INDIVIDUAL DIFFERENCES IN MAKING PERCEPTUAL INFERENCES mg (0.4 (DNOO '-—I I Thesis Ior ”19 Degree OI M. A. MICHIGAN STATE UNIVERSITY Charles A. Kiesler 1960 L 13 R A R Y Michigan Stan University INDIVIDUAL DIFFERENCES IN MAKING PERCEPTUAL INFERENCES BY Charles A. Kiesler AN ABSTRACT Submitted to the College of Science and Arts, Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF ARTS Department of Psychology 1960 ABSTRACT This study was primarily designed as a replication of an experi- ment by Binder in which a significant positive relationship was found between Lheperenoia scale- .efrthe. M.M.Pl....__end ceetieeseeee-.i.n,-.rea¥sie8__§__w ,pjrrfcgfeptualcinferenceshi.e., allowing a greater number of cues to ac- cumulate before making an inference about the identity of the stimulus. In addition to investigating the relationship obtained by Binder, hypothe- ses were formulated relating other variables to cautiousness in making a perceptual inference: introversion-extroversion and two cognitive variables, the latter derived from a task requiring the subject (S) to distinguish between descriptive and inferential statements. The procedure consisted of two phases, learning and test trials. In the former, 41 _S_s learned the names of a fixed set of alternative stimuli by the paired-associates method. Then, in the test trials, a sequence of slides was presented in which each successive slide pro- vided more information about the identity of the stimulus. The §_s were instructed to name the stimulus as soon as possible. As more cues ac- cumulated, the identity of the stimulus gradually, became less uncertain. As each slide was presented, the §_s had a choice of responding with one of the stimulus—names they had learned or of waiting until more cues had accumulated. Scores were assigned to the SS in accordance with the point in the sequence at which they were willing to respond, which was considered to be a measure of cautiousness in making perceptual inferences. None of the hypotheses was supported. A test of the significance of the difference between the correlation obtained in the present study ii and that obtained by Binder was performed. Although the difference between the correlations was not statistically significant (p = . 0534), it was concluded that the results of this study could not reasonably be interpreted as supporting Binder's finding. I .~ < - Approved V/,[31IJ%[)“A§3, Major Professor " / Id Date \— .4 C 1 #160 I/ ,7 .I iii INDIVIDUAL DIFFERENCES IN MAKING PERCEPTUAL INFERENCES BY Charles A. Kiesler A THESIS Submitted to the College of Science and Arts, Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF ARTS Department of Psychology 1960 ACKNOWLEDGMENT The author wishes to express his thanks and appreci- ation to Dr. Gerald F. King for his continuous advice and guidance in the preparation and completion of this research. His unselfish assistance and supervision as chairman of the master's committee helped to make this an interesting and invaluable learning experience. The constructive suggest- ions provided by the other committee members, Dr. Frank Restle and Dr. Terrence M. Allen, are also deeply appreci- ated. 3:: :3 >:: :k ::< :1: >I< >1: >;< II. III. IV. TABLE OF CONTENTS . INT RODUCTION ..................... METHOD ....................... RESULTS ........................ DISCUSSION. . . . . . . . ............... . SUMMARY. . . . . ................... REFERENCES . . ......... . . ........ APPENDIX.... ............... vi Page 12 19 22 24 25 TABLE 1. LIST OF TABLES Product-Moment Correlations Between Perceptual Inference (_(_3_) and Other Variables Stipulated in Hypotheses. . . . . .................. . Evaluation of Hypothesized Relationships: Log Transformation on C Scores ............. . Comparison of High and Low Groups (_P_a_, Extro- version, A, and Cir) on Perceptual Inference (E) . vii Page 13 15 17 LIST OF FIGURES FIGURE Page 1. Stimuli Shown During Successive Trials in the Test Phase (After Binder, 1958) . . . . . . . . . . . . . . . 8 Z. Stimulus Objects Shown During Learning Trials and Their Respective Names (After Binder, 1958) . . . . . 9 viii I. INTRODUCTION In the visual recognition situation where persons are asked to identify an object, they may differ in both the number of cues which they utilize and the class to which they assign the object. Binder's (1955) model for visual recognition implies that the person responds with a class name when he has enough cues available to satisfy him- self as to the identify of the object. For example, suppose a person sees an object approach him on the road. While it is still some dis- tance away, he may say it is a. car with some degree of assurance. As 'it approaches him and more cues become available, .he may identify it successively [as a Chevrolet, 3. Chevrolet sedan, and a Chevrolet sedan, Model 210. As the car gets closer and details are more apparent, a larger set of cues is available to him. However, the set of cues available when he identified the object as‘a car is still available when the response of "Chevrolet" was made but other cues had accumulated. If we label the set of cues available at each successive class-naming re5ponse as K1, K2, K3,and K4, respective- ly, then we may make the statement: Kl c KZ c K3 c K: As more cues become available, we can assign the object to a smaller class. A few more cues in our example could conceivably allow a unique specification, e. g. , the Chevrolet sedan, Model 210, that the Jones' own. It is obvious that K4C K5, where K5 is the num- ber of cues necessary for unique Specification. The assumption of monotonicity of cues implies an inverse monotonicity of possible classes, that is: KICKZC K3 ........ CKn —~> CnD ...... C3DCZDC1, where Kn is the set of cues necessary for unique specification and Cn is a class with 11 objects in it. Returning to our example, suppose another less experienced person is placed in the same situation, waits until the automobile is relatively close to him, and then calls it a.GeneralMot0rs car, unable to give a more specific response. Thus, we would have a series of responses, on the one hand, and a single response different from any in the preVious series, on 'the Other hand- None‘of the responses misidentified the object, and in that sense all are "correct. " Greater experience with the object allowed an assignment to a smaller class. In the first example, the person was able to use more cues to differen- , tiate this car from other cars, while the same cues could not serve this purpose for the less experienced person in the second example. However, this is a matter of conjecture, and we cannotmake the statement with any degree of certainty. Suppose, on the other hand, we present subjects (_S_s) with a series of line drawings which become increasingly complex and which come to look more and more like the automobile in question. The _S_s are asked to identify the object as soon as possible. Prior to this, in order to lessen the effect of experience, we could show each_S_ draw- ings of, say, eight automobiles with their respective names and in- form him that the series would end with one of these drawings. Thus, we have a fixed set of alternatives and have attempted to equate Ss' experience with each alternative. Each step in our series of drawings would eliminate some of the eight possibilities and leave others. After each step, é would have the choice of responding with one of the class names with which he had become familiar or of waiting until more cues had accumulated. Since the hypothetical series could be continued until there is no reasonable doubt about the identity of the object (in the fixed set of alternatives), there is no question of whether Es would .respond event- ually. Also, since incorrect responses could be ignored and the series- continued, there is no question of whetherSs respond correctly. Rather the question would be: When in this process of accumulating cues are Es willing to make an inference concerning the nature of the object? Binder (1958) refers to the point at which the individual is will— ing to make an inference as the ”recognition response level, " and the above example is similar to the technique he has devised to determine this point. The present author prefers to stress the inferential nature of perception and will refer to the variable as "cautiousness in making a perceptual inference" (g). A person who demands a large number of cues before making a perceptual inference would be considered high in 9. Using the above technique, Binder (1958) studied the relation- ship between C_ and. certain personality variables, as measured by the Minnesota Multiphasic Personality Inventory (MMPI). Only one pre- diction was supported, a positive relationship between E arid the paranoia (E) scale. The interpretation of this finding was that the distrust and suspiciousness shown by persons with high IE. scores tended to lead to a perceptual inference only after a relatively large number of cues had accumulated, where uncertainty was low. Predictions in Binder's study which were not supported were: (a) peo- ple with large differences between hysteria and psychasthenia scores (Hi-131;) are less cautious in making perceptual inferences than those with low Hy-Pi scores; (b) low Piscorers are less cautious in making perceptual inferences than highfl scorers. The purpose of the present research was twofold: (a) to replicate Binder's finding with the P3 scale, and (b) to investigate the relation- ship of other variables to C_. One variable which may relate to cautiousness in making a per- ceptual inference is introversion- extroversion. Himmelweit (1946) and Eysenck (1947) characterize the extrovert, as compared to the intro—- vert, as being quick but inaccurate in performance. Himmelweit pre- sented extroverts and introverts with tests which involved a choice between emphasizing speed or accuracy in completing the task. The approach of the extroverts showed a preference for Speed, while the introverts were oriented toward accuracy. On this basis, it might be expected that extroverts would be less cautious in making perceptual inferences than introverts. Bruner (1957) feels that the inferential a3pects of perception are very similar in general to the inferences characteristic of cognition. One might expect that a person who demands little information before making a perceptual inference, i. e. , low in 9, would evidence similar behavior in the cognitive sphere and View inferences as descriptions of a situation, i.e., as fact. This hypothesis was tested using Haney's (1953) Uncritical Inference Test, which consists of a short story, serv- ing as stimulus material, followed by a series of statements designed to test the ability to discriminate between descriptions and inferences. The present author derived the following two scores for this instrument: Acceptance (A), classifying inferential statements as descriptive; and CircurnsPection (C21), classifying descriptive statements as inferential. It seems reasonable to predict that high A scorers would'be less cau- tious in making perceptual inferences than low A scorers, and that people who score high in_C_i_r_ would be more cautious in making per- ceptual inferences than those who score low in 93.5. Specifically, the hypotheses formulated for testing were as fol- lows: 1, high 133: scorers are more cautious in making perceptual in- ferences than low_1:_’_a scorers; 2, extroverts are less cautious in mak- ing perceptual inferences than introverts; 3, high A scorers on Haney's test are less cautious in making perceptual inferences than low A scor- ers; 4, high _C_3_i_r_scorers on Haney's test are more cautious in making perceptual inferences than low Cir scorers. 11. METHOD Subjects The subjects (is) consisted of 53 students enrolled in an intro- ductor‘),r course in psychology. All _§_s were administered the paper and pencil tests during regular class hours. Of the original sample, 41 S_s (22 males and 19 females) volunteered for the necessary furtherpro- cedures which required individual administration. Paper and Pencil Tests Two paper and pencil tests were used, the first containing an introversion-extroversion scale and a paranoia scale, and the second was the Uncritical Inference Test (Haney, 1953). The introversion- extroversion scale was taken from the Maudsley Personality Inventory (Eysenck, 1959), and the paranoia scale was that contained in the MMPI.* The other fifty- six items of the Maudsley Personality Inven- tory, twenty-four of which comprise a neuroticism scale, were used as "filler" items for the first test. Appa ratus A Bell and Howell Robomatic projector was used to project 2 by 2 in. stimulus slides onto a rear projection screen 12in. square. An automatic timing device allowed control of intertrial interval as well as time of presentation of each slide. * TheEa; scale was not corrected forKalthough it was in Binder's study. The possible implications will be discussed later in the paper. Stimulus Mate rials The stimulus objects were taken from Binder (1958), and are presented in Figures 1 and 2. Note in Figure 1 that the progression in level from A to E is accompanied by an increasing number of cues concerning the identity of the object. Stimulus objects A1 and B1 con- tain elements common to all eight E objects, each object at level. C contains elements common to four of the E objects, and each object at level D contains elements common to two of the E objects. Care was taken to insure that each element common to a number of ob- jects occupied exactly the same position in the respective slides. The stimulus objects were first carefully drawn by draftsmen and then photographed for the slides used in the experiment. Procedure The part of the experiment designed to indicate the level' of perceptual inference consisted of two phases, learning trials follow- ed by test trials. During the learning phase, the names Of the ob— jects in Figure 2 were learned ,by the paired—assbciates method. Note that these objects are identical to thOse of level E in Figure l. The eight objects were presented in rahdom order, 'then the order was reversed. A sequence in which all eight objects appeared was considered to be a trial. Each slide. waspresented fOr-seven seconds with a seven-second interval between slides. During the first trial, _E_2 gave the name of the stimulus object immedi‘ately‘afterit appeared. During succeeding trials, _S_ was asked to attempt to give the name of the object. If§ had not responded after four seconds, _E_3,gave the name. The criterion of learning used was two consecutive trials in which all eight of the stimulus objects were correctly named by §. IA 5‘1 [:1 Level Figure 1. Stimuli Shown During Successive Trials in the Test Phase. (After Binder, 1958) Golf Flag Seaway Pennant Sail Vacuum Cleaner Tomahawk Battle Axe Hatchet W Banner Figure 2. Stimulus Objects Shown During Learning Trials and Their ResPective Names. (After Binder, 1958) __:a 4? —:.z P 9. Q—A +£- 10 After the _S_s had learned the names of the objects to criterion, the test trials commenced and the §_s were given the following instructions: "Now I will show you some other figures. The first figure in each trial will be one pa}: ofpie of the objects you have already learned. In each of the succeeding trials, there will be one other part of the same object added until, final- ly, at the end of the sequence you will be shown the whole object. As soon as you can recognize which of the objects we are leading up to, tell me its name. Try to determine which of the objects we are leading up to as quickly as pos- sible. Each of the figures will be shown only a very short time, so be alert! Remember that each slide will be pre- sented for only a brief exposure and that a slide with more information will follow it after a short interval. It is im- portant that you respond immediately upon recognizing the object represented. I will notify you when each new trial is to begin. At each new trial, a different object will be represented in the sequence. Do you have any questions before we begin the first trial? " Each _S_ was presented with four sequences of five slides each. Each successive slide in a sequence contained a greater amount of information about the stimulus object with which the sequence was to end. Each of the slides was on for two seconds with a seven-second . . an: . Interval between slides. For any particular _S, the four trials would consist of the following in random order: 2:: In Binder's experiment, each figure was presented for 1 1/4 seconds with a nine-second interval between figures. ll AIBICIDI and E1 or E2 AIBICIDZ and E3 or E4 A1B1C2D3 and E5 or E6 A1B1C2D4 and E7 or E8 The _S_ received a score on the sequence in accordance with the level at which he responded. If he responded at level A with only the minimum number of cues, he received a score of one on the sequence. If he responded at level E with the maximum number of cues present, then he would receive a score of five on the sequence. III. RESULTS On each sequence, S_was given a score between 1 and 5, depend- .ing on whether he responded on the A, B, C, D, or E levels. Each §_ was presented ~with four sequences and consequently had a total pos- sible score of 20. A score of 20 for a particular _§ would indicate he was highly cautious in making a perceptual inference and had withheld his inference until the maximum number of cues had accumulated. Conversely, a score of 4 would indicate _S_ was very low in E and had responded on each sequence with only the minimum number of cues available. The range of scores obtained was from 6 to 20, with a mean of 14. 5 and a standard deviation of 3. 33. In Binder's study, the mean was 11. 5 and the standard deviation4. 1. This indicates that in Binder's experiment the _§S were (significantly less cautious in making a per- ceptual inference than in the present experiment (_t_= 3.47; p < .001). The results will be presented as they bear on each hypothesis. In Table 1 can be found the correlations between "cautiousness in mak- ing a perceptual inferencef' (_(_3)and each of the predicted variables. " Hypothesis 1. The obtained correlation between the Pa scale and Ewas .08. While in the predicted direction, the _1_'_is obviously not significantly different from zero. In Binder's experiment, I the corre- lation .between Ba and g was found to be .49. Using Fisher's .23. transformation (Walker and Lev, 1953), where zr = 1/2 loge i + 3 and .- 42r - 42r 0 o o . — - E Z: " —-""'""z z , one may test the Significance of the difference be- r - r tween two correlations. Although the obtained difference between the two correlations does not quite reach a satisfactorylevel of significance 12 13 Table 1 Product -Moment Correlations Between Perceptual Inference (E) and Other Variables Stipulated in Hypotheses Pa Extroversion A Cir Perceptual Inference (E) .08 -.05 —. 15 -.07 14 (£2 .0534), the results of this experiment cannot reasonably be inter- preted as support for Binder's findings. In order to further explore the relationship between Pa and _C_, and since the assumption that the perceptual inference scores came from a normally distributed population was doubtful, each 9 score was sub- tracted frOmthe total possible score (20) and a log transformation per- formed. The correlation between_P_z_3_. scores and the transformed 9 scores was found to be .07, clearly insignificant. Thus, there is no indication that a skewed distribution masked a Significant linear rela- tionship between 133 and E. A summary of the product-moment corre- lations between the transformed _C_J_ scores and each of the predicted variables is presented in Table 2. I At this point it might be asked: To what extent does the paranoia scale really measure paranoia? Comrey (1958) factor-analyzed the MMPI _P_a scale and found thirteen items which he considered repre- sented a relatively "pure“ paranoia scale. The product-moment corre- lation between the "pure"paranoia scale and the full 23. scale was found to be . 396, a value extremely low considering that all of the items in the "pure" factor were also in the full scale. The ”pure" factor was also correlated with 9 resulting in an: of I . 02, which is very close to the _1; obtained between _P_a and £3. This further substantiates the previ- ously indicated lack of relationship between _l_3_a and _C_, even though the range of scores on the "pure“ factor would restrict the correlation co- efficient (twenty-six of the forty-one ES had a zero score on the "pure" paranoia factor). Hypothesis 2. The correlation between extroversion andC was -. 05, in the predi'cted’direction but clearly insignificant. The correlation be- tweenextr'oversiOn’and the transformedC scores was . 10. The change in 15 Table 2 Evaluation of Hypothesized Relationships: Log Transformation on 9 Scores Pa Extroversion A Cir Perceptual Inference (E) . 07 .10 .18 .16 l6 direction of the correlation coefficients is to be expected Since, in transforming the C scores, each score was subtracted from the total possible score. The results do not support the hypothesis that extro- verts are less cautious in making perceptual inferences than introverts. Hypothesis 3. The correlation between A(classifying inferential statements as descriptive) and E was -. 15, again in the predicted di- rection but not significant. The correlation between A and the trans- formed 2 scores was . 18. The results do not support the hypothesis that _S_s who tend to classify inferential statements as descriptive can Haney's Test are less cautious in making perceptual inferences. Hypothesis 4. The correlation between Cir (classifying descrip- tive statements as inferential) and C was -. 07, not significant and not in the predicted direction. The r between Cir and the transformed C, scores was .16. The results provide no support for the hypothesis that 83 who tend to classify descriptive statements as inferential on Haney's test are more cautious in making perceptual inferences. Additional Analys e s In an effort to arrive at a more sensitive test of each of the hy- potheses, a comparison was made between the meanE score of_S_s with high scores onthe other variables (1:2, Extroversion, _A_, and Cir) with those §S with low scores. For this purpose, each E was ranked on each variable. The _S_s were then divided on each variable as close to the median as possible. The meang score of thoseés falling in rough- ly the upper 50% of each distribution was compared with the mean 9 score of those falling in the lower 50%. Table 3 presents this analysis. For each variable, the two mean 9 scores were so close that tests of significance were not deemed necessary. These results further indi- cate the lack of relationship between (2 and each of the variables for 17 Table 3 Comparison of High and Low Groups (Pa, Extroversion, A, and Cir) on Perceptual Inference (C) Pa Extroversion A Cir 25* I; s a :1 I; I: 1; MeanCscore 14.6 14.4 14.6 14.4 14.2 14.8 14.7 14.2 :5: I_-l_, high group (upper half of distribution); 1;, low group (lower half of distribution). 18 which predictions were made. It might be noted that in Binder's experi- ment, he found a significant difference between the mean (_3_ scores of the high and low_l_33 groups (a = 2.43, _p_ < . 05). Hanley (1956) has investigated the role of response set "acquies- cence", the tendency to agree to test items independent of content, and its possible effects on correlations obtained in personality research. In the present study, a "yes" answer on the test including thefia and extroversion scales contributed to a high score 75% of the time. It was felt that "acquiescence" should be investigated for two reasons, the first practical and the second theoretical. If there were a relationship between E and "acquiescence", the correlation coefficients between E and each of the other variables could conceivably be depressed; and the variance attributable to this reSponse set should be partialed out when computing the predicted relationships. On theoretical grounds, one might posit a relationship between the tendency to acquiesce and a lack of cautiousness in making perceptual inferences (low in E). The 3 be- tween 9 and "acquiescence" was found to be . 06, clearly insignificant, indicating: a) that the re5ponse set "acquiescence" had little, if any, effect on the obtained relationships between E and the predicted vari- ables; and b) a lack of support for the hypothesized relationship between the tendency to agree to test items regardless of content and a lack of cautiousness in making perceptual inferences. For the reader's information, a table of intercorrelations among the variables considered in this study is presented in Appendix A. IV . DISCU SSION The results do not support Binder's finding of a positive relation- ship between the Ea scale of the MMPI and cautiousness in making per- ceptual inferences (C), nor do they support the hypotheses concerning the relationship of C_ to the other variables investigated in this experi— ment. The results will be discussed in the order that they were previ- ously listed. This experiment was not an exact replication of Binder's study. There are at least two differences in method between the present re- search and Binder's. In Binders experiment, each figure in the sequence was presented for 111- sec. with a 9 sec. interval between figures; in the present research each figure was presented for 2 sec. with a 7 sec. in- terval between figures. It is felt that this difference in timing is too small to account for the difference in results. In Binder's experiment, the as; scale was corrected for K; in the present research it was not. The K scale was designed by Meehl and Hathaway (1946)as a sup- pressor variable to detect test-faking behavior on the MMPI. Low scores represent faking bad, relatively speaking, and high Scores represent de- fensiveness, i.e., faking good. However, Meehl and Hathaway state that if the K scale measures the defensive, lying, or self-deceptive test- faking attitudes it was derived to measure, the correlations between K and each of the scales on the MMPI should be consistently negative and of some magnitude. They found the correlation between K and_l_33. to be -. 07 for normal males and -. 02 for normal females, leading one to sus- pect that for the Ea scale the influence of K as a suppressor variable is very slight. When one is correlating a particular variable withEa and 19 20 corrects for K, it is obvious that the correction would affect the ob- tained correlation coefficient. However, Since the K correction does not seem to be related to the Pa scale in the same manner that it is related to the other scales of the MMPI, it is difficult to see how such a coefficient could be interpreted. Correction for K could depress or inflate a correlation depending on whether K was related to the other variable. It is conceivable that the significant correlation obtained by Binder was entirely due to a relationship between K and_C_._ A possibility exists of a difference in experimental atmospheres. Smock (1955) has found that ES under psychological "stress" conditions reSpond earlier (are less cautious) than §S under "security" conditions. If§s found Binder's experimental situation more "stressful” (e. g. , due to the manner in which the instructions were given), then one would ex- pect the mean (_3_ score to be Significantly lower than in the present re- search. As indicated previously, this was true (_t_= 3.47; p < . 001). It may be that the relationship between Pa and cautiousness in making perceptual inferences is contingent upon a relatively high level of stress. The hypothesis regarding extroversion and C was based on data derived from the performance of hysterics and dysthymics, whom Eysenck considers to be equivalent to extroverts and introverts, re- spectively. Sigal, Star, and Franks (1958) tested this assumption by administering the Maudsley Personality Inventory to groups of hyster- ics and dysthyrnics. The groups were not differentiated by the intro- version- extroversion scale, which casts doubt on the practice of using the hysteric-dysthymic and extrovert—introvert concepts as interchange- able. The question becomes one of whether the MPI scale actually measures introversion-extroversion. The hypotheses concerning) the relationships of A and Cir to 9 rest on two assumptions: (a) Haney's test measures what it purports 21 to measure, and (b) the inferential aspects of perception and cognition are comparable (Bruner's position). Assuming the adequacy of the present research, then at least one of these is incorrect. Theoretical- ly one would expect a strong negative relationship between A and Cir. Appendix A shows that the correlation between A and Cir was . 08, which casts considerable doubt on assumption (a). Assumption (b) warrants further empirical exploration. V. SUMMARY This research was designed as a replication of an experiment by Binder (1958) in which a significant positive correlation was found be— tween the paranoia (33) scale of the MMPI and cautiousness in making a perceptual inference, i. e. , allowing a greater number of cues to ac- cumulate before making an inference about the identity of the stimulus. Subjects (is) learned the names of a fixed set of alternative stimu- li by the paired-associates method. They were then presented with se- quences of slides in which each successive slide provided more infor- mation about the identity of the stimulus with which the sequence was to end. The Es were instructed to name the stimulus as soon as possible. The identity of the stimulus gradually became less uncertain. As each slide was presented, the Es had a choice of responding with one of the stimulus-names they had learned or of waiting until more cues had ac- cumulated. A. score was assigned to each_§ in accordance with the level of uncertainty at which he was willing to respond, which was considered to be a measure of his cautiousness in making perceptual inferences. The following hypotheses were formulated for testing. 1. PeOple who score high on the Ea scale, because of a general distrust and sus- piciousness, are more cautious in making perceptual inferences than those who score low on the Pa scale. This was a replication of Binder's experiment. 2. Extroverts are less cautious in making perceptual in- ferences than introverts. This hypothesis was suggested by Himmel- weit's (1946) and Eysenck's (1947) characterization of the extrovert as being quick but inaccurate in performance. 3. People high in the tend- ency to classify inferential statements as descriptive (i. e. , view infer— ences as descriptions on Haney's test) are less cautious in .making 22 23 perceptual inferences than peOple low in this tendency. This hypothesis and the following one were suggested by Bruner's (1957) view that the inferential aspects of perception are very similar to the inferences characteristic of cognition. 4. People high in the tendency to classify descriptive statements as inferential are more cautious in making per- ceptual inferences than peOple low in this tendency. None of the above hypotheses was supported. A test of the signifi- cance of the difference between the correlation obtained in the present study and that obtained by Binfler was performed. Although the differ- ence between the correlation coefficients was not statistically signifi- cent (p .= .0534), it was concluded that the results of this experiment could not reasonably. be considered as supporting Binder's finding. REFERENCES Binder, A. A statistical model for the process of visual recognition. Psychol. Rev., 1955, 62, 119~129° Binder, A. Personality variables and recognition response level. _J_. abnorm. soc. Psychol., 1958, _5_7_, 136-142, Bruner, J. S. On perceptual readiness. Psychol. Rev., 1957, 64, 123-152. Comrey, A. L. A factor analysis of items on the MMPI Paranoia Scale. Educ. psychol. Measmt., 1958, _1_8, 99~108. Eysenck, H. J. Dimensions o_f personality. London: Routledge and Kegan Paul, 1947. Haney, W. Measurement of the ability to discriminate between infer- ential and descriptive statements. Unpublished doctoral disser- tation, Northwestern Univer. , 1953. Hanley, C. Social desirability and response to items from three MMPI scales: D, Sc, and K. ._I_. appl. Psychol., 1956, 42, 324-328. Himmelweit, H. T. Speed and accuracy of work as related to temper- ament, Brit. J: Psychol., 1946, _3_6_, 132—144. Meehl, P. E. and Hathaway, S. R. The K factor as a suppressor variable in the Minnesota Multiphasic Personality Inventory. _._I_. appl. Psychol., 1946, 39, 525-564. Sigal, J. J., Star, K. H., and Franks, C. M. .Criterion groups for introversion-extroversion, :1. abnorm. soc. Psychol., 1958, 57, 143-147. Smock, C. D. The influence of psychological stress on the "intoler- ance of ambiguity. "_J_. abnorm. soc. Psychol. , 1955, :0, 177-182. Walker, H. M. and Lev, J. Statistical inference. New York: Holt, 1953. 24 APPENDIX 25 Intercorrelations Among All Research Variables 1:: Cir A Extroversion E .08 -.07 ~.15 -.05 Pa -. 24 .12 -. ll Cir . 08 . 08 A .14 MICHIGAN STATE UNIVER 0 3 1293 I IIIIII’I‘I‘“ SI 1062 2389