Validity of self-reported fear

Validity of self-reported fear

VALIDITY OF SELF-REPORTED FEAR* RICHARD 1. LANYON and MARTIN MANOSEVITZ Dept. of Psychology, Rutgers University, New Brunswick, N.J.. U.S.A. (Recei...

412KB Sizes 26 Downloads 59 Views

VALIDITY

OF SELF-REPORTED

FEAR*

RICHARD 1. LANYON and MARTIN MANOSEVITZ Dept. of Psychology, Rutgers University, New Brunswick, N.J.. U.S.A. (Received

6 May 1966)

Summary-This study investigated the relationship between a self-report fear measure and several measures of fear made in the fear-arousing situation. Subjects at three levels of self-reported fear of spiders (mild, moderate and intense) were found to differ on most of the situational measures from subjects who had professed no spider fear. Subjects who initially reported different levels of spider fear did not differ among themselves on the situational measures. It was concluded that self-report paper and pencil measures of fear can be valid, though gross, indicators of fear behavior in an actual situation.

INTRODUCTION IN PERSONALITY assessment, self-report inventories and questionnaires have long been used to obtain information about a subject’s behavior, feelings, and thoughts. One criticism of such techniques is that the relationship between test scores and the criterion is usually far from perfect, while in some cases the nature of the relationship is completely unknown. For example, Azrin et al. (1961) studied self-reported fear symptoms during aerial combat, using college student subjects under two sets of instructions. They concluded that response bias accounted for most of the variability in the self reports, and that “. . . questionnaire responses may be independent of the actual behavior being studied” (p. 326). One implication of this finding is that when constructing self-report measures, independent behavioral validation should be made whenever possible. Only in this way can the relationship between self-report instruments and actual behavior be determined. An example may be seen in the work on the Taylor Manifest Anxiety Scale (MAS) (Taylor, 1953). Investigations of the relationship of the MAS with independent ratings of anxiety made by clinicians who had observed or interviewed the subjects have been summarized by Taylor (1956). Correlations between MAS scores and ratings ranged between .28 and .60, according to the population being studied, the rating procedure used, and the type of raters. A recent method for assessing specific fears has used a self-report Fear Survey Schedule (FSS) (Geer, 1965; Lang and Lazovik, 1963; Manosevitz and Lanyon, 1965; Wolpe and Lang, 1964). The schedule has been used, in its different forms, to screen and select subjects for participation in experimental studies of behavior therapy. In addition, changes in FSS scores have been used to assess pre- to post-therapy changes in the level of specific fears and also of general anxiety. However, in only one reported study (Geer, in press) has attention been focused directly on the relationship of FSS scores and other self-reports to independent measures of behavior. Geer used subjects who were either high in specific fear of spiders and low in snake fear, or who were low in both spider and snake fears. Subjects were shown slides of irrelevant stimuli, followed by either spiders or snakes.

* This study was supported in part by USPHS Grant MH 12313 from the National Institute of Mental Health. 259

260

RICHARD

I. LANYON

and MARTIN

MANOSEVITZ

Measures of autonomic arousal (GSR), taken after each slide, showed that the greatest autonomic arousal response was given by the subjects who were afraid of spiders, to pictures of spiders. Geer suggested that the subjects responded to the feared stimuli with increased emotional arousal. This study investigated the relationship between subjects’ predictions of their fear in a fear-arousing situation and several measures of fear taken in the situation itself.

METHOD Sut@ls and self-reported fears College females were asked to write a list of their irrational fears; i.e. objects and events which were fear arousing for them even though no objective danger existed. Each subject was asked, in addition, to rate each fear as mild, moderate, or intense. Lists of fears from 516 girls were obtained in introductory psychology classes and through the campus mail. After examining the responses, one particular fear, that of spiders, was chosen for investigation, and subjects who reported fear of spiders were asked to participate in the study. From a totaf of fifteen mild, twenty-eight moderate, and twenty-six intense spider fears reported, the number of subjects who actually took part in the experiment from each group was seven, fifteen, and fourteen respectively. In addition, ten subjects who reported a similar number of fears, but did not report spider fears, were included as a control group. Thus, forty-six subjects representing four different levels of self-reported fear of spiders participated in the experiment. Situat~o~~~fear measures Four scores or ratings representing each subject’s level of specific fear were obLained in the experimental situation, and in addition, each subject made a judgment of her fear at that time. Behavior sample test (BST). Four small jars, each containing a live spider, were placed on a table immediately beneath a one-way vision window in a large room in the Rutgers Psychological Clinic. A set of nine instructions was written to direct each subject through a series of tasks which brought her progressively closer to the spiders. The first instruction was “Go over to the table”, and the ninth was “Hold all four jars at once”. All subjects were told they could terminate the experiment whenever they wished. Each was given a BST score from l-9, corresponding to the last task she was able to perform. Each instruction was repeated twice and the test was terminated by the experimenter if any step was not The experimenter did not know the performed after repetition of a particular instruction. self-reported fear level of any subject. Task-oriented check list (TC). A check list was constructed of task-oriented behavior which presumably indicated fear. The list, similar to that used by Cooke (1965), included such items as “At first mention of spiders makes negative comment”, and “Needs repetition of instructions”. Two third-year clinical psychology graduate students observed from behind the one-way vision window and independently assessed each subject on the TC. The observers, who could hear the conversation in the experimental room through earphones, did not know the self-reported fear level of any subject. Each subject’s score on the TC was the total number of items checked, summed across the two observers. General check list (GC). A check list of general fear symptoms, based on Cooke’s (1965) revision of Paul’s (1966) check list, was also checked independently by the two

VALIDITY OF SELF-REPORTED FEAR

261

observers. This list included such items as “Covers mouth with hand”, and “Hand trembles”. Again, each subject’s score was the total number of items checked, summed across the two observers. IntervielY (IV). Each subject was rated for degree of fear evident during a post-experimental interview. Procedure

Each subject was ushered into the experimental room by the graduate student who had done the scheduling. The BST was administered, and immediately following it, the experimenter gave each subject a copy of Walk’s (1956) fear thermometer (FT), on which she was instructed to judge her level of felt anxiety. The FT ranged from “Completely relaxed” (l), to “Scared as I’ve ever been” (10). Each subject was then interviewed in another room by a second experimenter, who gave her an opportunity to express her feelings about the procedure and explained the nature of the research. The interviewer gave each subject a rating from 1-9, as indicated above, according to his assessment of her fear reaction. Interview (IV) ratings for the control or “no-fear” subjects were subsequently discarded because several of these subjects spontaneously indicated during the interview that they had not listed spiders on the initial survey. RESULTS Observer reliability

The Pearson product-moment correlation coefficient between the observers for the of TC items checked was .84, and for the GC it was .78.

number

Relationship between self-reported fear level and situational fear measures

The mean scores on each of the five situational measures of spider fear are shown in Table 1 for subjects from the four levels of self-reported fear. For three of these measuresthe TC, the GC, and the FT-separate simple randomized analyses of variance were performed. Results of these analyses are summarized in Table 2. All three F ratios were significant, and comparisons between individual means showed in each case that the significance was due to a difference between no-fear subjects and subjects with self-reported spider fear. TABLE 1. MEANSITUATIONALFEARSCORESFORSUBJECTSAT

FOUR LEVELSOF

SELF-REPORTED SPIDER FEAR Level of

Situational measure of spider fear

self-reported fear

No fear (control : .N=lO)

Mild

Moderate

(N=7)

(N= 15)

(N=14)

BST*

9.00

8.43

8.73

7.29

TC

2.80

8.29

10.27

9.07

CC

2.60

8.42

7.27

7.57

5.71

4.20

4.92

3.86

3.93

4.85

IV FT

1.70

* For the BST, a larger score indicates a smaller degree of fear.

Intense

262

RICHARD

and MARTIN

I. LANYON

MANOSEVITZ

TABLE 2. SUMMARIESOF THREESEPARATE ANALYSESOF VARIANCE AMONG FOUR LEVELSOFSELF-REPORTEDSPIDERFEAR

Measure

d.f.

F

P

TC

3142

3.48

<.025

GC

3142

3.93

<.025

FC

3142

5.95

1.005

An analysis of variance was not performed on the BST scores because the majority of subjects achieved the full score of 9. Instead a count was made, in each of the four groups, of the number of subjects scoring 9 and of the number who did not score 9. Fisher’s exact probability test showed that significantly more no-fear subjects achieved the full score of 9 than did subjects who originally reported intense spider fear (P=.O47). A similar comparison of no-fear subjects with the three self-reported fear groups combined gave a trend in the same direction (P=. 12). It should be noted that these are not independent probabilities. Intercorrelations were computed among the situational measures of fear, for all forty-six subjects and also for the thirty-six fear subjects separately. These correlations are reported in Table 3. With the exception of the IV/GC (. 10 > P> .05) and IV/TC (P< .05) correlations, all are significant beyond the .Ol level. TABLE 3. INTERCORRELATIONSAMONG SITUATIONALFEAR MEASURES FOR SPIDER FEAR SUBJECTS (N=36) AND FOR ALL SUBJECTS (N=46)* GC TC

.49f

IV .30:

t.61t) GC

FT .48t (.W)

.26

.sot

(.WJ .69t

IV * Intercorrelations 7 P<.Ol. : P< .05.

which include all subjects (N=46)

are given in parentheses.

DISCUSSION Consistent differences on measures and ratings made in the criterion situation were found between subjects who originally reported fear of spiders and those who did not. However, subjects originally reporting different levels of spider fear did not differ among themselves on any of the situational measures (Table 1). One interpretation of these findings could be that while gross discriminations are possible between girls who fear spiders and those who do not, finer discriminations within the fearful subjects cannot be made. If this interpretation is correct, the highly significant correlations among TC, GC, and FT ratings (N=46) would be due largely to the gross difference between fearful and nonfearful girls. However, when the same correlations were computed for the thirty-six fearful

VALIDITY

OF SELF-REPORTED

FEAR

263

girls only, they were still significant. Such findings suggest that it is possible reliably to assess degree of actual spider fear in subjects who profess this fear, and that such assessments tend to agree with the subjects’ own judgments made in the feared situation. While the BST elicited reliable signs and feelings of anxiety, it was not sufficiently extensive to be evaluated as an instrument in its own right. It is possible that most subjects yielded to situational demands in carrying out instructions which they would not normally have done. However, the data indicate that a set of tasks arousing a greater degree of fear might show distinct differences among all four groups. The fear ratings made by the interviewer (IV) were substantially related to the subjects’ own judgments in the feared situation (r=.69) and only minimally to the observers’ ratings (IV/TC r=.30; IV/GC r=.26). These findings would seem to suggest that the interviewer was largely rating the subjects’ own estimates of their reactions, which they had already reported on the FT. This study demonstrated that subjects can be assessed on gross levels of fear by means of paper and pencil self-report instruments in the absence of the feared stimuli. However, subjects who fear spiders can make a more valid estimate of their fear when they are placed in the fear-arousing situation. The difference between the present study and that of Azrin et al. (1961), who concluded that there was little if any relationship between experienced and reported fear symptoms, might be accounted for in terms of the degree of social desirability of the feelings and behaviors to be reported. While it is perhaps a little socially undesirable for a girl to fear spiders, it is considerably more undesirable for a combat soldier to report the loss of physical and psychological control reflected in the symptom list of Azrin et al. Thus, it is possible that the self-report method of assessing degree of fear is appropriate when the social desirability of the behavior or feelings to be reported is not an important factor. Acknowledgements-The authors are grateful to MERRITHHOCKMEYER, RUTH IMBER,and JAMESSMITHfor their assistance in the study.

REFERENCES AZRIN N. H., HOLZ W. and GOLDIAMOND 1. (1961) Response bias in questionnaire reports. J. consult. Psychol. 25,324-326. CNKE G. (1965) The Eficacy of Two Behavior Therapy Procedures. Unpublished masters thesis, University of Iowa. GEERJ. (1965) The development of a scale to measure fear. Behav. Res. & TheruDy _. 3,45-53. GEERJ. Fear-and autonot& arousal. J. ubnorm. Psychol. In press. LANQP. J. and LAZOVIKA. D. (1963) desensitization of a ohobia. J. &norm. sot. Psycho/. . , Exuerimental 66,519-525. MANOSEVITZ M. and LANYONR. I. (1965) Fear survey schedule: a normative study. Psychol. Rep. 17, 699-703. PAUL G. (1966) Insight Versus Desensitization in Psychotherapy: An Experiment in Anxiety Reduction. Stanford University Press, Stanford. In press. TAYLORJANETA. (1953) A personality scale of manifest anxiety. J. &norm. sot. Psychol. 48,285-290. TAYLORJANETA. (1956) Drive theory and manifest anxiety. Psychol. Bull. 53, 303-320. WALK R. D. (1956) Self ratings of fear in a fear-invoking situation. J. &norm. sot. Psycho/. 52, 171-178. WOLPEJ. and LANGP. J. (1964) A fear survey schedule for use in behavior therapy. Behav. Res. & Therapy 2, 27-30.