Psychological Repor&, 1991, 69, 551-560.
O Psychological Reports 1991
SYMPTOM-POSITIVELY A N D -NEGATIVELY WORDED ITEMS I N TWO POPULAR SELF-REPORT INVENTORIES OF ANXIETY A N D DEPRESSION ' JAAP MOOK WIM CHR. KLEIJN AND HENK M.
VAN DER
PLOEG
Vrije Univerriteit Amsterdam Summary.-Substantially higher mean scores on symptom-negatively versus symptom-positively worded items have consistently been reported in the literature for the balanced State-Trait Anxiety Inventory. In this study we aimed to replicate and extend these findings to Dutch trait versions of the inventory and Zung's similarly balanced Self-rating Depression Scale. Analysis indicated significantly higher mean subscale scores for symptom-negative as opposed to symptom-positive items of both measures, across sexes and age groups as well as across different levels of distress in nonclinical (n = 8631, subclinical (n = 450), and clinical subject samples (n = 96). Sex and age differences were mainly confined to symptom-positive subscales. Scale intercorrelations were lowest between symptom-positively and symptom-negatively worded scales both within and across measures. Factor analyzing the combined measures identified a symptom-negative and a symptom-positive factor, tentatively labeled "absence of positive affect" and "presence of negative affect." Several explanations of the findings among which item-intensity specificity, the response style of social desirability, and the trait model of posiuve and negative affectivicy are discussed.
Psychological assessment of affective states, traits, and clinical syndromes frequently occurs by means of self-report. Such inventories may contain a balanced number of symptom-positively and symptom-negatively worded items to counter certain response tendencies like acquiescence or "yea saying" from interfering too much with the measurement. Response formats of the symptom-negative items are subsequently reversed prior to scoring to ensure that all items are measuring the same underlying dimension or construct. O n e widely used affective self-report measure, the State-Trait Anxiety Inventory (Spielberger, 1983), for example, contains 10 items worded positively (e.g., "I feel nervous") and 10 items worded negatively (e.g., "I feel calm") in the State version, while in the Trait version the ratio is 11:9. Another example of a widely used instrument is the Self-rating Depression Scale (Zung, 1965, 1973, 1986) devised for the measurement of depression as a clinical syndrome and likewise consisting of 10 symptom-positively worded items (e.g., "I feel down-hearted and blue") and 10 negative statements (e.g., "Morning is when I feel the best").
'Address correspondence to Jaap Mook, Department of Medical Psychology, Vrije Universiteit Amsterdam, Van der Boechorststraat 7, # D-556, 1081 BT Amsterdam, The Netherlands.
552
J. MOOK, ETAL.
Empirically, research with the anxiety inventory has rather consistently yielded different results for the two sets of items. Spielberger, Gorsuch, and Lushene .(1970), in developing the inventory, had noted differences in mean state item scores under different severities and kinds of stress. In general, negatively worded items differentiated better when stress was lower whereas symptom-positively worded items did so when stress was higher (Spielberger, 1983). Differences in mean responses of the two sets for both the State and Trait version of the inventory have also been reported by Bernstein and Eveland (1982) who concluded that "its worst feature is that differences between positive and negative items are approximately as great as the differences between State and Trait items" (p. 372). Similar differences have subsequently been found by DeVito and Kubis (1983) for psychology students in a nonstressful situation and by Bonke, Smorenburg, Van der Ent, and Spielberger (1987) in a stressful situation (surgical patients on the day before the operation). In all these studies mean responses for the negatively worded items have consistently been substantially higher than that for the positively worded items. Moreover, in the Bonke, et al. (1987) study, differences were found irrespective of severity of State-anxiety which, as might be expected, for both men and women were appreciably higher in the pre- than in the postsurgical condition. The purpose of the present study is threefold. First, we wanted to know if the findings by Bonke, et al. (1987) on differences for positively and negatively worded items for the Dutch State version of the State-Trait Anxiety Inventory could be extended to the Dutch Trait version of the inventory and Zung's Depression Scale. Second, we examined the influence of sex, age, and general distress on the ~ ~ m p t o r n - ~ o s i t iand v e l symptom-negatively ~ worded subscale scores. Finally, to elucidate further the meaning of such differences in item scores, we intercorrelated the scales and explored their underlying structure by means of factor analysis.
Subjects The subjects participating in this study consisted of four groups: (a) 470 undergraduate students (39% men) in the social, medical, or health sciences (M age = 20.0 yr., SD = 3 3 , (b) 393 adults (45% men) from the general population of Leiden, The Netherlands (M age = 38.0 yr., SD = 14.0), (c) 450 medical patients (34% men) suffering from multiple sclerosis (M age = 48.0 yr., SD = 12.0), and (d) 96 psychiatric patients (42% men) of whom 46 were outpatients (M age = 35.0 yr., SD = 6.0) and 50 were inpatients (M age = 39.0 yr., SD = 13.0). Instruments The Trait version of the Dutch State-Trait Anxiety Inventory (Van der
SYMPTOM-POSITIVELY AND -NEGATIVELY WORDED ITEMS
553
Roeg, Defares, & Spielberger, 1980) was given. This has 20 items of which 10 are worded positively and 10 are worded negatively. In addition, a Dutch Trait-adapted version of Zung's Self-rating Depression Scale (Mook, Kleijn, & Van der Ploeg, 1989; Mook, Van der Ploeg, & Kleijn, 1990) was adrninistered. This scale also has 20 items of which 10 are worded positively and 10 negatively.
Procedure All subjects participated voluntarily and anonymously in the study. The students were asked at the beginning of a lecture and filled out the questionnaires at the end of the lecture. The adults and medical patients were invited by letter and completed the forms at home, returning them by mail. Lastly, the psychiatric patients completed the scales as part of their intake or during a break in their daily programs. Differences Between Positively and Negatively Worded Scales Means, standard deviations, and t ratios used in assessing differences in scores on positively and negatively worded subscales of the Trait-Anxiety Inventory and Zung's Depression Scale are presented in Tables 1, 2, and 3 . In Table 1 score differences are compared for the total sample and for men and women separately. As can be seen in Table 1, means for the positively worded scales on both the State-Trait Anxiety Inventory and the Depression Scale are substantially lower than those for the negatively worded scales, differences being significant for the total sample as well as for men and women separately. Differences are clearly larger for the Zung scale than for the anxiety inventory and somewhat larger for men than for women. It can also be observed in Table 1 that whatever differences there are between the two tests are predominantly confined to their subscale scores for TABLE 1 MEAN TOTALA N D SYMPTOM-POSITIVE VERSUSSYMPTOM-NEGATIVE SUBSCALE SCORES FORDUTCHT R A ~VERSIONS T OF STATE-TRAIT ANXIETYINVENTORY A N D ZUNG S E L F - R A ~DEPRESSION G SCALEM WHOLESAMPLE( N = 1409) AND MENA N D WOMENSWARATELY (ns = 556 AND 853) Subiects
AU
Scales
State-Trait Zung
Men
State-Trait Zung
Women
State-Trait
*All t ratios: p < ,001
Total M
SD
39.9 38.1 38.9 37.1 40.5
10.4 8.3 10.5 7.9 10.3
Symptompositive M SD 18.6 16.2 17.9 15.5 19.0
5.4 4.3 5.5 4.0 5.4
Symptomnegative M SD 213 21.9 21.0 21.5 21.5
5.9 5.2 5.9 5.1 5.9
I*
-12.8 -31.7 -9.1 -21.4 - 9.3
554
J. MOOK, ET AL.
positively worded items. This also holds for the small, though significant, sex differences in total anxiety and depression scores (t,,,, = 2.7, p < .O1 and tI4,,, = 3.5, p < .OOJ.) which appear to be based mainly on differences in subscale scores for positively worded items. In Table 2 means and SDs for the two sets of items are presented for different age groups by dividing the total sample into four categories. As may be observed, essentially the same score differences are found, subscale scores for positively worded items for both tests being appreciably and significantly lower than scores for negatively worded items across all age groups. Similar differences were found when scores within age groups were compared for men and women separately. I t may be observed also in Table 2 that scale scores generally increase for increasingly older groups. This, however, is largely confounded with the results for the subgroups, who differ both in age and distress, as presented in Table 3. TABLE 2 MEANTOTALAND SYMPTOM-POSITNELY VERSUSSYMPTOM-NEGATIVELY WORDEDS ~ S C A L E SCORESFORDUTCHTRAITVERSIONS OF STATE-TRAIT ANXIETY INVENTORY AND ZUNG SCALE IN DIFFERENTAGEGROUPS SELF-RATING DEPRESSION Age Groups (4
Scales
Total
M
SD
Symptompositive
M
SD
Symptomnegative
M
t*
SD
0-20
yr. (n = 366) State-Trait Zung 21-40 yr. ( n = 518) State-Trait Zung 41-60 yr. (n = 387) State-Trait Zung 61-90 yr. (n = 138) State-Trait Zung *AlI t ratios: p < ,001.
In Table 3 means and SDs are given for subgroups with different total scores on anxiety and depression (means for students and adults as well as for psychiatric in- and outpatients have been combined, differences in total Anxiety and Depression scale scores being small and nonsignificant). As can be seen in Table 3, again essentially similar differences are obtained for groups differing in total distress. Likewise, the same significant differences were found when results within groups were compared for men and women separately. I n fact, the only exception was observed for the psychiatric inpa-
SYMPTOM-POSITIVELYAND -NEGATIVELY WORDED ITEMS
555
TABLE 3 MEANTOTALAND SYMPTOM-POS~ELY VERSUSSYMPTOM-NEGATIVELY WORDED SUBSCALE SCORES FORDUTCHTRAIT VERSIONSOF STATE-TRAIT ANXIETYINVENTORY A N D ZUNGSELFRATING
DEPRESSION SCALEFORSUBJECTSWITH DIFFERENT LEVELSOF ANXIETYAND DEPRESSION
Subjects
Total
Scales
(n)
M
SD
Symptompositive
M
SD
Symptom-
t*
negative M SD
Student/Adult Sample (n = 863) State-Trait Zung Medcal Sample (n = 450) State-Trait Zung Psychiatric Sample (n = 96) State-Trait Zung
36.4 34.5
8.8 6.5
17.0 14.2
4.8 3.1
19.4 20.3
5.0 4.6
-10.4 -32.1
42.8 42.1
11.6 8.9
19.8 18.6
6.1 4.7
23.0 23.5
6.6 5.5
-7.6 -14.4
56.0 50.2
9.8 9.0
26.4 21.7
5.5 5.3
29.6 28.7
5.2 5.6
-4.1 -9.0
*All t ratios: p < ,001.
tient group, where the difference on the Anxiety inventory, although in the expected hrection, was not statistically significant.
Analyses of Variance To investigate the influence of sex, age, and groups separately as well as their possible interactions, three-way analyses of variance were performed on the scales as the dependent variables. As might be expected, by far the largest main effects were obtained for groups on all scales of the Trait-Anxiety Inventory (Fs = 152.6, 116.8, and 134.3, p < .0001) and the Zung Depression Scale (Fs = 194.6, 178.8, and 123.2, pc.0001). In addition, significant main effects for sex and age were observed on both the total scales and positively worded subscales of the Trait-Anxiety Inventory, and Zung's Depression Scale. For sex, Fs = 6.4 (p < .01) and 14.3 (p c .001) on the anxiety inventory, and Fs = 10.6 and 18.7 ( p < .001) on the Zung scale. For age, Fs = 3.9 ( P C .05) and 9.4 ( p < .01) on the anxiety inventory, and Fs = 6.8 ( P C .01) and 23.6 (pc.001) on the Zung scale. However, neither the main effects on the negatively stated subscales nor first- and second-order interaction terms were statistically significant. Scale lntercorrehtions and Factor Analysis Scale and subscale intercorrelations were computed for the two subgroups differing maximally in total amount of distress. Values are presented in Table 4, which shows that the patterns of scale intercorrelations for the studentladult and psychiatric patient groups are highly similar, generally being somewhat stronger in the latter though (and probably reflecting a difference in sample size). Total scale correlations between the anxiety inventory and Zung's scale
556
J. MOOK, ET AL.
TABLE 4 PEARSON INTERCORRELATIONS BETWEENSCALESA N D SUBSCALES OF DUTCHTRAIT VERSIONS OF STATE-TRAIT ANXIETY INVENTORY AND ZWNG SELF-RATMG DEPRESSION SCALE FORSTUDENT/ADULT GROW(n = 863, UPPER TRIANGLE) AND FOR PSYCHM-rruc PATIENTS (n = 96, LOWER TRIANGLE) Scales State-Trait Anxiety Inventory 1. Trait-total 2. Symptom positive 3. Symptom negative
Zung Self-ratingDepression Scale 4. Trait-total 5. Symptom positive 6. Symptom negative Note.-Decimal points omitted.
1
2
3
4
5
6
89
90 60
76 67 69
67 68 53
62 49 62
77
90 41
90 88
58
83 74 72
74 74 57
73
57 72
88 89
56
are high, suggesting that both measure essentially s i d a r constructs. As can be further observed in Table 4, positively and negatively worded subscales for anxiety tend to correlate higher with their like subscales for depression than they d o with each other, while the same holds in reverse for both depression subscales. I n fact, the lowest correlations are consistently found between the positively and negatively worded subscales both within and across measures, suggesting that whatever the differences in anxiety and depression, these are to be found in these subscales. To elucidate further the meaning of the positively and negatively worded subscales a factor analysis was performed on the pooled items of the two tests for the total sample. A two-factor principal components solution with varimax rotation yielded a complete overlap of anxiety and depression items, with all symptom-negatively worded items loading on the first and all symptom-positively worded items loading on the second factor. Factor 1 was tentatively labeled "absence of positive affect," while Factor 2 was called "presence of negative affect."
DISCUSSION Our results on consistent differences in response to positively vs negatively worded items agree with earlier findings for the anxiety inventory in both the State and Trait versions (e.g., Spielberger, et al., 1970; Bernstein & Eveland, 1982; DeVito & Kubis, 1983; Bonke, et al., 1987). Extending our findings to a similarly balanced measure of trait-depression like the Zung Self-rating Depression Scale attests to the robustness of the phenomenon when other item-content is involved. Differences in item-content also appeared to be mainly responsible for the larger positive-negative differences on the Self-rating Depression Scale as compared to the Trait-Anxiety Inventory. As was observed in Table 1, these larger differences were mainly due to the lower mean scores on the
SYMPTOM-POSITIVELY AND -NEGATIVELY WORDED ITEMS
557
positively worded subscale. Inspection of mean item-scores showed that the lower scores for positively worded depression items partly resulted from an imbalance in the eight somatic items of the scale (five of which are worded positively) and partly reflected the presence of a suicide item which is also worded positively. Likewise, the somewhat lower mean total scores for depression as compared to anxiety may partly reflect this presence of somatic items in the former as compared to the latter scale. I t should be noted, however, that while there is little overlap between the two measures in actual item-content, at the symptom-level the situation is quite different. Gotlib and Cane (1989), for example, on the basis of a content-analysis using DSM-111-R criteria for anxiety and depression, conclude that 40% of the items on the Zung Depression Scale actually measure symptoms of anxiety whereas almost half of the items on the State-Trait Anxiety Inventory rneasure symptoms of depression. Thus, the high correlations between the two inventories as found in our study as well as by others (Tanaka-Matsumi & Kameoka, 1986) may partly reflect these common symptoms (Gotlib & Cane, 1989). I n a similar vein our factor-analytic results for positively and negatively worded items loading on different factors tend to agree with a number of factor-analytic studies of the State-Trait Anxiety Inventory (Spielberger, Vagg, Barker, Donham, & Westberry, 1980; Vagg, Spielberger, & OJHearn, 1980; Bernstein & Eveland, 1982; Van der Ploeg, 1989) and to some extent also on the Zung scale (see, e.g., Blumenthal, 1975). I n factor-analytic studies of the anxiety inventory distinct symptom-positively and symptom-negatively worded factors have consistently been found for the State and Trait versions, both separately and combined. These results were initially interpreted as reflecting "item-intensity specificity," i.e., positively worded items supposedly being more sensitive with greater anxiety (Spielberger, et al., 1970, 1980). I t should be noted, however, that both our results and those by Bonke, et al. (1987) on consistent positive-negative differences irrespective of State and Trait anxiety levels would tend to argue against such an interpretation. Moreover, positively and negatively worded subscales of the anxiety inventory in our study correlated equally high with total anxiety and depression scale scores at different levels of distress, a finding which likewise would seem to contradict the notion of specificity of item-intensity. More recently, it has been suggested that the negatively worded items of the anxiety inventory may assess a "state of well-being," i.e., a positive emotional state inversely related to anxiety and reflecting something more than merely low anxiety (Spielberger, 1985). The notion of specificity of item-intensity having been found wanting, how can our results be explained? One possible explanation resides in the effects of response styles which generally may act as confounds in self-reports
558
J. MOOK, ETAL.
assessing a variety of stress-related characteristics including anxiety and depression (Langevin & Stancer, 1979; Linden, Paulhus, & Dobson, 1986; Tanaka-Matsumi & Kameoka, 1986). The most likely candidate here appears to be social desirability (Edwards, 1957; Crowne & Marlowe, 1960). Both the Trait-Anxiety Inventory and the Zung scale have correlated substantially with two measures of social desirability in college students (Tanaka-Matsumi & Kameoka, 1986). Social desirability is generally defined as the tendency to endorse items that are socially desirable while not endorsing items that are socially undesirable (Edwards, 1957). Thus, such a stylistic variable might be expected generally to exert a lowering effect on positively worded items which due to their negative affective content may be considered as socially undesirable. This could very well account for the sex and age differences found in our study (as well as in others, see, e.g., Bonke, et al., 1987; Blumenthal, 1975) which, as may be remembered, were predominantly due to differences on the positively worded subscales. For both women and older people it may be more acceptable to admit negative affect than for men and younger people. Conversely, social desirability might be expected to exert a heightening effect on negatively worded items which due to their generally positively toned affective content may be considered as socially desirable. The net effect of the social desirability variable then would be to produce more or less balanced scores on the positively and negatively worded subscales. However, this is not what we found; negatively worded subscales instead had significantly larger mean scores than positively worded subscales. I t is difficult to see how this imbalance can be reconciled with the social desirability notion unless we accept the possibhty that this response style operates differentially on positively and negatively worded items. In this respect it 'is unfortunate that Tanaka-Matsumi and Kameoka (1986) did not report separate correlations for positively and negatively worded subscales which might have shown different results. O n the other hand, it could be argued that social desirability operates more strongly in students (tested by Tanaka-Matsumi and Kameoka) than in psychiatric patients. Inspection of our raw data, comparing the students and adults with the psychiatric patients, indicated that this indeed might have been the case. The percentage of low scores (values 1-2) to high scores (values 3-4) on the State-Trait Anxiety Inventory for the students and adults was 54:46, and for the psychiatric patients 60:40, i.e., more balanced in the former. This result again seems to underline the difficulty of explaining our positive vs negative score differences solely in terms of social desirability, especially for the psychiatric patients. The second explanation, and the one we favour, asserts that our consistent positive-negative differences reflect the workings of two independent dimensions of positive and negative affect. Self-report inventories of affect
SYMPTOM-POSITIVELYAND -NEGATIVELY WORDED ITEMS
559
like those we used here, containing a balanced number of positively and negatively worded items, share an implicit assumption with regard to perand bipolarity of positive sonal affective structure, that of unidimen~ionalit~ and negative affect. This is implied by the reversal of response weights for the symptom-negative or construct-absent items prior to scoring, assuming construct-presence thereafter. Thus, in the Trait-Anxiety Inventory the weights for items like "I feel rested" and "I am happy" are reversed to "almost never" (4), "sometimes" (3), "often" (2), and "almost alwaysm(1). But surely, admitting that one only rarely feels rested or happy is not necessarily equivalent to feeling anxious most of the time. O n the Self-rating Depression Scale the weights for items like, "Morning is when I feel the best" and "I still enjoy the things I used to do," are similarly reversed. However, stating that one seldom feels at one's best during the morning or only rarely enjoys things the way one used to d o is not necessarily the same as feeling worst during the morning or not at all enjoying things most of the time. Research in personal affective structure underlying self-report measures of affect also suggests that positive and negative affect may be two unipolar and largely independent dimensions (Zevon & Tellegen, 1982; Tellegen, 1985; Watson & Tellegen, 1985). This implies that the negatively worded items of the anxiety inventory and Zung's depression scale may be not so much measuring the presence of trait-anxiety and trait-depression but rather the trait of (reversed) positive affectivity, i.e., the absence of positive affect. I n fact, the latter has been proposed as the distinguishing characteristic of anxiety and depression (Tellegen, 1985; Watson, Clark, & Carey, 1988; Watson & Kendall, 1989). As may be remembered it also turned out to be the first factor in our factor-analytic results. Furthermore, in this context, our findings on the consistent score differences for symptom-positively versus symptom-negatively worded items would not represent an imbalance of any sort but simply reflect the different positions of our subjects on the independent dimensions of positive and negative affect. REFERENCES
D. C. (1982) State vs Trait anxiety: a case study in confirmaBERNSTEIN, I. R., & EVELAND, tory Factor analysis. Personality and Individual Differences, 3, 36 1-372. BLUMENTHAL, M. D. (1975) Measuring depressive symptornatology in a general population. Archives of General Psychiatry, 32, 971-978. BONKE,B., SMORENBURG, J. M. J., ENT,C. K. VAN DER,& SI~IELBERGER, C. D. (1987) Evidence of denial and item intensity specificity in the State-Trait Anxiety Inventory. Personality and Individual Dzfferences, 8, 185-191. CROWNE, D., & -OWE, D. (1960) A new scale of social desirability independent of psychopathology. Journal of Consulting Psychology, 24, 349-354. DEVITO, A . J., & KUBIS, J. F. (1983) Alternate forms of the State-Trait Anxiety Inventory. Educational and Psychological Measurement, 43, 729-734. EDWARDS, A. L. (1957) The social desirability variable in personolio assessment and research. New York: Holt, Rinehart & Winston. GOTLIB,I. H., & CANE,D. B. (1989) Self-report assessment of depression and anxiety. In I?
560
J. MOOK, E T A L
C. Kendall & D. Watson (Eds.), AnxieQ and depression, distinctive and overlapping features. San Diego, CA: Academic Press. Pp. 131-169. R., & STANCER, H . (1979) Evidence that depression rating scales primarily measure LANGEVIN, a social undesirability response set. Acta Pr/chiatrica Scandinavica, 59, 70-79. LINDEN,W., PAULI-IUS, D. L., & DOBSON, K. S. (1986) Effects of response styles on the report of psychological and somatic distress. Journal of Consulting and Clinical Psychology, 54, 309-313. MOOK,J., KLEIJN,W. CHR., & PLOEG,H. M. V A N DER (1989) Depressiviteit als dispositie gemeten met de Zung-schaal: Interne structuur en relaties met angst, boosheid, coping en sociale steun [Measuring trait-depression with the Zung scale: internal structure and relations with anxiety, anger, coping and social support]. Nederhnds Tijdschrift voor de Psychologie, 44, 328-340. MOOK,J., PLOEG,H. M. V A N DER,& KLEIJN, W. CI-IR.(1990) Anxiety, anger and depression: relationships at the trait level. Anxiety Research, 3, 17-31. PLOEG,H. M. V A N DER (1989) A cross-cultural study of the dimensionality of the State-Trait Anxiety Inventory. In R. Schwarzer, H . M. van der Ploeg, & C. D. Spielberger (Eds.), Advances in test anxieo research. Vol. 6. Lisse: Swets & Zeitlinger. Pp. 195-202. PLOEG,H. M. VAN DER,DEFARES,P. B., & SPIELBERGER, C. D. (1980) Handleiding bij de Zev-Beoordelings Vragenligst, ZBV [Manual for the Dutch adaptation of the STAI-Yl. Lisse: Swets & Zeitlinger. SPIELBERGER, C. D. (1983) Manual for the State-Trait Anxiety Inventory, STAl-Form Y. Palo Alto, CA: Consulting Psychologists Press. SPIELBERGER, C. D. (1985) Assessment of state and trait anxiety: conceptual and methodological issues. Southern Psychology, 2, 6-16. C. D., GORSUCH, R. L., & LUSHENE, R. E. (1970) STAI-manual for the State-Trait SPIELBERGER, Anxiety Inventory ("Se[f-eualuation Questionnaire"). Palo Alto, CA: Consulting Psychologists Press. L. G. SPIELBERGER, C. D., VAGG,l? R., BARKER,L. R., DONHAM,G. W., & WESTBERRY, (1980) The factor structure of the State-Trait Anxiety Inventory. In I. G. Sarason & C. D. Spielberger (Eds.), Stms and anxiety. Vol. 7. Washington, DC: Hemisphere. Pp. 95-109. TANAKA-MATSUMI, J., & KAMEOKA,V. A. (1986) Reliabilities and concurrent validities of popular self-report measures of depression, anxiety and social desirability. Journal of Consulting and Clinical Psychology, 54, 328-333. TELLEGEN, A. (1985) Structures of mood and personality and their relevance to assessing anxiety with an emphasis on self-report. In A. H. Tuma & J. D. Maser (Eds.), Anxiety and the anxiety disorders. Hillsdale, NJ: Erlbaum. Pp. 681-706. VAGG,P. R., SPIELBERGER, C. D., & O'HEARN,T. P. (1980) Is the State-Trait Anxiety Inventory multidimensional? Personolio and Individrial Differences, 1 , 207-214. WATSON, D., CLARK,L. A,, & CAREY,G. (1988) Positive and negative affectivity and their relation to anxiety and depressive disorders. Journal of Abnormal Psychology, 97, 346-353. WATSON,D., & KENDAU,PH. C. (1989) Understanding anxiety and depression: their relation to nejptive,andgo+tive affective states. In Ph, C. Kendall & D. Watson (Eds.). Anxiety and epresslon, lstznchve and overlapping features. San Diego, CA: Academic Press. Pp. 3-26. A. (1985) Toward a consensual structure of mood. Psychological BuC WATSON,D., & TELLEGEN, letin, 98, 219-235. ZEVON,M. A,, & TELLEGEN, A. (1982) The structure ol mood change: an idiographic/nomothetic analysis. Journal of Personality and Social Psychology, 43, 111-122. ZUNG,W. W. K. (1965) A Self-rating Depression Scale. Archives of General Psychiatry, 12, 63-70. ZUNG,W. W. K. (1973) From art to science: the diagnosis and treatment of depression. Archives of General Psychiatry, 29, 328-337. ZUNG,W. W. K. (1986) Zung Self-rating Depression Scale and Depression Status Inventory. In N. Sartorius & T. A. Ban (Eds.), Assessment of depression. New York: Springer. Pp. 221-232.
Accepted September 1 7, 1991.
O Psychological Reports 1991
SYMPTOM-POSITIVELY A N D -NEGATIVELY WORDED ITEMS I N TWO POPULAR SELF-REPORT INVENTORIES OF ANXIETY A N D DEPRESSION ' JAAP MOOK WIM CHR. KLEIJN AND HENK M.
VAN DER
PLOEG
Vrije Univerriteit Amsterdam Summary.-Substantially higher mean scores on symptom-negatively versus symptom-positively worded items have consistently been reported in the literature for the balanced State-Trait Anxiety Inventory. In this study we aimed to replicate and extend these findings to Dutch trait versions of the inventory and Zung's similarly balanced Self-rating Depression Scale. Analysis indicated significantly higher mean subscale scores for symptom-negative as opposed to symptom-positive items of both measures, across sexes and age groups as well as across different levels of distress in nonclinical (n = 8631, subclinical (n = 450), and clinical subject samples (n = 96). Sex and age differences were mainly confined to symptom-positive subscales. Scale intercorrelations were lowest between symptom-positively and symptom-negatively worded scales both within and across measures. Factor analyzing the combined measures identified a symptom-negative and a symptom-positive factor, tentatively labeled "absence of positive affect" and "presence of negative affect." Several explanations of the findings among which item-intensity specificity, the response style of social desirability, and the trait model of posiuve and negative affectivicy are discussed.
Psychological assessment of affective states, traits, and clinical syndromes frequently occurs by means of self-report. Such inventories may contain a balanced number of symptom-positively and symptom-negatively worded items to counter certain response tendencies like acquiescence or "yea saying" from interfering too much with the measurement. Response formats of the symptom-negative items are subsequently reversed prior to scoring to ensure that all items are measuring the same underlying dimension or construct. O n e widely used affective self-report measure, the State-Trait Anxiety Inventory (Spielberger, 1983), for example, contains 10 items worded positively (e.g., "I feel nervous") and 10 items worded negatively (e.g., "I feel calm") in the State version, while in the Trait version the ratio is 11:9. Another example of a widely used instrument is the Self-rating Depression Scale (Zung, 1965, 1973, 1986) devised for the measurement of depression as a clinical syndrome and likewise consisting of 10 symptom-positively worded items (e.g., "I feel down-hearted and blue") and 10 negative statements (e.g., "Morning is when I feel the best").
'Address correspondence to Jaap Mook, Department of Medical Psychology, Vrije Universiteit Amsterdam, Van der Boechorststraat 7, # D-556, 1081 BT Amsterdam, The Netherlands.
552
J. MOOK, ETAL.
Empirically, research with the anxiety inventory has rather consistently yielded different results for the two sets of items. Spielberger, Gorsuch, and Lushene .(1970), in developing the inventory, had noted differences in mean state item scores under different severities and kinds of stress. In general, negatively worded items differentiated better when stress was lower whereas symptom-positively worded items did so when stress was higher (Spielberger, 1983). Differences in mean responses of the two sets for both the State and Trait version of the inventory have also been reported by Bernstein and Eveland (1982) who concluded that "its worst feature is that differences between positive and negative items are approximately as great as the differences between State and Trait items" (p. 372). Similar differences have subsequently been found by DeVito and Kubis (1983) for psychology students in a nonstressful situation and by Bonke, Smorenburg, Van der Ent, and Spielberger (1987) in a stressful situation (surgical patients on the day before the operation). In all these studies mean responses for the negatively worded items have consistently been substantially higher than that for the positively worded items. Moreover, in the Bonke, et al. (1987) study, differences were found irrespective of severity of State-anxiety which, as might be expected, for both men and women were appreciably higher in the pre- than in the postsurgical condition. The purpose of the present study is threefold. First, we wanted to know if the findings by Bonke, et al. (1987) on differences for positively and negatively worded items for the Dutch State version of the State-Trait Anxiety Inventory could be extended to the Dutch Trait version of the inventory and Zung's Depression Scale. Second, we examined the influence of sex, age, and general distress on the ~ ~ m p t o r n - ~ o s i t iand v e l symptom-negatively ~ worded subscale scores. Finally, to elucidate further the meaning of such differences in item scores, we intercorrelated the scales and explored their underlying structure by means of factor analysis.
Subjects The subjects participating in this study consisted of four groups: (a) 470 undergraduate students (39% men) in the social, medical, or health sciences (M age = 20.0 yr., SD = 3 3 , (b) 393 adults (45% men) from the general population of Leiden, The Netherlands (M age = 38.0 yr., SD = 14.0), (c) 450 medical patients (34% men) suffering from multiple sclerosis (M age = 48.0 yr., SD = 12.0), and (d) 96 psychiatric patients (42% men) of whom 46 were outpatients (M age = 35.0 yr., SD = 6.0) and 50 were inpatients (M age = 39.0 yr., SD = 13.0). Instruments The Trait version of the Dutch State-Trait Anxiety Inventory (Van der
SYMPTOM-POSITIVELY AND -NEGATIVELY WORDED ITEMS
553
Roeg, Defares, & Spielberger, 1980) was given. This has 20 items of which 10 are worded positively and 10 are worded negatively. In addition, a Dutch Trait-adapted version of Zung's Self-rating Depression Scale (Mook, Kleijn, & Van der Ploeg, 1989; Mook, Van der Ploeg, & Kleijn, 1990) was adrninistered. This scale also has 20 items of which 10 are worded positively and 10 negatively.
Procedure All subjects participated voluntarily and anonymously in the study. The students were asked at the beginning of a lecture and filled out the questionnaires at the end of the lecture. The adults and medical patients were invited by letter and completed the forms at home, returning them by mail. Lastly, the psychiatric patients completed the scales as part of their intake or during a break in their daily programs. Differences Between Positively and Negatively Worded Scales Means, standard deviations, and t ratios used in assessing differences in scores on positively and negatively worded subscales of the Trait-Anxiety Inventory and Zung's Depression Scale are presented in Tables 1, 2, and 3 . In Table 1 score differences are compared for the total sample and for men and women separately. As can be seen in Table 1, means for the positively worded scales on both the State-Trait Anxiety Inventory and the Depression Scale are substantially lower than those for the negatively worded scales, differences being significant for the total sample as well as for men and women separately. Differences are clearly larger for the Zung scale than for the anxiety inventory and somewhat larger for men than for women. It can also be observed in Table 1 that whatever differences there are between the two tests are predominantly confined to their subscale scores for TABLE 1 MEAN TOTALA N D SYMPTOM-POSITIVE VERSUSSYMPTOM-NEGATIVE SUBSCALE SCORES FORDUTCHT R A ~VERSIONS T OF STATE-TRAIT ANXIETYINVENTORY A N D ZUNG S E L F - R A ~DEPRESSION G SCALEM WHOLESAMPLE( N = 1409) AND MENA N D WOMENSWARATELY (ns = 556 AND 853) Subiects
AU
Scales
State-Trait Zung
Men
State-Trait Zung
Women
State-Trait
*All t ratios: p < ,001
Total M
SD
39.9 38.1 38.9 37.1 40.5
10.4 8.3 10.5 7.9 10.3
Symptompositive M SD 18.6 16.2 17.9 15.5 19.0
5.4 4.3 5.5 4.0 5.4
Symptomnegative M SD 213 21.9 21.0 21.5 21.5
5.9 5.2 5.9 5.1 5.9
I*
-12.8 -31.7 -9.1 -21.4 - 9.3
554
J. MOOK, ET AL.
positively worded items. This also holds for the small, though significant, sex differences in total anxiety and depression scores (t,,,, = 2.7, p < .O1 and tI4,,, = 3.5, p < .OOJ.) which appear to be based mainly on differences in subscale scores for positively worded items. In Table 2 means and SDs for the two sets of items are presented for different age groups by dividing the total sample into four categories. As may be observed, essentially the same score differences are found, subscale scores for positively worded items for both tests being appreciably and significantly lower than scores for negatively worded items across all age groups. Similar differences were found when scores within age groups were compared for men and women separately. I t may be observed also in Table 2 that scale scores generally increase for increasingly older groups. This, however, is largely confounded with the results for the subgroups, who differ both in age and distress, as presented in Table 3. TABLE 2 MEANTOTALAND SYMPTOM-POSITNELY VERSUSSYMPTOM-NEGATIVELY WORDEDS ~ S C A L E SCORESFORDUTCHTRAITVERSIONS OF STATE-TRAIT ANXIETY INVENTORY AND ZUNG SCALE IN DIFFERENTAGEGROUPS SELF-RATING DEPRESSION Age Groups (4
Scales
Total
M
SD
Symptompositive
M
SD
Symptomnegative
M
t*
SD
0-20
yr. (n = 366) State-Trait Zung 21-40 yr. ( n = 518) State-Trait Zung 41-60 yr. (n = 387) State-Trait Zung 61-90 yr. (n = 138) State-Trait Zung *AlI t ratios: p < ,001.
In Table 3 means and SDs are given for subgroups with different total scores on anxiety and depression (means for students and adults as well as for psychiatric in- and outpatients have been combined, differences in total Anxiety and Depression scale scores being small and nonsignificant). As can be seen in Table 3, again essentially similar differences are obtained for groups differing in total distress. Likewise, the same significant differences were found when results within groups were compared for men and women separately. I n fact, the only exception was observed for the psychiatric inpa-
SYMPTOM-POSITIVELYAND -NEGATIVELY WORDED ITEMS
555
TABLE 3 MEANTOTALAND SYMPTOM-POS~ELY VERSUSSYMPTOM-NEGATIVELY WORDED SUBSCALE SCORES FORDUTCHTRAIT VERSIONSOF STATE-TRAIT ANXIETYINVENTORY A N D ZUNGSELFRATING
DEPRESSION SCALEFORSUBJECTSWITH DIFFERENT LEVELSOF ANXIETYAND DEPRESSION
Subjects
Total
Scales
(n)
M
SD
Symptompositive
M
SD
Symptom-
t*
negative M SD
Student/Adult Sample (n = 863) State-Trait Zung Medcal Sample (n = 450) State-Trait Zung Psychiatric Sample (n = 96) State-Trait Zung
36.4 34.5
8.8 6.5
17.0 14.2
4.8 3.1
19.4 20.3
5.0 4.6
-10.4 -32.1
42.8 42.1
11.6 8.9
19.8 18.6
6.1 4.7
23.0 23.5
6.6 5.5
-7.6 -14.4
56.0 50.2
9.8 9.0
26.4 21.7
5.5 5.3
29.6 28.7
5.2 5.6
-4.1 -9.0
*All t ratios: p < ,001.
tient group, where the difference on the Anxiety inventory, although in the expected hrection, was not statistically significant.
Analyses of Variance To investigate the influence of sex, age, and groups separately as well as their possible interactions, three-way analyses of variance were performed on the scales as the dependent variables. As might be expected, by far the largest main effects were obtained for groups on all scales of the Trait-Anxiety Inventory (Fs = 152.6, 116.8, and 134.3, p < .0001) and the Zung Depression Scale (Fs = 194.6, 178.8, and 123.2, pc.0001). In addition, significant main effects for sex and age were observed on both the total scales and positively worded subscales of the Trait-Anxiety Inventory, and Zung's Depression Scale. For sex, Fs = 6.4 (p < .01) and 14.3 (p c .001) on the anxiety inventory, and Fs = 10.6 and 18.7 ( p < .001) on the Zung scale. For age, Fs = 3.9 ( P C .05) and 9.4 ( p < .01) on the anxiety inventory, and Fs = 6.8 ( P C .01) and 23.6 (pc.001) on the Zung scale. However, neither the main effects on the negatively stated subscales nor first- and second-order interaction terms were statistically significant. Scale lntercorrehtions and Factor Analysis Scale and subscale intercorrelations were computed for the two subgroups differing maximally in total amount of distress. Values are presented in Table 4, which shows that the patterns of scale intercorrelations for the studentladult and psychiatric patient groups are highly similar, generally being somewhat stronger in the latter though (and probably reflecting a difference in sample size). Total scale correlations between the anxiety inventory and Zung's scale
556
J. MOOK, ET AL.
TABLE 4 PEARSON INTERCORRELATIONS BETWEENSCALESA N D SUBSCALES OF DUTCHTRAIT VERSIONS OF STATE-TRAIT ANXIETY INVENTORY AND ZWNG SELF-RATMG DEPRESSION SCALE FORSTUDENT/ADULT GROW(n = 863, UPPER TRIANGLE) AND FOR PSYCHM-rruc PATIENTS (n = 96, LOWER TRIANGLE) Scales State-Trait Anxiety Inventory 1. Trait-total 2. Symptom positive 3. Symptom negative
Zung Self-ratingDepression Scale 4. Trait-total 5. Symptom positive 6. Symptom negative Note.-Decimal points omitted.
1
2
3
4
5
6
89
90 60
76 67 69
67 68 53
62 49 62
77
90 41
90 88
58
83 74 72
74 74 57
73
57 72
88 89
56
are high, suggesting that both measure essentially s i d a r constructs. As can be further observed in Table 4, positively and negatively worded subscales for anxiety tend to correlate higher with their like subscales for depression than they d o with each other, while the same holds in reverse for both depression subscales. I n fact, the lowest correlations are consistently found between the positively and negatively worded subscales both within and across measures, suggesting that whatever the differences in anxiety and depression, these are to be found in these subscales. To elucidate further the meaning of the positively and negatively worded subscales a factor analysis was performed on the pooled items of the two tests for the total sample. A two-factor principal components solution with varimax rotation yielded a complete overlap of anxiety and depression items, with all symptom-negatively worded items loading on the first and all symptom-positively worded items loading on the second factor. Factor 1 was tentatively labeled "absence of positive affect," while Factor 2 was called "presence of negative affect."
DISCUSSION Our results on consistent differences in response to positively vs negatively worded items agree with earlier findings for the anxiety inventory in both the State and Trait versions (e.g., Spielberger, et al., 1970; Bernstein & Eveland, 1982; DeVito & Kubis, 1983; Bonke, et al., 1987). Extending our findings to a similarly balanced measure of trait-depression like the Zung Self-rating Depression Scale attests to the robustness of the phenomenon when other item-content is involved. Differences in item-content also appeared to be mainly responsible for the larger positive-negative differences on the Self-rating Depression Scale as compared to the Trait-Anxiety Inventory. As was observed in Table 1, these larger differences were mainly due to the lower mean scores on the
SYMPTOM-POSITIVELY AND -NEGATIVELY WORDED ITEMS
557
positively worded subscale. Inspection of mean item-scores showed that the lower scores for positively worded depression items partly resulted from an imbalance in the eight somatic items of the scale (five of which are worded positively) and partly reflected the presence of a suicide item which is also worded positively. Likewise, the somewhat lower mean total scores for depression as compared to anxiety may partly reflect this presence of somatic items in the former as compared to the latter scale. I t should be noted, however, that while there is little overlap between the two measures in actual item-content, at the symptom-level the situation is quite different. Gotlib and Cane (1989), for example, on the basis of a content-analysis using DSM-111-R criteria for anxiety and depression, conclude that 40% of the items on the Zung Depression Scale actually measure symptoms of anxiety whereas almost half of the items on the State-Trait Anxiety Inventory rneasure symptoms of depression. Thus, the high correlations between the two inventories as found in our study as well as by others (Tanaka-Matsumi & Kameoka, 1986) may partly reflect these common symptoms (Gotlib & Cane, 1989). I n a similar vein our factor-analytic results for positively and negatively worded items loading on different factors tend to agree with a number of factor-analytic studies of the State-Trait Anxiety Inventory (Spielberger, Vagg, Barker, Donham, & Westberry, 1980; Vagg, Spielberger, & OJHearn, 1980; Bernstein & Eveland, 1982; Van der Ploeg, 1989) and to some extent also on the Zung scale (see, e.g., Blumenthal, 1975). I n factor-analytic studies of the anxiety inventory distinct symptom-positively and symptom-negatively worded factors have consistently been found for the State and Trait versions, both separately and combined. These results were initially interpreted as reflecting "item-intensity specificity," i.e., positively worded items supposedly being more sensitive with greater anxiety (Spielberger, et al., 1970, 1980). I t should be noted, however, that both our results and those by Bonke, et al. (1987) on consistent positive-negative differences irrespective of State and Trait anxiety levels would tend to argue against such an interpretation. Moreover, positively and negatively worded subscales of the anxiety inventory in our study correlated equally high with total anxiety and depression scale scores at different levels of distress, a finding which likewise would seem to contradict the notion of specificity of item-intensity. More recently, it has been suggested that the negatively worded items of the anxiety inventory may assess a "state of well-being," i.e., a positive emotional state inversely related to anxiety and reflecting something more than merely low anxiety (Spielberger, 1985). The notion of specificity of item-intensity having been found wanting, how can our results be explained? One possible explanation resides in the effects of response styles which generally may act as confounds in self-reports
558
J. MOOK, ETAL.
assessing a variety of stress-related characteristics including anxiety and depression (Langevin & Stancer, 1979; Linden, Paulhus, & Dobson, 1986; Tanaka-Matsumi & Kameoka, 1986). The most likely candidate here appears to be social desirability (Edwards, 1957; Crowne & Marlowe, 1960). Both the Trait-Anxiety Inventory and the Zung scale have correlated substantially with two measures of social desirability in college students (Tanaka-Matsumi & Kameoka, 1986). Social desirability is generally defined as the tendency to endorse items that are socially desirable while not endorsing items that are socially undesirable (Edwards, 1957). Thus, such a stylistic variable might be expected generally to exert a lowering effect on positively worded items which due to their negative affective content may be considered as socially undesirable. This could very well account for the sex and age differences found in our study (as well as in others, see, e.g., Bonke, et al., 1987; Blumenthal, 1975) which, as may be remembered, were predominantly due to differences on the positively worded subscales. For both women and older people it may be more acceptable to admit negative affect than for men and younger people. Conversely, social desirability might be expected to exert a heightening effect on negatively worded items which due to their generally positively toned affective content may be considered as socially desirable. The net effect of the social desirability variable then would be to produce more or less balanced scores on the positively and negatively worded subscales. However, this is not what we found; negatively worded subscales instead had significantly larger mean scores than positively worded subscales. I t is difficult to see how this imbalance can be reconciled with the social desirability notion unless we accept the possibhty that this response style operates differentially on positively and negatively worded items. In this respect it 'is unfortunate that Tanaka-Matsumi and Kameoka (1986) did not report separate correlations for positively and negatively worded subscales which might have shown different results. O n the other hand, it could be argued that social desirability operates more strongly in students (tested by Tanaka-Matsumi and Kameoka) than in psychiatric patients. Inspection of our raw data, comparing the students and adults with the psychiatric patients, indicated that this indeed might have been the case. The percentage of low scores (values 1-2) to high scores (values 3-4) on the State-Trait Anxiety Inventory for the students and adults was 54:46, and for the psychiatric patients 60:40, i.e., more balanced in the former. This result again seems to underline the difficulty of explaining our positive vs negative score differences solely in terms of social desirability, especially for the psychiatric patients. The second explanation, and the one we favour, asserts that our consistent positive-negative differences reflect the workings of two independent dimensions of positive and negative affect. Self-report inventories of affect
SYMPTOM-POSITIVELYAND -NEGATIVELY WORDED ITEMS
559
like those we used here, containing a balanced number of positively and negatively worded items, share an implicit assumption with regard to perand bipolarity of positive sonal affective structure, that of unidimen~ionalit~ and negative affect. This is implied by the reversal of response weights for the symptom-negative or construct-absent items prior to scoring, assuming construct-presence thereafter. Thus, in the Trait-Anxiety Inventory the weights for items like "I feel rested" and "I am happy" are reversed to "almost never" (4), "sometimes" (3), "often" (2), and "almost alwaysm(1). But surely, admitting that one only rarely feels rested or happy is not necessarily equivalent to feeling anxious most of the time. O n the Self-rating Depression Scale the weights for items like, "Morning is when I feel the best" and "I still enjoy the things I used to do," are similarly reversed. However, stating that one seldom feels at one's best during the morning or only rarely enjoys things the way one used to d o is not necessarily the same as feeling worst during the morning or not at all enjoying things most of the time. Research in personal affective structure underlying self-report measures of affect also suggests that positive and negative affect may be two unipolar and largely independent dimensions (Zevon & Tellegen, 1982; Tellegen, 1985; Watson & Tellegen, 1985). This implies that the negatively worded items of the anxiety inventory and Zung's depression scale may be not so much measuring the presence of trait-anxiety and trait-depression but rather the trait of (reversed) positive affectivity, i.e., the absence of positive affect. I n fact, the latter has been proposed as the distinguishing characteristic of anxiety and depression (Tellegen, 1985; Watson, Clark, & Carey, 1988; Watson & Kendall, 1989). As may be remembered it also turned out to be the first factor in our factor-analytic results. Furthermore, in this context, our findings on the consistent score differences for symptom-positively versus symptom-negatively worded items would not represent an imbalance of any sort but simply reflect the different positions of our subjects on the independent dimensions of positive and negative affect. REFERENCES
D. C. (1982) State vs Trait anxiety: a case study in confirmaBERNSTEIN, I. R., & EVELAND, tory Factor analysis. Personality and Individual Differences, 3, 36 1-372. BLUMENTHAL, M. D. (1975) Measuring depressive symptornatology in a general population. Archives of General Psychiatry, 32, 971-978. BONKE,B., SMORENBURG, J. M. J., ENT,C. K. VAN DER,& SI~IELBERGER, C. D. (1987) Evidence of denial and item intensity specificity in the State-Trait Anxiety Inventory. Personality and Individual Dzfferences, 8, 185-191. CROWNE, D., & -OWE, D. (1960) A new scale of social desirability independent of psychopathology. Journal of Consulting Psychology, 24, 349-354. DEVITO, A . J., & KUBIS, J. F. (1983) Alternate forms of the State-Trait Anxiety Inventory. Educational and Psychological Measurement, 43, 729-734. EDWARDS, A. L. (1957) The social desirability variable in personolio assessment and research. New York: Holt, Rinehart & Winston. GOTLIB,I. H., & CANE,D. B. (1989) Self-report assessment of depression and anxiety. In I?
560
J. MOOK, E T A L
C. Kendall & D. Watson (Eds.), AnxieQ and depression, distinctive and overlapping features. San Diego, CA: Academic Press. Pp. 131-169. R., & STANCER, H . (1979) Evidence that depression rating scales primarily measure LANGEVIN, a social undesirability response set. Acta Pr/chiatrica Scandinavica, 59, 70-79. LINDEN,W., PAULI-IUS, D. L., & DOBSON, K. S. (1986) Effects of response styles on the report of psychological and somatic distress. Journal of Consulting and Clinical Psychology, 54, 309-313. MOOK,J., KLEIJN,W. CHR., & PLOEG,H. M. V A N DER (1989) Depressiviteit als dispositie gemeten met de Zung-schaal: Interne structuur en relaties met angst, boosheid, coping en sociale steun [Measuring trait-depression with the Zung scale: internal structure and relations with anxiety, anger, coping and social support]. Nederhnds Tijdschrift voor de Psychologie, 44, 328-340. MOOK,J., PLOEG,H. M. V A N DER,& KLEIJN, W. CI-IR.(1990) Anxiety, anger and depression: relationships at the trait level. Anxiety Research, 3, 17-31. PLOEG,H. M. V A N DER (1989) A cross-cultural study of the dimensionality of the State-Trait Anxiety Inventory. In R. Schwarzer, H . M. van der Ploeg, & C. D. Spielberger (Eds.), Advances in test anxieo research. Vol. 6. Lisse: Swets & Zeitlinger. Pp. 195-202. PLOEG,H. M. VAN DER,DEFARES,P. B., & SPIELBERGER, C. D. (1980) Handleiding bij de Zev-Beoordelings Vragenligst, ZBV [Manual for the Dutch adaptation of the STAI-Yl. Lisse: Swets & Zeitlinger. SPIELBERGER, C. D. (1983) Manual for the State-Trait Anxiety Inventory, STAl-Form Y. Palo Alto, CA: Consulting Psychologists Press. SPIELBERGER, C. D. (1985) Assessment of state and trait anxiety: conceptual and methodological issues. Southern Psychology, 2, 6-16. C. D., GORSUCH, R. L., & LUSHENE, R. E. (1970) STAI-manual for the State-Trait SPIELBERGER, Anxiety Inventory ("Se[f-eualuation Questionnaire"). Palo Alto, CA: Consulting Psychologists Press. L. G. SPIELBERGER, C. D., VAGG,l? R., BARKER,L. R., DONHAM,G. W., & WESTBERRY, (1980) The factor structure of the State-Trait Anxiety Inventory. In I. G. Sarason & C. D. Spielberger (Eds.), Stms and anxiety. Vol. 7. Washington, DC: Hemisphere. Pp. 95-109. TANAKA-MATSUMI, J., & KAMEOKA,V. A. (1986) Reliabilities and concurrent validities of popular self-report measures of depression, anxiety and social desirability. Journal of Consulting and Clinical Psychology, 54, 328-333. TELLEGEN, A. (1985) Structures of mood and personality and their relevance to assessing anxiety with an emphasis on self-report. In A. H. Tuma & J. D. Maser (Eds.), Anxiety and the anxiety disorders. Hillsdale, NJ: Erlbaum. Pp. 681-706. VAGG,P. R., SPIELBERGER, C. D., & O'HEARN,T. P. (1980) Is the State-Trait Anxiety Inventory multidimensional? Personolio and Individrial Differences, 1 , 207-214. WATSON, D., CLARK,L. A,, & CAREY,G. (1988) Positive and negative affectivity and their relation to anxiety and depressive disorders. Journal of Abnormal Psychology, 97, 346-353. WATSON,D., & KENDAU,PH. C. (1989) Understanding anxiety and depression: their relation to nejptive,andgo+tive affective states. In Ph, C. Kendall & D. Watson (Eds.). Anxiety and epresslon, lstznchve and overlapping features. San Diego, CA: Academic Press. Pp. 3-26. A. (1985) Toward a consensual structure of mood. Psychological BuC WATSON,D., & TELLEGEN, letin, 98, 219-235. ZEVON,M. A,, & TELLEGEN, A. (1982) The structure ol mood change: an idiographic/nomothetic analysis. Journal of Personality and Social Psychology, 43, 111-122. ZUNG,W. W. K. (1965) A Self-rating Depression Scale. Archives of General Psychiatry, 12, 63-70. ZUNG,W. W. K. (1973) From art to science: the diagnosis and treatment of depression. Archives of General Psychiatry, 29, 328-337. ZUNG,W. W. K. (1986) Zung Self-rating Depression Scale and Depression Status Inventory. In N. Sartorius & T. A. Ban (Eds.), Assessment of depression. New York: Springer. Pp. 221-232.
Accepted September 1 7, 1991.
