<<

PREDICTING SELECTED BEHAVIORAL CHARACTERISTICS ON THE

BASIS OP OBSERVATION OF A GROUP PSYCHOTHERAPY

SESSION WITH MENTAL PATIENTS

DISSERTATION

Presented in Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy In the Graduate School of The Ohio State University

By

THOMAS MUN CHEW CHANG, B. A,, M. A.

*****

The Ohio State University 1957

Approved by:

Adviser Department of Psychology ACKNOWLEDGEMENTS

There are many individuals and groups of people to

whom I am deeply indebted for their various contributions

to this study.

Dr. Charles Lynch, as supervisor of my psycho­

therapy work and personal friend, contributed to the

fashioning of an investigation from the first nebulous

hypotheses conceived. Dr. Jack Basham, as Chief of the

Vocational Counseling Service at the Veterans Administration

Hospital, Chillicothe, Ohio, gave his full administrative

and professional support. The nurses and aides in the Exit

Service Program and the rehabilitation therapists in the

clinics gave generously of their interest and effort in making criterion ratings and behavior and therapy perform­

ance reports. The Registrar’s Office was always ready to provide record folders and other data on patients when requested. Mr. Robert Sone, Mr. Edward Polder, and Dr. Fred

Wright, as Counseling Psychology trainees, performed as observers and raters in the study and were actively inter­ ested in the proceedings.

The consultants from The Ohio State University con­ tributed significantly to the entire project. Dr. Collins

Burnett and Dr. Francis Robinson participated in the design of the study and acted as observers and raters in the group

ii H i

therapy sessions. Dr. John R. Kinzer also acted as an

obser ver and rater in the summer of 195b prior to his going

on l e a v e .

Dr, Robert J. Wherry of The Ohio State University

advised the writer in the statistical procedures in the

analysis of the data. Dr, Roy Reeves, Director of the

Statistical Computation Laboratory of the Ohio State Research

Foundation, and his staff gave unstinting help in teaching

the w r i t e r to prepare and analyze the data with I. B. M. methods on the Type 6 5 0 magnetic drum computer.

The psychiatric patients of the psychotherapy groups must understandably remain anonymous. But the roles they played, in the research are crucial and it is regretted that acknowledgment cannot be more specific. Furthermore, for the writer, working with them has contributed to a greater understanding and appreciation of psychotherapy and to an increased faith and admiration in the strength and resources that m a y lie within the individual.

A special acknowledgment is reserved for my adviser,

Dr. F r a n k M. Fletcher, Jr. , who guided the planning and execution of the study from its inception to the description in this dissertation. His patient and constructive criti­ cism, professional competence, and infectious enthusiasm are m o d e l s of research behavior that this writer will not forget a n d hope to take into his own work. lastly, to ray wife and two sons, ray humble apprecl tion for their many sacrifices that made my studies possible. TABLE OF CONTENTS

Chapter Page

I. INTRODUCTION ...... 1

Statement of the Problem Hypotheses Tested

II. REVIEW OF THE LITERATURE...... 12

Direct Observation of Behavior Rating of Behavior Research of Small Groups and Group Psychotherapy

III. PROCEDURES OF THE STUDY...... 42

The Subjects The Observers

The Prediction Items Description of the Criterion Measures and Scoring Procedures Analysis of the Data

IV. FINDINGS OF THE STUDY...... 92

Data on the Accuracy of Prediction Data on the Improvement in Accuracy

V. DISCUSSION OF THE FINDINGS ...... Ill

Accuracy of Prediction Improvement in Predictive Accuracy Some Related Findings

VI. SUMMARY AND CONCLUSIONS...... 151

The Problem The Present Study Findings Suggestions for Future Research

APPENDICES...... 169

BIBLIOGRAPHY ...... 191

AUTOBIOGRAPHY...... 203 v LIST OP TABLES

Table Page

I Items on which the Pairs of Raters had Complete Agreement on the Rating of Pacient B ...... 68

II Percentage of Items on which the Pairs of Raters had Complete Agreement on the Practice Rating of Pour Patients...... 69

III Intra-rater Reliability among Therapists .... 75

IV Intelligence Test Data on the Sample ...... 79

V Arrays of Means and Standard Deviations for 22 Measures Obtained from Psychologists, Nurses, Aides, Therapists, and other Sources ...... 93

VI Two Sets of Means and Standard Deviations on 22 Items Obtained Three Months Apart from Psychologists, Nurses, Aides, and Other Sources...... 96

VII Correlations between Predictions and Criterion Measures...... 98

VIII Correlations between Predictions and Criterion Ratings...... 100

IX Correlations between Predictions and Psycholog­ ical Test Measure, Sociometric Measures, Time Sample, and Hospital Clinical Records . . 101

X Two Sets of Correlations Between Predictions and Criteria Obtained Three Months Apart . . . 103

XI Seven Items with the Largest Increases in Accuracy...... 107

XII Items Significantly Predicted or. Both Trials . . 108

XIII Unsuccessfully Predicted Items ...... 109

XIV Means and Standard Deviations on the Items on Friendship and Leadership Preferences of the Two G r o u p s ...... l4l

vi CHAPTER I

INTRODUCTION

The Development of Psychological Counseling;

At the present time, professional counseling services are being carried on by psychologists, by physicians, priests, and lawyers, and by others whose counseling func­ tions are less clearly defined such as the social workers.

Though methods and emphases vary, each in its own way seeks the identical goal of assisting his client to achieve a greater degree of personal adjustment and harmonious Inter­ action with his environment.

The history of psychological counseling is a rela­ tively short one. Super (115) has traced the present status of counseling psychology back to three historical sources.

One of these origins lies in the work of Frank Parsons and the social agencies of Boston which provided occupational guidance to its youth. This interest In vocational adjust­ ment was joined by the efforts of the Minnesota Employment

Stabilization Research Institute during the Depression to develop psychological tests, occupational information, and vocational re-training to enhance the employment potential of individuals. The third movement was the interest and research In psychotherapy which took place primarily under

the auspices of clinical psychology and contributed materi­

ally to the understanding and refinement of the counseling

process.

Even within the confines of a vocational guidance

orientation of pre-World War II, counseling was regarded as

the crux of the process. The counseling interview was the meeting place between counselor and client where the latter was helped to understand and accept his abilities and weak­ nesses assessed by tests, to work through idealistic

aspirations and settle for a realistic goal, and to spell out a plan of action. At the present time, the importance of the counseling function in counseling psychology is reflected in the very title of this area of specialization,

In the existence of its own journal of publications, and in its inclusion as a topic In the Annual Reviews of Psychology.

Furthermore, the settings in which counseling may take place have been broadened to include industry, hospital, armed services, mental hygiene clinics, private practice, and, finally, the difference between the concepts of counseling and psychotherapy processes are narrowing.

Counseling Compared with Psychotherapy

Though this study is centrally concerned with the counseling process; the experimental task takes place in a therapeutic setting traditionally called group psychotherapy. Hence It may be profitable to spend a few paragraphs deline­

ating the similarities and differences between the processes

of counseling and psychotherapy as viewed by various repre­

sentatives in the field.

Pallor (30) for example, believes that psychotherapy

differs from counseling in that the latter concentrates on normal individuals, on breadth rather than depth of relation­

ship, and uses a broader range of tools and techniques.

Counseling also makes greater use of co-operative relation­

ships with community agencies, and in vocational counseling, the family’s socioeconomic status is an important variable.

Rogers, on the other hand, has been an early exponent

(1942) of the view that the two processes are highly related if not synonymous:

There has been a tendency to use the term counsel­ ing for more casual and superficial interviews, and to reserve the term psychotherapy for more inten­ sive and long continued contacts directed toward deeper reorganization of the personality. While there may be some reason for this distinction, it is also plain that the most intensive and success­ ful counseling is also indistinguishable from intensive and successful psychotherapy. (9 6 , pp.3“4)

T h o m e (125) regards counseling as a form of psychotherapy, and both were distinguished from "depth psychotherapy" which used intensive psychiatric or psychoanalytic methods.

Robinson's definition, which has an emphasis on student counseling and is similarly broad in its scope, considers therapy as an aspect of counseling: Counseling . . . covers all types of two person situations In which one person, the client, is helped to adjust more effectively to himself and to the environment. It includes the use of the Interview to obtain and give information, to coach or teach, to bring about increased maturity, and to aid in decision-making and therapy (95, P. 3).

An excellent example of the present ambiguity and tendency toward synthesis of these two terms is shown in the discus­ sion on counseling and psychotherapy in the editions of the

Annual Review of Psychology. In the 1954 volume, Wrenn

(author of chapter on counseling) and Saslow (author of chapter in psychotherapy) admitted that they would not attempt to define either term. They agreed to a categorical distinction that the counseling chapter would not Include a review of studies and papers dealing with a medical setting.

In the following year, the succeeding editors (Hobbs and

Seeman on counseling and Meehl on psychotherapy) again stated this distinction explicitly. However, in the 1956 volume, Shoben (writing on counseling) and Harris (writing on psychotherapy) did not enter into a discussion of defini­ tions or distinctions, though Shoben1s review definitely emphasized non-medical studies.

In the most recent volume, 1957, the trend has come to a full circle. Shaw (writing on counseling) has distinct­ ly recognized the emerging Importance of the counseling function in medical settings and has called for a rapproch- ment between the counseling and clinical specialities. If mental Illness is considered the province of the clinical psychologist and creative health the domain of the counseling psychologist . . . then (they) share . . . the common enterprise of what might be called developmental counseling. This developmental perspective can be discerned in papers . . . written by counseling psychologists vitally concerned with the rehabilitation of the mentally ill. Their interest . . . is not 30 much in “curing" or "treatment" as it is in opening up new opportunities for constructive life experiences and maturation cf personal resources. The impli­ cation of this kind of interest in psychotherapy . . . cannot be escaped even though (they) . . . do not profess to practice psychotherapy. The conceptual trends emerging . . . bypass distinc­ tions between counseling and psychotherapy and counseling psychology and clinical psychology . . . (105, P. 357).

Only time will reveal the outcome of this present propensity

for synthesis of therapeutic practice and areas of speciali­

zation. By then the forces promoting for a coalescence may

also be identified more clearly. At the present time, cer­

tain individuals, Super (114), for example, have spoken out

against a tendency for a loose distinction between the clinical and counseling areas.

In spite of the present predispositions, this discus­

sion would not be complete without presenting two repre­

sentative hypotheses which have affirmed the basic differ­ ences between counseling and psychotherapy. Bordin (13) has

suggested that the two processes may be distinguished on the basis of the "ambiguity-structuredness" dimension, with psychotherapeutic relationships being these containing a greater degree of ambiguity and hence being more intense and more encompassing of the client's life and experiences.

Mowrer (76), in an earlier and more complete formu­

lation, has differentiated them on the basis of anxiety

theory. He defines counseling as the process of giving

professional help to persons suffering from fully conscious

conflicts which are accompanied by normal anxiety; that is,

the anxiety is situational and has an objective basis for

reference.

Normal anxiety passes over into neurotic or clinical

anxiety when the anxiety is dissociated from the situational

context and enters a latent phase where the dissociations

are working and the painful materials are successfully repressed. However, the latent neurosis erupts into full­ blown manifest neurotic symptoms when the dissociations collapse and the repressions threaten the individual1s entire organization. Depression, mood swings, anxiety attacks, and inferiority feelings are all expressions of the manifest neurosis as the individual seeks to control or avert the neurotic anxiety.

The task of psychotherapy is one of reversing the dis­ sociating trend, replacing neurotic anxiety with normal anxiety and substituting conscious conflicts for unconscious ones, so that the client may bring normal rational problem- solving devices to bear upon them. Consequently, for

Mowrer, the latter stages of psychotherapy become "post­ neurotic counseling." Counseling In Psychiatric Rehabilitation

The establishment of the Vocational Counseling Ser­ vice within the Veterans Administration hospital system as an integral part of the treatment team represented the first major recognition of the special contribution that counseling may make in the rehabilitation of the mentally ill. As Super has stated, . . there is considerable evidence which shows that counseling can deal uniquely and effectively with abnormal persons by treating their normal­ ities rather than their abnormalities1’ (114, p. 2 7 3 ).

Fletcher (32) has sought to establish the principal roles that counseling may come to play in rehabilitation.

He suggested that the greater contribution may be made in the second half of the rehabilitation process, which he titled the "adjustment phase." In this period, the active treatment plans are mostly concluded and the disability Is considered stabilized, and counseling and other related therapies carried on are goal-directed toward eventual return to the community. Needelman (80) has also under­ scored the latter point stating that Counseling Psychology*s three major functions of evaluation, counseling, and place­ ment should be carried out with the ultimate goal of eventual re-integration of the patient Into a job and his family.

Recently these functions have been extended to a greater degree into the treatment aspects of psychiatric rehabilitation as various counseling philosophies proved

congruent with the therapeutic community approaches carried

out by Jones (59) and extended by others. Hence, where

Jones has demonstrated the successful rehabilitation of neurotics in his therapeutic community, V/right (136) has

applied the techniques and concepts successfully in the rehabilitation of long-term psychotics, employing a total push approach, which included intensive counseling, environmental manipulation, free use of incentives such as money and other privileges, and a rehabilitative emphasis in the development of the assets and positive goals of each individual.

One consequence of these broadening functions is that the counseling and therapy skills of the hospital counseling psychologist have come to play a more and more important part in his work. More than ever before counsel­ ing is being used as a treatment tool to bring the patient to a level of recovery where vocational and discharge plan­ ning become feasible. It is in the hospital setting that the distinctions between counseling and psychotherapy are disappearing most quickly.

Statement of the Problem

The present study was interested in assessing more precisely the accuracy of the hunches and guesses that the counselor made about bis client during the interview. The verification of relatively specific and concrete predictions and guesses has been a neglected area of research in the

study of the counseling process., and, in this regard,

FJosenwald has commented as follows:

In our assessment of people, we are seemingly at a loss predicting such elementary things as hov: the person gets along with other persons, whether a patient is liked or disliked by other patients, what sort of people he likes or dislikes, who the leader of the psychiatric ward may be, how the patient gets along with certain figures, etc, etc. It is not unreasonable to propose that validation studies should concern with such problems (98, P. 314).

In contrast to the kind of validation which Rcsenwald is advocating, most research on predictive accuracy has been devoted to the validation of the prediction of gross or molar events of a complex nature that took place In the

Immediate or distant future. Meehl's recent volume (70) comparing the relative merits of clinical versus actuarial predictions reported that the studies used mostly three types of criteria, (1) success in schooling or training,

(2) likelihood of recidivism, and (3) likelihood of recovery from a major psychosis. In contrast to the above types of molar or complex events, the present 3tudy was

Interested In validating the estimates and hypotheses that were "micro-events" fc-r which no statistical table cculd be prepared in time to be useful. These spontaneous predic­ tions are an integral part of the therapeutic counseling process and evolve naturally out of the give-and-take 10 between client and counselor, and they possess a therapeutic value that is generally agreed upon.

Nobody knows what the payoff rate is for these moment-to-moment guesses that comes to therapists; but the over-all success frequency might be con­ siderably less than 50?a and still justify the guessing, for . . . the time spent in exploration of poor guesses need not greatly distract from the positive contribution of successful ones. Presumably even the unsuccessful paths are rarely pure waste, since they contribute to such diverse concurrent aims as further getting acquainted, general desensitizatlon, and incidental support for quite unrelated constructions (7 0 , p. 121).

However, when these subjective predictions are employed for the purpose of diagnosis, then an inaccurate hypothesis simply lowers the success frequency. It was believed that the empirical assessment of the accuracy of these specific hunches and guesses would contribute to a greater under­ standing of the disappointing performance of clinicians in the prediction of complex events, since observational and interview data are usually obtained and incorporated into the clinical predictions of performance. This study did not investigate the characteristics of the predictors.

It restricted the task to evaluating the accuracy with which specific behavior and adjustment characteristics and more complex variables such as present Intelligence, interpersonal relationships, level of social Interaction, and likelihood of release could be predicted from a minimal amount of Information. The experimental setting was a group therapy session in a mental hospital, and the subjects were 11

schizophrenic patients.

Hypotheses Tested

No. 1 A group of psychologists who observed a session of group psychotherapy with mentally-ill patients whom they did

not know, will be able to make predictions on fifteen items concerning specific behavioral characteristics and psychia­

tric symptoms that will correlate significantly with the corresponding criterion ratings obtained from hospital per­

sonnel who worked with the subjects.

No. 2 A group of psychologists who observed a session of group psychotherapy with mentally-ill patients whom they did not know, will be able to make estimates of present func­ tioning intelligence, friendship and leadership standing in the therapy group, level of social interaction entered into on the ward, likelihood of release from the hospital, and adjustment on the outside which will correlate significantly with the corresponding psychological test measure, socio­ metric measures, time sampling record of behavior, hospital records of release, and social service report of adjustment on the outside.

No. 3 Given the same experimental task described in Hypoth­ eses No. 1 and 2, and using the same subjects, a second set of predictions made three months later will show a signifi­ cant increase in predictive accuracy when compared with the first set. CHAPTER II

REVIEW OF THE LITERATURE

The research literature that may appropriately be

classified under the aegis of study of prediction of behav­ ior is voluminous. This review has been selective and presented studies of prediction that were considered to be relevant to this study In one or more of the following ways:

Direct observation of behavior.— Since the estimates and predictions in this study were based entirely on the direct observation of behavior in group psychotherapy, some of the major investigations that have used this technique have been presented. These Included the earlier time sampling studies, the work of the Office of Strategic

Services Assessment Staff, the Character Education Inquiry, and Tindall's study on the comparability of adjustment indices.

Ratings of behavior.— The rating scale was the prin­ cipal instrument used in the present study. In this section, the inherent difficulties as well as the popularity of this measurement technique were discussed. The Veterans Adminis­ tration Clinical Psychology Assessment Study and the recent investigation of Hilton on the prediction of success in

12 13 industrial management were presented. A description of the

current use of ratings in psychiatric settings concluded

the discussion.

Research of small groups and group psychotherapy.—

First, the origins of small group research, as distinguished

from group therapy research, were briefly outlined, then

sociometric studies of normal groups employing variables

similar to the present study were reviewed, according to

the setting in which the study was performed; namely,

school, military, or industrial.

Since this study took place within a group therapy

setting, the concluding section of the review has been

devoted to a presentation of different variables that were

functions of the group therapy process or structure. These

were divided into the following five main areas: (1) out­

comes of therapy with different groups, (2) comparison of

effects of different therapies, (3) studies of the processes

of therapy, (d) role of the therapist, and (p) the selection

of members for therapy. It was believed that this brief

discussion would assist in better understanding and inter­ preting the findings obtained in this study.

DIRECT OBSERVATION OF BEHAVIOR

The technique of direct observation is employed con­

stantly in studies of behavior. Not only is this type of data more readily obtainable than others, but it also 14 provides information on the individual's typical or normal behavior and adjustment characteristics which cannot be ascertained by tests or other means.

There are two serious errors to accurate observa­ tion, and Cronbach has referred to them as "sampling” and

"observer" errors (24, pp. 386-37). Error in sampling occurs because of the specificity of situations which change from day to day, from moment to moment. Furthermore, the behavior elicited is always a function of the background situation. Wandt and Ostreieher (134), for example, have demonstrated that the behavior of school children is a func­ tion of the "social-emotional" climate of the classroom.

Observer errors are the result of human inability to take in the entire complexity of interaction at a given moment, and these errors may occur systematically in the form of either omission or over-emphasis.

The Time Sampling Technique

One of the techniques of direct observation is time sampling, which Arrington has defined as:

. . . a method of sampling the behavior of an individual or group in which the occurrence of specific overt acts is recorded in a series of time periods of uniform length in such a way as to yield quantitative measures descriptive of the characteristic frequency or variability of the behavior . . . (4, p. 3).

Use of this technique has diminished in recent years and the 15 investigations of Goodenough (3 6 ), Olson (84), Thomas (123), and the monograph by Arrington (4) are still the classic references. However, the recent volume by Barker and

Wright, One Bov's Day (8), reveals a continuing interest among small groups of Investigators.

Goodenough (3 6 ) reported success in recording behav­ ior samples of physical activity, talkativeness, laughter, compliance, social participation, and leadership. Olson

(84) studied the incidence of nervous habits In groups of children and the amount of these habits in individuals.

Olson was interested in highly specific behaviors, such as nallbiting and thumbsucking, and demonstrated the feasibil­ ity of a normative study of behavior when the characteris­ tics could be explicitly defined and easily observed in controlled situations. Olson's sampling was probably biased, however, since all of the observations were recorded In one day. A study of Olson's methodology contributed importantly to the present study, helping in the formulation of a list of specific activities that the hospital aide and therapist may check off in their observations. This simplification of the recording aspects made it possible to use this technique in this study.

Arrington's monograph reviewed the findings up to that time, 1939, and performed a definitive study of the varia­ bles unique to the method. She sought to establish the 16 optimal length of observation, number of observations, and number of behavioral characteristics to be observed at one time. Arrington also studied the effects of situational variability and the applicability of the technique to dif­ ferent age levels. She suggested the use of five-minute observation periods and demonstrated that good agreement between observers could be achieved, up to 83 per cent and yO per cent levels.

A study reported by Tindall (126) employed the time sampling method as well as sociometric measures and behav­ ior ratings, all of which were employed in the present study. Tindall sought to determine more precisely the degree of agreement or disagreement which there was among different adjustment measures. He used a sample of white male adolescents residing in an orphans' home and found that the two sociometric measures, a Guess Who Test and a

Companion's Choice Test, correlated more highly between themselves ( r = .625) than with the other indices, suggest­ ing that sociometric analysis may be one of the more accu­ rate methods of assessing adjustment and other personality characteristics of individuals. Tindall also found that the two sets of adjustment ratings made by supervisors were in significant agreement ( r = .5^5). Two different rating instruments were used, and this finding provided evidence for the validity and reliability of the observation method 17 and the rating technique. Furthermore, the ratings were

signficantly correlated with the sociometric measures

(.369 to .630) which indicated that the characteristics of

healthy adjustment perceived by the raters were also the

characteristics which 'were preferred by the boys themselves.

Lastly, the adjustment indices were also correlated

with intelligence, chronological age, and length of resi­

dence, which Tindall referred to as the "variables of

unknown effect." He found only four significant correla­

tions out of a possible forty-eight which were regarded as

due to chance.

Observation of Situational Tests

Situational tests differ primarily from direct obser­ vation in that the former use standardized task situations to elicit reactions which can be observed, recorded, and rated. The advantages of standardized tasks are that they make it possible to observe characteristics which may appear only infrequently, such as reaction to cheating opportun­ ities. This technique, furthermore, represents an experi­ mental refinement of direct observation in the greater degree of control attained. Practical use of the situa­ tional test has been limited because of its excessive cost, difficulty in design, and complexity in administration. Is has been employed moss often in military assessment; for 18 example, by the German military psychologists (3), the

U. S. Airforce (37), and the Office of Strategic Services

(128).

The following paragraphs present some of the find­ ings of two of the most comprehensive investigations employing the situational test technique. These are the

Office of Strategic Services (O.S.S.) study and the

Character Education Inquiry (43, 44, 45). Both of them examined variables that were common with the present study.

In the O.S.S. project, the general procedure of assessment was to use situational tests, intensive inter­ views, paper and pencil tests, projective techniques, questionnaires, personal histories, and sociometric measures to obtain primary data on each candidate from which the final ratings and recommendations were made. Of the eight variables rated, the following four had the most relevance to the present study: effective Intelligence, emotional stability, social relations, and leadership ability. The staff found a low positive correlation of .24 between intelligence and emotional stability, the latter being considered as a measure of personal adjustment. On the other hand, emotional stability correlated .54 with social relations. This finding was the highest correlation coefficient obtained and the staff interpreted it as sup­ porting their hypothesis that these two variables were related, that Is, good social relations provided an accept­ ance and approval of others which in turn enhanced personal emotional stability. Finally, the Assessment Staff's rating on social relations did not correspond with a sociometric measure of friendship choice among the candidates; the cor­ relation was a low .24. This finding suggested that the qualities of friendship rated by the staff were not those used by the assessees. On the other hand, the staff's per­ cept of leadership ability corresponded very closely with a sociometric measure of this variable (r = .6 8 ).

In 1924, Hartshorne, May, Mailer, Shuttleworth, and others began their investigation at the request of the

Institute of Social and Religious Research, and their three volumes, generally referred to as the Character Education

Inquiry. are still regarded as a landmark in the analysis of character traits by situational and other tests. The traits studied were Tendency to be of Service, Tendency to Deceive, and Self-control. The authors reported that there was a barely perceptible relationship between intelligence and service (r = .1 6 1 ) and that there was also a strong tendency for cheating to decrease as the level of intelligence increased (r =-.50 to -,6 0 ). Lastly, It was found that the soclometric measure of friendship choice and tendency to be of service were essentially noncorrelated, but friendship and deceit were highly correlated. For example, correlation 20

between deceit and friendship choice among members of the

same class was ,662, with friends in the same community

.652, and with friends in different communities, .486.

RATINGS OF BEHAVIOR

Rating is defined as the process of judging people

or things, and a rating scale is a standardized technique

to record judgments. This method of assessing behavior is

a popular one and is being used in all types of settings.

In Industry it is employed for selection, promotion, and

evaluation, in academic institutions, for admissions, and

in clinic and hospital settings, for the recording of

psychiatric symptoms and behavior characteristics. It is

used for prediction purposes, to provide criteria, and as basic sources of data in all types of research.

Patterson has reported briefly on the history of

this technique (85). In 1907, Pearson devised a rating

scale to obtain estimates of intelligence. Then World War I gave great impetus to its development. The Man-to-man com­ parison scales were devised and were later shown to be excessively unreliable in the judging of officers. Today, their use is limited to the rating of products. Graphic rating scales which are simpler and more straightforward in their procedure require only that the rater makes a check­ mark along a continuum of behavior description. An Improve­ ment of the graphic scale was the Richardson-Kuder Scale, in 21 which each step has a weighted value as determined by the

Thurstone and Chave method of scale construction. This

technique which enhanced the specificity, discriminative

capacity, and the consistency of the items yielded high

reliabilities.

Another rating scheme, named the forced choice method, was reported by the Personnel Research Section, AGO,

in 19^6. This technique provided the rater with a dyad,

triad, or tetrad of items, all apparently equally favorable

or unfavorable. It forced the rater to check the one item most descriptive (and least descriptive, if triad or tetrad)

of the ratee. Each item had a differently weighted value,

depending upon its discriminative power with the criterion

established. This method has demonstrated high validity and reliability. Its chief drawback is that a special

scale must be constructed for each rating situation, which is expensive and time-consuming.

Over the years, the graphic rating scale has con­ tinued to be one of the most popular methods of rating. It is easily constructed to meet specific requirements, ana the direct manner with which it refers to the rating situation appeals to the rater. The data are easily quantified for statistical analysis, and an apparent uniformity of ratings

Is secured. 22

However, these advantages are mostly ephemeral. The

difficulties Inherent with this method are well identified

now. Thorndike, for example, named these five factors that

affected the accuracy of the ratings: (l) ambiguity of

variables rated, (2) overtness-covertness of traits rated,

(3) opportunity to observe, (4) uniform standard of refer­

ence, and (5 ) specific rater idosyncracies. Other distor­

tions lowering validity were "generosity errors" which

occurred when the rater was unwilling to rate any individual

on the low end of the scale, and "halo errors" which

occurred when the rater tended to rate in terms of over-all

general impressions (124).

Reliability of Ratings

In 1931, Symonds (116) examined a number of studies

and concluded that the correlations between ratings given

by two independent raters clustered about the coefficient

of .55. At the present time, the reliability coeeficients

of ratings still vary over a wide range and must be estab­

lished for the individual study. Thorndike has recommended

the arithmetic pooling of independent ratings to Increase

reliability.

The phenomena of judgmental fatigue and judgmental

disorganization and their negative effects on rater relia- bilityhave been receiving increasing attention in the liter­

ature. Judgmental fatigue is cumulative, not affecting the first judgments of the rater, but increasingly reducing the accuracy of the judgments elicited. Like motor responses, this loss of judgmental adequacy has been attributed to boredom, decreasing motivation, or the accumulation of waste products. The U. S. Army, in a study of the validity of military ratings (1 2 7 ), has shown that there was an initial warm-up effect in a rise in validity between the first and second group of five ratings, then a subsequent decrement in validity for the third and fourth group of five ratings.

Bendig (11), on the other hand, has reported a study of this phenomenon on the rating of food preferences and found no decrement in reliability when the variables of length, list, and trial number were manipulated separately.

Judgmental disorganization occurs when the subject, in surveying the entire judgmental task he is to attempt, perceives it as being too extensive and complex for adequate handling. Cummings (25), in a clinical study evaluating the judgment of adjustment from Rorschach single-card perform­ ance, stated:

Many judges who participate in clinical judgment studies report feelings of dissatisfaction and of being overwhelmed by having to retain and manipulate so many complex variables simultane­ ously and having to continue to do so for what is perceived as a mountainous mass of protocols (25, p. 246).

Sorenson and Gross (111) have also observed the effects of judgmental disorganization in their study of rater reliabil­ ity. They also noted a diurnal inconsistancy in ratings 24 in spite of attempts to maintain objectivity:

Even more disturbing, at least to some observers, was the discovery of inconsistencies within themselves. Moods or transitory physical states changed from day to day the carefulness with which each observer made his ratings— changes his atten­ tiveness and the degree of tolerance or critical­ ness with which he regarded behavior (110, pp. 367-68).

Validation of Ratings

The validation of ratings to an independent measure is often neglected, especially when the ratings themselves are used as criteria. However, the fact that such independ­ ent verification is imperative is evidenced in the wide range of validation coefficients reported. Almy and

Sorenson (1) described a graphic rating scale of 20 traits rating teaching success which correlated .45 to .74 with practice teaching marks. Driver (28) reported data on the validity of ratings of employee performance and demonstrated significant difference in the rating of superior and infer­ ior groups. Examples of contrasting findings are those of

Johnson (58) in which the manager's ratings of 34 salesmen correlated only .16 with actual sales record. Wagenhorst

(112) found a correlation of .23 between practice teaching ratings and first year service ratings. Correlation of first year service ratings correlated .001 with intelli­ gence and .01 with grades. 25 Strang (112) has suggested the following precautions be taken to increase reliability and validity of the rating

process: (1) understand the limitations of the tecnnique,

(2) Improve the construction of the scales used, (3) use descriptive summaries to supplement ratings (4) create favorable conditions for ratings, (5) give instructions for ratings, (6) understand the nature of the situation under which the ratings are made, and (7) encourage practice by performing self-ratings.

Prediction Ratings

The prediction of future performance has been recorded in the form of ratings in various studies. The U. S. Mili­ tary Academy used ratings to record predictions of combat performance and found a correlation of .50 with criterion ratings made after observation of the ratee's effectiveness in combat in Korea (88).

The study reported by Kelly and Piske on the selection and evaluation of trainees in the Veterans Administration

Clinical Psychology Program (61,62) also chronicled the dif­ ficulties encountered in the use of ratings* With their large group of raters (over forty universities participated in the study) they found clear-cut variations among the ratings and interpretation of the traits rated. Seme uni­ versities consistently rated their trainees higher or lower 26

than other schools, and these differences were significant.

Furthermore, though the reliability of the criterion rat­

ings ranged widely from .27 to .75 and was regarded as

satisfactory, a closer inspection and a factor analysis of

the ratings revealed that this apparent good agreement was mostly in csrms of a general factor and not in terms of

specific erirerion skills.

The authors correlated a set of ratings of overt behavior with the Guilford-Martln Inventory of Factors

GAMIN and obtained a range of correlations from .58 to lower. These r*s were disappointing and corresponded with the findings of Tindall who also reported a lack of rela­ tionship among his indices of adjustment.

The V. A. study used the method of final pooled rat­ ings also, that Is, prediction ratings were made after a conference among the staff members and were based on all the data obtained. It was found that these ratings had a median validity value of .3^ (product-moment correlation) only, and this "grand slam" effort to achieve maximal pre­ dictive accuracy must be adjudged as primarily unsuccessful.

In fact, on one variable, the pooling conference resulted in a drop of the coefficient from .33 to .2 6 .

An ir.reresting on-going examination of the processes involved in arriving at predictions were portrayed when

Kelly and Fiske traced the alterations in validity coeffi- 27 cients when increasing amounts of data became available to

the assessors. They reported tnat the relationship between predictive accuracy and quantity of data was generally a non­

linear one, that is, with increasing amounts of data, the

level of accuracy may increase or decrease. In a sequence

of ten predictions, optimal validity was achieved at the

sixth step.

A prediction study performed in an Industrial setting was reported by Hilton, Bolin, et al. (47), who examined the accuracy of forecasting performance in engineering, sales,

sales training, and managerial positions. Predictions were recorded in the form of graphic rating scales on the vari­ ables of sociability, organizational ability in his work, drive, over-all performance on present job, and potential for advancement.

The predictions were validated against criterion rat­ ings on the same variables obtained from the subject's

"closest supervisor." The number of criterion ratings secured for the subjects varied from one to three, and were averaged in the correlational analysis. The validity coef­ ficients were low and ranged from .21 for sociability to

.38 for potential for advancement, all significant at the

.05 level or better for the sample of 100 subjects. A definite tendency to rate with a "halo" by both the predic­ tors and supervisors was reported, and the authors regarded 28

their findings as promising and as supporting the method of

clinical prediction.

In general, the validity coefficients of prediction

studies have been of the magnitude found in the V.A., O.S.S.,

and Hilton studies and have led to a questioning of the

efficiency of the clinical predictions by critics such as

Meehl (70), TIedman (71), and others. Cronbach (23), how­

ever, has argued that these low validities are of value if

the procedures cover many dimensions as compared with a

single accurate test which answers only one question.

Ratings in Psychiatric Evaluation

The use of rating scales in the present study to

record the predictions of behavior and the criterion obser­

vations of hospital personnel is in line with the long

popularity of this technique in the assessment of behavior

and behavior changes of psychiatric patients. The earlier

rating charts and scales have been in existence for some

fifty years now. These include the Phipps Psychiatric

Clinic Behavior Chart reported in 1915, a "Rating Scheme of

Conduct" for ward behavior (90) devised by Plant in 1922,

and Moore's "Schema for the Quantitative Measurement of

Abnormal Emotional Conditions," 1933 (73)•

The great impetus in interest in mental health and mental illness with the advent of World War II has also 29

spurred the development of more accurate and objective measurement of psychopathology and personality change. In the realm of rating methods, Lorr (6 7 ) has reported on some

20 scales developed between 1944 and 1954, and since the latter date, additional scales have appeared.

The reasons for the popularity of rating scales in the psychiatric setting are similar to those presented in the introductory section of this discussion. Not only are they easy to construct, to understand, and to perform, but they are also the most appropriate for the recording of observations of behavior and remain one of the crucial tech­ niques In the assessment of the effects of therapy.

In addition to the above reasons, the increasing interest and use of ratings have stemmed from a growing con­ cern for the need to improve on the objectivity and communi- cability of clinical interview findings and to develop a pattern of interviewing so that information is elicited adequately and in an orderly manner. The use of scales and checklists to record clinical judgments of behavior and inferences of attitudes obtained in interviews provides a record in more understandable terms and cuts through the differing and often amorphous, theoretical orientations of individual staff members. Such, a record also provides an objective framework and basis from which therapeutic plan­ ning may proceed. 30

The psychiatric rating instruments in extant may be

divided into two groups, those appropriate for less trained

personnel such as aides, nurses, and others, and those

scales suitable primarily for staff members with more ex­

tensive clinical training.

One of the best constructed scales for use by nurses

and aides is the Hospital Adjustment Scale developed by

Ferguson, McReynolds, and Ballachey (31). It consists of 91

statements that were selected empirically from a large pool

of items and can be completed in ten minutes. It yields

measures describing hospital adjustment, communication and

interpersonal relations, care of self, social responsibility,

work recreation, and other activities. Validity, reliabil­

ity, and norm data based on 518 patients from four hospitals

are available.

Forty-four scales for the evaluation'of behavior in

Physical Medicine and Rehabilitation (P.M. R.) activities,

such as Occupational Therapy, have been described by

Scherer (102). The value of these scales is enhanced by the

fact that they are also suitable for the rating of post-

hospital adjustment, thus providing a convenient Index of

comparison of behavior changes.

Examples of raring scales suitable for use by psychi­

atrists, psychologists, and others with more extensive clinical training, include the Elgin Prognostic Scale (130) constructed and validated by Wlttman and Sternberg. It is

designed primarily to predict recovery from schizophrenia

and has 20 rating scales which have different positive and

negative weights in the scoring. These scales contribute

to an algebraic total on which the prognosis is based. The

authors have reported or. the validation and cross-validation

of the Elgin Scale by comparing its predictive accuracy

(actuarial) with that of the staff"s judgment (clinical).

Findings indicate that the Elgin Scale was definitely sup­

erior. It achieved 81 per cent accuracy compared with a

median of 44 per cent accuracy reached by the psychiatric

staff (139).

The multidimensional Scale for Rating Psychiatric

Patients (M.S.R.P.P.) is a research product of the Veterans

Administration (137)* There are two forms, one for out­

patient use and another for hospitalized subjects, names the

"Hospital Form." The rating scales used in the present

study were derived from Items in the MSRPP, Hospital Form,

and these items will be described separately in detail in

the chapter on procedures.

Research on the MSRPP, particularly the Hospital

Form, has been extensive. Norms, standard scores, and pro­

files of psychiatric syndromes are available. Its reliabil­

ity has been demonstrated in a study by Rowell (99). Among

six teams of two raters, the best team agreed completely on 32

84.5 per cent of their ratings, and disagreed by one point

on 11.7 per cent of them. Tne poorest team of raters had

identical ratings on 54,5 per cent of tne items and dis­

agreed by one point on 24.4 per cent of them. Its validity was illustrated in Waites' study (133) in which a group of

78 open ward, chronic psycuotics was differentiated signifi­ cantly from a group of 238 closed ward patients. In another study, it was shown to discriminate between a group of pre-frental lobotomy cases and a group of control cases, with the lebotomized members exhibiting reduced severity of symptoms (5 6 ).

Factor analysis of psychiatric scales.— Recently, factor analytic techniques have been increasingly applied to psychiatric scales to determine the factors that were measured, and, in general, a high degree of communal!ty has been found among the Items of the scales in use. Guertin

(40) extracted three oblique factors in a factor analysis of the Hospital Adjustment Scale. The three factors, closely related, were lack of general interest, social with­ drawal, and personal unconcern.

A much more ambitious project was carried out by

Lorr and Rubinstein (68) who treated 73 scales to factor analysis. The original 73 x 73 correlation matrix was reduced tc a 58 x 58 table, then twiced analyzed to isolate ten first-order factors which were as follows; (l) a bi- 33 polar parameter of emotional responsiveness and energy

level, (2) distortion of reality in perception, thinking, and feeling, (3) a temperament continuum extending from manifest tension to relaxation, (4) sense of personal adequacy, (5) maturity, (6) a character trait of conscien­ tiousness, (7) obsessive-compulsive-phobic reaction, (8) car- dio-respiratory reaction, and (10) conflict between the patient's sexual impulses and moral standards.

RESEARCH OP SMALL GROUPS AND GROUP PSYCHOTHERAPY

The origins of the study of small groups may be briefly recalled first. In 1903, Taylor's studies (120) in

Industrial settings revealed that he was aware of the presence of group norms, and he considered techniques which may be used to deal with them. Puffer (92), in 1905, studied boys' gangs and discussed the relationship of mem­ bers to the leaders in the context of the situation. In another study that was strikingly modern in its design,

Terman, in 1904, used situational tests to rate leadership in children's groups and corroborated the results with friendship choices (121). In 1923, Moreno used the tech­ nique of interaction diagrams in his early work with the spontaneity theater (74). Another objective study of the interaction process was that by Riddle (93) on poker players in 1925. The effects of praise, blame, and razzing were studied by Hurlock (53) and others. Studies in the change 34

In personal opinion after group discussion were reported by-

Thomas (122), Bekhterev and Lange in Germany (10), and

Newcomb (82).

Since 1930, this area has attracted ever widening interest. The rise in bibliographic items from 1890 to

1953 reported by Strodtbeck and Hare (113) substantiated this in an emphatic manner. Prom 1890 to 1939, there was an increase of .5 to 21 items per year. During the period of

1950 to 1953, however, there was an average of 152.5 items per year.

Prediction of Performance with Soclometrlc Techniques

Sociometric devices to predict a variety of perform­ ance criteria are being increasingly used in industrial, military, and school settings. Military studies have sought to predict officer potential, leadership ability, and other factors. French (33) found significant negative correlations between peer group status and sick bay attend­ ance and disciplinary offenses. Sick bay attenders were less acceptable as liberty companions, but equally accept­ able as leaders. Disciplinary offenders were less accept­ able in all situations. The function of interpersonal relationships and bombing accuracy was studied by Hemphill and Secrest (46). They correlated the sociometric choices of members of a B-29 crew with bombing accuracy under actual conditions and found a correlation of .3 6 . 35 Sorgatta (14) conducted an intensive inquiry using

126 U. S. Airforce enlisted personnel and found a correla­ tion of .271 between sociometric measures of friendship and leadership. Relationship between friendship choice and intelligence was slightly lower, with a correlation of .198.

Leadership ratings by superiors did not correlate signifi­ cantly (.0 6 3 ) with friendship choice. However, the leader­ ship ratings did correlate significantly with the socio­ metric leadership nominations (.211) and with intelligence

(.287). These findings are reminiscent and similar to those reported by the Office of Strategic Services Assessment

Staff, and by Ven Zelst described below. There is usually a low positive correlation found between sociometric choices of friendship and leadership, and both of these variables are significantly related to intelligence level. Supervi­ sors' ratings of leadership do not correlate too highly with the candidates' own nominations to leadership. This suggests that the two groups use different criteria for leadership qualities.

■ In an industrial study, Van Zelst (129) reported that the ratings by co-workers were positively correlated with attitudes toward work, quality, and quantity. In two addi­ tional studies (130, 1 3 1 ), he also demonstrated that re­ grouping workers on the basis of socicmetrlc preference reduced construction cost as compared with a control group. 36

Speroff and Kerr (ill) showed that accident-proness was

inversely related to friendship choices received.

Studies in the educational situation have attempted

to predict choices and to correlate sociometric standing

with personal qualities, task proficiency, and observers'

ratings. Eng (29) successfully predicted at the .0^ level

of significance the choices of 22 fraternity residents for

three situations, a double date, a camping partner, and a

campus representative. Broderick (15), in a doctoral dis­

sertation which arrived at similar results, successfully

predicted the friendship choices of college girls. He found

that variables of status and value similarity and frequency

of contact entered significantly into the choices made.

One of the earliest studies was conducted by Moreno who asked teachers to judge the friendship status of the pupils in their classes. He found that they were unable to predict accurately, achieving 38 per cent to 48 per cent agreement only, with the level of accuracy varying inversely with grade level. Moreno commented on these findings as follows:

The teachers' judgments concerned only the extremes in position. The average positions . . . (were) . . . far more difficult to estimate accurately. The intricacies of the children's own associations prevented the teacher from having true insight. This fact appeared as one of the great handicaps in the develoDment of teacher-child relationships (75, P. 5^). 37 Gronlund (39) had forty teachers predict the choices

of sixth grade pupils and obtained significant correlations

that ranged from .268 to .838, with a mean r of .595* An

even higher level of accuracy was attained by Newstetter,

Feldstein, and Newcomb (83) who asked camp counselors to predict the sociometric status of fourteen year old boys.

They obtained a mean correlation of .756 between prediction and sociometric measure. They further found that the relia­ bility of the predictions was consistently high. It ranged from ,830 to .998 with a mean reliability coefficient of .9^5. Bass and White (9) found that the observer's evalua­ tion of a leadership group discussion and peer nominations for leaders yielded a range of coefficients from .25 to .6 0 .

These correlations were higher than those reported by

Borgatta, which might have been due to the more clinical and experimental conditions under which they were obtained.

Borgatta's leadership ratings were probably obtained from the service records of the subjects.

Group Psychotherapy Research

Since the present study examined the accuracy of esti­ mate and prediction in the group psychotherapy setting, any findings obtained will be a function of this particular therapeutic technique. Hence it was deemed worthwhile to present a brief general discussion of the variables that were 38 regarded as most relevant to the use of this technique.

The following review is divided into five sections with an emphasis on experimental contributions. The literature is generally characterized by a greater degree of subjectivity in approach and tentativenes3 in conclusions. The experi­ mental manipulation of groups and the comparison between groups are often crude and there are technical problems special to the area to be solved, such as controlling the effects of the presence of the observers.

Outcomes of therapy.— The outcomes of group therapy have not been investigated as intensively as other aspects of the method. This lag has been commented on by Harris

(42) and by Singer (106), who, in reviewing Bach's Intensive

Group Psychotherapy (5)* remarked on the total disregard of this important area in an otherwise excellent volume.

Gersten (35) > in a study of juvenile delinquents, found that group therapy contributed to school adjustment, emotional release, development of initial insights, and social maturity. Mehlman (7 2 ) compared the effects of play therapy with movie attendance in a population of mentally retarded children and found an increase in the experimental group only, as measured by the Haggerty-Olson-Wickman Scale.

Newburger and Schauer (81) studied the effects of group therapy upon the sociometric choices of reformatory immates, The data revealed there was a temporary increase 39 in the number of mutual choices after therapy, but this was

lost after three months of discontinuance of therapy. Sacks

and Berger (100) established that chronic schizophrenic

patients receiving group therapy did not differ from non­

treated patients in discharge rate. However, the patients

in the experimental group did move to and stay longer on an

improved ward in significantly greater numbers. Peters and

Jones (89) reported that veterans treated with psychodrama

responded with improved performance and motor reactions on

the Porteus Mazes and Mirror Tracing Tests.

Comparison of therapeutic techniques.— The study of

outcomes of therapy is closely related to a comparison of

different therapies. The success of different techniques,

such as activity group therapy, analytic group therapy,

activity-interview group therapy, and play therapy have not

been evaluated definitively (1 0 7 ), and Lipkin (66) has

stated that many different forms produce positive results.

The common factor underlying all of them was thought to be

the therapist's acceptance and tolerance of, and confidence

in, the patient.

Baehr (6) compared three kinds of therapy, individual

therapy, group therapy, and combined group and individual

therapy and studied their outcomes. He found that the com­ bined treatment was superior to either individual or group

therapy alone. Imber (5^)* in ^ Interesting study, compared 40

short-term directive therapy with non-directive therapy,

using a matched group attending the library as a control.

Significant improvement was found for the two treated

groups, when compared with the control group, but the two

types of therapies were not differentiated.

Studies of processes.— The Bales Interaction System

(7) and sociometric devices are being adapted more and more

to the study of group therapy processes. Talland (118)

analyzed the roles of members and the status structure of

the group, using the Bales technique, ratings, and socio­

metric measures. He reported a fairly stable pattern of

interaction evolving within the first few meetings in which

the members had characteristic roles which others expected

him to discharge* The sociometric ratings disclosed that

the members regarded leadership and popularity as almost

synonymous. The correlation between the two variables was

.70. Taylor (119) also used the sociometric technique to determine the pattern of friendliness and dominance in a group and was able to establish a provisional pecking order among the members.

Role of the therapist.— The role of the therapist was investigated in a novel manner by Cadman, Misbach, and

Brown (17) who removed the therapist from the room alto­ gether. They reported clinical improvement with their experimental group, but not in a non-treated control group.

The results of this study tentatively question the hypothesis 41 that the physical presence of the therapist Is necessary to manage group and Individual anxiety and to facilitate trans­ ference. Chance (20) had the group members rate their therapist and their parents on a modified form of the Q technique and found a correspondence between attitudes and feelings toward the therapist and the dominant familial figure.

Selection of members.— Kotkov and Meadow (6 5 ) com­ pared the continuing and non-continuing members of a therapy group on their Rorschach protocols and found that the PC / CF ratio was the most discriminative variable among the Rorschach indices. A prediction formula was con­ structed and separated the members significantly in a cross-validation study. There was also evidence that this formula discriminated among members in individual therapy.

In closing this section, the following statement by

Harris from the 1956 Annual Review of Psychology may be quoted to give an overview of the area:

. . . the current trends . . . indicate that group therapy may be regarded as the treatment of choice for many patients, and, combined with some indi­ vidual therapy, for most patients. It is not an inferior approximation of individual therapy, but a treatment method with unique characteristics. At the present time, both theory and practice . . . are highly diversified, but communications about groups are increasing, and correspondingly, convergence of ideas In important areas is taking place (42, p. 142). CHAPTER III

PROCEDURES OF THE STUDY

Tnis study evaluated the accuracy of the moment-to- mornent guesses, hunches, and hypotheses tnat a counselor or therapist may make during the course of counseling or therapy session. The experimental procedure required psychologists who were experienced therapists to observe a group psychotherapy session with mental patients and make individual predictions on the basis of these observations only. Tne observers did not know the subjects, and their performance was a function of their clinical skill, percept­ ual abilities, and interpretation of the therapy interaction.

A second part of the study examined she amount of increase in accuracy of prediction that was achieved with a second observation three months later. On tne second trial, the psychologists performed the same taks on the same subjects, and the accuracy of first and second sets of predictions were compared to determine the extent of the improvement.

Tne subJects.--The patients in the sample were made up of twenty-five males diagnosed as chronic schizophrenics in the following sub-categories: paranoids (N: 1 5 )> hebephrerics (N: 3), unclassified (N: I), catatonics (N: 4), and schizoaffectlve (N: 2). The age range was from 23 to 49,

42 with a median age of 3^-. Ten were married, fourteen were

single, and one was divorced. Educational background varied

from the eighth grade to four years of college, with the

median at completion of high school. Length of hospitali­

zation ranged from three to twelve years, with a median of

five years. Sixteen members had records of psychotic

episodes prior to current admission, and all had received

electric shock treatment, which had been discontinued prior

to this project. Twenty-one of the subjects were on chemo­

therapy treatment often referred to as the use of tranquil-

izing drugs. The effects of the drugs which were generally

regarded as facilitating psychotherapy reduced tenseness and

nervousness and made the individual more responsive to social

interaction.

The subjects were patients in the Exit Service Unit

of a V. A. Neuropsychiatric Hospital, The Exit Service

Unit was made up of a total of 50 hospitalized veterans, mostly chronic schizophrenics, who were judged by the hos­ pital staff to have achieved a level of recovery sufficient to benefit from a "therapeutic community" approach to

rehabilitation. The orientation of the Unit was a positive one and emphasized the assets of the patients. The atmos­ phere of the Unit was permissive and was directed toward the

realistic problem of returning the patient tc the community. 44

Rehabilitation activities carried on in the Exit

Program included bi-weekly group psychotherapy sessions for

all the patients, implemented by individual counseling

meetings on an arranged basis. There were weekly staff con­

ferences on individual patients to formulate therapy goals

and nursing-care plans. Incentives to return to the com­

munity were festered by routine weekend passes to visit the

nearby town, and each patient was given a weekly monetary

allowance in lieu of canteen books. The staff sought to

provide the patients with realistic success experiences and

personal attention to foster self confidence, self-esteem,

and a more adequate self-appraisal. The reader may refer

to Wright's dissertation (140) for a more detailed descrip­

tion of the concept of the therapeutic community as well as

his experimental work in this area.

Nature of the group therapy sessions.— The study

employed three therapy groups. All three were conducted by the same therapist (the writer) and had been meeting for

approximately three months. For the observation of the psychologists, the therapist sought, within the limits of conscious objectivity, to maintain an atmosphere and orien­ tation in tne group sessions comparable to the meetings held during the preceding months.

The conduct of the sessions was influenced by the fact that the groups were made up of chronic schizophrenics who manifested widely varying degrees of hostility, with­

drawal tendency, anxiety, somatic symptoms, and more obvious psychotic manifestations, such as delusions and hallucina­

tions. Some of them had only recently been transferred from

the Acute Intensive Service. Their personality re-organiza­

tion was still a very tenuous nature and was being

continually threatened to be overwhelmed by residuals of the

schizophrenic process. Furthermore, as it was characteris­

tic of this disorder, these patients were regarded as lonely and "hollow" individuals (91> p. 3 7 0 ) who had never been able to relate to others In a satisfying way.

At the time of the study, the groups had been meeting for three months, and the current aim of the meetings was to develop a level of social comfort in which members of this heterogeneous group could talk about themselves and their troubles. Coffey has pointed out that:

Anxieties cannot be dealt with unless social reality is preferred to autistic, unreal comfort. The therapist must help the psychotic patient exchange his autistic comfort for the comfort of social reality (21, pp. 599-600).

In these sessions there was a greater degree of directive­ ness on the part of the therapist as suggested by Klapman

(60) and Powdermaker and Frank (91), to supply a core of activity to which the members could relate. The paranoid members were more verbal and defensively hostile and were encouraged to ventilate their feelings which usually began with complaints about hospital services. All patients wno

were less resistive and willing to speak about themselves

were stimulated to express their present feelings and con­

cerns and to recall their previous experiences. Toward tne

more withdrawn and sometimes mute members, direct sugges­

tions were necessary to initiate a minimal degree of social

interaction. For example, the request for one of them to

relate about his current activities in Occupational Therapy

demanded a tremendous amount of effort of him. These

exertions usually elicited a highly sympathetic response

from the more adjusted members who were able to refer to

their own first stumbling efforts to communicate and relate

to others.

In summary, the emphasis of the sessions was on the

"support" and "stimulation" (21) aspects of group therapy.

The members were provided a non-punltive atmosphere in which

they could express themselves and discover a similarity in their difficulties. These activities were believed to counteract the effects of the psychotic residuals and rein­ force the reality perceptions of the individuals. Beyond this socialization objective, the next step of group therapy would be to provide the patients with an opportunity for reality-testing, to confront them with their distortions and misperceptions of themselves, which they must test and cor­ rect against the evidence obtained from their group exper­ iences. 47

The observers.--The predictions were made by psy­ chologists who observed the group therapy sessions. The observers had no previous contact with the subjects and knew nothing about their backgrounds except that they were patients in the Exit Service Unit. These steps were taken to minimize the possibility of contamination of predictions resulting from prior knowledge. It is acknowledged that in actual counseling and therapy situations, the counselor or therapist will make his "guesses11 and predictions from an accumulation of information previously obtained and sub­ jectively Integrated to arrive at a precise hypothesis. In the present study, the variable of prior knowledge and its accompanying possible biases were controlled. The psychol­ ogists were to make the predictions after observing the group proceedings, to draw from the therapy inter-action, individ­ ual behavioral clues, and any. other clues that they were able tc obtain and to interpret them on the basis of their training and experience in psychology and therapy.

Six psychologists participated in the prediction task. Four of them were Professors of Psychology at The

Ohio State University and Consultants in Counseling Psychol­ ogy tc the hospital. The other two members were Counseling

Psychology trainees stationed at the hospital and who were not acquainted with the patients In the Exit Program. The latter were advanced graduate students in psychology In their third and fourth years of training.

The therapy sessions were held in the visitors1

reception room, which was large, comfortable, and pleasant.

The psychologists were present in the same room with the members during the period of observation. A limit of three

observers attended any one session. The predictions were

recorded privately on the Prediction Blanks at the conclu­

sion of the session and prior tc leaving the room. Obser­

vation was limited to one session which was an hour in

length. The study of improvement in accuracy was carried

out by obtaining a second set of predictions on twelve of

the original group of 25 patients three months later. The procedures carried out in the second part of the study were

the same as in the initial phase. A tape recording was made of all of the sessions to have a record for reference or additional study.

This study was confronted with the problem common to all investigations of the process of therapy; namely, the impossibility of repeating an observation on the same phenomena. In the present case, serious consideration was given to the possibility that the observation of one session may be too limited to yield accurate predictions. A pro­ posal was considered to have the psychologists observe two consecutive sessions before recording their predictions.

This would have been within a week’s period, but the 49 arrangement could not be made. The consultants were not able to arrange their schedules to visit the hospital that frequently, nor were we able to employ hospital staff psy­ chologists to perform the prediction task since some or most of the Exit patients were known to them. Even the suggestion to hold two therapy sessions on the day of the consultants' visits was not feasible due to the interference of this procedure with the patients' care-routine. Simi­ larly difficulties encountered in carrying out therapy research in a field setting have discussed by others (9 1 >97).

Possible negative effects on the activity of therapy due to the presence of observers were offset by limiting their number to three and by preparing the patients with explanation beforehand. The degree with which the observers were accepted by the groups may be indicated by the fact that none of the group members withdrew, though permission had been granted to withdraw if they desired tc- do so.

Furthermore, there were several instances when a patient addressed a remark or question to an observer, apparently to invite him to enter into the discussion. These experiences with observers were typical of those reported in the litera­ ture. In general, the addition of a nominal number of observers to a group changed the situation very little, as compared with "the intrusion of a third person into a group of two." (79, P. 85). 50

Three months after the initial predictions, twelve patients from the original group of 25 were again observed by the psychologists in a group therapy session and a second

set of predictions on the same items was obtained. These predictions were correlated with a new set of criterion ratings, test and sociometric measures which were secured within the same week of the second predictions. The degree to which the psychologists were able to Improve in accuracy was determined by comparing the first set of prediction correlations with the second set.

There were only twelve patients out of the original sample of twenty-five available for the second part of the study. This was because, within the three months period, three subjects had been released from the hospital, and five had been transferred out of the rehabilitation program due to marked regression in behavior or demonstrated inability to assume the additional responsibilities and privileges required of them. Three other subjects were away from the hospital on pre-arranged leaves of absence to their homes, and the two last patients were lost to the study because the second observation coincided with unexpected requests by the families to take them home for a visit.

When the sample of 12 was observed, they were in two separate therapy groups. One of the groups contained eight of the subjects. The second therapy group contained the 51 other four subjects of the study and five other subjects who were observed originally but were absent on visits home at the time of the second rating. Hence, though the psychologists were observing and predicting for people he had seer, once, the group structure and relationships were changed. However, the effects of these uncontrolled vari­ ables ir. predictive accuracy and improvement cannot be determined at this time.

THE PREDICTION ITEMS

There were twenty-two items to be predicted for each subject. These were selected with the following considera­ tions in mind. First of all, the predictions required of the psychologists should be of a manageable number. This was tc avoid the negative effects of "judgmental fatigue" and "judgmental disorganization" which occur: when the judging task becomes too large or complex. Secondly, the items should deal with different aspects of the subject's behavior, adjustment, and other topics that were appropri­ ate for a hospitalized group of psychotic s. The items finally chosen were grouped into fifteen descriptive cate­ gories. These categories and their items will be described in the following section. A sample copy of the Prediction

Blank may be found in Appendix I.

The items for categories I to VIII were in the form of graphic rating scales, which were adapted from the 52

Multidimensional Scale for the Rating of Neuropsychiatric

Patients, Hospital Form (MSRPP), published by the Veterans

Administration. To make a prediction, the rater makes a checkmark along the line at the point that ne feels best describestne subject.

Category I: General Activity Level (3 Items)

Though this behavior parameter is also applicable to a normal population, it is of special pertinence In the study of psychctics. Guertln (40) recently reported in a factor analysis of the Hospital Adjustment Scale that one of the fundamental dimensions of schizophrenic adjustment was a general apathy, slowness of movement, restlessness, and lack of liveliness. Hence, it would be of research interest to determine if the counselor or therapist were able to predict this variable from the patient’s behavior in the group therapy session.

Item No. 1 : Compared with Exit Service patients, does he seem tired and worn out or lively and energetic? almost com- tired as lively livelier and pletely worn as mo 3t more ener- out getic

GThs numbering of the items follows tne sequence used in the Prediction Blank. Please see copy in .Appendix I. Item No. 8 : How much does he move around the hospital? marked- a little moves about a little moving ly under- under ac- as appro- restless about active tive priate most of the time

Item No. 9 : When in action (walking, talking, dress­ ing, eating) how does he compare with other patients in the Exit Service Program:

* • « • • — - — marked- a little at an average a little distinctly ly slow- slower rate faster faster er

Category II: Submissiveness vs. Hostility (2 items)

Research Interest in this category was at the "hostil ityn end of the continuum primarily. The expression of hostility is an expected and common occurrence in the course of therapy. The question under consideration was whether or not the degree of hostility expressed in group therapy was related to the level of hostility apparent in his be­ havior outside the therapy room.

A companion question to the above was the subject!s attitude toward authority figures. It was postulated that the correlation between the prediction and criterion rat­ ings in this item might be low and still reflected accurate observation by both rating groups. It was reasoned that the atmosphere of the group session was probably more per­ missive, and that the patients would be more willing to 54 exhibit his true feelings toward authority figures. The nurses, aides, and therapists, on the other hand, were more likely to assume authority roles with the patients, result­ ing in the latter inhibiting their authority attitudes to a greater degree.

Item No. 3: How hostile is he?

• * • « • A . ______— — — — .______— — _ — no hos- rela- as much as relatively extremely tllity tively average high hostile low

Item No. 10; What is his attitude toward authority figures: hostile passively reacts appro- looks up dependent and an- resistent priately to to them and ser- tagonls- the situation vile tic

Category III: Depression vs. Manic Excitement (3 items)

This category was concerned with the affective fac­ tors that may accompany the schizophrenic process, and the predictive task called for skill in detecting and differ­ entiating between the manic and depressive characteristics in the patients, such as elation, emotional responsiveness in interpersonal relationships, as compared with excessive concern, depression, and severe anxiety. Item No. 4: Consider his typical emotional tone or mood. deeply moder- neither moder­ highly depressed ately depressed ately elated depressed nor elated elated

Item No. 7: How emotionally responsive doe s he appear to be in interpersonal relationships? * * I « • little inadequate adequate a distinct a marked or no response feeling over-re- over-re- sign of sponse sponse feeling

Item No. 14; Does he exhibit concern, uneasiness, or apprehension to a degree that is not called for by external circumstances? • • • • unconcerned a little distinctly disrupting anxious anxious anxiety

Category IV: Withdrawal (2 items)

The other two factors identified by Guertin’s factor analysis were titled by him as, "Social Withdrawal" and

"Personal Unconcern" (40). In this study, they have been combined In accord with Jenkins1 concept of the "Schizo­ phrenic Sequence" (55), which began with a relative with­ drawal of attention and Interest in the outer environment ana was accompanied by an untidiness and disregard for per­ sonal appearance. 56

Item No, 5 : Does he stay by himself or does he like being with others?

• * • » •

always usually about as usually always In stays by much alone in com­ company by himself as with pany of of others himself others others

Item No. 6: Typically how clean does he keep himself?

* « • • * 4 • distinct­ more slop as neat neater and fastidi- ly sloppy py and and clean cleaner ous and dirty dirty than as most than most most

Category V: Conceptual Disorganization (l item)

This category represented the second phase of Jenkins'

Schizophrenic Sequence. In the Exit Program, most of the patients still retained residuals of their psychotic epi­

sodes, and in the rehabilitation effort, it would be valuable to establish accurately the degree of disharmony that still existed between thinking and feeling tone in the individual's behavior and speech. It was the interest of this study to determine what effects, if any, the unique social and psychological climate of the group therapy situa­ tion had upon the patient's wobbly attempts to maintain consistency In functioning. The patient who had learned to inhibit and suppress his paranoid thinking and ideation In interpersonal relationships to maintain a pose of adequacy may be doing so at a grear cost of anxiety and tension. In 57 the permissive climate of the therapy session where the

patient was encouraged to express his disturbing thoughts

ar.d urges, he might very well drop his pose and achieve a

measure of comfort by verbalizing about the dammed up con­

flicts within him. Therefore, the predictive accuracy on

this item may drop because the psychotic subjects might

have been less "on guard" and perceived as more maladjusted

by the psychologists compared with the other hospital per­

sonnel. In this study there was an opportunity to compare

the differences between the psychologists and the hospital personnel on their perception of the patient's adjustment.

The comparison may be made by examining the means of the

ratings of the two groups, with the higher means signifying

ratings of more adequate adjustment.

Item No. 12: Are his thoughts and feelings consist­ ent, or is there a discernible lack of harmony between them?

• • • * consistent a little distinctly almost disharmon- disharmon- totally ious ious unrelated

Category VI; Motor Disturbance (l item)

The schizophrenic process was often reflected in motor disturbances, primarily in the form of inappropriate gestures and mannerisms. The raters were required to pre­ dict to one item in this category. Item No. 13: Does he exhibit peculiar and inappro­ priate gestures, grimaces, or mannerisms?

none occasionally fairly repeatedly frequently

Category VII: Self-depreciation vs. Grandiose Expansive- ness (1 ItemT

An alternative to schizophrenic disorganization was a paranoid reorganization and stabilization in which a measure of reality orientation was retained. However, it was dis­ torted by ideas of grandeur, boastfulness, and conceit. The other end of the continuum would be represented by the more confused patient who exhibited more self-depreciatory tenden­ cies. The situation in the group session induced the patient to reveal distorted facets of his self-concept that he would not ordinarily show in social situations, thus again lower­ ing the correlation between prediction and observation of depreciatory or conceited attitudes in the patient's behavior in other situations.

Item No. 15: How favorably does he regard himself? distinct- inclined neither self- inclined to distinct- le self- to be depreciatory be self- ly con- deprecia- self-de- nor conceited important ceited tory predatory

Category VIII: Performance on Work Assignment (1 item)

In this study, the term "work assignment" referred to the therapy assignment of the patient. Tnis may have meant 59 two hours of activity in the morning, in the afternoon, or

both mornings and afternoons at some assignment. Most of

the patients were assigned to activities in the Physical

Medicine and Rehabilitation, or P.M.R., Program. Some of

the patients were assigned to Manual Arts Therapy, others to Incentive Therapy, Educational Therapy, and others.

Prediction of the patient’s performance from observation of a therapy session was admittedly a tenuous proposition.

However, the counseling psychologists often needed to establish the patient's quality of performance on the assignment in the evaluation of his level of recovery and readiness for release from the hospital.

Item No. 11: On his work assignment, he is regarded by his rehabilitation therapist as: diligent, good worker irregular avoiding reliable, by hospital worker his work ready for standards assignment working outside

This concludes the prediction items that were in the form of graphic ratings. There were six additional cate­ gories in the prediction task, and they involved the esti­ mate of present intelligence, the group's interpersonal relationships, likelihood cf release from the hospital, and adjustment on the outside after release. These Items will now be described. 6o

Category IX: Present Functioning Intelligence (1 item)

The estimation of intelligence level of psychotics is

usually difficult. The maladaptive behaviors and other

symptoms of disorganization that the patient manifests often

belie his capacities. Furthermore, prediction of intelli­

gence from observation of a group therapy session may be

handicapped by the fact that the person need not participate

if he so desired, thus revealing a minimum of himself. This

study was interested in determining the accuracy with which

intelligence level could be estimated in a deviant popula­

tion.

Item No. 16: His present functioning intelligence as measured by the Wechsler-Bellevue Intelligence Test is;

borderline (I.Q. below 80) below average (80-90) average (90-110) above average (110-120) very superior (above 120)

Category X: Present Friendship Relationships (l item)

From their observation, the psychologists were re­

quired to estimate how well each subject was liked by the other members of the group. The importance of friendship

or rivalry relationships among the members in facilitating or disrupting the therapeutic process has been pointed out by Powdermaker and Frank.

The rivals displaced the doctor, serving as com­ peting therapists to a third patient . . . Anxiety (in the other members) in the form of hostile reactions toward the rivals and playful 61

tittering spread through the group . . . The rivals settled nothing and became more anxious. . . (9 1 ,P. ^17).

These authors have also reported that rivalry situations could be utilized to foster therapy. This finding was accidental and was related to a special approach employed by one of their therapists (91, PP. 39^~^3^).

Item No. 17i How well does this therapy group like him as a personal friend? In a sociometric measure, they would rank him with:

the upper third of the therapy group (most liked) the middle third (median) the lower third (least liked)

Category XI: Present and Future Leadership Preferences (2 itemsT

The group therapist was often interested in knowing if the leadership preferences of patients corresponded with the staff's assumptions of whom the leaders were. Hence, it would be worthwhile to determine if the therapist con­ ducting his group were able to identify the individuals the group selected for leadership.

Item No. 18: At this time, for the position of leader in a work situation with this group, the patient will:

be selected by the majority — receive occasional mention only not receive any mention

Item No. 19: Four weeks from now the number of leadership choices given him will:

increase remain the same decrease I

62

Category XII; Level of Social Interaction (1 Item)

In psychiatric rehabilitation, one of the indication

of an individual's response to treatment was his increasing

association and interaction on a voluntary basis with other

patients. Smith, et. al. (109)., have demonstrated the suc­

cessful use of increased social interaction as a criterion

of progress in group therapy. This question of the patient

level of social activity often arises spontaneously during

the course of a therapy session, when the therapist may not

be able to verify adequately by personal observation. In

this study, the psychological raters had to predict to one

of four levels of social activity for each subject and the

estimate was validated against a time sampling measure of

the patient's social activity in the evening hours.

Item No. 20: In the ward during the evenings, he Is most likely:

to be sitting alone and not interested in the activities around him. to be alone but will converse when someone addresses him. to initiate some sedentary type of social Interaction himself. to be participating in more active games as billiards and pingpong.

Category XIII: Likelihood of Release by Medical Staff (1 item) ~

This item was concerned with the prediction of a com­

plex and critical Incident in the patier.c's life and was

analogous to the molar type criteria employed In the studies 63 examined by Meehl. In the Exit Program, one of the more

common concerns was the accurate estimate of likely medical

action on an individual patient, and actually, the thera­

pist's report of the patient's functioning in the therapy

session contributed to the medical staff's estimate of the

patient's readiness to be released.

Item No. 21: The medical staff will place this patient on trial visit status or discharge him:

within 3 months within 6 months not within 6 months

Category XIV: Level of Adjustment on the Outside (l item)

This was another prediction of a complex event, and

the question posed was similar to those found In the

recidivism studies. The accurate estimate of adjustment

outside the hospital was of special concern to the Exit

Service Staff because the patient population handled was mostly chronic psychotics and long-term patients, whose prognosis of outside adjustment was usually poor. To pre­ dict this Item from only an hour observation of a therapy

session was admittedly a strenuous task.

Item No. 22: Assuming that he will be placed on trial visit status or discharged K.H.B. anytime during the next three months;

he will be able to hold a job well. he will be able to hold a job with only marginal adequacy. he will not work and will depend upon his family. 64

DESCRIPTION OF THE CRITERION MEASURES AND SCORING PROCEDURES

The criterion measures used to validate the predic­ tion Items fell Into five general groups: (1) criterion ratings, (2) psychological test score, (3) sociometric analysis of the group structure, (4) time sample records, and (5) use of hospital records. The following section describes them in detail and relates them to their corres­ ponding prediction items. The description of the scoring procedures follows the discussion of each group of criteria and explains how both the criterion measures and the predic­ tion items were quantified.

Criterion Ratings

The graphic rating predictions were validated against the criterion ratings made by two groups of hospital per­ sonnel, the nurses and aides (ward personnel) and the reha­ bilitation therapists (rehabilitation clinic personnel).

Nurses and aides ratings.— The procedures used to obtain criterion ratings from the nurses and aides are pre­ sented first. This group made ratings on all of the graphic scale items except No. 11, which dealt with the subject's performance on his work or therapy assignment.

Five ward workers contributed criterion ratings.

These were two nurses and three psychiatric aides. All were well acquainted with the patients and have had five or more 65 years of work experience in a psychiatric hospital. The

two nurses employed in the study were the Chief Nurse or

Charge Nurse of the building and the evening nurse. Two of

the aides worked on the day shift; one of them was the

Charge Aide or aide supervisor. The other aide was his

assistant. The third aide worked on the evening shift, and

it was he who made the time sample of ward activities. All of the ratings were performed on an individual basis.

To prepare the ward personnel for the rating of the experimental sample, three training sessions totaling ap­ proximately three hours were held with them. They also had two opportunities to make practice observations and ratings on four patients who were not involved in the study. The training of the raters was conducted by the staff psychol­ ogist and supervisor of the Exit Program who had used the

MSRPP previously.

The first meeting was used to acquaint the raters with their Rating Form (See Appendix II) and to discuss their task. Each of the items was taken up in detail, dis­ cussing the behavioral characteristics relating to it.

Secondly, sample behavior descriptions were used to teach them to rate as accurately as possible along the continuum.

A similar meeting was held with the evening aide and nurse by the writer. The seven people (staff psychologist, two nurses, three aides, and the writer) then spent the next 66 two days making practice ratings on two patients (designated as A and B) who were not in the experimental sample. These two patients, and the next pair (patients C and D) were pur­ posely chosen for their differing behavior, personality and adjustment characteristics.

The completed practice ratings were collected and per­ centages of inter-rater agreement were calculated. A second meeting was then called to discuss the results. The data obtained on the first as well as second practice ratings will be dealt with in more detail in the next paragraph. However, it should be reported that the discussion with the raters revealed that many of the disagreements were due to two common errors inherent in the rating technique. One of the aides (A2) had many distorted ratings on patient B because he knew him from a previous ward and had been rating him on his knowledge of the man rather than on observed behavior.

Another example was that rater N2 tended to keep her ratings near the average on all the items. After this meeting, the group performed a second set of practice ratings on two other patients (designated as C and D). A third meeting was then held to discuss the second set of ratings and to clar­ ify other problems encountered. The results of the second set of ratings were regarded as yielding an adequate level of agreement among the ward personnel as well as between them and the two psychologists. 67 Tables I and II in the following pages summarize the results of the practice and training given the ward workers in preparing them for the rating task.

Table I gives an example of the type of analysis carried out and represents a breakdown of the ratings for patient B. The rating of this patient was chosen for fur­ ther discussion because the agreement among ratings on him was the lowest. He was a schizophrenic paranoid, highly verbal, often given to bizarre ideation and frequent mood swings. Prom Table I it can be seen that the agreement among the raters was lowest on Item No. 4, typical mood,

No. 9, rate of walking, talking, and eating, No. 13, inap­ propriate mannerisms, No. 14, excessive apprehension, and

No. 15, self-regard. Furthermore, It was noted in this first practice session that raters A2 and N2 were committing two rather common rating errors mentioned above which lowered the agreement further.

Table II presented the data on all the practice rat­ ings obtained. It revealed that, first of all, the two psychologists had good agreement in the rating of all four patients. Secondly, among the nurses and aides, there was evidence of improvement in rating accuracy from the first session to the second as shown in the increased percentages.

This was especially appai’ent for raters A2 and N2, though these two continued to have lower percentages. TABLE I

ITEMS ON WHICH THE PAIRS OF RATERS HAD COMPLETE .AGREEMENT ON THE RATING OF PATIENT B

Percentage of items Pairs of on which there was Ratersa Item No. complete agreement 1 2 3 1+ 5 0 7 8 9 10 12 13 lii i5 between pair of raters b Psy 1 + Psy 2 XXX XX XXXXXXX 87% Psy 1 + N 1 X XXXX X X X X X 72 Psy 1 + N 2 XX X X XX X XX 6k Psy 1 + A 1 XX XXX X XXX 6h Psy 1 + A 2 XX X XX 36 Psy 1 + A 3 XXXX XXX X X 6U N 1 + N 2 X XXX X XXXXX 72 N 1 + A 1 XX X X X XXX X X 72 N 1 + A 2 XXX X X X U3 N 1 + A 3 X X XXXXX 57 N 2 + A 1 X X X X X X XX 57 N 2 + A 3 XX X XXXX XX X 71 A 1 + A 2 X XX XXX 1*3 A 1 + A 3 XX X X X XX 50 P

^The raters are identified by the following code: Psy 1: Staff psychologist; Psy 2: The writerj N 1: Charge Nurse (day shift); N 2: Evening shift nurse; A 1: Charge Aide (day shift); A 2: Day shift Aide; A 3s Evening shift Aide,

^Complete agreement was achieved on this item. 69

TABLE II

PERCENTAGE OF ITEMS ON WHICH THE PAIRS OF RATERS HAD COMPLETE AGREEMENT ON THE PRACTICE RATING OF FOUR PATIENTS

First Practice Second Practice Pairs of Ses si on Session Raters3 Patient A Patient" B Patient C Patient D

Psy 1 and Psy 2 93 86 93 93 Psy 1 and N1 79 72 86 86 Psy I and N2 71 64 79 86 Psy 1 and A1 86 64 79 79 Psy 1 and A2 64 36 79 79 Psy 1 and A3 86 64 86 79 N1 and N2 79 72 74 86 N1 and A1 79 72 86 93 N1 and A2 64 43 79 93 N1 and A3 64 57 86 79 N2 and A1 64 57 71 79 N2 and A3 71 71 93 79 A1 ana A2 57 43 71 93 A1 and A3 71 50 79 71

The raters are identified by the following code:

Psy 1 = Staff psychologist N1 = Charge nurse Psy 2 = The writer N2 = Evening shift nurse

A1 = Charge aide A2 = Day shift aide A3 = Evening shift aide 70

The validity of the ward personnel criterion ratings

was established in terms of their agreement with the ratings

of Psychologist 1. Table II indicated that after the second

practice session, all five ward workers had 79 per cent or

more complete agreement with the psychologist. Hence it may

be assumed that they were able to observe for the same

characteristics that the psychologist considered as meaning­

ful and related to the items. A cross-check on the accuracy

of Psychologist l's observations was made by comparing his

ratings with those of Psychologist 2 and Table II showed

that these tv/o raters obtained the highest levels of agree­

ment. It was believed that the above procedures to estab­

lish the validity of the ward personnel's ratings were suf­

ficient and the most feasible under the given situation.

The reliability of the ratings was demonstrated in

the acceptable levels of agreement among the five ward

workers. Table II indicated that the percentages of complete

agreement on the ratings of the second set of patients

(C and D) ranged from 71 per cent to 93 per cent, with the

median at 79 per cent. It has been pointed out already

that the ratings of A2 and N2 were the most different and

were lower In the reliability percentages. However, their

rating services were retained In the actual study because

the data In Table II revealed that their ratings did have validity, since they were able to attain fairly good agree­ 71 ment with the ratings of the psychologist. Therefore, to

include their data with that of the other three ward people

was to actually enhance the validity of the ward ratings.

Therapists ratings.— Since the patients in the study

spent two to five hours each day, except weekends, on a

therapy or work assignment prescribed by the hospital staff,

one of the Interests of the study was to determine the

degree to which the predictions made in a therapy session

correlated with behavior observed in a work or physical

therapy situation. Furthermore, it was of secondary inter­

est to discover the amount of agreement there was between

the criterion ratings of the ward personnel compared with the therapists. For these reasons steps were taken to obtain a second set of criterion ratings on the graphic

scale Items from the rehabilitation therapists.

The therapy assignments were considered as an integral part of the individual's rehabilitation and were examples of the concept of Physical Medicine and Rehabilitation (86) in a neuropsychiatric setting. These therapeutic assignments varied widely in kind and in degree of mental and physical effort required. In the present study, two patients had janitorial assignments on wards; two went to Educational

Therapy which were school classes actually. Another group of patients was distributed among the following clinics: woodworking, metalwork, upholstery, photography, shoe 72

repair, radio and television, and printing. All of these

fell under the category, Manual Arts Therapy. Pour of the

more regressed members were in Occupational, Corrective,

Physical, and Incentive Therapy classes.

Finally there was a group of eight patients who

worked on the following assignments which were not formally

organized clinic situations, which were landscaping and out­

door work, hospital laundry, hospital machine and repair

shop, greenhouse, hospital x-ray laboratory, hospital medi­

cal technology laboratory, hospital library, and garden

detail. The patients were given these assignments only

after careful evaluation by the staff. The more important

criteria of patient suitability were his previous work

experience, present capacity, and level of recovery from his

illness at the time. For each patient so assigned, a

specific person (also called a therapist) was made responsi­ ble in working with the staff and carrying out stated

therapeutic goals.

There were 19 therapists involved in this study. Six of them supervised two patients. In all of the following discussion, they will be referred to as rehabilitation therapists or merely “therapists."

All the therapists participating in the observation and rating of the subjects were males. They varied wideiy in background and training for their present work. Some of 73 them were skilled workers previously; others had been in public school teaching. Finally a third group was distinc­ tive in the formal training they have had In their specialty, for example, in Occupational, Corrective, and Physical

Therapy. Many of the latter have earned their Master's degree. Regardless of their preparation for their work, however, all of them may be considered as sincerely inter­ ested in their work and their patients.

The efforts to prepare the therapists for the task encountered more difficulties than with the nurses and aides.

The rather large group of 19 people made it Impossible to arrange a meeting time suitable to the majority. They had different types of work which called for different schedul­ ing, and they were In clinics scattered widely over the hospital grounds.

The training of the rehabilitation therapists was conducted by the writer, meeting with them in small groups of one, two, or three therapists at their places of work and discussing the observation and rating task with them.

The original rating form was similar to that employed with the ward personnel except that it contained all 15 Items.

The Items and the behavioral characteristics associated wioh them were explained to the therapists in a fashion similar to that performed with the ward people. The tech­ nique of sample behavior descriptions was again used to give them practice in judging where to check along the graph. 74

Then each therapist was requested to make a practice rating on a selected patient who was in his class at that time. These were collected in two days and a second train­ ing session was held. Tneir questions and problems were discussed, and the writer 3ought to communicate to all of them the difficulties m a t individuals had encountered, thereby hoping to maximize the uniformity and accuracy of the ratings as much as possible.

Since the 19 therapists worked with different groups of patients, their practice ratings were performed on dif­ ferent patients and could not be compared. Hence a measure of individual reliability was obtained. One week after the first ratings had been gathered, the therapists were asked

(without previous announcement) to rate the same patients a second time. Percentages of agreement between the first and second sets of ratings were calculated and they are presented in Table III.

The table indicated that the therapists attained low levels of agreement in the rerating of the 15 items. The percentages of agreement on all the items (second to the last column) ranged from 40 per cent to 67 per cent with a median of 47 per cent. However, closer examination of the table indicated that the therapists’ accuracy was appreci­ ably higher on Item No. 5# stay by himself or liked being with others, No. 6, cleanliness, No. 8, movement around the hospital, No. 10, attitude toward authority, and No. 11, TABLE III

INTRA-RATER RELIABILITY AM0N3 THERAPISTS PERCENT A3 ES OF COMPLETE A3REEMENT OBTAINED BY THE THERAPISTS RAT 13X3 THE SAME PATIENT TWICE WITHIN AN INTERVAL OF ONE WEEK

Percentage of complete agreement Therapi st Item No. On Items No. On all No. 5,6, 1 2 3 It 5 6 7 8 9 10 11 12 13 ill i5 items 8,10,11 1 xa XXX X X X X XX 67% 100$ 2 X XXX X XX U7 80 3 X X X X X XX h7 60 h XX XX X XX k7 80 5 XX XXXX X X XX X 73 100 6 X XX X X X X X 53 60 7 XXX X X X X X 53 8o 8 X X X X XX Uo 60 9 X XX X X X X 1*7 80 10 XX X XX XXX X X X 73 80 11 XXXX XXX X XXX 73 100 12 XXX X X X X X 53 80 13 X XX XX XX h7 80 lU X XX X X X X XX 60 80 15 XXX X XXX hi 100 16 XXX XX XX hi 60 17 XXX X X X X X 53 80 18 X XXXXX Uo 60 19 XXX XXX XXXX 67 100 Median percentage 1*7 80 a Complete agreement on two ratings of the patient by the therapist. 76 performance on work assignment. The last column in the table listed the improved percentages calculated on these five Items only. Their range of percentages was from 60 per cent to 100 per cent, with a median percentage of 80 per cent. This finding, which provided a measure of assur­ ance that these five items were being reliably observed and rated, and it prompted the decision to require the therapists to make criterion ratings on them only. An example of the final rating form used by the therapists may be found in

Appendix III.

A measure of the validity of the therapists ratings was obtained by having them rate a new patient on the five items and comparing their ratings with those made by the experimenter. To perform this task, the experimenter spent at least two hours in the clinics observing a patient before completing the rating form. There were 19 different patients to be observed and rated by the experimenter, and the task took approximately two weeks to complete. The patients* ratings performed by the experimenter were compared with those performed by the therapists and percentages of agree­ ment were calculated. The range of complete agreement on the five items was from 20 per cent agreement (one item) to

100 per cent agreement (five items). The median was 60 per cent complete agreement (three items). If the ratings 'with one-step disagreement were also included in the above 77 analysis, then the percentages of agreement ranged from 60

per cent to 100 per cent, with the median still at 60 per

cent agreement. Consequently the therapists' ratings,

validated in terms of agreement with a psychologist's rat­

ings, were considered as adequate and comparable to those

attained in other studies, for example, the V. A. Clinical

Psychology investigation (62),

Scoring the rating scale items.--It should be made

clear at this time that both the prediction and criterion

ratings (Items 1 to 15) were scored in the same way. The items were quantified on an eignt or ten point scale, and the scores were arranged so that the ratings of greater degree of adjustment received the higher values. This pro­ cedure was similar to that of otner studies, such as the ones reported by Tindall (126) and Shatin and Fred, authors of the Albany Behavioural Rating Scale, who used the higher score to signify "the healthier direction or response . . .

(or) . . . healthier ward behavior" (104, p. 645). Below are Illustrations of the eight and ten point scoring and the differences in the direction of scoring to maintain a con­ sistency in rating the more appropriate behavior with the higher score. 78

Item No. 14, an eight-point scale: Does he exhibit concern, uneasiness, or apprehension to a degree that Is not called for by external circumstances?

876 5 4 3 2 1 ,pts. .pts. .DtS. .pts. . pts. pts. . pts. pt. ----- £------i------1.______i______i. unconcerned a little distinctly disrupting anxious anxious anxiety

Item No. 9, a ten point scale: When in action (walk­ ing, talking, dressing, eating) how does he'compare with other patients in the Exit Service Program? 123455432 1 . pt..pts, .pts. .pts. #pts. #pts. tpts. ^pts. .pts. ^ pt. marked- a little at an average a little distinctly ly slow- slower rate faster faster er

Psychological Testing of Intelligence Level

The prediction of present functioning intelligence was validated against the I. Q,. score obtained by testing the subjects individually on the Wechsler-Bellevue Intelli­ gence Test, full scale. Their I. Q. scores ranged from 79 to 125 (Table IV), with a mean I. Q. of 102.64. The data on the subgroup of 12 patients who participated in the study on the Improvement of predictive accuracy are also presented in the table. This latter group was tested a second time on the alternate form of the Wechsler three months later at the time of the second prediction. This was done to make sure that the predictions were validated against a con­ temporary measure. As it turned out, there was no 79

significant change, with the means and standard deviations

almost identical.

TABLE IV

INTELLIGENCE TEST DATA ON THE SAMPLE

Standard Mean Deviation

N: 25 patients 102.64 16.20

N: 12 patients

First measure 89.57 10.23 Second measure 90.90 10.16

Scoring— The prediction of intelligence on each sub­

ject was quantified by taking the mean of the three

estimates made by the psychologists. For example, assume

that a subject received the following three estimates of his

intelligence level: average (9 0 -1 1 0 ), above average (1 1 0 -

120), and above average (110-120). The middle score of

each of the ranges(ICO plus 115 plus 115) was taken and

summed to obtain the mean predicted intelligence score, which would be 110 In this instance.

Sociometric Analysis of Friendship and Leadership Standing

Predictions on Irems No. 17, 18, and 19 were validated

against a sociometric measure of the patients' status in the

group. Tne procedure was carried out in three steps. First, 80

a record of the present friendship standing was obtained in

the therapy session immediately following the one observed

by the raters. Since there were two meetings a week, this

measure was obtained within two to four days after tne

predictions. Then a record of present leadership standing was secured in the next meeting, which was one week after

the predictions were made. Four weeks after making tne pre­

dictions the group members were asked to state their

leadership preferences again. This constituted the measure of future leadership standing with which the prediction of

Item 19 was correlated.

At the beginning of the meeting, each member was presented with a list of the patients in the therapy group.

They were then instructed to write in on the Friendship

Choice Blank (Appendix IV) the names of two members with whom they would like to share an activity, and two members with whom they would not want to share an activity. There were three activities for companionship as follows:

1. To visit a nearby town on Saturday.

2. To room with in a double bedroom in the ward.

3. To accompany to a large city on a three day pass for a visit and to inquire about job possibili­ ties.

The two leadership measures were obtained in tne same manner. Each subject was asked to list two members trey recommended and two members they would not recommend for 81 three work situations. The cnoices were recorded on the

Leadership Choice Blank (Appendix V). The tnree work

situations devised for the patients to respond to were as

follows:

1. To direct the cleaning of the Day Room in preparation for a visit by the Gray Ladies.

2. To direct the painting of some lockers to be used by patients.

3. To direct the stapling of the hospital newspaper.

The friendship and leadership records were purposely obtained

in separate meetings to minimize possible generalization of

one characteristic to the other. Secondly, the patients were given the following simple description of work leader­

ship characteristics to guide them in making their choices:

The leaders you select should be people whom you feel can be depended upon to do the Job properly and to be able to get you and the. others to work with him (Appendix V).

Scoring.--Each favorable choice ascribed to a subject by his fellow members on the Friendship and Leadership Blanks was given plus one point and each rejection was given minus one point, regardless of whether tne nomination or rejection was in the first or second position. This method of equal weighting for all choices and rejections was used in accord- anae with the findings of Br-cnfenbrenner (l6 ) and Campbell

(18) that equal weighting gave an almost identical distribu­

tion of group members' status as compared wltn the more

complicated roetnod of differential weigntlng. The subject's 82

score was the algebraic total of his nominations and rejec­

tions. The highest possible score a subject could attain

was plus 21 points and the lowest possible score was minus

21 points. These extremes would occur if, in a group of

eight members, the other seven people either nominated him

to all three situations or rejected him in all three situa­

tions. To simplify the calculations, a constant large

enough to eliminate all negative values was added to each

subject’s score.

The predictions on the friendship and leadership

items were scored so the higher levels of preference pre­

dicted received larger scores. This corresponded to the

higher sociometric scores which reflected greater preference

by the group members. The score values given to each level

of preference predicted for a subject are given below:

Item No. 17: How well does this therapy group like him as a personal friend?

the upper third (most liked) 6 points the middle third (median) 4 points the lower third (least liked) 2 points

Item No. 18: At this time, for the position of leader in a work situation with this group, this patient will: be selected by the majority 6 points receive occasional mention only 4 points not receive any mention 2 points 83 Item No. 19: Four weeks from now, the number of leadership choices givenhim will:

increase 6 points remain the same 4 points decrease 2 points

Time Sample Records ,

The prediction of level of social interaction was

validated against a time sample record of each subject's

activities in the ward in the evenings. Another time

sample,recording the subject's activities at the clinic on

his work or therapy assignment, was also obtained and

correlated with the prediction of work performance as a

second criterion. It will be recalled at this time that

the prediction of work performance was also validated

against the therapists rating on this item.

Sampling ward activities.— In view of the excessive

amount of time usually required to make a time sample record,

an attempt was made to simplify the task as much as possible

for the evening aide. This was accomplished primarily by

providing him with a list of 20 activities that covered most of the activities usually entered into by the patients

In the evenings after supper. This period extended from

the hours of 6:30 p.m. to 10 p.m., when the patients went

to bed. This list enabled the aide to make a record by

checking off the appropriate activity at the time of obser­ vation. It was only when the patient was in an activity not

listed that the aide had to write down a description. A 84

copy of this time sample is presented in Appendix VI. The

20 activities are listed in the next section with a descrip­

tion of the scoring.

The procedure carried out by the aide was as follows.

He first observed and recorded the activities of twelve

designated patients on the hour (for example, at 7:00 p.m.).

He next observed and recorded the activities of the other

thirteen patients at the half-hour (for example, at 7:30

p.m.). Two records of behavior were made during each obser­

vation period. This was done by making a record for each

of the twelve patients consecutively, then returning to

make a second record immediately. Since the recording

involved little writing, he was able to make two records for

the group In five minutes. The same procedure was repeated

for the other half of the group at the half hour mark. On

the next evening, the observation periods of the two groups

were reversed. This procedure provided ten observations

for each patient per evening. Records were obtained on the week the predictions were made, alternate evenings, Monday,

Wednesday, Friday, and Sunday. This yielded forty observa­

tions per individual.

The reliability of the observations were ascertained

by the following means. By pre-arrangement, the writer ana

the evening aide separately observed and recorded the activities of two patients (not members of the study) during 85 each 15 minutes from 3:30 p.m. to 5:30 p.m. and from 6:30 p.m. to 8:3-0 p.m. Comparison of the two records showed agreement 6i times out of a possible 64 chances. This represented 92 per cent agreement which was comparable to the 86 per cent agreement out of 210 chances obtained by

Tindall.

Scoring.--The 20 activities were given values of one to five points, depending upon the degree of social interac­ tion and physical activity each required. The values were determined subjectively by the writer and were as follows:

Sitting, sleeping on ward, not inter- ested in activities 1 point Sitting, watching others 2 points Watching T. V. (listening to radio) 3 points Reading 3 points Playing cards alone 3 points Playing cards or checkers with someone 4 points Playing pingpong or billiards 5 points Taling with patient or ward personnel 4 points Taking medication 2 points Doing personal chore, e.g., shining shoes 3 points Taking a walk 3 points Sitting out on lawn 3 points Golfing, or attending ball game 5 points Attending movie or dance at Recreation Hall 5 points Attending movie on ward 4 points Visiting friend on another ward 5 points Bathing, shaving, readying for bed 3 points Doing ward chore, e.g., sweeping 3 points Other (please record) variable Donft know where he is 2 points

It will be recalled that the psychologists were pro- vided with four levels of activity from which they were to select the one considered to be most characteristic of the 86 patient. The3e levels of activity and their scores are presented below:

Item No. 20: In the ward during the evenings, he is most likely:

to be sitting alone and not inter­ ested in the activities 2 point3 to be alone but will converse when someone addresses him 4 points to initiate some sedentary type of social interaction 6 points _to be participating in more active games as billiards 8 points

Sampling work behavior.--A time sample of the sub­ jects' behavior on the therapy assignment was made in a pro­ cedure similar to that carried out by the evening aide. The time sample was completed during the same week the subject was observed in group therapy by the raters. The time sample covered five days, wita the subject's behavior being observed approximately two minutes every one-half hour dur­ ing his attendance. The number of observation per hour was doubled because the therapist had to observe only one or two patients, and the patients were in the clinic for a short period of time. The record was made on a form

{Appendix VII) that listed ten activities which were suggested by the therapists and covered most of the possibilities of behavior that was likely to occur. The therapist either checked or wrote in the activity the patient was observed to be doing during the two minute period. The clinic hours were usually as follows: 9 a.m. to 11 or 11:30 a.m. and 87 1:30 p.m. to 3 or 3:30 p.m.

Scoring;. - -There were ten activities in the work time

sample, and they were given score values of one to three points by the writer, with the higher scores being accorded to those activities that were Judged to be most appropriate in the clinic situation. The activities and their points are as follows:

Absent without excuse 1 point Idle (chatting with others, reading own materials, etc.) 2 points Moving about aimlessly 2 points Working on assignment 3 points Receiving instructions 3 points Attending class or group discussion 3 points Out on assignment or errand for clinic 3 points Doing maintenance work for clinic 3 points Washing up to leave 3 points Talking with therapist about a personal concern 3 points

It will be recalled that the patients spent a different num­ ber of hours in the clinics, varying from two to five hours per day. This meant that, over the week, some patients were receiving tv/ice as many observations as compared with others. To make the scores comparable, all of scores were prorated on a basis of five hours in the following manner:

5 hour patient: total score x 5/5 equals prorated score 4 hour patient: total score x 5/4 equals prorated score 3 hour patient: total score x 5/3 equals prorated score 2 hour patient: total score x 5/2 equals prorated score

Hospital Records

The hospital records on the patients were consulted to validate two of the prediction items, 2io. 21, likelihood 88 of release from the hospital, and No. 22, adjustment on the outside after release.

It will be recalled that Item 21 gave the psycholo­ gists three alternatives in predicting when the subject would be released, (1 ) in three months, (2 ) in six months, and (3) not within six months. This prediction was validated by examining the hospital records of each of the subjects at the end of three months, then at the end of six months to note what disposition had been taken on the case. The scoring of this item was the same for the prediction and criterion. The point values ascribed to the alternatives was subjectively arrived at by the writer and were as follows:

Item No. 2 1 : The medical staff will place this patient on trial visit status or discharge him M.H.B.:

within three months 6 points within six months 4 points not within six months 2 points

In the sample of twenty-five subjects used to study accuracy of prediction, three patients were released and five others were let off the program for regressed behavior within three months of the prediction. Two others were discharged by the end of the six months period, and fifteen remained hospitalized.

In the subgroup of twelve subjects used to study

Increase in accuracy, none was released within three months of the first prediction. Otherwise they would not have 89 been available for the second observation and prediction.

However, two of the members of this subsample were released within six months of the first prediction. In terms of the

validation of the second prediction of this item, No. 21,

these two same patients were considered to have been

released within the three months period, and the same ten were still nospitalized.

Item No. 22 dealt with the prediction of the patient's adjustment in the community after release from the hospital.

It required the psychologists to assume that the patients would be discharged within three months of the observation of the group, and to predict adjustment in terms of one of the following alternatives: (l) he will be able to hold a job well; (2 ) he will be able to hold a job with only mar­ ginal adequacy; and (3 ) :he will not work and will be depend­ ent upon his family. The above assumption was stipulated in order to complete the collection of data within a six months period, and to permit the discharged patients to live for a period of three months in the community before their adjustment was evaluated and scored as a criterion measure for this item. However, as events turned out, there were only five patients (in the sample of twenty-five) and two patients (in the subsample of twelve) available for the validation of this prediction. These N's were regarded as too small to be reliable. Consequently, Item 22,adjustment 90

after release, was omitted from the findings of the study

and will not be considered in the subsequent discussion.

ANALYSIS OF THE DATA

At the conclusion of collection of data, there were

twenty-one prediction scores and twenty-six criterion scores

on each subject. The twenty-six criterion scores included

a second set of criterion measures on five items. Product-

moment correlations were calculated between each prediction

item and its corresponding criterion measure or measures,

and if the validity coefficient attained the .05 level of

confidence, the item is considered successfully predicted

by the psychologists. The degree of increase in the predic­

tive accuracy was determined by comparing the two sets of

correlations and will be explained in detail in the next

chapter.

The computation was performed by I . B . M . machines. The

forty-seven scores were first punched on cards, verified

for correctness, and then programmed for calculation of

product-moment r ’s on the I . B . M . Type 650 magnetic drum

computer. The availability of this digital computer service made it feasible to obtain complete matrices of intercorre­

lations on the sample of twenty-five patients, and these are presented in the Appendix section. Appendix V I I I contains the prediction Inter-r's; Appendix I X presents the criterion inter-r's, and Appendix X, the inter-r's between predictions and criterion measures. These matrices contributed to a clearer understanding of the findings and revealed pat­ terns of interrelationships among the predictions and criteria. These related findings are discussed separately in the next chapter. CHAPTER IV

FINDINGS OF THE STUDY

Characteristics of the Original Data

Prior to presenting the findings related to the

central questions of accuracy of prediction and improvement in accuracy with a second observation, some of the charac­ teristics of the original data are presented. These are the means and standard deviations of each of the 21 items that had to be determined prior to calculating the product- moment correlations. An inspection and comparison of the means and S.D. *s of the 21 items will probably assist in a better understanding of the findings in the end.

Table V presents the means and standard deviations of the 21 items based on the larger sample of 25 subjects and provides a comparison of the rating tendencies of the psychologists, nurses and aides, and therapists.

Take Item No. 1 in Table V as an example. The mean rating obtained by the psychologists was 11.64. The nurses and aides, on the other hand, had a much higher mean rating of 17.08 on the item. The S.D.'s achieved by both groups were quite similar, however, and it may be concluded that the nurses and aides had a tendency to rate the subjects higher or. this item as compared with the psychologists.

92 93 TABLE V

ARRAYS OF MEANS AND STANDARD DEVIATIONS FOR 22 MEASURES OBTAINED FROM PSYCHOLOGISTS, NURSES AND AIDES, THERAPISTS, AND OTHER SOURCES

Nurses and Soclometrlc Psychologists Aides Therapists and Others Items Mean S.D. Mean S.D. Mean S.D. Mean S.D.

la 11.64 5.^4 17.08 4.77 2 6.84 1.40 9.64 1.62 3 15.20 4.52 22.12 6.95 4 6.40 1.26 10.44 1.72 5 12.32 5.10 15.76 5.69 14.12 5.22 6 9.16 2.75 13.04 3.19 10.24 2.87 7 6 .6o 2.44 9.40 2.00 8 6.60 2.41 9.88 2.82 7.80 1.98 9 11.92 5.53 17.72 5.22 10 6.60 2.52 9.40 2.77 6.96 2.37 11 14.84 3.84 15.76 4.55 12 16.44 4.78 25.32 5.98 13 18.68 3.94 28.64 5.30 14 15.44 4.34 24.92 5.91 15 6.36 1.32 9.88 1.82 16 98.60 8.55 102.64 16.20 17 11.12 3.88 13.88 4.56 18 11.44 4.77 13.12 4.48 19 13.36 2.57 13.84 4.25 20 7.60 2.78 22.08 5.84 21 11.84 3.75 2.24 1.51 Wk. -c T.S. 33.48 5.05

aNumber of Items corresponds to that in the prediction blank. bTime Sample of work activities. Item No, 5 in Table V may be used as an example of a

three way comparison among the ratings of the psychologists, nurses and aides, and therapists. In this case the mean rating of the psychologists was again definitely lower

(1 2 .3 2 ) than that of the nurses and aides and the thera­ pists (15.76 and 14.12). But the S.D.' s of the three groups were very similar, indicating that the psychologists again tended to rate the patients lower on the scales. Actually an inspection of the means and S.D.'s of Items No. 1 to 15 in the above manner reveals that the psychologists consis­ tently attained lower mean scores compared with the nurses and aides and the therapists. There Is no way to determine if this difference simply represents a divergence in rating tendencies or represents a genuine variance in the percep­ tion of the patients' behavioral characteristics by the rating groups. It is also interesting to note that the most marked differences among the mean scores achieved by the three groups of raters occurred in Items No.l, lively or tired, No. 3, how hostile is he, No. 9, rate of eating, etc.,

No. 12, thought and feeling disorganization, No. 13, inap­ propriate mannerisms, and No. 14, excessive apprehension.

Of these items all but Nos. 1 and 9 refer to traits or characteristics such as hostility and thought disorganiza­ tion which require a greater degree of subjective judgment.

In addition, the difference in experience and background 95 between the psychologists and the other hospital workers

may have been accentuated in these Items.

Table VI presents the means and S.D.'s of the 22

measures for the smaller group of 12 subjects. Since this

group was measured twice over an interval of three months,

both sets of original data are given.

A comparison of Tables V and VI yields some interest­

ing findings. Most apparent is that the present functioning

Intelligence of the subsample (item No. 16) is approximately

12 points lower than that of the larger group (I.Q.: 90 and

102). Secondly, the mean ratings of the psychologists on

all the items on the larger sample (Table V, Column l) were

consistently higher than the mean ratings on the correspond­

ing items on the subsample of 12 patients (Table VI, Column

1). The mean differences on some items are very slight, but

the trend is consistent. This suggests that the psycholo­ gists perceived the subsample as less well adjusted.

In the same fadHon a comparison of the S.D.1s of the

ratings made by the psychologists on each of the items for

the two samples may be made. Inspection of the two tables reveals that the S.D.’s of the subsample (Table VI, Column

2 and 4) are often much smaller than the S.D.’s of the larger sample of 25 subjects (Table V, Column 2). This finding suggests that the psychologists found the patients in the subgroup more homogeneous and alike in adjustment 96

TABLE VI

TWO SETS OF MEANS AND STANDARD DEVIATIONS ON 22 MEASURES ( PSYCHOLOGISTS, NURSES AND AIDES, THERAPISTS, A N*12

Psychologists Nurses and Aides 1 Item Mean0 Mean^ S.D. S.D. Mean Mean S.D. S.D. Mean V No.a 1 2 1 2 1 2 1 2 1

1 10.81 10.31 5 . 3 7 1+.99 16.33 1 3 . 3 3 It. 27 1+.95 2 6 . 7 5 6 .6 7 1 .1+2 1.37 9 . 6 7 7.83 1.81+ 1 .28 3 1 U . 6 7 11+.75 1+.66 1+.21+ 22.33 17.08 6.61+ 1+.77 1+ 6 . 1 7 7.17 1.3U 1 .1+0 9 . 75 8.17 1 .92 .8 0 5 1 1. 17 11.16 1+.1+5 1+.65 15.1+1 1 2 . 2 5 5 .2 1 1+.97 9.58 1 6 9 . 05 9.05 2.59 2.58 13.00 9.75 2 .21+ 2.36 7.83 7 6 .0 0 6 . 25 2.00 2 .1 7 9 .0 0 6 . 7 5 2.00 2 .00 8 6 .0 0 7.00 2.63 2.27 9 .8 2 8.67 1 .8 0 3.17 7.83 9 11.08 11.33 U.75 1+.35 19.67 11+.58 1+.36 I+.06 10 6 . 5 0 6 . 1 7 2.81+ 2.72 9.8 3 7 2.67 2.08 7.25 ( 11 1 U . U 2 1U.50 3.35 1+.39 17.17 i: 12 1 5 . 5 0 18.83 1+.1+8 3.13 21+.66 20.16 5 . 1 6 5.21 13 18.58 20.1+2 3.15 3.75 26.16 21.08 5.01+ 5.55 11+ 1U. 67 17. 83 3 .92 1+.37 26.91 19 . 3 3 6 . 3 8 5.01+ 15 6 . 3 3 7.00 2.25 2.12 10.33 7.92 2.09 1.71 16 8 5 . 6 7 89.66 8.81+ 10.26 17 10 . 6 7 11.16 1+.1+2 1+.79 18 1 0 . 6 7 11.50 1+.35 1+.01 19 12.5 0 11.16 3.18 3.87 20 6 . 9 2 5.5 8 2.1+3 2.33 2.92 21 b 11.00 9.67 2.77 Wk. T.S*

^Numbering of items corresponds to that in the Prediction Blank* Time Sample of work activities. c Mean 1 and S.D, 1: first measure* ^Mean 2 and S.D, 2: second measure* TABLE VI i DEVIATIONS ON 22 MEASURES OBTAINED THREE MONTHS APART FROM ES AND AIDES, THERAPISTS, AND OTHER SOURCES N=12 rses and Aides Therapists Socioaetric & Other Measures Mean S.D. S.D. Mean Mean S.D. S.D. Mean Mean S.D. S.D. 2 1 2 1 2 1 2 1 2 1 2

13.33 U.27 U.95 7.83 1.81* 1.28 17.08 6.61* U. 77 8.17 1.92 .80 12.25 5.21 i*.97 9.58 13.33 2.JU3 2.11 9.75 2.21* 2.36 7.83 8.50 1.57 2.29 6.75 2.00 2.00 8.67 1.80 3.17 7.83 7.83 2.61* 2.61* 11*. 58 ii.36 1*.06 7 2.67 2.08 7.25 6.17 2.11' 2.03 17.17 12.33 5.U1 5.01 20.16 5.16 5.21 21.08 5.oi* 5 . 5 5 19.33 6.38 5.ol* 7.92 2.09 1.71 89.57 90.90 10.23 10.16 6.83 6.67 1.57 1.1*9 6.00 6.83 1.53 1.52 ii.oo 1*.00 1.15 1.1*1 lli.86 13.50 3.57 U.53 2.33 2.66 0.91 0.82 2L.17 23.18 5.81 5.72

,he Prediction Blank. 97 and behavior traits, and were probably more difficult to differentiate along the rating continuum.

DATA ON THE ACCURACY OP ESTIMATE AND PREDICTION

Appendix X presents the complete matrix of correla­ tions between the estimates and predictions of the psychol­ ogists and the criterion measures on the sample of 25 subjects. From this appendix the coefficients that repre­ sent the validation of each item estimated or predicted by the psychologists have been extracted and presented in

Table VII. It will be observed that the correlations have been divided into three groups, according to the criteria against which the estimates and predictions were validated.

There were fourteen items correlated with the nurses and aides criterion ratings; five items correlated with the therapists' criterion ratings, and seven items correlated with psychological test, sociometric, time sampling, and other sources. The table shows that out of the 26 corre­ lations, there were twelve significant at the .01 level, three at the .05 level, making a total of fifteen signifi­ cant correlations.

Correlations With the Criterion Ratings

Inspection of the correlations presented in Table VII indicated that the predictions agreed best with the criter­ ion ratings of the nurses and aides. Seventy-one per cent TABLE VII

CORRELATIONS BETWEEN PREDICTIONS AND CRITERION MEASURES

N-25

Prediction Items Criterion Measures la 2 3 k 5 6 7 8 9 10 11 12 13 ll* 15 16 17 18 19 20 21

Nurses and Aides Ratings 559b 150 781 378 521; 823 693 792 552 1*31 1*09 371* 1*15 201

Rehabilitation Therapists Ratings 322 192 033 6oU 565

I.Q., Sociometric, Time Sample, etc. Measures 315 582 300 576 216 287 73U

Numbering of items corresponds to that in the Prediction Blank* d Underlined coefficients are significant: .505 “ .01 level •396 ■ .05 level

vo 00 of the correlations with this group were significant and the magnitude of the validity coefficients was higher, compared with those of the therapists. With the latter group, the predictions were significant on two out of five items, achieving a 40 percent level of accuracy. Table VIII below presents the correlations between predictions and criterions ratings In a descending order of magnitude.

First the correlations with the nurses and aides ratings are listed, followed by the correlations with the therapists ratings.

Among the non-significant correlations with the thera­ pists ratings, prediction Item 5 approached significance at the .05 level, the required r being .396. But the other two predictions (No. 6 and 8 ) were only slightly correlated with the therapists ratings, though they were successfully predicted to the ward personnel ratings. 100

TABLE VIII

CORRELATIONS BETWEEN PREDICTIONS AND CRITERION RATINGS

Correlations witn Nurses and Aides Ratings Items predicted at .01 level No. 6 . How clean does he ,-:eep nimself? .823 8 . How much does he move around the hospital? .792 p. How hostile is he? .781 7. How emotionally responsive is he in interpersonal relationships? .693 1 . Does he seem tired--or lively? .559 9. When in action (walking, etc.) how does he compare . . .? .552 5. Does he stay by himself or ... like being with others? .524 Items predicted at .05 level

1 0 . What is his attitude toward authority figures? .431 14. Does he exhibit apprehension . . . not called for? .415 1 2 . Are his thoughts consistent . . .? .409 Items non-significantly predicted 4. Consider his . „ . emotional tone . . . .378 13. Does he exhibit peculiar mannerisms . . ,? .374 15. How favorably does he regard himself . . .? .201 2 . Is he placid . . . or tense? .150

Correlations with Therapists Ratings Items predicted at .01 level No. 10 . What is his attitude toward authority figures? .604 11 . , . . how is he regarded on his work assignment? .565 Items non-significantly predicted 5 . Does he stay by himself or , , . like being with others? .322 6 . How clean does he xeep himself? .192 8 . How much does he move around the hospital? • ^O.J 101

Correlations with Intelligence, Sociometric, and Other Measures

Table VII reveals that out of the six predictions In

this group, three were significant, all at the .01 level,

achieving a 50 per cent level of success. The descending

order of magnitude of these coefficients is listed below in

Table IX.

TABLE IX

CORRELATIONS BETWEEN PREDICTIONS AND PSYCHOLOGICAL TEST MEASURE, SOCIOMETRIC MEASURES, TIME SAMPLE, AND HOSPITAL CLINICAL RECORDS

N = 25 Items predicted at .01 level

No. 21, Predict length of time before patient i3 placed on trial visit or discharged M.H.B. .734 18. What is his standing in nomination for work leadership at the present time? .571 16. What is his present functioning intelligence? .582

Items nonsignificantly predicted

17. Subject’s present friendship standing in group. .300 20. Level of social interaction in the ward. .287 19. Subject's work leadership standing four weeks from now. .216 102

DATA ON THE IMPROVEMENT IN ACCURACY

It will be recalled that for this phase of the study

there was a three-month Interval between the first and

second sets of observation and prediction, with the

psychologists having no contacts with the subjects In the

interim period. The number of subjects participating

totaled twelve people, and the reasons for the attrition

from the original twenty-five hare been presented. The first

set of predictions and criterion measures was obtained by

re-calculating the original scores on each item based on

the twelve subjects. The second set of predictions was

secured by having the psychologists observe the twelve sub­

jects in a second group therapy session which included new participants. The second set of sociometric measures was

also made on the basis of the new groupings. Since the

total number of members in the therapy groups remained the

same, from eight to nine people, the twelve subjects in this phase of the study had the same opportunity to receive nom­ inations making the first and second set of sociometric

standings comparable.

Table X on the following page presents a resume of the data on the degree of Improvement achieved by the raters.

The correlation coefficients for each item obtained on the first and second measures are presented together for easy comparison. The data in this table is examined in the TABLE X

TWO SETS OF CORRELATIONS BETWEEN PREDICTIONS AND CRITERIA OBTAINED THREE MONTHS APART N-12

Prediction Items Criterion Measures ia 2 3 U 5 6 7 8 9 10 11 12 13 m 15 16 17 18 19 20 21

Nurses and Aides Ratings First Set of r's 595* -350 73U 039 580 029 667 283 367 323 601 576 638 213 Second Set of r's W hh2 m -099 JE& -0£8 ToF 289 695 £92 W £66 622

Rehabilitation Thera­ pists Rating First Set of r's 123 l£0 227 190 606 Second set of r's £77 23k 250 £03 577

I.Q,, Sociometric, Time Sample, Etc, First set of r's 673 619 30£ 70£ 6£6 335 579 Second set of r's 929 82£ O T F 70 585 B06

Numbering of Items corresponds to that in the Prediction Blank, ^Underlined coefficients are significant: ,708 c ,01 level ,576 - .0$ level 104 following three ways to establish the degree of improvement attained: (l) the number of correlations that increased, though not necessarily to a significant degree, are counted;

(2) the number of significant correlations obtained in the first trial are compared with the number obtained in the second trial; and, (3) the median correlations of the two trials are compared to determine if a significant increase occurred. Finally, in the concluding part of this section,

Table X is studied in more detail to determine the items on which the highest increases in accuracy were achieved; the items that were most consistently predicted significantly, and the items that were essentially unsuccessfully predicted with this smaller sample.

Number of r's that increased with the second trial.—

Inspection of Table X showed that seventeen coefficients increased in magnitude and nine decreased, yielding a net of eight increases in prediction. Hence the predictions were in the direction of improvement, but not large enough to reach statistical significance. It was interesting to break down the increases and decreases and discover that twelve of the increases, or approximately two-thirds of theip, occurred among the criterion ratings. The other five increases, or one-third, were from the items on intelligence, sociometric standing, etc. An analysis of the decreases revealed an approximately similar apportionment. Seven out of nine 105 decreases, or slightly more than two-thirds, were from the

ratings, and the remainder came from the objective criteria

Items. Hence the psychologists achieved their successes and failures to the same degree in the two categories of criteria.

Comparison of number of significant correlations.--

There were thirteen significant correlations in the first trial as compared with fourteen In the secnnd attempt. Thus the second set of predictions may again be regarded as being in the direction of Increased accuracy, though not at a statistically significant level. A breakdown of these find­ ings also provided some interesting side-lights on the predictive process. It was found that the psychologists did not achieve any Improvement In the Items correlated with the ratings. On both trials there were seven significant pre­ dictions with the nurses and aides ratings and one with the therapists ratings. The only increase in significant corre­ lation occurred among the items on intelligence, interper­ sonal relationships, and other variables.

Comparison of median correlation coefficients.— A third indication of increased accuracy was evidenced in the higher median correlation coefficient obtained in the second trial.

The median r of the first trial was .462, and on the second trial, it was .522. A test of significance between them by the Fisher z transformation method revealed that they were not significantly different. However, the direction was io6

definitely toward an increase in agreement with the criteria.

The discussion so far may be briefly summarised by

stating that a comparison between the first and second sets

of predictions by three methods revealed a definite trend

toward an increase in accuracy, though none of them was at

a significant level. This positive trend is encouraging, and with a larger group, these tentative results may be demon­

strated at a significant level. However, they do substan­

tiate the contention that experienced therapists are able to predict with stability, if not with increased accuracy, from limited observation of interaction in group therapy.

It is not certain at this time if the psychotic population employed has facilitated or interfered with predictive accuracy. However, this population is more likely to change in their adjustment, behavior, and personality characteris­ tics over the three months period.

Items with the largest increases in accuracy.--The

seven Items, or approximately one-fourth of the number predicted that had the largest increases were placed in this category. Pour of them were not significantly predicted on the first trial, and all seven reached the .01 level on the second attempt. Among the seven items the magnitude of increase in coefficient value was .220 or more. Each of the pairs of coefficients In Table XI was tested for significant difference by converting them into Fisher z coefficients and 107 submitting to a t test. None of the pairs reached statist! cal significance. On the small sample of twelve subjects, a z difference (z^ - z2 ) of 1.038 (.05 level) or 1.464 (.01 level) would have been required. The largest z difference in the table occurred in Item 16 (present intelligence) which had a z difference of 0.93.

TABLE XI

SEVEN ITEMS WITH THE LARGEST INCREASES IN ACCURACY N = 12

Item First Second No. Trial Trial

1 6 . Present functioning intelligence .619 .929 17. Present friendship standing .304 •P 5. Does he like to stay by himself ...? *2.80 .8285 15. How favorably does he regards himself? (Correlated with ward rating) .213 .622 9. When eating, walking, etc., how does he compare ...? (Correlated with ward rating) .367 .695 20. Level of social interaction In evening..335 .585 21. Medical staff action taken. «579 .806

Underlined coefficients are significant. .708 = •01 level .596 = .05 level

Items most consistently predicted accurately .— An item was considered to have been predicted consistently and significantly if the psychologists were able to attain a significant correlation with the criterion measure on both trials. There were ten such items as shovrr. in Table XII. 108

TABLE XII

ITEMS SIGNIFICANTLY PREDICTED ON BOTH TRIALS N = 12

Item First Second No. Trial Trial

1. Does he seem tired . . . or lively? (Correlated with ward rating) .595a .577 3. How hostile is he? (Correlated with ward rating) .585. 5. Does he like to stay by himself? (Correlated with ward rating) .5SP_ .828 7. How emotionally responsive is he in interpersonal relationships? .667 .195. 1 2 . Are his thoughts and feelings consistent? .601 .749 1 6 . Present functioning intelligence. .619 1 1 . Performance on therapy assignment. .506 .5ZZ 18. Work leadership preference at present time-; .704 .684 19. Wprk leadership preference four weeks from now. .646 .670 21. Medical staff action taken. .579 .80F aAll coefficients are significant. .708 - .01 level .576 = .05 level

Five of the items in Table XII (No. 1, 3, 5, 1, and 12) dealt with behavioral characteristics that may have been elicited in the therapy session and thus facilitated accurate predic­ tion. The other five items demanded a greater degree of inference on the part of the psychologists, and their suc­ cessful prediction two times in two attempts suggests that each observer was employing valid clues of his own.

Items essentially unsuccessfully predicted.— Table X was examined to determine those items which had been signi­ ficantly predicted on the first trial but non-significantly 109 on the second trial, as well as those Items that were un­

successfully predicted on both trials. All of these Items are presented in Table XIII.

TABLE XIII

UNSUCCESSFULLY PREDICTED ITEMS N = 12

Item No. First Second Trial Trial Items with decrement in correlations 13. Does he exhibit inappropriate mannerisms ...? .576 .466 14. Does he exhibit ... apprehension not called for. .638 .482 11. Performance on therapy assignment. (Correlated with time sample of work activities.). t<573 .344 Items non-significantly correlated on both trials

2. Is he placid ... or ... tense...? -.350 .442 4. What is his typical emotional tone? .039 -.099 6. How clean does he keep himself? .029 -.048 8. How much does he move around the hospital? .283 .289 10. What is his attitude toward author­ ity figures? .323 .492

5. Does he stay by himself or like being with others? (Correlated with therapists ratings) .123 .477 6. How clean does he keep himself? (Correlated with therapists ratings) .140 .232 8. How much does he move around the hospital? (Correlated with thera­ pists ratings) .227 .250 10. What is his attitude toward authority figures? .190 .403 aUnderlined coefficients are statistically significant. 110

Apparently it was difficult to predict Items 6 and

10 to this particular sample since the psychologists were

not able to correlate their predictions with either the

ratings of the nurses and aides or with the therapists.

In the sample of 25 subjects, both items correlated succes-

fully with the ward ratings, and Item 10 correlated signifi­

cantly with the therapists rating. It should be noted also

that none of the items on intelligence, interpersonal relationships, and other variables validated against more objectively measured criteria, were in the above table of unsuccessfully predicted items. This suggests that the level of success achieved in the prediction of the behavioral items may have been lowered by the less accurate criteria employed. CHAPTER V

DISCUSSION OF THE FINDINGS

This chapter is divided into two main sections.

The first part discusses the findings related to the level of accuracy achieved in prediction, and the second section takes up the data on the improvement In accuracy attained with a second observation.

ACCURACY OF PREDICTION

Table VII revealed that the psychologists had pre­ dicted significantly a total of fifteen correlations out of twenty-six. There were twenty-six correlations because five of the items were correlated with double criteria.

Twelve of these significant predictions were at the .01 level, and three were at the .05 level. The range of corre­ lations, from the lowest to the highest r, was .033 to .823, with the median r at .423 (.05 level).

Prediction of Overt Traits

It will be recalled from the description of the pro­ cedures carried out in the study that there was reason to believe that the criterion ratings of the nurses and aides were probably more valid and reliable than those obtained

111 112 from the therapists. The ward personnel received more intensive training in rating and observation, and their practice ratings achieved greater agreement (reliability) and greater correspondence with the psychologists' ratings

(validity). Therefore, there was reason for confidence in believing that the 71 per cent level of significant predic­ tion obtained with the nurses and aides criterion ratings was more representative of the degree of success the psychologists had been able to attain in the estimation of diverse behavior and personality characteristics from the observation of a group session.

The correlations between predictions and ward ratings were presented in Table VIII and it indicated that the rat­ ing of the patient's personal cleanliness, Item 6, was most accurately predicted (r = .823). It seemed that the indi­ vidual's dress and personal appearance were adequate indices for making an estimate. The group members attended the sessions In their usual hospital garb and made no special efforts In personal hygiene. This was the same view of the patient presented to the ward personnel, and the high r indicated that they and the psychologists were observing and rating the same qualities on this item.

The therapists, on the other hand, evidently rated the patients from another frame of reference or on different qualities on this item. Their ratings correlated only 113

slightly with the predictions (r = .192). The correlation

of the therapists' ratings with the nurses and aides rat­

ings yielded an even lower r of .043. In the rehabilitation

clinics the subjects of the study associated primarily with

patients outside of their own wards and the latter presented

a wide range of behavior adjustment. The therapists were

probably making the ratings against a more heterogeneous

population, compared with the psychologists and ward per­

sonnel. Secondly, clinic activity usually entailed a natural

amount of soiling of clothes and reduced personal neatness.

These factors may partially account for the independence of

the therapists ratings.

Chapter Three had pointed out that Item 6, cleanli­

ness, and Item 5, how much does he stay by himself?, were

employed in the MSRPP as a measure of "Schizophrenic With­

drawal." However, the correlation matrices revealed that

the three groups of raters, psychologists, nurses and aides,

and therapists, rated these two items with a low positive

or an inverse relationship, that is, a greater degree of

cleanliness was perceived to be accompanied by a greater

tendency to remain by oneself. The psychologists' predic­ tion intercorrelation on these two items was low and negative (r = -.0 7 5 ); the nurses and aides' ratings even were more markedly negative (r = -.224). The therapists' intercorrelation of Items 5 and 6, on the other hand, was 114

low positive (r = .158). These findings do not correspond with the relationship postulated by the iMSRPP. One explana­ tion for these results is that the relationship between these two behavior traits is mostly applicable to the more regressed patients, those who are unable to care for them­ selves, are actively psychotic and are still under closed ward supervision. Perhaps after the patient achieves a minimal level of self-care, it ceases to be an important function of other psychotic tendencies. The patients in the present study were probably beyond this critical level, being able to assume responsibility for their own physical comforts and cleanliness. Lastly, it should be pointed out that the positive trends found in the correlations between

Item 6 and the other items also correspond with the slight positive relationships between cleanliness, persistence and tendency for service found in the Character Education Inquiry study.

The triad of items in Category I, G-eneral Activity

Level, were also significantly predicted at the .01 level.

These were No. 8, movement around the hospital, No. 1, tired or lively, and No. 9, rate of walking, talking, and eating.

The predictions in this category 'were most successful and were probably based partly on the individual’s verbal parti­ cipation, rate of talk, and general physical activity in the therapy session. Furthermore, It was found that these items, 115 supposedly dealing with activity level, really tapped two different aspects of behavior. The intercorrelations between Item 1 (tiredness-llvellness) and Item 9 (dressing, walking, etc.) were .747 (psychologists) and .550 (nurses and aides), and these two items may be considered as measur­ ing a closely similar area of behavior. However, these two items were negatively correlated with No. 8, How much does he move around the hospital?, in order of -.029 to -.2 7 5 .

This finding suggests that the patients find enough activity in and near their ward so that the tendency to move around the hospital was not related to energy or general activity level.

The two items in Category II, Submlsslveness-Bellig- erance, were predicted successfully. These were Item No. 3 j how hostile is he? and No. 10, what is his attitude toward authority figures? A criterion rating was also obtained from the therapists on this last item and it correlated .604

(.01 level) with the prediction, representing a markedly closer agreement than that with the nurses and aides. This r of .604 was the highest obtained between a prediction rating and the therapists criterion ratings. One hypothesis for this finding is that the therapists were comparatively more isolated in their work with the patients, and their relationship with the patients was more definitely a teacher- pupil one. Hence their rating of the patients on the 116

Attitude Toward Authority Item may have had personal rele-

vance, based on the Individual's perception of the patient's

reaction toward him as an authority figure. Furthermore,

the perception of the patients' authority attitudes by the

therapists was not similar to that perceived by the nurses

and aides, as indicated by their low criterion intercor­

relation of .090. The psychologists' perception of

patients1 authority attitude was a broader and more general

one, encompassing the perceptions of both the therapists and

the nurses and aides, as indicated by the dual significant

r ’s. However, it was correlated to a greater degree with

the therapists' ratings. This finding suggests that the

nurses and aides have a less clearly perceived authority

role for themselves in this situation as compared with the

therapists.

The estimate of hostility (item No. 3) by the psycho­

logists was highly accurate (r = .781). Apparently hostil­

ity feelings, If present in the patient, manifested them­

selves prominently in the person's therapy and ward behavior.

These feelings are more easily elicited and identified in a group therapy situation as compared with assessing one's attitude .toward authority. Therefore, of these two items in the Submissiveness-Belligerance Category, hostility was more accurately predicted. This confirms the hypothesis on their relative levels of predictability stated in Chapter III. 117 Though the MSRPP placed these two items in the same category, in this study, neither the psychologists nor the nurses and aides regarded them as closely related, as indi­ cated in their intercorrelations which ranged from .020 to

.205.

There were three items on Category III, Depression vs. Manic Excitement, and they were predicted with varying degrees of success. No. 4, typical emotional tone, was not significant; No. 7, emotional responsiveness in interper­ sonal relationships, was predicted at the .01 level, and

No. 14, exhibiting uneasiness or apprehension,was signifi­ cant at the .05 level.

The successful prediction of Item No. 7 (r = .683) indicated that the group therapy situation provided a valid opportunity for the observation of this behavioral charac­ teristic of emotional responsiveness In interpersonal rela­ tionships. The prediction of Item No. 14 at the .05 level has Importance In demonstrating that items dealing with less overt behavior characteristics, such as uneasiness, apprehension, anxiety, can be rated with reliability and validity, given training and practice. This favorable find­ ing is encouraging. The accurate rating or any other esti­ mate of manifest tension in psychiatric patients is often crucial in the diagnosis and treatment of psychiatric syndromes. It is one of the variables most often included 118

In rating scales. Lorr and Rubinstein (68), in a factor analysis of 73 rating scales, extracted a bi-polar factor which they titled, "Manifest Tension vs. Relaxation."

Examples of Items with heavy loadings on this factor were:

"sleep well vs. sleep difficult," Irritable vs, rarely irri­ table," "loses emotional control vs. rarely loses emotional control," "agitated vs. unworried," and "tense vs. relaxed."

The latter form was the one in which the item in this study was stated.

Item No. 12, are his thoughts and feelings consis­ tent, of Category V, Conceptual Disorganization, was pre­ dicted significantly at the .05 level (r = .409). This sug­ gests that the hour of observation was adequate in estimating the patients' harmony in thought and feelings. It was probable that the special situation of therapy with its emphasis on verbal interaction,provided more opportunities for the observation of stereotypy, irrelevance, incoherence, and blocking of speech.

Iten No. 11, which dealt with the patient's perform­ ance on his therapy assignment, correlated significantly with the therapists ratings at the .01 level (r = .5 6 5 ).

The high level of agreement is all the more significant when it is pointed out that the group session would hardly be expected to provide the observers with many direct clues on work or assignment performance. The matrix of prediction lntercorrelations (Appendix VIII) provided additional 119 information on this item. The psychologists regarded good work performance as being associated with the likelihood of

release and with present leadership standing. These are

logical relationships. That is, a person ready for release is also most likely to be ready for work and to be nominated as a work leader. However, why the psychologists postulated the above relationships and ignored relating them to present friendship choice and with leadership choice in the near future was not clearly discernible.

It should also be pointed out that though the psy­ chologists' prediction of work performance agreed signifi­ cantly with the criterion rating of the therapists on this item (Table IX), it missed correlating significantly with the time sample of work activities. That is, the psycholo­ gists were not successful in predicting work performance to a criterion that was probably more objectively measured.

The criterion ratings of the therapists on this item, on the other hand, did correlate significantly with the time sample, though not as high as would have been expected.

The Inter-criterion r was .422 (.05 level), and this signi­ ficant agreement provides evidence on the validity of the therapists ratings on this item.

There were four predictions not significantly cor­ related with the nurses and aides criterion ratings, ami three predictions not significantly correlated with the 120 therapists ratings. In the nurses and aides group, Item 4, typical emotional tone, has already been pointed out. The other three were No. 13j does he exhibit peculiar mannerisms,

(r = .374), No. 15>how favorably does he regard himself,

(r = .201), and No. 2,is he characteristically placid or tense, (r = .5 0 ).

Item No. 13 approached significance, but it was sslll a surprise that the psychologists and the nurses and aides were not able to reach a greater congruence in the estimate of the patients1 inappropriate gestures, grimaces, and man­ nerisms. It would seem that such motor disturbances, If manifested by an individual, would be exhibited generally in his behavior. Inspection of Table V reveals that the mean of the psychologists' ratings on this item compared with that of the nurses and aides was significantly lower (28.64 vs. 18.68). This may mean that the psychologists perceived more signs of motor disturbance among the patients. Perhaps the patient exhibited a greater amount of motor disturbance in the therapy situation, which may be due to the more per­ missive atmosphere, encouraging the patient to behave as he felt most comfortable. In the ward situation, the patient may be more guarded and inhibited especially in the presence of hospital figures, such as the nurses, aides, doctors, and others. 121

The other two predictions not significantly corre­

lated with the nurses and aides criterion ratings were also

the two more independent behavior traits measured in that

they intercorrelated the least with the other criterion

measures. These were Item No. 2, placid or tense, which

did not correlate significantly with any other item, and

Item No. 15, self-regard, which correlated negatively with

intelligence (r =-.423, significant at .05 level) and with

the following items (though not at a significant level);

friendship choice (-.288), leadership choice (-.2 0 3 ), time

sample of ward activities (-.2 6 3 ), and medical staff action

(-.075). It will be recalled that these two items had compara­

tively low S.D.‘s in the prediction and criterion ratings,

and it had been suggested that they were not clearly under­

stood by the raters. This may explain the lack of relation­

ship between Item 2 and any other prediction and criterion measure. For the same reason, the negative correlations

reported for Item 15 should also be open to question. The relationships found for this Item suggested that the person with a favorable self-concept was one who was neither excessively self-depreciatory nor self-conceited, and who tended to be significantly below-average in intelligence.

Further, he was not chosen for companionship and leadership, was less active inward activities, and had a lesser chance o being released from the hospital. These negative findings

are contrary to the findings reported in the literature.

Newsttetter, Feldstein, and Newcomb (83) and Hunt and

Solomon (51), for example, found positive relationships

existing between friendship standing, intelligence, and

personal adjustment. Greenblat (38), in a study of 7th

grade girls found good mental health associated with high

sociometric scores and greater degree of social participation

(comparable to this study's measure of social activity on the

ward). Van Zelst, studying carpenters and bricklayers (129),

reported that the better adjusted workers were more satis­

fied with jobs, more popular, had better attitudes toward management and co-workers, and exhibited more confidence in management and supervisors. It should be pointed out that these selected references did not correlate a measure of

self-concept specifically with other measures. Usually an over-all estimate of personal adjustment was employed and correlated with sociometric and other data. However, the trend of positive relationships is evident, and the present contrasting findings should be regarded in a tentative man­ ner pending cross-validation and improvement in the con­ struction of the items.

Only two out of the five predictions correlated with the criterion ratings performed by the therapists were sig­ nificant. These were Item No, 10, attitude toward authority, 123 and No. 11, performance on therapy assignment. They have been discussed already. The non-significantly predicted items were No. 5, stays by himself, No. 6, cleanliness, and

No. 8, movement around the hospital. (These were signifi­ cantly predicted to the nurses and aides ratings at the .01 level.)

One reason for the non-significant correlation of the latter three items may lie in the lower validity and relia­ bility of the therapists ratings. However, the low r ’s obtained may also be a function of actual behavioral dif­ ferences manifested in therapy and ward situations as com­ pared with a rehabilitation clinic setting. Some evidence for this was found in the low criterion intercorrelation of

.220 between the time samples of work activities and ward activities. These two measures were more objectively obtained, compared with the criterion ratings. If the rehabilitation clinic and the ward can be considered as familiar and stable settings to the patients, it may be hypothesized that their behavior and demeanor in the one situation will carry over into the other. However, from the findings, the two situations were different enough, perhaps in both physical and psychological dimensions, to evoke observably different behavioral and adjustment characteris­ tics. This finding was in accord with that of Wandt and

Ostrelcher (13^0 who reported that the behavior of school 124

children altered with different classrooms, and the authors

had postulated that eacn setting had its own particular

social-emotional climate.

Prediction of Intelligence

The present functioning intelligence of the patients

was significantly predicted at the .01 level (r = .582,

Table X ). The mean I.Q. predicted was 98.0 as compared with

the tested mean of 102.6-4 on the Wechsler Bellevue Intellig­

ence Test, full scale. Furthermore, an examination of the

prediction intercorrelations (Appendix VTII) revealed that

the psychologistfi were successful In rating the overt

traits (Items No. 1 to 15) largely independent of their

estimate of intelligence. There were only three significant

correlations between the traits and intelligence and these

may be considered as a chance occurrence. This finding

corresponds with that of Tindall (126), who considered

Intelligence as a "variable of unknown effect" and found it non-significantly related to the indices of adjustment. On

the other hand, the Character Education Inquiry (43,44,45)

and the study by the Office of Strategic Services (128) did

report a slight positive, but significant relationship between intelligence and general adjustment. The Office of

Strategic Services investigation, for example, found a cor­ relation. of .24 between intelligence and emotional stability rating which was significant for the large sample of 397 to 125 433 candidates. In the Character Education Inquiry, the

three character trait3 of Tendency to Serve, Self-control

and Honesty (as contrasted to Deceit) were positively related to intelligence as well as personal adjustment. The authors reported a series of coefficients ranging from chance level to significance, with the trend in the posi­ tive direction.

The successful prediction of intelligence from limited data in this study corresponded to the report by Hanna (4l).

His study was performed in a vocational counseling setting, predicting the counselee's intelligence from an interview and intake questionnaire. Correlations between prediction and test scores were .71 with the A.C.E. Psychological Examina­ tion (timed test) and .66 with the Ohio State Psychological

Examination (power test). Both coefficients obtained by

Hanna were significant at the .01 level. Wilson (136) pre­ dicted intelligence from a structured clinical interview, using an industrial sample of engineers, salesman, and foremen, and obtained a correlation of .74 (.01 level) with the Personnel Problems Test, a short form intelligence test.

Except that Hanna and Wilson required a prediction with a single face-to-face contact, there were some significant differences between their experimental procedures and the present study. Hanna and Wilson provided the observer with background data prior to the interview, and during the 126

meeting, the observer had an opportunity to Interact with

the subject, directing the interview as necessary to glean

additional data to enhance accuracy of prediction, Tne

psychologists in the present study had a much more diffi­

cult task and yet were able to attain an r of comparable

magnitude, .582 as compared with ,66 to .71.

Prediction of Soclometrlc Relationships

Present leadership preferences (Item 18) were success

fully predicted (r = .5 7 1 * .01 level), but not friendship

preferences and future leadership preferences (Items No. 17

and 19). One reason for these contrasting findings has been

alluded to when it was pointed out that the psychologists

were generally unable to disciminate among the patients in

Items 17 and 19 as seen from the low S.D.'s obtained. How­

ever, successful perception of the complex dynamics of pres­

ent leadership preferences under the stringent conditions of

the investigation compared favorably with those of other

studies reported.

In the O.S.S. Assessment Study, leadership rating correlated with sociometric measure of leadership among the candidates attained an r of .68 (.01 level). Borgatta reported a much lower r of .211 (.01 level with the large N used) between superiors' ratings and sociornetric measure in a study of airmen. The higher correlations obtained in the 127

O.S.S. Study and this study may be attributed partly to tne

background and training of tne observers, who were psycholo­

gists, and partly to the fact that they were probably mere

conscientious in their prediction task. That is, as stated

before, they were aware of the experimental nature of the

study and that their ratings would be validated against tne

group nominations. In the Borgatta study, the superior's

ratings were routine assessments from the subject's service

folder.

It is interesting to note that the psychologists in

the present study regarded leadership standing and intelli­

gence as highly related. In fact their prediction of

intelligence correlated more highly with leadership standing

(r = .618) than it did with the intelligence test score

(r = .582). Both correlations were at the .01 level of

significance. The relationship between these two variables has also been perceived as significant by other raters. In the O.S.S. Study, for example, intelligence and leadership rating correlated .6 3 .

The patients themselves also considered present work leadership and intelligence as positively related, as demon­ strated in the intercorrelation of these two criteria, r = .540, significant at the .01 level (Appendix IX). Sim­ ilarly, Borgatta‘s airmen nominated their leaders partly or. a basis of intelligence, r = .287* significant at the .01

level (14). 128 The psychologists' prediction of leadership standing

four v,reeks from the date of observation (Item 19) correlated

.216 with a contemporary sociometric measure that was

obtained a month after the predictions. This item had been

Included to study the changes that mignt occur in this

brief period of time and to study its predictability by the

psychologists. As it turned out, tr.Is 'was the lowest predic­

tion correlation among the objective measures achieved by the

psychologists. However, the psychologists did expect changes

to occur, as shown in the low interccrrelation of .111

between prediction of present and future leadership standing.

This judgment was supported by the lnfcercorrelation of

-.076 between the two leadership sociometric measures. The

lack cf stability in the leadership standings of the group

members is contrary to Talland's findings (118) who

reported that the role relationships cf group members tended

to be stable. However Talland's subjects were neurotics at­

tending an outpatient clinic. Murray (78), in a socio­

metric study of the stability of personal relations among

retarded children, concurred with Taliand. He described

the individual child as tending to elicit the same number of

choices and to maintain his position in the group when

measured at three months intervals.

One reason for the contrasting finding in this study

1 may lie in the population of schizophrenics. Perhaps the

nature of their Illness has given their leadership choices little permanence, being more an index of current interper­ sonal relationships. Furthermore, the Matrix of Inter­ correlations of Criterion Measures (Appendix IX), revealed that this future leadership measure correlated with intel­ ligence at the chance level only (r = .080), as contrasted with the correlation of .5^0 (.01 level) between intelli­ gence score and the first sociometric measure of work leadership. This suggests that the patients were nominating their leaders with different criteria this second time.

Consequently it may be concluded that Item 19 may have been unsuccessfully predicted due partly to unidentified contam­ inating factors entering into the sociometric validation of the item. A later section will relate the successful pre­ diction of this item at the .05 level for the subgroup of

12 patients on two estimates taken over a three months interval.

The prediction of friendship choice (r = .300) approached significance, with an r of .396 being required for the .05 level. This correlation coefficient is one of the lower ones compared with those reported in the litera­ ture. However, it should be pointed out that the present sample differs from those of other studies quite markedly.

It will be recalled that Bass and White (9) used an experi­ mental design similar to the one in this study when they had observers predict the nominations of the members of leader- less discussion group and attained a range of r's from .25 130 to .60 (.01 level). Gronlund (39) secured significant results with 40 teachers predicting friendship choices cf sixth graders. He obtained a median r of .595. Newscetrer,

Feldstein, and Newcomb's findings (83) with camp counselors were similar. Moreno (75) on the other hand, reported that the teachers he studied were essentially unable to estimate the friendship standings of their pupils.

It has already been suggested that the findings on this item may be a function of the deviant population. Per­ haps the patients had concepts of friendship about which the staff members and the psychologists predicting the items were unaware. Secondly, the hour of observation may have been too limited to provide the observers with an accurate esti­ mate of friendship status of such a diverse group where the typical clues of friendship may not be as valid as compared with a normal group.

From the Matrix of Intercorrelations of Criterion

Measures (Appendix IX), it can be seen that the patients considered intelligence as a significant variable in their selection of friends (r = .693) and present leaders (r = .540).

These significant (.01 level) correlations of intelligence with friendship and leadership resemble that of other studies. Borgatta's airmen, for example, selected their friends (r = .198, 05 level) and their leaders (r = .2 8 7 .,

.01 level) partly on the basis of intelligence. 131

The psychologists in the present study were also cog­ nizant of the significant relevance of intelligence to friendship standing. The Intercorrelation between predic­ tion of friendship and intelligence was .464 (.05 level), which approximated the r of .693 obtained in the intercor­ relation of the two criteria.

Prediction of Level of Social Interaction

This prediction, correlated with a time sample of activities on the ward, was not significant, with an r of

.287. Inspection of the prediction intercorrelations (Appen­ dix VIII) reveals that the psychologists perceived the patient's level of social activity as being a positive func­ tion of his energy level (Item l), preference for company

(Item 5 ), emotional responsiveness in interpersonal rela­ tions (Item 7), rate of talking, walking, (Item 9), and his being chosen for friendship and leadership qualities

(Items 17, 18, and 19). The logic behind the positive rela­ tionships posited by the psychologists in the above instance can be perceived. Yet the intercorrelations among the criterion measures indicate that the time sample was actually related to cleanliness (Item 6 ), movement around the hospital (Item 8 ) and attitude toward authority (Item 10).

These findings suggest that this Item had failed to measure level of social activity and consequently had not provided an accurate measure with which to validate the psychologists' 132

prediction. Instead, the list of activities selected for

observation was most indicative of personal cleanliness,

tendency to visit other areas cf the hospital, and an

appropriate attitude toward authority. The list of activ­

ities observed had been made up on an a priori basis, and

this finding again points up tne need to first experimentally

ascertain the validity of the selected criterion itself.

Prediction of Medical Staff Action

Item 21, estimate of length of hospitalization still

required before the subject was placed on trial visit or

discharged M.H.B., was successfully predicted at the .01

level (r = .734). This prediction also correlated signifi­

cantly v/ith one other criterion measure, the sociometric measure of present leadership, (r = .465). Hence, this one

item significantly identified the leaders of the group and predicted the patients' chances of being released.

IMPROVEMENT IN PREDICTIVE ACCURACY

With the small sample of twelve subjects providing the data for this section of the study, the stability of the correlations was lowered considerably, and the results must be interpreted in a tentative manner. For example, ir. the presentation of the findings in Chapter III, It was

shown that some of the correlations achieved a marked

Increase in magnitude from one first to the second predic­ tion. However, with the small sample, these; increases were 133 not statistically significant. It may be concluded only that the increases in accuracy, measured by three different comparisons, were in the positive direction.

Table :ci presented the seven items in which the greatest Increase in accuracy had occurred. These were

No. 5, tendency to stay by himself, No. 9, rate of walking and talking, No. 15, self-regard, No. 16, Intelligence level,

No. 17.i present friendship standing, No. 20, level of social inter-action, and No. 21, likelihood of release. The dis­ cussion has already reported on the successful prediction on the larger sample on four of these items: intelligence level, tendency to be by himself, rate of walking and talk­ ing, and medical staff action. Consequently it may be con­ cluded that these four Items were the most predictable ones.

Of these four, the estimate of intelligence had the great­ est Increase in agreement, from a correlation of .619 between prediction and test score on the first trial to .929 on the second trial. This increase, however, was not significant when the r's were transformed into Fisher 2 values and applied to the z test.

This finding plus those presented for this item in the previous section discussing predictive accuracy are in agreement with the studies reported in the literature and indicate that tne estimate of intelligence can be highly accurate, especially in the low-average ability range. 13^ The prediction of likelihood of release within a

stipulated period of time improved from .579 (.05 level) to

.806 (.01 level). The hospital records revealed that, in

terms of the first prediction, two patients were discharged within six months and ten remained hospitalized. In terms

of the second prediction, two patients (the same ones as

above) were discharged within three months and ten remained hospitalized.

At this writing It has been approximately nine months

since the first observations and predictions were made. Out of the original group of twenty-five patients, only these ten patients remain in the special rehabilitation program.

The other fifteen have been either released from the hos­ pital or removed from the program because they were found to be incapable of handling the extra privileges and responsi­ bilities. The ten remaining patients, all of whom were in the subsample of twelve subjects, may be regarded as the

"most chronic1' of the chronic schizophrenic group used in this study. That is, these ten patients have stabilized so well in their schizophrenic process that nine months of in­ tensive. rehabilitation effort were not sufficient to bring them to a level of recovery to warrant release from the hospital. At the same time they have demonstrated enough ability and resources to assume responsibility and handle privileges properly, so that they have been kept on the program. 135 The intelligence level of the subsarnple of twelve

patients was lower, compared with the total group, and an

inspection of the means and standard deviations of all the

items for this subsarnple (Table VI), revealed that they had

consistently lower means in the prediction ratings and

generally lower means in the criterion ratings. This might

mean that the psychologists and the ward personnel rated

them lower as a rating tendency, or they might nave regarded

them as in poorer adjustment and mental health compared with

the total group of 25 patients. Furthermore, the increasing

accuracy from first to second trial with which the psycholo­

gists were able to predict the slight chance that these

patients would be released was probably a function of their

perceptivity of the patients' greater maladjustment, and

this corroborates the postulated reason for the reduced means

on the ratings. Though the psychologists were probably not

fully aware of these relationships entering ir.to their pre­

dictions, this finding provides further evidence that the

psychologists did proceed with tneir task in a consistent

and logical manner.

The three other items that increased markedly in

accuracy from the first to the second trial were Items No.

1 7 , present friendship choice, ho. 1 5 > concept cf nimself,

and No. 20, level of social activities. It will be recalled

In the discussion of predictive accuracy, using the larger

sample of 25 patients, that these three items were not suc­ cessfully predicted. In this present Instance, they were 136 again unsuccessfully predicted in the first trial. But all were significant in the second trial, and all, except Item

No. 20, increased to the .01 level. This suggests that a second observation, thougn three months apart, may have provided enough additional information to predict success­ fully.

It will be recalled that the items were considered as consistently successfully predicted if they were beyond the chance level on both trials. There were ten such items and they were listed in Table XII. Pour of these Items were intelligence level, present and future leadership standings, and likelihood of release. These have been discussed already as the items that had attained the greates increases from the first to the second trial. The other six items regarded as consistently predicted were lively or tired

(item 1), hostility (Item 3), how much he stays by himself

(Item 5), responsiveness in interpersonal relationships

(Item 7), performance in therapy assignment (item 11), and consistency in thinking (item 12). Perhaps one reason for the consistently accurate prediction of these items was

■chat they dealt with behavior traits that were more likely to be evoked in a therapy session. For example, tiredness or liveliness can be observed from the patient's physical behavior and verbal participation. Hostility is detectable in both his actions and his speech, and thinking disorganiza­ tion can be detected from his verbalizations. 137

There were twelve predictions made on the smaller

sample that might be regarded as unsuccessful. That is, the correlations either decreased from significance to chance level, or were not significant on both trials. There were eight items in this category, but four items were correlated to a second criterion, the therapists ratings, and this brought the total to twelve correlations.

Examination of these items in Table XIII indicated that, generally, they were the same items that were more poorly predicted in the larger sample. These were Item 2, placid or tense, Item 4, typical emotional tone, Item 13, inappropriate mannerisms, Item 14, excessive apprehension,

Item 11, performance on therapy assignment, correlated with the work time sample, Item 10, attitude toward authority, correlated with the nurses and aides rating, and the follow­ ing items, correlated with the therapists ratings; Item stays by himself, Item 6 , cleanliness, and Item 8 , movement around the hospital.

Some of the reasons suggested for the unsuccessful prediction of various items in the study of predictive accuracy also seem to be appropriate in explaining the above findings. It was believed that Items 2, 4, and 13 might have been poorly constructed and these bahavioral charac­ teristics might hot have, been clearly understood by the hospital raters. The psychologists might have been less 138

certain of these predictions, and the criterion measures

obtained v/ere not adequate.

The finding on the prediction of Item 11, performance

on therapy assignment, resembled that of the first part of

the study; namely, that the prediction was significant when

correlated with the therapists rating, but not with the

work time sample. It was suggested in the previous discus­

sion that the time sample recorded observable activities

only, and the therapists had rated their patients on factors

besides those used in the time sample, such as general

adjustment and attitude toward their task. The psycholo­

gists' prediction agreed more closely with the therapists'

frame of reference as compared with the more objective

criterion.

The finding on Item 10, attitude toward authority, was also similar to the results obtained in the study of predictive accuracy. It was suggested that the non-signi­

ficant correlations between the prediction and the nurses

and aides ratings might be due to a lack of certainty they had about their role among the hospital personnel and the

effect this had in their relationships with the patients.

The nurses ana aides work very closely with the professional staff which includes psychiatrists, ward physicians, psy­ chologists, and other consultants. In this hierarchy, they tend to take the non-authoritarian and follower roles 139 mostly. These tendencies might have resulted in a reduced

level of skill and an ambivalence to assume authority and

leadership roles with patients and to have a less clear

perception of the patients* attitude toward authority. The

rehabilitation therapists, on the other hand, functioned

more individually. They worked in clinics and were usually

in less contact with the professional staff. The relation­

ship between therapist and patient was believed to be more

clearly defined and resembled a pupil-teacher association.

The unsuccessful prediction of Item No. 5, stay by

himself, No. 6 , cleanliness, and No. 8 , movement around the

hospital, correlated with the therapists rating, was also

a repetition of the findings in the previous section of the

study. First it was pointed out that the therapists rat­

ings might have been less accurate and hence affected the

results of the predictions negatively. However, there was also a likelihood that the therapists were validly

rating a situation that was different from that of the ward or group therapy. On the matter of cleanliness, for example, the work of clinic usually resulted in a greater amount of

soiling and reduced personal neatness. The therapists might have had an inadequate opportunity to discover how much the patient moved around the hospital (Item 8 ) because their contacts with tne patients were usually confined to the clinic building, and for a maximum of four hours a day, five days a week. Lastly, the rating of the subject's 140 tendency to remain by himself or to associate with others

(item 5 ) was made more difficult because the patients were generally given specif ic assignments and tasks to perform

In the clinics. Part of the therapeutic procedure consisted in the therapists' coaxing and encouraging the patient to attend to their assignments. Under these circumstances, the natural inclinations of the patients to be an isolate or to associate with others would be diverted.

In concluding this discussion of the results obtained in the Sbudy of the increase in predictive accuracy, the find­ ings on the prediction of friendship and leadership stand­ ings (Items 1 7 , 18, and 1 9 ) of the subgroup may be compared with that of the larger sample. First, it will be recalled that in Table X it was shown that only the prediction of present work leadership standing (Item 18) was significant on the sample of 25 patients. With the subgroup, however, the prediction of these items reached a higher level of success, with five out of a possible six correlations signi­ ficant. The prediction of present and future work leader­ ship standings (items 18 and 19) was significant on both trials, and friendship choice (Item 1 7 ) was significant on the second trial.

Table XIV presents a summary of the most pertinent data on the three sociometric Items from Table V and VI and provides some, clues to explain the above findings. It 141

should be recalled first that Chapter III had explained that the original scores of the sociometric items could range from the extremes of plus 21 to minus 2 1 , whicn would occur if a subject were either nominated or rejected to all three situations by the other seven members of the group.

In the calculations, a constant was added to the scores to eliminate negative values.

TABLE XIV

MEANS AND STANDARD DEVIATIONS ON THE ITEMS ON FRIENDSHIP AND LEADERSHIP PREFERENCES OF THE TWO GROUPS

Item 17 Item 18 Item 19

Present Present Future Friendship Leadership Leadershi Standing Standing Standing

Mean Scores for sample of 25 subjects: 13.88 13.12 13.84 for sample of 12 subjects:

first measure 6.83 6.00 4.00 second measure 6.67 6.83 4.00 Standard Deviations for sample of 25 subjects: 4.56 4.48 4.25 for sample of 12 subjects:

first measure 1.57 1.53 1.15 second measure 1.49 1.52 1.41 142

Table XIV lists the means and standard deviations obtained

from the larger and smaller groups on these items. It may

be seen, for example, that the mean or average score on

Item 17 for the 25 subjects was 13.88. On the other hand,

the sub-sample of twelve only attained a mean or average

score of 6.83 and 6.67 on this item. These averages of the

two samples can be validly compared since the subsample was

always a part of a group of eight or nine subjects and had

an equal opportunity to receive votes and rejections as the

other subjects. Hence the data in the table suggests that

the subjects of the subsample were the less popular members

of the group, as indicated by their consistently and signi­

ficantly lower mean scores on all three Items. Furthermore,

this group was highly homogeneous, as evidenced by the very

small standard deviations obtained.

The discussion on the improvement of accuracy may now

be recapitulated by first emphasizing again the tentative­

ness with which any of the findings must be regarded. Con­

fidence in the significant estimate of an item should be placed on those Items with the highest correlations only.

The procedures used in the study permitted a comparison between the significant r's obtained with this subsample and with that of the original group, and there was evidence to believe that the sample of twelve patients had distinctive characteristics as a group that were not obvious in the larger population. 143 In conclusion, these findings of the study may be discussed In terms of the three hypotheses originally pre­ sented. Hypothesis No. 1 stated that the psychologists would be able to predict certain specific behavior traits significantly, using the criterion ratings of the ward per­ sonnel and therapists as validation measures. The data revealed that the predictions of the psychologists were significant in ten out of fourteen items validated against the ward ratings and In two out of six items validated against the therapists ratings and work time sample, or predicting successfully a total of twelve out of twenty

Items at the stipulated levels of confidence. This degree of accuracy, tested by Chi Square with Yates' correction for continuity, yielded a value of 1 1 6 .1 7 * which, for one degree of freedom, was significant from chance beyond the

.01 level. Hence Hypothesis No. 1 is considered proven correct.

Hypothesis No. 2 averred that the psychologists would be able to estimate at stipulated levels of significance the subject's present- functioning intelligence, interpersonal relationships of friendship and leadership, level of social activity in tne ward, and likelihood of release from the hospital. Out of the.six items in this category, three predictions were successfully estimated. This degree of accuracy, tested by Chi Square with Yates1 correction for 144

continuity, yielded a value of 17.02 which, for one degree

of freedom, was again beyond the .01 level of statistical

significance. Hence Hypothesis No. 2 is also regarded as

proven.

Finally, Hypothesis No. 3 stated that a second obser­

vation by the psychologists three months later would yield

a significant Increase in accuracy of estimate and predic­

tion, using the same item and observing the same subjects.

The findings of this part of the study, evaluated by three

methods of comparison, revealed that the second set of esti­

mates were in the proper direction of increased accuracy.

But they were not of a magnitude to attain statistical

significance. Hence this hypothesis is rejected.

SOME RELATED FINDINGS

The preceding analysis of the results has already

reported on instances of high inter-relationships discerned

among the predictions of the psychologists and among the

criterion measures. Though these findings are not germane

to the central problem of this study, their examination will probably provide hypotheses on the dynamics of the

predictive process arid the successful performance of the

psychologists in the present study. Factor analysis would

be an appropriate technique for the isolation of the prin­

cipal syndromes or clusters in the intercorrelations. Such

a precise analysis, however, is beyond the scope of the 145

present study. For this discussion, the intercorrelations

will be inspected subjectively and concentrated on elicit­

ing the most prominent impressions from the data and making

tentative interpretations.

Among the intercorrelations of predictions, one of

the tight clusters Included Items 1, 2, 5* 1, 17j 18.

It indicated that the psychologists regarded the subjects

who were more likely to be elected by the other group mem­

bers for friendship and leadership positions at the present

time (items 17 and 18) also possessed the behavior and

adjustment characteristics of being lively as contrasted to

being tired (No. 1), relaxed rather than tense (No. 2),

associating with others rather than by himself (No. 5), and

emotionally responsive in interpersonal relationships

(No. 7). However, the criterion intercorrelations (Appendix

X) did not bear out these predicted formulations. It

revealed that the group's friendship and leadership nomina­

tions were highly correlated with only one behavioral trait,

and this was a significant negative relationship with atti­

tude toward authority (No. 10), as rated by the rehabilita­

tion therapists. The rating by the nurses and aides on this item was also negatively correlated with these two socio- metric measures, but not at a significant level. Consequent­ ly, the criterion intercorrelations suggested that the individuals tending to be servile or antagonistic were chosen more often for friendship or leadership. The logic 146

of this latter relationship is less readily perceived than

that of the prediction correlations and will require further

investigation.

The psychologists also regarded present friendship

and leadership choices as being significantly related to

functioning intelligence (No. 16), level of social inter­

action with others (No. 20), and likelihood of release

(No, 21). The criterion matrix agreed with this cluster

partially, indicating that intelligence, friendship, and

leadership were significantly related, and the correlation

with likelihood of release approached significance. The

inter-relationships among these variables have also been

reported by other studies, such as by Borgatta (14),

Van Zelst (129), and Newstetter, Peldstein, and Newcomb

(83). The reason for the low correlation found between

level of social interaction (No. 20) and the other criter­

ion measures (No. 16 and 21) in this cluster of items was

suggested to lie primarily in the low validity of Item 20.

The activities selected for observation as a measure of

social activity were believed to be inappropriate.

The matrix of prediction intercorrelations also revealed that Item 7, emotional responsiveness in inter­ personal relationships, was significantly correlated with eleven out of twenty intercorrelations, the highest number for any item. The ubiquitous halo aspects of this item suggested that for the psychologists, it did not possess clearly defined characteristics and the item was probably

perceived as an index of general adjustment. They regarded

emotional responsiveness as a function of being tired or

lively (No. 1), relaxed or tense (No. 2), typical mood

(no. 4), staying by hirnself or being with others (No. 5),

cleanliness (No. 6 ), movement around hospital (No. 8 ), rate

of walking, (No. 9), intelligence level (No. 16), present

friendship and leadership standings (Nos. 17 and 18) and level

of social interaction (No. 20). Furthermore the prediction

of Item 7 with attitude toward authority (No. 10) and like­

lihood of release (No. 21) was also high and approached

significance at the .05 level.

Perhaps even more interesting is the discovery that

Item 7, rated by the nurses and aides, did not correlate

significantly with any of the other criterion items. The highest intercorrelation obtained was with Item 10, atti­

tude toward authority, with an r of .362 approaching the

.05 level of significance. This disclosure suggests that, unlike the psychologists, the nurses and aides were able to perceive the characteristics of emotional interpersonal responsiveness distinctly and to differentiate them from the qualities of the other items. Finally, and perhaps most

surprising, is that the prediction and criterion rating on Item 7 correlated significantly at the .01 level (r = ,6 9 3 ), establishing from a statistical point of view the validity of the prediction. However, it cannot be determined from 148 the data If there were an actual congruence In the charac­

teristics observed on this item by the psychologists and

the nurses and aides.

The generality with which the psychologists rated

Item 7 was an extreme case and was not Indicative of the

entire rating performance. An example of the psychologists'

differentiation of the items and consistency in performance

was demonstrated in their restricting the prediction of work

performance (No. 11) to two significant intercorrelations,

future leadership standing (No. 19) and likelihood of

release (No. 21). The logic for the relationships hypoth­

esized by the psychologists among the three Items can be

seen easily, and the criterion measures confirmed their predictions. Moreover, the matrix of criterion inter­ correlations revealed that work performance was also a

significant function of future leadership; that is, the good worker also stood high In leadership choices. However, it further showed that the work performance criterion ratings

(Item 11) had a positive but non-significant correlation with release from the hospital as revealed by the hospital records (Item 21). This result agrees with clinical obser­ vations that the rehabilitation therapists' report of work performance sometimes are not relaced to adjustment or level of recovery. For example, chere is occasionally found the compulsive "one gait" worker in the clinics who receives above-average work ratings which are Inversely 149 related to adjustment. However, In general, these two vari­ ables are positively related to a slight degree, as shown in the obtained inter-criterion r of .315. Finally, the psychologists' predictions revealed a tight little cluster of Intercorrelations among the items of psychiatric symp­ tomatology, the lack of harmony between thought ana feeling

(No. 12), inappropriate mannerisms (No. 13), and excessive apprehension (No. 14). The median r among these items was

.604, significant at the .01 level, suggesting that the observers generalized markedly among these three items.

The matrix also indicated that the psychologists regarded psychiatric symptoms as inversely related to cleanliness

(No. 6) and movement around the hospital (No. 8).

The nurse and aides ratings on these three items also revealed a significant degree of Intercorrelation but of a smaller magnitude compared with the predictions. The median criterion intercorrelation for these items was .464. How­ ever, the "halo" effect of these psychiatric items v«*as also more extensive, being significantly correlated with lively or tired (No. 1), typical mood (No. 4), hostility (No, 3, and authority attitude (No. 10).

It may be concluded from the above discussion that the psychiatric symptoms were not clearly distinguished by both groups of raters. There was a tendency to race them all in a similar manner, with the psychologists showing a greater inclination toward generality compared with the 150

nurses and aides, This may infer that the descriptions of

psychiatric symptomatology were not well differentiated, or

that these symptoms were thought of as occurring together.

As a final observation, the criterion ratings of the

nurses and aides may be compared with those of the therapists

on the items that both groups rated. The criterion matrix

3howed that the therapists ratings on cleanliness (No. 6 ),

staying by himself (No. 5 ),and authority attitude (No. 10) were significantly intercorrelated. However, these three

items were not significantly intercorrelated for the nurses

and aides ratings. Even more important is that these two groups were evidently rating different qualities, since the two sets of criterion ratings were not highly correlated.

Lastly, the predictions by the psychologist on these items correlated significantly with the nurses and aides ratings, but not with the therapists ratings. It was suggested that this may be due to the fact that the therapists ratings were less accurate and valid compared with the nurses and aides ratings. Secondly, the lack of relationship may have been due to different situation and environment in the clinics. CHAPTER VI

SUMMARY AND CONCLUSIONS

The Problem

The present Investigation was a study of the counsel­ ing process. Specifically, it sought to ascertain more precisely the accuracy of the guesses and specific hypoth­ eses that the counselor or psychotherapist spontaneously makes and alters during the course of counseling or psychotherapy. These "moment-to-moment" guesses, as

Meehl (70) called them, are an integral part of the counsel­ ing process and evolve naturally out of tne counselor's listening to, interacting with, and interpreting of the client's productions. They are regarded as serving two functions in the therapy process. First cf all, they may he used by the counselor or therapist to facilitate treatment.

That is, as these hypotheses occur naturally to the counsel­ or, they are explored with the client and used for the pur­ pose of assisting him to arrive at a better understanding of himself and of his problem. When these moment-to-moment guesses are used in this fashion, they are regarded as serving a therapeutic purpose, and their level of accuracy is relatively unimportant.

Secondly, these impromptu hypothese may be used for the purpose of making predictive diagnoses. That is, they 151 are employed by the counselor or psychotherapist to construct

an explanatory structure or hypothetical model of the

client, and, on the basis of this construction, to make treatment plans and predictions of the clients behavior or performance in the future. Examples of predictions that are often made on the basis of diagnosis are likelihood of recovery from mental illness, success or failure in a course of training, and behavior while on parole. If these extem­ poraneous hypotheses are used in this fashion, their degree of accuracy becomes one of crucial importance. Incorrect hypotheses only serve to subvert the predictive function and to reflect on the competence of the counselor. It should be emphasized that these two functions of hypothesis- making during the course of counseling have been distinguished arbitrarily for the purposes of this discussion. In the actual counseling situation, they are probably related and perhaps discriminated only with difficulty by the counselor while in the midst of the interaction with the client.

Shoben, for example, has contended that, "Each time a ther- pist responds to a patient, he implicitly predicts that his response will produce or contribute to Some kind of behavioral change in the patient, either immediately or in some longer- range sense" (7 7 , p.1 2 3 ).

Research in the accuracy of these specific hypoth­ eses which occur on the spur of the moment has been sparse. Investigations of predictive accuracy have been concentrated on the study of the forecasting of gross or complex behav­ ior. However, the need to assess the accuracy of these moment-to-moment guesses is being increasingly expressed.

First of all, there is still little evidence that counselors and clinicians are able to predict elementary facets of behavior accurately. Secondly, their assessment would con­ tribute to an understanding of the disappointing results on the clinical prediction of performance and adjustment, since these predictions are usually based partially on the impres­ sions and hunches gained from interview and observation supplemented with psychological test and case history data.

Finally, even if these guesses were used only to facilitate therapy, it is reasonable to assume that the counselor would prefer that these impromptu hunches were correct and valid rather than faulty and spurious.

The Present Study

The present study undertook to determine the degree of accuracy with which observers of a group psychotherapy session were able to predict behavioral and other charac­ teristics of the members who were mental patients. The observers were counseling psychologists who were exper­ ienced therapists and who were not acquainted with the sub­ jects involved in the study. This requirement of non­ acquaintance was carried out to avoid contamination of the predictions due to prior knowledge. The psychologists were

instructed to make their ratings and predictions on the

basis of their observations and interpretations of the

group interaction. The predictions were made separately by

the observers for each subject in the group. They were pooled later in the statistical analysis to obtain a single

score for each subject on each item. The predictions were

validated against their corresponding criterion measures and

the degree of predictive accuracy was established in the

form of product-moment correlation coefficients. A second phase of the study investigated the increase in accuracy which was achieved with a second observation of the subjects after a three-month interval. The procedure required the psychologists to make a second set of predictions which were validated against a new and contemporary set of criterion measures. The first and second sets of validity coeffi­ cients were then compared to note if an increase had occurred.

The sutdy tested three hypotheses. Hypothesis No. 1 stated that a group of psychologists observing a group psychotherapy session with mental patients would be able to predict present behavior and psychiatric symptoms at a statistically significant level when correlated with the criterion ratings of nurses, aides, and rehabilitation therapists. 155 Hypothesis No. 2 stated that the psychologists in

the same experimental situation described above would be

able to estimate more complex characteristics about the

subjects; namely, their present functioning intelligence,

present friendship standing, present leadership standing,

future leadership standing, level of social activity on the

ward, and likelihood of release from the hospital. These

estimates were correlated with more objectively obtained

criteria, such as psychological tests and sociometric measures, time sampling records, and hospital clinical

records.

Hypothesis No, 3 stated that given a second observa­

tion of the same subjects in group therapy three months

later, the psychologists would be able to predict the same

items with a significant increase in accuracy..

The three hypotheses were tested by twenty-one items

that made up the prediction task. R sample copy may be

found in Appendix I. Hypothesis No. 1 was tested by fifteen

items which were in the form of graphic rating scalesvmich were grouped into eight categories of behavior traits and psychiatric symptoms.

Category I: General activity level

No. la . Does he seem tired and worn out or lively and energetic?

aNumbering of the items corresponds to that in the Prediction Blank. Please refer to copy in Appendix I. 156

8. How much does he move around the hospital? 9. How does he compare in his rate of walking, talking, eating.

Category II: Submissiveness vs. hostility No.3. How hostile is he? 10. What is his attitude toward authority figures?

Category III: Depression vs. manic excitement No. 2. Is he characteristically placid and relaxed or tense and nervous? 4. Consider his typical emotional tone or mood. 7. How emotionally responsive does he appear to be in interpersonal relationships? 14. Does he exhibit concern, uneasiness, or appre­ hension to a degree that Is not called for by external circumstances?

Category IV: Withdrawal No. 5. Does he stay by himself or does he like being with others? 6 . Typically how clean does he keep himself?

Category V. Conceptual disorganization No. 12. Are his thoughts and feelings consistent or is there a discernible lack of harmony between them?

Category VI? Motor disturbance No.13. Does he exhibit peculiar and inappropriate gestures, grimaces, or mannerisms?

Category VII: Self-depreciation vs. grandiose expansiveness No.15. How favorably does he regard himself?

Category VIII; Performance on work assignment No.11. On his work assignment, how Is he regarded by his rehabilitation therapist?

Fourteen of the above predictions, Nos. 1, 2, 3, 4,

5, 6 , 7 , 8 , 9, 1 0 , 1 2 , 1 3 , 14, and 15, were validated against' the nurses and aides criterion ratings. Four of the predictions, Nos. 5 , 6 , 8 , and 10, were correlated with a second set of criteria,the therapists ratings. Item No. 11 was also correlated with two criteria, the therapists rating 157

and a time sample of work activities. Thus the fifteen

items on behavior and psychiatric traits yielded a total

of twenty validity coefficients.

The criterion ratings made by the nurses and aides and by the rehabilitation therapists were recorded on the

same rating scales as the psychologists, and examples of the rating forms used by the hospital raters may be found in Appendices II and III. Both the nurses and aides and the rehabilitation therapists were given training in prep­ aration for their rating task. An analysis of the data obtained from the training sessions with the nurses and aides revealed that they were rating their fourteen items with satisfactory validity and reliability. The therapists, on the other hand, were found to rate only Items No. 5, c,

8 , 1 0 , and 11 with fair validity and reliability, and their rating services were retained for these Items only.

Hypothesis No. 2 was tested by six Items. These items and their criterion measures were: Item 16. present functioning intelligence, which was validated against the patient's test score on the Wechsler-Bellevue Intelligence

Test Form II, full scale; Item 17. present friendship preference, Item 18. present work leadership preference,

Item 19, future work leadership preference. Items Nos. 17,

1 8 , and 1 9 , were validated against sociometric measures of the group on these variables. Item 2 0 . level of social 158 activity on the ward, was validated against a time sample record of the patients1 activities, and Item 2 1 , likelihood of release from the hospital, was validated against the hospital clinical records of actions taken.

Hypothesis No. 3 postulated a significant increase in accuracy with a second observation of the subjects in a group psychotherapy session three months later. The psy­ chologists repeated the entire prediction task of 21 items, and the first and second set of prediction correlations were compared to determine if the hypothesis had been substantiated.

Findings

Hypotheses No. 1 and No. 2 were tested on a sample of twenty-five hospitalized schizophrenics who were members of a special rehabilitation program. These two hypotheses were substantiated at the .01 level of confidence by the results obtained. The findings on the prediction of the fifteen behavioral traits tested by Hypothesis No. 1 are reviewed first, followed by a discussion on the results on the estimate of the more complex characteristics of intelli­ gence, interpersonal relationships and other variables.

Prediction of behavioral traits.--The fifteen rating scale Items on behavioral and psychiatric traits provided twenty prediction coefficients. This was due to the fact that five of the items were correlated with a second set of 159 criteria. Twelve of the correlations were statistically

significant. Of these, twn were correlated with the nurses

and aides ratings, while two of them were correlated with

the therapists ratings. Among the eight non-significant

predictions, four were correlated with the nurses and aides

ratings, three predictions were correlated with the thera­

pists ratings, and one prediction had been correlated with

a time sample of work activities.

The ten items which correlated significantly with the nurses and aides ratings were Nos. 1, 3, 5, 6 , 7 , 8 , 9,

1 0 , 1 2 , and 14.

Category I: General activity level No. 1. How tired and worn out or lively and energetic is he? 8 . How much does he move around the hospital? 9. How does he compare with other patients in rate of dressing, talking, and eating.

Category II; Submissiveness vs. hostility No. 3. How hostile is he? 10. What is his attitude toward authority?

Category III: Depression vs. manic excitement No. 7* How emotionally responsive does he appear to be in interpersonal relationship? 14. Does he exhibit concern, uneasiness, or appre­ hension to a degree that is not called for by external circumstances?

Category IV: Withdrawal No. 5. Does he stay by himself or does he like being with others? 6 . Typically how clean does he keep himself?

Category V: Conceptual Disorganization No.12. Are his - thoughts and feelings consistent or is there a discernible lack of harmony between them? 160

The four behavioral items which correlated non-

slgnifIcantly with the nurses and aides ratings were Nos.

2 , 4, 1 3 , and 1 5 .

Category H I : Depression vs. manic excitement No. 2. Is he characteristically placid and relaxed or tense and nervous? 4. Consider his typical emotional tone or mood.

Catetory VI: Motor disturbance No.13. Does he exhibit peculiar and inappropriate gestures, grimaces, or mannerisms?

Category VII; Paranoid self-depreciation No. 15.How favorably does he regard himself?

Prom the above, it can be seen that the predictions

on general activity level (Category I), thinking and feeling

disorganization (Category V), and withdrawal symptoms (Cate­

gory IV) were the most successful. It was hypothesized that

these behavioral characteristics were the ones most likely

to be elicited in a group therapy situation, and hence the

psychologists had a better opportunity to predict to them.

Two out of four items in Category II, depression vs. manic excitement, and the single items in Categories VI

(motor disturbance) and VII (paranoid self-depreciation)

were unsuccessfully predicted. Two reasons were offered for

these negative findings. First, it seemed that these items

represented psychiatric symptoms that were less clearly

defined than the others, and might have been open to greater

individual interpretation and diversity of opinion among the

contrasting groups of psychologists, nurses, aides, and therapists. Second, the descriptions along the rating graph 161

for these items may not have been clearly stated or under­

stood by the raters.

The two items successfully predicted to the rehabili­

tation therapists ratings were No. 10 and 1 1 .

Category II: Submissiveness vs. hostility N c . 10. What is his attitude toward authority figures?

Category VIII: Performance on work or therapy assignment No. 11. How is he regarded by his rehabilitation thera­ pist on his work assignment?

The three items non-signlficantly predicted to the

therapists ratings were No. 5, 6 , and 8 .

Category I: General activity level No. 8 . How much does he move around the hospital?

Category IV: Withdrawal No. 5. Does he stay by himself or does he like being with others? 6 . Typically how clean dees he keep himself?

The final non-significantly predicted Item among the behavioral traits was No. 11, performance on work assignment, correlated with a time sample of work activities. This

finding was interesting in that the prediction did not cor­ relate significantly with the more objective criterion, namely, the time sample, but It did correlate significantly with the therapists rating on the item. Furthermore, the therapists rating on Item 11 did correlated significantly with the time sample. This suggested that the two criteria were actually measuring different aspects of the therapy assignment and the prediction was successful in correlating with only one of them. The therapists ratings on Item 11, 162 on the other hand, did Include the characteristics recorded

In the time sample, and hence their ratings were In signifi­ cant agreement with the second criterion.

It will be recalled from the discussion above that the three items non-signiflcantly correlated with the thera­ pists ratings, Nos. 5, 6 , and 8 , had been successfully pre­ dicted to the ward personnel ratings. A reason suggested for the non-significant relationships obtained with the therapists ratings was that the activity in the Physical

Medicine and Rehabilitation Clinics evoked different types of responses from the patients. That is, their behavior was sltuationally determined to the extent that it lowered to chance level the predictions made on the basis of obser­ vation of group therapy. Furthermore, it was suggested that the therapists might have been rating the subjects from a different frame of reference, perhaps comparing the subjects of the study with patients from other wards who demonstrated wider variations in behavior and adjustment.

Estimate of more complex characteristics.--The six items tested by Hypothesis No. 2 resulted in three signifi­ cant and three non-significant predictions. The successfully predicted items were No. 16, present functioning intelligence,

No. 18, present work leadership preference, and No. 21, likelihood of release from the hospital. The non-signifi- cantly estimated items were No. 17, present friendship preferences, No. 19, future work leadership preferences, and

No. 2 0 . level of social Interaction in the ward. These last

six items corresponded more to the variables of gross events

that have been employed in other studies of prediction, and

the level of accuracy attained in this study is comparable

to those which have been reported. It was pointed out also

that the results obtained on the items on interpersonal

relationships (Nos. 17, 18, and 19) were different from the

findings reported by other studies. This may have been due

to the particular population used in the study, namely, a group of hospitalized schizophrenics. The lack of stability

in the choices of work leadership (No. 18 and 19) was most

apparent, and it was suggested that the role relationships of the group members might have shifted radically within the short period of four weeks.

Hypothesis No. 3 rejected.— The second part of the

study investigated the increase in accuracy which was achieved with a second observation of the same members in a group psychotherapy session three months later. The data to test Hypothesis No. 3 were obtained from a sample of twelve subjects who had been members of the original group of twenty-five. In the analysis of the data, the first and second sets of prediction correlations were compared by three methods. A consistent trend uoward an increase in accuracy was found, but not at a statistically significant level.

Consequently Hypothesis No. 3 was rejected. This negative 164

finding was considered as partly the function of the small

number of patients employed in the study and partly a func­

tion of the composition of the group. The members of the

sample were characterized as leveled-off schizophrenics who had neither regressed enough to be let off the special

rehabilitation program, nor were they able to reach a level of recovery where discharge from the hospital could be

considered.

Related findings.— Inspection of the intercorrela­ tions of predictions and the Intercorrelations of criterion measures provided some pertinent clues to the predictive and rating processes of the psychologists and hospital personnel.

It was pointed out that some of the clusters of high inter­ correlations seemed to have a logical basis. For example, the psychologists rated the subjects so that work leadership, work performance, and likelihood of release from the hospital were positively and significantly correlated. On the other hand, there were clusters of Intercorrelations which had a less obviously rational basis, and these findings prompt further study. For example, the intercorrelations of criterion measures Indicated that the patients who were either servile or antagonistic were also the Individuals who were selected for positions of friendship or leadership.

The data revealed that the psychologists were able to rate most of the behavioral items independently. There was only one case of an extensive halo effect, and this was

in the rating of Item 7, emotional responsiveness in inter­

personal relationships, which was found to be highly related

to 13 out of the other 20 items. The intercorrelations of criterion measures revealed that the rating of the same behavioral traits by nurses and aides and the rehabilita­ tion therapists did not correspond with each other. Further­ more, the predictions of the psychologists of these items correlated significantly with the ratings of the nurses and aides, but not with the therapists. This finding also sup­ ported the hypothesis that the behavior observed on the ward differed from that observed in the clinics, and that the group therapy behavior resembled ward behavior.more than clinic behavior.

Suggestions for Future Research

The significant success with which the psychologists were able to perform their task under the experimental con­ ditions of the study immediately poses the problem as to the characteristics of the successful predictor and the nature of the clues or variables he employs in carrying out his task. Taft (117) has reported that some of the person­ ality characteristics associated with skill in making social judgments include having a hard-headed attitude toward peers, being task-oriented, serious, unemotional, and being a student of the natural sciences rather than the 166

social sciences. Mowrer has suggested that the therapeutic

skills may be different from predictive skills, . . I t

seems quite likely that therapeutic and diagnostic under­ standing are different processes, and that therapeutic understanding is not related to the therapist's knowledge about the patient" (7 7 , p. 3 0 8 ).

The present study has not studied the characteristics of the psychologists, but their successful performance points up the need to compare their personality characteris­ tics with those reported by Taft (117), who examined predic­ tive skill in more socially-oriented settings. Furthermore,

Mowrer's hypothesis can be suitably tested with the present group of observers to determine if predictive skill is nega­ tively related to therapeutic skill.

Additional research in another direction was suggested from a review of the findings on factor analysis of psychia­ tric rating scales. It will be recalled that most scales were found to have a great deal of commonality among its items, as reported by Guertin (40) and Lorr and Rubinstein

(6 8 ). In the present study, the same tendency was noted

In the ratings of the psychologists, nurses, aides, and therapists. The psychologists perceived likelihood of release as being positively related to the person's intelli­ gence level, being chosen by others for leadership, being relaxed rather than tense, and having a good record on his work assignment. A visual inspection of the Table of 167 Criterion Intercorrelations (Appendix IX) reveals similar clusters of high correlations. To establish the factors or variables the raters employed, factor analytic proced­ ures similar to those used by Guertin (40) and Lorr and

Rubinstein (6 8 ) might be profitable. The factors extracted for the prediction and criterion ratings might also be com­ pared to determine if they are similar, and if not, to discover or. what different bases each group had made its ratings.

Another area of research is related to the fact that accuracy of prediction has been shown to be inversely cor­ related with level of confidence of the validity of the prediction. Kelly and Flske, for Instance, have inquired,

"Why is it . . . that our staff members tended to make their best predictions at a time when they subjectively felt relatively unacquainted with the candidates, when they had constructed no systematic picture of his personality structure" (6l, p. 406).

In the present study it may be assumed that the psychologists’ confidence in their predictive accuracy was at the lower level since they were completely unacquainted with the subjects. Perhaps the experimental restriction was a blessing in disguise, preventing them from distorting their primary observations with biased associations and

Inaccurate weighting of the observed behavior In the 168

prediction. Additional study in this area may take the

form of systematically increasing the amount of Information

available to the observers and noting its relationship to

level of accuracy. Both the Veterans Administration Clinical

Psychology investigation and the Office of Strategic Ser­

vices study attempted to measure the accuracy of a sequence

of predictions, and, both found that accuracy of prediction was only slightly related to the amount of data used to make

the estimates. Lastly, the replication of this study with an­ other population is an important next step in the cross-

validation of the present findings. The use of a group of normal adult males would yield findings that may most appropriately be generalized and compared with the present

study.

The above suggestions are the most prominent ones pointed up by this dtudy. Others are perceived with addi­ tional review of the data and with reading of related studies reported in the literature. The assessment of pre­ dictive accuracy, therapeutic skill, and their relationship is crucial to psychology's claim to the privilege to help and counsel people. The science and the profession have recognized the responsibility inherent in this sanction and have dug into the task with vigor and ingenuity. The chal­ lenge is an adventure and the end is never in question. APPENDICES

1 6 9 APPENDIX I

PSYCHOLOGISTS PREDICTION BLANK

Name of patient:______Rater:

Date Today

1* Compared with Exit Service patients, does he seem tired and worn out or lively and energetic?

« • • • almost com- tired as lively livelier and pletely worn as most more ener- out getic

2. Is he characteristically placid and relaxed or tense and nervous? placid and as relaxed distinctly conspicuously relax as average tense tense

3. How hostile is he

* no ho s- rela­ as much as relatively extremely tility tively average high hostile low

4. Consider his typical emotional tone or mood.

• • • • » « » deeply moder­ neither moder­ highly depressed ately depressed ately elated depressed nor elated elated

5. Does he stay by himself or does he like being with others'

• • • • « * • always usually about as Sisually always In stays by much alone in com­ company by himself as with pany of of others himself others others 170 171

6. Typically how clean does he keep himself? distinct- more slop- as neat neater and fastldi' ly sloppy py and and clean cleaner ous and dirty dirty than as most than most most

7. How emotionally responsive does ne appear to be In inter­ personal relationships? little Inadequate adequate a distinct a marked or nc response feeling over-re­ over-re­ sign of sponse sponse feeling

8 . How much does he move around the hospital?

« • ■% i i • • marked- a little moves about a little moving ly under­ underac­ as appro­ restless about active tive priately most o: the time

9. When in action (walking, talking, dressing, eating) how does he compare with other patients in the Exit Service Program? marked- a little at an average a little distinctly ly slow- slower rate faster faster er

10. What is he attitude toward authority figures? hostile passively reacts appro- looks up dependent and an- resistent priately to to them and ser- tagcnis- the situation vile tic 172

11. On his work assignment, he is regarded by his supervisor as: t « • • • • • diligent, good worker irregular avoiding reliable, by hospital worker his work ready for standards assignment working outside

12. Are his thoughts and feelings consistent, or Is there a discernible lack of harmony between them?

• • • * * • • consistent a little distinctly almost disharmon­ disharmon­ totally ious ious unrelated

13. Does he exhibit peculiar and Inappropriate gestures, grimaces, or mannerisms?

• « • • • * « none occasionally fairly repeatedly frequently

14. Does he exhibit concern, uneasiness, or apprehension to a degree that is not called for by external circumstances? 1 _ i J : unconcerned a little distinctly disrupting anxious anxious anxiety

15. How favorably does he regard himself?

• • « • • • distinct- inclined neither self- inclined to distinct­ ly self- to be depreciatory be self- ly con- deprecla- self-de- nor conceited important ceited tory predatory

16. His present functioning intelligence as measured by the Wechsler-Beilevue Test is probably:

borderline (I.Q. below 8 0 ) above average (1 1 0 -1 2 0 ) below average (8 0 -9 0 ) very superior (above 1 2 0 ) average (9 0 -1 1 0 ) 173

17. How well does this therapy group like him as a personal friend? In a sociometric choice, they would rank him with:

the upper third of the therapy group (most liked)

the middle third (median)

the lower third (least liked)

18. At this time, for the position of leader in a work situa­ tion with this group, this patient will:

be selected by the majority

receive occasional mention only

not receive any mention

19. Four weeks from now, the number of leadership choices given him will:

increase remain the same decrease

20. In the ward during the evenings, he Is most likely:

to be sitting alone and not interested in the activities around him

to be alone but will converse when someone addresses him

to initiate some sedentary type of social interaction himself, e.g., conversation, to play cards

to be participating in more active social interaction, such as billiards, pingpong, dancing.

21. The Medical Staff will place this patient on trial-visit status or discharge him M.H.B.

’ within 3 months __ within 6 months NOT within 6 months 1 7 4

22. Assuming that he will be placed on trial-visit status or discharged M.H.B. anytime in the next three months:

he will be able to hold a job well

he will be able to hold a job with only marginal adequacy

he will not work and will be dependent upon his family APPENDIX II

NURSES AND AIDES RATING FORM

Name of patient:______Date today:

Personnel making ratings:______

Instructions: To rate, check along the line on each scale the behavior most characteristic or typical of the patient at the present time. Do not spend too much time on any one scale. If you do not feel able to reach a decision quickly,go to the next scale and come back to it later. Do not hesitate to give extreme ratings if they are warranted. Raters naturally tend to rate toward the middle of the scale and are often too timid about rating a patient as very high or very low. Thank you for your assistance.

1. Typically how clean aces he keep himself? distinct- more as neat, neater than fasti- ly sloppy, sloppy clean as most dious dirty than most most

2. How much does he move around the hospital? marked- a little moves about a little moves ly un- under- as appropri- restless about deractive active ate most of time

3. Does he stay by himself or does he like being with others? always usually about as much usually with always staysby by alone as with others with himself himself other others

175 176

4. What is his attitude towards authority figures?

• * * * * — — — — — — — — — — A.— — -.-.-. — - — — — — — — — I* — — 1 hostile passively reacts appro- looks up dependent and an- resistent priately to to them and ser- tagonis- the situation vile tic

5. Compared with Exit Service patients, does he seem tired and worn out or lively and energetic?

• * • • i ______— ______a. — ...______:l ______i, almost com- tired as lively livelier pletely as most and more worn out energetic

6. How hostile is he? no hos- relatively as much as relatively extremely tility low average high hostile

7. Consider his typical emotional tone or mood. deeply moderately neither de- moderately highly depressed depressed pressed nor elated elated elated

8. Are his thoughts and feelings consistent, or is there a discernible lack of harmony between them? consis- a little distinctly almost totally tent dishar- disharmon- unrelated monious ious

9. Does he exhibit peculiar and inappropriate gestures, grimaces, or mannerisms? none occasionally fairly fre- repeacedly quently 177

10. Does he exhibit concern, uneasiness, or apprehension to a degree that is not called for by external circumstances? uncon- a little distinctly disrupting cerned anxious anxious anxiety

11. How favorably does he regard himself? distinct- inclined neither inclined distinctly ly self- to be self- self- to be self- conceited deprecla- deprecia­ deprecia- important tory tory tory nor conceited

12. How emotionally responsive does he appear to be in inter­ personal relationships? little inadequate adequate a distinct a marked- or no response feeling over-re- ly over­ sign of sponse response feeling

13. When in action (walking, talking, dressing, eating) how does he compare with other patients in.the Exit Service Program? markedly a little at an aver- a little distinctly slower slower age rate faster faster

14. Is he characteristically placid and relaxed or tense and nervous? placid as relaxed distinctly conspicu- ana as average tense ously relaxed tense APPENDIX III

REHABILITATION THERAPISTS RATING FORM

Name of patient:______Date today:

Personnel making ratings:______

Instructions: To rate, check along the line on each scale the behavior most characteristic or typical of the patient at the present time. Do not spend too much time on any one scale. If you do not feel able to reach a decision quickly,go to the next scale and come back to it later. Do not hesitate to give extreme ratings if they are warranted. Raters naturally tend to rate toward the middle of the scale and are often too timid about rating a patient as very high or very low. Thank you for your assistance.

1. How Is the patient on the work assignment? •m •* • • diligent, good worker irregular avoids work reliable, by hospital worker assignment ready for standards working outside

2. What is his attitude toward authority figures? hostile passively reacts appro- looks up dependent and an - resistent priately to to them and servile tagonis- the situation tlc

3. Typically how clean does he keep himself? «• : : : • distinct- more as neat, neater fasti” ly sloppy, sloppy clean as than dious dirty than most most most 178 179 4. How much does he move around the hospital?

* *------• - A . ------: 4. marked- a little moves about a little moves ly under- underac- as appropri- restless about active tlve ate most of time

5. Does he stay by himself or does he like being with others?

• - _ ------A. • i • — — - — ------I— — ------A. • always usually about as usually always stays by much alone with with by himself as with others others himself others APPENDIX IV

Friendship Choice Blank

Group N o . ___ Date:

Members of the Group:

You have before you a list of the men In this discus­

sion group. It is in alphabetical order and contains your

name too.

Below, on this sheet, are three activities which two

people may do together. Please write down the names of two

friend3 from the list whom you would like to have join you.

Then write down the names of two men you would NOT want to

share this activity with you. You may use the same persons

or different persons for each activity.

Please spell names correctly. Your selections will be

known to Mr. Chang only and will help him to know each of you better. Thank you,

M r . Chang

Activity No. 1

You are visiting Chillicothe this Saturday to see the town, to take in a movie, then return to the hospital.

Whomwould you choose from Whomwould you NOT wish the listto go with you? to go with you?

1 . 1 .______2 . 2 . ______180 181

Activity No. 2

If this ward had a double bedroom vacant and you were asked to pick a roommate to bunk with you:

Whom wouldyou choose from Whom would you NOT want the list to room with you? to room with you? 1 . 1 .______2. 2 .

Activity No. 3

Everything has been arranged for you to go to Columbus for three days to visit the city and to inquire about job possibilities:

Whom would you choose from Whom would you NOT choose the list to go with you? to go with you? 1 . 1 . ______2 . 2 . APPENDIX V

Leadership Choice Blank

Group No. ______Date:______

Members of the Group:

Today you will help us with a task that Is very simi­

lar to the one you performed the last time. You have before

you a list of the men in this discussion group. It is in

alphabetical order and contains your name too.

Today we would like to have you select the people

whom you would recommend In leading the group to do three

jobs. You will write down the names of two people from the

list whom you feel will best fill the jobs and two names whom

you would NOT recommend for the job. You may use the same

persons or different persons for each of the activities.

We would like to remind you that the leaders you

select should be the people whom you feel can be depended to

do the job properly and to be able to get you and the other men in the graup to work with him.

Thank you,

Mr. Chang

182 Activity No. 1

We need a person from this group to lead the others in

cleaning the Day Room in preparation for a visit by the

Gray Ladies.

Whom would you Whom would you NOT recommend for the job? recommend for the job?

1. 1.

2. 2.

Activity No. 2

We need a person from this group to lead the others in the

painting of some lockers to be used by patients.

Whom would you Whom would you NOT recommend for the job? recommend for the job?

1. 1.

2. 2.

Activity No. 3

We need a member of this group to direct the others in the

stapling of the hospital newspaper this afternoon.

Whom would you Whom would you NOT recommend for the job? recommend for the job?

1. 1.

2. 2. APPENDIX VI

TIME SAMPLING RECORD OP WARD BEHAVIOR

Name of Patient: ______Date Today:

Personnel making observations:______

Please check what patient was doing at the time of observation Do not consult anyone

Activity Time of Observation 1. Sitting on ward, not interested in activities _ __

2. Sitting, watching others ______

3. Watching, T.V.__

li Reading —______

5. Playing cards a l o n e ______

6. Playing checkers or cards with someone — —____„__

7. Playing pingpong, B i l l i a r d s ______

8. Talking with patient or ward personnel ______

9. Taking medication

10. Doing personal chore (e.g. shin­ ing shoes! ______

11. Taking a walk __

184 185

Activity Time of Observation

12. Sitting out on lawn ______

13. Golfing or attending ball game ______

14. Attending movie cr dance at Recr, Hall ______

15. Attending movie on ward ______

16. Visting another w a r d ______

17. Bathing, shaving, readying for bed -

18. Doing chore on ward -______

1<3 Other (wr*ifce in^ —

20. Don't know where he is ______APPENDIX VII

TIME SAMPLING RECORD OF BEHAVIOR ON

THERAPY ASSIGNMENT

Name of* patient ______Date today

Name of personnel making observations______

Instructions; Please check the activity the patient is en­ gaged in at the time of the observation. Do not consult anyone. Thank you.

Activity Time of Observation 1. Absent without excuse ______

2. Idle (chatting with others, reading own materials,etc.)„

3. Moving about aimlessly ______

4. Working on assignment ___- __

5. Receiving in­ structions from therapist ..

6. Attending class or group dis-

7. Out on assign­ ment or errand for clinic ___

8. Doing maintenance work for clinic __

1 8 6 Activity Time of Observation 9. Washing up to leave ____

10. Talking with therapist about personal problem

11. Other (please write in) ______

12 ______

13 ______188

APPENDI

MATRIX OP INTERCORRELATIONS OP N:

Item N o .a 2 3 4 5 6 7 8 9 10 11 1 291 -131 335 759b 004 486 -181 747 -0 1 3 -003 2 433 441 314 338 55o 382 333 437 033 3 -133 -031 -032 -0 5 5 "128 002 320 -0 6 5 4 185 085 614 393 165 374 383 5 -075 485 -175 687 -138 -097 6 490 2 4 2 043 321 135 7 551 487 346 061 8 -0 2 9 239 303 9 -141 -200 10 0 98 11 13 13 14 15 16 17 18 19 20--- aNumbering of items corresponds to that in the Predict ^Underlined coefficients are significant .396 : .0 5 Level .505 : .Q&iLevel APPENDIX VIII INTERCORRELATIONS OF PREDICTIONS BY PSYCHOLOGISTS N: 25

8 9 10 11 12 13 14 15 16 17 18 19 20 2: -181 Z41 -013 -003 254 406 042 296 325 432 4^4 423. 130 1 ! 382 333 437 033 219 -103 -008 -012 228 5 g [ 55F 370 280 4t ' -128 002 320 -065 401 109 428 102 157 15b 205 307 -277 O', ■ 393 165 374 383 070 -015 -112 -182 392. 284 231 349 307 2: r -175 681 -138 -097 130 145 -081 285 185 564 558 217 122 2- ' 142. o53321 135 -169 -438 -210 017 198 230 22F 071 25F 31 551 487 346 061 136 -156 -055 O76 516 566- 561 297 422. 3 -029 239 303 -383 - 539 -304 -080 370 33532F 023 125 5 - l4 l -200 -16 190 -177 512 425. 600 653. 058 452 2 098 299 219 378 -127 095 -TTF 031 343-019 -0 028 -162 -058 -256 277 168 093 475 256 4 657 422 151-186 -109 2W 025 -3 161 006 -267 -195 340 -178 -2 289 327 313 -097 004-0 464 ' 105 292 4 825. 120 426 5 111 5§F 4 522 1 4 ;o that In the Prediction Blank lificant APPENE

MATRIX OF INTERCORRELATIOt W-J

[tern N o . a 2 3 ii 5 6 7 8 9 10 12 13 lit 15 1 071 123 652* 3U3 113 252 -196 550 2k3 573 575 352 172 2 373 3IE 39ft 03U 1U3 2104 U 7 8 260 T 9 3 T o O 202 162 3 313 298 -1 1 6 207 093 128 205 621 57U 599 577 it 15U 197 355 109 13U 652 SBC STB 5 3 7 3 3 5 5 -22U 18 U -210; 290 0 B £ r o m 111 6 323 765 0 51 U09 -156 -33B-127 -357 7 Oitl 353 257 206 0 6U 101 8 -198 it27 -2 5 1 -228 -073 -2 0 6 9 -T53 218 2it5 ■-0 2 8 329 10 506 hok U86 115 12 Boo B l 3 U52 13 75B 35T lit 23U 15 5 a 6 a fin Numbering of the items corresponds to that In -- the Prediction Blank* lua 11a Wo* 1 to 1%: ratings by nurses and aides* Wrk. T.S* No. 5a, 6a, 8a, 10a, 11a: ratings by therapists* 16 No* 16: Wechsler-Bellevue Intelligence Test score* 17 No* 17, 18, 19: Sociometric measures IS No* 20 and Wrk, T S: Time sample records* 19 No* 21: Hospital clinical records* 2° . uJnderlined coefficients are significant: 3 9 6-.0 $ 1< APPENDIX IX

X OF INTERCORRELATIONS OF CRITERION MEASURES N-25

Wrk. 12 13 15 15 5a 6a 8a 10a 11a T S 16 17 18 19 20 21 573 575352 172 067 278 053 219 258 518 -093 -101 055 216 -211 T73 ToO 202 162 256 -050 -035 058 -109 T82 -256 -112 101 090 278 -055 621 575 599577 176 -100 162 257 091 533-007 001 -157 150 -008 152 FT5 F3T w 083 003 -021 083 151 550 -231 005 -099 366 172 -132 335 515 o59 511 178 055 301 -081 -299 058 -175 -057 095 096 052 008 -156 -238 • -127 -257 206 053 -125 -037 355 103 007 025 211 082 706 550 257 206 065 101 386 227-030 -005 112 203 179 067 173 026 135 158 -251 •-228 -073 -206 376-O il -019 005 169 260 110 125 090 130 557 555 218 255 -028 329-151 170 053 -059 -200 -055 073 050 378 -215 F35 TO- 506 5o5 586 115 355-208 -023 090 156 365-5 i5 -288 -325 318 578-055 Boo m 552 197 -228 009 170 205 515 -998-283 -280 220 155 -238 m 351 105 -092 310 250 076 559 -019 -063 -209 52 5 -0 66 -137 235 095 -230 -039 262 216 £52 -102 -101 -315 HU 051 -162 058 182 182 268 -135 089 -523 -288 -203 -079 -263 -075 158 289 233 539 303-052 -235 ■^168 089 375 260 r%H c? 4* a 530 557 102 053 172 275 012 -122 -275 039 nas lo und,t»X XI w 075 -075 105 l6l -238 019 -161 305 517 376-120 -185 -589 320 -107 -156 and aid es. 511 -056 -207 -192 527 318 103 ngs bytherapists. -035 -028 -168 £55 220 097 lligence Test score. 693 550 080 -212 391 asures EE 072 -185 151 le records. -076 110 353 rds* -009 001 01cl a0 nificant: 396-.05 le v e l; .505--.01 le v e l APPEND

MATRIX OF INTERCQRRELATIONS BETWEEN N-2

Cri ter ion

Items lb 2 3 k 5 6 7 8 ? 10 12 13 11+ 1 lb 559c1 053-125 350 299-130 263 -225 329 -0l*8 265 214; 137 01* 2 37H 150 U20 3k3296 269 692 187 081 251 339 229 109 35 3 1U6 299 751 169 220 -117 205 -030 169 020 hlh 255 371 61* k 1*52 —lhh -0l9 378 058 23k 1*27 181 017 251 m 278 122 -11 5 325 208-027 157 52U -2 93 W -318 255 -068 238 178 063 23 6 -111; 058 410 095 37k 823 hio 755 -067 3U1 -197 -1*19 -169 -06 7 150 ohi -027 232 159 m W 3 313 -031 251 033 ^oIH -055 02 728 8 -205 -067 -209 •092 -222 3HI 792 -275 216 -i;78— 1>■■ —1*1*9 -361 -29 9 311 001 -075 037 227 -T2? 187 -I65 552 -21*2 -01S 020 -11*8 19 10 J|28 -010 167 3h2 oilii 316 525 166 092 k31 361 235 21k 31 11 127 029 455 089 -269 339 o5o 198 -303 021 —llih -115 065 -hi 12 217 129 368 210 073 -328 2h5 -385 -058 -08? 1*09 297 381; 33 13 U85 057 *0U2 192 20 U —UUi -055 -530 229 -111; m 37U 260 16 Ik 083 187 333 119 090 -296 132 -172 -175 050 270 353 1*15 29 15 ■-0?h 0l*2 453 433 065 -2l;0 -051; -192 19k-267 -252 -290 -W? 20 16 122 -097 *252 486 -Oitl 207 21*8 091 237-331 -330 -215 -288 -30 17 125 356 066422 hi$ 181* 3hh 173 256 -170 -129 -159 -213 05 18 211 333 061 410 TJJJ 19U h5h 129 362 -110 -059 -082 -185 02 19 hoi 251 ll*3 551-035 -016 295 01*5 -228 232 361 206 1*02 05 20 330 172-211; 295 312 132 158 050 028 155 183 099 129 -05 21 oil* 017-036 Oil 235 395 179 381; -019 -063 -251; -180 -272 -03

^ h e criterion measures were obtained from the following sources: No. 11a: ratings by therapists; No. 16: Wechsler-Bellevue Intelligence No. 20 and Wrk. T S: Time sample records; No. 21: Hospital clinical *Tiisabering of items corresponds to that' in the Prediction Blank* Underlined coefficients are significant: .396".05 level; .5o5*.01 le APPENDIX X

5 BETWEEN PREDICTIONS AND CRITERION MEASURES N-25

a ‘iter ion Measures Wrk. 3 lit 15 5a 6a 8a 10a 11a T S 16 17 18 1? 20 21 h 137 0U0 127 282 02 3 -227 086 297 077 160 319 -067 -013 -162 9 109 353 286 376 219 010 175 2 Hi 015 010 0U8 031 089 261 5 371 6U6 -091 il*U -089 281* -009 U02 -091 -071 -082 006 -166 01*2 8 122 -TIB 211 238 Xbk 099 190 270 169 0U5 -026 31*3 —12it 176 8 063 233 322 208 030 -263 -167 123 -099 -051 213 021 -091 -113 9 -169 -060 339 192 -156 -060 217 1U7 -087 0U2 153 -132 57ll U25 h -055 028 U21 361 Qk2-il*5 -051 Ilt2 215 175 266 009 oil? 3*5 9 -361 -293 m 261 033 -059 159 026 229 168 137 111 303 610 0 -1U8 190 0li3 U56 185 -332 -336-062 2?2 296 1*95 -278 - 20U o 76 CO 5 2lU 315 293 3C2 027 60U 373 386 - y y -U08 -263 092 139 097 5 065 -U15 115 051 -136 o5i 565 315 12U -057 172 376 lOit 233 7 381* 333 033 257—088 3L*0 TI5 208 012 061 -030 -135 -282 -358 it 260 162 -189 336 0I48 307 -165 116 -001 -019 -083 -156 -28JU -55U 3 1*15 296 027 162 -111 1*76 -057 32L 036 -117 -171 -017 -178 -5B5 0 - 2C7 201 058 398-003 - 0C6 -318 - l 8 l 0U9 196 2iilt -351 -2 lf7 -253 K -288 -30U Oltl 362 021 -231 -128 —11*5? 582 282 618 -260 -038 UU5 9 -2 1 3 053 230 256 008 -299 -121 038 300 575 070 017 3 oH 2 -185 025 19lt 263 017 -315 - i 6 l 031 300 152 ?7T 019 111 3li7 6 ii02 052 059 0 6k -308 oi*8 2 39 596-12 3 -179 020 2X6-01*5 -089 9 129 -096 380 077 087 -J lii 071 -162 -122 177 073 287 187 0 -272 -038 2h2 160 168 -369 ■ 073 00 3 21s? 117 hog 097 202 73U rces: No, 1 to 15: ratings by nurses and aides;; No, 0a, 6a* 8a* 10a, elligence Test score; No. 17, 18, 19: Socicsfte.tric measures; 1 clinical records# ank, 505*.01 level* BIBLIOGRAPHY

1. Almy, H. C., and Sorenson, H. "A teacher rating scale of determined reliability and validity," Education Administration and Supervision, 16: 179-1&&, 1930,

2. Allport, G. V/. Personality. New York: Henry Holt & Co., 1937.

3. Ansbacher, H. L. "German military psychology," Psychol. Bulletin. 38:370-392, 19*41.

4. Arrington, Ruth E. "Time sampling studies of child behavior," Psychol. Monographs. 51, No. 2, 1939.

5. Bach, G. R. Intensive Group Psychotherapy. New York: Ronald Press, 1954.

6 . Baehr, G. 0. "The comparative effectiveness of individ­ ual psychotherapy, group psychotherapy, and a com­ bination of these methods," J. Consulting Psychol.. 18, 179-183, 1954.

7. Bales, R. F. Interaction Process Analysis. Cambridge, Mass.: Addison-Wesley, 1950.

8 . Barker, R. G. and Wright, H. F. One B o y ’s D a y . New York: Harper and Brothers, 1951.

9. Bass, B. M., and White, 0. L. Jr. "Validity of leader- less group discussion observers’ descriptive and evaluative ratings for assessment of personality and leadership status," Amer. Psychol., 5: 311-312. 1950.

10. Beehterev, W. , and Lange, M. "The results of the expert ments in the field of collective reflexology," Zsch. f. angew. Psychol.. 24: 224-254, 1924.

11. Bendig, A. W. "Rater reliability and judgmental fatigue, J. Applied Psychol.. 3 9 : 451-453, 1955.

12. Bills, R. E. "Non-directive play therapy with retarded readers," J. Consulting Psychol., 14: .140-149, 1950.

13. Bordin, S. S. Psychological Counseling. New York: Appleton-Century-Crofts, Inc., 1955.

191 192

14. Borgatta, E. F. ’’Analysis of social interaction and sociometric perception," Soclometry. 17: 7 “31, 1954.

15. Broderick, C. B. Predicting Friendship Behavior: A Study of the Determinants of Friendship Selection and Maintainance in a College Population. Unpub- lished doctoral dissertation, Cornell University, 1956.

16. Bronfenbrenner, Urie. "A constant frame of reference for sociometric research," Soclometry. 6 : 363-397. 1943.

17. Cadman, W. H. , Misbach, L., Brown, D. V. "An assess­ ment of round-table psychotherapy," Psychol. Mono­ graphs . 6 8 , No. 13, 1954.

18. Campbell, D. T. "A rationale for weighting first, second, and third sociometric choices," Soclometry. 17: 242-243, 1952*.

19. Cattell, R. B, Personality. New York: McGraw-Hill Co., 1950.

20. Chance, E. "A study of transference in group psycho­ therapy," International J. of Group Psychotherapy. 2 : 40-53, 1952.

21. Coffey, H. S. "Group Psychotherapy," in An Introduc­ tion to Clinical Psychology (Eds. Pennington and Berg), New York: Ronald Press, 1954.

22. Cohen, L. H., Malmo, R. E., and Thale, T. "Measurement of chronic psychotic overactivity by the Norwich Rating Scale," J. General Psychol.. 30; 6 5 -7 4 , 1944.

23. Cronbach, L. J. "Assessment of individual differences," in Annual Review of Psychology, Vol. 7, Eds., P. R. Farnsworth and Q. McNemar. Stanford, Calif.: Annual Reviews, Inc., 1956.

24. Cronbach, L. J. Essentials of Psychological Testing. New Y o r k : Harper and Brothers, 19^9.

2 5 . Cummings, S. T. "The clinician as a judge: Judgments of adjustment from Rorschach single-card perfomance," J. Consulting Psychol.. 18: 243-250, 1954. 193 26. Davis, S. E. "An Investigation of Client Characteris­ tics Shown in Interview Behavior." Unpublished doctoral dissertation, The Ohio State University, 1953.

27. Driver, Helen I. Multiple Counseling. Madison, Wis­ consin: Monona Publications, 1954.

28. Driver, R. "Employee Performance Evaluation," Person­ nel Series, No. 93, New York: American Management Association, 1945.

29. Eng. E. W. "An approach to the prediction of socio- metric choice, Soclometry, 17: 329-339, 1954.

30. Pallor, C. "Distinguishing marks of counseling," Occupations. 30: 260-263, 1952.

31. Ferguson, J, T., McReynolds, P., and Ballachey, E. L. Hospital Adjustment Scale. Stanford, California: Stanford University Press, 1953.

32. Fletcher, F. M. Jr. "The role of counseling psychology in rehabilitation," J. Counseling Psychol.. 1: 240- 243, 1954.

33. French, R. L. "Sociometric status and individual ad­ justment among naval recruits," J. Abnormal Social Psychol. . 46: 64-72, 1951.

34. Fuller, E. M., and Baune, A. H. "Injury-proness and adjustment in a second grade," Soclometry. 14: 210- 225, 1951.

35. Gersten, C. "An experimental evaluation of group therapy with juvenile delinquents," International J. Group Psychotherapy, 1: 311-318, 1951.

3 6 . Goodenough, Florence L. "Measuring behavior traits by means of repeated short samples, J. Juvenile Research. 12: 230-231, 1928.

37. Guilford, J. p. (Fd.) Printed Classification Tests. AAF Aviation Psychol. Program. Research Report No. 5. Washington, D.C.: Gov‘t. Printing Office, 1947. 194

38. Greenblat, E. L. "Two additional studies In the dynamics of school structure of class room seating and school dances," J. Educ. Research. 47*. 261-270, 1953-1954.

39. Gronlund, N. "The accuracy of teachers' judgment con­ cerning the Sociometric status of sixth-grade pupils," Soclometry Monographs. No. 25, 1951.

40. Guertin, W. H. "A factor analysis of schizophrenic ratings on the Hospital Adjustment Scale. J. Clini­ cal Psychol.. 11: 70-73, 1955.

41. Hanna, J. "Estimating Intelligence by interview'," Educ. Ps?/-chol. Measurement, 10: 420-430, 1950.

42. Harris, R. E. "Clinical methods: Psychotherapy," In Annual Review of Psychology. Vol. 7. Eds. P. R. Farnsworth and Q. McNemar. Stanford, California: Annual Reviews, Inc., 1956.

43. Hartshorne, H., and May, M. A. Studies in Deceit. New York: MacMillan Co., 1928.

44. Hartshorne, H., May, M. A., and Mailer, J. B. Studies in Service and Self-Control. New York: MacMillan Co., 1929.

45. Hartshorne, H., May, M. A., and Shuttleworth, F. K. Studies in the Organization of Character. New York: MacMillan Co., 1930.

46. Hemphill, J. K., and Sechrest, L. "A comparison of three criteria of aircrew effectiveness in combat over Korea," Amer. Psychol.. 7: 3 9 1 , 1952.

47. Hilton, A. C., Bolin, S. F., Parker, J. W. Jr., Taylor, E. K., and Walker, W. B. "The validity of personnel assessments by professional psychologists," J. Ap­ plied Psychol., 39: 287-293, 1955.

48. Hobbs, N. "Group-Centered Psychotherapy," In Cllent- Centered Therapy by Carl Rogers. New York: Houghton- Mifflin Co., 1951.

49. Hoffman, A. E. "An Analysis of Counselor Sub-roles." Unpublished doctoral dissertation, The Ohio Stare University, 1956. 195 50. Hunt, J. McV., and Cofer, C. N. "Psychological Deficit," in Personality and the Behavior Disorders, Vol. II (Ed. J. McV. Hunt), N e w York: Ronald Press, 1944.

51. Hunt. J. M . , and Soloraan, R. L. "The stability and some correlates of group status in a summer camp of young boys," Amer. J. Ps?/chol. , 55, 33-45, 1942.

52. Hurlock, E. B. "An evaluation of certain incentives used in school work," J. Educ. Psychol., 16: 145- 159, 1925.

53. Hurlock, E. B. "The value of praise and reproof as incentives for children," Archives of Psychol., 16: 145-159, 1925. 54. Imber, S. D. "Short-Term Group Therapy: An Experimental Investigation of Effectiveness for Psychotics and A Comparison of Different Therapeutic Methods and Different Therapists." Unpublished doctoral disser­ tation, University of Rochester, 1953.

55. Jenkins, R. L. "The schizophrenic sequence, withdrawal, disorganization, psychotic reorganization," Amer. J. Orthopsychiatry. 22: 738-748, 1952.

5 6 . Jenkins, R. L., and Holsopple, J, Q. "Criteria and Experimental Design for evaluating results of lobotomy," Psychiatric Treatment, Vol. 31, Proceed­ ings of the Association for Research in Nervous and Mental Diseases. Baltimore: William and Wilkins Co., 1953.

57. Jennings, Helen H. "Structure of leadership - develop­ ment and sphere of influence," Soclometry. 1: 99-143, 1937. 58. Johnson, 0. R. "is the sales manager's opinion of that hew salesman worth anything?" Printer's Ink, 128: 25-28, 1924.

59. Jones, M . , et al. The Therapeutic Community. New York: Basic Books Co., 1953.

60. Klapman, J. W. Group Psychotherapy: Theory and Prac­ tice. New York: Grune and Stratton, Inc., 1946. cl. Kelly, E. L . , and Fiske, D. W. "The prediction of suc­ cess in the V, A. Training Program in Clinical Psy­ chology," Amer. Psychol.. 5: 395-406, 1950. 196

62. Kelly, E. L., and Fiske, D. W. The Prediction of Per­ formance In Clinical Psychology. Ann Arbor, Michi­ gan: University of Michigan Press, 1951*

6 3 . Kempf, E. J. "The behavior chart in mental disease," Amer. J. Insanity. 71: 761-7 7 2 , 1915.

64. King, G. P., Erhmann, J. C., and Johnson, D. M. "Exper­ imental analysis of the reliability of observations of social behavior," J. Social Psychol., 35: 151-160, 1952.

6 5 . Kotkov, B., and Meadow, A. "Rorschach criteria for con­ tinuing group psychotherapy," International J. Group Psychotherapy. 2: 324-333, 1952.

66. Lipkin, S. "Notes on group psychotherapy," J. Nervous Mental Diseases. 107: 450-479, 1948.

6 7 . Lorr, M. "Rating scales and checklists for the evalua­ tion of psychopathology," Psychol. Bulletin. 51: 119-1 2 7 , 1954.

68. Lorr, M., and Rubinstein, E. A. "Factors descriptive of psychiatric outpatients," J. Abnormal and Social Psychol. . 5 1 : 514-522, 1955.

6 9 . MacKinnon, D. W. "The Structure of Personality," In Personality and the Behavior Disorders. Vol. I (Ed. J. McV. Hunt), New York: Ronald Press, 1944.

70. Meehl, P. E. Clinical Vs. Statistical Prediction. Minneapolis, Minn.: University of Minnesota Press, 1954.

71. Meehl, P. E., Tiedman, D., and McArthur, C. "Symposium on clinical and statistical prediction," J. Counsel­ ing Psychol.. 3 : 163-173, 1956.

72. Mehlman, B. "Group play therapy with mentally retarded children," J. Abnormal and Social Psychol. , 48: 53-60, 1953.

73. Moore, T. V. "The essential psychoses ana their funda­ mental syndromes," Studies in Psychology and Psychi­ atry, III. Baltimore: Williams and Wilkins Co., 1933. 197

74. Moreno, J. L. Das Stegreiftheater. Potsdam: G. Klepenheuer Verlag, 1923.

75. Moreno, J. L. Who Shall Survive? Washington, D.C.: Nervous and Mental Disease Publishing Co., 1934.

76. Mowrer, 0. H. "Anxiety theory as a basi3 for distin­ guishing /een counseling and psychotherapy," Minnesota Studies in Student Personnel Work, No. 1 (Ed. R. H. Berdie), 195^.

77. Mowrer, 0. H. Psychotherapy Theory and Research. New York: Ronald Press, 1953.

78. Murray, H. "The sociometric stability of personal relations among retarded children," Soclometry Mono­ graphs , No. 28, 1953.

79. Nash, Helen T., and Stone, A. "Collaboration of thera­ pist and observer in guiding group psychotherapy," Group Psychotherapy. 4: 8 5 “93, 1951.

80. Needelman, S. D. "Helping patients achieve occupational reintegration," Personnel and Guidance Journal, 33: 448-450, 1955.

81. Newburger, H. M., and Schauer, G. "Sociometric evalua­ tion of group psychotherapy," Group Psychotherapy, 6 : 7 -20, 1953.

82. Newcomb, T. M. "The consistency of certain extrovert- introvert behavior patterns in 51 problem boys," Teachers1 College Contributions to Education, No. 382, 1929.

83. Newstetter, W. J. , Feldstein, M. J. , and Newcomb, T. M. Group Adjustment: A Study in Experimental Sociology. Cleveland, Ohio: School of Applied Science, Western Reserve University, 1948.

84. Olson, W. c . "The measurement of nervous habits in normal children," Institute of Child Welfare Mono­ graphs , No. 3, Minneapolis, Minnesota: University of Minnesota Press, 1929.

85. Paterson, D. G. "Rating," In Handbook o.f Applied Psy­ cholog?/ . Vol. I (Eds. D. H. Fryer and E. R. Henry), New Yo rk: Rinehart and Co *, 1950. 198 8 6 . Pennington, L. A. "Rehabilitative Approaches," in An Introduction to Clinical Psychology. (Eds. L. A. Pennington and I . a VB erg.) New York: Ronald Press, 1954.

87. Peplnsky, H. B. "Research Notes from Here and There - The Strategy of the Counselor," J. Counseling Psychol. 3 : 222-226, 1956. 88. Personnel Research Section, A. G. 0. Survey of the Aptitude for Service Rating System of the U. S. Mili- tary Academy, West Point, New York. Washington,!).C. : Adjutant General's Office, 1953.

89. Peters, H. N., and Jones, F. D. "Evaluation of group therapy by means of performance tests," J. Consult­ ing Psychol.. 15: 3 6 3 -3 6 7 , 1951.

90. Plant, J. S. "Rating scheme for conduct," Amer. J. Psychiat.. 1: 547“572, 1922.

91. Powdermaker, Florence B., and Frank, J. X). Group Psy­ chotherapy. Cambridge, Mass.: Harvard University Press, 1953.

92. Puffer, J. A. "Boys' gangs," Ped. Sem., 12: 175-212, 1905.

93. Riddle, Ethel M. "Aggressive social behavior In a small social group," Archives Psychol.. 12: No. 78, 1925.

94. Roberts, B., and Strodtbeck, F. L. "Interaction process differences between groups of paranoid schizophrenic and depressed patients," International J. Group Psychotherapy, 3*. 29-41, 1953-

95. Robinson, F. P. Principles and Procedures in Student Counseling. New York: Harper and Brothers, 1950.

96. Rogers, C. R. Counseling and Psychotherapy. New York: Houghton Mifflin and Co., 1942.

97. Rogers, C. R., and Bymond, Rosalind. Psychotherapy and Personality Change. Chicago: University of Chicago Press, 1954.

98. Rosenwald, A. K. "An assessment of the current role of psychodiagnostic testing," J. Consulting Psvchol.. 18: 311-315, 1954. 199

99. Rowell, J. T. "An objective method of evaluating mental status," J. Clinical Psychol., 7: 255“299, 1951.

100. Sacks, J. M . , and Berger, S. J. "Group therapy tech­ niques with hospitalized chronic schizophrenic patients," J. Counsultlng Psychol.. 18: 297-302, 1954.

101. Saslow, G., Gressel, G. C., Shobe, P. 0., Dubois, P. H., and Shroeder, H. A. "Possible etiologic relevance of personality factors in arterial hypertension," Psycho­ somatic Medicine. 12: 2 9 3 -3 0 2 , 1950.

102. Scherer, I. V/. "A behavior rating scale for use in activity therapy situations," Information Bulletin. Dept, Medicine and Surgery, Psychiatry and Neurology Division, Veterans Administration, Jan., 1951.

103. Shaskan, D. A. and Joseph, Miriam. "War and group psychotherapy," Amer. J. Orthopsychiat., 14: 571“577, 1944. '

104. Shatin, L. and Freed, E, X. "A behavioural rating scale for mental patients," J. of Mental Science, 101: 644-653, 1955.

105. Shaw, F. J. "Counseling," in Annual Review of Psychol­ ogy. Vol. 8. Eds., P. R. Farnsworth and Q. McNemar. Palo Alto, Calif.: Annual Reviews, Inc., 1957.

106. Singer, J. L. Review of Bach*s Intensive Group Psycho­ therapy . J. Abnormal and Social Psychol., 50: 153- 154, 1955.

107. Slavson, S. R. "Group psychotherapy," Scientific American. 183: 42-45, 1950.

108. Slavson, S. R. "Common source of error and confusion in group psychotherapy," International J. Group Psychotherapy. 3: 3“28, 1953.

109. Smith, Marion R., Bryant, J., and Twitchell-Allen, Doris. "Sociometric changes in a group of adult female psychotic patients following an intensive socializing program," Group Psychotherapy, 4: 145“ 155, 1951. 110. Sorenson, A, G., and Gross, C. F. "inter-rater relia­ bility from the viewpoint of the rater," Personnel and Guidance J .. 3 5 : 365-3 6 8 , 1957. 200

111. Speroff, B. , and Kerr, W. “Steel mill ‘hot strip' accidents and interpersonal desirability values,” J. Clinical Psychol., 8 : 8 9 “91, 1952.

112. Strang, Ruth. "Seven ways to improve the rating process," Occupations. 29: 107-110, 1950.

113. Strodtbeck, F. L., and Hare, A. P. "Bibliography of small group research {from 1900 through 1 9 5 3 )," Sociometry. 17: 107-178, 1954.

114. Super, D. S. "Comments on current books and the pass­ ing scene," J. Counseling Psychol. , 1: 2 7 3 -2 7 6 , 1954.

115. Super, D. 2. "Transition from vocational guidance to counseling psychology," J. Counseling Psychol., 2: 3-9, 1955.

1 1 6 . Symonds, P. M. Diagnosing Personality and Conduct. New York: Century Press, 1931.

117. Taft, R. "The ability to judge people," Psychol. Bulletin. 52: 1-23, 1955. 118. Talland, G. A. "Role and status structure in therapy groups," J. Clinical Psychol., 13: 27-32, 1957-

119. Taylor, F. K. "The pattern of friendliness and domin­ ance in a therapeutic group," J. Mental Science, 96: 407-425, 1950.

120. Taylor, F. W. "Group management," Trans. Soc. Mech. Eng. , Nc. 24, 1903.

121. Terman, L. X. "A preliminary study of the psychology and pedagogy of leadership." Ped. Sem., 11: 413“ 451, 1904.

122.Thomas, Dorothy. Some new techniques for studying social behavior. New York: Teachers College, Columbia University, Bureau of Publications, 1929. Also as Child Development Monographs, No. 1, 1929.

123. Thomas, Dorothy S., Loomis, A. M., and Arrington, Ruth E. Observational Studies of Social Behavior: Vol. I. Social Behavior Patterns. Institute of Human Relations, Yale University, 1933. 201

124. Thorndike, R. L. and Hagen, Elisabeth. Measurement and Evaluation In Psychology and Education. New York: John Wiley and Son, 1955.

125. Thorne, P. C. Principles of Personality Counseling. Brandon, Vermont: J. Clinical Psychology, 1950.

126. Tindall, R. H. "Relationships among indices of adjust­ ment status," Educ. Psychol. Measurement, 15: 152- 162, 1955.

127. U. S. Dept. Army. A. G. 0. P. R. B. "A study of officer methodology: III. Order of rating and validity of rating," Personnel Research Bureau Report. No. 902, 1952.

128. U. S, Office of Strategic Services Assessment Staff, Assessment of Men. New York: Rinehart and Co., 19^8.

129. Van Zelst, R. H. "Worker popularity and job satisfac­ tion," Personnel Psychol., 4: 405-412, 1951.

130. Van Zelst, R. H. "Sociometrically selected work teams Increase production," Personnel Psychol.. 5: 175-186, 1952.

131. Van Zelst, R. H. "Validation of a sociometric regroup procedure," J. Abnormal and Social Psychol., 47: 299-301, 1952.

132. Wagenhcrst, L. H. "The relation between ratings of student teachers In college and success in first year of teaching," Educ. Admin, and Supervision, 16: 249-253, 1930.

133. Waites, J. A. "The assessment of techniques for esti­ mating the behavioral adjustment of chronic neuropsychiatric patients," Amer. Psychologist, 8: 271, 1953. 134.Wandt, Z., and Ostreicher, L. M. "Validity of samples of classroom behavior for the measurement of social- emotional climate," Psychological Monographs, Vol. 68, 1954.

135. Whitten, E. 3. "Programs for rehabilitation of the handicapped," Employment Security Review, 22: 19“ 21, 195b. 202

136. Wilson, J. W. "Correlation of clinical estimates with test scores on mental ability and personality tests," J. Clinical Psychol.. 10: 97-99, 1954.

137. Wittenbom, J. R. "A new procedure for evaluating mental hospital patients," J. Consulting Psychol., 14: 500-501, 1950.

138. Wittman, Phyllis, "A scale for measuring prognosis in schizophrenic patients," Elgin Papers, 4: 20-33, 1941.

139. Wittman, Phyllis and Sternberg, L. "Followup of an objective evaluation of prognosis in Dementia Praecox and Manic-Depressive psychoses," Elgin P a p e r s , 5: 216-227, 1944.

140. Wright, F. H. "An Evaluation of the Candidate Employee Program in the Rehabilitation of Psychiatric Patients." Unpublished doctoral dissertation, The Ohio State University, 1956.

141. Zeleny, L. D. "Selection of compatible flying part­ ners," Amer. J. Sociology. 52: 424-431, 1947. 203

AUTOBIOGRAPHY

I, Thomas Mun Chew Chang, was born in Honolulu,

Hawaii, on March 2, 1923. I received my primary and

secondary school education in Honolulu, am a graduate of

the Apprentice School at the Pearl Harbor Naval Shipyard,

and worked there during the entire period of World War II,

first as an apprentice machinist, then as a journeyman.

In 19^+3 I began undergraduate studies at the Uni­

versity of Hawaii, earning a Bachelor of Arts degree in

19^7* From 19^7 to 19^9 I taught in the high schools in the

Hawaiian Islands, then entered Teachers College, Columbia

University, and completed work for the Master of Arts

degree in 1950. From 1951 to 195^# I was Director of Test­

ing and Guidance in a private school in Honolulu, at which

time I left to begin doctoral studies at The Ohio State

University,

While in residence at this university, I have held

the positions of Research Assistant in Psychology for three

quarters and Assistant In Psychology for four quarters. I

also completed an intemeshlp in Counseling Psychology at

the Veterans Administration neuropsychiatric hospital In

Chillicothe, Ohio. Xn 1956 I was a recipient of the Honolulu

Foundation Scholarship, and during the 1957 school year, I was an Opportunity Fellow of the John Hay Whitney Foundation.