Sample

changeable

Emotion focused

Seeks Social Supports Not

Changeable

Z

changeable

Not

Changeable

Z

changeable

Changeable

Z

Distressed psychiatric samples Sex offenders Anger/dyscontrol

-.31 -.48

- . 15 -.36

-0.75 - 1.31

-.50 -.27

- . 17 - . 19

- 1.70 -0.63

.45 .48

.18 .38

1.38 0.95

Suicide patients

-.27

-.47

0.73

.01

-.30

0.97

.15

.42

-0.92

-.35 -. 11 -.23

0.41 -0.02 0.35

.27 .17 .21

.57 .44 .55

-1.35 - 1.44 -1.29

Distressed samples Physical health problems Spouses of AD patients Camp counselors

-.22 -. 10 -.20

-.63 -.52 -.63

1.88" 2.25** 1.73"

-.25 -. ! 2 -.12

Note. A D = Alzheimer's disease.

*p