anxiety, were minimally related to locus of control and social desirability, and were essentially unrelated to the MMPI-168"s measures of depression and anxiety ...
Motivation and Emotion, Vol. 6, No. 4, 1982
Irrational Beliefs as Predictors of Negative Affective States I Damaris J. R o h s e n o w 2 University of Wisconsin-Madison Ronald E. Smith University of Washington
One effort to specify cognitive mediators related to dysfunctional emotional responses has been made by Ellis, who proposed that certain irrational beliefs elicit maladaptive emotional reactions. In order to test Ellis's hypotheses in a nonclinical sample, the Irrational Beliefs Test, two anxiety inventories, the MMPI-168, and a locus o f control scale were completed by 32 male heavy social drinker undergraduates, who then kept daily records o f their anxiety, anger, unhappiness, and drinking behaviors for 7 months. Four factors underlay the 10 beliefs but did not necessarily correlate with dysphoric affect. Certain irrational beliefs predicted daily unhappiness ratings over the 7-month period and daily anger ratings over 3 months, but the beliefs exhibited only a weak relationship with daily general anxiety ratings. Irrational beliefs, especially worrying about approval and competence, were strongly related to test anxiety and Spielberger's trait anxiety measure, were somewhat related to social anxiety and general anxiety, were minimally related to locus o f control and social desirability, and were essentially unrelated to the MMPI-168"s measures o f depression and anxiety and to drinking behaviors. Although 2 o f the irrational beliefs failed to exhibit relationships with negative affect, the other 8 show the 1This research was partially supported by grants from the Alcoholism and Drug Abuse Institute, University of Washington, Seattle, from a National Institute of Mental Health small grant AA03947-01, and from a National Institute of Alcohol Abuse and Alcoholism postdoctoral training fellowship No. 1T32AA07171-01. We would like to thank Jean Chapman, Ph.D., for her advice on the statistics and Dorothy Knapp, Ph.D., for her painstaking scoring and coding of all the data records. 2Address all correspondence to Damaris J. Rohsenow, Department of Psychology, University of Wisconsin-Madison, 1202 W. Johnson Street, Madison, Wisconsin 53706. 299 0146-7239/82/1200-0299503.00/0 © 1982 Plenum Publishing Corporation
300
Rohsenow and Smith
predicted relationships to moods, providing support for a cognitive mediational model of maladaptive emotions in a nonpathologieal population.
The cognitive mediation of emotional states has received increasing theoretical and empirical attention in recent years. Psychodynamic theorists have traditionally emphasized the role of cognitive and symbolic activities in affective arousal, as have social learning formulations (Bandura, 1969; Dollard & Miller, 1950; Staats, 1975). Bandura (1977), for example, has suggested that emotional responses may be elicited by two different sources of stimulation: environmental cues that directly elicit conditioned emotional responses, and cognitive events, such as thoughts and images, which more typically evoke emotional responses. Mediational models of emotion have played a prominent role in recent attempts to understand and change dysfunctional emotional responses such as depression, anxiety, and anger (Beck, 1976; Ellis & Grieger, 1977; Goldfried & Goldfried, 1980; Novaco, 1977; Sarason, 1978; Smith, 1980). Such models have stimulated research demonstrating the ability of covert verbalizations and symbolic activities to elicit emotional reactions. Rimm and Litvak (1969) found that subjects who read affectively toned sentences (e.g., "I might get injured or crippled") exhibited stronger psychophysiological responses than did subjects who read neutral (nonemotional) sentences, and later studies also found increased emotional arousal in response to internally evoked thoughts of an affective nature (May & Johnson, 1973; Russell & Brandsma, 1974; Velten, 1968). From a clinical perspective, a critically important issue is the manner in which individuals differ in their habitual affect-eliciting thought patterns and the consequences these individual differences have for their functioning and adaptation. Beck (1976) and Ellis (1962) have made important efforts to specify the content of thought patterns as they relate to specific classes of dysfunctional emotional/behavioral responses. In particular, Ellis has identified a number of irrational beliefs hypothesized to underly covert verbalizations that elicit maladaptive emotional responses, including the ideas that it is essential to be successful and approved of by virtually everyone in order to be worthwhile and that one should be terribly upset when misfortune occurs. Jones (1968) developed the Irrational Beliefs Test (IBT), consisting of factored measures of t0 irrational beliefs described by Ellis (see Table I). Goldfried and Sobocinski (1975) found significant relationships between most of the IBT scales and trait measures of social anxiety, debilitating achievement anxiety, and speech anxiety. They also found that subjects who
301
Irrational Beliefs and Mood Table
Scale 1 2 3 4 5 6 7 8 9 10
I. BeliefsAssessedby the Irrational BeliefsTest Belief
An adult needs to be loved and approved by almost every significant other person. One should be completely competent and achieving in all ways to be worthwhile. Certainpeople are thoroughly bad and should be severelyblamed and punished for their villainy. It is awful and upsetting when things are not the way one would very much like them to be. Unhappiness is caused by external events and people cannot control their emotional reactions. If something unpleasant might happen, one should keep dwelling on the possibility of its occurring. It is easier to avoid difficulties and responsibilities than to face them. One should be dependent and rely on stronger others. One's past history irrevocably determines one's present and future behavior. There always is a perfect and precise solution to human problems and it is awful if this solution is not found.
subscribed to the belief that it is necessary to be loved and approved of by others to be worthwhile scored higher in anxiety and anger when imagining situations involving social rejection than did subjects who did not subscribe to this belief. This same irrational need (as well as several others) was found to differentiate between socially anxious individuals undergoing social skills training and competent daters (Gormally, Sipps, Raphael, Edwin, & VarvilWeld, 1981). Several other studies have assessed relationships between irrational beliefs and trait measures of depression. Lapointe and Crandell (1980) reported that depressed college students scored higher than a nondepressed comparison group on all of the irrational beliefs, while Nelson (1977) found scores on the Beck Depression Inventory related to the IBT scales Dwell on Negatives, Need for Competence, Frustration Intolerance, and D o o m e d by Past. The purpose of the present study was to extend previous research concerning the relationship of irrational beliefs to negative affect in several ways. Virtually all previous research has related IBT scores to other trait measures, the exception being the Goldfried and Sobocinski study, in which one o f the beliefs was related to affect adjective checklist scores in a contrived laboratory setting. The present study tested the hypothesis that even in nonclinical populations, the acceptance o f irrational ideas increases the likelihood o f experiencing aversive emotional states in actual life situations. This was done by relating the IBT scales to daily m o o d ratings recorded by college males over a 7-month period. Both the frequency and the intensity o f anxiety, anger, and depression reactions were assessed so as
302
Rohsenow and Smith
to examine relationships between these two dimensions of emotional responsiveness and to relate them to irrational beliefs. It was also possible to test hypotheses derived from Ellis's formulations linking certain emotions to specific beliefs (e.g., that anger is most strongly linked to beliefs involving the blaming of others for one's frustrations). While the major focus of the present study was on IBT predictors of daily mood ratings, a number of trait measures of emotionality were also examined in relation to irrational beliefs. Self-report rating scales of trait general anxiety, test anxiety, social anxiety, and aggression anxiety were administered together with the trait scale of the State-Trait Anxiety Inventory (Spielberger, Gorsuch, & Lushene, 1970) and the Psychasthenia scale of the MMPI to determine if different forms of anxiety are related to different patterns of irrational beliefs. Relationships between IBT scales and the Depression, Paranoia (presumed to reflect anger), and Social Introversion scales of the MMPI were also assessed. Finally, Rotter's (1966) locus of control scale was administered to test the hypothesis that the tendency to endorse certain IBT beliefs (namely, External Causes, Blame Others, and Doomed.by Past) is related to an external locus of control. The focus of this study was on a normal population, and the MMPI was used to exclude college males who exhibited pathological profiles. Nevertheless, it was of interest to study subjects who, despite their appropriate level of psychological adjustment, had a behavior pattern that might be functionally related to the negative affective states assessed. The tendency to use alcohol within the heavy social drinker range is characteristic of about 40% of male students at the University of Washington (Marlatt, no date). Subjects falling within this range on the Drinking Practices Questionnaire (Cahalan, Cisin, & Crossley, 1969) were selected and kept daily records of their alcohol intake over the 7-month period. It was thus possible to relate both IBT scores and mood ratings to this specific behavioral index. Finally, although the response bias of acquiescence was controlled during scale development, the possibility of a social desirability response bias was never assessed. The availability of MMPI's K scale permitted an assessment of, and statistical correction for the potential role of, a social desirability response bias in the IBT scores and mood data.
METHOD
Subjects The subjects were 36 male undergraduate college students preselected on the basis of nonpathological scores on the MMPI-168 (a short form of
Irrational Beliefs and Mood
303
the Minnesota Multiphasic Personality Inventory; Overall & Gomez-Mont, 1974) and scores on the Drinking Practices Questionnaire (Cahalan et al., 1969) indicating at least 45 drinks per month and often 5 or more drinks at a time (M = 126 alcoholic drinks per month, SD = 98). The mean MMPI168 profile of the subjects yielded no T scores above 65 and all of the scale means were below 60, with the exception of the Masculinity-Femininity (M = 64.8) and Hypomania (M = 63.1) scales. Subjects were screened to rule out alcoholism, drug abuse, and major health problems. The subjects were offered $4 per week plus $50 for completing the entire study to encourage compliance. Of the 40 subjects who agreed to participate, 4 dropped out within the first 3 weeks because they found the data recording too onerous and 1 dropped out during the final month. Due to missing or unscorable data, complete data were obtained from 32 subjects at all points except 6 to 7 months, when data were obtained from only 27 subjects. Questionnaires
The subjects completed a packet of questionnaires within 1 week. Each subject completed the 100-item IBT (scored for 10 beliefs plus a total score), the State-Trait Anxiety Inventory-Trait form (STAI; Spielberger et al., 1970), the MMPI-168 (Overall & Gomez-Mont, 1974), and the InternalExternal Locus of Control Scale (I-E; Rotter, 1966). Additional measures of the frequency and intensity of general anxiety, test anxiety, social anxiety, and aggression anxiety were developed. On 6-point rating scales, ranging from never to always, subjects indicated how frequently they experienced an affective state (e.g., "How frequently in your daily life do you experience worry and tension? .... How frequently do you experience anxiety while you are taking a course exam? .... How frequently do you worry about how other people are judging and evaluating you?"). Each frequency scale was paired with a 6-point intensity rating scale ranging from not at all to extremely (e.g., "When you do experience worry and tension, how intense is it?"). Two pairs of items were used to assess general anxiety, three pairs for test anxiety, two pairs for social anxiety, and one pair for aggression anxiety. Data from the scales were analyzed using the sum of the frequencyintensity pairs. We will refer to these as the Social, Test, and Aggression Anxiety Scales (STAAS). Daily Mood and Drinking Records
Subjects were asked to rate on lO-point scales, at bedtime each night, the amount of time that day during which they felt angry, anxious, or depressed, the maximum intensity of these feelings when they occurred, and the number and types of alcoholic beverages consumed. The daily mood
304
Rohsenow and Smith
and drinking data were recorded on cards, which were returned to a secretary each week, at which time the subject was paid $4.
RESULTS
Reliability o f IBT Scales As the possible size of a correlation is limited by the reliability of the measures, Hoyt's reliability (coefficient alpha) was calculated for each IBT scale and for the STAAS scales. The alphas for each IBT scale are shown in Table II. The alphas for the STAAS scales range from .80 to .86 for social, test, and aggression anxiety, and is .70 for general trait anxiety.
Relationship o f Defensiveness to Irrational Beliefs and Daily Moods To assess the potential role of a form of defensiveness, the social desirability response bias, in the relationships between the IBT and mood ratings, these data were examined in relation to the K scale of the MMPI168. The K scale was unrelated to 8 of the 10 IBT scales and to the total IBT score. Significant relationships existed with only 2 of the scales: Dwell on Negatives (r = - . 3 8 , p < .05) and Problem Avoidance (r = - . 3 4 , p < .05). The K scale was unrelated to daily mood ratings of depression and anger at all time periods sampled, but was significantly related to anxiety mood ratings sampled at 6 weeks (r = - . 3 3 , p < .05), 3 months (r = - . 4 3 , p < .01), and 7 months (r = - . 4 8 , p < .01). In subsequent correlational analyses involving the IBT and anxiety mood ratings, the effect of K was therefore partialed out to control for this influence.
Irrational Beliefs and Mood Ratings The daily mood scores were sampled during the first 2 weeks of the study, at 6-7 weeks, 14-15"weeks, and 6½-7 months. The mean frequency and intensity ratings ranged from .5 for anger frequency (SD = .5) to 1.6 for anxiety intensity (SD = 1.3), with individual ratings ranging from 0 to 10 on each. Preliminary analyses revealed that the frequency and intensity scores were highly correlated for each emotion at all periods sampled. The analyses to be reported were thus based on frequency × intensity product scores for each mood. The product-moment correlations between the IBT scales and the mood product scores at each of the four time periods are presented in Table
.37 ~ .20 --. 15 .12 .18 -.13
,72 .62 .57 .55
.62 .72 .66 .45 .59 .59
2 wks
.33" .25 .39 a .29 .25
c~
.23 .35 a -.03 .07 .14 - .04
.31 a .17 .30" .33 a .08 .39" .22 - . 12 .12 .28 - .17
.32 a .27 .33" .20 .23
3 mos
Depression
6 wks
~N = 32 except at 6 months, when N = 27.
~p < .01.
~p < .05.
Total Approval Competence Blame others Frustration intolerance External causes Dwell on negatives Problem avoidance Rely on others Doorned by past Precise solutions
Irrational belief
-.13 ,07 .16 -.11
.20 ,42" -.17
.34" .16
.24 .17 .25 .20 .24
2 wks
-.15
.40 ~ .29
.47 b .37" .44 ~ .26 .33 ~
7 mos
.13 .01 -.02
-.0t
.04 .39"
.27 .19 .20 .34" .04
6 wks
.14 .33" .03 -.13 .11 -.17
.18 .18 .24 .16 .01
3 mos
Anxiety
.25 .13 .02
-.02
.01 .27
.20 .11 .01 .35" .15
7 mos
Table II. C o r r e l a t i o n s o f I r r a t i o n a l Beliefs with Daily Moods over Time ~
.18 .26 -.13
-.12
.47 b .25
.42 b .31" .39" .28 .36"
2 wks
-.11
.18 .33 ~
-.25
.42 ~ .25
.34" .29 .26 .26 .25
6 wks
-.12
.22 .31~
-.13
.42 ~ .33"
.41 b .35" .34" .30 a .20
3 mos
Anger
-.11
.32 .38"
-.11
.19 .09
.19 .10 -.04 .20 .09
7 mos
e~
306
Rohsenow and Smith
II. Since, with a large number of correlations, some could occur by chance, single correlations were not interpreted. The relationship of a belief to a mood was considered to be significant if it correlated significantly with the mood at two of the four time periods and to be strong if it correlated at three of the four time periods. The relationship of irrational beliefs to a mood was considered to be stronger if the total score correlated with the mood than if only a few beliefs were significantly correlated, although those individual beliefs could be important predictors of mood in themselves. The total IBT score was found to be consistently and positively related to depression and anger ratings, but not to anxiety ratings. To explore this finding further, the relationships between the affective states and each irrational belief were examined. Daily ratings of depression were correlated at all sampled time periods with IBT Scale 2 (defining self-worth in terms of competence) and at three of the four periods with Scale 5 (external causes are the determinants of one's fate). These two beliefs, particularly the latter, were also related to the experiencing of anger. Anger was also significantly correlated with another externally oriented belief, that one is doomed by past misfortunes (Scale 9). The only other belief related to either affective state at more than one time period was the conviction that it is necessary to gain the approval of others, which was significantly correlated with anger ratings at 2 weeks and 3 months. As shown in Table II, several other beliefs were significantly correlated with anger or depression at one time period, but were not consistently so related. In contrast with anger and depression mood ratings, anxiety ratings exhibited no consistent relationships with any of the IBT scales, although correlations were significant at half the time periods with Blame Others and Dwell on Negatives. Because the MMPI K scale was found to be related to anxiety ratings as well as to several of the IBT scales, partial correlations were calculated to control for the possible contribution of defensiveness. The pattern of nonsignificant relationships was not appreciably altered except for Blame Others, which was significantly correlated with anxiety at the final three time periods (pr = .34, .39, and .45, p < .05, .05, and .01, respectively). Although the frequency and intensity dimensions were highly correlated (r > .80) for all three of the mood states, supplementary analyses were performed to determine whether the frequency and intensity dimensions were differentially related to irrational beliefs. These analyses disclosed that while the relationships between specific irrational beliefs and depression, anger, and anxiety remained essentially unchanged, 12 significant correlations were obtained with frequency scores, compared with only 4 significant intensity-belief relationships.
Irrational Beliefs and Mood
307
Relationship to Trait A n x i e t y Scales
The relationship of irrational beliefs to general trait anxiety was studied by examining the univariate and multiple correlations of the IBT with the Speilberger STAI, the STAAS's general anxiety scales, and the P t scale of the MMPI (see Table III). In interpreting the results that follow, single correlations are, once again, not considered important. If the regression of the 10 IBT scales on a personality measure or the correlation of the IBT total score with the measure was significant, the relationship was considered to be strong. When the regression was not significant, at least 3 of the 10 scales had to have significant correlations to be considered significant. The STAI was highly correlated with the IBT total score, accounting for 24% of the variance. A regression of the IBT scales on STAI was significant (R 2 = .652, p < .01), with the significant predictors (p < .05) being Need for Competence (sr 2 = .09), Doomed by Past (sr 2 = .09), and Dwell on Negatives (sr 2 = .07). (The semipartial squared correlation, sr 2, reflects the amount of variance accounted for by the variable, and R 2 reflects the amount of variance accounted for by the regression equation.) In univariate correlational analyses, Need for Approval and External Causes were also significantly related to trait anxiety. The belief in Precise Solutions was significant in the opposite direction from that predicted by Ellis (sr 2 = .09, p < .05). Partial correlations controlling for K decreased the correlation with Need for Approval (pr = .28, p < .07), but the remaining relationships were not significantly affected. Table I i i . Correlations of Irrational Beliefs with Other Concurrent Measures Anxiety scale
Total Approval Competence Blame others Frustration intolerance External causes Dwell on negatives P r o b l e m avoidance Rely on strong others D o o m e d by past Precise solutions "p < .05.
~p < .01. ~p < .001.
STAI
General
Test
Social
Aggression
IE
.49 b .34 ~ .63 c .11 ,35 ~ ,31" ,43 b .18 - .03 .38" -.31"
.16 .07 .32" -.12 .30 ~ .24 .25 .38" - . 19 -.17 -.37"
.60 c ,53 ~ .40" ,29 .64 ~ .25 .61 ~ - ,09 .37" .17 .19
.36 ~ ,48 b .35" .10 .21 .1 t ,35 ~ .17 - .03 .23 -.30"
.09 ,322 .14 -.10 .11 - .03 ,31 ~ .06 .00 -.30" -.24
.30 a .36 a .15 .18 ,29 .14 .36" .07 .08 .06 -.27
308
Rohsenow and Smith
The STAAS general anxiety scale had a weaker relationship to the IBT. The regression of IBT scales on general anxiety was of borderline significance (R z = .495, p < .07). The univariate correlations showed that the strongest relationships were with Problem Avoidance, Need for Competence, and Frustration Intolerance. Again, Need for Precise Solutions was significant in the opposite direction. When K was partialed from the correlations, the relationship with Problem Avoidance decreased to borderline significance (pr = .25, p < . 10), but the other correlations were essentially unaffected. The P t scale of the MMPI was virtually unrelated to the IBT. The regression equation was nonsignificant, and only two correlations were significant (Precise Solutions, r = .31, Need for Approval, r = - . 3 4 , p's < .05), the latter in the opposite direction from prediction. The regression of IBT on the social anxiety scale did not reach significance. However, Need for Approval was highly correlated, and Need for Competence and Dwell on Negatives were also significantly correlated with social anxiety in univariate analyses. When K was partialed, the relationship with Dwell on Negatives became only borderline (pr = .27, p < .10). The Social Introversion (SO scale of the MMPI can be considered indirectly to reflect social anxiety. The regression equation was significant (R 2 = .573, p < .05), with Doomed by Past contributing positively (sta = • 12, p < .05) and Precise Solutions contributing negatively (sta -- . 10, p < .05) to the equation. Si was also significantly correlated with Need for Competence (r = .39, p < .05) and Problem Avoidance (r = .38, p < .05) in univariate analyses. Test anxiety exhibited the strongest relationship with irrational beliefs. The correlation of trait test anxiety with the IBT total score was very high, accounting for 36% of the variance, and the multiple regression of the IBT scales on test anxiety was highly significant (R 2 = .645, p < .01). Within the regression, only Frustration Intolerance attained significance (sr ~ --- . 12, p < .02), but in the univariate correlations, Dwell on Negatives and Need for Approval were highly significant, and Need for Competence and Rely on Strong Others were also significantly correlated. When K was partialed, all of these remained significant, and Blame Others attained significance as well (pr = .32, p < .05). Anxiety about aggression was only minimally related to the IBT. The regression equation was nonsignificant and, of the three significant univariate correlations (Need for Approval, Dwell on Negatives, and a negative correlation with Doomed by Past), the former two became of only borderline significance when K was partialed.
Irrational Beliefs and M o o d
309
Relationship to Depression and Paranoia on the M M P I The IBT was virtually unrelated to the MMPI's Depression scale (D). The D scale was significantly correlated only with Problem Avoidance (r = .35, p < .05), and when K was partialed this correlation became only borderline (pr = .29, p < .06). The only IBT scale correlated with the Pa scale was Frustration Intolerance (r = .38, p < .05). This relationship is in accord with predictions, but two other scales that one might expect to be related, Blame Others and External Causes, were not significantly correlated with Pa. It should be pointed out that since subjects showing significant psychopathology were excluded from the study, virtually all the subjects scored in the normal range on the MMPI scales discussed here. Relationship to Locus o f Control External locus of control was predicted to be related to three beliefs: Blame Others, External Causes, and Doomed by Past. However, it was found to be essentially unrelated to these three beliefs in this study. External locus of control was positively correlated with Need for Approval and Dwell on Negatives, and had a borderline correlation with Frustration Intolerance and a borderline negative correlation with Precise Solutions. When K was partialed, the significant correlations decreased to borderline significance (pr = .27 and pr = .22, respectively; p < . 10). Relationship to Drinking The daily drinking data were converted to standard drink units equivalent to 118 ml (4 oz) of 12% ethanol wine. The subjects drank a mean of 85.1 drinks per month (SD = 49.1). The data were summed over consecutive 2-week samples for analysis. The IBT total was not significantly correlated with drinking at any time period except weeks 8-9 (r = .30, p < .04). The IBT scales were not consistently correlated with drinking; the few scattered positive and negative significant correlations appeared to be random. "Drunk days" were defined as days during which the subject had 6 or more drinks (~I = 17.1 days per month, SD = 6.0). No significant correlations were found between the IBT and number of drunk days, Factor Analysis o f the 113T Scales In order to investigate the degree to which the various irrational beliefs may be reflecting common underlying concerns, in spite of the original
310
Rohsenow and Smith
factor-analytic basis of the scale's development, the factor structure of the scales was analyzed with a principal components factor analysis with orthogonal Varimax rotation. The correlations of each scale with the four resulting factors are presented in Table IV. Factor 1 consists of worrying about possible negative events combined with a strong need to be approved and to be competent. Factor 2 is most strongly determined by the belief that one cannot change the way one is, combined with blaming others for being bad and insisting that people must be competent. Factor 3 reflects a need to rely on others and blame and punish others for their failings along with an intolerance of frustration. Factor 4 involves preferring to avoid problems rather than to seek precise solutions for them. The relationship of the factor scores to the other measures was investigated in order to determine whether the correlations reflected these common underlying factors. The factor scores showed little correlation with the mood ratings but appeared instead to obscure the relationships of the moods to individual beliefs. The correlations of irrational belief factors to the other scales were more strongly related (see Table IV), although, in places, correlations with individual beliefs were obscured by using factor Table IV. Varimax Rotated Factors of the IBT: Correlations with IBT AND Other Scales Factors Scale IBT scale 1. Approval 2. Competence 3. Blame others 4. Frustration intolerance 5. External causes 6. Dwell on negatives 7. Problem avoidance 8. Rely on others 9, Doomed by past 10. Precise solutions % of variance Personality scale STAI General anxiety Text anxiety Social anxiety Aggression anxiety Locus of control Social introversion "p < .05, bp < .01. ~p < .002.
1 ,74 .50 ,08 .36 .22 .93 - .05 .10 .12 .02 51.7 .49 c .33 ~ .63 ~ .43 b .35 ~ .39" .22
2
3
4
.30 .40 ,50 ,19 .32 .00 -.11 .13 .92 -.02 19.8
.16 -.02 .41 .39 .t0 .10 -.21 ,90 .09 -.08 15.4
-.15 .28 -.09 .20 .07 -.15 .63 -.12 -.16 -.49 13.0
.41 b -.12 .13 .24 -.33 ~ ,05 .36"
-.06 -.15 .34 ~ -.08 .05 .04 -.09
.33" .53 c .11 .22 .17 .15 .30 ~
Irrational Beliefs and Mood
311
scores. The various anxiety scales and external locus of control were all significantly correlated with Factor 1, worrying about approval and competence. General anxiety was additionally associated with avoiding seeking solutions to problems. Trait anxiety was correlated with all factors except relying on and blaming others; that latter factor was associated with test anxiety specifically. Aggression anxiety was correlated both with the worry factor and with seeing oneself as able to change. Social introversion (MMPI's Si scale) was correlated with seeing people as unchangeable, needing approval and competence, and avoiding problems. DISCUSSION
The total number of irrational beliefs predicted subjects' daily selfratings of depressive mood up through 7 months later, and their daily anger ratings through 3 months later. Thus, the tendency to endorse irrational beliefs appears to have a consistent relationship with daily levels of dysphoria and anger. For depression self-ratings, the need to be perfectly competent and the belief" in external causes of unhappiness were most strongly related, although five other beliefs also showed a pattern of correlations in the predicted direction. The pattern suggests that overly strong expectations of self and others, combined with a sense that problems are not controllable, may increase the tendency to experience unhappiness. Largely the same beliefs are involved in daily levels of anger, but the belief that problems are caused by external events or people and that one's life is not changeable have the strongest influence on anger. While the frequency and intensity mood rating dimensions were highly correlated, the frequency scores were more frequently correlated with the IBT scales, suggesting that irrational beliefs may increase the likelihood of experiencing depression or anger, but not necessarily the intensity of the reactions when they occur. Irrational beliefs had relatively little effect on daily ratings of state anxiety, which is somewhat surprising in light of the strong relationship they had to some of the trait anxiety scales. The one belief that tended to be most consistently correlated with daily anxiety ratings was the tendency to dwell on possible negative events. The weak relationship between the daily ratings of general anxiety and irrational beliefs may reflect the possibility that irrational beliefs are more strongly related to certain specific forms of anxiety than to others; a general anxiety rating could obscure this relationship. As can be seen in the relationship of the anxiety scales to the IBT, test anxiety is strongly related to irrational beliefs, whereas the General anxiety scale was the least strongly related. The STAI scale was quite strongly related to the IBT, but as the STAI appears to be predominantly associated with social anxiety rather than anxiety in general, while
312
Rohsenow and Smith
the daily anxiety ratings are most strongly correlated with general anxiety rather than social or test anxiety (Rohsenow & Smith, 1981), the conclusion that the IBT relates most strongly to specific sources of anxiety still seems warranted. Test anxiety was most strongly related to the tendency to catastrophize when hopes are not satisfied and to dwell on possible negative events. The tendency to dwell on task-irrelevant thoughts has also been found in other research on test anxiety (Sarason, 1978). Other people's opinions are important to test-anxious students, as the need for approval and tendency to rely on stronger people shows. Not surprisingly, the need to be perfectly competent is also important in test-anxious subjects. These findings provide additional support for the view that maladaptive cognitive styles are important determinants of test anxiety and suggest some specific beliefs to explore in the treatment of test-anxious students. Social anxiety was most strongly related to the need for approval from others, as one would expect, as well as to the need to be perfectly competent and dwelling on possible negative events. These same beliefs were found to differentiate students with heterosexual anxiety in a previous study (Gormally et al., 1981), so this relationship is apparently reliable. Similarly, social introversion was related to the need to be competent and tended to be related to the need for approval, and was also related to avoiding problems and seeing oneself as unable to change. Anxiety about having expressed anger (aggression anxiety) was related only to worrying about approval and seeing oneself as able to change. However, this relationship may reflect an underlying need for social desirability. A factor analysis of the IBT scales revealed four underlying factors: worrying about approval and competence, seeing people as unchangeable and blamable, relying on and blaming others for frustrations, and avoiding seeking precise solutions to problems. Although these factors obscured the relationship of irrational beliefs to the daily mood ratings, the beliefs that correlated with the trait scales tended to lie along these factors. The one factor that underlay all of the trait anxiety measures was the tendency to worry about approval and competence. General trait anxiety was also associated with the factors of avoiding seeking solutions to problems and believing people cannot change. It is interesting that the MMPI's depression and anxiety scales were virtually unrelated to the irrational beliefs: It may be that the MMPI does not have much predictive power in the normal range. The IBT appears to be relatively unaffected by a social desirability response bias, except for scales 6 and 7. The virtual lack of relationship between locus of control and irrational beliefs was more surprising in that locus of control is usually correlated with anxiety and depression, and several of the beliefs would be specifically predicted to be associated with helplessness. The generalized
313
Irrational Beliefs and Mood
expectancy may not necessarily correspond with specific expectancies, as Rotter (1975) has noted. Thus, there appears to be fairly consistent support for Ellis's hypothesized relationship between certain of the irrational beliefs and maladaptive emotional reactions. Across the studies, two beliefs are most often associated with measures of depression, anxiety, and anger: Dwell on Negatives and Need for Competence. Two other beliefs are commonly associated with social and test anxiety and anger: Need for Approval and Doomed by Past. Two additional beliefs appear related to anger and show occasional correlations with other measures as well: Blame Others and External Causes of Unhappiness; and two more are occasionally related to measures of depression and anxiety: Frustration Intolerance and Problem Avoidance. However, one of the beliefs has received virtually no support as an irrational belief, as Rely on Strong Others has been essentially unrelated to any measure except test anxiety, and in previous studies has been negatively related to depression (Nelson, 1977) and emotional steadiness (Jones, 1968). However, this lack of relationship could easily be due to the low reliability of this one scale. One additional belief has consistently been found to act more as a "rational" belief instead: Scale 10, Need for Precise Solutions, is significantly associated with lower trait and social anxiety scores in this study, and with less depression and anxiety in the Nelson (1977) and Jones (1968) studies. Probably, a desire to search for precise solutions is a more effective coping strategy than a willingness to flounder in vague uncertainties. In conclusion, it appears so far that eight of Ellis's irrational beliefs in fact are associated with maladaptive emotional states and that four factors underlie the beliefs. In future research, it would be of interest to study those who have sought treatment for depression and anger, and to investigate these relationships in nonstudent populations as well.
REFERENCES Bandura, A. Principles of behavior modification. New York: Holt, Rinehart & Winston, 1969. Bandura, A. Social learning theory. Englewood Cliffs, New Jersey: Prentice-Hall, 1977. Beck, A. T. Cognitive therapy and the emotional disorders. New York: International Universities Press, 1976. Cahalan, D., Cisin, I. H., & Crossley, H. M. American drinking practices: A national study of drinking behavior and patterns. New Brunswick, New Jersey: Rutgers Center of Alcohol Studies, Monograph No. 6, 1969. Dollard, J., & Miller, N. E. Personality and psychotherapy. New York: McGraw-Hill, 1950. Ellis, A. Reason and emotion in psychotherapy. New York: Lyle Stuart, 1962. Ellis, A., & Grieger, R. (Eds.). Handbook of rational-emotive psychotherapy. New York: Springer, 1977. Goldfried, M. R., & Goldfried, A. P. Cognitive change methods. In F. H. Kanfer & A. P. Goldstein (Eds.), Helping people change: A textbook of methods. New York: Pergamon Press, 1980.
314
Rohsenow and Smith
Goldfried, M. R., & Sobocinski, D. Effect of irrational beliefs on emotional arousal. Journal of Consulting and Clinical Psychology, 1975, 43, 504-510. Gormally, J., Sipps, G., Raphael, R;, Edwin, D., & Varvit-Weld, D. The relationship between maladaptive cognitions and social anxiety. Journal of Consulting and Clinical Psychology, 1981,49, 300-301. Jones, R. G. A factored measure of Ellis' irrational belief systems with personality and maladjustment correlated. Wichita, Kansas: Test Systems, 1968. LaPointe, K. A., & Crandell, C. J. Relationship of irrational beliefs to depression. Cognitive Therapy and Research, 1980, 4, 247-250. Marlatt, G. A. Unpublished data, University of Washington, Seattle, no date. May, J. R., & Johnson, H. J. Physiological activity to internally elicited arousal and inhibitory thoughts. JournalofAbnormalPsychology, 1973, 82, 239-245. Nelson, R. E. Irrational beliefs in depression. Journal of Consulting and Clinical Psychology, 1977, 45, 1190-1191. Novaco, R. W. Stress inoculation: A cognitive therapy for anger and its application to a case of depression. Journal of Consulting and Clinical Psychology, 1977, 45, 600-608. Overall, J. E., & Gomez-Mont, F. MMPI-168 for psychiatric screening. Journal of Educational and PsychologicalMeasurement, 1974, 34, 315-319. Rimm, D. C., & Litvak, S. B. Self-verbalization and emotional arousal. Journal of Abnormal Psychology, 1969, 74, 181-187. Rohsenow, D. J., & Smith, R. E. Frequency and intensity dimensions of anxiety. Manuscript in preparation, 1981. Rotter, J. B. Generalized expectancies for internal versus external control of reinforcement. PsychologicalMonographs, 1966, 80(1, Whole No. 609). Rotter, J. B. Some problems and misconceptions related to the construct of internal versus external control of reinforcement. Journal of Consulting and Clinical Psychology, 1975, 43, 56-67. Russell, P. C., & Brandsma, J. M. A theoretical and empirical integration of the rationalemotive and classical conditioning theories. Journal of Consulting and Clinical Psychology, 1974, 42, 389-397. Sarason, I. G. The test anxiety scale: Concept and research. In C. D. Spielberger & I. G. Sarason (Eds.), Stress andanxiety (Vol. 5). New York: Wiley, 1978. Smith, R. E. Development of an integrated coping response through cognitive-affective stress management training. In I. G. Sarason & C. D. Spielberger (Eds.), Stress and anxiety (Vot. 7). Washington, D.C.: Hemisphere, 1980~. Spielberger, C. D., Gorsuch, R. L., & Lushene, R. E. Manual for the State-Trait Anxiety Inventory. Pato Alto: Consulting Psychologists Press, t970. Staats, A. W. Social behaviorism. Homewood, Illinois: Dorsey, 1975. Velten, E. A laboratory task for induction of mood states. Behaviour Research and Therapy, t968, 6, 473-482.
