IRRATIONAL BELIEFS IN THE ARTICULATED ... - CiteSeerX

Journal of Rational-Emotive Therapy, Vol. 5, No. 4, Winter 1987

IRRATIONAL BELIEFS IN THE ARTICULATED THOUGHTS OF COLLEGE STUDENTS WITH SOCIAL ANXIETY Gerald C. Davison and Vivien Zighelboim University of Southern California

ABSTRACT: Cognitive conceptualizations of social anxiety emphasize the role of negative self-statements, unrealistic expectations, and irrational beliefs in the development and maintenance of anxiety in social-evaluative situations. Research into these cognitive factors has entailed administration of questionnaires and instructions to subjects to write down their thoughts during a preceding or impending real-life encounter. These methodologies are criticized on several grounds, such as their assessment of abstract, generalized views by the subject of his/her typical way of thinking over a broad range of circumstances: constraints on responses because of experimenter-provided alternatives: and delays between the eliciting events and self-talk such that recall is subject to distortion and forgetting. Reported here is the use of a recently developed paradigm for uncovering thoughts in complex social situations. In the articulated thoughts during simulated situations method, subjects role-play participation in an audiotaped interpersonal encounter and, at predetermined points, verbalize thoughts elicited by a short segment of the fictitious event. The think-aloud data are taped for later content analysis. The purpose of this study was to explore the relationship between irrational beliefs and social anxiety. Results confirm those of two previous related experiments in that subjects articulated thoughts of greater irrationality when confronted with a stressful social-evaluative situation than with a neutral one. In addition, individuals with a tendency to become anxious in social situations articulated more irrational thoughts than did control subjects, confirming the basic assumption of cognitive-behavioral approaches that certain patterns of unrealistic thinking are associated with psychological distress. Experimental psychopathologists have relied on three theoretical models to explain the etiology and maintenance of social anxiety (CurII is a pleasure to acknowledge the collegial cooperation of Gloria Porter. Barbara Cadow. Anita Siegman, and the other staff of the USC Counseling Center for screening and referring clients as potential subjects. Thanks go also to Susan Ellison and Toni Wells for their skillful content analysis of the data. Valuable statistical consultation came from Rand Wilcox of the University of Southern California and Cora Lee Yale of UCLA. Marcia Johnson offered useful comments on the manuscript. Requests for reprints should he sent to Gerald C. Davison, Department of Psychology, Seeley G. Mudd Building, University of Southern California, Los Angeles, CA 90089 10151. •

1957 Ilunlun

11:n, (Sen

Press

2:38

GERALD C. DAVISON AND VIVIEN ZIGHELBOIM

239

ran, 1977). The first model conceives of social anxiety as resulting from classical conditioning episodes during which previously neutral social situations are associated with aversive stimuli and thereby become fearsome. The second model views anxiety in social-evaluative situations as resulting from a deficit of social skills or from an inappropriate behavioral repertoire. The present study examines social anxiety in the context of the third perspective. A cognitive conceptualization stresses the role of negative or maladaptive self-statements, unrealistic expectations, irrational beliefs, negative self-evaluations or faulty cognitive appraisal as determinants of anxiety in social-evaluative situations (Glass & Merluzzi, 1981). This conceptualization is part of a growing tendency in theory and research to focus on cognitive factors in psychotherapy and psychopathology (Bandura, 1977; Beck, 1967; Davison, 1966; Ellis, 1962; Goldfried & Davison, 1976; London, 1964; Mahoney, 1974; Meichenbaum, 1975; & Mischel, 1968). The main assumption of cognitive behavioral approaches to psychopathology is that the way a person thinks is a major factor in how s/he feels and behaves; a corollary is that alteration of thought will contribute to therapeutic change. For example, Ellis (e.g., Ellis, 1962, 1977; Ellis & Bernard, 1985) holds that certain beliefs cause maladaptive emotional reactions, and that changing them in certain ways can alleviate a broad range of psychological distress.

Cognitive Assessment of Social Anxiety In a review of cognitive assessment techniques, Merluzzi, Glass and Genest (1981) note that the Jones Irrational Beliefs Test (IBT) (Jones, 1969) is one of the most frequently used measures for research on the cognitive nature and treatment of social-evaluative anxiety. This questionnaire is designed to assess a person's degree of endorsement of various irrational beliefs within the theoretical framework of Albert Ellis. Newmark, Frerking, Cook, and Newmark (1973) conducted one of the first studies using the IBT. They administered this test to subjects diagnosed as neurotic, personality disordered, or normal, and found that many more of the neurotics than the other two groups endorsed irrational beliefs. Goldfried and Sobocinski (1975) found that women who held a strong belief in the importance of social approval on the IBT reacted with greater self-reported anxiety and hostility when they imagined themselves being rejected in a social situation than did women for whom this belief was less important. More recently, Gormally, Sips,

240

JOURNAL OF RATIONAL-EMOTIVE THERAPY

Raphael, Edwin, and Varvil-Weld (1981) compared competent daters to a clinical sample of anxious men and found that the latter endorsed irrational beliefs more strongly on four of the five IBT subscales. The IBT, moreover, has been used in cognitive therapy outcome research in the treatment of social anxiety. Although some researchers (Kanter & Goldfried, 1979) found that subjects who had cognitive interventions showed significantly lower IBT total and subscale scores relative to waiting list controls, others (Derry & Stone, 1979) have uncovered no posttreatment IBT differences. More research is thus needed to establish the validity of the IBT (Merluzzi et al., 1981). Critical limitations exist in the use of any questionnaire or inventory to assess irrational beliefs. For example, to answer questions on the IBT " such as "There is no such thing as an ideal set of circumstances, a subject must reflect on past experiences and then draw general conclusions that exemplify his or her characteristic way of thinking. The experimenter thereby assesses subjects' general or abstracted views, not what they may think in a particular situation. Another limitation of ' questionnaires is that they tend to restrict subjects thoughts to experimenter-defined categories of thinking or even to specific thoughts. As a result, the experimenter might miss some important thoughts by omitting categories from the questionnaire. Furthermore, and perhaps of greater importance, categories or exemplars can be leading or suggestive, and the experimenter may thus unintentionally cue certain thoughts in the subject. Because little is known at this time about cognition and psychiatric disorder, it seems preferable to construct a paradigm that places as few constraints as possible on the data the subject may provide. Several studies have allowed for subjects' open-ended responses. Sutton-Simon and Goldfried (1979) correlated IBT scores with subjects' positive and negative self-statements and with measures of social anxiety and fear of heights. The tendency to hold irrational beliefs was positively associated with social anxiety and tended to correlate with negative self-statements emitted in social situations. Craighead, Kimball, and Rehak (1977) found that subjects high on the importance of social approval scale of the IBT made more negative self-statements while visualizing social rejection scenes than did subjects low on the scale. Cacioppo, Glass, and Merluzzi (1979) told male subjects who were either high or low in social anxiety that they would soon be interacting with a woman in a conversation about college life. The experimenter then left the room, returning three minutes later to ask the subject to write down what he had been thinking during the waiting period. Both the subject and the experimenter then rated the thoughts as positive,

GERALD C. DAVISON AND VIVIEN ZIGIIELBO1M

241

negative or neutral. More negative self-statements and a more negative view of themselves and the impending conversation were emitted by subjects high on social anxiety than by less socially anxious subjects. Similarly, Halford and Foddy (1982) found that socially anxious individuals emitted more self-statements of disapproval by others in social interactions than did less anxious people. The paradigm used by Cacioppo, et al., (1979) appears to be a better vehicle for eliciting realistic answers than are the textual and typically hypothetical situations described on questionnaires. Nevertheless, arranging lengthy and complex social situations would be costly and often impractical. Moreover, whenever there is a noticeable interval be' tween the occurrence of thoughts and feelings and subjects report of them, there is the risk that important reactions will be forgotten or distorted during recall. Finally, studies such as Cacioppo, et al., require subjects to write down their thoughts and feelings. This approach has serious drawbacks. Not only do written reports take longer than verbal, but it is generally more difficult to write about one's thoughts than to talk about them, especially since speaking is probably closer to thinking. There is a good chance that subjects will attempt to express more complicated or subtle thoughts with verbal reports than with written reports; and thinking aloud probably leads to more nearly complete reporting than writing. (For reviews of historical and contemporary think-aloud research, see inter alia Ericsson & Simon, 1981; Meichenbaum & Cameron, 1981; and Kendall & Hollon, 1981.)

Articulated Thoughts During Simulated Situations Paradigm An improved paradigm for examining thoughts in a variety of situations of relevance to cognitive behavioral workers seems to require a number of features. It should permit open-ended verbal responding that reflects, as much as possible, ongoing thought processes rather than retrospective reporting. Subjects should be constrained as little as possible in what they report, and they should be enabled to make their reports with ease and a minimum of delay. The experimenter should also be able to specify and manipulate the situations to which subjects are reacting while at the same time being able to present events that are sufficiently realistic and complex. Situations unlikely to be bothersome to subjects should be presented along with troubling situations; hitherto, direct comparisons have not often been made. Finally, the procedure should not be prohibitively expensive in time or money. The study reported here utilizes a paradigm that provides virtually

242


"on-line" assessment of cognition during complex experimenter-controlled interpersonal situations. Called articulated thoughts during simulated situations (ATSS), the procedure has a subject pretend that he/she is a participant in a role-played interaction, such as being the object of someone ' s critical remarks. At predetermined points in the audiotaped scene, there is a pause, during which the subject is instructed to verbalize whatever is on his/her mind in reaction to what has occurred just seconds earlier. Subject ' s tape-recorded verbalizations are later content-analyzed and compared both to his/her articulated thoughts to contrasting stimulus tapes and to thoughts collected from subjects differing on personality dimensions of interest to the experimenter. In two previous experiments from our laboratory (Davison, Robins, & Johnson, 1983; Davison, Feldman, & Osborn, 1984), it was demonstrated that college undergraduates found the procedure credible, that articulated thoughts could be reliably content: analyzed, and that the technique pulled from subjects verbalized congitions that varied meaningfully with situations to which they were exposed (to wit, subjects thought more irrationally in stressful situations than in neutral ones). Unsuccessful in both experiments were attempts to demonstrate relationships between verbalized thoughts and several personality inventories, most particularly the Fear of Negative Evaluation Scale (Watson & Friend, 1969) and the IBT. These relationships were sought in light of predictions from the contemporary cognitive behavior theory of Ellis that people who are sensitive to criticism have a tendency to think in demanding, absolutistic ways and will manifest catastrophizing and "irrational" thinking when criticized by others. Both experiments used subject analogues, that is, unselected undergraduates who scored at the extremes of the FNE and the IBT. Overlooking for the sake of argument the likelihood of significant error variance in the self-report inventories, a sample of therapy clients with social anxiety was employed in this study for a better test of the basic hypothesis that people with social anxiety generate more irrational thoughts than controls when exposed to social criticism situations. METHOD

Subjects Two groups of college students participated in this study. The clinical group consisted of 8 undergraduates (4 men and 4 women) in therapy at the USC


243

Counseling Center. They had been screened by their therapists, who used the following criteria in referring them for the experiment:

244


Procedure Subjects were told that they would be participating in an experiment on The things that people say to themselves, " and that the interests of the experi menter were the thoughts and feelings people have in a variety of situations. The experiment took place in a sound-proofed laboratory containing a comfortable armchair with ultrasensitive stereo microphones on either side to record subjects' comments, a pair of high-fidelity stereo speakers for presenting the recorded situations, and a desk for completing questionnaires. The experimenter sat in a separate control room that contained a reel-to-reel tape recorder for playing the stimulus tapes and a precision cassette deck for recording subjects ' responses. A prerecorded cassette tape instructed subjects to listen to the tapes and to i magine as clearly as they could that the event was actually occurring. They were presented with a description of the situational context for the control tape, asked to tune into their thoughts and feelings and to say those thoughts aloud each time the tape stopped. Then the stress tape was presented, and subjects' responses to this tape recorded. Subjects were exhorted to be as frank and complete in their comments as possible. After articulating their thoughts to both tapes, each subject completed the FNE and IBT. Watson and Friend (1969) have presented evidence for the acceptable reliability and validity of the FNE. The IBT has been used frequently in studies to measure cognitive content (Merluzzi, et al., 1981; Woods, 1984; Woods & Coggin, 1985). Jones (1969) has reported acceptable internal validity for this scale. Merluzzi, et al., however, have uncovered conflicting results in the literature and have suggested that there is a need for further validation of the IBT. "

1. The person experienced anxiety in social situations, was shy, inhibited—in general overly sensitive to social evaluation or criticism. 2. The person was not psychotic or borderline. 3. The person spoke English reasonably well (a consideration only for foreign students). Clients who volunteered were paid $6 for participation. The control group consisted of 15 introductory psychology undergraduates (8 women and 7 men) who participated in this study to receive extra credit toward course requirements.

Simulated Situations Students from the University of Southern California Drama department were hired to play the roles of speakers in two taped simulated conditions; these tapes had been used previously in the Davison, et al., (1984) experiment. Each was composed of seven spoken segments of 15–25 seconds duration with a 30 second blank interval interspersed between each segment to permit subjects to articulate their thoughts.

Overheard Conversation Tape (Stress Condition). On this tape, subjects were asked to imagine overhearing two acquaintances (a man and a woman) talk about them and the impression they had made at a recent social function. The speakers were not aware of the subject's presence. Six of the seven segments contained statements designed to be critical of the individual subject's appearance, attitudes and behavior, such as "He (she) just blurts out the first thing that comes to his (her) mind," and ... he (she) looked so out of place," and "You have to be blind to the facts of the universe to talk like that." Segment 4 was in defense of the student, acknowledging that we all "get pretty argumentative when it comes to certain things." This noncritical segment was included to increase the credibility of the situation. Separate tapes were used for male and female subjects, differing only in the gender of the pronouns used.

Control Tape (Neutral Condition). On this tape, the subject was asked to imag-

ine sitting in the school cafeteria, listening in on two strangers (a man and a woman) discuss an unindentified class they are taking and the professor who teaches it; both the class and the professor are unfamiliar to the subject. Segments 1, 2, 3, and 5 contained statements that were critical of the class and/or ' the teacher, such as "He doesn t seem to care very much about what he is teaching," whereas segments 6 and 7 were complimentary, containing statements such as: "He is well respected in his field." Segment 4 was somewhat in defense of the professor.

RESULTS

Articulated Thoughts The content analysis employed in the present experiment was virtually identical to that used by Davison, Feldman, and Osborn (1984). Each subject provided 14 thirty-second response segments, 7 to the Stress tape and 7 to the Control. All segments were independently scored directly from tape by two raters, undergraduates carefully trained to detect the presence of each of Ellis's (1962) 10 Irrational Beliefs on a scale from 0 (not present) to 6 (extremely irrational). During three training sessions, raters learned to content analyze subjects' articulated thoughts. Raters were blind to subjects' scores on the FNE and IBT measures as well as to whether subjects were controls or therapy clients. To familiarize them with the ATSS procedures, raters were themselves run through the experiment (exposed only to the instructions and tapes, but not asked to fill out questionnaires). The data thus


245

collected were used as material for training and then coded conjointly by the raters and experimenter during the training sessions. Following these sessions, the two raters proceeded to code the subjects' data. Each rater worked separately on the same material. Raters coded all response segments of each subject before proceeding on to the next subject. ' Raters looked for implied endorsement of Ellis s irrational beliefs in ' articulated thoughts. A Sum Irrational Beliefs (SIB) score was subjects then computed for each subject on each stimulus tape by adding together the 10 Irrational Beliefs scores for all 7 tape segments. Therefore, every subject received from each rater a SIB score for the Control tape and another one for the Stress tape. In this fashion we hoped to capture the essence of a subject's irrationality as defined by Ellis. Because assessing the degree of irrationality of subjects' responses requires a degree of inference, reliability of scoring is important. Correlations were computed for the Sum Irrational Beliefs scores assigned to the Stress tape and to the Control tape by the two raters. rX ~ for the Stress tape was + .76, and for the Control, + .96. Both Pearson coefficients are highly significant (p < .001). Table 1 contains the means and standard deviations of the sums of 10 Irrational Beliefs scores (:SIB) for therapy clients and control subjects as they listened to neutral and stressful tapes, separately by each rater as well as by their average. A 2 x 2 ANOVA with a repeated measures within-subjects factor of Condition (control vs. stress tape) and a between-subjects factor of Group (therapy clients vs. control subjects) performed on the SIB scores showed significant main effects for the withinsubjects (p < .001) and between-subjects factor (p < .01) for the average of Rater 1 and Rater 2, see Table 2. These findings hold for Rater 1 and Rater 2 separately as well. No significant interaction effects were found. Thus, therapy clients with social anxiety articulated overall more irrational thoughts than did controls, an encouraging finding because it represents our first demonstration of a significant between-group difference in an ATSS experiment. Secondly, subjects overall articulated more irrational thoughts to the stress than to the control tape, a finding that replicates our two previous studies. It should be noted that although therapy clients were drawn from a highly restricted population (therapy clients who met the threefold criteria mentioned above), it can be assumed (due to the central limit theorem) that their SIB summary scores approach normality, since these scores were based on the sums of 10 Irrational Beliefs scores for all 7

246


Table 1 Means and Standard Deviations of Sum Irrational Beliefs Scores (SIB)

Average of Raters Simulated Situation Controls

Neutral 14.99 (SD = 12.22)

Stress 39.63 (SD = 11.901

27.0 (SD = 11.92)

46.37 (SD = 10.43)

Subjects Therapy Clients

Rater 1 Simulated Situation Controls

Neutral 16.47 (SD 12.95)

Stress 46.6 (SD = 14.18)

27.5 (SD = 12.39)

(SD


–

50.0 11.76)

Rater 2 Simulated Situation Controls

Neutral 13.53 (SD = 11.61

Stress 32.66 (SD = 10.49)

26.5 (SD = 11.95)

42.75 (SD = 11.27)


segments. The use of ANOVA to analyze these scores is thus appropriate.

Personality Measures Table 3 contains the means and standard deviations of IBT and FNE scores for therapy clients and control subjects. Student t-tests to assess the significance of the difference between the means of the scores of the


248

247


Table 3

Table 2 j

Repeated Measures ANOVA Performed on Sum Irrational Beliefs Scores

Sources of Variation Average ofRater 1 and Rater 2

Rater 1

df

MS

F

Means and Standard Deviations of IBT and FNE Scores

Probability Controls

Group (Therapy clients vs Control Subjects) Condition (Neutral vs Stress tapes) Group by Condition

1

Group Condition Group by Condition

1

Group Condition Group by Condition

1

Rater 2

1

1 1 1

1

916.3

8.41

5052.34 29.96

72.13

0.43