Anda di halaman 1dari 9

Journal ol Consulting and Clinical Psychology 1976, Vol. 44, No.

1, 83-91

General Versus Specific Trait Anxiety Measures in the Prediction of Fear of Snakes, Heights, and Darkness
Martin Mellstrom, Jr., George A. Cicala, and Marvin Zuckerman University of Delaware
The relations between general and specific trait anxiety tests and fear measures in three actual situations were investigated. Both types of test were administered to 76 undergraduate females early in the semester. Later, each subject was exposed to each situation, where observer's ratings, behavioral, and subjective fear measures were obtained. The results indicate that the specific tests were clearly superior to the general ones in predicting fear of snakes but only slightly superior in predicting fear of heights and darkness. It was concluded that the overall superiority of the specific measures supported the current trend toward situation specificity in personality assessment. The roles played by threat to self-esteem, the form of the general trail anxiety measure, and sensation seeking are also discussed.

Most attempts to measure anxiety have begun with the assumption that it is a trait, that is, an enduring disposition of a person to act in a reliable manner in a wide variety of situations (Allport, 1937). For example, Taylor's (1953) Manifest Anxiety Scale (TMAS) was intended to measure a person's characteristic level of anxiety, or drive, which was thought to be relatively constant over time and across situations. Recently, the trait concept of anxiety has been questioned. Mischel (1968) has argued that trait tests lack predictive validity because they do not take into account the situational specificity of behavior. Hodges and Spielberger (1966, 1969), Houston and Hodges (1970), and Katkin (1966) have found trait anxiety (A-Trait) to be unrelated to autonomic response and performance measures in stressful situations. Spielberger, Gorsuch, and Lushene (1970) noted that his A-Trait scale (State-Trait Anxiety Inventory; STAI) predicted state anxiety (AThis report is based on a thesis submitted by the first author in partial fulfillment of the requirements for the MA degree from the University of Delaware. The authors would like to thank the following students for their work as experimenters in the present study: Elmer Bouman, Chip Carpenter, Richard Dark, Scott Eiler, Jeff Gibson, Terry Herbein, John Jones, and George Laskaris. Requests for reprints should be sent to Marvin Zuckerman, Department of Psychology, University of Delaware, Newark, Delaware 19711.

State) in situations in which there was some threat to self-esteem but did not predict in situations in which there was threat of pain or physical harm. Some researchers have done studies in which specific trait items, referring to the feared object or situation, were compared with general A-Trait scales in the prediction of responses to specific fear situations. Hodges and Spielberger (1966) found that a single fear of shock item correlated .50 with changes in heart rate made in response to threat of shock, but there was no relationship between the TMAS and the heart rate measure. Mellstrom, Zuckerman, and Cicala (1974) found that items from a new specific A-Trait test, the Zuckerman Inventory of Personal Reactions (ZIPERS), and a questionnaire using snake items predicted fear reactions to the exposure of a snake, whereas general A-Trait measures did not. Similar results have been found for public speaking (Lamb, 1973) and test anxiety (Sarason, 1961) situations. Although the previous studies all seem to suggest that specific A-Trait measures are better predictors of situational responses than general ones, a finding by Zuckerman (in press) raised doubts about this important generalization. Using the ZIPERS he found that general and specific trait measures were equally predictive of A-State responses in an actual classroom examination situation. The
83

84

M. MELLSTROM, JR., G. A. CICALA, AND M. ZUCKERMAN

implication of this finding is that reactions to some anxiety-provoking situations may be equally predictable from general and specific tests. If general trait measures have predictive validity for certain situations, it is of theoretical and practical importance to identify the situations and discover how they differ from other situations. The current movement toward situation specificity in personality assessment (viz., Endler, Hunt, & Rosenstein, 1962; Mischel, 1968; Zuckerman, in press) is critically dependent on the answer to this question. The purpose of the present study was to determine the differential predictive power of general and specific A-Trait measures, with the addition of two types of general A-Trait measures. The two types, distinguished by Endler and Shedletsky (1973), are (a) those that rarely specify stimulus situations, such as the TMAS and STAI A-Trait scales, and (b) those that do specify situations, with a person's total score across all situations representing a general A-Trait measure. Most criticisms of "general A-Trait" scales have been directed toward the first type of measure, and the comparative validity of the second type, relative to the first and to specific A-Trait measures, has not been adequately established. Fear situations involving snakes, heights, and darkness were used because they involve fairly common fears that are amenable to behavior measurement techniques. A finding by Segal (1973) suggests that sensation seeking might be related to anxiety in these situations. The Thrill and Adventure Seeking subscale of the Sensation Seeking Scale (SSS; Zuckerman, 1971, 1974) meaures the approach tendency as opposed to the avoidance tendency in situations involving moderate danger. Although the scale does not correlate with general trait anxiety scales, Segal (1973) showed that it does correlate with responses to hypothetical situations in Endler et al.'s (1962) S-R Inventory of Anxiousness, which contains scale items with elements of physical threat from inanimate sources. Since the SSS was shown to correlate with responses to hypothetical situations, it was expected that it would predict re-

sponses to real situations that contained some elements of danger. METHOD Subjects
Seventy-six female students enrolled in introductory psychology at the University of Delaware in the fall of 1973 were subjects. They participated in the study in order to fulfill a course requirement.

Predictor Tests
Three of the pretests have been regarded traditionally as measures of general trait anxiety. The STAI A-Trait Scale (Spielberger, Gorsuch, & Lushene, 1970) was originally intended as a research tool for selecting subjects who vary in their disposition to respond to psychological stress with different levels of state anxiety. The TMAS (Taylor, 1953) was designed to measure manifest anxiety as defined by Cameron's (1947) description of chronic anxiety reaction. It has been the most widely used measure of A-Trait. The Neuroticism scale of the Eyscnck and Eysenck (1964) Personality Inventory (EPI) was used as the third measure of general A-Trait. Prior studies have found very high correlations between the TMAS and the EPI Neuroticism scale, and both tests may alternately be considered as measures of "anxiety" or "ncuroticism." Two pretests provided indices of both general and specific trait anxiety. The Geer Fear Survey Schedule (FSS) consists of self-rated fear reactions to 51 commonly feared objects and situations and has been shown to be related to behavioral and selfreport measures of fear (Geer, 1965). A subject's total fear score across all 51 items might be taken as a general Tearfulness or A-Trait measure; her score on any one particular item can be used as a specific A-Trait measure. Another test that provides measures of both general and specific A-Trait is a modified form of the ZIPERS. It consists of 12 positive- and negative-affect-arousing situations in which subjects indicate, on a 5-point scale, the degree to which each situation elicits in her each of 13 reactions. Factor analysis of the response dimensions of the scale has yielded several factors including the fear arousal factor used in the present research. By summing a subject's scores across all situations on the responses comprising the fear factor, a measure of general A-Trait may be derived. A subject's score on those reactions to 1 of the 12 situations yields a specific A-Trait measure. The test was modified by substituting a "heights" and a "darkness" situation for the two "success-acceptance" situations in the usual form. The test already contained a "snake" situation. To provide additional specific trait tests, the first author devised three fear questionnaires, one for each fear used in the study. The snake questionnaire asks the subject to indicate, on a scale from 1 to 8, the degree to which he/she would, or would not, like to be in 27 situations involving snakes. This scale

GENERAL VERSUS SPECIFIC TRAIT ANXIETY MEASURES was used in an earlier study (Mellstrom et al., 1974) and appears to be a valid index of snake fear. Additional fear of heights and fear of darkness questionnaires were constructed to be similar to the fear of snakes questionnaire. The total questionnaire is called the Situation-Specific Questionnaire (SSQ). The Thrill and Adventure Seeking subscale of the Zuckerman (1971) SSS was described in the introduction. It is a forced-choice scale. A sample item is as follows: 43. A. I would like to try parachute jumping. B. I would never want to try jumping out of an airplane, with or without a parachute. ratings, and behavioral measures of fear were obtained. Subjects were selected randomly, with the restriction that they not reside in either of the two 16-story dormitories where the height situation was located. It was felt that living there might affect the subjects' behavior in that situation. In addition, subjects were randomly assigned to the six possible temporal sequences of fear situations to determine if sequence, a "nuisance variable" (Kirk 1968), had any effect on fear in the situations. To minimize such carry-over effects from one situation to the next, subjects were scheduled for no more than one situation on a given day. Snake situation procedure. When the subject arrived, with the snake in view, she was seated and given her instructions by tape recorder. They instructed her to "perform the task" of approaching the snake and lifting it up into the air. As in the height and darkness situations, she was told that if she could not perform the task it was all right. After the instructions, the subject was administered the STAI A-State and ZIPERS State scales. After the state tests were completed, the behavioral fear test was begun. The experimenter told the subject to stand on a particular spot that served as a standard starting point for all subjects and then said, "You may now begin your task." The time it took the subject to touch the snake, if she could, was recorded by the experimenter as latency. If she could not touch the snake, she was given the maximum latency (300 sec). The experimenter rated the subject's behavior on a 1-12 rating form describing a continuum of behavior from 12 ("cannot approach snake closer than 6 feet") to 1 ("picks up snake and holds for 30 sec"). The subject's score on this task was called the task score. The experimenter also recorded the snake's behavior, another nuisance variable, during the subject's approach and then gave the subject a fear thermometer to fill out as she stood at the location of her closest approach to the snake. After the subject departed, the experimenter rated her on a 1-7 scale of anxiousness, which was called the observer's rating. Height situation procedure. The taped instructions told the subject that her "task" was to step out onto the landing, record the letters that were on the ground on a piece of paper, till out the questionnaire (the fear thermometer), and return into the building. The subject was then administered the STAI A-State and ZIPERS State scales. After completion of the state scales, the behavioral heights test was begun. The subject started from a standard location. The latency was the lime it took for her to reach another standard location and begin looking down from the landing. The distance of her toes, in IS-cm intervals, from the edge of the landing comprised her task score. The total amount of time spent by the subject on the landing was recorded as total time. After the subject left, the experimenter made the 1-7 observer's rating and verified that she had correctly recorded the letters that were on the ground.

Fear Situations
In the snake situation, a 1-m long Ball Python was housed in a plywood box with a sliding glass front. The height situation was on the 16th story of a dormitory at the university. On each floor, a 2 m X 1.5 m landing connected the main building to a fire tower. These landings were fenced in, except for 8-c gaps where the landing joins the building and tower, and were impossible to fall from. Situated on the ground below were four 3S X 27 cm poster boards on which the letters L, V, T, and Z were painted. In the darkness situation, the 3 X 4 m room was windowless and became totally dark when the lights were turned off. A small wooden box with a toggle switch was constructed so that pushing the switch turned off the lights and simultaneously activated a timer located in an adjoining room, where the experimenter was present.

Subjective Situational Fear Measures


To measure subjective fear in each situation, three state measures were used. Behavioral measures used are described in the Procedure section. 1. The A-Statc scale of Spielberger ct al.'s (1970) STAI was devised to measure the instantaneous anxiety level of persons at any given time. It is assumed that anxiety fluctuates over time, and this scale is designed to be sensitive to such changes. It requests subjects to describe how they feel "now." 2. The A-State form of the ZIPERS (Zuckerman, in press) was administered with the STAI A-State scale to provide evidence of their concurrent validity. This test consists of the items of the 13 reactions on the ZIPERS, with instructions to indicate the reactions being experienced "now." 3. A third state measure was provided by the fear thermometer (Walk, 1956). Subjects are asked to place a check on a 1 to 10 scale to indicate the amount of fear they are experiencing at that moment.

Procedure
During the first 4 weeks of the semester, subjects were administered the predictor tests. Six weeks later, subjects were scheduled for the three fear situations. Each subject individually participated in each fear situation once, and self-report, observer's

86

M. MELLSTROM, JR., G. A. CICALA, AND M. ZUCKERMAN

It was thought that weather conditions might influence the subjects' behavior on the landing since it was open, except for metal grillwork, on two sides. To observe the effect of this nuisance variable, the experimenter recorded the outdoor temperature, precipitation, and wind conditions, each on separate scales of intensity, after each subject departed. Darkness situation procedure. Upon arrival, subjects heard taped instructions telling them that their task was to sit in the adjoining room, turn off the light, and remain in the darkness until the experimenter returned. The STAI A-State and ZIPERS State scales were administered at the end of the tape. After completing the scales, the darkness fear test was initiated. The subject was left alone in the room, with the experimenter in the adjoining room. The time from the door being closed to the subject's turning off the light was recorded as the latency. The length of time the subject spent in darkness, up to a limit of 300 sec, was recorded as total time. At the end of the behavioral test, the subject completed the fear thermometer, describing the fear she had experienced in the darkness. After the subject left, the experimenter made the observer's rating of her anxiety in the situation.

Relations Among the Predictors High intercorrelations between the TMAS, STAI A-Trait, and EPI Neuroticism scales indicated that there is considerable overlap in what is being measured by these three tests. The correlations for these scales ranged from .73 to .86. Since they intercorrelated so highly, these three tests will be called, for purpose of discussion, the "ncuroticism tests." The subjects' total scores on the FSS and ZIPERS, which will be called the "omnibus" measures, correlated more highly with each other (r = .68) than with the neuroticism tests, suggesting that two different forms of general A-Trait were measured. Though basically unrelated to the neuroticism tests, the Thrill and Adventure Seeking Scale showed moderately high negative correlations with the FSS and ZIPERS. The intercorrelations of the specific trait measures of the FSS, ZIPERS, and SSQ (not shown in these tables) revealed good convergent validity (Campbell & Fiske, 1959) for all three fears. The correlations between measures of the same specific fear from different tests were high, ranging from .48 to .78. Thus, the various A-Trait measures fell into three classes: (a) neuroticism measures, consisting of the TMAS, STAI A-Trait, and EPI Neuroticism scales; (b) omnibus measures, comprised by total scores on the FSS and ZIPERS; and (c) specific A-Trait measures, represented by the specific trait measures of the FSS, ZIPERS, and SSQ. Since comparisons among the individual measures were not as important as comparisons among classes of measures, the standard scores of the measures comprising a given class were summed to provide a composite measure representing that class. The intercorrelations among these composites and the Thrill and Adventure Seeking Scale, the fourth predictor, are shown in Table 1. The omnibus composite measure showed moderate to strong correlations with all the other predictors, probably as a result of its "general" nature. The correlations between the neuroticism composite and most of the other predictors were low, whereas the Thrill and Adventure Seeking Scale correlated highly with everything but neuroticism.

RESULTS There was a truncation of range in the behavioral measures of the heights (task score) and darkness (total time) situations. In the heights test, all but 10 subjects received the lowest possible fear score, whereas in. the darkness test, all but 4 subjects remained in darkness the maximum time, receiving the lowest possible fear scores. As a consequence of the restricted range, the correlations of these variables with the predictors were expected to be low. Fortunately, the other measures taken in the fear situations were not as seriously restricted in range, permitting higher correlations with the predictors. It should also be noted that the total time measures taken in the darkness and heights situations were subjected to a reciprocal transformation so that high scores on all fear measures indicated greater fear. Only one of the recorded nuisance variables affected subjects' behavior in the situations. A correlation of .28 (df 75, p < .01) was found between precipitation and task score in the heights situation, so the relations between the task score measure in the heights situation and all other measures were assessed by partial correlations, thereby removing the effect of the precipitation variable.

GENERAL VERSUS SPECIFIC TRAIT ANXIETY MEASURES

87

Though all three specific A-Trait composites were significantly correlated, the relation between the composites of heights and darkness fear was the strongest. Relations Among the Situational Fear Measures Overall, the response measures taken in each situation intercorrelated quite highly. STAI A-State and ZIPERS-State correlated between .43 and .49 with the main behavioral measure in each situation and between .48 and .64 with the clinical rating. The fear thermometer correlated between .53 and .62 with the behavioral measures and around .70 with STAI A-State and ZIPERS-State. The last two measures correlated .86, .85, and .77 in the snake, heights, and darkness situations, respectively. Some of the situational measures, for example, total time, simply did not correlate with the other measures. However, most of the response measures did intercorrelate highly within each situation, increasing confidence in their construct validity as measures of fear or anxiety states. Somewhat similar to the findings for the specific trait predictors, responses in the heights and darkness situations were correlated, whereas responses in these situations

TABLE 1 INTJORCORREI-ATIONS AMONG THE FOUR CLASSES PREDICTOR MEASQRES


Scale Neuroticism SSS-TAS
Snake A-Trait

Omni- Ncurot- SSSSnake bus icism TAS A-Tiait

Height A-Trait

Height A-Trait Darkness A-Tiait

.45* -.57* .55* .6.!* .74*

-.21 .l.i .29* .20

-..14* -.57* --.56*

..TO* .42*

.59*

Ntilt:. Omnibus Summed standard scores of the total scores on the Geer Fear Survey Schedule and Zuckernian Inventory of Personal Reactions. Neurotieisni Summed standaid scores on the Taylor Manifest Anxiety Seale, State-Trait Anxiety Inventory Tiait Scale, and Kysenck Personality Inventory Neuioticism scale. SSS-TAS = Thtill and Adventure Seeking subscale of the Sensation Seeking Scale. Snake A-Ttait Summed standard scores of the snake A-Trait measures of the Geer Fear Survey Schedule, Xuckciman Inventory of Personal Reactions, and SSS. Heights and darkness A-Tiait composites weie similarly derived.
*J> < .01.

did not correlate with responses in the snake situation. Relations Between the Predictors and Situational Fear Measures Table 2 shows the correlations between the four classes of predictor measures and the situational fear measures. The validity coefficients of the specific composites represent the relations between the situational fear measures and the specific composite designed to predict that fear. To provide a single mea-

TABLE 2 CORRELATIONS AND HETA WRIGHTS FOR FOUR CLASSES OF PREDICTORS


P rcdictor

/-test comparison SSS-TAS

Situational fear measure Snake situation A-State Fear thermometer Observers' rating Behavioral index Heights situation A-State Fear thermometer Observers rating Behavioral index Darkness situation A-Statc Fear thermometer Observels' rating Behavioral index % Significant r"s

I : Omnibus

II : Neuioticism

III : Specific

\ vs.

Ill

II vs.

Ill

,42**(.00) ..18** (.04) .42** (.00) .23*(-.22) .2S*(-.10 .25* (-.24) .16(-.19) ,29**( --.!(>) ..18**(-.02) .4 !**(-. 02) .28*(.04) .09 (.00)

,25*(.I6) .1.1 (.04) .15 (.05) .04 (.03) .3/)** (.32") ,27*(.21) .12 (.06) .15 (.04) ..17**(.2H") .25**(.14*) .08 (-.01) .07 (.0.1)
25%

.61**(.57) ,62**(.59) .62** (.50") .64**(.73) .29**(.I3) ,4X**(.48) ,37**(.4M .44** (.34") .35** (.13) ,49**(.2'>) .30** (.12) .06 (-.09)
92%

-.28** (-.05) -.22* (.00) .36**(-.l6) -.20 ( . 0 8 ) ~.3t**(-.23) -,34**(-.t6) -.24*(-.10) -.**(-.27") -.4.1**(-.32'0 -.45**-(.28") -.36**(-.28) -.21 (-.25)

I > II III > I III > II 1 > 11 III > I 111 > II
HJ III >
1

.s

III > I

III > II

III > I I

.6.1 .62 .64 .66


.4 4 .53 .39 .9 4 .53 .53 .58 .22

III > [ III > I 111 > II

50%

67%

Note. The values in parentheses are beta weights. Omnibus = Summed standard scores of the total scores on the Geer Fear Survey Schedule and Zuckeiman Inventoiy of Personal Reactions. Neuroticism Summed standard scores oti the Taylor Manifest Anxiety Scale, State-Trait Anxiety Inventory Trait Scale, and Eysenck Personality Inventoiy Neurotieisni scale. Specific = For each of the three fears, summed standard scores on the specific A-Trait measures of the Geer Fear Survey Schedule, Zuckei man Inventoiy of Personal Reactions, and SSQ. SSS-TAS = Thrill and Adventure Seeking subscale of the Sensation Seeking Scale. * The regression of the criterion variable on this predictor is significant (p < .05) after taking the effect of the other predictors into account.

88

M. MELLSTROM, JR., G. A. CICALA, AND M. ZUCKERMAN

sure of A-State in each situation, the standard scores of the STAI A-State and ZIPERSState measures were summed for each situation. In the snake situation, the two behavioral measures were similarly combined. The behavioral indices in the heights and darkness situations were task score and total time, respectively. The latency and total time measures obtained in the heights situation were deleted because they did not correlate with either the predictors or other sitnational measures. The latency measure of the darkness situation was similarly deleted. To compare the validity coefficients of the three types of A-Trait measures a t test for the significance of the difference between two correlation coefficients for correlated samples (Ferguson, 1971, p. 171) was used. The right-hand portion of the table shows the results of these comparisons. For the snake situation, the f-test comparisons show that the specific predictors correlated significantly higher with all four situational measures than either the omnibus or neuroticism measures. In addition, for the fear thermometer and observer's rating, the omnibus composite had significantly higher correlations than the neuroticism composite, which had the lowest validity coefficients of all predictors for all situational measures. For the heights situation, the specific composite predicted the behavioral index significantly better than the neuroticism composile, and it also predicted the fear thermometer and observer's rating significantly better than the omnibus composite. No other differences were significant, although the specific measures had the highest validity coefficients for all situational measures but A-State. The omnibus and neuroticism measures seemed to be equally predictive of the measures in this situation. in the darkness situation, no differences between the A-Trait composites were significant, and no patterns of superiority were readily apparent. However, an overview of the correlations for all three situations reveals some important differences. First, the only situational measures the neuroticism composite predicted were self-reports. In contrast, the specific predictors showed significant cor-

relations with all types of situational measures. The percentages of significant correlations (p < .01) for each type of measure, shown in the bottom row of Table 2, show that 92% of the validity coefficients for the specific A-Trait measures were larger than would be expected by chance. The corresponding percentage for the neuroticism composite was only 17%. The specific measures showed significantly higher validity coefficients than the omnibus measures for the snakes and heights situations but not for the darkness situation. The overall validity seems somewhat better for the specific measures, as demonstrated by the percentages of significant correlations (92% versus 58%), but this difference is not nearly as great as that for the specific versus neuroticism comparison. Only for the snake situation did the omnibus measures show significantly higher correlations than the neuroticism measures. However, in all situations the omnibus composite, in contrast to the neuroticism composite, correlated with other situational measures, in addition to A-State. Table 2 also shows that the Thrill and Adventure Seeking Scale, which does not measure anxiety directly, shows considerable validity in the three situations. It seems that the validity of the scale was as good as any other predictor, with the exception of the specific A-Trait measures in the snake situation. To shed more light on the composition of each situational measure and the relative contributions of the predictors, 12 stepwisc multiple regression analyses using the four types of pretests as predictor variables and each situational measure as a criterion variable were performed. For the measures in the snake situation, the beta weights of the specific A-Trait predictors were universally the largest, suggesting that anxiety and fear in this situation were caused primarily by fear of snakes, thus permitting the specific composite to be most predictive. For the heights situation measures, the results were not as simple. For the A-State criterion variable, the beta weight of the neuroticism composite was the only one to reach significance, although the Thrill and Adventure Seeking scale approached signifi-

GENERAL VERSUS SPECIFIC TRAIT ANXIETY MEASURES

89

cance, suggesting that these traits, not fear of heights, were related to A-State. For the behavioral index, the weight associated with the Thrill and Adventure Seeking scale was as large as that of the specific A-Trait composite, and both were significant. In the darkness situation, the weights for the behavioral measure showed that the Thrill and Adventure Seeking scale was the variable that best explained the subjects' behavior. Subjects' levels of A-State in this situation may be explained by Thrill and Adventure Seeking and netiroticism, with a small component (ft .13) contributed by fear of darkness. For all three situations, the weights of the omnibus composite were near zero or negative, suggesting that when the other three predictor variables are in the regression equation the omnibus measures provide only redundant information. DISCUSSION The results of the present study suggest that measures of specific A-Trait permit more accurate prediction of behavior in specific situations than do measures of general A-Trait. Particularly supportive of this interpretation were the findings for the snake situation, in which for all four situational measures, the correlations of the specific composites were significantly higher than both the neuroticism and omnibus general A-Trait measures. For the heights situation, three significant differences were found between the validity coefficients of the specific and general predictors. Also, the percentage of significant correlations for the specific A-Trait composite measures was considerably higher than that for either general composite. Although there was little doubt about the superiority of the specific measures relative to the neuroticism measures, the difference in predictive power between the specific and omnibus measures was not as great. Comparisons between the neuroticism and omnibus composite measures revealed slightly greater validity for the latter measures. For the snake situation, two differences between validity coefficients were significant, and across all three situations; the neuroticism measures predicted self-report measures only,

whereas the omnibus measures predicted other measures as well. Thus, for the situations of the present study, the specific A-Trait measures showed the most validity, the neuroticism measures the least validity, and the omnibus measures intermediate validity. However, the fact that for the heights and especially the darkness situations both types of general measures were nearly as valid as the specific ones, demonstrates the potential usefulness of the general measures. It is theoretically important to determine which aspects of these situations permit the general measures to be nearly as predictive as the specific ones. One such aspect of these situations may be threat to self-esteem. Spielberger (1966) and Spielberger et al. (1970) have pointed out that tests such as the TMAS and the STAITrait scale are maximally predictive of situational anxiety when the situation poses some threat to self-esteem. If the heights and darkness situations of the present study were more threatening to self-esteem than the snake situation, this would explain the nearequal accuracy of the neuroticism and specific trait composites in predicting responses to those situations. Since the omnibus composite correlated .45 with the neuroticism composite, the omnibus would also be predictive in situations posing threat to self-esteem. The heights and darkness situations could have posed more threat to self-esteem if subjects felt .the tasks involved were so simple that inability to perform them indicated that the subject was childish, foolish, or timid. This attitude could easily develop in the heights and darkness situations because the tasks were very simple and involved absolutely no risk of physical pain or injury. Support for such a view comes from the fact that only 4 subjects could not complete the darkness task and only 10 could not maximally complete the heights task. In the snake situation, however, subjects may have experienced little or no threat to self-esteem because of an attitude that the task involved was difficult and dangerous enough that no one, including themselves, really expected them to perform it. The cultural stereotype of women's fearfulness of snakes may also have helped pre-

90

M, MELLSTROM, JR., G. A. CICALA, AND M. ZUCKERMAN

vent loss of self-esteem for failure to perform approach rather than avoid a feared situation in this situation. The fact that only half of to obtain arousal. Thus, much behavior usuthe subjects touched the snake and even ally thought to be under the control of fear fewer picked it up suggests that many sub- or anxiety, like that in the present fear situjects did find the task difficult to perform. ations, may also be a function of other variThus, we might attribute anxiety aroused ables such as sensation seeking. Isolation in the snake situation primarily to the pres- of the effects of these other variables enence of the snake, whereas anxiety in the hances our ability to describe and understand heights and darkness situations may have anxiety phenomena. been a product of threat to personal adequacy and a fear of heights or darkness. This idea REFERENCES is supported by the relative sizes of the Allport, G. W. Personality, a psychological interprebeta weights of the neuroticism and specific tation. New York: Holt, 1937. A-Trait composites for the three situations. Cameron, N. The psychology oj behavior disorders: A bio-social interpretation. Boston: Houghton This interpretation helps explain the findMifflin, ings of the present study. First, assuming Campbell, 1947. & Fiske, D. W. Convergent and D. T., differential threat to self-esteem in the fear discriminant validation by the multitrait-multisituations explains why the general measures mcthod matrix. Psychological Bulletin, 19S9, 56, 81-105. were almost as predictive as the specific measures in the heights and darkness situations. Endler, N. S., Hunt, J. McV., & Rosenstcin, A. J. An S-R inventory of anxiousness. Psychological It also suggests that if the tasks of the Monographs, 1962, 7(5(17, Whole No. 536). heights and darkness situations had been Endler, N. S. & Shcdletsky, R. Trait versus state anxiety, authoritarianism, and ego-threat versus more difficult, the differences in predictive physical threat. Canadian Journal oj Behavioural power between the general and specific preScience, 1973, 5, 347-361. dictors might have been as great as was found Eysenck, H. J., & Eysenck, S. B. G. An improved for the snake situation. Second, the interpreshort questionnaire for the measurement of extratation explains why the responses in the version and neuroticism. Lije Sciences, 1964, 3, 1103-1109. heights and darkness situations were correlated: They both threatened personal ade- Ferguson, G. A. Statistical analysis in psychology and education. New York: McGraw-Hill, 1971. quacy. Third, the interpretation explains why Geer, J. H. The development of a scale to measure the neuroticism composite predicted the selffear. Behaviour Research and Therapy, 1965, 3, reports but not the behavioral measures in 45-53. the heights and darkness situations. The be- Hodges, W. F., & Spielbcrger, C. D. The effects of threat of shock on heart rate for subjects who havioral measures, by their very nature, were differ in manifest anxicy and fear of shock. confined to the specific fear of heights or Psychopkysiology, 1966, Z, 287-294. darkness, whereas the self-report A-State tests Hodges, W. F., & Spielberger, C. D. Digit span: An indicant of trait or state anxiety? Journal of measured any anxiety being experienced by Consulting and Clinical Psychology, 1969, 33, the subject, including that aroused by "ego 430-434. threat." Itouston, B. K., & Hodges, W. F. Siluational denial Another important finding was the surprisand performance under stress. Journal of Personality and Social Psychology, 1970, 16, 726-730. ing degree of predictive power of the Thrill and Adventure Seeking scale. The predictive Kalkirt, E. S. The relationship between a measure of transitory anxiety and spontaneous autonomic power suggests that persons scoring high on activity. Journal of Abnormal Psychology, 1966, this scale are less likely to report anxiety in 71, 142-146. anxiety-inducing situations. The multiple re- Kirk, R. E. Experimental design: Procedures for the behavioral sciences. Bclmont, Calif.: Brooks/Cole, gression analyses corroborate this finding and 1968. suggest that fear of a given situation is not Lamb, D. H. The effects of two stressors on state anxiety for students who differ in trait anxiety. the only variable controlling a person's apJournal of Research in Personality, 1973, 7, 116proach toward that situation; thrill and 126. adventure seeking is also important. Conceiv- Mellstrom, M. Zuckerman, M., & Cicala, G. A. ably, someone high on this trait may even Anxiety: General versus specific trail in predicting

GENERAL VERSUS SPECIFIC TRAIT ANXIETY MEASURES fear of smite. Psychological Reports, 1974, .75, 317-318. Mischel, W. Personality and assessment. New York: Wiley, 1968. Sarason, I. G. Test anxiety and the intellectual performance of college students. Journal of Educational Psychology, 1961, 52, 201-206. Segal, S., Sensation seeking and anxiety: Assessment of responses to specific stimulus situations. Journal of Consulting and Clinical Psychology, 1973, 41, 135-138. Spiclberger, C. D. Theory and research on anxiety. In C. D. Spiclberger (Ed.), Anxiety and behavior. New York: Academic Press, 1966. Spielberger, C. D., Gorsuch, R. L., & Lushcnc, R. E. STAI manual for the State-Trait Anxiety Inventory. Palo Alto, Calif.: Consulting Psychologists Press, 1970.

91.

Taylor, J. A. A personality scale of manifest anxiety. Journal of Abnormal and Social Psychology, 1953, 48, 28S-290. Walk, R. D. Self-ratings of fear in a fear-invoking situation. Journal of Abnormal and Social Psychology, 1956, 52, 171-178. Zuckerman, M. Dimensions of sensation seeking. Journal of Consulting and Clinical Psychology, 1971, 36, 45-S2. Zuckerman, M. The sensation seeking motive. Progress in Experimental Personality Research, 1974, 7, 79-148. Zuckerman, M. State and trail anxiety: New approaches to assessment with implications for motive measures. In M. Zuckerman & C. D. Spielberger (Eds.), Emotions and anxiety: New concepts, methods, and applications, in press. (Received May 27, 1975)

Anda mungkin juga menyukai