Multivariate Behavioral Research, 1985, 20, 3-26 The Ways of Coping Checklist: Revision and Psychometric Properties Peter P. Vitaliano, Joan Russo, John E. Carr, Roland D. Maiuro and Joseph Becker Department of Psychiatry and Behavioral Sciences, University of Washington ‘This study examined the psychometric properties of the “original” seven factored scales derived by Aldwin et al. from Folkman and Lazarus’ Ways of Coping Checklist (WCCL) versus a revised set of scales. Four psychometric properties were examined including the reproducibility of the factor structure of the original scales, the internal consistency reliabilities and intercorrelations of the original and the revised scales, the construct and concurrent validity of the scales, and their relationships to demographic factors. ‘These properties were studied on three distressed samples: 83 psychiatric outpatients, 62 spouses of patients with Alzheimer’s disease, and 425 medical students. The revised scales were consistently shown to be more reliable and to share substantially less variance than the original scales across all samples. In terms of construct validity, depression was positively related to the revised Wishful Thinking Scale and negatively related to the revised Problem-Focused Scale consistently across samples. Anxiety was also related to these scales, and in addition, it was positively related to the Seeks Social Support Scale across samples. The Mixed Scale was the only original scale that was consistently related to depression and anxiety across the three samples, Evidence for concurrent validity was provided by the fact that medical students in group therapy had significantly higher original and revised scale scores than students not participating in such groups. Both sets of scales were shown to be generally free of demographic biases. The Ways of Coping Checklist (WCCL) is a relatively new mea- sure of coping that was derived from Lazarus’ transactional model of stress (Aldwin, Folkman, Shaefer, Coyne & Lazarus, 1980; Folkman & Lazarus, 1980). In this model an event is considered stressful when a person appraises it as potentially dangerous to his/her psychological well-being (Lazarus, 1966). Such an appraisal may be influenced by a person’s beliefs or personality and generate cognitive expectancies which affect both emotion and behavior. Once a situation is perceived as potentially harmful, a person decides how dangerous it is and what kind of coping strategy to use to reduce the potential harm. Furthermore, Aldwin et al. (1980) have noted “. . . coping is affected not only by the appraisal (and vice-versa), but also by the results of coping efforts (p. 1).” Although, researchers have paid considerable attention to relationships between coping strategies and psychological, physical, and social well-being (Andrea- son & Morris, 1972; Antonovsky, 1979; Averill & Rosenn, 1972; Carson, 1969; Cohen & Lazarus, 1979; Lipowski, 1970), problems in ‘This research was supported by three grants: Biomedical Research Support Grant RR05432; Mental Health Grant 33779, National Institute of Mental Health; and, the Graduate School Research Fund, University of Washington. ‘The authors would like to thank Dr. Susan Folkman for her advice on the original manuscript. JANUARY, 1985 3Vitaliano et al, the measurement of coping per se have not been as vigorously pursued. The stress and coping paradigm on which the 68-item WCCL was developed requires that the subject focus on a current serious stressor. In its original form the WCCL contained two rationally derived scales: problem-focused coping—the management of the source of stress and emotion-focused coping—the regulation of stressful emotions. In a more recent analysis of the 68 items, Aldwin et al. (1980) used a principal components analysis with varimax rotation to empirically elucidate coping strategies. In addition to a problem-focused coping factor, six kinds of emotion-focused coping factors were derived. Thus, in its present form, the WCCL contains seven scales: Problem-focused (“made a plan of action and followed it”), Wishful Thinking (“wished you could change the situation”), Growth (“changed or grew as a person in a good way”), Minimize Threat (“making light of the situation”), Seeks Social Support (“talked to others and accepted their sympathy”), Blamed Self (“felt responsible for the problem”); the Mixed Scale contains both avoidant strategies (“refused to believe it had happened”) and help-seeking strategies (“sought advice”). In addition to the 68 coping items, the WCCL contains four items which allow the subject to appraise his or her current serious stressor in terms of four dimensions. This research began as an attempt to examine the psychometric properties of the WCCL on three different types of samples. However, prior to the empirical examination of the WCCL, we encountered several methodological problems. The original scales were developed by factor analyzing 68 items on only 100 middleaged subjects drawn from the general population; this raises questions regarding the stability of the factors, as well as the clinical generalizability and construct validity of the scales. In keeping with this concern, certain scales contained items which appeared to lack face validity. The Mixed Scale, in particular, was difficult to interpret because it contained both avoidant and help-seeking strategies. In addition, the intercorrela- tions between the scales (when corrected for attenuation) were in many cases very high making it difficult to assess coping multidimen- sionally. Finally, from a pragmatic perspective, subjects complained that the checklist was too long. For these reasons, a revised version of the WCCL was developed and compared with the original version in terms of its psychometric properties. Method Subjects Three groups participated in the study: (1) outpatients at the Harborview Community Mental Health Center, Seattle, who had been 4 MULTIVARIATE BEHAVIORAL RESEARCHVitaliano et al, referred to a specialized treatment program for anger and dyscontrol problems (n = 83)!, (2) spouses of patients with senile dementia of the Alzheimer’s type (SDAT), who were being seen supportively at the University Hospital, Seattle (n = 62)?, and (3) students who were in their first or second year at the University of Washington Medical School in 1980 (n = 235) and 1982 (n = 190). anoacaphic and teristics of Medical sty Characterietice _Madlcal students Spouses of SDAT patients Psy. ta Gender Mate (9) 50.0 36.8 oL.0 Perale w.0 oz 90 re 1 (years) 2.8 65.8 22.5 2 at 9.2 6 Bducation 2 16 years (2) 100 23.5 3.7 12 yoars ° 60.3 37.0 Macétal statue Married (a) a 100 ae Not maceieg 38 ° 75.6 Anxiety, 4 44.087 55.68" 3.39 30 2.97 1.7 13.88 Depression « 20.2%, 10.09% 82 @ 5.73, 4.90 na Note. “sci~9 outpationt T-scores. SCL-90 non-patient T-scoces, spt raw scores, HDR ‘Psychiatrie outpatients were part of an ongoing study on the “Assessment and Management of Anger Problems in Clinical and Offender Populations.” Approximately 92% of the sample had independently documented histories of coping problems as evidenced by incidents of assault or property damage. All subjects met DSM-IIl criteria for adjustment reactions, impulse control, or personality disorder. "Spouses of SDA patients were part of a study of “Stress Vulnerability in Alzheimer Patients’ Families.” Forty-five percent met current Research Diagnostic Criteria for minor or intermittent depression and an additional 44% had met these criteria during the most distressful phase of the spouse's illness. Those meeting current, diagnostic criteria had mean Beck Depression Inventory scores of 13.3 (SD = 6.2); those with past, but not currently diagnosable depressions had a mean BDI of 9.0 (SD ~ 4.9), and those who never had been depressed had a mean BDI of 6.8 (SD = 2.9). ®Medical students were part of an ongoing study identifying “Students at High Risk for Distress in Medical School.” Some indication of student distress can be inferred by comparing medical student anxiety scores on the Symptom Checklist-90 to the norms for non-patients and outpatients. While 90% of the students obtained anxiety scores above the mean for non-patients, the mean T-score based on outpatient norms, was 44.84 (SD = 8.97), indicating that a substantial number of medical students fell in the same range as outpatients on anxiety. JANUARY, 1985 5Vitaliano et al. Table 1 indicates that the samples differ in gender, age, education- al level, and marital status. The vast majority of the psychiatric outpatients are male; the majority of the spouses of SDAT patients are female. The spouses of SDAT patients are also older than the medical students and psychiatric outpatients. The medical students all have bachelor’s degrees and more education than the psychiatric outpa- tients and spouses of SDAT patients. Finally, the majority of the psychiatric outpatients and medical students are not married; mar- riage was a subject criterion in the SDAT study. Demographic differ- ences in these samples make it possible to determine how generaliz- able psychometric relationships are both within and across the WCCL scales. Procedures Following the paradigm used by Folkman and Lazarus (1980), each participant was asked to respond to the WCCL with respect to a current serious stressor. Within each sample, the current stressor was grouped according to its source (ie., context or person involved, depending upon the sample). The source of stress varied as a function of the sample; among the outpatients the current stressor was always a person (at home or in work) who the patient regarded as the “cause” of his/her anger/violence; in the spouses of SDAT patients, 80% listed stressful experiences that were related to their spouse’s illness; all medical students listed experiences that were related to medical school. Four psychometric properties of the WCCL were studied: (1) the reproducibility of the factor structure of the originally derived scales (Aldwin et al.), (2) the internal consistency reliabilities and intercorre- lations of the original and revised scales, (3) the construct and criterion-related validities of the scales, and (4) the relationships of the scales to demographic factors. Factor reproducibility/scale revisions. In examining the psycho- metric properties of the WCCL, a principal components analysis with varimax rotation was performed first to simultaneously examine the reproducibility of the Aldwin et al. items’ factor structure and to revise the scales as appropriate. The revised scales were developed using a combination of factor analytic and rational approaches. Because the medical student sample was the largest (n = 425), it was used for this analysis. The other two samples then were used to determine the degree to which the resulting versus the original scales were internal- ly consistent and intercorrelated. 6 MULTIVARIATE BEHAVIORAL RESEARCHVitaliano et al. The principal components analysis resulted in six factors with As greater than 1. Before the factors could be interpreted, a decision had to be made about the meaning of the loadings for each item. Using the factor loading matrix, the highest loading for each item was recorded and the items were ranked according to the magnitude of their highest loading. The loadings ranged from a high of .81 to a low of .07, with the mean and median being .35 and .39 respectively. Because one of our objectives was to develop a measure with fewer items, a decision was made to consider only those items having loadings of .35 or greater. Using this criterion, the 46 items that remained were used to label the six factors. As in the Aldwin et al. analysis, Factor 1 could be labeled Problem-Focused coping. Eight of the original 15 problem-focused items had their highest loading (all greater than .39) on this factor, with a mean loading of .58 (A = 13.29, 40% of variance). This factor also included four items from the original Wishful Thinking Scale (Table 2) and three items from the Growth Scale, with respective mean loadings of .48 and .52. Factor 2 was labeled Blamed Self (\ = 5.03, 15.2% of variance) because its three highest loadings came from the three items on the original Blamed Self Scale (M loading = .70). Ten other items also had their highest loadings on this factor: five from the Mixed Scale, three from the Wishful Thinking Scale (both with M loadings of .39), and two from the Minimize Scale (M = .45). After examining these ten items, a rational decision was made to combine them into a separate scale from the Blamed Self Scale, which was labeled Avoidance (Table 2). This was done because of the manifest content of these ten items and because their loadings were so much less than the items retained from the original Blamed Self Scale. The third factor is similar to Aldwin et al.’s second factor— Wishful Thinking (d = 2.72, 8.2% of variance). Seven of the original 19 items on the Wishful Thinking Scale had their highest loading on this factor (all with loadings greater than .39). The revised scale for Wishful Thinking consisted of these seven items (M loading = .50) and one item from the Mixed Scale (loading = .66). Factor 4 (A = 2.06, 8.3% of variance) was labeled Seeks Social Support because the two highest loadings resulted from two original items on the Seeks Social Support Scale (M loading = .66). Three items from the Mixed Scale (Table 2) also had their highest loading on this factor (M loading = .40), as did one from Problem-Focused with a loading of .37. This latter item also loaded .35 on the Problem-Focused Scale. JANUARY, 1985 7Vitaliano et al. Factors 5 and 6 together accounted for 6.2% of the variance. Factor 5 contained only four items with loadings greater than .35. These items were from the original Minimize Scale and had a mean loading of .45. Factor 6 contained two items with loadings above .35; however, these items had equal or higher loadings on the Problem-Focused Scale. Table 2 summarizes the scales that resulted from the items that loaded on Factors 1 to 4. Although a scale for Minimize Threat could have been developed from Factor 5, it was decided that this factor did not add much to the explained variance (4%). Internal consistency reliabilities and intercorrelations of the scales. Certain researchers have argued that coping behavior is situation- specific (Billings & Moos, 1981; Hartmann, Roper & Bradford, 1979; rable 2 Revised Scales and Source of Itens Revised scale Original item source Problem-Focused 1. Bargained or compromised to get something P positive from the situation. 2. Concentrated on something good that could P come out of the whole thing. 3. Tried not to burn my bridges behind me, but left things open somewhat. Changed or grew as a person in a good way. Made a plan of action and followed it. Accepted the next best thing to what I wanted. Came out of the experience better than when I went in. 8. Teled not to act too hastily or follow my ‘own hunch. 9. Changed something so things would turn out all right. 10, Just took things one step at a time. 11! I know what had to be done, so I doubled my efforts and tried harder to make things work. 12, Came up with a couple of different solutions to the problem. 13. Accepted my strong feelings, but didn't let them interfere with other things too much. 14, Changed something about myself so I could deal with the situation better. 15. Stood my ground and fought for what 1 wanted. aswa = my ws sas % Seeks Social Support 1. Talked to someone to find out about the situation. 2. Recepted sympathy and understanding from someone. 3. Got professional help and did what they recommended. 4. Talked to someone who could do something about the problem. 5. Asked someone I respected for advice and followed it. 6. Talked to someone about how I was feeling. oz wma Blamed Self 1. Blamed yourself 2. Criticized or lectured yourself. 3. Realized you brought the problen on yourself. wom 8 MULTIVARIATE BEHAVIORAL RESEARCHVitaliano et al. Wishful Thinking 1. Hoped a miracle would happen. a 2. Wished I was a stronger person -~ more optimistic W and forceful. 3, Wished that I could change what had happened. w 4! Wished 1 could change the way that I felt. Ww 5. Daydreaned or imagined a better time or place than Ww the one I was in. 6. Had fantasies or wishes about how things might turn w out. 7, ‘Thought about fantastic or unreal things (Like perfect ” revenge or finding a million dollars) that made me feel better. 8. Wished the situation would go away or somehow be @ finished. Avoidance Ment on as if nothing had happened. Min Felt bad that I couldn't avoid the problem. Kept my feelings to myself. a Slept more than usual. M Got mad at the people or things that caused the problem. 6 Tried to forget the whole thing. Mas Tried to make myself feel better by eating, drinking, ™ ™ W ™ smoking, taking medications. , 8. Avoided being with people in general. 9. Kept others from knowing how bad things were. 10: Refused to believe it had happened, Note. Abbreviations for scales are: ?, Problen-focused; W, Wishful Thinking; G, Growth; M, Mixed; Min, Minimized; B, Blamed Self; S, Seeks Social Support. Zuckerman, 1979) and, as such, within-subject consistency and inter- item correlations (i.e., coefficient alpha; Cronbach, 1951) of coping behavior should not be high (Cone, 1977; McFall, 1977). In this study, however, we believed it was appropriate to examine inter-item associa- tions because both Aldwin et al. (1980) and we had already grouped items into relatively homogeneous scales (based on principal compo- nents) making the items within a scale quasi-equivalent. More impor- tantly, the only way that one could ignore the assessment of coefficient alpha would be to assume that every coping item was free of measure- ment error, an untenable assumption. Construct validity. Cronbach and Meehl (1955) have argued that one form of evidence for the construct validity of a measure is the successful prediction of associations between theoretically related variables. As noted above, the “transactional model of stress” would predict that coping strategies should be related to one’s appraisal of stressors and one’s responses to stressors (e.g., anxiety, depression). In this sense, the construct validity of the original and revised scales was assessed by examining the relationships of coping to the source of the stressor, appraisal, and distress. Source of current serious stressor. Statistical analyses examining the relationships of the source of current stressor to coping were JANUARY, 1985 9Vitaliano et al, performed within each sample. Among the psychiatric outpatients there were three source categories: 54% of the subjects listed a person within the family to be the source of their stress, 23% listed a person at work, and 23% listed persons both in the family and at work. Among the spouses of SDAT patients there were also three general sources of stress: 44% of the subjects listed an event/experience that was directly attributed to their spouse (e.g., patient soils himself), 36% listed an event not directly attributed to their spouse but which resulted from the disease (e.g., filling out applications to secure medicaid for nursing home fees); and, 20% listed problems unrelated to the patient’s disease (e.g., break up of child’s marriage). Five general sources of stress were identified within the medical dent samples, concerns about: endur- ing long hours/clinical responsibilities (19%), mastering the pool of medical knowledge (18%), limited time for personal interests (39%), financial problems (9%), and the general medical school environment (e.g., class ranking, faculty, competition) (15%). Measurement of appraisal of the. stressor and of the distress response. In the three samples, appraisal was measured by the four general appraisal items from the WCCL. These binary items read, “In general, is the situation one: (1) that you could change (changeable) or do something about?; (2) that must be accepted or gotten used to?;. (3) that you needed to know more about before you could act?; and, (4) in which you had to hold yourself back from doing what you wanted to do?”. Two types of distress were studied—anxiety and depression. In the three samples, anxiety was measured by the SCL-90 Anxiety Scale (SCL-A) (Derogatis, 1977). The Beck Depression Inventory (BDI) (Beck, Ward, Mendelson, Mock, & Erbaugh, 1961) was used to.assess depression in the psychiatric outpatients and spouses of SDAT pa- tients. In the latter group, the Hamilton Depression Scale (HDS) (Hamilton, 1960) was also employed. In the medical student sample, depression was measured by the SCL-90 Depression Scale (SCL-D). Within each sample the subject was asked to report symptoms of anxiety (or depression) in response to his/her major stressor. Such procedural directions were an attempt to control for variability in distress due to stressors unrelated to the identified stressor. Statistical analyses of coping with appraisal, and coping with appraisal and distress. Two sets of analyses were performed within each sample. In the first set, Hotelling’s T? tests examined the multivariate relationship between each appraisal item and the origi- nal and revised scales. In this scheme subjects’ responding yes or no (on the appraisal item) served as the respective samples on which the five 10 MULTIVARIATE BEHAVIORAL RESEARCHVitaliano et al. coping scales could be compared. Such a procedure lowers the probabil- ity of type I errors. When indicated, these global tests were followed by univariate analyses in order to specify the relation between an appraisal item and a coping scale. In the second set of analyses, regression analyses were used to examine the degree to which apprais- al and coping were predictive of anxiety and depression. Because demographic variables had the potential for accounting for variance in distress, independent of appraisal and coping, they were allowed to enter the equations first. Demographics therefore were partialled out of the relationships when they were associated significantly with distress. Within the samples of psychiatric outpatients and spouses of SDAT patients, anxiety (SCL-A) and depression (BDI) were predicted simultaneously, using multivariate regressions, and the results were compared using the original versus the revised coping scales as predictors. Multivariate regression differs from multiple regression in that it involves the regression of multiple criteria on to multiple predictors, Such a global test not only accounts for the associations between multiple criteria, but it also minimizes the number of tests that need to be performed. Because the SCL-D was not available on the first and second year medical classes in 1980 (nm = 235), multivariate regressions (on both SCL-A and SCL-D) would have been restricted to the 1982 samples. For this reason, multiple regressions were per- formed separately on the SCL-A and the SCL-D. For the SCL-A, models using the original versus the revised scales were developed on the first and second year classes in 1980 (n = 235) and cross-validated on the first and second year classes in 1982 (n = 190). For the SCL-D no cross-sample-validation was possible; however, models using the original versus the revised coping scales were developed and com- pared. Concurrent validity. Criterion-related validity was assessed in the 1980 medical school sample by examining the degree to which the original versus revised coping scales were related to membership in a vertical support group (VSG), a form of group therapy established at the medical school. Sixty-seven (29%) of the first and second year students (1980) belonged to such groups. As will be shown below, membership in a VSG reflects distress and/or illness behavior and, as such, is a valid behavior criterion to relate to coping strategies. In addition, both our pilot study and previous research (Cadden, Flack, Blakesee, & Charlton, 1969; Dashef, Epsey & Lazarus, 1974; Seguin, 1965) suggest that students participate in such groups in order to reduce the distress they experience in medical school. JANUARY, 1985 "Vitaliano et al. Relationship of demographic factors to WCCL Scales. Since associ- ations between a psychometric scale and demographic variables can confound a scale’s interpretation, we examined the degree to which the coping scales were related to gender, age, education, and marital status. When the number of subjects within a particular demographic stratum was large enough (n > 10), MANOVAS were performed to determine whether gender, marital status, and education were signifi- cantly related to the coping scales. Student’s t-tests were performed to test the significance of the relationship of age with the coping scales. Folkman and Lazarus (1980) have noted that in order to properly assess demographic differences in coping strategies, the source of the stressful situation should be held constant. In this study both one’s current serious stressor and anxiety (SCL-A) were partialled out in the tests of the relationships between demographic factors and coping because distress is also part of the psychological feedback loop between the person and the environment. As noted above, the demographic makeup of the three samples made it difficult or impossible to study marital status in the spouses of SDAT patients, gender in the outpa- tient sample, and education in the medical school sample. Results Internal consistency reliabilities and intercorrelations of scales. The relative merits of the revised versus original scales were examined by comparing their internal consistency coefficients and scale intercorrelations on the samples used in this study as shown in Table 3. The medical school sample was used only for illustrative purposes since it served to establish the revised scales and therefore could not be used to compare the merits of these and the original scales. For the purpose of SDAT patients the coefficient alpha was greater on the revised Problem-Focused Scale, unchanged on the revised Wishful Thinking Scale, and much greater on the revised Seeks Social Support Scale than on the respective original scales. Both the Blamed Self Scale, which is identical to the original, and the newly created Avoidance Scale yielded reliable alphas. For the psychiatric outpa- tients the pattern was similar; the coefficient alpha was higher on the revised Problem-Focused and Wishful Thinking Scales and substan- tially higher on the revised Seeks Social Support Scale. The scales for Blamed Self and Avoidance yielded respectable alphas. Within the 12 MULTIVARIATE BEHAVIORAL RESEARCHVitaliano et al. abe 2 Coping sesles ReLiabilinies Propiew-Foouses 32 88 16 os ary 08 Wisnfar Tinking 86 88 86 86 36 ” Sees octal Support «78 a5 60 ” 60 ae ianed self ” ry 80 00 a6 6 Intarcorsalations eat oss 38 48 a 186 as Ps -# 7 “ 66 7 6 oxy as a a “6 < a 32 0 aa 37 0 ae 70 6 -.07 02 ™ on wa 3s 7 2 2 93 .” 5.3 4 mn 7 00 6 8 sia 38 a “10 a 20 36 ba 2 ” 00 00 86 6 Nore. “Abbe ne for sesles are: P, Problen-Fooused) W, Wiahlul Thinking: Seeiai Supports A, Avotzance, "corrected for attenuation. medical student sample the mean alpha (across the four common scales) was .81 on the original and .82 on the revised scales. Among the spouses of SDAT patients and the psychiatric outpatients the mean alphas were 9% higher than the means of the original scales (.83 versus .76). Table 3 indicates that the revised scales had substantially less overlap than the original scales. The percent of variance shared by any two scales is obtained by squaring the correlation for that pair. When the mean of these squared correlations was calculated across the six correlations (of the four common scales), a measure of the average shared variance was obtained. Within the medical student sample the shared variance was 29% on the original and 23% on the revised scales. Among the spouses of SDAT patients there was a 40% drop in the average shared variance of the revised scales versus the original scales (ie., from 20% to 12%); among the psychiatric outpatients there was a 33% drop (42% to 28%). Overview of results for construct validity Because of the large volume of analyses, an overview of the results is provided which is followed by a detailed report of the multivariate and univariate findings. Table 4 summarizes the results of the two sets JANUARY, 1985 13Vitaliano et al. Direction of Relationships of Coping Seales vith Appraisal and pets: Spouses of SDAP patiente Problem-Fecused-R - oe HLenfod Tinking-® + te Peyoniatric outpatients Problen-Focused + misnfol Thinking + Mixed . + + + Mininine Bhrest + problen-Focueed-R . WishEul Thinking-® + ? Mesical students Problen-Pocused + + Bianed Solt + Wienfol Thinking + wined + . + ra Minimize Mmeeat + Soeks Social Support-R + + Problan-Fooveed-R + - “ Avatdance + + Wieneul Thinking-R Note, "indicates direction of relationship. with abr. with wor, ‘With sci-90 of analyses with coping versus appraisal, and coping and appraisal versus distress. Only significant relationships that occurred simulta- neously in at least two samples are included in the table. For example, because a significant relationship occurred between problem-focused coping and changeable among the spouses of SDAT patients and medical students, this relationship was included in the table. In contrast, because a significant relationship between the Mixed Scale and changeable only occurred in the medical student sample, it was not included in the table. The criteria used to establish Table 4 follow the belief that construct validity cannot be confirmed by a single predic- tion on different occasions or by many predictions in a single study. Instead, “construct validity ideally requires a pattern of consistent 14 MULTIVARIATE BEHAVIORAL RESEARCHVitaliano et al. findings involving different researchers using different theoretical constructs across a number of different studies (Carmines & Zeller, 1974, p. 24).” If a matched pair is defined as the occurrence of the same relationship in two samples, the original coping scales contain seven matched pairs across the four appraisal items, while the revised coping scales contain five matched pairs. For distress, the original scales contain two matched pairs (i.e., the Mixed Scale predicts anxiety and depression), while the revised scales contain four matched pairs and one matched triplet: Wishful Thinking significantly predicts depres- sion in all three samples. Univariate and multivariate analyses of construct validity Current serious stressor versus coping. Within the samples of psychiatric outpatients and the spouses of SDAT patients, one-way ANOVAS were performed to examine the relationship of the source of stress (the grouping variable) with each of the original and revised scales (the criterion variable). Because of the large number of univar- iate tests a rejection level of .01 was used. None of the original or revised scales was significantly different across the sources of stress among outpatients or spouses of SDAT patients. Because the medical student samples were large, MANOVAS were performed on the original and the revised scales. None of the original or revised scales was significant across the source of stress. Appraisal versus coping. Within each sample, Hotelling’s T? analyses were performed to examine the multivariate relationship between each appraisal item and the original and the revised scales. Among the spouses of SDAT patients, the T? test of changeable was significant for the original coping scales, F (7,57) = 2.55, p < .02, and for the revised scales, F (5,59) = 3.02, p < .05. No differences in the original or revised scales were observed across the appraisal dimen- sions of accepted, know more, or hold yourself back. Table 5 contains the results of the follow-up univariate analyses only for the appraisal dimensions which yielded a significant T?. Among the psychiatric outpatients no differences in coping were observed on the appraisal dimensions: changeable and accepted. The T? test for know more, however, was significantly different on the original scales, F (7,76) = 5.18, p < .001, and on the revised scales, F (5,77) = 6.06, p < .001. On the appraisal item hold yourself back, the original coping scales were globally significant, F (7,75) = 2.92, p < .01, as were the revised scales, F (5,76) = 3.03, p < .05. The results of JANUARY, 1985 15Vitaliano et al. the follow-up univariate analyses are presented in Table 5. The T* analyses on the medical students were performed with gender as a covariate because, as will be shown below, in these samples gender was related to coping. On the appraisal dimensions of: change- able, the T? test of the original scales was significant, F (7,396) = 2.09, p <.05, and on the revised scales it also was significant, F (5,399) = 11.56, p < .001; on accepted, the T* test was significant for the original coping scales, F (7,394) = 3.24, p < .005, but not for the revised scales; on the appraisal dimension, know more, the T? test was significant on a1 Verous Coping Seater ‘Appraisal coping seaies Spouses of SOAT patients’ Probler-Pocused 6.5608 Diened seit ao.s4nee Seeks Social Support-R 4.62" Paychiateic outpatients Wiantul Tinking Miniatze Threat o.a6en Seeks Social support Branca Se1e Wsentur Thinking-R aovaaner Seeks Social Support- aa,ageee Avoldance 20,2600 sass Medical students Problen-Focused ‘soa 1a.o4eee WishEu1 Thinking anvaseee crown ean minimize theeat s.a8e Btaned seit 5.380 ae.o2see Problen-Fooutad-R sou aasizeee Wishfol Tinking-® 1.93 1,63 for spouses of SDAT patientsy - begeees of freedom for all tests ari ptychiatefe ovepatients; and, 1,402 for medical students. +p < 05, Mp © 01. MAP < .005, 16 MULTIVARIATE BEHAVIORAL RESEARCHVitaliano et al. the original scales, F (7,397) = 3.48, p < .001, and on the revised scales, F (5,400) = 3.31, p < .01; on the dimension, hold yourself back, the T° test of the original scales was significant, F (7,395) = 3.58, p< .001, and it was also significant on the revised scales, F (5,398) = 10.83, p < .001. Appraisal and coping versus distress. For the spouses of SDAT patients, the multivariate regression revealed that appraisal and the original coping scales were significant predictors of distress, F (22,94) = 2.29, p < .005, accounting for 58% of the variance in SCL-A and BDI. The multivariate regression of appraisal and the revised coping scales was also significant, F (18,98) = 3.43, p < .001, accounting for 62% of the variance in the set of distress measures. These two significant multivariate regressions allowed us to perform multiple regressions to determine which specific scales were responsible for the significant global relationships. The results of these analyses are presented in Table 6 for the prediction of anxiety and depression. Because both the SCL-A and the BDI are based on structured self- reports, an attempt was made to examine the degree to which the original and the revised scales were predictive of less structured ratings of distress (i.e., Hamilton Depression scores). The results of these analyses are presented in Table 6. For the psychiatric outpatients, the multivariate regression was significant, F (22,84) = 3.18, p < .001, with the appraisal and original coping scales accounting for 70% of the variance in the set of distress measures (SCL-A and BDI), Appraisal and the revised coping scales were also significantly related to the set of distress measures, F (18,88) = 2.69, p < .005, accounting for 57% of the variance in SCL-A and BDI. The results of the multiple regressions predicting anxiety and depres- sion are presented in Table 6. Table 6 also contains the results of the multiple regressions of the coping scales predicting first anxiety, and then depression within the medical school samples. The models for anxiety, using the original and revised scales, were developed on the 1980 samples (n = 235) and then cross-validated on the 1982 samples (n = 190). The 1980 model based on the original coping scales was cross-validated by forcing the significant variables in the order that they appeared in the 1980 model. The same variables as in the 1980 model (Table 6) were significant, accounting for a total of 29% of the variance in SCL-A on the 1982 samples. When the same coefficients (as in the original 1980 model) were used to cross-validate the model, the r? was .22. The model based on the revised scales was cross-validated on the 1982 sample by entering in the same variables as in the 1980 model (Table 6). The JANUARY, 1985 7Vitaliano et al. same variables as in the 1980 model were significant, accounting for 23% of the variance in SCL-A on the 1982 samples. When the coefficients from the 1980 model were used on the 1982 sample, the r? was .20. For depression no cross validation was possible; however, the 1982 regression models of the coping variables that significantly predicted depression are presented in Table 6. Results for concurrent validity Because gender was shown to relate to coping in the medical student samples, it was controlled for in the MANOVAS that exam- ined the relationships between coping and membership in a vertical support group. After gender was partialled out of the relationship between coping and VSG, F (7,216) = 6.71, p < .001, the original coping scales were still significantly related to VSG, F (7,216) = 4.19, p < .001. The same result occurred for the revised scales: gender was significant, F (5,219) = 4.61, p < .001, as were the coping scales tencon mineioa® same ays er i" fom ash nist owe ash on Swees sootat Support ua.petr 458s 18 MULTIVARIATE BEHAVIORAL RESEARCHVitaliano et al. analyzed as a set, F (5,219) = 3.51, p < .001. The original scales that were significant were: Mixed, F (1,222) = 11.94, p < .001; Seeks Social Support, F' (1,222) = 5.42, p < .05; and Blamed Self, F (1,222) = 8.24, p < .005. The revised scales that were significant were: Seeks Social Support, F (1,222) = 9.17, p < .005; Wishful Thinking, F (1,222) = 4.15, p < .05; and Blamed Self, F (1,222) = 8.24, p < .005. Inspection of the significant coping means across the two groups revealed that members of VSG’s had higher scores than non-group members on: the Mixed Scale (M = 15.59, SD = 2.29 vs. M = 5.45, SD = 2.35), Blamed Self (M = 5.15, SD = 2.43 vs. M = 3.97, SD = 2.53), and Seeks Social Support (M = 6.29, SD = 2.29 vs. M = 5.45, SD = 2.35). Members of VSG’s had higher scores on two revised scales in addition to Blamed Self, these were: Seeks Social Support (M = 9.72, SD = 3.87 vs. M = 7.79, SD = 3.68), and Wishful Thinking (M = 13.93, SD = 6.39 vs. M = 11.90, SD = 5.87). Demographic relationships with the coping scales As noted in the procedures section, source of current severe stressor and SCL-A were partialled out of the analyses which exam- ined the relationship of coping to demographic factors. The source of stress was not a significant correlate of the original or revised scales in the three samples. SCL-A was a significant correlate of the revised scales in the three samples; however, on the original scales, SCL-A was only a significant variable in the medical school and outpatient samples. Gender. Among the spouses of SDAT patients, no significant associations occurred for gender with either the original scales, F (7,53) = 1.33, or the revised scales, F (5,55) = .38. Among the medical students, the 7? test of gender with the original scales was significant, F (7,401) = 11.86, p < .001. Univarite F-tests (all with 1,407 degrees of freedom) revealed that the significant original scales were: Problem- Focused, F = 6.67, p < .01; Wishful Thinking, F = 3.92, p < .05; Mixed, F = 36.73, p < .001; Growth, F = 11.15, p < .001; Seeks Social Support, F = 61.64, p < 001; and Blamed Self, F = 4.50, p < .05. In every case, females had higher coping means than males had. On the revised scales the T? test of gender was also significant, F (5,404) = 9.23, p< .001; univariate F tests (all with 1,407 degrees of freedom) revealed that the significant revised coping scales were: Problem-Focused, F = 7.24, p < .01; Wishful Thinking, F = 6.36, p < .01; Seeks Social Support, F .58, p < 001; Avoidance, F = 4.79, p < .05; and Blamed Self, F = 4.60, p < .05. Females had significantly higher scores than males on these scales. JANUARY, 1985 19Vitaliano et al. Age. Within the samples of psychiatric outpatients and SDAT patients, no significant partial correlations were observed between age and the original or the revised scales (with SCL-A and source of stress partialled out). However, among the medical students, age was signifi- cantly related to the original Problem-Focused Scale (r = .12, p < .01) the original and revised Wishful Thinking Scales (r = .11, p < .01) and (r = .16, p < .001) respectively, and the original and revised Seeks Social Support Scales (r = .13, p < .005) and (r = .17, p < .001) respectively. When corrected for attenuation, the correlation between age and the revised Seeks Social Support Scale became .23. Even this (the highest of the significant relationships) represented a very small percent of shared variance with age (5%), indicating that the signifi- cant associations of coping with age are the result of large samples and not of strong associations. Education. Among the spouses of SDAT patients, neither the original set of coping scales, F (14,104) = 1.02, nor the revised set of scales, F (10,108) = 1.36, were significantly related to education. Among the psychiatric outpatients, the original set of scales, F (14,92) = .58, and the revised set of scales, F (10,96) = .44, were not related significantly to education. Marital Status. Among the psychiatric outpatients, neither the original set of scales, F (7,46) = .57, nor the revised set of coping scales, F (5,48) = .40, was related significantly to marital status. Among the medical students, the original set of scales, F (7,388) = 1.65, and the revised set of scales, F (5,391) = .89, were not related significantly to marital status. Discussion The purpose of this study was to examine the psychometric properties of the original and the revised WCCL scales. Specifically, the factor structure, internal consistency reliability, shared interscale variance, construct and concurrent validity, and demographic bias of the WCCL were studied. In an attempt to determine the generalizabili- ty of the factor structure of the WCCL, these properties were assessed on three different samples and the reliability and validity of the revised scales were demonstrated consistently across the samples. The present results indicate that the original scales have respect- able reliability coefficients, but that our revised scales have alphas that are higher; more importantly, these scales share substantially 20 MULTIVARIATE BEHAVIORAL RESEARCHVitaliano et al. less variance. Because the WCCL items were rationally derived from a theory that views coping as a multidimensional rather than unidimen- sional process (Folkman & Lazarus, 1980; Lazarus & Launier, 1978), it is important that the WCCL scales “approximate” independent dimen- sions. In the transactional model of stress, coping is defined in terms of what a person does or thinks, and not in terms of adaptation and distress (Vaillant, 1977). As such, one is better able to obtain an unconfounded assessment of the relationship between coping and distress (cf, Cohen & Lazarus, 1979) and thereby evaluate construct validity. In this study, construct validity was assessed using the guideline that associations between theoretical constructs be replicat- ed across at least two different samples. Overall, there were ten replicated relationships of the revised scales with appraisal and distress. These relationships would have been anticipated theoretically from the transactional model of stress (Lazarus & Launier, 1978) and empirically from the results of Folk- man and Lazarus (1980). The latter researchers found that when subjects appraised situations as changeable and when they thought they needed to know more the subjects did significantly more problem- focused than emotion-focused coping. In contrast, appraisals of accept- ance and holding oneself back yielded significantly more emotion- focused than problem-focused coping. It is appealing that the appraisal of changeable was related to the revised Seeks Social Support Scale; this scale contains problem-solving strategies through social contacts (Table 2). It also is appealing that the revised Problem-focused Scale was related to the appraisal—know more. The relative associations of the coping scales with anxiety and depression provide the strongest evidence of the construct validity of the revised versus the original scales. Using the original scales, Coyne et al. (1981) found a significant positive correlation between the Wishful Thinking Scale and depression. In the current study the original Wishful Thinking Scale was not shown to be a significant predictor of depression in more than one sample. The revised Wishful Thinking Scale was replicated as a predictor of depression in three samples that measured depression in different ways and this scale accounted for sizable amounts of variance in depression (14%-21%). A significant negative association was found in this study between the revised Problem-Focused Scale and depression in both medical students and the spouses of SDAT patients. Within the latter sample, the same pattern occurred for the prediction of both the BDI and the HDS (which had a .62 correlation, p < .001). That is, among the JANUARY, 1985 21Vitaliano et al. spouses of SDAT patients, both structured and unstructured measures of depression had significant negative correlations with the revised Problem-Focused Scale and significant positive correlations with the revised Wishful Thinking Scale. Citing the learned helplessness model of Abrahamson et al. (1978), Coyne et al. (1981) predicted that problem-focused coping should be related negatively to depression. Using the original Problem-Focused Scale, Coyne et al.’s study and the current study, did not find this to be true. Because of the moderate-to-high correlations of depression with anxiety in the spouses of SDAT patients (r = .62), psychiatric outpa- tients (r = .61), and medical students (r = .80), it is not surprising that the revised scales should be related to anxiety and depression in a similar way (Table 4). The relationships between the revised coping scales (Wishful Thinking and Problem-Focused) and anxiety were, however, not nearly as strong as those with depression. In addition, one revised scale, Seeks Social Support, was replicated as a significant predictor of anxiety, but not of depression. This result is appealing given the classic work of Schacter (1959) who demonstrated that subjects who are anxious are significantly more likely to seek affili- ations than subjects who are less anxious. Schacter concluded that misery likes company. More recently, Janis (1983) discussed the role of social supports in relation to anxiety provoking situations (i.e., career decisions, marital difficulties, health problems, and other personal dilemmas). Nine of the relationships between the original coping scales and appraisal and distress were replicated. Some of these relationships are consistent with the results of Folkman and Lazarus. (1980). For example, the original Problem-Focused Scale was significantly related to the appraisals of change and know more in two samples. However, know more was also significantly related to such original emotion- focused scales as Minimize Threat and Wishful Thinking; acceptance was not related to any original or revised emotion-focused scales. The Mixed Scale accounted for four of the nine replicated relation- ships that involve the original coping scales with appraisal and distress. This scale is the only original scale that significantly predict- ed anxiety and depression in more than one sample. Although Coyne et al. (1981) also found the Mixed Scale to be significantly related to depression, this result is difficult to interpret because the scale contains both avoidant and help-seeking strategies. A case in point involves the replicated relationship between the Mixed Scale and the appraisal hold self back. Although the results of Folkman and Lazarus (1980) lead one to expect that: hold self back should be related to 22 MULTIVARIATE BEHAVIORAL RESEARCHVitaliano et al. emotion-focused coping (i.e., Avoidance and Wishful Thinking), it is not clear from the Mixed Scale which strategy (avoidant or help- seeking strategies) is related to this type of appraisal. In contrast, the replicated association between hold self back and the revised Avoid- ance Scale indicates that at least avoidance is related to such an appraisal. The strongest evidence for the concurrent validity of the scales is provided by the finding that VSG members have significantly higher scores than non-VSG members on the original and revised Seeks Social Support Scales. Vertical Support Group members also have signifi- cantly higher Mixed scores than non-VSG members, however, this finding is difficult to interpret. Finally, although the two groups do not differ on the original Wishful Thinking Scale, VSG members do have significantly higher scores on the revised Wishful Thinking Scale. This can be partially attributed to the fact that VSG members are signifi- cantly more distressed than non-members (on the SCL-A, F (1,217) = 12.47, p < .001), and that anxiety, while related to the revised Wishful Thinking Scale, was not related to the original scale. Because VSG members are more distressed than non-VSG members, they may be engaging in more cognitive distortions (i.e., Wishful Thinking) for defensive purposes. It is encouraging that both the original and the revised coping scales were not related to demographic factors in the spouses of SDAT patients and the psychiatric outpatients. Among medical students, the relationships of the coping scales to marital status and age were either nonexistent or minimal. However, gender was significantly related to the original and the revised coping scales in the medical student samples (even after the adjustment of SCL-A). This result is consistent with the findings of Folkman and Lazarus (1980), in which females had significantly higher scores than males on both the problem- and emotion-focused coping scales. Billings and Moos (1981) also found that females were more likely to use avoidant strategies than males. The current results could be attributed to the fact that covariance did not completely equate females and males on anxiety. An increasing body of literature has reported that females in medical school report more problems of role conflict (Adsett, 1968; Notman & Nadelson, 1973; Roeske & Lake, 1977), drop out at higher rates than males and have more difficulty in finding satisfying social relationships (Edwards & Zimet, 1976; Goldstein, 1975), and are more likely to seek psychiat- ric counseling (Adsett, 1968). In examining the current study’s findings one should note that our primary goal was to assess the psychometric properties of the original JANUARY, 1985 23Vitaliano et al. rabie 7 Medical students Spouses of SDAT patients Peychiatric outpatients Problem-focused 24.70 0.37 20.71 9.51 23.07 9.75 Wishful Thinking 12.72 5,92 9.36 5.89 13.56 6.67 Seeks Social Support 8.87 3.69 8.65 4.80 8.62 4,93 Blamed Self 43s 248 2.07 24d 409 2.89 Avoidance 12.82 5.37 535.23 1a6s 6.69 Rppeaiaal ieens Percent responding "yes" could change or do sone- 48 45 n thing about the problem: Accept or get used to 6 so M4 the problen Need to know more before 52 28 59 Tean act Had to hold myself back 51 26 B versus the revised WCCL scales, and not to consider the substantive implications of the relationships between coping with appraisal and distress. Given this goal, no attempt was made to compare coping across the three samples. These comparisons would have required detailed substantive considerations; because of the major demographic and psychosocial differences in the three samples, such comparisons would be uninterpretable. The means and standard deviations for the scales are presented in Table 7 for descriptive purposes. Considering the above results, the WCCL holds promise as a measure of a wide range of coping strategies. When used on three samples of subjects who are different from the samples used to establish the WCCL, the revised scales had respectable internal consistency reliabilities, and construct and criterion-related validity. In addition, these scales were generally unconfounded by demographic differences across the three samples. For these reasons, the revised WCCL should be a valuable measure of coping in response to environ- mental stressors. References Abrahamson, L. Y., Seligman, M. E., & Teasdale, J. P. (1978). Learned helplessness in humans: Critique and reformulation. Journal of Abnormal Psychology, 87, 49-74. ‘Adsett, C.A. Psychological health of medical students in relation to the medical education process. (1968). Journal of Medical Education, 43, 728-734. Aldwin, C., Folkman, S., Shaefer, C., Coyne, J., & Lazarus, R. (1980, September). Ways of Coping Checklist: A process measure. Paper presented at the annual American Psychological Association Meetings, Montreal, Canada. 24 MULTIVARIATE BEHAVIORAL RESEARCHVitaliano et al, Andreasen, N. J., & Norris, A. S. (1972). Long-term adjustment and adaptation mecha- nisms in severely burned adults, Journal of Nervous and Mental Disease, 154, 352— 362. Antonovsky, A. (1979). Health, stress, and coping. San Francisco: Jossey-Bass. Averill, J.R., & Rosen, M. (1972). Vigilant and nonvigilant coping strategies and psychophysiological stress reactions during the anticipation of electrical shock. Journal of Personality and Social Psychology, 23, 128-141. Beck, A. T., Ward, C. H., Mendelson, M., Mock, J. & Erbaugh, J. (1961). An inventory for measuring depression. Archives of General Psychiatry, 4, 561-571. Billings, A. G., & Moos, R. H. (1981). The role of coping responses and social resources in attenuating the stress of life events. Journal of Behavioral Medicine, 4, 139-187. Cadden, J.J., Flack, F., Blakesee, S., & Charlton, R. (1969). Growth of medical students through group process. American Journal of Psychiatry, 126, 862-868. Carmines, E. G., & Zeller, R. A. (1974). On establishing the empirical dimensionality of theoretical terms: An analytic example. Political Methodology, 1, 75-96. Carson, R. I. (1969). Interaction concepts of personality. Chicago: Aldine-Atherton. Cohen, R., & Lazarus, R. 8. (1979). Coping with stresses of illness. In G. C. Stone, F. Cohen, and N. F. Adler (Bds.), Health psychology. San Francisco: Jossey-Bass. Cone, J.D. (1977). The relevance of reliability and validity for behavior assessment. Behavioral Therapy, 8, 411-426. Coyne, J.C., Aldwin, C,, & Lazarus, R. 8. (1981). Depression and coping in stressful episodes. Journal of Abnormal Psychology, 90, 489-447. Cronbach, L. J. (1951). Coefficieiit alpha and the internal structure of tests. Psychome- trika, 16, 543-548, Cronbach, L. J., & Meehl, P. (1955). Construct validity in psychological tests. Psycholog- ical Bulletin, 62, 281-302. Dashef, S. S., Epsey, W. M., & Lazarus, J. A. (1974). Time limited sensitivity groups for students. American Journal of Psychiatry, 151, 287-292. Derogatis, L. R. (1977). The SCL-90-R administration, scoring and procedures Manual I. Baltimore: Clinical Psychometric Research Unit, Johns Hopkins University. Edwards, M. T., & Zimet, C. N. (1975). Problems and concerns among medical students. Journal of Medical Education, 51, 619-625. Folkman, S., & Lazarus, R. 8. (1980). An analysis of coping in a middle-aged community sample. eJournal of Health and Social Behavior, 21, 219-239. Goldstein, M. Z. (1975). Preventive mental health efforts for women medical students. Journal of Medical Education, 50, 289-291. Hamilton, M. (1960). A rating scale for depression. Neurology, Neurosurgery, and Paychiatry, 23, 56-62. Hartmann, D.P., Roper, B.L., & Bradford, D. C. (1979). Some relationships between behavioral and traditional assessment, Journal of Behavioral Assessment, 1, 321. Janis, I.L. (1983), The role of social support in adherence to stressful decisions. American Psychologist, 38, 143-160. Lazarus R. 8, (1966) Paychologcal stress and the coping process. New York: MeGraw- il. Lazarus, R. S., & Launier, R. (1978). Stress-related transactions between the person and environment. In L.A. Pervin & M, Lewis (Eds.), Perspective in interactional psychology. New York: Plenum, Lipowski, Z. J. (1970). Physical illness, the individuel and the coping processes. Psychiatry in Medicine, 1, 92-102. ‘McFall, R. M. (1977). Analogue methods in behavioral assessment: Issues and prospects. In J.D. Cone & RB. P. Hawkins (Eds), Behavioral Assessment. New York: Brunnel azel Notman, M.T., & Nadelson, C.C. (1973), Medicine: A career conflict for women. American Journal of Psychiatry, 130, 1123-1127. Roeske, N. A., & Lake, K. (1977). Role models for women medical students. Journal of Medical Education, 52, 459-466. Schacter, S. (1959). The psychology of affiliation. Stanford, California: Stanford Univer- sity Press. Sequin, C.-A. (1965). Groups in medical education. Journal of Medical Education, 40, JANUARY, 1985 25Vitaliano et al. Vaillant, G. (1977) Adaptation to Life. Boston: Little, Brown. Zuckerman, M. (1979). Traits, states and uncertainty. Journal of Behavioral Assessment, 1, 43-54. 26 MULTIVARIATE BEHAVIORAL RESEARCH