A FACTOR ANALYTIC STUDY OF THE SOCIAL BEHAVIOR OF SCHIZOPHRENICS 3y Harry W. Blair ubmitted to the School of Graduate Studies of Michigan State College of Agriculture and Applied Science in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Psychology 1952 A FACTOR ANALYTIC STUDY OF THE SOCIAL BEHAVIOR OF SCHIZOPHRENICS By Harry W. Blair AN ABSTRACT Submitted to the School of Graduate Studies of Michig State College of Agriculture and Applied Science in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of psychology 195 2 Year 7 Approved_____ ^ H. V. Blair From the time that the d i s o r d e r known today as schizo­ phrenia has been i d e n t i f i e d , investigators have recognized impairment in social b e h a v i o r as a prominent aspect of both the etiology and c l i n i c a l m a n i f e s t a t i o n s of the disorder* A diversity of c o n c e p t u a l t r e a t m e n t of the observations in this regard suggested a n e e d for concepts or principles which can facilitate f u r t h e r systematic exploration in the area of social b e h a v i o r * alternatives to w h i c h a factor analysis, Ajnong the various methodological clinical considered offers many advantages, psychologist may resort, as a branch of statistical method, p a r t i c u l a r l y in the establishment of classificatory p r i n c i p l e s . S e v era l ii-ctor analytic approaches are specifically a p f l i c u b l e to the problems encountered in clinical psychology research.* The present I n v e s t i g a t i o n concerned itself with the study of the empirical items of personal a no relationships social elicitea in a s t r u c t u r e a b ehavior of schizophrenics situation. factors or concepts w h i c h w o u l d this observed behavior, descriptive usefulness ana of The test items w e r e the the test performances w e r e normal distribution. The more parsimoniously describe factors roughly The It attempted to aerive to then provide- a test of the c ic notorious scoring c a t e g o r i e s choric coefficients. among various specific so derived. scaled and later reduced to for the commutations of tetra- (tI* Olie*lity variables underlying a s s u m e d to be essentially of c o r r e l a t i o n s were based on the test H. V. Blair score distributions of luO hospitalized schizophrenic patients who were highly variable w i t h regard to subtype classification, age, duration of illness, education, etc. The final matrix of the intercorrelations of 4k variables was subjected to a multiple-group centroid type of factor analysis. Preliminary factorization aiaed in the most meaningful placement by graphic rotation of the first two factors, w h i c h were of the greatest size and import. Six oblique factors were extracted w h i c h in some cases were highly related. They were named as follows: A E C D E F General social orientation and capacity Socio-motoric responsiveness Restitutional curiosity nonchalance and lack of preoccupation Reactive assertion and reproduction Interpersonal confronting The residual correlations after the extraction of these six fa.ctors failed to reveal any remaining group factors of s i g n j fieu n c o . Computations of factor l oadin w for case history v: ri- ableo were made, as well as factor loading indices for d i a g ­ nostic categories. This aiuecl not only ir th' g-. rt ral ii.aer- \rotation of factors, but also provided a 'minor test of their use!uIness. /, group of 41 ho spit.- lized non-: chizophrenics was cou^ ■ reel sith the schiz ophroni r* cubfypf c . - * . ♦ t ^ -V -irtrea , y „ a. - -u J _> ~ t to no •.r. factor n C' . . . . . . -■ J. .1. l. h- a. r h-' hi J- >_ i v - • i-> J •tj.iC', - W V,, - * ,-r ' ***io The sohizo; I.rer.ic subtype a ■•Iso d if Cored signifi­ cantly in several import: rt respects, ^ r t icularl;r in general K. B. Blair social capacity. Two second — order factors v:ere extracted correlation Taatrix of the 11 *c .. l- Oj i oil . a g si.. foli;uc hj.^ r o 1 '.t e a from the inter factors. ui*x U Q c 01..^ Tie first o o l*c.li o_l j- '.c tors j , B, and r, ana was interpreted as a very general so­ cial ability which may be dif f ereutia j.ly invn-.ired. by schizo phrenic r e a c t i o n s . The entire investigation was viewed as exploratory i.na the relationships observed were regarded as tentative in their usefulness. ever, may open ir. this area, The results of this investigation, how­ the "ay for future empirical investigations as well as assist in the conceptual thinning tb rot schizophrenia. ii ACKNOWLEDGMENTS The author wishes to express his sincere thanks to Dr. Donald M. Johnson, under whose lofty inspiration, con­ stant supervision, and unfailing interest this investigation was undertaken and to whom the results are herewith dedi­ cated. He is also greatly indebted to Dr. Albert I. Rabin for his kind guidance and valuable help in all phases of the investigation. Grateful acknowledgment is also due to Dr. W. R. VanDenBosch, Superintendent of the Dr. Norman M. Beatty Memorial Hospital, the late Dr. Herbert G. McMahan, the Board of Trustees, and the Indiana Council for Mental Health for making it possible to carry out the research while in their employ. The author deeply appreciates the assistance and con­ stant support of Dr. Wilson H. Guertin, mathematical phases of the study. particularly in the The many others who readi­ ly lent their assistance are hereby acknowledged and the author expresses his sincere gratitude. iii Harry Walter Blair candidate for the degree of Doctor of* Philosophy Final examination, Department Dissertation: November 2b, 1952, 2:00 P. M . , Psychology A Factor Analytic Study of the Social Behavior of Schizophrenics Outline of Studies Major field: Clinical Psychology Minor fields: Collateral work in Sociology and Physiology Biographical Born, Items July 5, 1921, Hudson, New York Undergraduate Studies, Michigan State College, 1 9 I+I-I+2 , cont. 19*+5-*+7 Graduate Studies, Michigan State College, Experience: 19V7-52 Member United States Army Air Force, 191+3-1+5; Clinical Psychology Trainee, Veterans A d m inis­ tration, 19I+7- 51; Staff Psychologist, Beatty Memorial Hospital, Westville, Indiana, 1951-52, Supervising Psychologist, 1952- Member of American Psychological Association iv TABLE OF CONTENTS Page ACKNOWLEDGMENTS ii VITA ill LIST OF TABLES vi INTRODUCTION Social Behavior in Schizophrenia Factor Analysis in Clinical Psychology Research 1 8 PROBLEM 15 METHOD Sample Characteristics 16 The Test Scale 26 Procedure 31 Treatment of Data 33 0CP^T~r T1*o Pest Scores and Correlations 1 d !j j o L M+ Freliminary Factorization 1+9 Final Factorization 5^ T’ ne Obtained Factors and Interpretations 66 Factor Loadings for Individuals 76 The Higher-order Factors 89 Suggestions for Future Research 93 SUMMARY AND CONCLUSIONS BIBLIOGRAPHY 97 100 APPENDIX A APPENDIX B Scale of Tests Used in Test Instructions 10^ Analysis 113 123 vi LIST OF TABLES ?able No. Page No., 1. Fsychiatric Subtype Diagnoses by Sex 19 2. Age of Sample by Sex 20 3. Duration of Illness of Sample by Sex 22 *+. Education of Sample by Sex 23 5. Marital Status of Sample by Sex 2b b. Comparison of Schizophrenic and Non-schizophrenic Groups with Respect to Case History Variables and Ratings of Cooperativeness 10*+ 7. Percent of Plus Scores by Tests 105 S. Chi Square Comparison of Test Score Distributions for Schizophrenics and Non-schizophrenics bj 9. Interccrrelation Matrix of b2 *+8 Tests 10. Loadings on the Preliminary Factor Extracted by Complete Centroid Through Correlation Matrix 11. Clusters Determining Group Centroids 106 12. Average Intercorrelations Within and Between Clusters 107 13• Test Loadings on Six Oblique Factors 1 m-. Cosines of the Angular Separation Between the Six Oblique Factors 15 . Orthogonal Rotation Matrix for the Six Oblique a 50 56 108 59 Factors lo. Test Loadings on Six Orthogonal Factors 60 vii 17. Coramunality Accounted for by Six Oblique Factors 109 18. Residual Matrix After Extraction of Six Factors 110 19. Orthogonal Loadings on the Seventh Factor E x ­ tracted by a Complete Centroid Through Residual Matrix 6b 20. Factor Loadings of Names for Factors .M-O or Higher that Led to the 68 21. Six Factor Oblique Matrix for Schizophrenic SubType 77 Oblique Factor Loadings for Case History V a r i ­ ables and Cooperativeness Rating 79 23. Mean Factor Loading Indices for Schizophrenic Subtypes and Non-schizophrenics 8 *+ 2 ^-. Weighted Mean Factor Loading Indices for Schizo­ phrenic Subtypes and Non-schizophrenics 25. t Ratios for Differences Between Mean Factor Loading Indices among Diagnostic Categories 86 26. Factor Loadings for Two Orthogonal Factors Derived from Intercorrelations of Six Oblique Fac tors 92 22 . 111 1 INTRODUCTION Social Behavior in Schizophrenia Social aspects of etiology and clinical m a n i e e s t a t i o n s . During the past fifty years, from the time that Kraepelin first identified the disorder known then as dementia praecox and today as schizophrenia, a deficiency in social behavior has been included in almost every description of the disorder. This was particularly so after Meyer (33) placed emphasis upon the environmental influences in the etiology of schizo­ phrenia, and the psychoanalytic school speculated upon the role of early infantile experiences in its formation. Many investigators who have concerned themselves w i t h the nature of social behavior in schizophrenic reactions have elaborated upon the loss of social skills, role-taking. Cameron especially that of (6,7*3,9) characterized the behavioral disturbances in schizophrenia as a "social disarticulation" with a progressive loss of the ability to communicate. further hypothesized He that a failure to develop adequate role- taking skills results in a poor adaptation to the cultural pattern. Hunt and Cofer (25) interpreted the apathy of schizo phrenics as consisting essentially of a "loss of socially ac­ quired drives", and in general looked upon the symptoms of schizophrenia as "extinctions of socially learned responses". It is to be noted that while these investigators were ap ­ ,;i parently observing and making interpretations of similar phenomena, namely the overt behavior of schizophrenics, there is an apparent disagreement about the etiology of the social deficiency. Cameron speaks of a "failure to develop" while Hunt and Gofer speak of a "loss" or "extinction" of social r e s p onses. Whether the social deficiencies in schizophrenia are themselves etiological in nature or the result of a more fundamental causative process prior to socialization is an important question waich deserves careful study in the future. One possible approach is that of Gerard and Siegel (21) who studied the family backgrounds of 71 schizophrenics and 3 0 controls. findings, In general, it seemed considering both positive and negative that parental relationships and their social implications were the most important factors. parental attitudes, Through they found, patients were often prohibited from early, or even later social experiences. Though Myerson O 1*) dealt with a later stage in the eti­ ological process, he advanced an interesting theory concerning the role of scrutiny and social anxiety in the development of schizophrenia. Presumably this investigator, others, would hold as well as that the Individual who later develops schizophrenia brings to his social situation a set of at­ titudes or reactions arising out of his early family relation­ ships which so structure his interpersonal relationships as to bring on a social maladjustment, and later, matic reactions of a schizophrenic nature. specific sympto­ 3 Another possible approach to the understanding of the social deficiencies in schizophrenia, is the analysis of the social reactions of the patient, w h ich may or may not refer to his social development. Various descriptions of schizo­ phrenic reactions have been made, and some similarity is ap ­ parent. Wittman and Steinberg (*+7) spoke of the '‘s h u t - i n 1* personality and Slotkin (37) pointed out that the private symbolism of the schizophrenic makes communication with him ineffective. These descriptions seem to suggest a rejection of outside realities, of the social world without, which cur­ tails social interaction. Levin (30) rejected the concept of "splitting", since he felt that only mature adults could exhibit such a phenomenon wherein parts could act independently of one another. He pre ferred to think of schizophrenia as involving a loss of cere­ bral versatility, o r a failure in the development of differ­ entiation. This seems to be in accord with Kasanin (2 8 ) who considered the disorder as a failure in maturation and de­ velopment. To maintain equilibrium or security, no matter how poor the level, the schizophrenic engages in defensive techniques which contribute to his general social inadequacy. Sullivan (*+1) pointed out active attempts on the part of d e ­ teriorated schizophrenics to avoid interpersonal relations, and the use of "quasi-communicative gestures" seen in the manneristic hebephrenic. The defensive use of stereotypy is brought out by Fromm-Reichmann (20), who held that 14- stereotypes screen true emotional reactions. and Bartlett Deutsch (18) (2) commented upon the suggestibility seen in schizophrenics, which may be a technique for superficial adaptation to social demands. Similarly, Deutsch pointed out that passivity is a common feature seen in schizophrenia, but she felt that it may serve to cover the more basic ag­ gressive tendencies. An interesting approach to the study of spontaneous social behavior in schizophrenics has been made by several investigators who studied the ward behavior of these patients. Rabin (35 ) and later Maas, Varon, and Rosenthal (32) de­ veloped techniques whereby observations are recorded over a specific period of time. Rabin dealt with social responsive­ ness as it is reflected by the initiation of social contacts by a patient as well as that patient*s response to contacts initiated by others. than responsive, He reported more spontaneous contacts but the two types of reactions were cor­ related positively in any given patient. Such a measure did not, however, distinguish between schizophrenic and non-schizo­ phrenic patients on a ward, but his technique, as well as others of the observational type, would bear further investi­ gation. The need for precise conce pts. ature pertaining actions, In examining the liter­ to the social aspects of schizophrenic re­ it becomes obvious that a common set of concepts is needed in order that research and communication can be facili­ 5 tated. At the present time one must resort to either very stereotyped descriptions, have done. or invent a new terminology— as some Much of what the clinical psychologist observes in a schizophrenic patient, and later incorporates into his im­ pression, is not verbalized and often even intuitive. One often hears the clinicial say that a certain patient "looks l i k e ” or "seems" like a schizophrenic, without being able to specify the particular observations on which he is making this evaluation. Perhaps one major reason for the difficulty in setting forth just what it is that we do observe is that the behavioral acts are a function of the interpersonal re­ actions of w h ich the observer himself is a part. that we call a schizophrenic This person seems to behave differently toward us than does the average person reared in our culture. He lacks many of the social graces that one anticipates from the other, and is frequently unable to deal in superficiali­ ties common to social interplay. And, as has been said before, he has little recourse to flexible social roles. It is not meant that the clinician could not utilize con­ cepts concerning the social deficits of schizophrenics, but at the present time his thinking is to a large degree bound by inadequate, concepts and rather rigid, classificatory concepts. Such tend to concentrate his observations in narrow areas so that a diagnosis might be arrived at with maximum efficiency. Thus, other behavior, particularly that of a social nature, may either be overlooked or "unconsciously" perceived and 6 incorporated into "intuitive" judgments. To investigate these less formally treated areas would be to reduce the narrowing of vision fostered by Kraepelinian diagnostic classifications. To investigate this area would be a further step toward -under­ standing etiology and treatment, for a complete description of an illness at all levels almost automatically indicates the treatment. The clinical interview situation provides an excellent opportunity to study the social behavior of schizophrenics, but to make use of such a situation for systematic research, a certain amount of structure is needed. structure needed, Not only is general but the use also of certain manipulatory techniques is desirable so as to invoke responses w h i ch may or may not occur if they depended on the chance of spontaneous elicitation. Jones (27) cites several studies of structured interview techniques, and reports that particularly when specific problems are to be answered is this technique suc­ cessful. Much work has been done in the area of structured interview techniques in psychiatry, with the results that the mental status examination is relatively standard in most clinical settings. Cohen and Flower (17) have presented a proposal for the psychiatric examination which employs a slightly different nosological principle than that of the classical mental status examination. The gross concept of inattention is subdivided into various areas such as patterns of behavior, thought expression, and affect. These attempts at refinement of observation in psychiatric practice make for progress, but it is advantageous to begin with classificatory principles that have been demonstrated to be more independent and factorially pure with respect to one another. The es­ tablishment of such functional wholes depends upon the empiri­ cal investigation of the interrelatedness of the variables which comprise the domain of observation. As pointed out previously, a large portion of the clinical interview consists of the interaction between the interviewer and the interviewee. Each is reacting to the other as a person, and each is "testing” the other at some level of awareness. To understand more of these interactions is an extremely dif­ ficult problem, particularly those reactions which the patient sets up and observes in the psychologist. Such understanding could come about only through insight into the patient*s personal frame of reference, and vice versa, insight into the patient*s personal frame of reference would be gained by an understanding of the way in which he is attempting to structure the interview. The latter might be said to be one of the purposes of an interview, cation. It is, however, and is little subject to objectifi­ both possible and desirable to study the patient with a more objective and standardized procedure. This not only would have clinical implications, but would lend itself to research on the social responsivity, or lack of it, on the part of the schizophrenics. 8 Factor Analysis In Clinical Psychology Research The a l t e r n a t i v e s . Particularly over the past few decades have applied psychologists become aware of pressing social responsibilities. To meet these needs of society the prac­ titioner has gone forth armed with a pitifully Inadequate theoretical framework into many and varied fields. He has developed specialized techniques and tests for specialized purposes, and has gradually cultivated a specialized set of concepts--concepts limited in their communicability and tied down to the techniques from which they arose. Research in these applied areas, and especially in clinical psychology, has been based on these concepts of limited usefulness and it too suffers from lack of communicability. There has been some resort to classical experimental methods making use of vari­ ables which are treated by theoretical systems, particularly those of the behavioristic type. While such theories may someday encompass human behavior of a very complex sort, they fail to prove presently useful in the description or prediction of the particular behavior with which the clinician must con­ cern himself. Thus, the clinician who involves himself in such research leads a double life and enjoys only a minimum of inter-facilitation of purposes. At the other extreme the clinician may avail himself of psychoanalysis. resist) To the degree that he "accepts" (or does not the tenets of this school he may effectively employ them in psychodiagnosis and therapy. If, however, he wishes 9 to utilize psychoanalytic principles in his research, the clinician is faced with ill-defined and gross concepts which do not easily lend themselves for this purpose. The alternative to these extremes lies somewhere in be­ tween, and here, two possibilities are open. The first, which is more in the nature of deductive approach, is the develop­ ment of a very broad and flexible theory of human behavior and personality from which testable hypotheses flow. The second possibility is the inductive approach in which the statistical method comes to its fullest realization. useful in controlled experiments, Although statistical methods are chiefly used by the clinician in field studies wherein vari­ ables are dealt with in their natural setting, subject manifold influences of nature. In general, to the the function of statistical operations is to abstract and measure variability due to the particular influences in which the investigator is interested. Despite the availability and use of statistical methods, however, there remains the paucity of conceptual de­ velopment and laws of interaction. The investigator must choose variables somewhat arbitrarily and his results are dependent upon what may prove to be badly-chosen variables. The place of factor a n a l y s i s . Factor analysis can be considered as a technique within the general statistical method referred to above. There are, however, certain dif­ ferences from elementary statistical techniques which place factor analysis in a rather unique methodological position. 10 This derives from the fact that one of the major goals of science is to observe lawful events and to derive therefrom laws of interaction w h i c h facilitate more refined description and prediction. While techniques are available, notably that of analysis of variance, for studying simultaneously the re­ lationship among several variables, they fail to yield in­ formation as to the magnitude of such mutual influences. On the other hand, while simple correlational methods reveal the magnitude of relationships, they fail to distinguish the functional wholes of which the observed relationship is an abstraction. In a sense, the investigator who attempts to explain an observed relationship between A and B Is searching for a factor, or a source of variation. The same investi­ gator learns more w hen he establishes the relationship between B and C, and then A and C. This piecemeal approach, however, is Inefficient since the inclusion of many more variables in­ to a single investigation can establish the mutual relation­ ship of all of the variables included, in addition to infor­ mation as to the common sources of variance. To demonstrate the relationship between variables, tests and factors, or it is helpful to visualize the so-called test space, wherein test vectors are projected from a common point of origin. They are so projected that the cosine of their angular separation between any two tests is equal the coefficient of correlation between those tests. to Factors, then, are merely a system of coordinate axes which have been 11 systematically imposed upon the structure of test vectors. If the situation is such that test vectors may be represented in two dimensional space, then the entire problem is resolvable into two factors or coordinate axes. ance of any test, The common-factor vari­ then, can be completely accounted for by the numerical value of its projections on the two coordinate axes. From this derives the factorial or specification equation w h ich expresses the numerical relationship between a test and the factors from which the test variance derives. Factor analysis by no means represents a new idea in scientific methodology. Burt (5) pointed out that Pearson was concerned w ith the underlying structure which could ac ­ count for the correlation of two events, and in a sense paved the way for factor analysis. Spearman (39) then demonstrated that the mass of intelligence measures employed at that time could be resolved into a general intellective factor. quently fhurstone Subse­ (*+3) proposed three types of factors: gener­ al, group, and specific, which could more adequately account for the variance in a given test. Further refinements in technique and application have been made by investigators too numerous to mention. The main value of factor analysis lies in its contribution to the parsimony of description, in that a few descriptive categories can encompass a wide variety of variables. A cor­ relation matrix may contain thousands of values which are 12 needed to express the relationships among the variables, w h e r e ­ as a factor matrix can yield almost as m u c h information in a matrix w h i c h has as many rows as there are tests, and as many columns as there are factors. Providing that factorization has been complete, and the axes are orthogonal w i t h respect to one another, any original correlation may be reproduced by multiplying the corresponding row of the factor m a tri x by the corresponding column of the transposed factor matrix. A further contribution to parsimonious description has been made by Thurstone simple structure. (M+) who originated the concept of This refers to the location of the factor axes in such a way that some tests are highly loaded on a factor, while others have low or negligible loadings. The latter tests then define a hyperplane through w h i c h the factor axis may project. To achieve simple structure in a factor problem it is necessary to carry out a number of rotations of the factor axes until a structure is achieved which meets the criteria. This, of course, may proceed from orthogonally located factors or from those which are already in some sort of oblique reference to one another due to the use of cluster­ ing methods of extraction. It is almost certain that the rotated matrix will be n o n - o r t h o g o n a l , but at the other ex­ treme, it should not be so oblique that some degree of functional independence is not preserved. then, attempts which, Simple structure, to define the functioning unities of nature, it is important to note, are rarely completely 13 independent of one another, nor all-embracing with respect to the totality of variables in a particular domain. This is particularly true in the area of personality as Cattell 1 3 ) has (11, so well demonstrated. Factor analytic a p p r o a c h e s . The alternative experi­ mental designs offered by factor analysis provide for wide applicability in the social sciences. Of particular interest to the clinical psychologist, and those researchers in the area of personality are the P technique devised and demonstrated by Cattell (12,m-), and the Q technique originated by Burt and Stephenson (M-0). (M-) These, of course, are in addition to the more basic R technique employed in the present investi­ gation in which tests are correlated. In Q technique, which is in transposed or inverted relation to R, people are cor­ related with respect to performance on a wide variety of tests. Thus, such an approach develops descriptive types of individuals who have in common certain test performances. technique, on the other hand, P involves the correlation of a set of variables on one person which is repeated on many oc­ casions, the N being the number of occasions. The use of P technique and certain Q technique designs require the use of ipsative measurements described by Cattell (10) and Guilford (23), as opposed to normative scores used in the R analysis. In ipsative measurements, of standard the scores are expressed in terms scores with respect to the variability within a person instead of a population of persons. I1* When, as in the case of R technique, made in terms of the group distribution, discuss individuals, evaluations are it is precarious to though it may be more defensible to con­ sider groups of individuals. The precariousness of such an analysis depends upon the degree of disparity between the factor structures arising out of the R type of analysis versus the Q type of analysis on similar data. The number of factors common to the two techniques in a given area is relative to, in the case of Q technique, the wideness of the variety of tests chosen, or, in the case of R technique, individuals chosen. the variety of 15 PROBLEM The purpose of this investigation is to study the re­ lationships among various specific items of behavior of schizophrenics in a structured social situation. It will attempt to derive factors or concepts which will economi­ cally describe this observed behavior. The problem includes comparisons to be made among the various schizophrenic subtypes as well as hospitalized n on ­ schizophrenics with respect to the obtained factors. study, then, The strives to test the descriptive usefulness on the group level of the concepts developed by it. It pro­ poses to test whether or not the factorial concepts d e ­ veloped can discriminate among different schizophrenics, and between schizophrenics and non-schizophrenics. 16 METHOD Sample Characteristics A rather heterogeneous group of schizophrenics who ex­ hibited a wide range of symptomatic reactions was employed in the study. This was felt to be desirable in order to in­ sure sufficient variability on the various obtain such a sample, test items. To selections were made within a wide age range with marked variability in durations of illness. Since a certain min imum of cooperation was necessary in order to ar­ rive at valid ratings of behavior, extremely uncooperative patients were necessarily excluded. Ratings of general co­ operativeness were made for all patients included. Those schizophrenic reactions complicated by neurological disorders, senile changes, Shock therapies, and involutional reactions were excluded. including insulin and electric, must have been completed or otherwise terminated at least one month prior to testing in order effects. Finally, to rule out temporary neurogenic only American born subjects were used, for conceivably early cultural learning is an important determinant of social and interpersonal behavior. No exclusion was made on the basis of sex or education, and separate factor loadings were computed for several of these conceivably pertinent case history variables. The inter­ correlations of these variables with the test items proper did 17 not, however, contribute to the location of the group centroids or the derived factors. The purpose of this technique was to provide a further aid in the general interpretation of the ob­ tained factors, especially with regard to generalizations other populations. Further, to since such factor loadings d e m o n ­ strate the relative importance of various case history v a r i ­ ables, pre-conceptions regarding necessary. their importance were u n ­ It is perhaps more meaningful to allow them free range rather than to assume their significance and thereby to control them without evidence that they are important d e ­ terminants in the behavior under consideration. Such as­ sumptions may severely curtail variability and impose seri­ ous limitations upon general interpretation. An attempt was made to select only those cases who d e m o n ­ strated active symptoms of schizophrenia, situation at times failed even though the test to elicit evidences of such symptoms. The entire sample of one hundred schizophrenics were so d i a g ­ nosed by the psychiatric staff at Beatty Memorial Hospital. Not all cases used had been formally staffed and diagnosed schizophrenic at the time of this study, but in all cases the Investigator agreed with the preliminary psychiatric diagnosis of schizophrenia. The group was composed of transfer patients from another institution as well as direct commitments, whom resided in northern all of Indiana. A total of 6 l males and 39 females were Included in the sample, which approximately represents the sex distribution .. 18 in the hospital population drawn upon. While the aforementioned statistical control of sex difference will indicate to what d e ­ gree sex is associated with each factor, a significant loading for sex on any factor will raise the question as to whether the sex difference is an inherent one or a situational re­ action to the sex of the investigator. That such a question could not be answered is not considered a weakness of the present study, but a problem beyond its scope. Table 1 presents the schizophrenic subtype diagnoses by sex, and it is to be noted that the distribution of subtypes between the sexes is somewhat disproportionate. Since there are approximately two-thirds as many females as males, it would be desirable to have that relationship hold in each subtype category. This observation applies more to the Catatonic and Hebephrenic subtypes, distributed more proportionately. avoid since the Paranoids are No attempt was made to the selection of those patients in the Mixed category, and the small number merely reflects the hospital's policy of classifying according is a concomitance A wide to predominant symptomatology where there of symptoms range of ages, as for reasons already discussed. from different subtypes. seen in Table 2, was employed The median age for females was greater than that for males, which again reflects population drawn upon. the fact the The reason for this may be related to that mental disorders become obvious in males earlier in life as a result of their more active participation in the 19 TABLE 1 PSYCHIATRIC SUBTYPE DIAGNOSES BY SEX Subtype Male Female Total Paranoid 2 >+ 19 *+3 Catatonic Sb- 8 32 Hebephreni c 6 9 15 Simple p 0 if Mixed 0 3 3 Unclassified 3 0 3 20 TABLE 2 AGE OF SAMPLE BY SEX Age Males Females Total m~i8 3 0 3 19-23 5 1 6 24— 28 13 3 16 29-33 13 4- 17 34— 38 8 7 15 39-4-3 i+ 9 13 4-4-4-8 4- 6 10 ±9-53 3 4- 7 54- 5 8 4- 2 6 59-63 2 2 4- 64— 68 2 1 3 61 39 Po t a l 100 21 environment, and curity, their necessity for attaining economic se­ A pre-schizophrenic female, or even one with a chronic schizophrenic process may adjust in a limited fashion for years before an acute reaction to stress necessitates hospitalization. The durations of illness for the sample is reported in Table 3- These figures do not represent merely the duration of hospitalization, but the amount of time intervening be­ tween the first outbreak of a schizophrenic psychosis and the present, in so far as it could be determined from the avail­ able anamnestic data. Despite the fact that the median number of years of illness is greater for females than that Tor males, port. the above discussion concerning age receives sup­ When the median duration of illness for each sex is subtracted from the median age for that respective sex, females appear to have been older at age of onset. findings are reported by Haas Health Service (2*+), and Similar the U. S. Public (V5)? although the latter is based on the rather selected populations of 2 2 5 private mental hospitals. The extremely large samples, however, lend significance to the trends reported in these and other sources. Table b shows the years of education for the sample by sex. No important differences seem to exist between the sexes, and in general a wide range was obtained. A marked sex difference in marital status is apparent in Table 5 . This discrepancy, i.e. the predominance of single 22 TABLE 3 DURATION OF ILLNESS OF SAMPLE BY SEX Years Male Female Total < 1 13 2 15 l-*f 17 7 2k- 5-8 13 9 22 9-12 7 8 15 13-16 2 8 10 1 5 17-20 > 20 Total 9 5 61 39 100 23 TABLE *+ EDUCATION OF SAMPLE BY SEX Years Male Female Total < 8 7 2 9 8 19 10 29 9 2 10 3 2 5 11 7 2 9 12 17 l*f 31 >12 6 5 11 39 100 Total 61 6 2b TABLE 5 MARITAL STATUS OF SAMPLE BY SEX Status Male Female Total Single b2 7 »+9 Married 13 18 31 Separated 5 13 18 Widowed 1 1 o 61 39 100 Total 25 males, is not easily explained, but one might hypothesize that the classical schizophrenic syndrome, including withdrawal and apathy is more foreign to the male role in our culture. A schizoid male would be ill-equipped to compete with his more aggressive brothers, to obtain. and he may find a mate rather difficult Most likely he does not want a mate for the same reason that he is indifferent toward others regardless of sex. On the other hand, the passive and withdrawing nature of schiz­ oid symptoms may enhance the female role in our culture, and such a female would be a desirable marital partner for a male whose aggressive needs require fulfillment. Lastly, no exclusion was made on the basis of race, and the six Negroes included approximate the distribution of Negroes in the population drawn upon. This is a weakness of the study in that the small number of Negroes prevents any reliable statistical analysis of them in comparison to nonNegroes. Their contribution to the total factor structure, regardless of deviation due to cultural background, would, however, be negligible. Table 6 , Appendix A shows a comparison of the 100 schizo­ phrenics and the ^ 0 non-schizophrenics with respect to the case history variables and cooperativeness ratings. The ab­ sent category, duration of illness, was meaningless for nonschizophrenics. This group was composed of hospitalized jects who were diagnosed as character disorders, sub­ or some form thereof, and it was not possible, nor practicable, to fix the 26 date or onset due to the insidious nature of the disorders. All of the subjects in the non-schizophrenic group were Ameri­ can born, n o n - p s y c h o t i c , non-organical ly impaired, non-toxic, and were not receiving physiological type. As regards sex distribution, therapies of a convulsive it will be noted that too few females were included in the non-schizophrenic group, w hich may impair the validity of the interpretations of com­ parisons w i t h schizophrenics. This discrepancy resulted from the non-av a i l a b i l i t y of non-schizophrenic females, since females who are not psychotic are frequently kept at home by relatives. Further, such disorders in females may go u n ­ noticed and be less disabling because of their lesser need to earn a living or adjust to the competition of the working world. rhe qualifications upon interpretation made necessary by this and other differences between the groups will be con­ sidered later in connection with the obtained factor structure. The Test Scale The general purpose in the construction of the test scale was to devise items which would divide the group equally with respect to presence or absence of the various behavioral re­ actions. Included in the test scale (see Appendix B) are certain symptom manifestations w h ich are prominent in the classical clinical syndromes of schizophrenia, as well as some items of behavior w h ich are less formally treated. Among these latter responses are included reactions to problem­ 27 solving tasks of a social nature, and some which deal with the schizophrenic *s response to some social act on the part of the examiner. Bales* Particularly useful in the selection of items was (1) categorization of the interaction process. his system is designed group situation, Since to record spontaneous interaction in a the categories served to suggest items, than lend them as such. rather To insure a standardized interview situation, with good continuity, requires a careful selection of items that will, or can be made to occur quite naturally in the course of the interview. Of further use was Lewis* examinations, (31) manual for psychiatric particularly with regard to suggestions for observing appearance, posture, and movement, as well as other general Many items were designed by the investi­ suggestions. gator and arose out of general clinical experience. At this point a legitimate question arises as to the general hypotheses the investigator had in mind in selecting from his clinical experience those bits of behavior which he chose to include. To answer this question it is necessary to describe the method for devising items. First of all, the domain or gross con­ text from which variables could be abstracted for study con­ sisted of the sequence of behavioral events occurring when a hospitalized schizophrenic is summoned from his ward, met by a representative of the psychiatric staff, and interviewed by this representative for a short period of time. A further limitation or dimension of this general domain of behavior was 28 the cross-sectional aspect which prevented longitudinal study of the behavior under investigation. defined, With the situation thus it was necessary then to list all possible behavioral acts which were both subject to measurement and typical of the interview situation. To attempt to rule out the possibility that the investi­ gator was biased in his listing of possible items for study, several other competent clinical psychologists were asked to contribute to the list. Thus, the final test scale, rather than flowing out of specific hypotheses, consisted of a re­ tained assortment of behavioral items which were not excluded previously because they seemed to be measurable, not too foreign to a typical interview situation, and not too depend­ ent upon interpretive judgment for scoring. Some bias in the selection of items cannot be avoided but it was felt that the selection technique employed held the operation of such in­ fluences to a minimum. Certain items were included which arise out of more formal test material. The Kohs blocks were employed, not to measure the ability to build a pattern as is conventional as a vehicle for interactive behavior. (2 9 ), but The use of these and other such items represented an attempt to study quantitatively that which is often used qualitatively in the interpretation of formal test results. Certain items attempted to gauge the desire and ability of the subject to imitate the examiner. In one test (Bead Stringing), which was adapted from the 29 Stanf ord-Binet Scale (^2) the subject was asked if he would like to try it, without the suggestion that he imitate the examiner. The subject was then free to imitate the examiner in varying degrees. In another task he was asked to imitate the examiner, first in a gross fashion (Task Sequence A), and then in a more minute, abstract way (Task Sequence 3). tasks may involve a number of abilities, memory, social acuity, designed to partial Such including attention, etc., and the treatment of data was these out by examination of clusters of i terns. The Verbal Fluency test was adapted from the verbal fluen­ cy factor developed by Johnson and Reynolds (26). The vocabu­ lary test which was also cited by these investigators as a good measure of verbal fluency was not used. was devised that purports Instead, a test to measure the i n d i v i d u a l s value judgment of words in terms of their social desirability. This test, then, undoubtedly measured something different from, or in addition to, likely, the usual vocabulary test. Most it measured a more restricted area than that of the general vocabulary test— an area of socially oriented word facility. To measure a conceivably important variable in social behavior a test of attention was devised, tention factor derived by Wittenborn to simplify Wittenborn's sample, but despite (*+6 ). based on the at­ It was necessary test so as to adapt it to the present this modification and despite the faet 30 that other tests were loaded on his attention factor to a les ser degree, it was felt that attention was measured fairly adequately. Since many other items in the scale used a p ­ parently involve attention it was anticipated that informa­ tion would be obtained as to its importance in general behavior. social 31 Procedure Each subject was seen individually in the same room, and the situation presented him, consisting of a planned interview and testing, was standard for the entire sample. The test items were administered in approximately the same order as listed. Conversation interspersed among items was, so far as possible, at maintaining test of an innocuous nature, aimed continuity of the relationship. conversation of the subject, learning the various especially Spontaneous that directed toward the nature of the interview, was handled either non- directively or responded to with remarks that had a minimal in­ fluence on the test results. The general situation, had sufficient warmth and permissiveness however, to elicit the best possible cooperation and social spontaneity. It was deemed important to standardize not only the pro­ cedure and behavioral sequences of the investigator, but also the physical aspects of the office used for interviewing. this purpose a check list was employed, object, including For and each pertinent the desk and chairs, was placed in the same position prior to each interview. The rating of each subject was based on that which was most typical for, or representative of the patient for the period studied, unless the rating was based on a response to a specific stimulus. An attempt was made to avoid inferences from the subject*s diagnostic classification or history, and the ratings were based on the actual responses during the 32 period studied. To avoid correlating the investigator *s bias as to w h ic h items were related, an attempt was first of all made to define each item carefully and to clearly differenti­ ate it from the others. To check on such biases which are bound to exist, however, the degree of inter-rater agreement was computed. The investigator was fortunate to have available the services of a competent clinical psychologist with special­ ized experience in behavioral observation and rating pro­ cedure. These two functioned as an observer team in the simul taneous ratings of test items with schizophrenic patients selected randomly from the same population as the original sample. The complete procedure consisted of, first, a ses­ sion devoted to definition of test items, and the special problems created by two observers. It was necessary to have the observer in the testing room, despite the fact that a one­ way screen and monitoring device were available. Many of the items measure responses so subtle that an observer outside the room would be unable to observe them adequately. The in­ vestigator noted no special reactions to the other observer, and the situation was so structured that the observer was re­ moved from the immediate test setting and out of the direct line of vision of the subject. The observer acted the role of attendant in that he procured patients from the wards and delivered them to the investigator. From this point on he played an impassive role, and in no case was communication between him and the subject. there further 33 The entire scale of test items was subject to such dual ratings except for item numbers 3 and 1 5 , *strength of grip in handshake* aminer*. and ’number of seconds patient stares at ex­ These items, although involving some slight judgment on the part of the investigator, must remain untested for bias. Agreement between observers was computed in terms of dichotomous scoring of test items. Of a total of 193 test items so observed the percentage of agreement was 9 1 .2 , with a standard error of 2 .QU-%. This type of data was well suited for the computation of a phi coefficient by way of its in­ trinsically dichotomous nature. lation so calculated was The coefficient of corre­ .8 2 , which is equivalent to the highly significant chi square value of lll.O^. Treatment of Data The test scale and correlation s. The test items in the scale consisted of from two to five response categories which were designed to facilitate scoring. The measurements are at best very gross, and in a relatively unexplored area such as this good scaling procedure is all but impossible. The re­ sponse categories for each item might be said to represent verbal descriptions of those regions along the continuum of a response variable which are easily perceived as different from an adjacent or previously defined region. For example, test item three attempts to gauge the strength of the s u b j e c t ’s grip during the handshake. This continuously distributed 3^ response could be measured by an apparatus (which, incidental­ ly, would remove it from the category of a social response) which would assign it some value along a scale of innumerable points. The number of differences in pressure perceptible to the human observer would, however, be far less, and the number of verbally-differentiable points would be even fewer. in the item under consideration, Thus, it is a simple matter to recog­ nize whe n the handshake pressure exceeds and when it is approximately equal the examiner*s pressure to it, and the two response differences become adjacent categories for scoring Because of such rough measurement, the item. there can be no com­ parison either withi n or between items with respect to the numerical score obtained. Comparison can only be made after each item is reduced to dichotomous scoring categories, a form which is suitable for the computation of tetrachoric cor­ relations. Numerical throughout the study ratings on each item were, however, made in order that the reduction to dichotomous values could be based on the distribution of scores for each item. The use of the tetrachoric correlation technique re­ quires that the cutting point be approximately at the point of the median score, and thus the preservation of numerical score became highly desirable. The same applies to the sever­ al tests administered on w hich a fairly wide range of scores was obtained. A further requirement when employing tetrachoric corre­ lations is the assumption of a normal distribution of the 35 variables under consideration. This assumption was more dif­ ficult to make for some items than for others. For example, the score for test item six depended upon which one of two available chairs the subject chose upon entering the testing room. The overt behavior upon which the scoring was based was obviously of a discrete and dichotomous nature, but the assumption of a non-discrete and normal distribution was based on the underlying personality variable responsible for the choice. Were no such variable to exist, would not be possible. the factor analysis This variable of personality would presumably cause the choice of a sitting position, for ex­ ample, somewhere between one inch and twenty-five feet from the examiner. The actual placement of the two chairs repre­ sents neither extreme, but something closer to the mean and, it is hoped, on different sides of the distribution. Due to the relative unreliability of the tetrachoric correlation coefficient, it was decided to exclude from further analysis any test which failed to meet the minimum criterion of 25 percent in either category after dichotomization. failing Of the original 1+6 tests, two were excluded for to meet this criterion (numbers 39 and *+6). other tests (numbers 19 and 2*+) were excluded earlier for reasons of i m p r a c ti c al it y. of M-2 tests. Two The final scale, then, consisted Table 7, Appendix A shows the percent of plus scores for each test, and includes the two tests excluded from further analysis due to poor distribution. After reduction to dichotomous values for each test the data were transferred to McBee cards with cut-outs being made for plus scores. The case history data for wh ich factor load­ ings were desired were also entered on the cards in similar fashion. The sorting of these cards, then, yielded the raw frequencies w h ich w ere entered in the four-fold computation of tetrachoric coefficients. tables for Since the N was 100, these raw frequencies represented proportions w it h no further computation being necessary. The graphic solution for tetra­ choric correlations was adapted since it resulted in consider­ able saving of labor and yields two-place accuracy. made of the Thurstone computing diagrams Use was (16) for this pur­ pose. The f a c t o r i z a t i o n . The factoring method employed was the multiple group centroid as described by Thurstone and Cattell (15). As in any factoring method, (*+*+) there are advantages and disadvantages which must be weighed with re­ spect to the material being factored and the purposes of the factorization. The time-saving feature of the multiple group method derives from the fact that successive residuals are not computed after the extraction of each factor as in the n o n ­ clustering methods. However, there is the danger of having to repeat a large part of the process due to faulty communality estimates or due to errors which become obvious later than in the more mechanical centroid process with its regular checks. 37 The most desirable advantage to the multiple group centroid method is the probability that the obtained oblique or unrotated factor matrix will be close to the simple structure position of the rotated matrix. This state of af­ fairs is more true, however, when clustering is tight and relatively independent so that non-cluster variables will have low or negligible loadings on a given factor and will already constitute a hyperplane. The danger in assuming that the unrotated matrix is close to simple structure arises from the fact that some factors may not appear as clusters in the correlation matrix, or that some clusters may represent more than a single factor. however, The number of such factors should, be in the minority, and it is assumed that the ma­ jority of the factors are extracted in a correctly rotated position. The method for cluster search in the correlation matrix was a conventional one. First, a pivot test was selected which showed the highest number of correlations over .40. A cluster of tests was then assembled which correlated highly with both the test used as a pivot and the other tests in the cluster. The second and succeeding clusters were formed by selecting a new pivot test from those not already in another cluster. To qualify for inclusion in a cluster a test had to show a high average correlation with same-cluster tests and a low average correlation with each group of tests included in the other clusters. This process was continued until no 38 further clusters could be established wh ich would meet the criteria of tightness of clustering and of relative inde­ pendence from other clusters. The average intercorrelations within and between the clusters chosen for final factoring may be seen in Table 12, Appendix A. It was deemed desirable to perform a complete trial factoring prior to the final factoring in order to provide better estimates of communality, as well as some insight in­ to the number of factors operating. This process was begun by extracting two factors simultaneously and then plotting the test vector space with respect to these two oblique factors. While such a geometric plot could not yield in­ sight into the test space beyond two dimensions, it did pro­ vide a clue as to the major clustering of tests. At this point a complete centroid was placed through the correlation matrix in order to give further information as to the test space. With this' preliminary knowledge, the process was be­ gun again by extracting two large oblique factors, the first of which was through the largest clustering of test vectors. A residual matrix was then calculated after rotating two oblique factors to orthogonality. matrix, the first From this residual in which some high correlations remained, four more factors were taken in multiple group fashion. Frior to this last step, however, a complete centroid was put through the residual matrix in order to provide information as to the re­ maining variance to be accounted for. preliminary factorization, On the basis of this six oblique factors were extracted 39 from the original correlation matrix wh ich had, at this point, adjusted communality estimates based on the factors known to exi s t . The multiple group centroid method yields an oblique factor matrix which must be rotated to an orthogonal matrix before residuals can be computed or the amount of communality accounted for can be calculated. diagonal method of factoring or the matrix expressing This is done by applying to the factor correlation matrix, the cosines of the angular separation between the non-orthogonal factors. tained triangular matrix yields The inverse of this ob ­ the rotation matrix w h ich when multiplied by the oblique matrix yields thogonal matrix. the the desired or­ The sum of the squared factor loadings on the orthogonal factors, then, provides a measure of the amount of estimated communality accounted for by each of the oblique factors. hesiduals may be computed by multiplying matrix by itself and subtracting the products from the re­ spective correlations in the original the residual correlations are correlation matrix. toe high, bility of another factor being present. tnerefore, the orthogonal to put a complete centroid there exists If the proba­ It was felt desirable, through the residual matrix to provide an estimate of the largest remaining factor, he amount of communality accounted for by this factor is easily obtained since it is orthogonal with respect to the other factors extracted before the residual calculations. *fO Due to the fact that oblique factors are always corre­ lated to some degree, it is an informative procedure to factor the matrix expressing rations of the factors. the cosines of the angular sepa­ This matrix is equivalent to the intercorrelations of the factors, and communality estimates should be based on the highest correlation in each column to avoid working with specific factors or error factors which would arise were the communalities expressed as unity. This procedure aids in the understanding of the derived factors and may result in concepts which are more parsimonious for purposes of description. expected These nigher-order factors would be to be more functionally independent than the lower order factors developed directly by the study. It is gener­ ally true that the higher the order of the factor system the lower the general correlations will be among those factors. Conversely, high correlations among single variables can oc­ cur because they can mutually share the effect of the same factors to a great degree. Factor loading i n d i c e s . While the factors derived by this study represent concepts and not persons or types of persons, a technique is available for considering the re­ lationship between the factors and persons who have received scores on the tests which comprise the factors. This is similar to the procedure for calculating factor loadings for a particular case history variable such as age, diagnosis of paranoid, for example, sex, or a except that separate scores for each individual may be obtained. For any given individual the procedure consists of summing the squared factor loadings for those tests which are significantly loaded on a particular factor and for which this individual receives a plus score. In this study a significant factor loading was considered to be AO or greater. The difference between this technique and that for obtaining factor loadings for case history variables, aside from the fact that the former per­ tains to individuals, is that factor loading indices are based on all tests which show a significant loading on a factor, including non-cluster tests. Since, then, each indi­ vidual is assigned a factor loading index for each factor, various criterion groups may be selected out for study. Or, it may be that the investigator would find it desirable to proceed from the opposite end, i.e. selecting extreme groups with respect to factor loading indices on a given factor for the purpose of investigating possible differentia. of this technique in this study, however, The use is limited to spe­ cific purposes which are within the defined scope of the study. Factor loading indices were computed for schizophrenic subtype groups Mixed, (excluding the total of ten subjects in the Unclassified and Simple groups), and comparisons be­ tween means were made among these. The operations involved the calculation of the standard error of the mean for each group and the standard error of the difference between the means in each comparison. The t. value was obtained by *+2 dividing the latter into the obtained difference between the means involved in the operation. The significances of the £ values were obtained from Guilford (22). Since the mean factor loading index for any factor was dependent upon the magnitude of the sum of the squared factor loadings, the mean index was divided by the sum of the squared factor loadings for all significantly loaded tests. This done, the mean factor indices became comparable for all factors, regardless of the number of tests involved or the magnitude of their loadings. Included in the above analysis among subtypes was the non-schizophrenic group, for which mean factor loading in­ dices were computed. It was deemed desirable to consider non-schizophrenics in relationship to each subtype sepa­ rately since there is no reason to believe that all types of schizophrenics will be uniform with respect to social be­ havior, or even as regards any general factors. This method of comparing differences is not only more economical, but it also takes account of and weights items according to their significance in terms of the derived factors. It makes use of the concepts developed by the study, and in a sense tests their usefulness for describing differences among the afore­ mentioned groups. Since it would be desirable to have some notion as to the discriminating power of each test in the scale, a chi square analysis was done which included the entire samples of *+3 schizophrenics and non-schizophrenics. The analysis was aimed at the examination of differences in the distribution of scores for each test between the two groups. Although this technique yields data on the ability of each test to distinguish between schizophrenics and non-schizophrenics, it fails to indicate the meaning of any particular test in terms of its relation­ ship to other facets of behavior. RESULTS Test Scores and Correlations The dichotomization of the raw data was based on the ob­ tained distribution of scores for each item. The cut-off point for assignment of plus-minus values was placed as close­ ly as possible to the point of the median score for each test. Table 7, Appendix A shows the percent of plus scores for each test after the cut-off point was established, and it is obvi­ ous that those tests for which a wide range of scores was ob­ tained were more adequately divided into near-equal halves. Only two tests failed to meet the minimum criterion of 25 per­ cent in the smaller category, namely tests numbers thirty-nine and forty-six. These were excluded from further consideration. The initial comparison between schizophrenics and nonschizophrenics was made with regard to the distribution of test scores. Table 8 reports the results of a chi square a- nalysis of the distributions of scores for these two groups which was designed to test the null hypothesis of no d if ­ ferences between them. Thus the expected cell frequencies were based on the assumption that the plus-minus scores would be distributed proportionately according to the number of sub­ jects in each group. frequencies Due to the fact that some of the cell in each chi square table were less than 50 it was TABLE 8 CHI SQUARE COMPARISON OF TEST SCORE DISTRIBUTIONS F OR 100 SCHIZOPHRENICS AND b O NON-SCHIZOPHRENICS 1C J ^ 1 3 1+ 5 6 7 8 9 10 12 18 lb 15 lb 17 18 Per Cent of Scores Plus 3c h i z . Non-s c h i z . 72 90 83 57 ^3 16 11 50 57 18 71 57 15 98 95 90 83 73 83 95 98 88 75 93 58 bb 85 62 61 11 0 tr S fl •'3 S ' - 20 28 22 ■ 3c b) 3b P, P ) , 0 ■ “T^_y .1' b 27 5 b 28 73 b p 1? • - 8u 31 -33 ■i ' J -J r"•3 s y \ \ U-i-L 2 >q _9. r.'s it 'n J, / 29 < 28 39 10 , 100 iO 15 68 70 72 O 7 ^C s O P ^ w‘ u✓ q 80 100 95 18 68 90 100 93 0 -1 Chi Square .0 3 1 5 .5 9 2 1 .9 3 5 b . 51 1 2 .5 7 9 .8 3 15.5 9 5 .9 8 1 5 .5 9 2 3 .3 7 1 1 .3 6 1 6 .0 1 0 .0 1 l l . 36 1 .7 1 1 8 .7 8 1 0 . 21 2 .9 0 2 2 .7 1 1 1 .1 1 2 9 .9 s 1 .0 5 .01 .01 .01 .01 .01 0.03 21.93 12 .1 1 3 0 .0 9 0 . 00 ■1. vQ _yO /’ 16 .1 9 2 1 .7 8 26.12 8 .1 1 2 8 .1 5 58 1 .x*y'i 19.67 P.c yr (?) 1.01 b’i 0 =r 0 Significance .s -•_5J •Q1 10. l l 1 .1 9 1 8 .7 6 .0 1 .0 2 .01 .01 .0 1 .01 .83 .01 .1 9 .0 1 .0 1 .0 9 .0 1 .01 . 01 .30 .8 6 .0 1 .0 1 .01 .9 9 .1 0 .0 1 .0 1 .0 1 .0 1 .0 1 .0 1 .0 1 .01 . 29 .0 1 TABLE 8 continued. rest hi b2 63 bb b5 bo Per Cent of Scores Plus 3chi z . Non- s c h i z . 57 •25 31 6L 52cr o cO 100 30 r cr yj 100 95 85 Chi Square 21.93 0A2 6.30 16.73 20.28 0.00 Significance (P) .01 .52 .01 .01 .01 .99 b7 necessary to employ Yates* involved a subtraction of values representing correction for c o n t i n u i t y .1 This .5 from each of the discrepancy the differences between observed and ex­ pected cell frequencies. The tests of significance for each chi square value revealed that 32 tests or seventy-six p er­ cent discriminated well between schizophrenics and non-schizo­ phrenics. These 32 chi square values were of such magnitude that they could be expected in 100 times or less. to occur by chance alone only once Thus, as regards these 32 tests, null hypothesis was rejected, and the the assumption may be made that the groups arose from different populations with respect to the test performances involved. Table 9 shows the obtained correlation matrix for the b2 tests employed in the analysis. Four tests of the original *+6 had been previously eliminated due either to impracticality or failure to meet the distribution criterion of 25 percent in the smaller category. Tests 12 and 29 have been reflected in the matrix to avoid negative column suras, and their names were changed so as to agree with the signs of the correlation coef­ ficients in their respective rows and columns. The communality estimates which appear in the diagonals are based on the re­ sults of preliminary factoring and will be explained later. Preliminary factoring made use of communalities equivalent to the highest correlation coefficient in each column. ^Snedecor, G. W . , Statistical Methods. legiate, 1937, p. l6l. Ames, Iowa: Col­ r# fi (4M*i z O fte n j Usa Of fifitM /A n t 0 / e s / w n e r» £■ fitessv** //y #4 HPSfifixe t LottoAt £, SUM- MA*bsM*m / Z 3 1 s 6 7 11 ion ititniejitritW JU i& i& iffltStotoyu& yttsibW tm m K to m a m ? soa n tu s x u -o /m n o t n *1s* u p * m w - tw s w u f n m s i u m t **« tS[tl)fS5d fl Of 19S ttt 57 II - 4 * 4f 577* %U ft 15 t t * It %52 5 H f f t 35 51Hi 5721 31% 0 f 57 00 OJ U vjKtfjaun nun ftftnetititsonnifH slotu sshowu*6tssthiit» tftm simi it% 6\ » r i ^ t f t / I HI U 61 Z W M W S t t S t t M l 17 12 0t% 6/S tH 11 77 J / » 5563 5f ic i t Si tin V» tfi itt v n ltfu ita n n i/ *-*ssu,nussutiH7untwur*iinsi'iinsitwituii-btm 5 SrAres * * * £ PKJtf£*&**. cC ito ilS tii^tt U N # UtH-fbtA 13HU V M tlm ttc fU N b llW # 2 S t5 lliL M H 'tS - l6 i( > 1 t 6 5t TS 0£AA £ • 63» k h n oitf « u « n tt net UH tt is-rnt tt if-tiu it a «*/ u » » is w s M / t it n 4 « 7 R fM cv/S J/rrfiStf’ OSSTACL* ot?61 tits*HIS 6H&ISHU /t'lf 4 HI 5715554 tl%V>n Hio] 66 tl S~ CMA/fi a SI 504 511/ K ffto tu r n t f fe s fitw fis T9 TACA 0 0 £ % litotes Mttrwu. M t A l /£$*>*%£ I 1721nulla* jifrfln*it'lltsuunistuttosototl94 6tuunutt sstiusttrustv n 0 5 7 7357 9 4 4 f / 2 ^ J 7 / / l f * 4 If t7V 67i? f i 5 r * r ‘# a f l T » t t # J 7 # t t i i 4 51 II K ei* * /J* 7t> c ts to fg r r e o reg e tS If 9 » f U 4 U 65 77(0) M 4 HRH i t V 4 U I T t t H Ifit'OSSXlfSI St HU St 37/SSW tS 13 3) U’tt 1331 It tt t1"H lll3 tc (^ H //•/*315 9H f f 61 H 'tl 139 I i6 l3 jjf 351311415 V 3 5/ 354 If % ll C rosses 4 / 6 $ U - 0 Z t f » l W U 417 H I itiftfit-tttl-toDlrn Z1)l'6i'3l/5/f Sifif 63I f 65 2t 66 H 4 111706611 tS/i IS RClAXti * » » / f c s f i w o s ro c m K t t o e t i r 75 77 If 175561111713*5 U lM !(/falff3l SfU SfH tSIS #51 i f * 26513* 31V W 3**1913 H c l a n If IS Covf/p/n £ s srfiic \W7 * U U Hi I* niA IQH’$ jnit% )0Zlt 1955'lfif O llflfilH 'd -n M lS M I 1413 0013bX IS 6Z 22H U I t56 St 1161 631317u 6i t m m t f p i v s i 17U 'f t t i n StPm\H0331ltt2iS527% 0l 2ilfC72!fi(t 4 F o u o rV s G-ask oe£. 17 R s s W X r§ # s At/ £ / fieHA/K if H 3i h u ii oe 15m i h u n i n f u & p xm tu m a m p h to n 4st u hi 11n 55» 4 u & It h r te n ts TS AHtni COHteiiATttH 27 U %53551113 66 355717ItH S !» 5 S W fr HlXM H5S5W.104S15I & I S % 5 7 51* iSZfiil 0714 3511WIS-mi7l1H*1lSt33n3055bi)6HlH1i U6S7g‘M5fO it 65U 55S U M it U l t V U 10 f e i f i s £■ riN P M A rtn R e s f i t t f t s to MtnUL&TMto 4*1 nett 5733/2 31m i6li5H >71455119'67,lft1U t fW j 9t)f %11151774 VUH SU M iS-t 11 ll M« KoCOMOTOK /£TAAMTtcH IX CoNTMts sropy Afire*mftiprm \5u w i n i u 6ii\X6HHiininnawi$to$uv%HH6t%MKoiMnibiUifi 23 fofi***s AOpvt AA$y n c re * * 2f £aP**SS*S fiCdJTKAL OPOfMM Sin0%/iU» i5so9 559M 5M mill 004 (0toSOU4 3M 1M Ofp m ton3ti30fOh7i$MifrOMt-imisHtoUli tomMsmuuioiZixiowoU/SiSsi-m t4%3 S6U 4 to52i7ll//X/5*/7flS15StotS20llll($3(3tn/i36to531A3llittlltS3!/3ii2i 27 Hf.Lfis m 251 R e r fiifv e s p / p / p / p fit/tc /i £ , ffC A U SAME % il U H * w a n W im % * t l 5 l t 5 n t i W l W T ik s i i 6 U 5 0 S > 5 t S ? 3 5 7 7 / W » 11 ts fiS/U/HP TS A rrem n toffiKMT THi 27 fe r e tr s £>s FtLfi H3533Stutt4 i%4 H-H4%itl6ttt7illf06UtSSiati(f4%Cin5H 5t7llSIII1711171921il naunnoi Ptim-iom H-ohinn tttnpisniWHtfis-istsn'iasisim302701/on 30 RfACTT TO CttrtCiSM 31 If*>TA7£s £i seAp sTtoupme32 fanes Peasms /A re m re £ 35 Re-EXACTS TASK S B H W ti(H m s) 3V fa-EKAtrs rAftc 6eQ0t*tf(HWTs) 35 O oes A rre M m n U 1h HU 17 61 21 M U 2 | 5X31 132013 164 M n » 6 )S fn 4 4 0 flS (l^ $ 3 2 S ff1 t6 S 3 l 16133X22113/ » 'JetBAUi TLUSMT 57 Goes s o o a l faCA&*AKy 30 P e r e c ts P o l l P kh c/l 70| H e a t P e r s e * a l A P fieA /4 * cs. V/ feerenee op fiu tt Acpviataw HJ\ CM /bvs A f m TAH. VHP** B os f f M t H B f / S £ .\ AAM £ (jHoeesrAHts struAno* #5 G ives PAPriho. i t s P M s e 5Hssnnsniuii5uiti5iinawnsiit not ittiniiwmsi soastsim isnto*56 H S t $111 nu*><7361151 M W f m f i f c f l 4 m S H 4 U 6651133lSl(j^7Sll51561031W H I f H atommoummaiiunnuiiseesintptimii’ftnsnsmiisnnsisnisnsi H51115551U 1595to552646631311*511*143/<4tnt/SOlMHit1\%(f05l67VZ5%246t$$Sf mi 3665S3mi 4 U 31U liHWHSSSI <43511254 *956* tSUSI515132(5^661173t15-05fl66 1tVI3t5117’lt/5ilWltltIf4 3313till5JHHit336216515105If5150516766(60)6141326IIII61 mmitsii6w i t 3iimi%o6i6iii6teiimiio%nn6ii\s6irun uHolsbsttotibimo to0116tm 1W 303133I5H #7ml toH 6fH55Z362flt1llUVf'ttO-HWttn * ^Siusolt* 33 nustsib Hit2111165151002562Z63f55If065151/5X57111301355666556301566SftyUll21fl 62313536U15It11U 11U 55011615/1/0H21554 H’IS35It21tilt21133126(5241150S%S^610510 36002111-OflO0001SfU2500II01tlOf/615ItISft555113OlUOltXOO’IO156f% 114 20/V'##7/7 ’463171026161600571063MS 03tZ21U ftIVH2155'ft1651ft10UU It1633H HI1316K 0321(11)53 36UHH111053X1115557Hlt%-6510506llS1l15ltl1tHirklM\S6l3595t606llOl56610lltslit 17HI511651Of673711$ tt11-1156IfU1t7&U3Qft26U 36ItttIO353151S356151016%1fSI’lt061l( *+9 Preliminary Factorization Factoring was begun by locating two clusters by methods previously described, and extracting two oblique factors based on these clusters. At this point no attention was directed toward the content meaning of the factors in terms of the clusters of tests. A graphic representation of the test space was then made with respect to the two oblique factors, in order that more perspective might be had for the entire problem. limited, While such a two-dimensional representation was it nevertheless seemed quite clear that the entire test space was rather circumscribed, and that perhaps a large group factor or even a general factor existed. To aid in the estimation of the magnitude of such a first factor, a complete centroid was placed through the original test correlations, and the loadings on this factor are presented in fable 10. factoring was then begun anew by repeating the process of extracting two oblique factors in multiple-group fashion. The first of these was close to the location of the complete centroid, but both were based on clusters that seemed to have apparent meaning. Further clusters were difficult to dis­ tinguish in the original correlation matrix due to the domi­ nating effect of these two rather large group factors, and so it was decided to calculate a residual matrix. This would re­ move the effects of these two factors and would allow the ef­ fects of additional factors to appear more sharply. Since the first two factor reference axes were not orthogonal with respect 50 TABLE 10 LOADINGS ON THE PRELIMINARY FACTOR EXTRACTED BY A COMPLETE CEL l'ROID THROUGH CORRELATION MATRIX Mo. 32 37 27 33 66 G ll 60 3 IS 35 36 2 10 1 5 lb 28 26 •'+5 17 31 8 16 bl 23 21 9 22 12 30 ll 38 53 20 7 62/ to Test ---------------------------------Content Varies reading rate with S. Good social vocabulary Looks at S. during handshake Helps 2. recall name Re-enacts task sequence (gross) Understands situation Re-enacts task sequence (minute) Neat personal appearance Uses pressure in handshake Attends to phone conversation Good attention Verbally fluent Offers hand in response to F.. No locomotor retardation Responds to E.'s change of Dace States name preference Responds to compliment Willing to attempt difficult task Expresses political opinion Gives parting response Responds to weather remark Imitates E.'s bead stringing Responds to tack on E.'s chair Follows gaze of E. Presence of mood modulation Remarks about baby picture Responds to interrupting question Makes initial verbal response Continues story after interruption Crosses legs Reacts to criticism Responds to cigarette offer Rejects dull pencil Remembers E.'s name Helps S. find match Removes sitting obstacle Curious about tail under box Sits near E. Loading .80 .79 .77 .77 .77 .77 .70 .69 .67 .65 .65 .65 .66 .66 .63 .61 .60 .59 .56 .56 .55 .55 .56 .51 .50 .69 .68 .66 .66 .63 .62 .61 .61 .38 .32 .26 .26 .26 51 TABLE 10 continued. Test ----------------------------------------------No. Content 25 15 29 13 Retrieves dropped pencil Confronts E . ’s stare Rejects E.'s nelp Relaxes body Loading .21 .17 .11 .06 52 to one another, it was necessary to rotate them to orthogo­ nality before calculating the residuals. This was done by multiplying by the inverse of the matrix derived by diagonal­ ly factoring the matrix expressing the cosines of the angular separation of the oblique factors. The orthogonal matrix so derived was then simply multiplied by its transpose, which yielded the product matrix which was subtracted from the original correlation matrix. The residual matrix so obtained revealed the presence of significant correlations, factors. and thus the presence of further After reflecting the many negative column sums, a complete centroid was extracted from the matrix, which pro­ vided information as to the largest remaining factor. This factor accounted for 10 percent of the estimated communality of the original matrix and 20 percent of the new estimated communality of the residual matrix. Subsequently four ob­ lique factors were extracted from the residual matrix in multiple group fashion. Since the first of these was already in orthogonal reference to the factors derived before the calculation of the residual matrix, it was necessary to in­ clude only the last four factors in the process of rotation to orthogonality. The cluster search in the residual matrix was complicated by what appeared to be doublet factors based on the inter­ relationships of only two tests. Since the tetrachoric corre­ lation coefficient may be unreliable, these relationships were 53 Ignored so far as locating a new factor was concerned. How­ ever, if a third test could be incorporated into the cluster without causing the loss of too much in the way of tightness of clustering and relative independence from other clusters, a cluster of three tests was formed. This occurred in all of the remaining clusters each of which was composed of three tests. Cattell (15) warns the investigator of the pitfalls to be encountered as a result of the location of centroids on the basis of small clusters. The danger lies in the fact that the clustering may be due to error variance which is carried through to the factor space with no way available to discover its nature. hered to. To guard against this, several criteria were ad­ In the first place, it became more obvious after prolonged cluster search that groups of no more than three tests would make acceptable clusters mathematically, regard­ less of meaning. Only those clusters were accepted which not only were important in terms of variance accounted for, but also in the meaning which was interpreted from the clustering. Further, an attempt was made to work with those tests which showed the higher residual communalities, and whose original estimates of communality arose from the correlations with tests included in the clusters being formed. The clusters were determined in decreasing order of im­ portance in terms of the magnitude of the average intercorre­ lations, and it appeared that beyond the four clusters decided upon little of a worthwhile nature remained. Further cluster ,4 5V extractions would most probably have resulted in tests with communalities-accounted-for in excess of unity, and thus factors probably based on correlated errors. Such spurious factors based on error will probably be small in relation to real factors and thus it is advisable to regard less important clusters with suspicion, especially if they have little mean­ ing in terms of experience. Final Factorization One of the more advantageous products to be harvested from the preliminary factorization process was the informa­ tion concerning each t e s t ’s communality, or the amount of variance it was sharing with other tests due to common factors. At this point it was possible to re-estimate com­ munality by summing the squared factor loadings for each test in the orthogonal matrix. This resulted in the increase of some estimates of communality over the original estimates, while others were reduced, and many remained unchanged. The communality estimates of ten tests were markedly reduced (ranging from .05 to .1 8 ) due to the fact that preliminary factorization failed to reveal common variance to the extent suggested by original communality estimates. Most of these tests had significant relationships with very few other tests, or as little as one other test. In such an exploratory project as this, very small group factors, and specific factors had to be ignored. Further, the danger was present that correlated 55 errors were manifesting themselves. Thus, in Table 9 the diagonal entries are based on this re-estimation process which, it is felt, resulted in more accurate communalities, better designed for the specific purposes of this investi­ gation. Table 11, Appendix A shows the specific test clusters which were employed to locate the factor axes in the final factorization. At this point the names for the factors had not yet been devised, but they appear in this table for pur­ poses of convenience. The clusters are those which were identified in the preliminary factorization. In Table 12, Appendix A are seen the average intercorrelations within and between clusters used in the final factorization. The rather circumscribed test space has been previously alluded to and it is obvious in this table that the clusters, in some cases, are highly related to one another. ihe diagonals of this matrix express the average intercorrelations among withincluster tests, and provide a measure of the tightness of the clustering. It is desirable to locate clusters whose average correlations within the cluster are twice as great as the average correlations with other clusters, this was not possible. but in this problem This must qualify the interpretations made, and in itself can be interpreted, as will be seen later. The test loadings on the six oblique factors extracted in rnultiple-group fashion are arranged in clusters for convenient reference in Table 13. In this oblique matrix the interre- 56 TABLE 13 TEST LOADINGS ON SIX OBLIQUE FACTORS Tests No. 37 33 32 66 36 2? 6 36 35 60 31 17 11 63 7 1 10 8 hi Ih 65 23 Factors Name D B Good social vocabulary Re-enacts task sequence (gross) Varies reading rate with E. Understands situation Re-enacts task sequence (minute) Helps E. recall name Looks at E during handshake Verbally fluent Good attention Neat personal appearance Imitates E*s bead stringing Responds to weather remark Responds to cigarette offer Remembers S fs name Removes sitting obstacle .86 .80 .80 .80 .77 .76 .76 .73 .73 .72 .68 .50 .66 .6l.28 Responds to E*s change of pace Offers hand in response to E. No locomotor retardation Responds to tack on E's chair Presence of mood modulation Responds to compliment Gives parting response Remarks about baby picture .61 .55 .59 .56 .36 .52 .66 .62 .69 .26 .58 .31 .68 27 .29 . 53 .58 .21 .59 .16 .61 .31 .62 .33 .35 .23 .35 .36 .69 .23 .29 .63 .26 .05 -.01 18 .10 06 . . . .91 .85 .77 .66 .65 .59 .58 .52 .12-.02 .06 .12 .51 .38 .36 .36 .28 .29 .06 .00 12 21 13 30 Crosses legs .35 .18 .08 Responds to interrupting question.37 .37 «6l Relaxes body .16 .00-.12 Reacts to criticism .35 *29 .26 .60 .62 .65 *66 .59 .37 .68 .56 .68 .26 .28 .28 .13 .07 .09 .16 .27 .75-.02 .67 .18 .59 .20 .39 .07 .39 .36 .22 .01 9 38 29 62 28 .66 .51 .66 .53 .61 .57 .67 .50 .22 .61 .30 .50 .62 .29 .23 .15 .26 .17 .03 .03 .05 .35 .18 .13 .16 .11 .36 .26 Retrieves dropped pencil Curious about tail under box Attempts difficult task Makes initial verbal response Rejects dull pencil Rejects E's help 25 E .80 .72 .56 .38 .36 .20 A 3 .3^ .66 .25 .66 .20 .38 .35 .20 .22 .36 .31 .21 .22 .66 .72 .39 .69 .39 .58 .62 .16 .12 .32 .29 .36 .11 .69 .26 .18 .15.19.65 .25 .17 .16- .02 .39 .26 .08 .36- .09 .16 .29 06 29 29 •37 .05 .08 if 57 TABLE 13 continued. Tests i'iO • 26 3 Name Factors A B C D F A 6 .28 .37 .62 .22 .2 5 18 20 Expresses political opinion Uses pressure in handshake Follows gaze of E. Attends to phone conversation Helps E. find match hi 55 h5 65 2b 15 6 5 22 Confronts E*s stare Sits near E, States name preference Continues after interruption 03 . 0 9 . 0 2 - - . 2 5 19 . 0 5 . 0 2 . 0 3 6 1 .hi •15 . 1 8 3h . 1 9 • 3h . 2 0 16 S •37 . 7 7 .h2 . 1 9 •75 ,b2 . 1 7 . 2 3 •73 . 2 2 .h8 . 3 8 . 6 7 . 2 5 . 3 1 - . 0 8 •hi .h6 .20 .09 .63 .08 .62 .52 . 6 1 .2 0 • 51 A 58 lationships may be seen in Table l1*-, Appendix A which gives the cosines of the angular separation among the factors. These values are equivalent to the correlations among the factors. Since this matrix is symmetrical it may be treated as a triangular matrix which simplifies the process of diago­ nal factoring. The inverse of the resultant matrix, after the diagonal factorization of the oblique factor correlations, the rotation matrix which is seen in Table 15. is Through matrix multiplication of the oblique and rotation matrices, the or­ thogonal factor matrix was obtained. Table 16 which presents the test loadings when the factor axes are rotated to orthogonality also shows, column, in the last the amount of communality accounted for in each test by the six factors. A consideration of these values is a guide in the determination of the adequacy of the factoring. It is, however, more convenient to gauge this from an exami­ nation of the amount of communality accounted for by the factors. Table 17, Appendix A shows the amount of estimated communality accounted for by the factors, both in trial factoring in which the highest column sums were employed, and in the final factoring which made use of adjusted communality estimates. An indication is also gained as to the relative importance of each factor in terms of the amount of test variance it influences. The sum of the original communality estimates was 25.80, while that for tne adjusted estimates was 23.73* Since the 59 TABLE 15 ORTHOGONAL ROTATION MATRIX FOR THE SIX OBLIQUE FACTORS A B c D E F A 1.000 -.909 -.338 -.>+85 -.330 -.616 3 0 1.351 -.028 ,0>+l -. 652 .168 C 0 0 1.062 - •0>+6 -.280 .115 D 0 0 0 1.107 -.157 .3^1 E 0 0 0 0 1 .>+80 -.335 F 0 0 0 0 0 1.188 60 TABLE 16 TEST LOADINGS ON SIX ORTHOGONAL FACTORS Factors Name • o' Te s t s I II III IV n V VI Good social vocabulary Re-enacts task seq. (gr.) Varies read, rate with E. Understands situation Re-enacts task seq. (min.) Helps E. recall name Looks at E. during handsh. Verbally fluent Good attention Neat personal appearance Imitates E . ’s bead str. Res. to weather remark Res. to cigarette offer Remembers E . ’s name Removes sitting obstacle .89- .10- .02 .11 .01 .06 .80 .06 .09- .11- .01 .11 .80 .19 .00 .08 .00 .19 .80- •39- .28 .12 .06 .16 .77 .08- .05- .12- .12 .29 .76 .11- .10- .02 .01 .02 .76 .13 .06 .17 .02 .22 .73- .10 .09- .06- .12- .19 .73- .19- .01- .09 .06- .01 •72- .18 .13- .09 .07- .10 .68 .01+ .00- .18 .05- .18 .50- .06 .28- .17 .08 .10 .99- .05- .10- .11- .05 .00 .91- .39 .05- .05 .33 .06 .28- .12- .05 .17 .10 .13 Responds to E . ’s pace Offers hand in res. to E. No locomotor retardation Res. to tack on E.'s chair Presence of mood modulation Res. to compliment Gives parting response Remarks about baby picture .61 .55 .59 . 56 .39.52 .96 .92 .68 .21 .09- .31.65 .03- .08 .29 .50 .03- .08- .21•38- .18- .21 .10 .57- .08 .2 5- .03 .32 .00- .09 .23 .37 .00- .08 .03 .32 .23 .09- .91 ^5 1+2 28 9 98 29 Retrieves dropped pencil Curious about tail in box Attempts difficult task Makes Initial verbal res. Rejects dull pencil Rejects E.'s help .12.09 .51 .36 .28 .09- .11+ .13 .05 .13 .li+ .09- 12 21 13 30 rosses legs es. to interrupting que s t Relaxes body Reacts to criticism .35- .07-■.09 .37 .16 .30 .16- .15--.18 .35 .07 .15 37 33 32 99 3927 36 35 90 31 17 11 93 7 10 3 M ± ‘- r \ rr '• y >0 7- ? .73 .67 .70 .88 .71 .60 .63 .59 .57 .58 .53 .38 .22 .9-9 .15 .08 .02 .07 .09 .02 .23 .09 .13 .98 .79 .65 . 55 .51 .99 .36 .52 .01-•.07 .03 .00 .05 .08 .09 .16 .19 .92 .16--.11 .63 .52 .97 .27 .96 .09 .72 .20 .07 .61- .21 .23 .53 •"\rg .01--.30 —i— •*■— J .10- •< .69 .72 A5 .27 .76- .12.69 .15.91+- .09 .28- .10.31 .23.22- .02 61 TABLE 16 continued. Tests No. Factors Name I II III IV h2 V VI 16 13 ^0 Expresses political opinion .6-1 .25 .15 .22 .57 lb Uses pressure in handshake .55 .3*+ .26--.05 .38-•.11 follows gaze of 2. .*+5 .16 .02 .05 .57-■.06 Attends to phone conv. .65-•29 .28 .09 .66 .16 Helps E. find match .26 .12 •26— .21 .38-•.02 .65 .63 . 56 .81 .32 15 6 5 22 Confronts E.'s stare Sits near E. States name preference Continues after interrup. .50 .6-3 .5^ .6-0 26 “S . .03 .09 .01-.29 .10 .19-.11- .06-.06 .01 .61 .00- .06-.09 .23 .36-.05 .26- .05-. 07 .63 .61 .32 .67 62 total variance of the correlation matrix was ^2.00, these figures constitute sixty-one percent and fifty-seven percent of the total variance, respectively. The decrease in the amount of total variance under consideration was due to the adjustment of communality estimates so as to exclude con­ sideration of very small group factors and specific factors. Thus, while the six factors accounted for 95*7 percent of the re-estimated communality, they accounted for only of the total variance of the correlations. percent The remainder of the variance was due largely to the operation of specific factors which would not be disclosed in this analysis. The operation of error, however, would considerably reduce the number of meaningful specific factors. The further purpose of placing the factors in a position of functional independence with respect to one another was to allow for calculation of residuals. When the orthogonal matrix was multiplied by its transpose (rows and columns transposed) the product matrix was obtained. These test loading products were then subtracted from their respective positions in the correlation matrix to yield the residual matrix seen in Table 18, Appendix A. The great majority of these values were less than .10, indicating that at least the large group factors had been extracted. There were, however, several significant correlations remaining, which raised the question of the presence of another factor. The diagonals of this matrix were those actually calculated, and further 63 factoring had to employ new estimates based on the highest re­ sidual correlation in each column. The process of test vector reflection was lengthy due to the many negative column sums, with the result that almost half of the tests have negative loadings on the complete centroid placed through the residual matrix, as seen in Table 19. Only two of the tests appear to have significant load­ ings on this factor, and over naif of them are loaded less. .20 or These findings are, of course, partly due to the nature of the complete centroid method which deals in the overall relationships. While not necessarily a meaningful factor, the complete centroid does provide an estimate of the largest remaining factor in the residual matrix. The complete centroid accounted for 7*6 percent of the estimated communality, and this figure plus the 95-7 percent accounted for by the other six factors total more than 100 percent. This discrepancy was caused by re-estimating communalities in the residual matrix in accordance with the highest residual corre­ lation in each column, which, in most cases resulted in an in­ crease in value for each communality. The complete centroid check, having suggested the presence of another factor, resulted in the formation of a new test cluster which included the tests of appreciable loading on the complete centroid. After elimination from the cluster of those tests which correlated poorly with it, the cluster de­ cided u p on included the four most highly loaded tests. Further, 69- TABLE 19 ORTHOGONAL LOADINGS ON THE SEVENTH FACTOR EXTRACTED BY A COMPLETE CENTROID THROUGH RESIDUAL MATRIX Test ----------------------------------------------No. Content 9-0 23 30 11 18 31 1 6 29 33 21 32 33 20 jo 7 --3 3p 27 -*-5 5 22 hf b2 b c .0 o 3 .2 \7 28 17 -+R 3- Neat personal appearance Remarks about baby picture Reacts to criticism Social response to cigarette offer Attends to phone conversation Imitates E.'s bead stringing Responds to E.'s change of pace Sits near E. Rejects E.'s help Re-enacts task sequence (gross) Responds to interrupting question Varies reading rate with E. Rejects dull pencil 'Helps E. find match Verbally fluent Removes sitting obstacle Remembers E.'s name Good attention Helps E. recall name Gives parting response States name preference Continues story after interruption Presence of mood modulation Curious about tail under box Looks at E. during handshake Responds to tack on E.'s chair Hc locomotor retardation Relaxes oody Confronts E.'s stare Uses pressure in handshake Crosses legs Good social vocabulary Follows gaze of E. Willing to attempt difficult task Responds to compliment Responds to weather remark Understands situation Re-enacts task sequence (minute) Loading .9-9.b2 .36 .35 -.33 -.29 -.26 .2b .29.23 -.23 .2 .2 i -.21 -.21 .20 .20 -.20 .18 .18 -.18 .17 .16 .16 -.15 .19-.19.13 -.13 .12 .12 .12 -.12 -.12 -.11 .10 .10 -.09 65 TABLE 19 continued. Test No 2 25 9 26 . Loading Content Offers hand in response to E. Retrieves dropped pencil Makes initial verbal response Expresses political opinion -.08 -.08 .07 -.03 66 they seemed to have some meaning and empirical justification for clustering. To arrive at the oblique loadings for this factor it was merely required to extract it from the original correlation matrix. The orthogonal loadings were derived by extracting it from the residual matrix which would locate the factor axis in orthogonal reference to the other six factors. The sum of the squared factor loadings equaled only 2.6 per­ cent of the estimated communality, and thus the factor seemed of little importance. The oblique matrix showed high loadings, but the reason for this became obvious when it was learned that this seventh oblique factor was highly related to factor A. The cluster tests were related to A-cluster tests almost as much as they were interrelated among themselves, and some of the highly loaded tests on the seventh factor were even more highly loaded on factor A. would not be meaningful Thus, it appeared that it to distinguish this factor, but to treat it as a special aspect of factor A. meaning may be applied That some specific to it will be discussed below. The Obtained Factors and Interpretations Even without regard to the special limitations of the present investigation the problem of factor naming is an ex­ tremely complex one. The task is not made easier by the lack of concepts in this area, either through theoretical formu­ lations or empirical studies such as this one. The special limitations of this study are the relatively small number of cases, and perhaps more serious, the lack of a sound theoretical 67 framework with which to integrate the results. however, The latter, should not act as a deterrent to research in this domain, for the need of concepts is pressing. search in a new area is necessarily exploratory of the group of variables chosen for study. That any re­ is a function It is a natural result that some factors will be well tested while others will be viewed with lesser clarity, and may even mislead the investi­ gator. factors, This fact Is intimately associated with the naming of and emphasizes a entire project. plea for the tentativeness of the Finally, it should be borne In mind that the unrotated factors are in unknown proximity to reference axes which would best meet the criteria for simple structure. As has been previously discussed, the multiple-group method, with its special heed to the experiential meaning of clusters, usually results in a solution close to that of simple structure. In this problem it is felt that the preliminary factorization accomplished B. this in regard to the locations of factors A and The other factors, seemingly of less importance and of less clarity, might be better measured by tests other than those included in the present study. The present matrix is a transient one which may never be employed again, even though other matrices may take account of the present one in their design. With these limitations upon interpretation in mind the factor meanings may now be considered. Reference is made to Table 20 for this purpose since the emphasis here is upon the qualitative content of factors. 68 TABLE 20 FACTOR LOADINGS OF .^O OR HIGHER THAT LED TO THE NAMES FOR FACTORS Factor Symbol D Name Test (By Magnitude of Loading) General social orientation and capacity Good social vocabulary Re-enacts task sequence (gross) Varies reading rate with E. Understands situation Re-enacts task sequence (minute) Helps E. recall name Looks at E. during handshake Verbally fluent Good attention Neat personal appearance Imitates S.'s bead stringing Responds to weather remark Responds to cigarette offer Remembers E . ’s name Socio-motoric responsiveness Responds to E.*s change of pace Offers hand in response to E. No locomotor retardation Responds to tack on E.'s chair Presence of mood modulation Responds to compliment Gives parting response Remarks about baby picture Restitutional curiosity Retrieves dropped pencil Curious about tail under box Willing to attempt difficult task Nonchalance and lack of pre­ occupation Crosses legs Responds to interrupting question Relaxes body Reac tive assertion and reproduction Expresses political opirion Uses pressure in handshake Follows gaze of E. Attends to phone conversation Helps S. find match 69 TABLE 20 continued. Factor Symbol F Name Interpersonal confronting Test (By Magnitude of Loading) Confronts E.*s stare Sits near E. States name preference Continues story after interruption 70 Factor A . This is by far the largest and most important factor derived in this study, and alone it accounts for almost half of the estimated communality of the entire matrix of correlations. Caution is needed in interpretation, however, due to the extreme obliqueness of some of the factors with re­ spect to each other. This may result in a somewhat artificial differentiation of concepts which only some knowledge of their descriptive and predictive usefulness can resolve. Factor A seems to involve a general social orientation and capacity, but not necessarily (in this sample) of social acts on a motoric basis. the actual carrying out Involved in this factor are the abilities to make social value judgments and to learn social concepts. The ability under consideration seems further to involve the comprehension of social events, and the imitation of another. Verbal fluency seems to be in­ volved, as does the capacity for attention and comprehension. More along interpersonal lines are the awareness of self in respect to others, a readiness to react pleasantly, and to play a role in conformity to the other person. important, at least in terms of this factor, to be superficial. Somewhat less is the ability The recognition of social obligations or "manners" may also be involved. In summary, factor A is interpreted as involving the capacity to react socially, and to bring intellectual re­ sources to bear with some objectivity as a result of at­ tention and understanding. It further involves the capacity 71 to exert a social neutrality with reserve potential and readi­ ness, The name invented for this factor, then, is General social orientation and c a p a c i t y . Factor 3 . in nature, This factor seemed to be essentially motoric i.e., with reference to its unique aspects. The correlation of .67 between the reference axes of factors A and B attested to the difficulty of differentiating between them semantically. It is probable that it requires the abili­ ty measured by factor A for an individual to be highly loaded on B. The two most highly loaded tests were obviously motoric responses which have to do with sociability. The next most highly loaded test also had this significance when it was realized that more than pure psychomotor performance was being measured. The subject was asked to put on the room light, and his score was the number of seconds taken to return to his chair--or to re-enter the interpersonal or social situation. Further, it would seem that this factor is getting at social sensitivity, and the finer modulations of motoric and af­ fective responses. nuances. Involved also are responses to social This factor, therefore, was named Socio-motoric responsiveness. Factor C . The naming of this factor required much in the way of speculation, as opposed to the more direct in­ terpretation of the above factors. While the amount of com- munality it accounted for was slightly greater than the re­ maining factors, too few tests were significantly loaded to 72 clearly demarcate its identity. however, In this or any other factor, it was necessary to examine the loadings of all tests in the battery on the factor under scrutiny. an important factor, necessary. It may well be but more complete testing of it would be The most highly loaded test, 'Retrieves dropped pencil', was probably impure in its measurement. of the other tests comprising the factor, the subject picked up, or looked for, osity. In the light it would seem that the pencil out of curi­ Since the factor was derived from a group of schizo­ phrenics, many of whom were in the active and early stages of the disorder, there would seem to be some restitutional nificance present. Fenichel sig­ (19) spoke of the type of resti- tutional activity designed to regain lost object relation­ ships, particularly on the part of catatonics. Such patients, fearing the loss of objective reality, and of relationships to objects, attempt to cling to, and seek objects. A certain amount of automatic obedience may also be involved, which may partially account for the retrieving of the dropped pencil. It will be noted that the tests with loadings above volve some sort of concrete object. .*+-0 in­ Whether or not cata- tonics are more heavily loaded on this factor will be seen later, but such a situation is not anticipated. Catatonics are a heterogeneous group and it is not likely that this specific activity would be common to all, or even to many. Pursuing terpretation, the theme of caution with regard to factor in­ it must be emphasized that significant loadings 73 on this factor by non-schizophrenics will neither prove nor disprove the interpretation of restitution. This refers particularly to the situation where non-schizophrenics do not contribute to the location of the factor by way of test correlations. Moreover, if such a factor did arise from the test performances of non-schizophrenics, its interpretation would not take into account the possible restitutional aspects of curiosity seen among some schizophrenics. It was with caution, and with due appreciation of the probable re­ stricted value of this factor that it was named Restitutional curiosity. Factor D . The above statements concerning the tenta­ tiveness of factor C's interpretation, apply to this factor as well, and more so to these than factors A, 3, S, or F. The two most highly loaded tests here seemed to be related to tension, and may have been reactions against tension. third test The ‘relaxes b o d y ’ in this context might be interpreted also as a reaction to felt tension, in which the subject gave the appearance of being bodily relaxed. and factors, Examining other tests it appeared tha c some social capacity was part of the picture, but that ’'sociability" may have been lacking. For a tense person to give the appearance of relaxation, it would be almost prerequisite to have some understanding of the situation. Some tests having to do with interaction between subject and examiner had low or negative loadings on this factor, which pointed toward nonchalance on the part of the 7*+ subject highly loaded on this factor. Again it must be empha­ sized that these speculations apply to schizophrenics and may be of an entirely different significance in other populations of people. This factor was named Nonchalance and lack of preoccupation. Factor 5 . This factor along with A and B form a highly related triad which must be considered when interpreting the significance of each. Based apparently on the capacity rep­ resented by factor A, and related to the socio-motoric re­ sponsiveness of B, this factor seemed to involve a special kind of response. Though 'uses pressure in h a n d s h a k e 1 and ‘follows gaze of E . * were motoric responses, the others, they, as well as required somewhat more in the way of spontaneity. They were initiated by the examiner, but the response was not delimited to the extent of B-factor tests. assertiveness seemed to enter the picture, Here also some particularly since the responses were not as clearly prestructured. Imitation and reproduction of the examiner*s role was felt to be in­ volved, which seem to de-emphasize the role of spontaneity. Some subjects with high loadings on this factor may have been attempting to gain or retain control of the situation, pos­ sibly due to anxiety, and the reactions may be quite super­ ficial as opposed to a consistent personality trait. however, is purely speculative. This, The name Reactive assertion and reproduction was invented for this factor. 75 Factor F . Although related to the above mentioned triad of factors A, 3, and E, this factor seemed to be getting at a somewhat more independent aspect of social behavior which had to do with confronting tention as such, the other. Not highly related to at­ there seemed to be a type of social at­ tentiveness involved, particularly when the entire pattern of test loadings was considered. At the same time the individual highly loaded on this factor seemingly maintained a certain degree of independence as seen in the loading of ‘Continues story after interruption*, and the low negative loadings on the two most highly loaded “curios ity” tests. was named, therefore, Seventh f a c t o r . This factor Interpersonal confronting. The difficulty of distinguishing this factor has been previously discussed in connection with the treatment of residuals. was so highly related fied In the oblique position this factor to factor A that it did not seem justi­ to consider it separately. An examination of the oblique test loadings suggested that this factor involved a special type of orientation to social interaction and reality in gen­ eral, which may be on a higher level than the general capacity represented by factor A. from and The most significant loadings came 'Neat personal appearance', 'Reacts to criticism'. 'Remarks about baby picture', These were not the three most highly loaded tests on the factor, loaded on the seventh factor but they were more highly than on any other. to the lack of clarity of this factor, Due, however, it was not named. 76 Factor Loadings for Individuals Schizophrenic s u b t y p e . Since the factors were derived from the test performances of the schizophrenic group, it was important to have some notion as to the relationships of the major subtypes to the factor structure. Table 21 shows a breakdown into three major subtypes of schizophrenia which includes 90 of the 100 subjects. Ten other subjects were classified in groups too small for comparative purposes. In this comparison it should be pointed out that factor loadings were computed on the basis of the correlations between subtype and cluster tests only. An examination of the matrix reveals only two loadings with probable significance, and two others which approach significance. All of these occurred on factors A and B, and in general confirm the conventional psychi atric appraisal of the subtypes with respect to social behavior Paranoids are more highly loaded on these two factors than either of the other subtypes, suggesting that they retain a relatively good social capacity, and are relatively more re­ sponsive motorically to social stimuli. tonics, however, The trend of cata- is toward greater social understanding (Factor A) than their behavior (Factor B) would indicate. This trend is consistent with the notion that stuporous catatonics most often comprehend the events in their environment, despite a lack of overt response to them. atric text will point out, And, as any psychi­ they often are able to reconstruct 77 TABLE 21 SIX FACTOR OBLIQUE MATRIX FOR SCHIZOPHRENIC SUBTYPE Diagnosis J+2 .36 Catatonic type -.09 -.30 Hebephrenic type -. 56 .12 Paranoid type E F .05 .23 .13 .07 00 r-* • 1 B 0 OJ • 1 A -.22 -.10 -.13 .11 I • O Oblique Factor -.13 C D 78 the events occurring about them after the stupor has termi­ nated. The fact that the trend in the factor loadings is not pronounced to the extent of significance is probably due not only to the heterogeneity of the catatonic group studied, but also to the necessity for excluding patients in severe stupor­ ous states. The finding for hebephrenics with regard to the d is­ crepancy in loadings for factors A and B tends toward the re­ verse of that found for catatonics. The trend here is for hebephrenics to show some motoric responsiveness in the presence of poor social understanding and orientation. findings tend to confirm the speculations of Sillman who pointed out that with hebephrenics pression become confused. These (36), thought and motor ex­ With catatonics, he believed, motor expression is not regulated by ideation, excessively released or inhibited. so that it is Thus, he felt, and the findings of this study tend to agree, that schizophrenia in­ volves a loss of integration of ideation and motility. Case history variables and c o o p e r a t i o n . While negligible factor loadings by case history variables would have added to the generality of the factors in terms of other populations, the fact that several significant loadings were obtained, as seen in Table 22, can add to the validity of interpretation. The first to be considered is the sex variable. all loadings are negative may exceed but generalizations concerning The fact that the expectations of chance, the relationship between sex and 79 TABLE 22 OBLIQUE FACTOR LOADINGS FOR CASE HISTORY VARIABLES AND COOPERATIVENESS RATING Oblique Factor A B D C E rh OJ • -r Case History Variable F 15 Living with spouse 13 -.09 .00 -.17 ~ . 2 h - . 1 2 Education, 30 - .01+ .11 - .01 .28 .12 Duration of illness,<7 years ,22 -.03 .bl .11 .37 .12 Very cooperative .79 .U6 .11 .02 .67 .55 > 1 0 years .06 .51 .17 1 Age, < 36 • 21 - .01+ -.29 1 Sex, female .39 -.12 .01 80 social behavior in schizophrenia cannot be made. Rather, the specific correlations responsible for this must be examined. For example, the correlation between female sex and political opinion* ’Expresses is significantly negative, and another test ’Follows E*s gaze* almost approaches negative significance. These correlations account for the negative loadings for females on factor E. Sex differences in the expression of political opinion is a cultural truism, which apparently schizophrenia does not destroy. On examining other specific correlations it was found that the females in this study were slightly less attentive, frequently, ence. picked up the dropped pencil less and were more reluctant to state a name prefer­ As regards factor loadings, however, it may be said that these schizophrenic females were less assertive than the males. On the seventh factor, discussed above, it was found that female sex was positively related to tests important to the factor: ’Helps E. recall name' and baby picture'. ’Responds socially to The latter also seems to be a cultural phe­ nomenon which may be preserved in schizophrenia. As regards age, the fact that those subjects under 36 were fairly highly loaded on factor C, Restitutional curiosi­ ty, seems to support the interpretation of that factor. tutional symptoms are more in evidence during of the illness according to Fenichel grasps for reality. the early stages (19) where the ego still There was also a trend for the younger subjects to be more assertive, s ign ificance. Resti­ but the loading did not reach 81 Education showed no significant loadings, noting but it is worth that this variable was more highly loaded on factor A than the other case history variables under consideration. the one hand On this finding suggests that the amount of edu­ cation a subject received is related to his general social capacity, but that general social capacity involves more than mere academic achievement* Examining the loadings for duration of illness, it will be noted that those schizophrenics whose illness was of less than seven years duration were more highly loaded on factor C, which again tends to support the resti­ tutional aspect of that factor. have, in general, The fact that younger patients been ill for a shorter period does not de ­ tract from the relationships of these variables to factor C considered separately. Approaching loading of duration on factor E. significance also is the It would seem generally that the shorter the illness the greater is the preservation of the ilan vi tal which characterizes the normal social being. The ratings of general cooperativeness which were made on all subjects proved to have significant loadings on four of the six oblique factors. Most striking was the relation­ ship between cooperation and the triad composed of factors A, B, and E. It is to be admitted at the outset that cooperation is a rather vague concept which defies denotative definition, at least as employed here. Rather, the rating depended on the investigator *s clinical impression of cooperation and there­ fore was not included in the somewhat more definitive test 82 scale. Thus, it was highly subject to the halo effect carried over from test ratings having to do with compliance, at­ tentiveness, and general responsivity. In view of this, the loadings on the above factors were not unexpected, and little is gained in the way of interpretation. lation should be made. However, one specu­ If a given patient had a high loading on factor A, and related factors, it was probable that his comprehension of the situation caused him to cooperate, i.e., the examiner represented an official of the hospital from which, let us assume, he desired release. If this were true, a minor test of the validity of factor A was had. In regard to the moderately high loading of cooperation on factor F it can be said that "confronting" as meant here does not apparently involve a hostile approach to the situ­ ation. Evidently a subject was able to confront the examin­ er, maintain some independence, and yet be rather high-level social ability. In reference to the discussion of the cooperative--a loadings for sex, it was brought out that females had consistently low negative factor loadings on all factors except E, which was signifi­ cantly negative. Since proportionately more males are in­ cluded in the non-schizophrenic group, the examination of dif­ ferences on factor E should be qualified in light of this discrepancy. At this point the question might be raised as to whether schizophrenic and non-schizophrenic females can be viewed as homogenous with respect to sex differences in 83 behavior. It is felt that, due to the heterogeneity of schizophrenics in general, the variable of sex may be ab­ stracted from the diagnostic context for purposes of ex­ ploring sex differences. The dichotomous distributions for age were matched perfectly for the two groups, and the other case history variables can be ignored, not only because of the slight differences, but also because they have no significant factor loadings. The most striking difference between the groups in this table is in cooperation, for which all nonschizop'nrenic subjects received high ratings. This variable could not be controlled since no relatively uncooperative non-schizophrenics were available in the hospital population. Further, such patients, if they were available, would proba­ bly not be representative of hospitalized non-schizophrenics in general. Cooperation is most probably, however, a result, and not a cause of the ability represented by the factors to which it is related. be artificial To the extent that this is so, it would to attempt to control this variable. Factor loading indices. A factor loading index was com­ puted for each subject, of both groups, on each of the factors. The method employed in this operation has been previously dis­ cussed. Table 23 shows the obtained results in terms of the mean factor loading indices for schizophrenic subtypes and non-schizophrenics. It will be remembered that this analysis makes use of all tests significantly loaded on a factor as 89 TABLE 23 MEAN FACTOR LOADING INDICES FOR SCHIZOPHRENIC SUBTYPES AND NON-SCHIZOPHRENICS Diagnosis A Oblique Factor ■ ■ ---------------------------------3 C D E F 2.60 0.95 Catatonic 3.13 1.87 0.57 Hebephrenic 1.89 1.92 Hon-schi z . 6.98 3.50 rn H .58 0 • 1 —i Paranoid 1.18 0.8 1 0.71 0.99 0.67 0.95 0.70 0.92 0.69 0.72 1 .27 1.96 1.19 85 opposed to the computations of factor loadings for diagnostic categories which considered only those tests in clusters which located the centroids. ported, In this table only raw means are re­ and comparisons can be made only within columns since the means are not comparable from factor to factor. Table 2*+, Appendix A shows the means of the previous table which have been weighted by simply dividing by the respective sums of the squared factor loadings of the significantly loaded tests or. each factor. The divisor for each factor was, of course, equivalent to the maximum obtainable factor leading index for that factor. Figure 1 following terial graphically, this table presents the ma­ and aids in grasping the total picture. The differential factor loadings among types have been previously interpreted, the two foregoing schizophrenic sub- and in this regard tables and Figure serve only to lend clari­ fication to 'hose trends. Fne s a y r purpose for computing the factor loading indices was to compare non-schizophrenics in as accurate a manner as possible within tne context of the present study. schizophrenics and The number of non-schizo­ phrenics was only ^C, but there seemed little value in ex­ ceeding this number when the schizophrenic cluded only ^3 paranoids, Further, 32 catatonics, subtype groups in­ and 15 hebephrenics. there was little variability in factor loading indi­ ces for non-schizophrenics, whicn resulted in extremely small standard errors of the means for each factor. Table 25* which shows tne t. ratios for tne differences between mean factor 86 TABLE 25 t RATIOS FOR DIFFERENCES BETWEEN MEAN FACTOR LOADING INDICES AMONG DIAGNOSTIC CATEGORIES Oblique Factor Diagnostic Categories d.f. A B C D ♦* ** 2 . 58 1.13 2.68 56 ** 2.95 ** 6-.63 1.71 0.00 Cat. vs . lieb. 05 1.99 0.12 44 ** F a r . vs . Non- schiz. Si 5.95 b.bQ 0.75 ★* 3.10 C a t . vs . Non- schiz. 70 ::e b . vs . Non- schiz. 53 8 .6 l ** 9 .26 F a r . v s . Cat. 73 F a r . vs . Heb. 44 ** 6 .U-5 lAl ♦* b.18 1.52 .05 leve 1 of ccnfid ence * * ^ fnif‘i can t at .01 leve 1 of confidence • F 1.33 1.38 2.b2 1.09 1.18 0.07 ★* 2.59 ♦* 5.3'+ ** >+. 59 0.08 0.21 44 ** '-+.03 ** 5.70 ** * * Si v"’"1"f1'icant at - E 6.11 ** S.h-9 ** ' h .71 3 .82 87 loading indices among the diagnostic categories, includes the number of degrees of freedom for each comparison. Fer'naps the most important finding reflected by this table is the importance of factor A for differentiating between the diagnostic categories which were paired for comparison. In each case the t. ratio was highly significant except for the comparison between catatonics and hebephrenics which met only tne minimum criterion for significance. This suggests that general social orientation and capacity is significantly im­ paired by a schizophrenic psychosis, and further, that the extent of this impairment is relative to the type of reaction syndrome developed. The fact that hospitalized non-schizo­ phrenics were employed suggests that the ability measured by factor A transcends mere residency in a mental hospital. What is not known, however, is whether this impairment seen in schizophrenics is due to psychosis or schizophrenia, such a distinction can be made. psychotic if Data obtained from non- schizophrenics or non-schizophrenic psychotics would aid in the clarification of this unknown aspect of factor A. All of the comparisons of the differences between means among non-schizophrenics and schizophrenic subtypes were high­ ly significant in favor of non-schizophrenics, C. except on factor While the mean factor loading index for non-schizophrenics on factor C was higher than for any schizophrenic subtype, only in the comparison with paranoids did this difference reach significance. The possible meaning of this factor, Restitutional curiosity, in schizophrenia has been discussed at length, and it was suggested that the behaviors en­ compassed by this factor may have a qualitatively different significance outside of schizophrenia. The trends of the data suggest that the meaning of this factor may even be spe­ cific to catatonics, is forthcoming. but no reliable evidence to support this These findings emphasize the danger of in­ terpreting behavior in isolation without regard to its rela­ tivity. The i n t e r p r e t e r s point of view is important here in much the same way as in the case of severe anxiety symptoms which, on the one hand suggest psychopathology, otner, but on the a favorable prognosis. The tenuousness of this investigation has been frequently emphasized, particularly with reference to the factor analytic solution and the interpretation of the obtained factors. caution is urged even more This strongly in the interpretation of factor loading indices beyond very broad generalizations. The concepts represented by the factors may or may not be those which differentiate usefully among individuals, poses of description. even for pur­ The fact that significant differences were obtained so consistently suggests that the factors may be useful for description, but little can be said here beyond the observation that schizophrenics and non-schizophrenics be­ have differently in this situation. The behavior measured was largely social and interpersonal in nature, and it is likely 89 that its impairment in schizophrenia is intimately associated with the etiology of the disorder. Within schizophrenia there were differences in this behavior and it is likely that this too is related to etiology, but in a more specific way. The Higher Order Factors The entire problem of the organization of variables is a complex one, and one which transcends the technique of factor analysis. Factor analysis has, however, taken upon itself the classical problem of level of analysis and has much to offer toward its clarification. Psychology as a science has had much difficulty in deciding upon levels of analysis and units of measurement which would prove useful for purposes of de­ scription of behavior of the sort in which it is interested. Many and varied attempts have been made to employ units ranging along a size continuum from the minute units of measurement employed by the physical and biological sciences to the gross units used by sociology. The variables employed in this study, for example, may be in some cases very complex as to composition of lcwer-order variables, while others are more pure with respect to the variables which they organize. And so it is with factors which may show extreme variability as to their makeup, and the variable constellations which they represent. From the point of view of the things which are organized the organizer is a relatively independent in­ fluence, but the organizer is itself organized into an even 90 larger context upon which it may be highly dependent. The analogy may be made to the construction of an outline wherein the sub-categories are organized by the larger categories, and the latter by even larger ones. Factor analysis attempts to build up toward the larger categories as opposed to a strictly hypothetical-deductive system which begins with them. At the present stage of research in personality it is more desirable to discover the larger functional unities, and both approaches may complement one another in this respect. One empirical investigation employing factor analysis cannot, however, establish these larger wholes with any sense of finality. Many investigations within the same and related domains are necessary, particularly if the choice of vari­ ables is not guided by theoretical formulations. factors, Higher order then, must be regarded as being as tentative as those factors which they organize. To establish the second-order factors in the present study, the matrix of oblique factor intercorrelations was factored. This was done by inserting the highest value in each column in the diagonals as communality estimates. one cluster could be clearly identified and tracted as a group centroid. Only this was ex­ Based on the cluster of factors A, B, and E, this second-order factor accounted for 76 per­ cent of the estimated communality. Residuals were computed in the usual manner, and a complete centroid was placed through the residual matrix. This latter factor accounted 91 for only 11 percent of the original estimated communality, and suggested that little remained in the way of group factors. Table 26 shows the orthogonal matrix for these second-order factors. Factor II seems to have little meaning, particu­ larly in its unrotated, orthogonal location with respect to factor I. Factor I has been anticipated in previous discussions at which time the usefulness of the triad of factors A, B, and E was examined. Not only do these factors show high loadings on factor I, but the other factors are also loaded to the point of significance. Thus, this second-order factor would appear to be general in nature, served variability. tentative, in that it organizes much of the ob­ "While such a factor must be considered certain speculations may be made. This factor may broadly define an area of general social ability which arises from the socialization process and early cultural learning common to al] human beings in our society. this same study, It may be that if repeated on normals, would not reveal separate factors similar to A, B, and E, but would derive a larger general factor, similar to factor I directly from the data. It may be further speculated that this broad general factor is differentially impaired in schizophrenia, and some evidence for this is reflected in the differential loadings for schizophrenic subtypes among the A-B-E triad of factors. This raises interesting questions concerning the etiology of 92 TABLE 26 FACTOR LOADINGS FOR TWO ORTHOGONAL FACTORS DERIVED FROM INTERCORRELATIONS OF SIX OBLIQUE FACTORS Oblique Factors Orthogonal Factors ,2 II 8l -.10 .67 B ,82 -.17 *70 C AO .2^ .22 D A+3 .32 .29 0 .80 .10 .65 F .*+3 -.37 .32 0.3 5 2.85 ai j2 2.50 93 this impairment in schizophrenia, and the roles of the parents who act as agents for the transmission of much of what is learned in the way of social and cultural responses. It may well be that factor I repreeents an over-abstraction which has little value for description, but on the other hand it may prove useful for higher-level categorization. stantiation would be necessary, however, Sub­ for whatever purpose it may serve. Suggestions for Future Research A factor analytic study such as this represents an at­ tempt to discover new relationships. This was an exploratory study because the observed relationships, reliability, are of unknown usefulness, regardless of their except for the circum­ scribed descriptive comparisons made earlier. the variables themselves is open to question, wide variety of things may be observed. The choice of since such a The use of specific hypotheses in further studies can give meaning and organi­ zation to the variables employed. Thus, at all stages of the present research, many questions are raised. The present method should be contrasted with some of the more conventional procedures in clinical psychology, ample, the use of objective tests, suggested. for ex­ since further research is The present study made no a priori assumptions about the interrelatedness of variables and set out to study the relationships of items of behavior which had little 9^ conceptual treatment. It would be important to bridge the gap between the more time-honored categories, such as those of the Wechsler-Bellevue Scale subtests and the MMPI divisions, and the present categories represented by the factors. Many objective tests have not been shown to be factorially pure, and a breakdown of their items with the inclusion of such items as were employed in the present investigation, would result in test procedures which transcend those of the present. Assuming that the social behavior of schizophrenics is a fruitful area for research, and the literature concerning eti­ ology would suggest that it is, other methods might be devised for observation. This may take varied forms, whether it be observations of spontaneous ward behavior, or behavior in a planned group situation, cr even techniques similar to the one employed here in which a good deal of deliberate structure was had. Each of these techniques would involve different sets of variables which is desirable from the point of view of enlarging our understanding. Methodology should by no means be limited to the use of factor analysis, since more deductive methods may help to build sorely needed theories ap­ plicable in clinical psychology in general, and psychopatholo­ gy in particular. There would analytic studies for be some value, in further factor in the same general area in order to test the presence of the factors Thepresent factors however, mignt serve derived by the present study. to suggest variables which 95 would provide more extensive tests of the factors. This would serve to purify the factors by exclusion of poor tests and in­ clusion of tests which are better measures of the source traits represented by the factors. On the basis of such further ex­ ploration and purification, the complex procedure of rotation to simple structure would be justified. Even more practical in this regard would be an analysis employing Q technique in which the emphasis would be on types of individuals who share similar patterns of reaction. If relatively pure types of individuals were thusly established, the way would be open for an intensive study of etiological factors associated with each type. Jince the etiology of schizophrenia seems to be closely associated witn the early social vidual, it would be more logical life of the indi­ to study criterion groups which are more pure witn respect to the behavior being in­ vestigated. While some homogeneity of social behavior is seen in the Kraepelinian diagnostic groupings, it is not un­ reasonable to attempt better categorization in this regard. 5ecause of the extensiveness and probable validity of she second-order factor I, which arose out of tne clustering of first-order factors A, 3, anu E, in testing its usefulness. there would be much value It would be desiraole to trace the etiological factors associated with high and low loadings on cm s actor. formation as Of more immediate practical to response import would be in­ to treat nor. t and prog nos 1s of sc hi zo- riirenic criterion grouts comoosed on trie oasis of Lneir factor 96 loadings. In this regard, such criterion groups would permit a study of post-hospital adjustment, which is associated close­ ly with social behavior. In general, there would seem to be little value in furtner investigations designed to demonstrate differences the factor loadings or factor loading in indices between an a- massed group of schizopnrenics and any other diagnostic group. This study deliberately placed approximately half of the sub­ jects in each of t.ne dichotoraous categories, ferentiate amonr scnizopnrenics. are not homogenous, in order to dif­ dchizophrenics as a group either in symptomatic reactions, sponse to treatment, or in post-hospital adjustment. clinician must concern himself with which; may be related reactions. re­ The the possible differentiae to, and aid in trie prediction of these i’o this end ..e needs conceptual categories wnicu will aid in the construction of criterion groups with which to study these reactions. fh is study has empirically d e mon­ strated certain relationships among variables which may not only aid in such investigations, atic approach to categorical but lead to a more system­ thinking about schizophrenia. 97 O A review of previous work in sc'nizopnrenia revealed that impairment in social behavior is a prominent aspect of the clinical manifestations of the disorder, closely associated with its etiology. ceptual and that it may be A diversity of con­ treatment of these observations suggested a need for concepts which can facilitate furtner systematic exploration in this area. Among the various methodological alternatives to wnich the clinical psychologist may resort, sis, factor analy­ considered as a branch of statistical method, advantages. offers many leveral factor analytic approaches are partic j - larly applicable to the problems encountered in clinical psy cncloy; research . inis investigation commenced with a selection of uavicral items vhi cu seemed be- to be of social import, and sub­ ject to some sort of measurement. .111 of the responses were elicited in an interview situation, whica included some moreformalized reduced tests. Ate test items were roughly scaled and later to dicnotomous scoring of tetracnoric correlations. categories for the calculations ine correlations were based on tne test score distrimutions of 100 scnizophrenic suojects who were nighly variable witn regard to subtype classification, ace, duration of illness, education, etc. line final matrix of tne intercorrelatinns of On variables was subjected to the 1 98 multiple-group centroid type of factor analysis. Preliminary factorization aided in the most meaningful placement of the first two factors, which were of the greatest size and import. Six oblique factors were extracted which in some cases were highly related. A B C D E F rhey were named as follows: General social orientation and capacity Socio-motoric responsiveness Restitutional curiosity Nonchalance and lack of preoccupation Reactive assertion and reproduction Interpersonal confronting The residual correlations after tne extraction of these six factors failed to reveal any remaining group factors of signif icance. Computations of factor loadings for case history variables were made, as well as factor loading indices for d iag­ nostic categories. of factors, This aided not only in the interpretation but also provided minor tests of their usefulness. A grout of L D non-schizophrenics was compared witn the ;:r.iztpnre:»1 c subtypes as to mean factor loading indices, most differences were found to be significant. and Significant differences between scnizcpnrenics and n o n - schizophrenics were attained on all factors except G, but were more apparent on factors k snrenics, and 1. laranoids were more similar tna.o eitner cats tonics or neoepnreuics. a.mo t g tne scnizopnrenic c u m : d ered to non-schizc- subtypes suggested setara tel" in retards Differences tr.at they must be to social a t i_ 11 v and socio - 99 Two second-order factors were extracted from the intercorrelation matrix of the oblique factors. The first of these was based on the highly related triad of factors A, B, and E, and was interpreted as a very general social ability which may be differentially impaired by schizophrenic reactions. The entire investigation was viewed as exploratory and the relationships observed were regarded as tentative in their usefulness. The method of direct observation as employed in this study, however, is to be emphasized, and it lends a certain face validity to the relationships found. of this investigation, pirical The results thus, may open the way for future em­ investigations in this area and assist in conceptual thinking about schizophrenia. 100 BIBLI0GRAPHY 1. Bales, R. F. Interaction process a n a l y s i s . Addison-Wesley Press, 1950. Cambridge: 3artlett, M. R. Suggestibility in paranoid patients. jI» g e n . P s y c h o l .. 1 9 6 6 , 3 0 , 97- 1 0 2 . 3. Beliak, L. Dementia p r a ecox. Stratton, 1 9 6 8 . New York: Grune and 6. 3urt, C. L. Correlations between persons. P s y c h o l .. 1937, 28, 59-96. J. Burt, C. L. Alternative methods of factor analysis and their relations to P e a r s o n ’s method of principal axes. E r i t . J. P s v c h o l .. 1969, 2, 98-1, 6. Cameron, N. Reasoning, regression, and communication in schizophrenics. P s v c h o l . K o n o g r .. 1938, 50, No. 1. 3rit. Cameron, P. A s tudy of thinking in senile deterioration and schizophrenic disorganization. Amer. P svcho 1 . . 1938, 51, 650- 0 6 5 . 8. Cameron, N. Schizophrenic thinking in a problem-solving situation. J,. m e n t . S c i . . 1939, 35, 1012-1035. 9 . Cameron, N. The functional psychoses. In J. M c V . Hunt (E d .), Personality and tne behavior d i s o r d e r s . Vol. 2. New York: Ronald Press, 19^+6". P p . B o l - 9 2 1 . 10. Cattell, R. B. Psychological measurement: ipsative, normative, and interactive. P s v c h o l . R e v .. 1 9 b b , Hi, 292— 3 O g . 11. Cattell, R. 3. The description and measurement of person­ ality .New York: World Book Co., 1 9 6 6 . 12. Cattell, R. B . , with Cattell, A. K. S., and Rhymer, R. M. P-technique demonstrated in determining psycho-physiological source traits in normal individuals. Psychome trika. 196-7, 1 2 , 269-283. 13. Cattell, study. R. B. Personality: A systematic and factual New York: McGraw-Hill, 1950. 101 1*+. Cattell, R. B. On the disuse and misuse of P, Q, Qs, and 0 techniques in clinical psychology. c l i n . P s v c h o l .. 1951, 7, 203-21^+. 15. Cattell, R. B. Factor ana lysis: an introduction and manual for the psychologist and social scientist. Mew York: Harper, 1952. 16. Chesire, L., Saffir, M. , and Thurstone. L. L. Computing diagrams for the tetrachoric correlation coefficient. Chicago: Univer. of Chicago Press, 1933- 17. Cohen, B. and Flower, 5. H. Clinical findings in d e ­ mentia praecox. A m e r . J. P s v c h i a t .. 1990-91, 97, 319- IS. Deutsch, Helene. Forms of emotional disturbance; re­ lationship to schizophrenia. P s v c hoanal. Q u art. . 1992, 1 1 , 301- 3 2 1 . 19. Fenichel, 0. The psychoanalytic theory of n e u r o s i s . New York: Norton, 19^+5. 20. F romm-R eichmann, Frieda. Preliminary note on emotional significance of stereotypies in schizophrenics. Bull. F o r e s t . S a n .. 1992, 1, 17-21. 21. Gerard, D. L. and Siegel, J. The family background of schizophrenia. Psychla t .Q u a r t . . 1.950, 2*+, '-+7-73 • 22. Guilford, J. P. Psychometric m e t h o d s . McGraw-Hill, 1936• New York: Guilford, J. P. V/hen not to factor analyze. B u l l .. 19 52, 6 ?, 26-37. Psvchol. 29. Haas, M. Sex incidence, annual curves and incidence of schizophrenia. A r c h . f,. P s v chiat. . 1938, 108, 552-561. (Ci ted in (3 ) ) . 2>. hunt, J. McV., and Gofer, C. N. Psychological deficit. In J. McV. Hunt (Ed . ) Personality and the behavior disorders.. Vol. 2. New York: Ronald PresTJ 19^+T Pp. 971-1032. 26. Johnson, D. M. , and Reynolds, F. A factor analysis of verbal ability. P s v c h o l . R e c ., 1991, 9, 183-195* _7. Jones, E. S. Subjective evaluations of personality. In J. McV. Hunt (Ed. ) Personality and the behavior disorders. Vol. 1. New York: Ronald Press, 1999. Pp. 139-159. 102 28. Kas&nin, J. 3. Developmental roots of schizophrenia. A m e r . j;. P s v c h i a t .. 1995, 101, 770-776 . 29- Kohs, S. C. Intelligence m e a s urement. Macmillan, 1923. 30. Levin, M. Misunderstanding of the pathogenesis of schizophrenia, arising from the concept of "splitting". A m e r . J. P s v c h i a t . . 1937, 9*+, 877. 31. Lewis, N. D. C. Outlines for psychiatric examinations. (3rd Ed.) Utica; State Hospital Press, 19^+3• 32. Maas, H. S., Yaron, 2., and Rosenthal, D. A technique for studying the social behavior of schizophrenics, j;. a b n o r m . s o c . P s y c h o l . . 1951, 96, 119-123. 38. Meyer, A. Fundamental conceptions of dementia praecox. B r i t . M e d . J . , 1906, 2 , 757. (Cited in (3) ). 3*+. Myerson, A. Scrutiny, social anxiety, and inner turmoil in relationship to schizophrenia. A m e r . J. P s v c h i a t .. 1998, 105, 901-909. 35. New York: Rabin, A. I. Social contacts of psychiatric patients-some preliminary observations. P s v c h i a t . Q u a r t . S u p p I .. l91+5, 19, 177-183. 3 6 . Sillman, L. R. The dynamics of schizophrenia. m e n t . Dis., 1997, 105, 61-72. 87. J_. n e r v . 3 1 o t k i n ? J. S. Nature and effects of social interaction in schizophrenia. J_. a b n o r m . s o c . P s v c h o l . . 1992, 37, 305-3 6 8 . 33. Snedecor, G. W. Statistical m e t h o d s . Collegiate, 1937. 39. Spearman, C. General intelligence objectively determined and measured. A m e r . jJ. P s v c h o l . . 1909, 1 5 , 201-293- 90. Ames, Iowa: Stephenson, ':!. The inverted factor technique. P s y c h o l .. 13 36, 2 6 , 399-361. Bri t . J. 91. Sullivan, H. S. Conceptions of modern psychiatry, the first William Alanson ’White Memorial Lectures. Psvch i a t .. 1 9 9 0 , 3 , 1-117. 9.;. Terman, L. M . , ana Merrill, M. A. Measuring intelligence. New YcrP: houghton Mifflin, 1937. 103 ^3. Phurstone, L. L. The vectors of mind. 193 ^, *+1 , 18 -2 1 . bb. Thurstone, L. L. Multiple factor analysis. Univer. of Chicago Press, 19^-7 • b?. U. 3. Public Health Gervice Publication ho. 2 33* Patients in mental ins t i tutio'ns 196-9 . 'Hashing ton: U. 3. G o v ’t. Printing Office, 1952. bo. Wittenborn, J. R. Factorial equations for tests of at­ tention. Ps vchonetrika. 1 9 6 3 , 2, 19-35. by . P s v c h o l . R e v .« Chicago: Wittman, P. and Gteinberg, D. L. Prodromal factors in schizophrenia. A m e r . J_. P s v chiat. . 196-6, 100, 811-8 1 6 . APPENDIX A 10lf TABLE 6 COMPARISON OF SCHIZOPHRENIC AND NON-SCHIZOPHRENIC GROUPS WITH RESPECT TO CASE HISTORY VARIABLES AND RATINGS OF COOPERATIVENESS Category Percent of Cases ---------------------------Schizophrenic Non-schizophrenic Sex, female 39 20 Age, 50 50 Living with spouse 31 38 Education, 51 1+0 52 100 < 36 > 10 years Very cooperative 105 TA3LE 7 PERCENT OF PLUS SCORES BY TESTS Percent Plus Test No. Percent Plus 1 72 25 bQ 2 bb 26 3>b 3 57 27 73 b bQ 28 b2 5 62 29 b7 6 61 30 b9 7 • bb 31 52 8 bQ 32 53 9 bb 33 10 50 3*+ 56 11 57 35 51 12 bQ 36 V9 18 71 37 51 lb 57 38 38 15 b5 39 10 lb bb bo -r Test No. 17 36 bl 57 18 52 b'2 25 20 28 bl 31 21 69 bb bb 22 3o b5 52 23 68 b6 85 106 TABLE 11 CLUSTERS DETERMINING GROUP CENTROIDS Cluster Jymbol Test Name A General social orientation and capacity B Socio-motoric responsiveness No. 27 31 32 33 35 36 37 1 2 Name Helps E. recall name Imitates E.'s bead stringing Varies reading rate with E. Re-enacts task seq. (gross) Good attention Verbally fluent Good social vocabulary M-i Responds to E.*s pace Offers hand in res. to E. Res. to tack on E.*s chair Fresence of inood modulation 8 C Restitutional curiosity 25 ^8 h2 Retrieves dropped pencil Attempts difficult task Curious about tail in box D Nonchalance and lack of preoccupation 12 13 21 Crosses legs Relaxes body Res. to interrupting question E Reactive assertion and reproduction 3 16 Uses pressure in handshake Follows gaze of E. Expresses political opinion 5 6 15 States name preference Sits near S. Confronts S . fs stare F Interpersonal confronting 107 TABLE 12 B C D E F A .58 .39 .17 .22 .36 .21 B .39 .59 .12 .l*f 00 .1*+ C .17 .12 A 5 .08 .20 .Ob D .22 .1^ .08 •b? .18 -.01 E .36 • 00 .20 .18 .56 .17 F .21 .lb .Ob -.01 .17 • 00 A • AVERAGE INTERCORRELATIONS WITHIN AND BETWEEN CLUSTERS 108 TABLE 1*+ COSINES OF THE ANGULAR SEPARATION BETWEEN THE SIX OBLIQUE FACTORS A B C D E F A 1.00 .67 .3^ A3 .63 .*+5 B .67 1.00 ,2b .27 .66 .30 C .3^ .2V 1.00 .18 .39 .10 D A3 .27 .18 1.00 .35 -.02 S .63 .66 .39 .35 1.00 .38 F A5 .30 .10 -.02 .38 1.00 109 TABLE 17 COMMUNAL ITY ACCOUNTED F O R BY iS IX OBLIQUE FACTORS Oblique Factor Percent of Estimated Trial factoring Final with adj. A *+5.76 *+9.69 B 11.05 12.05 C 8.72 9 .*+8 D 7.17 7.9*+ E 7.52 8.30 F 7.33 8.20 87.55 95.66 Total 110 TABLE 18 RES IDJAL MATRIX AFTER EXTRACTION OF SIX FACTORS / Z 5 1 £ 6 7 t 4 * t / A t i to * * i 7 i t t o i i M S X l f l l l t X / ft z is oi 3 ’otOtoH f o io o o tts svtos-o/ioot 0 0X03 0Z 00-0300 TOlOl i3'0i't0'060\ MTZ'Ot WOW OO'OSO/ f '/I'oS~H-*Ot~tiZ50*00 to oo vx-tntf oi o t-+ -ey* o t // *7 10-6VK, 10of-or-otof / t /x in~/soi-oirtf*-ot-H ot a os on *3 Won Oi tilt 06 iC oS-ff’M 01 »H VI'IX01TW /O'H'H 01 it *• 'OS't£'0Ol /fyf'Orvi t0 0! 03 *73 OS OS *~*X0i * OH * *V74*05 0/ O iV l'13 OtoS-OHTOfOX’tf-OSVI Oi li^to H~C6 'IS~Cf06'150° 00 0! ts on 'frof of-Of if it 16-V2'0700 03 vt'iS'UOt 23030$ 77-0300os 13ot 20 10*7* 00 i/'03-/t'070flZ'Og'Otl2’2J 062113*0001 21 ~/33 Vl'li U - t t W U ~ 0i~f2/303ot if'OScl ot of ofof x i v s t s ti'oHs - tf - f i - q * X 0H -ct 18 t% H 'H TU 'O bit’H ’jO 0%, 23 -3 ttf-2 0 -2 fo t-01 of V f - i t l s / 1 2f-ot-ol-aZci UOtroX’T T 'tfof isyo f/x cf oo 07'3fV f-03Otoe -oxas-# if o s - o s /^ ix - o iir o tx o i 20 0l - * 00'0t v i 030t Oi to 18 ii'V f O70X on OSOfonS’Ot-ti l&-0f 03 Z) -ff*3-S30fi£'tf 10 i7 OltX'/S 6507t0~O3Ol~13ot~13'tiliznoeo3 Of 21 03 Oi -of-o£'O70forl2'/7/1 '30 Of-0117-03*2~MS-*i l / a i vy~clffj'of 06 vt[t6T8 o6 tfc v jijo i'iS r tf'g -o ftfo fito t to os'is'tt'fo 2o~io~2Zl?~t3i f if 5*1 C i'ti^ i'O fo f 1) M>o%!0 0612 Oi ’i f O S -tftf *072 '/7 t i l * 13 Of-J0 21 TV O S /i M i f 'tf 'c f o t ’o t'o jix oo o 3 o n x o p & t y o y v / 1 3 i t - a /9 i l - o f o f c s - o r t T M l o o J0lI t 0 2 1 0 1 7 'Oi i f 0 7 0 6 0 * 1 ! 2/ o ti7 '2 o % o 3 o i o i -o^ofoCrf.)0 5 -tH -Qa a of too* u o w n ‘I f i 7 ID II -Ol'OtdO t s 07 0013 ' I f / S 00 03 i S ' i f * H O tH O7~Ot'fZOS'0l 0*61 TfrtSII 07 Ot’OZ- (7 '* OS7 5 OS-lO'OH 03-11 07 01 I7 ’HIO 01 Ot'i3'Ur0t'GS‘07’l2'l2'O\ O f' 1$00 33 $1-01 -00 ts 07~li’24 ’i f 12 Oi ' K ' t i i i H 01 Ot'OTVf 17 07 01 'OS’Oi OS'OIOStO 01 Oi OOVS~Ol OX % M S o t OS o f f t-OL'Oi-OS 01 OS'03 01 01 20 0X01 12 t l ' o r l l 01 -JT’OlOt I t OO-OtOO-01 -OSll OS 01 n '0067 OX H -tt i t -il OH’OS'OS Ol'iz Ot-Of’OlOfSI 'H 01 OS tX'01tt-H -Oi’OI OS oi 01 OL'Sl 07 01 05 o t 51 ' / f i t Oi '/s 02-or3130 07 Of Of-lZ II i f '01 -OXrOS-if-OiOl-OXIS OS00 iS-05-0S OO’IfO I 03 -0901-07 01 0/ % V4-1011 03 -XL iS'H'OS’i t 07 U Z7'l3'Ot0i'Ot'O5Vf-Ol’t r o f / 6 ’Of-OStiOS XI iS'IC i f tO -if I f l S ’Ol tZ Ot OX'H Oi O t-4i-4Stf H> Of 00 i f 0 f'f7 /3 iS'OSOZOI I7'030l If-OZOXinii s i IX O iV S -o iV lil O f 'O lO tm t tt 03-11 01 Oi O O 'ttl l i H ' t S i f O t 17 0305 OS-0Z-O3O3V14*11 OS’Oti iO-OH-Hoi 00-17 07 Of-Vi 07-0! fOOS-Ot Oi a Oi VSVi/X-IS 00'O$'/ZXn ' t ' / r v i / t i s * V 7 - 1 1 ' 3 f o r i 3 0 l fo -3f If X5 Oi 0/ O lX V O iif tio S li OS'0 7 17 1 1 * 0 1 U tiO L iOUD-0113 0 3 -V i-iltS OS01V I 01’iO 06 CO'/S0100 007X'01’IS-0740/(,Oi 4 1 /Z-Ot-Ol'Of 0/ Ot-O2-0i01-03-17Oi ¥S VSVllt'OfOOOO’Oi’O S X iV l/o O S 'itlttb 'iS IZ 0 5 O t a i f '01/1 0 ! -0 tU - l7 lf -OSOj-Oi-Ol f t - i f o i * 2 0 /0 - 1 3 0 0 & Ill TABLE 2b WEIGHTED MEAN FACTOR LOADING INDICES FOR SCHIZOPHRENIC SUBTYPES AND NON-SCHIZOPHRENICS Oblique Factor Diagnosis B C D E F Paranoid .65 .66 .33 .60 .50 .57 Catatonic •b-5 • b7 ,b2 .60 .b Q Hebephrenic .27 ,b& .33 .68 .39 •b 5 Non-schiz. .92 .88 .53 .88 CO .81 00 • -r A Index Loading ted Mean F'actor no ,00; A Fir. 1. C B Oblique Lac tor B Weighted mean factor loading indices by diagnostic group. F APPENDIX B 113 SCALE OF TESTS USED IN ANALYSIS 1. Response to E.^auickening his p a c e . While walking with P. to examining room, E. quickens his pace. 1. P. keeps up with E. with little or no lag. 2. P. attempts to keep up to E. but with notice­ able lag. 3. F. makes some attempt to keep up, or at least notices E.'s quickened pace, but shows marked­ ly inadequate response. W . P. makes no apparent attempt to change his pace. 2. Response to E.'s offer of h a n d . E. offers hand immediately upon P.'s entrance. 1. P.either precedes E. or offers hand in response to E. with no delay. 2. P. delays offering hand, but not markedly, and with no obvious sign of lack of comprehension. 3 . P. delays markedly in offering his hand, with or without signs of non-comprehension. b. F. does not respond to offer of E.'s hand. 3• Strength of 1. 2. 3. S. b. grip in handshake. Strength of grip exceeds E.'s. Strength of grip is about equal to E.'s. Strength of grip is noticeably less than E.'s. Little or no grip is exerted by P. Orientation of S.'s glance when shaking h a n d s . When shaking hands, E. looks at P. 1. P. returns E.'s glance entirely, or almost en­ tirely, throughout handshake. 2. P.'s glance is directed part of the time at E . , and partly at self or room. 3. P.'s glance only fleetingly directed toward E. b . P.'s glance directed away from E., or down­ ward, almost constantly. The designation r,E . " is employed throughout as an ab­ breviation for Examiner, as is the designation "P" for P at ie nt . 11 *+ 5. Name patient wants to be called After E. introduces self and confirms P.'s name, E. asks what P. wishes to be called. 1. P. responds with either first name, last name (with or without Mr., Mrs., or Miss), or with nickname, 2. P. attempts to give some answer but with d e ­ lay or indecision. 3. P. says "I don't care" or something similar, or does not respond. 6. Seat taken by p a t i e n t . fwo identical soft chairs are available to P., one being obviously closer to E.'s chair behind the desk. If P. asks which seat, after being invited to sit down, he is told "either one." 1. Closer seat chosen. 2. Further seat chosen. (For purpose of a standard seating arrangement, P. is asked to sit in the seat closer to E. if he chooses the further one, by means of the following statement: M0h, I guess it would be better if you sat here," indicating closer chair.) 7. Obstacle to s i t t i n g . On each of the chairs is a small box arranged that it is possible to si t without moving the but in so doing the sitting position would be cramped. 1. F. removes the box, either giving it to or putting it on the desk. 2. P. pushes box to one side so that there more sitting room. 3. P. sits without disturbing box. 4-. P. sits on box. 8. Response to so box, E., is tack on E.'s c h a i r . Fastened to E.'s chair Is a thumb tack (arranged so that E. apparently sits on it and can produce it'.) E. jumps slightly, picks up tack and remarks "I wo n ­ der how that tack got on my chair." 1. P. laughs or consoles E. in some manner. 2. P. laughs very mildly, smiles, or shows some consolation of non-verbal nature. 3. P. notices situation, but shows no specific reaction of enjoyment or consolation. b , P. shows no apparent response. 115 9. Initial verbal response of P . After P. Is seated and E. orients toward P., he says nothing for about ten seconds. 1. P. gives socially constructive response: Starts conversation, raises E . ’s status, etc. 2. P. asks a question. 3. F. says nothing. b. P. is socially destructive in remarks. 10. Speed of l o c o motion. Time in seconds to walk specified distance. asked to put on room light.) 11. (P. is Response to offer of cigar ette. E. asks "Would you care for a cigarette?", at the same time offering an open pack in his hand. 1. P. accepts with "thank you" or equivalent indication, or declines with similar nicety. 2. P. accepts without response, or declines with inadequate response. 3. P. grabs, or accepts in some other way ne­ glecting recognition of gift from other, or merely refuses by no action at all. 12. P.'s lees crossed or "open" during Initial stages. During the early stages of the interview S. observes predominant position of P.'s legs. 1. Separated and uncrossed. 2 . Touching and uncrossed. 3. Loosely crossed, i.e. foot, ankle, or lower calf of one leg crossed over other leg. U-. Tightly crossed, i.e. upper calf or knee of one leg crossed over other leg. 13• Bodily position of P. during initial stage s. 1. Slouched, or overly-relaxed. 2 . Normally relaxed posture. 3. Stiff posture. compliments P.'s personal appea rance. S. says neat." 1. P. 2. P. 3. P. "I notice that you keep yourself looking thanks E., or apologizes for appearance. nods or otherwise recognizes E.'s comment. reacts inadequately or not at all. 116 15• Response to E . ’s staring at P.'s eves E. looks at P . ’s eyes, waits until P. fixates on E . fs e y e s ,measures number of seconds until P. looks elsewhere. 16. Response to E.'s sudden looking o u t s i d e . E. suddenly turns head at a time when P. is oriented toward him, at the same time turning swivel chair slightly, and looks out the window to his back. 1. P. follows E. with his glance with little or no delay. 2. P. slowly follows S. 3. P. looks at E., or shows some reaction other than above. b. P. shows no noticeable reaction. 17. Response to remark about w e a t h e r . After looking outside, E. returns and orients to P., and gives an opinion on the weather. 1. P. agrees or disagrees, and elaborates with own opinion or comment. 2. P. accepts F,. *s comment, but does not elabo­ rate further than a word or two of agreement or disagreement. 3. P. gives some non-verbal recognition of E . ’s c o mment. '-+. P. shows no noticeable reaction. 18 . Memory of E . ’s statement of number over p h o n e . E. dials artificial number on phone and says in nor­ mal voice (loud enough for P. to hear.) "Oh, say, After E. that number you wanted was 213------ O.K." hangs up he pauses a moment, chen asks P. in puzzled tone "What number was that I told him?" 1. P. tells number correctly. 2. P. tells number incorrectly. 3. ?. d o e s n ’t know number. 19• Response to E . ’s staring at P . ’s f e e t . E. stares at P . ’s feet at a moment when P. is orient­ ed toward E. 1. P. makes some comment re: E . ’s act. 2. P. looks at own feet or at E. 3. P. shows no apparent reaction. 117 20. Response to E . fs need of m a t c h . With an open book of matches on desk right beside P., so that P. could see them clearly, E. puts a ciga­ rette in his mouth and feels his pockets, obviously in search of a light. 1. P. lights E.'s cigarette. 2. P. indicates location of matches or pushes them toward E. 3. P. reacts by watching E. but does not seem to comprehend, or does nothing. *+. P. shows no reaction. 21. P. tells storvf E. interrupts with question. E. asks P. to tell him what he does in the hospital. When P. starts talking, E. asks the following ques­ tion in the middle of one of P.'s sentences: "How old are you?" 1. P. responds immediately with age, correct or not. 2. P. delays noticeably, but responds with age or "I don't know". 3. P. does not respond with age. 2 2 . Continuance of story after interruption. After E. interrupts F.'s story with a question, E. remains silent allowing P. to continue unfinished story (or sentence). 1. P. picks up story in coherent fashion after E . ’s question. 2. P. continues story but resumption has poor continuity with last section before interrup­ tion . 3. P. makes an attempt to continue but can't re­ member where he was. '+. P. makes no attempt to continue the same story. 23. Response to baby p i c ture. E. shows a photo of a baby to P., saying it is his n iece. 1. ?. examines the photo, giving some appropri­ ate response verbally. 2. P. looks at the photo, but makes no comment or a very inadequate one. 3. P. shows no interest response. 118 Sympathy response. 2^. E. has a b a n d a g e on his finger, and shows finder to P. saying that he just got a bad cut. 1. F. shows consoling reaction, or asks for d e ­ tails. 2. P. mav look at finger but shows no sympathy reaction. 25. Response to pencil dropped on f l o o r . E. arranges pencil so as to be able to unobtrusive­ ly cause it to drop on the floor between S. and P., but closer to P. 1. P. picks the pencil up and gives it to E.or puts it on desk with no delay. '2.- F. delays picking it up but does so. 3. P. notices pencil drop, or even searches with glance, but does not p i c k it up. b. P. shows no reaction, except perhaps normal response to noise. . Response to E . fs political o p i n i o n . E. says, after leading the discussion around to current events, "Well, there*s a lot of political news iatel'r. Personally, I*d like to see __________ e l e c t e d ." 1. ?. agrees -or disagrees, and states his own opinion. 2. F. merely agrees or disagrees. 3 . ?. acknowledges by verbal comment, neither his own.opinion, agreement nor disagreement. b. P. shows no verbal -reaction. . P. asks for verbal assistance in recalling something. E. says "Let's see— what's the attendant's name on your ward?" 1. P. responds with little delay with name, cor­ rect or incorrect, or says "I don't know," with an explanation. 2. P. does not respond or does so very inade­ quately, such as "I don't know", with no qualification. 2 po . hehsDesign A . Before showing task, E. asks "I have a very diffi­ cult task here that most people can't do. Do you want to try it?" 1. Yes, or like response. 2. P. says "I don't know" or hesitates. 3 . he, -or like response. 119 19- 5,. After F. accepts task, the construction ci Koh's design .CI, 2. asks "Do you want me to help you?" 1. Yes, or like response. 2. P. says "I d o n ’t know" or hesitates. 3. No, or like response. ;C. hohsDesign C. •After working with P., E. disrupts two of F . ’s blocks saying "these d o n ’t belong there," with a firm voice. 1. F. asserts self with verbal comment, or ac­ cepts the criticism verbally. 2. F. reacts, but does not respond verbally. 3. F. does not react. 31. E. strings beads from Stanford-uinet Scale and asks F. if he would like to try it. Z. 3-. strings 7 beads, square, round, alternating. 1. P. responds with correct number and pattern. 2 . P. uses same pattern, p. ?. uses same number. m-. P. strings beads, but uses neither E . ’s num­ ber nor pattern. 5. P. does not respond. E. reads a list of numbers varying his rate, after F. is asked to read with E . r as closely as he can. (See 1. 2. 3. Pest Instructions.) P. varies closely with E. P. attempts tc vary rate, but inadequate. P. makes slight attempt to vary rate, out with little or no success. 9-. P. does not attempt to vary rate. 33• Task Sequence A . says "Watch closely what I do." (1) E. then takes paper from table between E. and P . , (2 ) puts on window sill, (3) opens window to left of paper, (9-) takes p a p e r ? (5) closes locawindow, and (6) returns paper to o r ’ginal ~ tio n . says "Do what I did." 1 point for each error of omission, addition, and sequence. 120 3^. Task Sequence B . E. says ,rWatch closely what I do." E. puts his hand on the desk, (1) palm up, (2) turns palm down, (3) makes a fist on the desk, (U-) extends two fingers on the desk, and (5) re­ turns to the palm down position. E. says "Do what I did." 1 point for each error of omission, addition, and sequence. 35. Attention T est. (See Test Instructions.) 36. Verbal F l u e n c y . (See Test Instructions.) 37• Social Vocabulary. (See Test Instructions.) 38. Pencil task with dull p o i n t . E. presents a simple design, and asks P. to copy it. The pencil given is so dull that it will make a mark if held upright, but not at an angle. 1. P. reaches another pencil which is closer to E. 2. ?. asks for another pencil, or the one near E, or says "I c a n ’t write with this pencil" (no­ tices before starting.) 3. P. complains of dull point, or that pencil w o n ’t write. b. P. goes ahead with dull point. 39. Clianse of instructions. S. presents a simple design asking P. to copy it (above), and as soon as he has well begun, E. puts another simple design alongside and asks P. to copy it. "Now copy this one." 1. P. changes immediately to new design. 2. P. continues with old design for a while, out changes to new design before completion of first. 3. P. does not change to new design, or net until completing the other. 121 . Personal Appearance. (Person and dress) 1. Clean, well-kempt. 2. Careless, but not slovenly. 3. Untidy, unkempt.. u. Dirty and slovenly. Modulation of M o o d . 1. 2. 3. U-. . Labile, over-reactive. Normal modulation, moderate mood changes. Tendency to flattening, dullness. Plat, apathetic. 2^ box is sitting in obvious view of P . f out of which an a n i m a l 1s tail is exposed. 2 . makes no attempt to direct F.'s attention to it, and it is present throughout the interview. 1. F. physically investigates box. 2. P. asks about box. 3 . F. notices box and makes attempt to discover it;s nature by repeated glances, but does not ask about it. h. p. obviously notices it but does not ask about it. 3. P. apparently does not notice box. Memory of E . ’s name. 2 . introduced self at the beginning of the interview, giving last name. Asks if P. remembers his name. gives correct: name, gives wrong name, m t similar, gives wrong name, and dissimilar, does not remember 2 ,'s name. Understanding of total situation. 2. asks "1Pny do you think I called you down here?” 1. P. gives a logical, realistic response. 2 . P. gives an improbable, but fairly realistic re spcnse. 3. P. gives an unrealistic response or says "I don't knew". 122 h^. Farting response of P . E. says MI guess that's all for today--thanks For your cooperation." 1. P. offers hand, or otherwise spontaneously gives a parting response. 2. F. gives a minor non-verbal parting response, such as nod or smile. 3. P. gives little or no apparent parting re­ sponse . ho. Passage through d o o r . E. and P. approach doer together at the termination cf the interview. 1. P. precedes E. through door. 2. P. waits for E. to go through first. Test Instructions K o n s Design: A. - (Before design is presented) "I have difficult task that most people can't do. want tc try it?" E. - (After P. accepts task) help you?" a veryDo you "Do you want me to C. - (ATter working with F.) E. disrupts two of F.'s correct blocks, saying "these don't belong there", with a firm voice. Bead Stringing; E. strings beads, asking F. "Would you like it?" (7 beads, square, round alternately.) to try lumber R e a ding: "I am going to read seme numbers. I will g ive you the same list and I want you to read along with me. Keep up with me as closely as you can." Task S equence(A and B) "'Watch closely what I do" -- "Do what I did." Attention T e s t : "After each group of 3 numbers in the first, column is a plus or minus mark, or a blank space. The idea is tc put a plus mark if the first number is the largest of the three, and a minus mark if the third number is the largest. Otherwise it is left blank in the answer space. I will show you. now it works. (E. gives six samples.) Now you wil l begin with column 1 and put in the right marks. I will read the letter first sc you can keep track of where we are, and then I will read the number. Re­ member, plus if the first number is largest, and minus if the last number is largest. Otherwise leave the answer space blank. Ready?" 12*+ 3o. Verbal Fluency : "I want you to write as many different words as you can on this paper. Any kind of words will do. I will tell you when to stop." p7. Social Vocabulary; "I am going to read a list of words and I want you to tell me after I read each one whether they are good or bad socially. Say yes if they are words that are friendly or desirable socially and say no if they are unfriendly or undesirable socially." * * * * * * * * * * * * * * * 1. 2. 3. *+. 5. 6. 7S. 9. 10. 11. 12. 13. 1 *+. lp. 16. 17* IS. 19* 20. 21. o2. 23. 2*+. 25. 26. 27. 28. 29. 30. Friend Ridicule Disagree Greet Withhold Conversation Trust Compliment Reject Congratulate Argue Aloof Approve Secluded Group Help Criticize Glare Thank Smile Sympathize Accuse Wave Ignore Force Nag Admire Disturb Shove Respect * Words considered socially positive