<<

THE EFFECTS OF INSTRUCTIONS AND REINFORCEMENT VERSUS A MULTIFACETED SELF CONTROL PROCEDURE ON THE MODIFICATION AND GENERALIZATION OF BEHAVIOR IN SCHIZOPHRENICS

Robert C. Gresen

A Dissertation Submitted to the Graduate School of Bowling Green State University in partial fulfillment of the requirements far the degree of DOCTOR OF PHILOSOPHY

March 1975 ii

ABSTRACT

The present investigation examined the treatment efficacy of a multifaceted self control procedure relative to behavior modification and yoked control procedures with chronic, nonparanoid schizophrenics. Further, the potential differential effectiveness of the same proce­ dures, depending upon the experimenter (E) who administered them, was examined. Forty-two subjects were pretested on five dependent measures. Subsequently, each subject was randomly assigned to one of two training Es and to one of the three experimental procedures. Regardless of B and treatment assignment, each subject received four hours of training on two of the five dependent variables- digit symbol and proverb inter­ pretation. After training, the subjeots were posttested using parallel forms of the five dependent measures. Four weeks later, a follow-up assessment was conducted, using a third set of parallel forms. Analysis of the pretest measures and five individual difference variables suggested that the groups were comparable prior to training. Individual analyses of the posttest and follow-up data, however, failed to reveal significant effects due to the E or methods or £ by method interaction. Separate multivariate trend analyses were conducted on each of the five dependent variables. Except for the digit symbol analysis, which may have reflected a rather transitory difference in treatment effectiveness, significant differential trends were not found among the three training methods, between the two training Es, or on the E by method interaction. Significant trends over time for the 42 subjects combined were found on both training variables but only on one of the three variables employed to assess generalization across responses the Holtzman inkblot test (HIT). For digit symbol, the trend reflected an increasing performance from pretest through the follow-up session. For proverb Interpretation, subjects significantly improved on the post- test relative to pretest levels. In addition, posttest performance was maintained on the four week follow-up. On the HIT, subjects improved significantly from pre- to posttest{ follow-up performance was sig­ nificantly lower than either pre- or posttest levels. The remaining measures of generalization across responses did not reveal significant change over time relative to pretest levels. It was concluded that for chronic, nonparanoid schizophrenics, the training procedures employed and/or the E administering the procedures were not differentially effective. This conclusion was equally applica­ ble to generalization across situations and responses. The research further demonstrated the complexity of the area and the necessity far conducting a series of experimental studies to isolate important in­ dependent variables. iii

ACKNOWLEDGEMENT I wish to express ray appreciation to Donald Leventhal, Ph.D., Pietro Badia, Ph.D., Robert Warehirae, Ph.D., and Peter Facione, Ph.D. far their valuable contributions to the current research endeavor as well as their support and guidance throughout irçy career at Bowling Green State University. In addition, I wish to acknowledge the Psychology Department and the Research Services Office at Bowling Green State University, and the Research Service at the Veterans Administration Center, Wood, Wisconsin, for the funding made available to the present investigation. I also want to thank Mr. Paul DeLisle and Stan Simon, Ph.D., whose active assistance was indispensible to the implementation and completion of this study. iv

TABLE OF CONTENTS Page INTRODUCTION...... 1

METHOD...... 19 Subjects...... 19 Apparatus and procedure ...... 19 Digit symbol test...... 20 Auditory distraction test...... 21 Proverbs test...... 22 Word Association test...... 22 Holtzman inkblot test...... 23 SI training group ...... 25 Operant training group ...... 29 Practice control group...... 31

RESULTS...... 33 Interrater reliability ...... 33 Pre-, post-, and follow-up test analyses ...... 3^ Trend analyses...... 36 Digit symbol test...... 36 Proverbs test...... 38 Holtzman inkblot test ...... ^3 Auditory distraction test ...... 46 Word association test ...... 46

DISCUSSION...... 51

REFERENCES...... 60 V

APPENDIX I. DEPENDENT MEASURES AND TRAINING TASKS ...... 66

APPENDIX II. BECKER’S SCORING SYSTEMS ...... 87

APPENDIX HI. E GUIDE FOR PROVERB INTERPRETATION INSTRUCTIONS . . 106

APPENDIX IV. DATA...... 109

APPENDIX V.S UBJECT ASSIGNMENT SCHEDULES...... 122 vi

LIST OF TABLES Table Page 1. Multivariate trend analysis far digit symbol ...... 37 2. Tukey’s HSD post hoc test results for digit symbol. All pairwise comparisons of means within each treatment group ...... 40 3. Multivariate trend analysis for proverb interpretation . . . 42 4. Tukey’s HSD post hoc test results for proverb interpreta­ tion. All pairwise comparisons of means for all subjects combined...... 43 5. Multivariate trend analysis for the HIT...... 44 6. Tukey’s HSD post hoc test results for the HIT. All pairwise comparisons of means for all subjects combined ...... 46 7. Multivariate trend analysis for the auditory distraction task...... 47 8. Multivariate trend analysis for the word association task . 48 9. Tukey’s HSD post hoc test results far the ward association task. All pairwise comparisons of means for all subjects combined ...... 50 1- A. Univariate analyses of variance. Pretest data...... 110 2- A. Univariate analyses of variance. Posttest data ...... 112 3- A. Univariate analyses of variance. Follow-up data ..... 113 4- A. Correlation matrix for the 10 pretest variables ...... 114 5- A. Means and standard deviations for each group. Pretest data.115 vii

LEST OF FIGURES Figure Page 1. Pretest, posttest, and follow-up mean performance for the SI, operant, and control groups on digit symbol...... 39 2. Pretest, posttest, and follow-up mean performance for all subjects combined on digit symbol ...... 41 3. Pretest, posttest, and follow-up mean performance for all subjects combined on proverb interpretation ...... 41 4. Pretest, posttest, and follow-up mean performance for all subjects combined on the HIT...... 45 5. Pretest, posttest, and follow-up mean performance for all subjects combined on the word association task ...... 49 1- A. Mean performance for each group on digit symbol...... 117 2- A. Mean performance for each group on proverb interpretation . 118 3- A. Mean performance for each group on the HIT...... 119 4- A, Mean performance for each group on the auditory distraction task...... 120 5- A, Mean performance for each group on the word association task 121 I

Behavior modification, defined by Liberman (1972) as the applica­ tion of operant conditioning principles to abnormal behavior, is currently among the most widely used treatments with schizophrenics. This approach has provided the clinician with a variety of intervention techniques and, even more importantly, has attempted to substantiate the effectiveness of those techniques with systematic, empirically de­ rived data. As the following overview of the literature will indicate, experimental investigations support the notion that many effective operant techniques are available and appear to be applicable to a wide range of behaviors in schizophrenics. On the other hand, however, for­ midable problems exist in obtaining generalization of operantly develop­ ed behavioral change. The issue of generalization is an extremely im­ portant one, especially from a therapeutic viewpoint, as treatment goals are seldom confined to a single response or to the client’s behavior only during treatment sessions. One set of procedures has recently been shown to facilitate generalization of behavior change in schizo­ phrenics (Meicheribaum & Cameron, 1973). The present investigation was a further examination of the efficacy of the Meichenbaum and Cameron technique. That operant procedures can be utilized effectively with schizo­ phrenic individuals has been most convincingly demonstrated by research­ ers using a token econony system. The token econony was first introduced in a paper by Ayl ion and Azrin (1965) and more fully described in a sub­ sequent book by the same authors (Ayllon & Azrin, 1968). This system, which is currently employed in a large number of psychiatric hospitals, is typically implemented on a ward-wide basis and administered by a 2

staff of psychiatric nurses and aides. Although many variations in specific details are possible, three major components can be found in any well designed token program. 1) Behaviors to be shaped or strengthened are objectively and concretely specified to both the ward personnel and the patients. 2) When a specified desirable behavior occurs, the patient is reinforced by receiving a predetermined number of tokens or some other comparable unit of exchange. 3) The tokens are exchangeable for items and/or activities which are desired by the patients. Behaviors that have been most often targeted in such a system in­ clude personal hygiene and self-grooming responses, social interaction, eating and other self-maintenance behaviors, and various on- and off­ ward work behaviors. Thus, the emphasis of most token programs has been upon building and maintaining adaptive and constructive behaviors. On the other hand, behaviors usually described as symptomatic have sel­ dom been dealt with directly. In fact, most token programs ignore symptomatic behavior completely. The rationale for this procedure is twofold. As the probability of appropriate behavior increases due to the operating reinforcement contingencies, the probability of inappro­ priate behavior, e.g., symptoms, should decrease. Secondly, symptoms maintained by their environmental consequences should extinguish if completely ignored. Occasionally, however, a particular behavior be­ comes intolerable and cannot be ignored. In such cases, deducting tokens from a patient’s earnings and making time-out periods contin­ gent upon unacceptable behavior have been found to be effective proce­ dures (Steffy, Hart, Crow, Torney, & Marlett, 1969). 3

The effectiveness of token economies in developing appropriate be­ haviors in schizophrenic inpatients has been well documented by the original authors as well as by a multitude of comparable programs that have recently emerged (Allen & Magaro, 1971} Atthowe & Krasner, 1968; Cohen, Florin, Grusche, Meyer-Osterkamp, & Sell, 1972} Ellsworth, 1969} Gorham, Green, Caldwell, & Bartlett, 1970} Lloyd & Abel, 1970} Schaefer & Martin, 1966). In addition, one investigator (Hendersen, 1971) has reported the successful application of this approach with adult psychotics in a community-based environment. The success of positive reinforcement in modifying the behavior of schizophrenics has also been demonstrated in studies not involved with the token econoiry approach. In many of these studies, the behavior focused upon has been mare directly related to schizophrenic sympto­ matology. For example, a number of investigations have reported that schizophrenics can be conditioned to increase the frequency of common word associations emitted on a word association task (e.g., Sommer, Witney, & Osmaond, 1962} Ullmann, Krasner, & Edinger, 1964). Similar studies have further demonstrated that the performance of schizo­ phrenics can be dramatically enhanced using contingent positive rein­ forcement on proverb interpretation (Meichenbaum, 1966} Meichenbaum, 1969)» the amount of “healthy talk” emitted during a structured inter­ view (Meichenbaum, 1969} Ullmann, Forsman, Kenny, McInnis, Unikel, & Zeisset, 1965). attending and abstracting responses (Wagner, 1968), psychomotor tasks (D’Alessio & Spence, 1963). and the frequency of affective self-references (Salzinger & Pisoni, 1958). Both primary and secondary positive reinforcers have been found to be effective with 4 with this clinical population. In addition, techniques other than positive reinforcement have been successfully employed to modify the behavior of schizophrenic individ­ uals. Wolff (1971) reported a substantial reduction in the rate of de­ lusional expression using a satiation procedure with a chronic, para­ noid, female schizophrenic. Tactile hallucinations have been suppress­ ed by making time-out from recreational activities contingent upon their occurrence (Richardson, Karlsalas, & Lal, 1972). When the onset of auditory hallucinations was followed by self-administered electric shock, the frequency of the hallucinatory behavior dramatically de­ creased (Bucher & Fabricatore, 1970). By combining modeling, shaping, and positive reinforcement procedures, a number of investigators have successfully reinstated verbal behavior in mute or near-mute schizo­ phrenic inpatients (Baker, 19715 Isaacs, Thomas, & Goldiamond, 1960} Wilson & Walters, 1966). The underlying rationale common to most behavior modification pro­ grams is that behavior can be maintained or modified by manipulating its environmental consequences. The literature reviewed thus far strongly supports this basic assumption regarding the behavior of schizophrenics. Thus, the success or failure of a particular behavior modification effort depends, in part, upon the degree of control that the experimenter can exert upon the environment of the subject. The experimenters in the above studies were able to obtain a great deal of environmental control, as the studies were conducted primarily on psychiatric hospital wards and in experimental laboratories. The ability to generate behavioral change in a well controlled 5

environment is an extremely important characteristic of a useful therapeutic technique. However,, the behavior ‘therapist is concerned', with modifying the client’s behavior in extra-treatment situations as well. As Meichenbaum and Cameron (1973) have indicated, however, it is seldom feasible far the therapist to engineer the extra-treatment en­ vironment of a client. Thus, techniques need to be developed which can be used in one situation (i.e., therapy sessions), but which will be effective in changing the schizophrenic’s behavior in other situations as well. The problem being raised here is the issue of generalization. Al­ though it is an extremely important problem, it has received little attention in the behavior modification literature (Liberman, 1972J Yates, 1970). In a recent review of the token econony literature, Kaz- din and Bootzin (1972) described the current status of the problems

”... experimental studies have focused on the functional relation­ ship of the reinforcement procedures to the dependent variables, rather than on the maintenance of the altered behavior. Generali­ zation of treatment effects to extra-treatment settings...is usually not assessed directly... Although token economies have been dramatically effective at changing behavior within the psychiatric hospital, there is little that improvement is maintained outside of the institution... Generalization should be planned rather than depended upon as an inadvertant consequence of the token program.”

Although the above quotation refers exclusively to the token approach, the statement is equally applicable to the entire operant literature using schizophrenic subjects. In response to the suggestion of Kazdin and Bootzin, Liberman (1972) noted that a number of conditioning techniques are available which would appear to be useful in planning generalization of operantly de­ 6 veloped behavior change. To date, however, few studies have attempted to utilize these techniques in order to promote the generalization of treatment effects. Further, results from the few available studies suggest that employing such procedures does not guarantee that adequate generalization will, in fact, be obtained. Operant principles suggest, for example, that reinforcing a given behavior in a variety of environments will increase the probability of that behavior's occurrence in other situations as well. To test this hypothesis, Lyon (1972) employed three chronic, male schizophrenics re­ siding in an ongoing token economy ward. Each subject was seen four times weekly by Experimenter 1 at mealtime in Experimental Setting 1. During each session, the experimenter asked a series of questions and recorded the length and appropriateness of each response. After ob­ taining a reliable baseline for each subject on both dependent measures, a reinforcement procedure was initiated. Now, the patients' meal was withheld and small portions of food were delivered contingent upon giv­ ing appropriate answers to the experimenterIs individual questions. During the third phase of the study, the contingencies were reversed- food was made contingent upon giving inappropriate or no answers to the questions. After reestablishing the original contingency, food was made contingent upon both appropriateness and increasingly longer responses. The results of this portion of the study were consistent with other operant research using schizophrenic subjects. Except far one subject, whose frequency of appropriate responses did not decrease during the reversal phase, the three subjects responded as predicted during all experimental conditions. In some instances, however, prompts 7

were needed in conjunction with reinforcement in order to develop the desired response. In the context of the present discussion, the most interesting com­ ponent of the study was its deliberate effort to facilitate generaliza­ tion. An initial test for generalization was conducted at the end of the first phase of the study. This was done by having three experimen­ ters asking each subject five standard questions in each of three hospi­ tal settings. Length of response was the only measure taken. Two of the three subjects did exhibit a minimal increase in response length on the generalization test. However, even this small increment disappeared within a few test trials. To increase generalization, additional proce­ dures were instituted with one of the original subjects. Experimenter 2 began prompting answers to questions and reinforced increasingly longer answers in Experimental Setting 2. Experimenter 1 continued the same procedures he had used during the first part of the study. Thus, the subject was given additional training trials by two experimenters, each in a different stimulus situation. A second generalization test, con­ ducted after this extended training period, revealed substantially greater generalization than was obtained on the initial generalization test. Further, the effects did not dissipate over trials. The largest improvement on the second test was obtained on answers given to Experi­ menter 2 across the three hospital settings. Little generalization to Experimenter 3. however, was obtained. In a final effort to increase generalization, Experimenter 3» in the third hospital setting, began prompting and shaping longer responses while the other experimenters continued their activities without change. As expected, the final 8

generalization test reflected more generalization than either of the first two tests. The largest increase over the second test was ob­ served on answers to Experimenter 3 in all three hospital settings. Although generalization was obtained, the results of the study were somewhat disappointing due to the large number of trials that were need« ed to produce generalization. Further, the effects were confined to generalization across situations with experimenters who had previously reinforced the individual. Generalization from one experimenter to another, however, was negligible in all cases. Thus, the study suggest« ed that generalization across settings may be easier to develop in schizophrenics than generalization across people. Two other reports were located which employed similar procedures to promote generalization in schizophrenic subjects. The first study (Patterson & Teigen, 1973) reported that only slight evidence for generalization was obtained. Kale, Kaye, Whelan, and Hopkins (1968), using an extensive treatment procedure, reported somewhat stronger evi­ dence for generalization, though the results remained discouraging con­ sidering the elaborate procedures employed. Of particular importance regarding the issue of generalization is a study done by Meichenbaum (1969). This study investigated the effects of contingent reinforcement in conjunction with Instructions explaining both the reinforcement contingencies and the requirements of the ex­ perimental tasks. The dependent measures used were: level of abstrac­ tion on a proverbs test, percentage of "sick talk" emitted during a structured interview, and three measures to test for generalization. The three generalization measures were percentage of "sick talk" emitted 9 during an interview with a second interviewer, number of common word associations, and the similarities subtest from the Wechsler Bellevue Adult Intelligence Scale. Thus, the Meichenbaum (1969) study examined generalization to a second experimenter as well as to other response measures. One group of schizophrenic subjects was trained (given in­ structions plus contingent positive and negative social reinforcers) to emit “healthy talk” during an interview} a second group was trained to give abstract proverb interpretations} and a third group received training on both tasks. Three additional groups, comparable to the first three groups, were given token reinforcement rather than social approval and disapproval. Two control groups were also employed, con­ trolling far extraneous variables such as intercurrent life experiences, personal contact, and repeated testings. The experimental procedures not only resulted in improved perform­ ance on tasks explicitly trained but, in addition, evidence of generali­ zation was obtained as well. Generalization was reflected by improved performance on proverb interpretation by subjects trained only to emit “healthy talk”, by improved posttest scores on the “healthy talk" measure of subjects trained exclusively on proverb interpretation, and by improved functioning in all groups on the word association and simi­ larities tests. In addition, evidence for generalization to a second interviewer was obtained from the token reinforcement group trained to emit "healthy talk". This group emitted a significantly higher percent­ age of "healthy talk" on the posttest interview relative to pretest levels with an interviewer not involved in the training sessions. Thus, the effects of the procedure had apparently generalized to responses not 10 trained directly as well as to an interviewer who was not involved in the training. Since procedures to facilitate generalization were not deliberately included in the Meichenbaum (1969) study, the possible reason for its occurrence was of special interest. It was observed that several in­ dividuals trained to emit ’’healthy talk” repeated aloud the experimental instructions ’’give healthy talk, be coherent and relevant” when being tested for abstract level of proverb interpretation. Meichenbaum hy­ pothesized that the experimenter’s instructions had become discrimina­ tive stimuli for these subjects, signalling the subjects to utilize a general set for problem solving before responding to the experimental task. Further, the use of self-instructions may have aided the subjects in attending to the relevant demands and requirements of the task at hand. In any case, Meichenbaum suggested that training schizophrenics directly to use self-instructional (SI) statements could be a useful intervention technique for the modification of schizophrenics’ be­ havior. Moreover, the SI procedure appeared to be a most promising technique in achieving generalization of obtained behavioral change. The potential application of SI training to schizophrenics was pur­ sued in a later set of two studies (Meichenbaum & Cameron, 1973). Study I of this report was essentially a pilot study to determine the feasibility of training schizophrenics to use self-instructions effec­ tively. Having established that such a procedure could be implemented, a second study was undertaken to investigate the efficacy of the approach using a more adequate methodology. Two groups of five male hospitalized schizophrenics were used in li the second study. The two groups were matched on clinical diagnosis, length of current hospitalization, and pretest performance. Two, one- half hour pretesting sessions were conducted by Experimenter 1 with each subject. During the first session, the subject was given a 10 minute structured interview followed by a proverbs test and a digit recall task with and without auditory distraction. During the second pretest session, another 10 minute interview was given along with 12 cards from the Holtzman Inkblot Test. At this point, each subject was introduced to a second experimenter who administered either the treat­ ment or control procedures, depending upon the group assignment of the subject. Each subject in the SI training group received three, 45 minute sessions in order to develop a general set of "talking to himself". In general, this involved training the subjects to pause and think be­ fore responding to a given task requirement. To familiarize the sub­ jects with the procedure, each subject was trained to use self-instruc­ tion while performing a number of sensorimotor tasks (e.g., Porteus Maze, Trail Making, etc.). During the second phase of training, the subjects’ use of self-instructions was extended to more cognitively de­ manding tasks. Also, this part of the training instructed the sub­ jects to monitor and evaluate their behavior in order to recognize when to use self-instructions. Phase III (two and one-half sessions) was concerned with having each subject attend to interpersonal cues emitted by others when the subject’s behavior was bizarre, incoherent, or irrelevant. Subjects were trained to use the reactions of others as cues to begin evaluating their behavior and to use the SI procedure. 12

The SI training procedure is a multifaceted approach, using tech­ niques such as instructions, modeling, behavioral rehearsal, shaping, and discussion. Since the SI procedure was used in the present investi gation, descriptions of specific procedures will be deferred to the Method section of this paper. Subjects in the SI group received direct training on a large number of tasks including two of the four dependent measures- proverbs and in­ terview behavior. Thus, the two remaining dependent variables (ink­ blots; digit recall) served as measures of generalization. Subjects in a yoked practice control group received the same amount of time with the experimenters, the same number of testing sessions, and the same number of trials on each task as subjects in the SI group. During their training sessions, examples of adequate and inadequate re­ sponses for each training task were given and the subjects were provid­ ed opportunities to generate their own appropriate answers to the tasks as well. Further, the SI and control subjects received equal amounts of social reinforcement for task performance. Thus, procedures given to the controls were comparable to those given to the SI subjects, ex­ cept that explicit modeling and practice in self-instruction were not included. Statistical analyses indicated that both groups improved signifi­ cantly over trials on all dependent measures. In addition, SI sub­ jects improved significantly more than the control group on all meas­ ures except the digit recall task during the no distraction condition. These results were maintained at a three week follow-up session. Thus, the SI procedure was consistently mare effective than control proce- 13 dures in improving the performance of schizophrenics on tasks directly trained as well as on tasks not used for training. Further, evidence for generalization across people was found, as the testing was conducted by an experimenter not involved in the training sessions. It appears that the SI procedure may be a promising new technique for the modification and generalization of the behavior of schizophren­ ic subjects. It must be acknowledged, however, that the technique has been investigated in only one research effort using schizophrenic sub­ jects. Therefore, much research has yet to be done before a statement concerning the efficacy and range of application of the procedure can be made with confidence. Due to the multitude of variables that need to be considered, the rigorous investigation of any behavior modification technique is a time consuming and arduous task. Paul (1969) has discussed the complexity of such a task and has outlined the major domains of variables that re­ quire consideration. He concluded that the ultimate question to be answered in the field of behavior modification is "what treatment, by whom, is most effective for this individual with that specific problem, under what set of circumstances, and how does it come about?”. Obvious­ ly, no individual study could possibly answer such a detailed, . However, as Paul (1969) has indicated, the meaningful inte­ gration of knowledge across studies would be greatly facilitated if in­ dividual researchers explicitly delineated the aspect of the question being investigated. This would include the description, measurement, or control of each major variable domain of importance within this re­ search area. Since the goal of the present study is to contribute in­ 14 formation concerning a particular behavior modification technique (i.e., the SI procedure), each variable domain will be considered at this time. The client variable alone poses a formidable problem far researchers interested in studying schizophrenics. This diagnostic label defines a heterogeneous group of individuals in terms of both task performance and response to experimental manipulation. Due to this high degree of intersubject variability, researchers have advocated the division of schizophrenics on various dimensions in order to form more homogeneous samples (Shakow, 1963s Silverman, 1964; Cash, 1973). The dimensions that have been most commonly employed include the following: chronic- acute, good premorbid- poor premorbid, process- reactive, and paranoid- nonparanoid. Although a significant amount of overlap exists among the dimensions, the use of these dimensions has facilitated the integration of the schizophrenic literature (Silverman, 1967; Neale & Cromwell, 1970). Hence, it appears to be highly unlikely that a given intervention procedure, including the SI technique, would be equally effective across all schizophrenic individuals (i.e., due to intersubject variability). In other words, the efficacy of a given behavior modification tech­ nique must be separately assessed far each subgroup of schizophrenics. In the Meichenbaum and Cameron (1973) study, half of each group was composed of acute, reactive, paranoid schizophrenics while the other half consisted of chronic, process, nonparanoid schizophrenics. Due to the heterogeneity of this sample, the study did not suggest which schizophrenic subgroups might profit most or, for that matter, least from the SI training. The present investigation examined the effective­ 15 ness of the SI procedure with a particular subgroup of schizophrenic individuals- i.e., chronic, nonparanoid, male schizophrenics. Paul (1969) listed two additional variable classes of importance under the client domain. The first subsumes variables such as sex, in­ telligence, socio-economic background, etc., including the full range of individual differences not directly related to clinical diagnosis. The second class involves the physical- social life environment of the sub­ jects. Variables listed here include any additional circumstances which could affect therapeutic outcome. For researchers of schizophren­ ia intervention programs, the variable of most concern is psychotropic drug dosage (see Spohn, 1973). Future research with the SI procedure could attend to these variables, as none was investigated or controlled in the Meichenbaum and Cameron (1973) report. Since the present study assigned schizophrenics to groups on a random basis, the variable classes listed in this paragraph were not investigated directly. How­ ever, statistical analyses relating to these variables were performed. The second major variable domain concerns therapist characteristics. The first class of variables under this heading consists of the par­ ticular behavior modification technique being employed. In the present study, a group receiving the SI procedure was contrasted with a group given other operant procedures which, as the above review indicated, have been successfully applied to schizophrenics’ behavior. Analogous to the client domain, the effects of individual differences among thera­ pists and the physical- social life environment of the therapist must also be evaluated. This relationship between therapist characteristics and schizophrenic response to behavior modification programs remains to 16 be investigated. In fact, a search of the literature dealing with be­ havior modification in schizophrenics did not locate a single study- concerned with the therapist as an independent variable. One method of investigating therapist effects involves the use of multiple thera­ pists, each administering the same procedures to comparable groups of schizophrenics, and assessing the differential effectiveness of the individual therapists. This procedure was implemented in the present investigation. Time was listed as the third major domain in behavior modification research. In relation to the SI procedure, for example, the amount of time spent in training the procedure to schizophrenics could be an ex­ tremely important variable. Also, administering follow-up assessments at various time intervals would be useful in determining the limits of the procedure regarding the maintenance of acquired behavioral change. The time frame employed in the present study was quite similar to that of the Meicheribaum and Cameron (1973) study. Thus, the differential effectiveness of an orthodox operant procedure versus SI training was assessed on both immediate posttest data as well as on data obtained at a four week follow-up. The final domain of variables discussed by Paul (1969) was criteria of treatment effectiveness. One interesting question to be asked of ary behavior modification technique concerns the range of behaviors to which it can effectively be applied. Thus, the measures used in the present investigation will not overlap completely with those used in the Meicheribaum and Cameron (1973) study. However, since the SI pro­ cedure has only been studied by one research team, using a number of 17 the same or similar measures seemed appropriate. Therefore, the pre­ sent study was a partial replication of the Meichenbaum and Cameron (1973) study as well as an effort to extend, to a small degree, the application of the SI procedure. A second aspect of the criteria question concerns the issue of gen­ eralization. In the Meichenbaum and Cameron (1973) study, an unspeci­ fied number of tasks were employed when training the SI procedure. Further, the intercorrelations among the training tasks and dependent measures were not provided. Thus, the extent of generalization to other responses cannot be assessed by the available information. In other words, the higher the intercorrelations among the training tasks and the dependent variables used, the smaller the amount of generalization that was demonstrated in the Meichenbaum and Cameron (1973) report. The present study investigated the interrelationships of the tasks em­ ployed in order to minimize this problem of interpretation. A second important aspect of the generalization problem is best in­ troduced by quoting Paul (1969). "The minimum requirement for any ex­ periment on behavior modification, therefore, is that the dependent variables include changes in the clients’ distressing behaviors from pretreatment to posttreatment, assessed external to treatment proper.” The concept being stressed in the above quotation is generalization to extra- treatment situations. In the present investigation, this aspect of generalization was evaluated by having all dependent measures obtained by an experimenter who was not involved in the administration of the treatment procedures. Further, all testing was done at a loca­ tion far removed from the-area where the treatment proper was conducted. 18

The present experiment consisted of an SI training group, an op­ erant training group, and a practice control group. Further, half of the subjects in each group were trained by one experimenter while the other half was seen by a second experimenter. Pre-, post-, and follow­ up testings were conducted by a third experimenter who did not have knowledge concerning the group assignment of the subjects. Five de­ pendent measures were administered during each testing session. Two of the five measures were used as training tasks during the experimental sessions. These two variables were the most direct measures of treat­ ment effectiveness, whereas the remaining three variables served as es­ timates of generalization. Concerning the former measures of treatment efficacy, it was predict­ ed that both the SI and operant groups would perform significantly high­ er on posttest and follow-up measures relative to their respective pre­ test levels. Further, it was hypothesized that posttest and follow-up measures would be significantly higher for the two experimental groups than for the practice control subjects. Since the dependent measures were obtained by an experimenter not involved in the experimental sessions, it was predicted that the SI group would perform significant­ ly higher on all measures than either of the other two groups (i.e., generalization across situations). Also, it was hypothesized that the SI group would exhibit greater generalization across responses than the other two groups, reflected by significantly higher performances on the three dependent measures not employed during training. Great care was taken in training the experimenters to administer each experimental procedure. Assuming, then, that both experimenters administered virtual­ 19 ly identical procedures, differences in effectiveness between the two experimenters was not predicted. METHOD Subjects Forty- two chronic, nonparanoid, male schizophrenics were employed in the present study. Each subject was randomly assigned to one of the six experimental conditions, resulting in six groups of seven schizo­ phrenics each. Identification of schizophrenics was accomplished by reviewing the hospital records of the Veterans Administration Center at Wood, Wisconsin. Determination of nonparanoid status was made from psychiatric diagnosis and from the absence of major paranoid symptoma­ tology in each subject’s hospital chart. In order to meet the defini­ tion of chronicity, each subject’s first hospitalization for schizo­ phrenia had to have occurred a minimum of three years before the study began. Patients with known or suspected organic impairment were not employed. Further, all subjects were 55 years old or younger. Statis­ tical analyses were performed to assess the comparability of the groups on each of the following dimensions: age, socio- economic level, in­ telligence, year of first psychiatric hospitalization, and current medi­ cation level. Apparatus and Procedure Each subject was seen by Experimenter 1 (E|) for a pretesting and information gathering session. At this time, the subjects were given Form I of the five dependent measures- a proverbs test (Gorham, 1956a), Brown’s (1969) digit symbol test, an auditory distraction test (McGhie, Chapman, & Lawson, 1965), a word association test, and a subset of the 20

Holtzman Inkblot Test (Holtzman, Thorpe, Swartz, & Herron, 1961). In addition, other information concerning each subject was obtained at that time. An estimate of intellectual level was obtained from the Shipley Institute of Living Scale (Shipley, 1939)» which has been found to correlate between .73 and .90 with the Full- scale score on the Weehsler Adult Intelligence Scale (Bartz & Loy, 1970). Socio-economic level was determined by using the two- factor index of social position (ifyers & Bean, 1968). Age, medication level, and year of first psychia­ tric hospitalization of each subject was obtained from hospital chart data. Digit Symbol Test The digit symbol test employed in the present study was developed by Brown (1969). This task, which consists of eight parallel farms, is quite similar to the digit symbol subtest of the Weehsler Adult In­ telligence Scale. Each test form consists of three component parts. The first line of the form presents a series of seven digits, each digit being assigned a particular geometric shape. The second part of the test form provides a series of boxes, each containing one of the seven digits and a space far the placement of the appropriate geometric figure. Each subject was given one line of practice in order to clarify the nature of the task and to provide the subject with practice in us­ ing the code. Finally, a series of 120 boxes, identical to the practice boxes, was presented to each subject. Subjects were instructed to supply as many correct symbols as possible in the time alloted (90 seconds). The test was scared by summing the total number of correct symbols supplied during the 90 second period. Copies of the parallel forms used 21 in the current study can be found in Appendix I. Auditory Distraction Test This measure, which was used in Study II by Meichenbaum and Cameron (1973). has been employed by researchers concerned with attentional de­ ficits in schizophrenics (McGhie, Chapman, & Lawson, 1965). The task consists of 16 sequences of either six letters or digits, presented at one second intervals, which are recorded by a female voice. The sub­ ject’s task was to reproduce each sequence immediately after the tape recorded presentation. Eight of the 16 sequences also contain a male voice, serving as a distractar, reciting an irrelevant number or letter during the silent one second intervals of the female’s presentation. The subjects were instructed to ignore the male voice and only to report the sequence recited by the female voice. Three forms of this measure were generated by the present author by random selection of digits and numbers. The generated sequences and their order of presentation are provided in Appendix I. Each subject was given four practice trials before completing the task- two with the male distractor and two without distraction. During the test proper, ten second rest intervals were given between each of the 16 sequences. To enhance the potency of the distractor stimulus, the 16 sequences were presented in a random order. In this manner, the subjects could not know when the distractor sequences would be presented. Following the McGhie et. al. (1965) procedure, each sequence was scored on a six point scale, one point being subtracted for each error (each omission, addition, or order deviation was considered to be an error). Each subject’s performance on the measure was determined by the 22

sum of his scores on the 16 sequences. Proverbs Test Gorham’s proverbs test (1956aj 1956b) was the third dependent measure used. Each of its three parallel forms contains 12 proverbs on a single page with instructions for the subject and, of course, space to write an interpretation for each proverb. The subject’s task was to in­ terpret ”what the proverb means”. Appendix I contains a copy of each form of the test. Proverb interpretations were scored for level of abstraction using Becker’s (1956) six point system. This scoring system rates the ab­ straction level of a given interpretation from highly abstract and appropriate (6) to absurd (1). Inter- rater reliability of this rating scale has been found to be .98 by Becker (1956) and .90 by Meicheribaum and Cameron (1973). Thus, past research suggested that the system can be reliably employed. However, inter- rater reliability was assessed in the present investigation as well (see Results). Specific details con­ cerning the Becker system can be located in Appendix II. Word Association Test In the present study, number of common word associations was measured by administering a 50 item word association test. Three parallel farms of 50 items each were developed for this investigation using the Palermo and Jenkins (1964) word association norms. The three ward lists were matched on the frequency of common associations (i.e., associates given by at least 10$ of the Palermo & Jenkins 12th grade normative group). Correlations among the three forms substantiated that the forms were parallel in terms of the matching criterion (r^3 .931» r13= .938} 23 r23= *967). Appendix I contains the instructions used in administering the test as well as a copy of each form. Instructions were adapted from those used by Palermo and Jenkins (1964). The test was scared by counting the number of common associations (see definition above) given. Thus, a subject’s potential score on this measure ranged from 0 to 50. The definition used is consistent with other measures of commonality found in the literature (Jenkins & Palermo, 1964). Holtzman Inkblot Test (HIT) The final dependent measure employed was derived from the HIT (Holtzman, Thorpe, Swartz, & Herron, 1961). Following Meichenbaum and I Cameron’s procedure (1973) » three parallel forms, each consisting of 10 inkblots, were generated from Form A of the test. This was accom­ plished by matching the forms in terms of a weighted frequency of path­ ognomic verbalizations. The normative data provided by the test develop ers (Holtzman et. al., 1961) were used for this procedure. In the normative data, pathognomic verbalizations were scored on a 0 to 4 scale, and frequencies for each level were reported far every card. The weighted frequency for each card was obtained by multiplying the fre­ quency of pathognomic verbalizations times the level of the deviant verbalization (0- 4). Correlations among the three farms in terms of the matching criterion yielded the following: »972; r13= .946; 1*23- «983. Thus, the three forms were found to be virtually identical regarding the level and frequency of pathognomic verbalizations obtain­ ed from the normative sample. Appendix I lists the cards used for each form of the test. 24

The subjects were instructed to provide only one response per card. Each response was scored on a one to six basis, again using a scoring system developed by Becker (1956). Specific details of this scoring procedure are deferred to Appendix U. In general, the scale attempts to define differential levels of perceptual functioning, ranging from low level perceptual responses (1) to highly developed and integrated responses.(6). Although excellent inter- rater reliability data have been obtained with this system (Becker, 19565 Meichenbaum & Cameron, 1973), the present study examined inter- rater reliability as well (see Results). After completing the pretesting session, each subject received four, one hour training sessions regardless of his group assignment. Subjects were assigned on a random basis to a) either Eg or E^ and b) one of the three experimental training procedures. The first experimental session occurred within one week of the pretesting session. The four sessions were administered over a period of ten days, although the exact distri­ bution of training sessions varied from one subject to another. E*, who administered all testing sessions, was not involved in any of the experimental sessions. In addition, he attempted to remain unaware of the group assignment of the subjects. Regardless of either the experimental procedures administered or the experimenter involved, each subject was given four hours of experi­ ence with two of the five dependent measures. Two hours were spent us­ ing parallel forms of the digit symbol test which were not employed during any of the testing sessions. The other two hours were spent us­ ing a set of proverbs not included in Gorham’s test. These proverbs 25

were taken from tests developed by Benjamin (1944) and Kaufman (i960). This list of training proverbs can be found in Appendix I. Within one week of the final experimental session, each subject was given Form II of the five dependent measures. Four weeks later, a follow- up testing session was administered using Form III of each de­ pendent variable. Thus, the posttest and follow- up data yielded meas­ ures on tasks that were trained directly (digit symbol} proverbs) and three measures on tasks that were not trained directly (auditory distrac tion} word assoication} HIT). Moreover, all testing was done ly an experimenter not involved with the training procedures in a separate part of the hospital. Thus y. differential generalization across per sons and situations among the six groups was also examined. Procedures carried out in each group are described in the following sections. SI Training Group The specific procedures used with this group corresponded to the Meichenbaum and Cameron (1973) SI procedures as closely as possible. Initially, the E modelled the SI procedure using one of the training tasks, talking aloud, while the subject observed. Secondly, the sub­ ject performed the same task while the E verbalized the SI instructions aloud. The next step required the subject to perform the task while self- instructing himself out loud. At this point, the E monitored and, when necessary, helped the subject to generate appropriate SI state­ ments. The fourth step of the procedure was identical with the preced­ ing phase, except that the subject’s self instructions were whispered by the subject. Finally, the subject performed the task while covertly 26 instructing himself- i.e., the subject’s self instructions were not audible to the E during this phase of the training. The specific SI verbalizations employed can also be categorized} 1) questions concerning the nature and demands of the task} 2) answers to these questions, thereby focusing the subject’s attention on the re­ quirements of the task? 3) SI and guidance statements, aiding the sub­ ject in maintaining his attention and minimizing irrelevant behavior} 4) coping statements designed to minimize the subject’s frustration and feelings of failure} and 5) self reinforcement to maintain his perform­ ance. The following dialog taken from Meichenbaum and Cameron (1973) is an example of what the E might say while modelling the SI procedure during an early part of the training period using the digit symbol task:

"What is it I have to do? I’m supposed to fill in these boxes with symbols. Now look up at the top code of symbols and numbers. Good. The first symbol I have to look for goes with number 94. It’s three lines. That’s it. Now quickly to the next one, number 24 has a circle with a dot in it. Just continue this way until I finish the line. I’m getting it. Let me see how many I can get. Remember, I must go quickly, but also carefully."

Four parallel forms of Brown’s (1969) digit symbol test were used during the treatment sessions. These four forms were not among those used in the testing sessions. For each form, the E modelled the SI pro­ cedure for the first line of 20 digits. The subject did the second line while the E verbalized SI statements aloud. The third line was done by the subject while he was verbalizing SI statements himself. The last lines were used to gradually fade out overt verbalizations com­ pletely. For all groups, including the SI group, the first and third train­ 27 ing sessions employed the digit symbol task. During the last half of the third session, the difficulty level was increased by playing an audio tape of random numbers and symbols while the subject was perform­ ing the task. With the introduction of the tape distraction, the E again modelled the use of SI statements. For example:

•'That tape recorder is trying to distract me. Just pay attention to what I have to do. Number 56 gets a circle. Good. Number 12, two lines. I can disregard the distraction. Number 15» a triangle. No, that’s wrong. That is okay, even if I make an error I can go on carefully and quickly. Good, I’m getting it. If I make up ray mind, distractions won’t bother me.’1 (from Meichenbaum & Cameron, 1973).

The two remaining training sessions were used to teach the subjects how to employ SI statements when generating abstract proverb interpre­ tations. The general strategy of the training was essentially the same as for the digit symbol training. The subjects were taught to develop the same problem- solving strategy via the SI technique, although the specific SI statements were tailored to the proverbs task. During these sessions, more emphasis was placed upon monitoring and evaluating res­ ponses (i.e., interpretations) by self questioning. Deficient responses, revealed by self questioning,-were to become signals for the use of self instructions to generate more adequate interpretations. Training subjects to generate abstract proverb interpretations re­ quired techniques that were not needed when using the digit symbol test. Far example, the definition of a proverb and the mechanics of interpret­ ing proverbs had to be fully described to the subjects. Using Benjamin’s (1944) conceptualization, a proverb is a statement ”built up of one or mare substantive symbols, usually objects of nature or everyday life 28

(cat and mice, milk, bridge, stone, chickens, ...), combined with predi­ cates which have varying degrees of figurative or literal value (play, cry, cross, ...).” Elaborating further on a proverb’s characteristics, Benjamin stated that:

’’the literal meaning of the proverb can be a simple statement of real or supposed fact, sometimes based on observations in nature, at other times containing a cause-and-effect relationship. Or it may consist of a piece of homely advice. In all cases, however, the objects as well as the actions are supposed to represent (sym­ bolize) something else, the objects usually either types of human beings or mare or less abstract categories (cat- person in authori­ ty, rolling stone- restless or changeable person, ...).”

After explaining the above characteristics, an explanation of the mechanics of interpretation was also provided. Interpretation involves: i) translating the symbols into the figures, categories, and actions which they represent (desymbolization), 2) expressing the meaning of the proverb (varying from concrete to highly abstract meanings), and 3) evaluating the adequacy of the interpretation. As Benjamin (1944) has indicated, however, correct desymbolization can only be aceonplished if the proverb’s meaning is understood at the same time. Thus, the steps outlined above represent a set of interdependent processes rather than a discrete, step-by-step procedure for proverb interpretation. This information was presented in a manner which was compatible with the in­ tellectual level of the individual subjects. Thus, the above discussion contains the information conveyed to the subjects, but it does not represent the manner in which it was communicated to the subjects. A brief manual was developed to help the Es explain these concepts to each subject. The manual can be found in Appendix IH. Following the above explanations, the E modelled the use of the SI 29 procedures in order to fulfill the above requirements of interpreta­ tion. For the proverb "When the cat’s away, the mice will play", the E’s modelling sounded like the followings

"What am X supposed to do? I’m supposed to interpret the meaning of this proverb. This requires that I figure out what the words in the proverb stand for. Now, let’s see. The cat is stronger and more powerful than the mice. Many times, the major words in the proverb stand for a type of person. So, if the cat and mice stand for people, the proverb means ’When the more powerful person isn’t around, the weaker people can play around- they don’t have to be on their guard’. Now, how can I make that more general? How about ’When authority is gone, responsible behavior decreases. Goodt I did it. If I stop and figure it out slowly, I can come up with a good interpretation."

Throughout training, much discussion necessarily ensued, emphasizing basic interpretation principles, discussing the adequacy and/or in­ adequacy of a given response, discovering the reasons for the adequacy or inadequacy of a response, etc.. Regarding the SI training proper, the sequence of events was exactly the same as described for the digit symbol test. Operant Training Group Subjects in this group were seen for the same amount of time and re­ ceived the same amount of experience with each training task as subjects in the SI group. However, training for these subjects did not include any reference to the SI procedure. Rather, training procedures for this group consisted of detailed instructions in conjunction with contingent positive social reinforcement. Operant training on the digit symbol task proceeded as follows. Parallel forms of Brown’s (1969) digit symbol test were presented to the subject. In addition,to receiving a description of the task re­ 30

quirements, the subjects were informed as to the nature and goals of the training sessions. Each subject was told that his job was to improve his performance during the training. Improving one’s performance on this task required that the subject produce a decreasing frequency of incorrect geometric figures over trials and an increasing speed of accurate performances. The subjects were also informed that the E would reinforce performances that were improvements over the preceding trial. Reinforcement took the form of the E saying "mmh-hmm", smiling, nodding his head, and verbalizing an encouraging ’'good”. Also, the sub­ ject was told if the reinforcement given was due to a lower number of errors, a faster speed, or both in relation to the preceding trial. The E’s task involved monitoring the accuracy of the subject’s perform­ ance and recording, trial by trial, the following: a) the number of in­ correct geometric figures produced; b) the time interval of each ex­ perimental trial; and c) whether or not the subject was reinforced on each trial. Points a) and b) were recorded so that the E would be able to reinforce the subjects accurately. The utility of c) will be de­ monstrated when describing the procedures used with the practice con­ trol group. Operant training of proverb interpretations included the same elabor­ ate descriptions given to the SI subjects concerning the nature of a proverb and the mechanics of its interpretation. Again, the subjects were informed of the reinforcement contingencies in effect. Here, the subject was reinforced for giving interpretations classified as Ab­ stract I, II, or HI according to Becker’s (1956) scoring system (see Appendix H). Social reinforcement was given in the manner described 31

for the digit symbol task. During the proverb training, the E recorded, verbatim, each interpretation given by the subject. Also, the E in­ dicated which interpretations were reinforced. Practice Control Group Subjects in this group received the same amount of time with the E and with each training task as subjects in the operant and SI groups. Instructions given to this group were exactly the same as the instruc­ tions given to the operant group, except that information concerning the reinforcement contingencies were deleted. Each subject in this group was yoked with a subject from the operant training group regarding the number of trials given on each training task and the pattern of social reinforcement. Thus, the only difference between this group and the operant training group was that reinforcement for this group was ad­ ministered in a noncontingent manner. Appendix V contains the random schedule used in assigning each subject’s group placement. In addition, this table shows which control subject was. yoked to each operantly trained subject. During each session with a subject from this group, the E knew the number of trials and the pattern of reinforcement that were given to the operant subject with whom the control subject was yoked. For ex­ ample, if the operant subject was reinforced on trials 5» 9, 12, 15, and 16- 30» the control subject was reinforced on the same triàls. In order to make the procedures comparable to the operant procedures (ex­ cept for the reinforcement contingency), the E recorded the same infor­ mation that was recorded during the operant training sessions. This group was employed in order to control for such variables as the time spent with the E, intercurrent life experiences, familiarity with the training tasks, and repeated testings. One final aspect of the procedure should be delineated at this time- 2 the training of Eg and E^. The author of the present paper, Ep was responsible for training the Es to administer each experimental condi­ tion. Thus, before Eg and E^ were trained, E^ ran three pilot subjects in order to resolve unforeseen procedural difficulties and to familiar­ ize himself with the specifics of each experimental procedure. At that point, the training of Eg and E^ was initiated. During an introductory session with the present author, the Es were given a des­ cription of the methodology for the present investigation. Also, each E was given the Method section from both the current proposal and the Meichenbaum and Cameron (1973) study. They were required to reread both papers until they had become quite familiar with the experimental proce­ dures. Further discussions were needed in order to clarify specific de­ tails of the procedures to the Es. In order to minimize experimenter bias, the rationale and hypotheses of the study were not revealed. After becoming familiar with the procedures, each E was required to role play the procedures with each other and with E . When all three “1 Es agreed that the procedures were well learned, Eo and E_ each ran one —j pilot subject with each set of procedures. During these sessions, E* monitored and, when necessary, corrected the behavior of the Es. Since E| tried to remain unaware of the group assignment of the sub­ jects, the experimental sessions were not observed directly. B* , how­ ever, remained available for consultation when questions arose. In addition, both Es tape recorded each condition at two points during the 33

study. These tapes were monitored by an independent rateP to evaluate

consistency of administration during the course of the study. Accord­ ing to the rater’s report, the procedures were implemented consistently throughout the study by both Es. RESULTS All measures were scared such that the rater was unaware of the sub­ ject’s identity, his group assignment, and the session being evaluated (i.e., pre-, post-, or follow-up session). Interrater reliability was established for two of the scoring systems used- proverb interpretations and inkblot responses. The remaining scaring systems were quite ob­ jective and did not require this evaluation. Interrater reliability was established between Eg and E^ for the proverb interpretation scoring system. This was essential in order to ensure that both Es evaluated and, thereby, reinforced interpretations in a similar manner during the training sessions. As E^ scared all in­ terpretations collected during the testing sessions, it was also necessary to obtain adequate scoring reliability between E^ and the training Es. One hundred proverb interpretations generated by pilot subjects were in­ dependently rated by E|, Eg, and E^. Interrater reliability coefficients were r12= .849, 2^= .837, and ^3= .873. Interrater reliability of the inkblot scoring system was evaluated by having E* and a second, independent rater*score two sets of 120 ink­ blot responses gathered during pre-, post-, and follow-up testing sessions. Interrater reliability for this system was found to be .925 on the second set of responses. Thus, scoring reliability of both systems was quite adequate for the present investigation. 34

A 3 (Method) x 2 (E) multivariate analysis of variance (Manova) was performed on the five dependent and five individual difference variables obtained during the pretest session. Far the 10 variables combined, there were no significant differences among the groups re­ ceiving the various treatment procedures (multivariate F= 1.3106; df= 20, 54; £>.05), between the groups seen by each E (multivariate F= 1.0603; df= 10, 27; £>.05), and on the test for a Method by E interac­ tion (multivariate F= 0.6315» df= 20, 54» £>.05). This analysis was conducted in order to evaluate group comparability prior to the intro­ duction of training. As the groups were not significantly different on the pretest measures, reliable posttest differences could not have been attributed to initial, pretreatment differences among the groups. r Appendix IV, Table 1-A, summarizes the corresponding univariate analyses for each of the 10 pretest variables. None of the 10 univariate tests was statistically significant. The second 3x2 Manova was conducted on the five dependent measures collected during the posttest session. Differences among the three treatment groups were not significant (multivariate F= 1.3102; df= 10, 64; £>.05). This result failed to support the hypotheses regarding the differential effectiveness of the three treatment methods. The analysis further revealed no reliable posttest differences due to the E administer­ ing the procedures (multivariate F= 0.7120; df= 5, 32; £>.05). This result confirmed, in part, the hypothesis of no differences between the groups seen by the two Es. The test for a Method by E interaction also 35 resulted in an X <'1.00 (multivariate F= .9072; df= 10, 64; £>.05). This suggested that levels of either factor were not dependent upon the level of the second factor. Appendix IV, Table 2-A, presents the univariate analyses of variance for the five posttest variables. Again, none of the univariate X ratios reached a statistically significant level. The purpose of the third 3x2 Manova, conducted on the five follow­ up measures, was to discover which, if any, posttest differences were maintained over a four week period. As no statistically significant differences resulted from the posttest data, the follow-up results were not surprising: Method (multivariate_F= 1.2229; df= 10, 64; £*> .05); E (multivariate F= 1.7739; df= 5. 32; jo > .05); and Method x E (multi­ variate F= .8733» df= 10, 64; p >.05). The follow-up analysis parallel ed the pre- and posttest results- no reliable differences for either factor or their interaction. Appendix IV, Table 3-A exhibits the uni variate F- ratios obtained from the follow-up data. On the follow-up univariate tests, a significant effect was obtained on the E factor for the HIT (F= 4.6339; df= 1. 36; £.< .05). Had the Manova not been conducted, this result would have been reported as a potentially reliable finding. This type of situation demonstrates the utility of the multivariate approach. Considering the follow-up analy­ sis, the multivariate F, testing for the equality of mean vectors, was not significant on the E factor. In other words, the multivariate dis­ tributions of the five variables, considered together, were not reliably different for the two groups. The purpose of the univariate is to answer the following question: which variable(s) account(s) for 36

the reliable differences between the multivariate distributions?. When the multivariate_F is not significant, as in the present case, questions directed to the univariate analyses become meaningless. Hence, the HIT result cannot be viewed as a statistically significant event. It may be helpful to review the hypotheses that relate to the above analyses. The SI and operant techniques were viewed as the primary treatment procedures in the current investigation. Thus, it was pre­ dicted that the SI and operant groups would perform significantly higher on the posttest and follow-up than the yoked control subjects. The analyses, however, failed to support this hypothesis. It was further predicted that the SI group would perform significantly higher on all measures than either of the other two groups (due to greater generaliza­ tion across situations). Also, the SI group was expected to exhibit the greatest amount of generalization across responses. In other words, the SI group, compared to the other treatment groups, was expected to perform significantly higher on the three variables not used in training. These predictions included both the posttest and follow-up results. Again, the'Manova results did not support the hypotheses. Thus,far, the only hypothesis supported concerns no differences between the groups seen by the two training Es. Returning to the statistical analyses, the above Manovas revealed that the groups were not reliably different at the pre-, post-, or follow-up sessions. However, these analyses offered no information con­ cerning group*trends over time. In order to obtain this information, multivariate trend analyses were conducted separately on each of the five dependent variables. Table 1 summarizes the trend analysis for 37

TABLE 1 Multivariate Trend Analysis for Digit Symbol

Source df MS F

Trend (multivariate) 2, 35 - 26.0604 ** Linear 1 565.7618 45,7695 ** Quadratic 1 13.3493 1.0644 Method x Trend 4. 70 - 5.1001 ** (multivariate) Linear 2 54.5119 4.4100 * Quadratic 2 58.0278 4.6270 *

E x Trend 2, 35 1.2373 ^multivariate) Linear 1 23.0476 1.8645 Quadratic 1 16.2540 1.2961 Method x E x Trend 4, 70 - 1.6897 (multivariate ) Linear 2 19.0833 1.5438 Quadratic 2 18.2659 1.4565 Within Cells Linear 36 12.3611 Quadratic 36 12.5410

* P_< .05 ♦* p < .01 38

the digit symbol task. The Method x trend interaction was significant on this analysis (multivariate F= 5.1001; df= 4, 70; _£ < .01). The univariate statis­ tics further revealed that this difference included significant linear and quadratic trends (see Table 1). Figure 1 represents these data, exhibiting the trends over time for the three treatment groups. Tukey’s HSD post hoc test was used to perform all pairwise comparisons of the means within each treatment group. Table 2 summarizes the findings of the post hoc analysis. It was discovered that both the operant and con­ trol subjects performed significantly higher on the posttest than on the pretest. Also, posttest levels were maintained by both groups at the four week follow-up. For the SI group, however, there was no signif leant difference between pre- and posttest levels. Yet, there was a significant improvement on the follow-up relative to pretest and post­ test observations. The test for trend, combining all 42 subjects regardless of group assignment, was also significant for the digit symbol task (multivariate _F= 26.0604; df= 2, 35» £ < .01). The subsequent univariate analyses re­ vealed that a simple linear equation adequately represented the form of the trend over time (Flinear= 45.7695» df= 1, 36;_£_<.01). Figure 2 demonstrates that the significant linear trend reflected an increasing performance on this task from pretest through the follow-up session. Table 3 presents the trend analysis for proverb interpretation. Here, the only significant result consisted of linear and quadratic trends over time for the 42 subjects combined. Figure 3 graphically depicts this result; Table 4 presents the HSD test results on the com- 39

Figure 1. Pretest, posttest, and follow-up mean performance for the SI, operant, and control groups on digit symbol. 40

TABLE 2 Tukey’s HSD Post hoc Test Results for Digit Symbol All Pairwise Comparisons of Means Within Each Treatment Group

Comparison df a

SI

Pretest vs. Posttest 3. 78 1.5434 Pretest vs. Follow-up 3. 78 4.5018 ** Posttest vs. Follow-up 3. 78 6.0453 ** Operant Pretest vs. Posttest 3. 78 12.0911 *♦ Pretest vs. Follow-up 3. 78 14.5350 ** Posttest vs. Follow-up 3. 78 2.4437 Control Pretest vs. Posttest 3. 78 7.2031 ** Pretest vs. Follow-up 3. 78 9.0039 ** Posttest vs. Follow-up 3. 78 1.8006

** p<.01 41

Testing Sessions Figure 2. Pretest, posttest, and follow-up mean performance for all subjects combined on digit symbol.

Figure 3. Pretest, posttest, and follow-up mean performance for all subjects combined on proverb interpretation. 42

TABLE 3 Multivariate Trend Analysis for Proverb Interpretations

Source df MS F

Trend (multivariate) 2, 35 - 14.5555 ** Linear 1 5.0519 28.7702*** Quadratic 1 2.6825 7.1892* Method x Trend 4, 70 - 1.1611 (multivariate) Linear 2 0.1837 1.0461 Quadratic, 2 0.7027 1.8831 E x Trend 2. 35 - 1.3820 ^multivariate)

Linear 1 0.2976 1.6949 Quadratic 1 0.1429 0.3829 Method x E x Trend 4, 70 - 0.8753 (multivariate) Linear 2 0.0158 0.0902 Quadratic 2 0.5665 1.5183 Within Cells Linear 36 0.1756 Quadratic 36 0.3731

*P<.05 **p<.01 43

TABLE 4 Tukey’s HSD Post hoc Test Results for Proverb Interpretations All Pairwise Comparisons of Means for All Subjects Combined

Comparison df a

Pretest vs. Posttest 3, 82 6.7642 *♦ Pretest vs. Follow-up 3. 82 5.9805 ** Posttest vs. Follow-up 3, 82 0.7841

**p<.01 bined means. Statistically, the subjects improved significantly on the posttest and follow-up over their pretest level. In addition, posttest performance was maintained on the four week follow-up. The following prediction had been made concerning the digit symbol and proverbs measures on the trend analyses: both the SI and operant groups were expected to perform significantly higher on posttest and follow-up measures relative to their respective pretest levels. The results suggested, however, no differential trends over time for the various treatment groups. The third trend analysis was conducted on the HIT. Table 5 presents the results of this analysis which revealed only a significant change over time for all subjects combined. Both linear and quadratic compo­ nents were significant on the univariate tests. Figure 4 exhibits the specific form of the trend. Tukey’s analysis (Table 6) of the-combined subject means revealed that the subjects improved significantly from 44

TABLE 5 Multivariate Trend Analysis for the HIT

Source MS -FT ■

Trend (multivariate) 2». 35 10.9542 ** Linear 1 0.8201 5.8070 *

Quadratic 1 2.2289 16.1917 ** Method x Trend 4, 70 « 0.3682 (multivariate) Linear 2 0.0062 0.0438 Quadratic 2 0.1005 0.7299 E x Trend 2, 35 - 2.4745 "(multivariate ) Linear 1 0.1458 1.0326 Quadratic 1 0.5432 3.9461 Method x E x Trend 4, 70 - 2.0270 (multivariate) Linear 2 0.2062 1.4600 Quadratic 2 0.4148 3.0130 Within Cells

Linear 36 0.1412 Quadratic 36 0.1376

*p<.05 **p <.01 45 l e v e L

e s n o p s e R

T I H

n a e M

Testing Sessions Figure 4. Pretest, posttest, and follow-up mean performance for all subjects combined on the HIT. 46

TABLE 6 Tukey’s HSD Post hoe Test Results far the HIT All Pairwise Comparisons of Means for All Subjects Combined

Comparison df a

Pretest vs. Posttest 3, 82 3,0400 * '•' Pretest vs. Follow-up 3, 82 3.2786 ♦ Posttest vs. Follow-up 3. 82 6.3184 ♦*

*2 <.05 **p< .01 pre- to posttest; however, follow-up performance was significantly low­ er than either pre- or posttest levels. None of the trend interactions was significant on the repeated measures analysis of the auditory distraction task. Table 7 summarizes these results which reveal no reliable change over time on this variable. Table 8 presents the trend analysis of the ward association measure. A significant trend over time was found for the 42 subjects combined. Figure 5 graphically reflects this result; Table 9 presents the results of Tukey’s analysis on this variable. The only significant difference occurred on the posttest vs. follow-up test comparison. Posttest and follow-up performances were not reliably different from pretest levels. This concludes the major statistical analyses of the current in­ vestigation. However, two additional results are presented which may be of interest to other researchers, in this field. Table 4-A (Appendix IV) presents the intereorrelations of the 10 pretest measures. The five lv?

TABLE 7 Multivariate Trend Analysis for the Auditory Distraction Task

Source df MS F

Trend (multivariate) 2. 35 «ft 3.1791 Linear 1 11.4404 0.3190 Quadratic 1 211.7506 6.5327 * Method x Trend 4, 70 - 2.3070 (multivariate) Linear 2 126.6190 3.5308 * Quadratic 2 33.7619 1.0416 E x Trend 2, 35 - 1.5613 (multivariate) Linear 1 10.0119 0.2793 Quadratic 1 81.1468 2.5034 Method x E x Trend 4, 70 - 0.1181 (multivariate) Linear 2 5.9048 0,164? Quadratic 2 1.5873 0.0490 Within Cells Linear 35.8611 Quadratic 36 32.4140

*P<.05 48

TABIE 8 Multivariate Trend Analysis for the Word Association Task

Source df MS F

Trend (multivariate) 2,.35 3.9194 ♦ Linear 1 22.0119 1,1800 Quadratic 1 90.4800 6.6546 * Method x Trend 4, 70 - 0.7011 (multivariate) Linear 2 18.3690 0.9847 Quadratic 2 5.9563 0,4381 E x Trend 2, 35 - 0.3618 Tmultivar iate )

Linear 1 10,0119 0.5367 Quadratic 1 2.4803 0.1824 Method x E x Trend 4, 70 - 0.7393 (multivariate)

Linear 2 11,0833 0.5951 Quadratic 2 12.9087 0.9494 Within Cells

Linear 36 18.6548 Quadratic 36 13.5966

*p <.05 49

Figure 5. Pretest, posttest, and follow-up mean performance for all subjects combined on the ward association task. 50

TABLE 9 Tukey’s HSD Post hoc Test Results for the Word Association Task All Pairwise Comparisons of Means for All Subjects Combined

. Comparison df

Pretest vs, Posttest 3, 82 2.1171 Pretest vs. Follow-up 3 , 82 1,6858 Posttest vs. Follow-up 3. 82 3.8014 **

**p<.01 individual difference variables were examined due to their possible rela tionships to the dependent measures. Table 4-A reveals correlations greater than .30 between each individual difference variable and at least one of the dependent variables employed. Thus, depending upon the task being used, it may be important to control for one or more of the individual difference variables examined in this study. The second potential function of the information contained in Table 4-A regards the selection of dependent variables for research purposes. In the present study, variables were selected on the basis of their apparent dissimilarity, rather than on the basis of previous correla­ tional data. Selections were made in this manner due to a lack of in­ formation regarding variable relationships in the psychological litera­ ture. The selection of dependent variables, when investigating chronic, nonparanoid, schizophrenics, may be facilitated by the availability of this type of information. 51

Table 5-A (Appendix IV) presents the means and standard deviations for each of the six groups, on each of the five dependent variables, far the three testing sessions. In addition, the means and standard deviations of the five individual difference variables are included. Figures 1-A through 5“A in Appendix IV present the mean data for the dependent variables in graphic form. The most striking feature of these data was the magnitude of the obtained standard deviations. By traditional diagnostic and research standards, the subjects were a rela­ tively homogeneous group of schizophrenics. However, the high level of variability suggested that a rather heterogeneous group of subjects were employed. These results indicated that the current methods em­ ployed to form homogeneous groups of schizophrenic subjects may be quite inadequate. This will be discussed further in the next section of the report. DISCUSSION The major hypotheses of the present investigation were not supported by the empirical data obtained. Separate analyses of the pre-, post-, and follow-up testing data revealed no significant differences among the three treatment methods, between the two training Es, or on the Method by E interaction. The hypothesis of no differences between the groups seen by the two training Es was the only hypothesis supported by these analyses. The repeated measures analyses revealed significant trends over time on four of the five dependent variables. On the digit symbol task, a significant Method by trend interaction reflected different learning curves across time for the SI, operant, and control groups. Specifically, 52 posttest performance was significantly higher than pretest performance for both the operant and control groups. Both groups sustained this improvement on the follow-up. The performance of SI subjects, however, was not reliably different on the pre- and posttest. Yet, the follow­ up data was significantly higher than pre- and posttest levels for the SI group. The repeated measures analysis of the digit symbol task also revealed a significant linear trend over time for the 42 subjects com­ bined. This reflected an improved performance from the pretest through the follow-up when the treatment and E assignments of the subjects were disregarded. The results suggested a significantly improved performance over time on the digit symbol task for both the operant and control groups. One possible explanation of the SI data required a closer examination of the technique in relation to the nature of the digit symbol task. The SI procedure was administered in an extremely slow and deliberate man­ ner. However, even if the training had been helpful to the subjects, two hours of practice at a deliberately reduced rate may have impeded efforts to ’’perform as quickly and as carefully as possible” on the posttest. Thus, no differences between pretest and posttest were dis­ covered. At the four week follow-up, the effects of the training ex­ perience may have emerged as the temporary, practice effects of perform­ ing the task at an artifically lowered rate diminished. Hence, improved follow-up performance, relative to pretest levels, resulted. This type of ’’delayed training effect” would only be expected to occur on timed tasks, such as the digit symbol measure. This was the only timed task employed and, in fact, this pattern was not found for ary of the other 53 variables. The following conclusion was reached concerning the digit symbol task. Except for the ’’delayed training effect’’ exhibited by the SI subjects, no differences were found among the three treatment groups. This conclusion was also supported by the pre-, post-, and follow-up test Manovas. Therefore, procedures in excess of those administered to the control groups did not contribute significantly to the improved per­ formance of the subjects over time. Performance on the second training task- proverb interpretation- exhibited significant linear and quadratic trends over time for the 42 subjects combined. More specifically, the subjects improved signifi­ cantly on the posttest relative to pretest levels, and the posttest performance was maintained on the follow-up evaluation. No differences were discovered among the three treatment procedures, between the two training Es, or on the interaction of the two factors. In other words, these results were quite similar to the digit symbol findings. The conclusion reached for the digit symbol data can be generalized to the proverb interpretation data as wells no significant differences existed among the groups regarding their performance on the two train­ ing variables. Procedures in excess of administering the test sessions and providing the subjects with instructions, practice, noncontingent reward, and contact with the Es did not contribute significantly to the im proved performance of the subjects over time. The remaining data were primarily concerned with the issue of gen­ eralization across responses. For the HIT, the only significant trend observed involved the 42 subjects in combination. Improvement from pre­ & test to posttest was reliable; however, this improvement was not sus­ tained on the four week follow-up. In fact, follow-up performance was significantly lower than pre- and posttest levels. The HIT was the first, and last variable to demonstrate generaliza­ tion across responses. The remaining variables- auditory distraction and word association- were also employed to examine this type of gener­ alization. Neither variable, however, exhibited significant change rel­ ative to pretest levels. Thus, generalization across responses was dem­ onstrated on only one of the three response generalization measures. In addition, the effect was not enduring, as it was not maintained over a four week period. To summarize, the only difference among treatment groups was found on the trend analysis of the digit symbol task. As discussed above, this may have reflected a rather transient difference in the effectiveness of the three training procedures. In all the remaining analyses, no differences were discovered among the six experimental groups. Thus, except for the null hypothesis concerning no E effect, the major hy­ potheses of the study were not supported. Differential effectiveness of the various procedures was not demonstrated on tasks trained directly or on tasks designed to test for generalization across responses. Also, as all measures were collected by an E not involved in training, no differences were discovered among the groups regarding generalization across situations. The above conclusion-regarding.the results of the two •braining measures is equally applicable to the present findings con­ cerning generalization across situations and responses. Procedures in excess of the control procedures (i.e., administering the test sessions, 55 providing the subjects with instructions, practice, noncontingent re­ ward, and contact with the Es) did not contribute significantly to the improved performance of the subjects over time. The results of the current investigation appear to be at variance with the Meichenbaum and Cameron (1973) findings. However, numerous differences between the studies suggest that their results may not be directly comparable. One of the most striking differences between the two studies regarded the subject samples employed. In the current in­ vestigation, all subjects were chronic, nonparanoid schizophrenics from the Veterans Administration Center, Wood, Wisconsin. No significant differences existed among the groups on the five individual difference variables measured. The subjects were drawn from the Domiciliary and Day Treatment Centers at the V.A. Hospital. The former location is primarily a residential facility, requiring adequate self care abilities and, at most, a minimum amount of psychopathology. Day Treatment members reside in the community and attend hospital activities on a daily basis. Members of this facility must be capable of independent living and can­ not function effectively on the unit with severe behavioral difficulties. In the Meichenbaum and Cameron (1973) report, half of the subjects were chronic, nonparanoid schizophrenics; the other half consisted of acute, paranoid schizophrenics. The report did not reveal the groups* compara­ bility on the individual difference variables measured in the current study. In addition, subjects were selected from both acute admission wards and chronic back wards. Thus, it seems extremely likely that the samples employed in the two investigations were quite different in many respects. Other differences between the studies included: amount of time 56

spent in training, number and type of training tasks employed, and the period of time between the posttest and follow-up evaluations. Another important difference between the two investigations regarded the comparison groups employed. The Meichenbaum and Cameron (1973) study compared the SI group with a yoked practice control group; Sub­ jects in both groups received the same number of training sessions, assessments, and trials on each task. In the current investigation, subjects in the operant group received many more trials on each training task than subjects in the SI group. As the present control subjects were yoked to the operant subjects, both comparison groups received substantially more trials than the subjects in Meichenbaum and Cameron’s (1973) control group. Each of the above noted differences is important, as they relate to crucial independent variables in the behavior modification literature (Paul, 1969). In retrospect, it is not surprising that the two studies, having a multitude of critical differences, reported disparate results, Meichenbaum and Cameron (1973) reported significantly more improvement for the SI group than for the practice control group. The present in­ vestigation, using different subjects, procedures, and comparison groups, failed to demonstrate ary differential effectiveness of the various treatment techniques. The immediately preceding discussion suggests a large number of fu­ ture research directions. For example, the SI and operant procedures may each be the most effective with particular subtypes of schizophrenic individuals. Thus, various schizophrenic subtypes could be employed to investigate differential responsiveness to the treatment methods. This 57 research area becomes even more complex, as differential effectiveness can be assessed in terras of direct training effects, generalization a- cross situations, and/or generalization across response measures. An­ other area of investigation could attempt to replicate the present study using the same, or different dependent variables. It may be dis­ covered that the variables exhibiting change may depend upon the par­ ticular schizophrenic subtype employed as well as the type of change being examined- i.e., direct training effects, generalization across situations, or generalization across responses. The present research was a first effort to investigate the E factor using operant techniques with schizophrenic subjects. Further examina­ tion of this variable is required to evaluate the reliability of the current empirical findings. Again, interactions between the E variable and the factors listed above may be important. A final comment regarding future research directions relates to the area of schizophrenia investigation in general. By current research and diagnostic criteria, the present study employed a relatively homogeneous sample- i.e., chronic, nonparanoid schizophrenics. let, as Table 5-A attests, variability among the subjects was extremely high. This suggested that traditional classification methods may be unsatisfactory to specify a homogeneous subject sample. A schizophrenic sample could be more adequately defined by matching subjects on pretest or preselec­ tion measures to reduce intersubject variability. Of course, this raises another methodological difficulty, as the final sample would no longer be representative of the schizophrenic subtype being studied. This po­ tential criticism appears to be a minor one, however, considering the 58 increased subject specificity that would result. An alternative approach to the problem consists of the application of operant methodology to this area of investigation. This methodology has been described in detail by Sidman (i960). Gottman (1973), dis­ cussing the approach in the context of psychotherapy research, summar­ izes its potential advantages over the more traditional strategy employed in the current study. In any event, it is believed that the problem of intersubject vari­ ability requires additional research consideration. As Paul (1969) has indicated, it is important to accurately identify the particular subjects employed in behavior modification research. Until this classification difficulty is adequately acknowledged, investigated, and resolved, re­ search in the area will be greatly impeded. 59

FOOTNOTES * Dr. Donald Meichenbaum, personal communication, August, 1973. 2 The training Es were senior undergraduate psychology majors from the University of Wisconsin- Milwaukee: Greg Jurenec and Mike Halmo. 3 Dr. Robert Hall monitored the tape recordings of the training sessions. 4 Dr. Donald Leventhal served as the second HIT rater. references

Allen, D. J., & Magaro, P. A. Measures of change in token economy programs. Behaviour Research and Therapy. 1971, 2» 311-318. Atthowe, J. M., & Krasner, L. Preliminary report on the application of contingent reinforcement procedures (token econony) on a ’’chron­ ic” psychiatric ward. Journal of Abnormal Psychology. 1968, 73(1).

37-43. Ayllon, T., & Azrin, N. H. The measurement and reinforcement of be­ havior of psychotic s. Journal of the Experimental Analysis of Be­ havior, 1965. 8, 357-383. Ayllon, T., & Azrin, N. H. The token economy« a motivational system for therapy and rehabilitation. New York: Appleton-Century-Crofts, 1968. Baker, R. The use of operant conditioning to reinstate speech in mute schizophrenics. Behaviour Research and Therapy, 1971, 2» 329-336. Benjamin, J. D. A method for distinguishing and evaluating formal thinking disorders In schizophrenia. In J. S. Kasanin (Ed.), Language and thought in schizophrenia. Los Angeles: University of California Press, 1944. Pp. 65-90. Bartz, W. R., & Ley, D. L. The Shipley-Hartford as a brief I.Q. screen­ ing device. Journal of Clinical Psychology. 1970, 26. 74-75. Becker, W. C. The relation of severity of thinking disorder to the process-reactive concept of schizophrenia. Unpublished doctoral dissertation, Stanford University, 1955. Bucher, B., & Fabricatore, J. Use of patient-administered shock to suppress hallucinations. Behavior Therapy. 1970, 1_, 382-385. 61

Brown, M. A set of eight parallel forms of the digit symbol test. Unpublished test, University of Waterloo, Waterloo, Ontario, 1969. Cash, T, F. Methodological problems and progress in schizophrenia re­ search« a survey. Journal of Consulting and Clinical Psychology. 1973. 40, 278-286. Cohen, R., Florin, I., Grusche, A., Meyer-Osterkamp, S., & Sell, H. The introduction of a token economy in a psychiatric ward with ex­ tremely withdrawn chronic schizophrenics. Behaviour Research and Therapy. 1972, 10(1), 69-74. D’Allesso, G. R., & Spence, J. T. Schizophrenic deficit and its rela­ tion to social motivation. Journal of Abnormal and Social Psychol­ ogy. 1963, 66, 390-393. Ellsworth, J. R. Reinforcement therapy with chronic patients. Hospital and Community Psychiatry. 1969, 20. 238-240. Gorham, D. R. A proverbs test for clinical and experimental use. Psychological Reports. 1956, 2, 1-12. (a) Gorham, D. R. Use of the proverbs test for differentiating schizophren­ ics from normals. Journal of Consulting Psychology. 1956, 20(6). 435-440. (b) Gorham, D. R., Green, L. W., Caldwell, L. R., & Bartlett, E. R. Effect of operant conditioning techniques on chronic schizophrenics. Psychological Reports, 1970, 27, 223-234. Gattman, J. M. N-of-one and N-of-two-research in psychotherapy.' Psycho' logical Bulletin, 1973. 80(2),- 93-105. Hendersen, J. D. A community-based operant learning environment. Is Overview. In R. D. Rubin, H. Fensterheim, A. A. lazarus, and C. W. 62

Franks (Eds.), Advances in behavior therapy. New York: Academic Press, 1971. Pp. 233-237. Hollister, L. E. Choice of antipsychotic drugs. American Journal of Psychiatry. 1970» 127. 186-190. Holtzman, W,, Thorpe, J., Swartz, F., & Herron, W. Inkblot perception and personality. Austin« University of Texas Press, 1961. Isaacs, W., Thomas, J., & Goldiamond, I. Application of operant con­ ditioning to reinstate verbal behavior in psychotics. Journal of Speech and Hearing Disorders, I960, 25. 8-12. Jenkins, J. J., & Palermo, D. S. A note on scoring word association tests. Journal of Verbal Learning and Verbal Behavior. 1964, J, 158-160. Kale, R. J., Kaye, J. H., Whelan, P. A., & Hopkins, B. L. The effects of reinforcement on the modification, maintenance, and generaliza­ tion of social responses of mental patients. Journal of Applied Behavior Analysis. 1968, 1.(4), 307-314. Kaufman, L. N. The development of a proverb scale for the measurement of thinking pathology in schizophrenia and a further investigation of the process-reactive dimension. Unpublished master’s thesis, University of Illinois, I960. Kazdin, A. E., & Bootzin, R. R. The token economy: an evaluative re­ view. Journal of Applied Behavior Analysis. 1972, ¿(3)» 343-372. Liberman, R. P. Behavioral modification of schizophrenia: a review. Schizophrenia Bulletin. 19?2, No. 6, 37-48. Llqyd, K. E., & Abel, L. Performance on a token econony psychiatric ward: a two year summary. Behaviour Research and Therapy. 1970. 8, 63

1-9, Lyon, V. L. Conditioning and generalization of verbal response in hospitalized schizophrenics. Unpublished doctoral dissertation, University of Kansas, 1971. Meichenbaum, D. H. Effects of social reinforcement on the level of abstraction in schizophrenics. Journal of Abnormal Psychology. 1966. 21(5), 354-362. Meicheribaum, D. H. The effects of instructions and reinforcement on thinking and language behavior of schizophrenics. Behaviour Re­ search and Therapy. 1969, 2» 101-114. Meicheribaum, D. H., & Cameron, R. Training schizophrenics to talk to themselves« a means of developing attentional controls. Behavior Therapy. 1973, 4, 515-534. McGhie, A., Chapman, J., & Lawson, J. S. The effect of distraction on schizophrenic performance. (1) Perception and immediate memory. British Journal of Psychiatry. 1965, 111. 383-390. M^ers, J. K., & Bean, L. L. A decade later: a follow-up of social class and mental illness. New York« Wiley, 1968. Neale, J. M., & Cromwell, R. L. Attention and schizophrenia. In B. A. Maher (Ed.), Progress in experimental personality research, Vol. 5. New York« Academic Press, 1970. Pp. 37-66. Palermo, D. S., & Jenkins, J. J. Word association norms. Minneapolis» University of Minnesota Press, 1964. Patterson, R. L., & Teigen, J. R. Conditioning and post-hospital generalization of nondelusional responses in a chronic psychotic patient. Journal of Applied Behavior Analysis. 1973, 6, 165-170. 64

Paul, G. L. Behavior modification research* design and tactics. In C. M. Franks (Ed.), Behavior therapy« appraisal and status. New York« McGraw-Hill, 1969. Pp. 29-62. Richardson, R., Karlsalas, Y., & Lal, H. Application of operant pro­ cedures in treatment of hallucinations in chronic psychotics. In R. D. Rubin, H. Fensterheim, A. A. lazarus, & C. M Franks (Eds.), Advances in behavior therapy. New York« Academic Press, 1971. Pp. 147-150. Salzinger, K., & Pisoni, S. Reinforcement of affect responses of schiz ophrenics during the clinical interview. Journal of Abnormal and Social Psychology. 1958, 57. 84-90. Schaefer, H. H,, & Martin, P. L. Behavioral therapy for "apathy” of hospitalized schizophrenics. Psychological Reports, 1966, 19, 114? 1158. Shakow, D. Psychological deficit in schizophrenia. Behavioral Science 1963, 8, 275-305. Shipley, W. C. Shipley Institute of living Scale for measuring in­ tellectual impairment« manual of directions and scaring key. Hart­ ford« The Institute for Living, 1939. Sidman, M. Tactics of scientific research. New York« Basic Books, I960. Silverman, J. The problem of attention in research and theory of schizophrenia. Psychological Review, 1964, 21(5), 352-379. Silverman, J. Variations in cognitive control and psychophysiological defense in schizophrenics. Psychosomatic Medicine, 1967, 29, 225- 251. 65

Sommer, R., Witney, G., & Osmond, H. Teaching common associations to schizophrenics. Journal of Abnormal and Social Psychology, 1962, 65(1). 58-61. Spohn, H. E. The case for reporting the drug status of patient sub­ jects in experimental studies of schizophrenic psychopathology. Journal of Abnormal Psychology. 1973. 82(1). 102-106. Steffy, R. A., Hart, J., Crow, M., Torney, D., & Marlett, N. Operant behaviour modification techniques applied to a ward of severely regressed and aggressive patients. Canadian Psychiatric Associa­ tion Journal. 1969» 14. 59-67. Ullmann, L. P., Forsman, R. G., Kenny, J. W., McInnis, T. L., Unikel, I. P., & Zeisset, R. M. Selective reinforcement of schizophrenics’ interview responses. Behaviour Research and Therapy. 1965, 2, 205- 212. Ullmann, L. P., Krasner, L., & Edinger, R. L. Verbal conditioning of common associations in long-term schizophrenic patients. Behaviour Research and Therapy. 1964, 2, 15-18. Wagner, B. R. The training of attending and abstracting responses in chronic schizophrenics. Journal of Experimental Research in Per­ sonality. 1968, J, 77-88. Wilson, F. S., & Walters, R. H. Modification of speech output of near­ mute. schizophrenics through social learning procedures. Behaviour Research and Therapy. 1966, 4, 59-67. Wolff, R. The systematic application of the satiation procedure to de­ lusional verbiage. Psychological Record, 1971, 21.(4), 459-463. Tates, A. Behavior therapy. New lark: Wiley, 1970. 66

APPENDIX I. DEPENDENT MEASURES AND TRAINING TASKS. 6?

DIGIT SYMBOL TASK The following pages present the forms of Brown’s (1969) digit sym­ bol task used in the current study. Forms I, II, and III were used as dependent measures. The remaining forms were employed as training tasks during the experimental sessions. The width of the boxes has been reduced by one-sixth to conform to the required format standards. 68

26 29 42 53 64 89 93 COLS FORM I J © □ < î

64 93 ¡53 26 29 64 93 ¡89 29 42 PRACTICE r ■'T" ’ i ? 1

93 ^89 26 42 26 29 4 U1 8.9.. 42 29 53 93 29 42 89 42 26 Î „—L™

i ] ¡26 53 64 93 29 ¡93 29 26 89 29 64 29 93 42 53 29 64 89 53 29 j

""""" 53 93 26 42 89 29 64 53 64 26 64

42 93 42 26 29 53 26 89 29 89 29 93 29 89 64 26 42 29 64 89

lu 1 69

25 36 48 62 69 72 91 CODE a|ì © U h]O iÛ

62 j 91 Ì69 72 ¡36 25 36 72 69 48 ima 1 PRACTICE A üLj LL

r" ’ 62 ¡72 T~" 69 48 62 72 69 36 48 36 91 62 36 48 ¡72 48 69 48 62 69 i !

---- , 91 25 62 36 69 72 36 25 62 36 62 36 48 91 36 25 72 91 36 91

36 91 1 36 72 62 25 91 62 25 36 25 91 69 25 48 ¡91 62 72 91 36 j 1

48 62 62 91 ¡72 62 48 ¡62 48 62 48 25 72 36 72 36 _ 69 36 91 69 1

IJ " ■llr 25 36 25 36 62 69 91 62 36 62 25 48 36 25 48 36 62 25 69 36 ¡IIWMWÏkLV —

91 48 91 36 72 69 72 91 48 91 62 36 91 36 25 25 25 25 --2-5-- L 70

jn 37 38 41 56 71 CODE FORM III Bc 00 □ X z°

r-"™ (■ '" 98 41 15 98 56 38 15 71 37 41 PRACTICE — □

3« I15 98_ 56 38 56 37 41 38 71 38 98 41 98 ¡71 | 37 56 15 56 41 *—i ... . L L

38 37 15 56 38 98 15 38 98 15 38 15 56 71 37 41 98 71 98 37

---- J

37 98 56 15 41 j 38 71 98 15 37 56 37 ¡71 98 38 15 37 71 37 15 |

41 37 71 37 71 56 ¡71 56 98 37 15 56 37 56 38 98 71 15 56 r HU.,-.. i 71

28 39152 54 68 82 95 CODS ¿Si FORM IV □ilfeo' * •

28 1 52 28 68 82 54 82 39 95 95 i PRACTICE □

54 52 28 82 52 54 95 39 52 95 54 52 28 52 95 39 54 68 28 54

28 82 f 28 39 28 52 54 28 95 82 68 54 68 39 68 39 154 39 82 52 j ■-J 1 ■ - -

28 95 54 95 54 28 52 68 52 28 68 95 52 82 95 52 j 82 39 52 '54 L±

82 39 52 95 52 54 39 68 28 68 52 54 82 28 82 68 54 68 28 | 52 1 - ■ L L

54 j 39 39 82 39 54 68 39 82 95 54 ¡52 68 68 39 95 52 95 68 28 1

□7 5 54 68 54 82 95 39 95 82 95 52 28 82 39 28 39 28

l_ 72

19)24 47 78 84 CODE FORM V =M 0 Ö

24 k 19 24 47 24 78 94 87 19 PRACTICE V

94 84 87 78 84 24 19 87 ¡78 94 84 87 19 47 84 24 19 ¡87 78 84 j j

87 47 94 87 78 94 84 24 87 94 24 47 24 94 78 19 94 78 47 24

• 19 47 87 84 24 94 19 24 78 84 87 78 84 78 94 94 94 ..... 19 87 47

¡MJ"«»“ 19 87 24 84 19 78 84 94 87 94 19 87 24 84 19 78 19 87 24 47

19 47 94 78 47 94 24 87 24 84 94 24 78 47 84 78 87 47 ¡84 24

1 84 24 19 4? 94 78 84 78 87 84 47 94 24 19 94 24 87 84 78 19 I 73

ü»hcm, r~T '1 12 31 45 ¡51 73 751 97 j CODE Ao|r □N

75 ¡12 97 31 12 75 45 97 51 73 PRACTICE sA d j

12 h ’—“"I 45 12 97 51 75 31 ¡12 75 12 31 51 73 51 31 12 51 31 45 ffl r

12 ¡97 31 45 45 12 51 73 12 51 45 31 97 73 51 12 ¡73 12 97 51 dj

12 75 51 73 12 45 12 31 73 75 45 73 97 73 45 51 12 73 75 12

51 75 12 97

73 ¡97)51 12 73 12 75 45 12 31 97 73 31 1 73 12 45 45 51 75 73 L ,----L---

---- j— r~~ 97 75 12 51 73 45 |75 31 97 51 97 75 97 31 12 51 97 31 45 12 i 7^

WKUWH unu’j uw 14 17 23 59 67 7 CODE FORM VII n B> 1 -o<

r™"1 23 J 76 17 J14 ¡23 17 76 59 86 67 PRACTICE 4™" f— 1 ! UL s L___

r—f““- i?h6 86 j 23 67 17 23 14 67 17 76■ ■ 17 ¡76 14 59 17 14 23 Ll? ¿0 j kbcmmt asasa»» iii'in i.iiji.

76 ¡59 23(76 ~~r—■ 17 86 86 17 86 23 67 76 86 14 L 17! 59 59 23 67 1, i

23 176 23(59 59 67 17 67 23 14 86 76 14 14 17 67 14 86 23 67 .... JiL..— 1 '.... |....

r~”r 14 86 76 14 59 23 14 23 76 67 14 67 59 76 17 76 23 14 86 59

■ r

17 59 76 75

AUDTWRX DISTRACTION TEST The sequence of numbers and digits for each form of the test is presented below. Underlined units are the distractor stimuli- i.e., stimuli presented by a male voice which the S was asked to ignore. FORM I FORM II FORM HI 1) D2Q2H2V8C1Z 1) Z6CJX2G6K6A 1) M2V401Z8S2T 2) G8TOT8C2G6D 2) DBJEXT 2) 120289 3) 297378 3) DEMVQO 3) ZQRQTK 4) K1T2T^A6POJ 4) TQFXNV 4) Q0KT4J2H8H 5) DUJCCE 5) 2G3N7G8Q6P0 5) CTIEAZ 6) PAPJZH 6) 496027 6) 554988 7) 182057 7) 3F7K4W4F4T3 7) V2H2X2K1N4F 8) 2H8F6G8A8P0 8) 258496 8) VTGZBM 9) Q2P2P8C202S 9) 067826 9) 508214 10) 780936 10) ^K6F6F1G1D 10) 3B4U0R8Q.6B6 11) 5DOK5ROI8B7 11) Z2G2F1W4Y

PROVERBS Forms I, II, and III of Gorham's proverbs test are presented. A listing of the 23 proverbs used in training are also provided 71

PROVERBS TEST I DIRECTIONS« Below are some proverbs. You are to write out the mean­ ing of each one in the space below it. Far example, the proverb "Large oaks from little acorns grow" means that great things may have small beginnings. Now fill in the meaning for those given below. Remember- You are to tell what the proverb means rather than to just tell mare about it. Try to answer every one.

1. Where there’s a will, there’s a way. 2. Rome was not built in a day. 3. When the cat’s away the mice will play. 4. Barking dogs seldom bite. 5. A stream cannot rise higher than its source. 6. Don’t swap (trade) horses when crossing a stream. 7. The used key is always bright. 8. Gold goes in at any gate except heaven’s. 9. One swallow (bird) doesn’t make a summer. 10. The wife is the key to the house. 11. Riches serve a wise man but command a fool. 12. Don’t cast pearls before swine (pigs). 78

PROVERBS TEST II DIRECTIONS« (see PROVERBS TEST I)

1. He who stumbles twice over one stone deserves to break his shins. 2. Don’t judge a book by its cover. 3. The proof of the pudding is in the eating. 4. One may ride a free horse to death. 5. A rolling stone gathers no moss. 6. Strike while the iron is hot. 7. All is not gold that glitters. 8. Quickly come, quickly go (easy come, easy go). 9. All’s well that ends well. 10. Let sleeping dogs lie. 11. Great bodies move slowly. 12. Better be happy than wise. 79

PROVERBS TEST III DIRECTIONS» (see PROVERBS TEST I)

1. The sun shines upon all alike. 2. The grass is always greener in the other fellow’s yard. 3. Words cut more than swords. 4. One today is worth two tomorrows. 5. A drowning man will clutch at a straw. 6. Too many cooks spoil the broth. 7. The worst spoke in a cart breaks first. 8. Speech is the picture of the mind. 9. It never rains but it pours. 10. Don’t throw good money after bad. 11. There’s many a slip twixt (between) the cup and the lip. 12. The more cost, the more honor. 80

TRAINING PROVERBS The following list of proverbs was used far training during the experimental sessions. The first nine proverbs were taken from Ben­ jamin (1944) while the other 14 were drawn from Kaufman’s (I960) test.

1. Don’t cry over split milk. 2. The burnt child dreads the fire. 3. Don’t cross your bridges till you come to them. 4. Discretion is the better part of valor. 5. To fiddle while Rome burns. 6. Don’t count your chickens until they’re hatched. 7. He who laughs last, laughs best. 8. New brooms sweep clean. 9. He travels swiftest who travels alone. 10. What is sauce for the goose is sauce for the gander. 11. Birds of a feather flock together. 12. A bird in the hand is worth two in the bush. 13. Still waters run deep. 14. Don’t burn your candle at both ends. 15. If the shoe fits, wear it. 16. Blood is thicker than water. 17. Don’t put the cart before the horse. 18. One rotten apple spoils the barrel. 19. Don’t put all your eggs into one basket. 20. A bad penney always returns. 21. Mighty oaks from little acorns grow. 81

22. ’Tis the last straw that breaks the camel’s back. 23. There are many ways of skinning a cat. 82

WORD ASSOCIATION TEST The word lists for the parallel forms of this measure can be located on the following three pages. The instructions below were used for each form of the test. INSTRUCTIONS» On the following pages, you will find a list of words. You are to write next to each word the first ward that it makes you think of. It doesn’t make any difference what word you write as long as the word on the paper makes you think of it. There are no right or wrong answers. For example, suppose that Coat is the word you see. It will look just like this in the list with a space beside it for you to write the first word that you think of. Your job will be to write down the first word that Coat makes you think of in the space next to Coat. For exam­ ple, Coat might make you think of Hat or Man or Wear or Brown or Big or Warm or you might think ofs ome other ward. Whatever the first word is that you think of write it down in the space next to Coat. You are to do the same thing for each word in the list. Be sure to write the word clearly but don’t worry if you are not sure how to spell the word. Spell it as best you can. We are not interested in spellings we are just interested in the word you think of each time. When you finish a page, go right on to the next one. Write only one word on each line but do not skip any words. Remember, be sure to write (or print) clearly. Be sure to write the first word that you think of. And be sure to write just one word for every word on the list. 83

FCRM I Salty Sweet Eating Dark Bey Whistle Stem Cry Deep White River Get Yellow Comfort Light Come Spider Always Trouble Bath See Cabbage Fingers Priest Child Soldier They Slow Stand On Foot Live Sour Lift Carry Bible High Lion Religion Then Haz*d Head Afraid Short Square Shoes Butterfly Sit Hungry Younger 84

FCRM n Dream Find Blossom Go Stomach Doctor Salt long Children Bitter Street Girl Window Quickly Quiet Earth Dogs Health Hammer Quietly Bread Thirsty Cars Jump Citizen Smooth Running Man Memory Rough Carpet Playing Tell Loud Speak Hotter Over Appear People Woman Sheep Anger Moon City Scissors Beautiful Black House Slowly Whiskey 85

FORM HI Sickness Buying Chair Justice Eagle Thinner Faster Butter Where Hardly Music Lamp Fruit Cottage Hand Stove Command Blue Needle Only Numbers Easier Swift Guns Soft Kittens Closer Baby Tobacco Make Working Bed Mountain Cold King Thief Cheese Sleep Table To Sell Ocean Joy Heavy Green Doors Red Broader Here Wish 86

HIT The HIT cards listed below delineate the three parallel forms of the test that were generated for the present study. All the cards were taken from Form A of the original test.

FORM I FORM II FORM HI 37 27 35 41 10 6 9 12 14 17 16 28 29 39 21 22 45 32 25 1 20 7 26 15 31 4 23 11 13 3 &7

APPENDIX II. BECKER’S SCORING SYSTEMS 88

PROVERBS For proverbs, Becker’s (1956) scoring system will be presented in three parts. Below, Becker’s definition for each scoring category is reproduced. Secondly, examples of each level are given for proverbs that were used on both Gorham’s (1956) test and either Becker’s or Kaufman’s (I960) test (i.e., proverbs used as dependent measures in the current study). Finally, the same format will be used to present exam­ ples of each interpretation level for the 23 training proverbs. All examples have been taken directly from either Becker or Kaufman (I960).

Abstract Hit A correct generalized interpretation without detracting elements (Weight of 6). Abstract II« A correct example with reference to human behavior} an­ other proverb meaning the same thing; a response partly generalized, partly restricted to a specific example; a lower level generalization (Weight of 5). Abstract 1» A response tinged with the literal; a response which would be acceptable at Abstract U but far some minor inaccuracy, overstate­ ment, or alternative explanation which is false (Weight of 4). Vague Response» An attempt ¡at interpretation which is on the right«? track, but is left too vague to be adequate, or fails to account for part of the proverb (Weight of 3). False Interpretation« The interpretation is very inaccurate, yet an attempt was made to interpret. The error is usually due to faulty desymbolization or faulty generalization (Weight of 2). General Literal« The interpretation is literal in effect, though stated 89

in general terms. At first glance these responses do not appear to be literal, but they can only be understood as stemming from a literal in­ terpretation (Weight of 2). Literal» The proverb is interpreted literally (Weight of 2). Absurd t‘ The response indicates a failure to interpret axA/ar is logical ly absurd in terms of the task at hand. These responses are usually based on concrete associations to some aspects of the stimulus (Weight of 1). Literal-Abstract» Response which gives both a literal and an abstract interpretation. The tendency to be drawn into a literal interpretation is strong, but the subject is able to counteract it (Weight at best abstract level reaohed). 90

Examples of Becker’s Scoring Categories Using a Subset of Gorham’s Proverbs (e.g., 1-3= Form I, no. 3). 1. When the cat’s away the mice will play. (1-3) Abstract Ills When people are not under supervision they are apt to do just about as they please. When authority is out of the way, people are inclined to be mis­ chievous. Abstract IIi When the parents are gone, the children will raise cain. If there isn’t someone to control the situation you get into mis­ chief. Abstract It When the daddy or mother are away, that’s when the children can play (literal tinge). If the wife’s away the husband will play (literal tinge and in­ correct élément in that the relationship is not a subordinate one). A woman when her husband goes away, she will perhaps commit adult­ ery (overly specific interpretation). Vague Response» If you ain’t watching something, something is liable to happen. Like someone trying to get away with something. Literal» I guess the mice came out. Go after the cheese and other stuff. Absurd» When you’re away the cats must play (absurd over-generaliza tion). Cheese is spread in the right places, the horses will drink the water and the mice will eat the fleas. 91

2. The used key is always bright. (1-7) Abstract III; Activity promotes freshness. Abstract Ht The man who is active is usually in good health. Abstract 1« If you offer yourself to someone else, you don’t get old... like in friendship. Something that’s used is better than something that’s not used. Vague Response« Experience is the key to success. False Interpretation« The thinking man always has the final say. Literal« It reflects and it’s used in opening doors. The used key is always fresh. Absurd« Carry it around in your pocket. Doctor has a key for me to do something, where another person would not have that key. 3. Don’t cast pearls before swine (pigs). (1-12) Abstract Hit Consider the receptivity of the person before making an overture. Only say something to someone who’ll understand what you mean. Abstract Ht Don’t offer something that’s too good for a person who isn’t worth it. Vague Response« Don’t be too quick to judge. False Interpretationt Don’t throw your money away, you may need it later on. Don’t try to tempt people. Literal« Pearls are valuable... the swine don’t want them. Absurd« Holds meaning to whoever receives an answer or something. Does it mean that one is a male and one is a female? 4. The proof of the pudding is in the eating. (H-3) Abstract Hit Don’t go by appearances. You must experience something to know about, it. When you try something, then you know if it’s all right. Abstract II» (see general criteria) Abstract It If it’s good, it’s good. The value of anything is in if it serves its intended purpose. False Interpretation» The finished product is proven by its appre­ ciation. Experience is the best teacher. You ain’t sure of what you got until you get it. Anything that would be good, would be worth testing. General Literal» You don’t know how good something tastes until you try it. It might look like a beautiful cake and yet not taste good. You have to eat something before you can tell if you like it or not Literal» If the pudding is good you’ll eat it. The only way you can prove it’s good is by eating it. The pudding could look very good, but if it were wrongly cooked, wrongly mixed, you might get a stomach ache. Absurd» Man's taste is man's likes and dislikes. The taste. (Q) How you make it. Literal- abstract» The flavor counts more than the looks or appear' ance. Not right. Ifeans the best doesn't always show on the surface. 5. A rolling stone gathers no moss. (H-5) 93

Abstract III» If a person doesn’t settle down and have roots and establish himself, he will never profit. Abstract II» If you’re always travelling around the country, you don’t have a home, don’t have friends. Abstract 1« Moving from one place to another, you don’t gain any­ thing by it. If you keep on moving, you don’t accumulate anything by it. False Interpretation» Speed is the best policy. Idleness breeds idleness. literals If it gathers moss it can’t roll. A stone can’t gather moss. It's never still long enough to accumulate. Absurd« Either roll a rock and wouldn’t gather any moss or roll a Swede and don't get any moss on him. 6. All is not gold that glitters. (H-?) Abstract III» Everything that seems good may not be as good inside as it seems on the surface. All that looks well need not be so. Abstract Ut Just because something's beautiful doesn't mean it's good. Abstract I, Anything that glows isn't necessarily valuable. All that sparkles isn't really what it’s made up to be. Vague Response» Everything that looks good isn't. False Interpretationt Means maybe mud doesn't glitter, but from it you can get gold. Absurd» Has no value to gold, like it says. 94

7. Let sleeping dogs lie. (11-10) Abstract III; Don’t look for trouble. Don’t invite troubles upon yourself which haven't invited them­ selves. Abstract Hi It’s better not to bring up something in the past be­ cause it might cause hard feelings or make you upset. Abstract I: Anything not worthwhile, you might as well leave it be. Vague Response« Stay out of other people’s business. False Interpretation« You can never convert a crook, let him be. Don’t feel sorry for lazy people. literal« Dogs that sleep, let them be... they have sleeping times. Absurd« Means a person is attending a store. 8. It never rains but it pours. (IH-9) Abstract III« Calamities occur in series. When your luck starts running bad, it never seems to stop. Abstract Ht Things happen all at one time. Abstract Is A small start of trouble ends up in a large downpour. Vague Response« Referring to your troubles, your sorrows. False Interpretation« Good things happen all at once. Literal» Rain’s real hard. When it does rain, it doesn’t pour. Absurd« Stormy distance... evaluation of products of different types. 95

Examples of Becker’s Scoring Categories for the Training Proverbs. 1. Don’t cry over split milk. Abstract III» Brooding over past mistakes is futile. Abstract 11» What’s done is done. Abstract 1» Don’t cry over something that has gone wrong, can’t be helped. Vague Response» Forget the past. False Interpretation» Don't let defeat stop you. Literal» Don’t be sad over split milk. Absurd» (Due to the obvious nature of this scoring category, further examples will not be given) 2. The burnt child dreads the fire. Abstract III» Experience is a good teacher. Abstract Hi If you get hurt you’re careful. Abstract Is If you’ve been scorched, you don’t usually go back for more trouble. Vague» The child would be aware. False» If you play with matches you’re going to get burned. General literal» If you were burnt once, they’re afraid of the fire. Literal» You get scared if you get burned, 3. Don’t cross your bridges till you come to them. Abstract III» Don't worry about the future until it’s necessary. Abstract II» Solve your problems when you come to them.

Abstract It Don't wottry about things ahead of time. Vague» Don’t worry about things ahead. False» Don't make decisions alone. 96

Literal« Don’t go on bridges if you can help it. 4. Discretion is the better part of valor. Abstract HI« Think before you act. Abstract Ut Thinking for yourself is the better part of bravery. Abstract 1« Caution is the better part of bravery (less adequate). Vague« Think before you say anything. False« A wise choice is the way to be brave. literal« Not used with this proverb- score as False. 5. To fiddle while Rome burns. Abstract HI« Not do anything about something important. Abstract Hi You play when someone else is in trouble. Abstract 1» To be playing around while someone else does your work. Vague« To play, fooling around. False« It’ s time to do something in a hurry. Literal« To amuse oneself while something is on fire. 6. Don’t count your chickens until they’re hatched. Abstract HI» Don’t be too sure of anything until it happens. Abstract II» Don’t count your money before you get it. Abstract 1» Don’t count on what you hope is going tb be. Vague« Don’t count on anything. literal» Don’t count the chickens if you haven’t got them. Literal- abstract» About the same as don’t cross bridge before you come to it. You can’t count chickens before hatched cause you don’t know how many are going to hatch. 7. He who laughs last, laughs best. Abstract lilt He who comes out on top in the end is the true victor 97

Abstract lit The person who’s patient may win out in the long run. Abstract 1« It’s the end result that really counts. Vague» I guess somebody thinks he’s smart, he’s liable to be the one that’s fooled. False» The one who laughs last gets the joke best. 8. New brooms sweep clean. Abstract III» Something new works better than something old. Abstract II» A new instrument works better. Abstract 1» New interests will make the past fade. Vague» Get rid of something old. Falset The new is best. Literal» ’Cause the new ones have new material. 9. He travels swiftest who travels alone. Abstract Hit The person alone has no burdens to hold him back. Abstract II» He who ventures forth on his own, progresses the fast­ est. Abstract Ii A person goes faster if he is by himself. Vague» Not to be too dependent on other people. False» He who walks fast travels alone. 10. What is sauce for the goose is sauce for the gander. Abstract Hit What’s applicable to one person is equally applica­ ble to another... it serves each equally well. Abstract Hi What’s good for one is good for another. Abstract 1» If one person likes a kind of food, the other will, too Vague» It’s all the same. False» Both the male and female should have equal rights. 98

Literal» What’d be sauce for you would be sauce for/'me. 11. Birds of a feather flock together. Abstract HI» People who like the same things tend to associate with one another. Abstract II» Similar people stay together. Abstract 1» Five or ten whiskey drinkers or gamblers stay together. Vaguet Two of a kind stay together. False» You’re better off living with your own kind of people. Literal» Sparrows stay with sparrows and robins with robins. 12. A bird in the hand is worth two in the bush. Abstract Hit Achievement is more valuable than the mere prospect of that achievement. Abstract lit If you already have something it’s better than having to hunt far something. Abstract Is If you have something in your hand, it’s worth more than something that’s not in your hand. Vague» Don’t expect more than you have. False t A dollar saved is worth more than two spent unwisely. Literal» What could it mean unless you’re catching birds. 13. Still waters run deep. Abstract Hit Generally, a person who is quiet may have a lot of good or bad characteristics which you don’t recognize or see at first Abstract lit A person who is quiet might have the greater ability. Vague t There are lots of undercurrents. False» We can never really escape ourselves. Literal» Deep waters run slow. 99

14. Don’t burn your candle at both ends. Abstract lilt Don’t use up your energy too rapidly. Abstract Hi Don’t try to do two things at once. Abstract It Take it easy, don’t rush life so much. Vague« Use your head. literal« You can light a candle only at one end. 15. If the shoe fits, wear it. Abstract III« If something is so and applies to your situation or to you, accept it. Abstract Ut If you are told something about yourself that fits, take it. Abstract 1« If a name fits, wear it. Vague« If it belongs to you, keep it. False« Try something out before you buy it. Literal« If it fits, wear it. 16. Blood is thicker than water. Abstract HI« Your family is more important to you than friends. Abstract II« You should think more of your family than your friends Abstract 1« Blood relatives are closer than other kinds. Vague» If you have a relation, he’s a better deal. False« The good people can always outsmart the bad. Literal« Well, it is thicker than water. 17. Don’t put the cart before the horse. Abstract HI« Do the job right, not backwards. Abstract II« First things first. Abstract 1» Don’t accept a position until you know you can do it. 100

Vague» People should put things in their natural order. False» Don’t put something in a situation before the finals of it. Literal» Don’t have the horse push the cart. 18. One rotten apple spoils the barrel. Abstract HI» If there’s one bad person in a group he’ll make every­ one else bad. Abstract lit If a group of people has one in the crowd that’s ig­ norant or bad, if he stays, the rest will develop ideas like his. Abstract It Something bad in a bunch of good can spoil all the good. Vague» One person can influence allkthe people a lot. False» One stupid guy can spoil it for the others. 19. Don’t put all your eggs into one basket. Abstract lilt Diversify your prospects. Abstract II» Spread your talents and friendship around. Abstract 1» Don’t plan on too many things to come good or true at one time. Vague» Don’t depend too much on one problem. False» Don’t deliberately mix all the races together. Literal» All the eggs will not fit in one basket. 20. A bad penny always returns. Abstract HI» Our deceptions deceive only ourselves. Something you do that’s wrong always hangs onto you or comes back. Abstract Ht Bad people are always coming back. Abstract 1» Someone who’s a bad egg always returns to where they became bad. Vague» You can’t always hide mistakes. 101

False» A thief always returns. Literal» It does because it’s no good. 21. Mighty oaks from little acorns grow. Abstract III» Somebody great usually starts from something small. Abstract II» No matter how little a person may be, he eventually may become someone important. Abstract 1» Everything has to start small. Vague» Humble beginners sometimes get the best jobs. Falset It takes a part to make a whole. Literal» Has to do with the birth of trees. 22. ’Tis the last straw that breaks the camel’s back. Abstract III» An accumulation of minor setbacks may cause a major disaster. Abstract II» Too much of a strain or load causes things to go wrong Abstract Is Some people think they can’t take any more. Vague» A person can take so much and no more. False» It’s all that’s needed to finish. Literal» The camel is abused by overwork. 23. There are many ways of skinning a cat. Abstract HI» There is more than one right answer. Abstract II» There are many ways of doing something. Abstract Is There are many ways of doing things you really want to do. Vague» Do the job to the best of your ability. False» A person can do things he plans out ahead. Literal» You can skin ’em more than one way. 102

HIT The major components of Becker’s (1956) rating system for inkblot perceptions are presented here verbatim» Level one is characterized by diffuse, global, undifferentiated per­ ceptions. Wa is a direct indicator of the diffuse, global nature of perception at this level. W- indicates the lack of differentiation of inner and outer worlds and the syncretism in mental organization. DW is a clear example of a type of diffuse perception (that) has been des­ cribed by the phrase ’’pars pro toto”- any part has the quality of the whole. The contaminated and fabulized responses reflect both the con­ crete and syncretic nature of primitive perception. In these responses there is an absurd fusion of percepts on the basis of spatial identity and spatial contiguity. Perseveration is assumed to reflect the dyna­ mic rigidity of the personality at this level. Empirically, (it has been found that) these perceptual classifications (are) most character­ istic of ... severely regressed schizophrenics. Level two is characterized by attempted differentiation in which the diffuse and syncretic nature of perception is still apparent. (It has been found that) Da. DdD. and Dv responses (are) relatively rare in children... . In psychopathology, however, one might expect to find a greater frequency of this combination of discreteness with diffusion and syncretism where there is regression from higher levels of percep­ tual differentiation. A similar explanation is offered to account for the differences in relative frequencies of the Wv and Dv responses. The greater frequency of vague responses in the schizophrenic group may be the result of regressive diffuseness and interfering 103

with percepts which might have been Wm responses. Because Wy shows some integrative effort with consideration of the formal aspects of the blot it is placed at level three. Dy lacks integrative effort and is therefore placed at level two. D- and Dd- again are indicative of an unsuccessful attempt at differentiated perception. Level three is characterized by the achievement of fair differentia tion with only rudimentary integrative efforts. In some ways, Adx-Hdx responses are like the Dm of the next level. However, they are placed at this lower level because they clearly indicate a failure at inte­ gration where Integration is usually easily achieved. Dd+ was placed at level three, while the more immature Dd- was placed at level two. Level four is a stage of accurate differentiation with the ability to make simple integrations. The Mediocre responses scored at this level indicate the ability to meet certain constant typical require­ ments in form necessary for adult perception. level five is indicative of clearly integrative activity with the ability to subordinate differentiated parts to the whole. Level six is characterized by the highest form of differentiation and hierarchic integration which is found only in mature perception. The score given to each HIT response ranges from 1 to 6, corres­ ponding to the perceptual level of each response. The following chart presents more specific definitions for each perceptual level. 104

Definitions of Becker’s Perceptual Scoring Categories

Level Classification Definition 1 Amorphous Shape plays no determinable role. Whole (Wa) Minus Whole Content requires specific form not provided (W-) by blot. * Confabulato^ A single detail is basis for interpretation Response (DW, of the whole. Contaminated Fusing of two interpretations of the same Response blot area. (Con R) Fabulized Com­ Absurd combination on basis of spatial con­ bination tiguity. (Fab C) Perseveration Same content to three or more cards with (Per) little regard to form requirements. 2 Amorphous Analogous to Wa. Detail (Da) Confabulatary Analogous to DW. Detail (DdD) Minus Detail Analogous to W-. (D-) Vague Detail Farm element is so unspecific that almost any (Dv) blot area could encompass content. Minus Unusual Analogous to W-. * Detail (M-) 3 Vague Whole Analogous to Dv. (W) Oligophrenic Response to part of an A or H percept usually Detail seen as a completed figure. (Adx-Hdx) Plus Unusual Content is a reasonable match to blot area iso­ Detail (Dd+) lated. 105

Level Classification Definition 4 Mediocre Detail Form implied in outline and articulation matches (to) blot area. At level of "populars". Mediocre Whole Analogous to to. but applies only to unbroken (Wm) blots. 5 Plus Detail Two or more D areas are combined into one (to) "good form" percept. * Plus Whole All D portions of a broken blot are combined (W+) into one "good form" percept. * 6 Plus-Plus An unbroken blot is perceptually articulated Whole (W++) and reintegrated into a "good form" percept. * Plus-Plus A D area is articulated and reintegrated into Detail (to+) a "good form" percept. * * Norms reported by Holtzman et. al. (1961) were used as a scoring guide. 106

APPENDIX III. E GUIDE FOR PROVERB INTERPRETATION INSTRUCTIONS 107

The training Es used the following dialog as a guide when adminis­ tering the instructions during the proverb training sessions. ”A proverb is a statement such as ’When the cat’s away, the mice will play* or ’Don’t cry over split milk’ or ’Don’t cross your bridges until you come to them’ and so on. These and similar statements are called proverbs because they have a more general meaning than what they say directly. The more general meaning of a proverb can be a simple statement of fact or it may be giving you some advice. A proverb has a more general meaning because the major words in the proverb stand for something else- in other words, they are symbols. Mary times, the major words represent (stand for) types of human beings. In all proverbs, however, you can be sure that the major words can be replaced with more general words or ideas to obtain the proverb’s more general meaning. In some eases, you may not know the mare general meaning of a pro­ verb right away. So I want to explain how to figure out the more gen­ eral meaning of a proverb. To interpret (figure out) what the proverb means, you must do FOUR things, l) You must decide which words in the proverb are the major words. 2) Once you find the major words, you must figure out the relationship among them, taking the major words at their face value. By doing this, the next step will be a little less diffi­ cult. 3) The third step involves making up a number of statements which are similar to the original proverb, using different words and situations 4) Finally, find a general word for each symbol (major word) in the pro­ verb that will include all the statements you have made. Maybe this will be easierito understand by giving an example. Let’s take the proverb ’When the cat's away, the mice will play’. What do 108 you think this proverb means?" If the subject gives an adequate interpretation, says "Good. Now let’s see how you could have figured that out if you hadn’t known its meaning". If the subject says he doesn’t know or if he gives an in­ adequate Interpretation, say» "Good. Now, let’s see if we can figure out an even better interpretation". In either case, proceed as follows, a) Paraphrase the definition of a proverb, b) Repeat the four steps of proverb interpretation, c) Say: "Let’s try to interpret this proverb using the steps I’ve listed". Go through each step for the sample proverb, providing the solution if the subject has difficulty. Be sure that the subject understands your instructions before initiating the remainder of the training session. Repetition will be necessary for most subjects. 109

APPENDIX IV. DATA Ilo

TABLE 1-A Univariate Analyses of Variance. Pretest Data.

Source Variable df MS F

Method Digit Symbol 2 41.809 0.664 Proverbs 2 2.034 2.311 HIT 2 0.032 0.257 Auditory Distraction 2 18.952 0,116 Word Association 2 67.88I 1,304 Age 2 45.167 0.611 IQ 2 55.143 0.501 Socio-economic Status 2 350.594 3.199 First Hospitalization 2 52.952 0.902 Medication Level 2 1422.035 0.019 E Digit Symbol 1 0.857 0.014 Proverbs 1 0.747 0.848 HIT 1 0.200 1.622 Auditory Distraction 1 337.168 2.056 Word Association 1 2.381 0.046 Age 1 77.357 1.047 IQ 1 27.524 0.251 Socio-economic Status 1 .54,857 0.500 First Hospitalization 1 160.095 2.726 Medication Level 1 76288.000 1.036 Ill

TABLE 1-A Univariate Analyses of Variance. Pretest Data (Continued).

Source Variable df MS F

Method x E Digit Symbol 2 30.857 0.490 Proverbs 2 0,124 0.142 HIT 2 0.059 0.476 Auditory Distraction 2 41.238 O.252 Word Association 2 37.881 0.728 Age 2 23.643 0.320 IQ 2 53.238 0,486 Socio-economic Status 2 20.786 0.190 First Hospitalization 2 3.534 O.O6O Medication Level 2 96546.875 1.311 Within Cells Digit Symbol 36 62.954 Proverbs 36 0.880 HIT 36 0.123

Auditory Distraction 36 163.971 Word Association 36 52.048 Age 36 73.892 IQ 36 109.517 Socio-economic Status 36 109.606 First Hospitalization 36 58.738 Medication Level 36 73657.375 112

TABLE 2-A Univariate Analyses of Variance. Posttest Data

Source Variable df MS F

Method Digit Symbol 2 64.381 0.597 Proverbs 2 2.054 1.923 HIT 2 0.354 1.566 Auditory Distraction 2 171.714 0.858 Word Association 2 54.166 0.780 E Digit Symbol 1 6.095 0.056 Proverbs 1 0.886 0.829 HIT 1 0.034 0.152 Auditory Distraction 1 737.521 3.685 Word Association 1 1.524 0.022 Method x E Digit Symbol 2 23.238 0.215 Proverbs 2 0.597 0.559 HIT 2 0.720 3.183 Auditory Distraction 2 30.095 0.150 Word Association 2 11.310 0.163 Within Cells Digit Symbol 36 107.898 Proverbs 36 1,068 HIT 36 0.226 Auditory Distraction 36 200.147 Word Association 36 69.485 113

TABIE 3-A Univariate Analyses of Variance. Follow-up Data.

Source Variable df MS F

Method Digit Symbol 2 25.167 0.284 Proverbs 2 2.366 2.458 HIT 2 0.082 0.388 Auditory Distraction 2 410.380 3.231 Word Association 2 4.786 0.086 E Digit Symbol 1 34.381 0,387 Proverbs 1 0.009 0,009 HIT 1 0.975 4.634 * Auditory Distraction 1 192.857 1.518 Ward Association 1 8.595 0.154 Method x E Digit Symbol 2 62.167 0.700 Proverbs 2 0.271 0.281 HIT 2 0.174 0.829 Auditory Distraction 2 62.000 0.488 Word Association 2 49.310 0.885 Within Cells Digit Symbol 36 88,787 Proverbs 36 0.962 HIT 36 0.210 Auditory Distraction 36 127.028 Word Association 36 55.723 *P< .05 TABLE 4-A -d- Correlation Matrix for the 10 Pretest Variables

Socio- First Medica- Digit Auditory Word Variable Age IQ Economic Hospital­ ition Symbol Proverbs HIT Distrac­ Associa­ (1 (2 Status(3 ization^ Level(5 (6 (7 (8 tion (9 tion (10 1 1.000

2 -0.169 1.000

3 -0.045 -0.402 1.000

4 0.738 -0.238 0.016 1.000

5 0.119 -0.216 0.158 0.212 1.000

6 -0.566 0.426 -0.270 -0.453 -O.I74 1.000

7 -0.283 0.457 -O.229 -0.179 -0.042 0.482 1.000

8 0.216 -0.048 -0.044 0.236 0.410 -O.O7I 0.118 1.000

9 -0.265 0.458 -0.174 -0.228 -0.198 0.431 0.509 -0.105 1.000 10 -0.493 0.Q08 0.318 -0.413 0.278 0.239 0.268 0.100 0.039 1.000 TABIE 5-A Means and Standard Deviations for Each Group. Pretest Data.

Socio- First Medica­ Digit Auditory Word Age IQ Economic Hospital­ tion Symbol Proverbs HIT Distrac­ Associa­ Status ization Level tion tion SI-^

X 45.429 114.143 50.143 20.714 207.143 28.429 3.571 3.000 71.714 22.286 SD 7.807 6.257 8.395 6.775 228.087 4.826 1.081 0.361 9.928 7.610 SI-E-j

X 41.286 111.857 50.429 15.857 457.143 26.429 3.343 3.043 65.857 19.000 SD 7.847 15.551 12.621 7.244 332.201 12.259 1.071 0.223 15.816 8.737 Operant Ê2 _ X 45.857 108.143 50.286 16.000 389.286 25.143 2.757 2.971 71.143 20.286 SD 7.081 10.684 14.716 6.218 225.397 9.155 0.774 0.309 8.533 4.990 Operant y X 46.143 110.714 55.286 13.143 307.143 23.143 2.657 3.057 62.143 19.000 SD 9.118 11.295 8.281 9.856 235.281 5.956 0.957 0.509 8.375 8.775 Control ®2 - X 44.714 112.286 59.143 17.429 285.714 23.286 3.200 2.957 68.000 14.857 SD 9.995 9.394 7.603 7.678 186.446 6.897 1.030 0.282 18.690 5.047 Control

X 40.429 107.143 60.714 13.429 373.571 26.429 2.729 3.243 65.857 18.000 SD 9.361 6.793 9.214 7.700 372.343 6.106 0.624 0.355 11.950 7.118 \o TABIE 5“A (Continued) Means and Standard Deviations far Each Group. Posttest and Follow-up Data, Respectively

Digit Auditory Word Digit Auditory Word Symbol Proverbs HIT Distrac­ Associa­ Symbol Proverbs HIT Distrac­ Associa­ tion tion tion tion SI-Eg POSTTEST FOLLOW-UP X 25.714 3.886 2.943 78.714 19.714 28.143 3.971 2.528 77.714 22.143 SD 8.655 0.801 0.761 10.996 6.701 8.255 0.901 0.419 9.827 6.094 si-e3

X 27.429 4.043 3.228 69.571 17.428 31.714 4.043 3.071 71.143 18.714 SD 16.821 1.026 0.457 18.573 10.261 II.87O 1.193 0.298 14.369 9.759 Operant h - * X 32.571 3.957 3.143 75.857 18.286 33.714 3.228 2.671 68.143 18.571 SD 11.602 0.991 0.321 13.898 6.343 12.298 0.941 0.774 10.684 8.039 Operant h - > X 29.143 3.3OO 3.257 65.OOO 19.571 30.714 3.414 2.943 61.286 21.571 SD 6.568 0.927 0.369 8.583 10.533 5.678 1.085 O.3O5 5.851 7.871 Control “* X 29.143 3.386 3.686 69.714 15.428 27.428 3.443 2.886 65.143 17.714 SD 8.O3O 1.098 0.501 15.564 6.477 6.161 0.818 0.372 15.104 5.122 Control - ~ X 28.571 3.014 3.114 64.571 15.286 32.286 3.100 2.986 65.714 20.857 SD 6.876 I.279 0.279 15.032 8.577 10.111 0.898 0.410 9.087 7.010 117

Testing Sessions Figure 1-A. Mean performance for each group on digit symbol. 118 n o i t a t e r p r e t n I

b r e v o r P

n a e

Figure 2-A. Mean performance for each group on proverb interpretation 119

Figure 3-A. Mean performance for each group on the HIT. 120 s e pons s e R

t c e r r o C

f o

r e b m u N

n a e M

Figure 4-A. Mean performance for each group on the auditory distraction task. 121

Testing Sessions Figure 5-A. Mean performance for each group on the word association task. 122

APPENDIX V. SUBJECT ASSIGNMENT SCHEDULES 123

SUBJECT ASSIGNMENT SCHEDULE Group 1= Eg; SI training Group 4= E3f SI training Group 2= Egj Operant training Group 5= E3» Operant training Group 3= 5.2» Control group Group 6= E^f Control group

Subject number Group Assignment 1 1 2 4

3 5 4 5 5 6 6 1 7 6 8 2 9 1 10 4 11 2 12 4

13 3 14 2 15 3 16 5 17 2 18 1

19 5 20 6 124

Subject Number Group Assignment 21 5 22 4 23 5 24 3 25 5 26 6 27 4 28 6 29 3 30 1 31 4 32 1 33 2 34 2 35 6 36 1 37 2 38 4 39 6 40 3 41 3 42 3 125

OPERANT- CONTROL PAIRINGS

Operant Subject Control Subject 8 13 11 15 E 14 24 Subject 17 30 Numbers 34 44 35 46 38 47

3 5 4 7 E„ 16 20 -J Subject 19 27 Numbers 21 29 23 36 25 43