Person. indiuid. I?@ Vol. 8, No. 2, pp. 185-191, Printed in Great Britain
0191s8869/87 53.00 + 0.00 F’ergamon Journals Ltd
1987
EVIDENCE OF DENIAL AND ITEM-INTENSITY SPECIFICITY IN THE STATE-TRAIT ANXIETY INVENTORY BENNO BONKE,’ JACQUELINE M. J. SMORENBURG,’ C. KORS VAN DER ENT,’ and CHARLES D. SPIELBERGER~ ‘Department of Medical Psychology and Psychotherapy, Erasmus University, School of Medicine, Rotterdam, P.O. Box 1738, 3000 DR Rotterdam, The Netherlands *Department of Psychology, College of Social and Behavioral Sciences, University of South Florida, Tampa, (Received
Florida,
U.S.A.
5 May 1986)
Summary-The Dutch form of the State-Trait Anxiety Inventory (STAI) was administered to 188 surgical patients the day before their operations and 3 days later. Scores for the state anxiety-present (S-Anx+) items were consistently lower than for the state anxiety-absent (S-Anx-) items; scores on the S-Anxf items before and after surgery were close to minimum for males. Tentative explanations for these findings and consequences for scoring the STAI State Anxiety scale are indicated. The effects of item-intensity specificity, social desirability, and denial as a mechanism for warding off the distressing emotion of anxiety are discussed.
The Stute-Trait Anxiety Znoentory (STAI: Spielberger, 1984; Spielberger, Gorsuch and Lushene, 1970) is currently used more frequently than any other psychometric instrument to assess anxiety in experimental research and clinical practice (Spielberger, 1983). State Anxiety (S-Anxiety) and Trait Anxiety (T-Anxiety) are assessed with two separate STAI scales. A further distinction has also been made between anxiety-present (Anx+) and anxiety-absent (Anx-) factors (Bernstein and Eveland, 1982; Spielberger, Vagg, Barker, Donham and Westberry, 1980). Both types of items are included in the STAI and were previously used in the Afect Adjective Checklist, the first psychometric measure specifically constructed to assess state and trait anxiety (Zuckerman, 1960). Agreement with Anx+ items and disagreement with Anxitems indicate higher levels of S-Anxiety and T-Anxiety in scoring the STAI. The results of a number of studies of the factor structure of the STAI have consistently identified anxiety-present and absent factors, as well as state and trait anxiety factors (Spielberger et al., 1970, 1980). The Anx+ and Anxfactors were initially interpreted as reflecting ‘item-intensity specificity’, a psychometric concept that refers to the fact that Anxitems are more sensitive at low to moderate levels of anxiety, whereas Anx + items are better measures when the anxiety level is relatively high (Spielberger et al., 1970, 1980). More recently, it has been suggested that the Anx - items may assess a ‘state of well-being’, a positive emotional state inversely related to anxiety, but reflecting something more than merely a low level of intensity of S-Anxiety (Spielberger, 1985). In a study in which the factor structure of the STAI was confirmed to contain state and trait Anx + and Anx - factors, Bernstein and Eveland (1982) raised questions about how Anx + and Anx - factors should be interpreted. They suggested that the endorsement of Anx+ items and disavowing Anx - items was not equivalent in the assessment of S-Anxiety (p. 361). They reported higher mean scores (higher S-Anxiety) for the STAI S-Anxitems than for the S-Anx+ items in Spielberger et al.‘s (1980) Air Force and High School samples, The mean Anx - items were 0.66 greater for the Air Force sample, and 0.89 greater for the High School sample, out of a total scale range of 3.0 points. Devito and Kubis (1983) reported similar differences between S-Anx + and S-Anxitems for introductory psychology students. Thus, in these studies, S-Anx - items consistently contribute more to S-Anxiety scale scores than S-Anx+ items. In this study, differences in responses to the S-Anx + and S-Anx - items of a Dutch version of the STAI (STAI-DY; Van der Ploeg, Defares and Spielberger, 1980) were evaluated under stressful conditions. The STAI-DY S-Anxiety scale, which contains a balanced number of anxiety-present 185
&two
186
and absent surgery.
items, was administered
to surgical
B~NKE
et
patients
al.
immediately
before and several days after
METHOD Subjects
Two groups of consenting surgical patients were the subjects of the study: 89 in Group I (56 female. 33 male), and 99 in Group II (59 female, 40 male). Most patients had plastic or reconstructive surgery and a few had dental surgery, all under general anaesthesia. The operations were performed at the University Hospital Rotterdam, between March and June 1983 (Group I), or during the same period in 1984 (Group II). The patients ranged in age from 18 to 80 yr (median age = 34.0 yr).
The STAI S-Anxiety Scale consists of 20 self-descriptive items, e.g. “I feel tense”. Each item is scored on a 4-point scale to which patients respond by rating the intensity of their feelings at a particular moment (“Report how you feel ‘right now’.“). For the S-Anx+ items (e.g. “I feel tense”), scores of 1, 2, 3, and 4 correspond with intensity ratings of ‘not at all’, ‘somewhat’, ‘moderately so’, and ‘very much so’. For the S-Anxitems (e.g. “I feel calm”), the same response alternatives are used, for which the scoring weights are 4, 3, 2, and 1, respectively. S-Anxiety scale scores are computed by simply summing the weighted scores for each item, and can range from 20 to 80. Higher scores indicate higher levels of S-Anxiety. Procedure
During the afternoon of the day before surgery, all patients completed the STAI-DY S-Anxiety scale. Three days later, on the second day after surgery, the inventory was readministered. The data for the pre-surgery and post-surgery S-Anxiety measures of any patient who failed to respond to three or more items were eliminated from the study. When one or two items were omitted, the mean item-values for the remaining S-Anx + or S-Anx - items were inserted as the score for each omitted item. This procedure was used in deriving S-Anx+ and S-Anxsubscale scores, but scores for the substituted items were disregarded in the item analysis. RESULTS
AND
DISCUSSION
Item means for the S-Anxand S-Anx+ items, computed separately for males and females in the two patient groups, are reported in Table 1. Pre-surgery S-Anxiety scores were higher than post-surgery scores for both patient groups. Moreover, with only a single exception (‘secure’ for males), the mean pre- and post-surgery item scores for men and women in both groups were higher for the S-Anxitems than for the S-Anx+ items. The average score of about 1.6 for the S-Anx+ items (range of I .0-4.0) revealed that most patients acknowledged the presence of less than ‘some’ anxiety as measured by these items on the day before an impending surgical event. On the other hand. the patients reported discomfort, as measured by the S-Anxitems, with average scores above 2.0. The results of the present study are similar to those of Devito and Kubis (1983), who administered the STAI (Form X) to 302 subjects in a neutral (nonstressful) condition. These investigators reported S-Anx+ item means of 1.56 and 1.61 for males and females, whereas the means for the S-Anxitems were 2.13 and 2.12, respectively. Although these means cannot be compared directly with our results because the STAI-DY is based on the revised STAI (Form Y), there was a similar tendency for S-Anx + items to yield lower scores than the S-Anx - items. Thus, irrespective of the presence of a stressful event such as surgery, the means for the S-Anxf items tend to be lower than scores on the S-Anxitems. The findings in the present study were also consistent with those of Bernstein and Eveland (1982), and can be interpreted as reflecting differences in item-intensity specificity between S-Anx+ and S-Anx - items, as suggested by Spielberger et al. (1980). The higher means for the S-Anxitems confirm earlier findings that these items are more sensitive at lower intensity levels, whereas most
Denial in the State-Trait Anxiety Inventory Table
1. Pre- and post-operative item means for each item of the STAI S-Anxiety two groups of male (M) and female (F) surgical patients Group
I
Group
Pre STAI Item No.
Keywords
2 5 8 10 11 15 16 19 20
Post F 56
secure at ease satisfied comfortable self-confident relaxed content steady pleasant
2.00 1.67 2.36 2.38 2.39 2.18 2.45 2.18 2.48 2.45
2.45 2.05 2.48 3.16 2.91 2.68 2.89 2.49 2.73 2.68
mean s.d.
2.25 0.26
2.65 0.31
tense strained upset misfortunes frightened nervous jittery indecisive worried over-excited
1.94 1.64 I .27 1.52 1.30 1.67 1.85 1.28 I .50 I .39
2.02 I .84 1.16 1.78 1.73 1.95 I .93 I .27 1.77
1.41 1.21
1.55 1.17 I .45 1.32 1.60 1.57 1.51 I .62 I .23
mean s.d.
1.54 0.24
1.69 0.30
1.47 0.17
scale for
II
Pre
M 33
N’
Post
M 30
F 41
M 40
F 59
M 28
F 55
1.83 2.07 1.97 2.17 2.00 2.13 2.07 2.20 2.07
2.09 2.04 2.40 2.60 2.49 2.30 2.45 2.23 2.53 2.34
2.20 1.73 2.16 2.47 2.58 2.08 2.26 2.30 2.33 2.53
2.61 2.24 2.66 3.11 2.84 2.63 2.83 2.45 2.71 2.78
1.93 1.57 2.00 2.30 2.32 2.00 2.00 2.00 2.36 2.14
2.13 1.96 2.22 2.44 2.43 2.38 2.40 2.02 2.55 2.35
2.02 0.16
2.35 0.18
2.26 0.25
2.69 0.24
2.06 0.23
2.29 0.20
1.66
1.93 1.83 1.20 1.43 1.33 1.74
1.54
1.55 1.28
2.14 1.88 1.29 1.67 1.62 2.07 I .96 1.29 1.81 1.61
1.25 1.29 1.32 1.32 1.32 I .25 1.11
I .65 1.75 I .20 I .43 I .28 1.60 1.56 I .37 I .41 1.40
1.53 0.25
1.72 0.32
I .2-l 0.13
I .47 0.17
(Anx -)
Anxiety-absent
1
187
calm
I .67
Anxiety-presenl (Anx +) 3 4 6 I 9 12 13 14 17 I8
1.50
I .43 I .03 1.40 I .03
1.20 1.30 I .40 I .43
1.30 0.17
1.66 1.38
aNumber (N) of patients in the groups on which each mean was based; “Post” due to discharge of patients before the second STAI administration.
I .25 I .07
N’s are smaller,
S-Anx+ items only come into play at moderate to high levels of anxiety. Almost all patients who scored high in S-Anx+ also scored high in S-Anx-, and few patients with low S-Anx- scores had elevated S-Anx+ scores (see Figs 1 and 2). Thus, the findings indicate that both types of items measure a similar anxiety construct. Differences in the responses to S-Anx+ and S-Anx- items may also be due, in part, to denial. Denial as a defence against distressing emotions may explain the finding that some patients fail to report the presence of any anxiety in responding to the S-Anx+ items under highly stressful 40
A
1
A
35.
A
0
A
A 30 -
A A
I ; a
I v)
OA
0 25-
:O 000 0
OA
AOQ) OA
00 OOQ,
20 -
A
m
00 A
OAW
0:ooo &Y 00 m 0
15.
0
0
0
0
0
0
0 0 10 .
ih
m” 9
A
0 15
20
S-Anx
25
30
35 36
+
Fig. 1. S-Anx+ scores (item 9 deleted) and S-Anx- scores of 72 male patients who report being “not at all” frightened (O), or at least “somewhat” frightened (A), before an impending surgical event.
188
40.
A 0 0
35 -
30
-
I 5 U I v,
25-
00
A
A A A
0
A A
A
A o A A AAOLAA AA A A-AA A 0 Am A A MA Am A o A AA A oaa 00 m ca m AOA:AA aoao cm A
Q)O
mo
20
0
104 -G/,
m
A
A
A A
0 A
A
0
0
000 OA
0
3
15
20
S-Anx
25
30
35 36
+
Fig. 2. S -Anx + scores (item 9 deleted) and S-Anx - scores of 114 female patients who report being “not (0). or at least “somewhat” at all” frightened frightened (A). before an impending surgical event.
circumstances of imminent surgery. Denial is most often observed in persons experiencing intense anxiety (Freud, 1936/1968). Patients who deny their feelings, typically do not acknowledge the presence of such feelings when someone inquires about them, though the emotion itself is presumed to remain active at an unconscious or subconscious level. Denial is looked upon as a mechanism stemming from early life, in which a child protects himself from the realities of the outside world by denying these realities (Freud, 1936/1968, p. 70). In a similar way, adults may use denial as a last resort to cope with anxiety. Both its early origin, and its regular occurrence in situations of severe distress. make it likely for denial to occur as an ‘ultimum refugium’ in states of high anxiety. Evidence of defensiveness in responding to the STAI in a stressful environment was noted by Redfering and Jones (1978) who concluded from their findings in a study of aviation cadets, that “there is the conspicuous possibility that defensiveness, or an attempt to ‘look good’ was mediating responses to appear less concerned and more competent to deal with the stressful conditions” (p. 87788). The possibility that denial was used as a defence against anxiety in the present study was evaluated by examining patients’ responses to two particular S-Anx+ items. In responding to item 9, which inquired directly about feelings of fear (“I am frightened”), 76.4% of the males and 49.1% of the females reported that they were ‘not at all’ frightened (score = 1). These patients were in hospital, right before surgery with general anaesthesia, and nevertheless denied being frightened. Post-operatively. the percentages of patients who were ‘not at all’ frightened were even higher. The pre-operative S-Anxf and S-Anxscores of patients with minimum scores on item 9 are compared with the remainder of the total sample in Figs 1 and 2 which report the results for males and females, respectively. Quite a few patients, who reported they were not at all’ frightened, nevertheless showed moderate to high anxiety as measured by the S-Anxsubscale, and moderate anxiety as measured by the S-Anxf subscale. Although the findings for males and females were similar, the females’ scores were higher on both the S-Anxf and S-Anx - subscales. Patients with low, moderate or high anxiety, as measured by the full S-Anxiety scale, are compared in Table 2. Moderate anxiety was defined by scores in the range p - 0.430 to ,U + 0.430 (,u = 36.4, (r = 10.3 for males. p = 38.8. 0 = 13.2 for females) based on a Dutch reference-group of randomly selected inhabitants of Leiden (Van der Ploeg, 1981, p. 3). Anxiety was considered to be low in patients scoring below p - 0.430 (lower 33%) and high in patients with scores over p -+ 0.430 (upper 33%). Inspection of Table 2 reveals that 98.7% of the patients who preoperatively reported that they were at least ‘somewhat’ frightened, showed moderate to high levels
Denial
in the State-Trait
Anxiety
Inventory
189
Table 2. Numbers of patients reporting that they were “not at all” or at least “somewhat” frightened or worried before a” impending surgical event S-Anxiety moderate
IOW
high
M
F
M
F
M
F
I 20
0 19
2 20
16 26
14 15
42 11
2 19
3 16
6 16
24 18
19 10
45 9
“fiiahrened” at least “somewhat” “not at all” “worried” at least “somewhat”
“not at all”
of anxiety as measured by the STAI S-Anxiety scale, but 65% of those who were ‘not at all’ frightened also had moderate to high S-Anxiety scores. Thus, a ‘not at all’ response to this item was associated with a wide variety of anxiety-levels. The finding that half of the male patients with high anxiety-levels described themselves as ‘not at all’ frightened suggests that some of these patients were denying their fear. Such denial apparently occurred less often in women: only one in five high-anxiety women patients reported they were ‘not at all’ frightened. Does ‘concern’ or ‘worry’ rather than fear contribute to high anxiety scores? The responses to S-Anx + item 17 (“I am worried”) are also tabulated in Table 2. When answered in the affirmative, item 17 was a strong indicator of anxiety: 92.6% of the males and 95.8% of the females who reported that they were at least ‘somewhat’ worried showed moderate or high S-Anxiety scores. In contrast, those who denied being worried (score = 1) showed a variety of anxiety scores, much the same as was the case for item 9. Approximately 30% of the males and 15% of the females reported that they were neither frightened, nor worried, but nevertheless had moderate to high S-Anxiety scores (Table 3). Such patients may be experiencing an emotional state involving discomfort or pain, an anxiety state which is partially suppressed by the patient (Hiifling and Butollo, 1985, p. 277), or an anxiety state that is unrecognized as such. Therefore, classification of these patients as experiencing moderate or high anxiety would not seem to adequately describe their emotional states, particularly with regard to the consequences for intervention in clinical situations (cf. Janis, 1958). The finding that patients with extremely low S-Anx + scores (e.g. 10 or 11) showed relatively wide ranges of S-Anxsubscale scores (Figs 1 and 2), raises further questions about the validity of the S-Anxiety scale for patients scoring extremely low in S-Anx+, and moderate to high in S-Anx - . Since our patients faced the threat of imminent surgery, one would expect elevations in at least some of the S-Anx+ items. If such elevations are not observed, this might be interpreted as evidence of denial. Further evidence of denial may be seen in the item-remainder correlations for the S-Anx + items, which ranged pre-operatively for males and females combined from + 0.45 (item 6) to +0.74 (item 3), and post-operatively, from +0.62 (item 14) to +0.85 (item 12). Since higher S-Anx+ scores are generally associated with higher S-Anxiety scores, especially in more stressful situations (pre-operatively), these findings are contrary to what would be expected on the basis of known differences between S-Anx + and S-Anx - items in item-intensity. In more stressful situations, one would expect higher item-remainder correlations for S-Anx+ items, than in less stressful situations. Such findings are consistent with the interpretation that some patients, especially males, appear to use the mechanism of denial to ward off the distressing emotion of anxiety as they await surgery in the highly stressful pre-operative situation. Table 3. Numbers of patients reporting that they were neither frightened nor worried, either frightened or worried, or both frightened and worried before a” impending surgical event S-Anxiety moderate
IOW
M fear and worry fear or worry no fear. no worry
I
I 19
high
F
M
F
M
F
0 3 16
I 6 15
IO 20 12
IO 12 6
38 10 5
BENNO
190
BONKE
et ul.
r-_-_J____., I Non-anxiety
’ emotional state :/cl +yF-;
: , 1
---------7
r--l High manifest anxiety
I?__________>
A
’
B
C
C________.‘_l_
-.
-
____
1 I : Ir
D
Fig. 3. Tentative algorithm for differentiating between persons experiencing anxiety and related emotional states from those who suppress, mislabel or deny anxiety, based on the pattern of responses to the S-Anx + and S-Anxsubscales of the STAI S-Anxiety scale.
In Fig. 3, a tentative schema for interpreting STAI scores is proposed which takes denial into account. This schema differentiates between anxiety states that are experienced and acknowledged by the patient (manifest anxiety) and the denial of anxiety symptoms. Patients who do not readily acknowledge anxiety, nor clearly deny such feelings, can also be identified. Using this algorithm, four different outcomes can be distinguished, labelled A, B, C, and D in Fig. 3, which are achieved via six different routes. Although cut-off points for S-Anxiety, and for S-Anx+ and S-Anx -, are arbitrary, one may use a tripartition based on the distribution of scores in a normative reference population. as was done in the present study. The schema presented in Fig. 3 was used in the present study, to compute the percentages of male and female patients for each of the four outcomes. Cut-off points for ‘low’ versus ‘moderate/high’ were based on the previously mentioned criteria. Minimal S-Anx+ was operationalized as scores of 10 or 1 I. The distribution of S-Anx - scores, for which no normative data were available in the literature, was divided into ‘low’ and ‘moderate/high’ scores on the basis of the 33rd percentile for the total sample (males and females separately). The percentages (and route) for each outcome for both sexes are reported in Table 4. As can be seen, males and females differed in outcome, and these differences were especially striking between the percentages of patients in categories A and B: 30.5% males versus 14.8% females. Approximately one of every five male patients appeared to deny anxiety (outcome B), as contrasted with only one in eight females. It should be noted, however, that these findings require replication, preferably with an independent measure of denial, or with an external criterion of denial, such as a stressful situation in which anxiety is considered absent by the patient, whereas objective observers (close relatives and/or friends) affirm its presence. While differences in the distribution of responses for the anxiety-present and anxiety-absent items in the present study support Spielberger et ul.‘s (1970) concept of item-intensity specificity (see
Denial
in the State-Trait
Anxiety
Inventory
191
Table 4. Percentages of 187 surgical patients for differentiating between persons with different “anxiety”-outcomes. based
on the presurgery
STAI
State
Anxiety item scores, with the algorithm described in the text Outcome’
Route
Males
Females
81 BZ Cl c2
9.7 19.4 1.4 22.2 2.8 44.4
2.6 12.2 0.0 x.7 5.2 71.3
99.9
100.0
A
B B C C D
‘See Figure
3 for details of letterings.
Table 2, Figs 1 and 2) denial of anxiety in responding to the S-Anx+ items also seems to have occurred quite frequently in this study, especially for the males. Assuming that denial was responsible for the low S-Anx+ scores of some of our patients, why should more males than females use denial? Possibly, social desirability motivates some males to suppress reporting anxiety symptoms because the content of the S-Anx+ items is less socially desirable for men than for women. To describe oneself as nervous, frightened or upset is likely to be experienced as psychologically more unacceptable and threatening or harmful by men than by women. Acknowledgements-The authors wish to thank the following medical students for their help in gathering the data: Frank Flier, Sandra Lako, Jorina Langerveld, and Marie-Anne Portengen. The departments of Anaesthesiology and Plastic- and Reconstructive Surgery of the University Hospital Dijkzigt. Rotterdam were kind enough to cooperate with us in every respect. Dr. H. J. Duivenvoorden gave valuable statistical advice, and our colleagues of the department of Medical Psychology and Psychotherapy of the Medical School of Rotterdam carefully read and reread earlier drafts of the manuscript and suggested indispensable alterations of the text. All those who helped us gather the data, explain the results and prepare the manuscript in one way or the other, are cordially thanked for their assistance.
REFERENCES Bernstein I. R. and Eveland D. C. (1982) State vs Trait anxiety: a case study in confirmatory factor analysis. Person. indiuid. 013 3, 361-372. Devito A. J. and Kubis J. F. (1983) Alternate forms of the State-Trait Anxiety Inventory. Educ. psychol. Measurmf. 43, 729-734. Freud A. (1968) The Ego and fhe Mechanisms of Defence. Hogarth, New York. (Originally published, 1936.) Hofling S. and Butollo W. (1985) Prospektiven einer psychologischen Operationsvorbereitung [Prospects of a psychological preparation to surgery]. Anaesfhesisf 34, 273-279. Janis I. L. (1958) Psychological Sfress. John Wiley, New York. Redfering D. L. and Jones J. G. (1978) Effects of defensiveness on the State-Trait Anxiety Inventory. Psychol. Rep. 43, 83-89. Spielberger C. D. (1983) Manual for fhe Slate-Trait Anxiety Inventory. Revised edn. Consulting Psychologists Press, Palo Alto, CA. Spielberger C. D. (1984) State-Trait Anxiefy Inventory: A Comprehensive Bibliography. Consulting Psychologists Press, Palo Alto, CA. Spielberger C. D. (1985) Assessment of state and trait anxiety: conceptual and methodological issues. Sourhern Psychol. 2, 616. Spielberger C. D., Gorsuch R. L. and Lushene R. D. (1970) Tesf Manualfor fhe Stare Trait Anxiety Inventory. Consulting Psychologists Press, Palo Alto, CA. Spielberger C. D., Vagg P. R., Barker L. R., Donham G. W. and Westberry L. G. (1980) The factor structure of the StateTrait Anxiety Inventory. In Stress and Anxiety, Vol. 7. (Edited by Sarason I. G. and Spielberger C. D.). Hemisphere, Washington, D.C. Van der PIoeg H. M. (1981) Manualfor the Self-Assessment Questionnaire, STAI-D Y; Addendum. Swets & Zeitlinger, Lisse. (In Dutch.) Van der Ploeg H. M., Defares P. B. and Spielberger C. D. (1980) Manualfor fhe Self-Assessment Questionnaire, STAI-DY. Swets and Zeitlinger, Lisse. (In Dutch.) Zuckerman M. (1960) The development of an Affect Adjective Check List for the measurement of anxiety. J. consult. Psychol. 24, 457462.