University of Southern California Dissertations and Theses

Methodological Problems In The Identification Of Suicidal Behavior By Means Of Two Personality Inventories

(USC Thesis Other)

Methodological Problems In The Identification Of Suicidal Behavior By Means Of Two Personality Inventories

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content T his d isserta tio n has been 64— 6237 m icrofilm ed exactly as received DEVRIES, Alcon G ysbertus, 1927— METHODOLOGICAL PROBLEMS IN THE IDENTI FICATION OF SUICIDAL BEHAVIOR BY MEANS OF TWO PERSONALITY INVENTORIES. U n iversity of Southern C alifornia, P h.D ., 1963 P sych ology, clin ica l University Microfilms, Inc., Ann Arbor, Michigan Copyright by ALCON GYSBERTUS DEVRIES 1964 METHODOLOGICAL PROBLEMS IN THE IDENTIFICATION OF SUICIDAL BEHAVIOR BY MEANS OF TWO PERSONALITY INVENTORIES ty Alcon Gysbertus Devries A Dissertation Presented to the FACULTY OF THE GRADUATE SCHOOL UNIVERSITY OF SOUTHERN CALIFORNIA In Partial Fulfillment of the Requirements for the Degree DOCTOR OF PHILOSOPHY (Psychology) August 1963 UNIVERSITY O F S O U T H E R N CALIFORNIA G R A D U A TE S C H O O L U N IV E R S IT Y PA R K L O S A N G E L E S 7 . C A L IF O R N IA This dissertation, written by A ^ . 9 ^ ® tu® . . P . ® . Y -T .i ® . ® . under the direction of h.^...Dissertation Com mittee, and approved by all its members, has been presented to and accepted by the Graduate School, in partial fulfillment of requirements for the degree of D O C T O R OF P H I L O S O P H Y D m * Date t... DISSERTATION COMMITTEE Chairman The Neuropsychiatric Hospital system of the Veteran Administration and the Suicide Prevention Center at Los Angeles "If a thing exists, it exists in same amount, and if it exists in seme amount, it can be measured." E. L. Thorndike ACKNOWLEDGEMENTS It Is with a sense of deep gratitude that the writer acknowledg es the cooperation of the many persons who made the completion of this Investigation possible. Among all those who have been involved with this study, one per son in particular stands out. To Dr. Norman L. Farbercrw a debt is owed that cannot be repaid. It was through his counsel, encouragement and guidance that this research came about. This study emerged as a nat ural outgrowth of the work that he and Dr. Edwin S. Shneidman have done in the area of suicide. The many years the writer had the privi lege to be associated with them, professionally and personally, have been an enriching and invaluable experience. Special thanks are due to Dr. J. Paul Guilford, the writer’s chairman, advisor, and teacher. Dr. Flqyd L. Kuch, who so kindly served on the doctoral committee, offered encouragement with his coun sel and support. The author acknowledges his debt to Dr. Harry M. Grayson, chief, psychology department, Brentwood V.A.H., to Dr. John H. McCormick, chief, psychology department, Sepulveda V.A.H., who made possible the special testing in their departments; to Dr. Charles D. McCarthy and Dr. Matthew W. Buttiglieri of the Sepulveda V.A.H. and to Mrs. Opaline Jackson and Miss Celia Bloom of the Brentwood V.A.H., who assisted in the testing; and the other hospital staff members and volunteers who gave freely of their cooperation and time. The analysis of the data would not have been possible without the generous assistance of several computer centers. The Honeywell computer center at the University of Southern California made their equipment available for much of the preliminary work. Mr. Rugby Mathur of this center was kind enough to write a special Honeywell computer program for use with these data. The Western Data Processing Center was most Involved in the ana lysis of the data. It was through the kind cooperation of Mr. William Anderson, the chief of user services, and Mr. Gall Montgomery, the chief processor of the input and output data, that the analysis was efficiently managed. It was especially through the assistance of the telecomputer center of the Graduate School of Business Administra tion, University of Southern California, that the analysis was ef fected quickly. Special recognition is due to Mr. Tom Dudley of that center, who spent long hours transferring the data from cards to tape and printing much of the output. Many others were In one way or other associated with this project. Of those, special thanks are due to Mrs. Helene Hoffman, and Miss Kim Wallin. TABLE OF CONTENTS CHAPTER PAGE I. THE PROBLEM........................................... 1 A. Purpose ..... .................................. 1 B. Methodological Problems............................. 1 1. Normative versus Clinical Approach............. 2 2. Replacement versus Non-Replacement Sampling ... 5 3. Chance Expectancy............................. 8 * 4 - . Profile versus Item Analysis......................12 5. Control Variables...............................1^ 6. The Definition of Suicide....................... IT II. REVIEW OF THE LITERATURE...................................22 HI. METHODS AND PROCEDURES.................................... 35 A. The Minnesota Multlphasic Personality Inventory .... 35 B. The Development of a Suicide Potential Personality Inventory..........................................37 1. The Critical Incident Technique................. 37 C. The Population...................................... 50 D. The JWFX Analysis.................................... 5^ 1. Suicide Variable............. 5^ 2. Control Variables...............................61 3. Chance Significances.............................66 E. The SFFI Analysis.................................... 68 vill CHAPTER PAGE iv. results................................................. 72 A. Chance Findings In Sampling Without Replacement .... 72 B. Chance Findings In Sampling With Replacement...........Jk C. Analysis of the Diagnosis, Religion, Education, Occu pation, Marital Status, Age, and Number of Admissions Variables .................. j6 D. Analysis of the Suicide Variable.................... 80 E. Significant Comparisons for M1PI Items..............81 * F. SFPI Analysis....................................... 85 V. DISCUSSION...............................................103 A. No Replacement Sampling......................... 103 B. Sampling With Replacement..........................105 C. The * M P I .......................................... 106 1. Diagnosis ........................... 107 2. Religion...................................... 108 3. Education .............. ......... 108 If. Occupation.................................... 109 5. Marital Status.............. 110 6. Age .................................112 J. Number of Admissions........................... 113 8. General Conclusion............................. UA 9. Suicidal Behavior............................. 115 D. The SPFI..........................................119 VI. SUMMARY AND CONCLUSIONS...................................12^ ix CHAPTER PAGE REFERENCES....................................................130 APPENDIX......................................................Ikl LIST OF TABLES TABLE PAGE I. Classification of Types of Overtly Suicidal Patients ... 20 H. Causes of Suicide Obtained from the Suicide Literature by Means of the Critical Incident Technique..........^0 HI. Critical Behaviors by Frequency of Causes............. ^7 IV. Number of Causes Selected from Each Critical Behavior Area ^9 V. Suicide Potential Personality Inventory Items ........... 51 VI,1 Information Obtained for Each Subject................. 53 VII. Comparisons of Groups in MMPI Analysis............... 59 VHI. Subcategories of Variables Analyzed In the JMPI Analysis . ^3 IX. Comparison of Non-Suicidal Variables Analyzed in the MMPI . ^ X. Comparison of "No Control" Samples with "Complete Con trol" Samples......................................^7 XI. Number of Comparisons for the Establishment of Chance Significances for Samples of Different Size. Without • Replacement Sampling...............................69 XH. Number of Comparisons for the Establishment of Chance Significances for Samples of Different Size. With Replacement Sampling...............................7° XU I. Number of Significant Items at the .05, the .01 and the .001 Levels of Significance for Samples of Dif ferent Size Obtained through Sampling Without Re placement ..............................................73 xi TABUS PAGE XIV. Number of Significant Items at the .05, the .01 and the .001 Levels of Significance for Samples of Different Size Obtained through Sampling With Replacement .... 75 XV. Expected and Obtained Number of Significant MMPI Items for Comparisons of Different Groups at the .05, the .01 and the .001 Levels of Significance................... 77 XVI. Expected and Obtained Number of Significant MMPI Items for Comparisons of Different Suicidal and Non-Suicldal Groups at the .05, the .01 and the .001 Levels of Significance XVII. Code for Comparisons Shown in Table 1 8 .................. 1^1 XVUI. Table of fttfPl Item Numbers with those Comparisons which Were Found to be Significant for that Item .... 1^3 XIX. List of MtPI Items and their Obtained Chi Square Values which Attained the .05, the .01 and the .001 Levels of Significance ......................................86 XX. Significant Items at the .05, the .01 and the .001 Levels of Significance for "No Variable Control'1— NPC Group (n=13) versus "No Variable Control"— Combined S Group (n=»8).................. 92 XXI. Significant Items at the .05, the .01 and the .001 Levels of Significance for "Complete Variable Control" — NPC Group (n=13) versus "Complete Variable Control"— Com bined S Group (n=8)....................................93 XXII. Significant Items at the .05, the .01 and the .001 levels xli TABLE PAGE of Significance for "Complete Variable Control" ^IPC Group (n=13) versus "No Variable Control" — NFC Group (n=13)............................................... 9^ XXHI. Significant Items at the .05, the .01 and the .001 Levels of Significance for "No Variable Control"— NPC Group (n=13) versus "Complete Variable Control"--Combined S Group (n=8)..........................................95 XXIV. Significant Items at the .05, the .01 and the .001 Levels of Significance for "Complete Variable Control"— Com bined S Group (n=8) versus "No Variable Control"— Combined S Group (n=*8)................................ 96 XXV. Significant Items at the .05, the .01 and the .001 Levels of Significance for "Complete Variable Control"— NPC Group (n=13) versus "No Variable Control"— Combined S Group (n=8)..........................................97 XXVI. Expected and Obtained Number of Significant SFPI Items for Comparisons of Suicidal and Non-Suicldal Variables at the .05, the .01 and the .001 Levels of Signifi cance ............................................... 98 XXVH. List of SFPI Items and their Obtained Chi Square Values which Attained the .05, the .01 and the .001 Levels of Significance. NPC non S (n=*83) versus A H S (n=130) . . 102 LISE OF FIGURES FIGURE 1. 2. PAGE Paradigm for the 2 x k MMPI Item Analyses................. 57 Paradigm for the 2 x 2 JtlPl Item Analyses................. 58 CHAPTER I THE PROBLEM Purpose The primary purpose of this investigation is to determine whether or not it is possible to differentiate neuropsychiatric non-suicidal male patients from neuropsychiatric suicidal male patients using two personality inventory measures. A second Mm is to ascertain whether or not these same measures are also able to discriminate between various types of sui cidal behavior such as committed suicide, attempted suicide, threat ened suicide, etc. A third goal is to attempt to obtain specific test items which might be used in deriving propositions about specific person ality patterns that are indicative of suicidal individuals. The fourth purpose consists of investigating certain methodo logical problems which arise from engaging in exploratory research in an area as complex as suicide. Methodological Problems Scientists are of course well aware of the fact that many things may influence the results of their research. A careful con sideration of the various steps which are planned for the investi gation of the problems under study and an awareness of the decisions to which these steps lead may prevent much difficulty In the analysis of the result s. In this study, the Influence of such problems as a normative versus a clinical approach, replacement versus non-replace ment sampling, chance findings, profile versus Item analysis, control variables, and the definition of suicide, will be considered. Normative versus nnnio.H.1 A-nnroach There are two general approaches to the investigation of personality characteristics. One school of thought believes that abnormal behavior Is an extension of normal behavior. The other school believes that normal and abnormal behavior lie on two differ ent dimensions. These two approaches reflect themselves in the field of clinical testing. In the first line of thought, tests are developed in which responses of normal Individuals are compared with those of neuropsychiatric or non-normal individuals. The Minnesota Multi- phasic Personality Inventory*(43) Is a good example of this type of application. With the second type of approach on the other hand, tests are constructed in such a manner that it allows the clinician to analyze each person in light of the person’s own test responses, viewed in a broad rather than specific theoretical framework. Projective tech niques tend to apply this type of approach. ’ ’ The Minnesota Multiphasic Personality Inventory will here after be referred to as the M4PI. The decision as to which type of approach to take in the in vestigation of suicide is of seme importance from a methodological point of view. Clinical psychologists in general are apt to think of suicidal behavior as a reflection of specific suicidal personality patterns or syndromes. The use of a normative approach involves thinking about a suicidal personality pattern as existing in all persons. Suicides would possess more of this pattern than non-suicides. This approach involves the comparison of suicidal behavior with non-suicldal be havior. Behavioral norms could be established and the degree of inclination toward suicidal behavior can then be estimated from these norms. The clinical approach disagrees with this. This latter point of view believes that suicidal behavior should be judged in its own light and that evaluations against non-sulcldal norms are apt to be disappointing and non-frultful as far as the understanding and pre diction of the phenomena is concerned. As a result of these different points of view, two types of test development approaches are possible in the establishment of a personality inventory which would categorize the suicidal personality. In the one instance, one would collect data from a large number of behavior areas which are found in normal individuals and develop a test from this information. The same test would then be given to a group of suicidal individuals. Responses of the normal persons are then contrasted with those of the suicidal individuals. When using the second approach, one would determine what out standing characteristics are indicative of suicidal persons. From this knowledge, one would then develop a test that would measure these characteristics. This latter type of application might prove to be more efficient for the identification of suicidal persons than the first so called "shotgun" "type of approach. The question has never been settled whether or not suicide is a disease entity like scarlet fever, that is either present or not present as well as time bounded, or that it is an entity that is present throughout all of life but may show only under particular circumstances. It is also questionable whether or not mere deviation from those behavior areas which are found in the normal non-suicldal person is the distinguishing charac teristic of suicidal persons. It is very possible that these devia tions from normal behavioral patterns would only classify him as a non-normal or mentally ill individual without distinguishing him as specifically suicidal. If one accepts the proposition that suicidal!ty is not common to everyone, then one would look for a suicidal pattern of behavior over and beyond behaviors found in a non-sulcldal popula tion, whether normal or mentally ill. It has became the custom among clinical psychologists to de scribe patients in terms of psychiatric classifications rather than in terms of psychologically derived personality traits. Often, psychia- trlcally derived behaviors have become associated with certain per sonality patterns. For Instance, different personality syndromes are attached to general neurotic, general psychotic and to even specific schizophrenic behavior. As a consequence of this type of thinking, one is apt to look for specific personality patterns or syndromes whenever someone talks about "suicidal" behavior. However, until it has been shown that there is a specific suicidal personality pattern, one should be aware of the fact that "suicidal" behavior may be just another method of response, that is not necessarily associated with any specific personality syndrome. It may be that suicidal behavior can be exhibited by any one possessing any type of personality pat tern. On the other hand, it may also be that persons with certain personality patterns are more likely to respond with suicidal behavior than others with different personality patterns. The resolution of this problem, is important in the scientific investigation of suicide. The application of only one approach may mean a fruitless concentra tion on a non-existing entity, or an overemphasis on external stimuli without taking an adequate account of internal personality dynamics. This study attempts to determine which of the two test development approaches might be the more productive in identifying suicidal behavior. Replacement versus Non-Replacement Sampling Frequently it is necessary to draw one or more samples from the same population for any one experiment. In many instances such samples are chosen at random from a certain defined population. It is often overlooked however, that the theory of random sampling re quires that whenever a subject has been drawn from the population, he should be returned before the next subject is chosen (22, p. 33)* With this definition of random sampling any one Individual could he selected more than once for a particular sample. It Is also possible that he could be chosen as a subject for different sample groups In the same experiment. However, because of the nature of many psycho logical experiments, it is not always feasible to select an Individual more than once. As a result, the experimenter usually purposely neg lects to return selected subjects to his population before he makes his next choice. In other words, he samples without replacement. When the population Is large and the samples drawn are small, the chances that a person will be selected more than once are rela tively small. In sampling without replacement, the sample size can of course never exceed the size of the population* In this type of sampling, therefore, there are N possible choices for the selection of the first subject, but only N - 1 for the second and N - 2 for the third, etc. This results in a probability of selecting N ( N - 1 ) ( N - 2 ) ..•••• ( N - n-^ + 1 ) different samples of size n^ from a population of size N. This Is In contrast to sampling with replacement. Here, the probability that a person Is chosen remains constant. The number of possible samples of size n^ that can be drawn with this procedure is n -i N x. In random sampling it is assumed that each sample has the same chance of being selected, regardless of whether the method of sampling has been with or without replacement. The ratio between the proba bility of selecting a sample by means of replacement sampling and the probability of selecting a sample by means of non-replacement sampling gives an indication of the discrepancy between the number of samples that can be selected using the two methods. It follows that this ratio is close to unity when N is large and n^ small. In that case It does not make much difference which type of sampling the experimenter uses ( 31, p. 29). Especially in clinical psychology, however, the population sizes available are often fairly small. Differences in method of sam pling could, therefore, make a difference in any findings. Not too much attention has been paid to this problem. In addition, it is often forgotten that the method of sampling is likely to Influence the sampling distribution when the population sizes are small,. Al though the mean of the sample means will in both sampling procedures equal the population mean, differences in method of sampling will affect the variance of the sample means. While the variance of the sample means for sampling with replacement is CT /n^, the variance of the sample means for sampling without replacement should be cal- N — n culated by C*2/nn ( 1 - nl ~ 1 ) or by (<^/nn ) ( w ^ ). J - N - 1 x N - l The type of sampling also affects the mean sample variance. Here too, a correction factor needs to be used when the sampling has been done without replacement. In this case, the correction factor which is applied is calculated by(N/N-l) . Since the researcher is often dealing with finite populations in clinical psychology, he should be aware of the fact that the type of sampling can affect his sampling distribution. It follows from the above that the average mean of sampling without replacement is smaller than the average mean of sampling with replacement. The apposite is true however, for the variance. When sampling without replacement, the average variance can he expected to he larger than the average variance of sampling with replacement. In other words, there seems to he more dispersion in those sampling distributions where the sampling has been done without replacement. This fact is of special importance when the investigator is dealing with a large number of not necessarily independent variables and he has to estab lish the number of significant outcomes which could be the result of chance. This aspect of the method of sampling and its Influence on the number of expected chance significances will be Investigated in this study. Chance Bgnftntancv* Block discusses this problem in his article "On the number of significant findings to be expected by chance." (13) In attempting to determine which particular items among a large set of items will differentiate two groups, one can expect to find a number of so- called significant items which occur purely by chance. This problem baa long been recognized. The most common solution has been to as siime that all items in the set are uncorrelated. In order to cal culate, for instance, the number of items that one might expect to find to reach chance significance at the ten per cent level of *For a more detailed discussion of this topic, the reader should refer to Block, . i960, 25# 369-380 (13) On the Number of Significant Findings to be Expected by Chance. significance, it has been customary to multiply the number of items in the set by .10, by .05 for the five per cent level of significance, and by *01 for the one per cent level of significance. However, if one makes the assumption of Independence among the items, it is often overlooked that another and possibly somewhat more appropriate solu tion would be to calculate the probability of finding at least n significant items from the binomial expansion. For many of our psychological tests, of course, the assump tion of independence between items Is not warranted. Interdependence or covariation of Items will affect the number of spuriously signi ficant items obtained. It is generally agreed that it will increase the number of chance findings when the variables used happen to be discriminating variables. The rationale is that when a variable is discriminating, all. Items which are related to this variable will also tend to be differentiating. This logic however, works also in the opposite direction. When items covary, but the variable is a nondiscriminating one, the number of significant items which will be found could be far less than the number expected by chance. Another aspect that is mentioned by Block is the fact that fewer discriminating items may be found than can be expected by chance when Independent variables are used to differentiate homo geneous groups. This is due to the fact that the dispersion of a measure decreases when homogeneous samples are used. It should not be forgotten that whenever this is the case, the reliability of that measure for those samples will decrease correspondingly. 10 There is no easy solution to these problems. In many in stances, the proposition that the obtained results are significant, when the number of discriminating items are more numerous than the number of items which one would expect to occur by chance as cal culated from independence of variables, is not always warranted. Especially when objectively developed psychological tests are used, this approach may often lead to the rejection of what may have been discriminating items or variables, or conversely the inclusion of nondiscriminating items. Block proposed an empirical solution to the problem of chance findings. He suggests that instead of combining samples according to some meaningful criterion, samples should be selected on a purely random basis and then be subjected to item differentiation proce dures. In this fashion, any item found to be discriminating, will only have reached a significance level due to chance processes. If one takes enough samples and makes enough comparisons, a stable sam pling distribution of items, significant by chance only, will be found. Such a distribution can then be used as a criterion against which discriminating items found in the analysis of samples selected by means of meaningful variables can be compared. In his article, Block gives an example of the use of his solution. Although his exact experimental procedure was not com pletely clear, It seems that pseudo-random samples were drawn from two papulations of approximately 300 subjects each. For each appli cation of his solution, two groups were compared with each other. 11 One group was drawn from one population, the other group was drawn from the other population. In six applications, twenty subjects were used in each of the two comparison groups. In one application, there were IT subjects in each of the two groups and in another application, two groups of 100 subjects each were compared. Four of the eight applications involved the use of the California Psychological Inven tory (472 items), one involved the use of the Inner Life Inventory (228 items), two involved the use of the Welsh Figure-Preference Test (400 drawings), and one application involved the use of a set of self-descriptive adjectives (200 items). Both men and women were used. Block concluded from his analysis that many fewer items emerged as significant than previous convention had presumed. Moreover, the conservative error was considerable, with factors of 2 or 3 or even 4 involved. The average mean per cent of items reaching significance at the .0 1 level is .36, for the .05 level it is 2 .0 8 and for the .10 level it is 4.37. Unfortunately, it is not clear on what basis Block* s popula tions were collected. Following his discussion of chance findings, it would appear to be a necessity that the population consist of the same type of subjects as those who would, be selected twr the samples to be drawn from that population according to seme meaningful criter ion. In order to find the number of chance findings, one would, then sample this population completely at random and compare the resulting samples. In Block’s example, it appears that his populations are 12 morip* up of a collection of -widely assorted subjects, from college men and women, to Air Force officers, to prison Inmates. It does not seem that this constitutes a meaningful population such as a general population or a college population or a neur©psychiatric hospital population. What effect, if any, this had on the number of significant findings remains speculative, but this may be an Im portant variable In the determination of the number of chance sig nificances in his study. Another Interesting finding was the fact that when groups of 17 subjects each, were ccmpared with one another, the smallest mean per cent of Items reaching chance significance was obtained. (.21 for the .0 1 level; 1 .^ 7 for the .05 level and 3*19 for the .1 0 level). These findings are not surprising since we know that In larger samples the probability of selecting more extreme cases is greater than in smaller samples. It appears then that sample size may be another variable that will affect the number of chance signi ficances. This problem too will be Investigated In this study. Profile versus Item Analysis In many clinical tests, the clinician Is more Interested in the particular response pattern that groups of items evoke than In the individual Item responses. In tests where the results are ana lyzed by the profile method, various items are grouped according to different criteria and scores are obtained for each group of items. The Investigator then arranges these sub-test scores into a pattern or prpfile. Different combinations of sub-test scores and their 13 re stilting patterns lead to different conclusions. Patterns can be analyzed In various fashions. Several types of analyses have been proposed. Some of these are the least square technique (64), the dual pattern method (67); the reversed dual pattern method (36), reductive techniques (66), difference method (78), cumulative methods (l), and analysis of variance techniques (96). Michael (71) > however, concluded after a review of the dif ferent methods that: Although Important strides have been made in the develop ment and refinement of statistical models In pattern and profile analysis, the results from empirical studies have been disap pointing. (p. 122) In some psychological tests, such as the MMPI, profile con figurations have been given specific meaning. This type of analysis has been used In conjunction with the MMPI In the Investigation of suicide by such investigators as Farberow (28) and Rosen et al (83). In exploratory research however, the use of profile analyses may not be the most efficient approach since profile analyses do not gener ally use all available information. In the above studies for In stance, only two or three of the nine clinical scales contributed significantly to the overall differentiation. In view of the fact that many items are Included in each scale, it is not improbable that certain other scales contain a few Items which would, discrim inate between the groups if they were used separately instead of in conjunction with particular scales. An Item differentiation analysis of all test Items would Immediately tell which Items are discrimi nators and which are not. Once this fact has been established, a I l l - special so-called "suicidal” scale could be developed which could be incorporated into existing profile patterns. It is this latter approach, which will be applied in this study. Control Variables - Religion, marital status, education, number of hospital admissions, age, and occupation One of the difficulties that was encountered in the previously mentioned research studies was the fact that the MMPI was originally constructed to differentiate Individuals according to psychiatric classifications. When this test is used for other purposes such as for the differentiation of individuals according to suicidal behavior, one bftfi to control the various groups for those variables that will influence test response. In this case, diagnosis should be held constant for the different groups. Were this not done, obtained results could be due to proportional differences in psychiatric classifications within the various groups. The use of control groups in research has been widely recog nized. One common procedure is to select a sample on some meaningful basis and separate it into two groups. One group is then given a particular treatment while the other group is not. This is the familiar paradigm A, B, A for the experimental group and A, -, A for the control group. In clinical psychology a conmon procedure is to select from a certain defined population a group of subjects exhibiting a certain characteristic X and another group of subjects from the same population not exhibiting that desired characteristic. These two groups are then tested in order to determine whether or 15 not these measuring Instruments will discriminate between the group exhibiting the desired characteristic X and the group that does not. It is of course very likely that other variables besides the characteristic X may influence the testing response. 'Whenever such variables are known to exist their influence can be accounted for by means of special experimental research designs. For instance, de signs such as treatment by levels, three-way analysis of variance, factorial designs and Graeco-Latin square designs have been widely used as such controls (62). However, when the existence of such variables is not known but is nevertheless suspected, a common pro cedure to eliminate any possible effects has been to match the ex perimental and control group on these variables. That is, one at tempts to select groups in such a fashion that the control variables are distributed in identical proportions within each group. Such a procedure leads of course to considerable difficulty in trying to find exact proportions when more than one or two variables are in volved. Even if it would be possible to match the groups there remains the distinct possibility that subjects who were easily matched will react somewhat differently to the treatment variable than those subjects who were difficult to match. Freedman (3*0 discusses this problem, stating that: The nature of the difference found (between pairs most easily matched and those difficult to match) Is apparently a function of something connected with characteristics matched most easily. (p. kST) His comments about the usable number of subjects which remain, after matching, from an original large sample, is of pertinence. 16 The proportion of matched may determine a system of undesigned weights which attach to the particular values of control vari ables most easily matched, (p. l j - 87) Only a few investigations of the Influence of control vari ables have been conducted. The results have been controversial. For example, Brower (l6) found that a negative correlation existed between intelligence and the Hysteria, Hypochondriasis and Psycho pathic Deviate scales of the MMPI. He concluded from the results of his study that in his sample of college students, lower Intelli gence was related to responses to the fysteria and Hypochondriasis scales. Winfield (99) 0 1 1 the other hand, found no significant corre lations between Intelligence and the Hypochondriasis, Hysteria and Psychopathic Deviate scales, nor to the Depressive scale. He did obtain a small ( .2 8 at the .01 l.s.) correlation between Intelli gence and the Masculinity-Femininity scale. The difference between his findings and those of Brower were thought to be due to differ ences In populations which the two Investigators used. In the one study, students had been used while in the other study, veterans had been used who were neuropsychiatric patients. Wexner ( 97) vho also correlated Intelligence with the 1MPI scales only obtained a significant relationship between Intelligence and the Paranoia scale. Brozek and Keys (17) Investigated the changes of personality patterns with age. They analyzed the the "L", the "F", and the Ifypochondriasis scales of the M4PI. They found that over a twenty- 17 nine year time period the frequencies of the yes-no responses in 184 items out of 566 were significant at the ten per cent level or "better. The variability of MtPI scores, upon successive testing ses sions was studied by Layton (59). He administered the M'fPT once a week for eighteen weeks to fifteen subjects. He deteimined that the intra- and inter-individual variation was a function of individual scales as well as the time sequence of the trials. The overall rank of Individuals within the groups did not change significantly, nor did individual profile patterns. It appears from these findings that although the MMPI is a relatively reliable instrument, psychiatric diagnosis is not the only variable that influences the response to this inventory. There seems to be a need for further research in this particular area. This study will investigate the influence of various so-called control variables. The Definition of Suicide It is generally understood that the term suicide refers to the act of intentionally killing oneself. We have become used to talk not only about committed suicides, but al so about threatened and attempted suicide. Many Investigators have used this classifi cation in the past as a means of comparing such groups with each other or with a nan-suicidal group. In many respects researchers have come to think of suicide in the same manner as the psychiatrist thinks about schizophrenia. The latter recognizes nTn<^ r | g the general classification of schizophrenia different types such as hebephrenic, 18 paranoid, catatonic and simple schizophrenia. In the same manner the suicide researcher has become used to thinking about the general classification of suicides as consisting of threatened, attempted and committed suicides. There is however, a question whether or not this classification Is actually meaningful. Shneicbnan, Farberow, and McEvoy have paid attention to this aspect of suicide. Shneidman (88) has derived a different taxonomy of death. He views death (including suicide) In terms of Intentional, sub-intentional and un-intentional death. Farberow ( 27) looks an the problem of suicide classification in terms of degree of lethal In tention of "to be," "not to be," and "to be or not to be." McEvcy (65) investigated the possibility of deriving an Index of Intention. It would appear that the Intention approach might Indeed be a more meaningful way of classification, provided one Is able to make appropriate measurements. In the meantime, this is not an easy task, as was demonstrated by McEvoy who found a .8 correlation between his Intention scale and measures of overt suicidal behavior. In many respects, this Is a problem of subjective versus objective indices. It seems analogous to the theoretical approaches of such learning theorists as Tolman and Skinner. Whereas Tollman Is ntuch concerned about intervening variables, and their measurement (26), Skinner only concentrates on antecedent stimuli and the con sequent response (1+7) • Both these approaches have proved to be of value. Depending on the scientist’s frame of reference, Intervening variables such as intention may play a more or less prominent 19 position in the investigation of suicide. Even if one would concentrate on overt suicidal behavior alone such as threatened, attempted and ccnanltted suicide, the picture is fax from straightforward. As has been mentioned before, several investigators have made comparisons between groups of threatened, attempted and committed suicides. They concluded that there were indications of personality differences between threatened and at tempted suicides. One can also deduce that both threatened and attempted suicides are different from committed suicides simply by the mere fact that the former axe still alive but the latter are not. If one therefore would propose, as seems to have been done, that there are three possibly different types of overt suicidal behaviors, then it appears logical that one should also consider that there are various possible combinations of these three basic types. Assuming that an emotional disturbance is concomitant with suicidal behavior, one can postulate the following types of suicidal categories: (see table l). Such a classification scheme may have merits since it may aid in making a so-called basic group of threatened suicides or attempted suicides more homogeneous. It is not unreasonable to assume that a person who makes suicidal threats at one occasion and suicidal at tempts at another, has different personality dynamics from someone who engages only in suicidal threats or only in suicidal attespts. On the other hand, one can also argue that noboty makes a suicidal attempt without first thinking about it. The question then 20 TABLE I CLASSIFICATION OF TTFES OF OVEETLI SUICIDAL PATIENTS* E.D. s emotional, disturbance - non-sulci dal T.S. a threatened suicide A. S. a attempted suicide C.S. a ccnmltted suicide ♦Double headed arrows Indicate the possibility that one can change frcm one condition to a previous one arises whether or not someone who thinks out loud, such as a suicidal threat, Is basically different from someone who thinks silently and then makes a suicidal attempt. This line of thought would also lead to the conclusion that someone who would be classified as a combina tion threat and attempted suicide is not different frcm the "pure" threat, and "pure" attenrpt. However, since some investigators have found differences between threatened and attempted suicides, and between attempted and completed suicides, the proposed classifica tion continues to be both logical and meaningful and will be used in this study. CHAPTER II REVIEW OP THE LITERATURE A still small voice spake unto me, "Thou art so full of misery, Were it not tetter not to "be?" The Two Voices, Stanza 1 Lord Alfred Tennyson The literature about suicide is of special interest* In con trast to such fields as mathematics or physics it shows that people of many different fields of interest have paid attention to this prob lem. The phenomenon has been approached from the standpoint of law, philosophy In general, ethics, morality, religion, medicine, sociol ogy, psychology, as well as prose and poetry. Writings about suicide go back as far as recorded history. The gradual change and emphasis in the various investigations, especially during the last one hundred years, gives an interesting picture of the various Influences that played a part. Suicide itself can be considered as a universal phenomenon, now as well as in early history. For instance, in early Eastern civilizations, suicide was practiced frequently. The Japanese phi losophy from that time considered suicide as an act of special vir tue, at least when society would not suffer as a result of the act. This point of view has remained until relatively recently. Among the Tartars it was the custom for the wife of a deceased warrior to hang herself* The voluntary suicide by fire by the wife of deceased Brahmins has been practiced for several centuries. Strangely enough, according to Moore (73)9 Gentoo religion did not originally approve of suicide. He states that ..•■the more rational and original tenets of oriental philosophy and religion did not countenance self murder, yet that thorough contempt of life, which early possessed the Brahmins, in opposi tion to the real principles of their faith, tended much that way, and led them very early to encourage and applaud the volun tary sacrifice of life on many occasions, (v. 1, p* 112) Although practiced widely, this type of suicide behavior was not officially approved by the Brahmins until 1773* At that time "A Code of Gentoo Laws or Orientation of the Pundits" was drawn up which stated that "it was proper for a woman after her husband1 s death to bum herself in the fire with his corpse." (73# p. 112) In Western civilization, suicide has been practiced from early times. In Scandinavia, during the period of the Vikings, it was the custom that the wife was burned together with her dead hus band. The Bible mentions such suicides as Samson, Saul, Ahltcphel, Zimri, Ablmelech and Judas Iscariot. Many of the Greek philosophers wrote about suicide. There was no general agreement among them, however, regarding its merits. Philosophers such as fybhagoras and Socrates argued against the com mission of suicide because it was an offense against the authority of God. Plotinus and Macroblus believed in the union of the physical body with a non-physical soul and disapproved of killing the body by oneself. Aristotle looked upon suicide as a crime in general but tended to condone it In cases of extreme poverty or extraordinary 2k misfortunes* The Stoics such as Cato, Seneca, Epicetus and Antoninus, on the other hand, approved of suicide* To them, man was only answerable to himself and therefore could live or not live as he desired. For the Epicureans, suicide was the only thing to do when the difficul ties of life became too great and irreversible. Although same of the philosophers argued in favor of suicide and same against It, many laws were aimed at discouraging the prac tice. For Instance, a Theban law stated that no honors could be paid to the memory of anyone who had committed suicide. A somewhat similar law existed In Athens where the hand that had committed the suicide and therefore had betrayed the body, had to be burled apart from the body. Some of the Greek philosophy that condoned suicide carried over to the Homan empire. For a period of time, the practice became so widespread that it was called the "Roman Death." In the Western world, suicide had not been extensively men tioned In the teachings of the early religions. It probably was the early Greek philosophers, especially Seneca, who first paid con siderable attention to the phenomenon. The Influence of religious thought, however, and especially Christian thought, became pronounced during the period of the first martyrs. It had become a tendency among early Christians to glorify martyrdom as the greatest expres sion of their belief. For a time, this tendency led to a purposeful seeking of martyrdom by great numbers of Christians. In many respects 25 it was an intentional act of seeking death* It was looked upon as such by St. Augustine who condemned suicide as against the laws of God. The Council of Arles in A.D. made this standpoint the official position of the Christian Church* In Western Europe, no special widespread attention seemed to have been paid to the topic until after the Renaissance and the peri od of the English philosophers* Both Hume and Donne wrote about sui cide and argued in favor of it. Robeck, a Swedish philosopher (73), went one step further* After arguing In favor of suicide and show ing, at least to his own satisfaction, that suicide was lawful, he killed himself* The attention to the phenomenon of suicide by the English philosophers started a flurry of writings, especially by the clergy. Hoffman (1*8) cited a few of the early English expositions of suicide during the 19th Century. For instance, in 1700, the rector of Saint Alban, the Rev. J. Adams, published "An Essay Concerning Self- Murther (sic), Wherein is Endeavored to Prove that It Is Unlawful according to Natural Principles, With Some Considerations upon What Is Pretended from the Said Principles*" In 1716, another clergyman published a paper on "Cato Con demned: Or, The Case and History of Self Murder, Argued and Dis played at I*rge, on the Principles of Re an on, Justice, Lew, Religion, Fortitude, Love of Ourselves and our Country, and Example" together with "A Solution of this Problem, Whether a Man of Sense, Goodness and Courage, Ever Did, or Can, Kill Himself? Particularly Calcu- 2 6 lated to Prevent It in the English Nation." The English however, were not the only ones who could discourse about the subject. In 17^7* the Rev. Simon Phillips, of Boston, pub lished a sermon on "The Sin of Suicide, Contrary to Nature. A Plain Discourse Occasioned by the late Perpetration of That Heinous Crime: Published for a Warning to Survivors; at the General Desire of the Hearers: Unto Wham It Is Dedicated." % the end of the 18th Century, the phenomenon of suicide had been discussed by mary disciplines. In the late 1800*8 and early 1900* 8, a more scientific attitude towards the investigation of the phenomenon of suicide had replaced the philosophic and religious approach. Medical practitioners had started to pay attention to the problem. The attempt at understanding this phenomenon was through a concentration on specific examples of suicidal behavior. Sullivan in 1899 (94) describes three cases of attempted suicide where the effort was made while the person was under the influence of alcohol or as he put it, "in conditions of cerebral automation." According to Sullivan in a 1900 paper (95)/ Huss had been an early proponent of the theory that alcoholics are especially prone to develop suicidal ideation. This point of view was opposed to that of Durkheim (24) who did not believe that there was any correlation between the incidence of al cohol consumption and suicide among his statistics from French re gional districts. Sullivan who had described the phenomenon of cerebral automation in cases of attempted suicide, investigated this theory ( 95). He compared the increase in alcohol consumption during 27 the period of 1842-1898 with the rate of male and female committed and attempted suicides during approximately the same period. He came to the conclusion that .•..the balance of evidence and argument leans to observers who look upon alcohol els one of several causes of suicide, Its ab solute and relative Importance differing In different places and at different epochs, (p. 262) This type of approach where theorizing was banked up by descriptive statistics was a great step forward In the application of scientific principles* It was In marked contrast to the "arm chair type" of theorizing that was done for example by Ireland In 1899 (90) who speculated on the causes of the Increase of suicide. He believed, in contrast to Durkheim, that the same motives played a part In the commission of suicide In the sane an in the insane. He advocated the belief that the Increase in the number of suicides was due to a greater strain upon the nervous system (modem times?). Due to this strain, people tended to become more Irritable and tired of living. When this occurred concomitant with a decline In religious belief, suicide was much more likely to occur. As was the case with the above two studies, many others were the direct result of Durkhelm's well known work on suicide (24). His work was probably the earliest most extensive exposition of a socio logical theory of suicide. He tried to understand as well as explain the phenomenon and paid special attention to possible causative fac tors. Another early sociological treatise on suicide was that by Morse 111 (74). He discussed suicide in terms of biological, psycho* 28 logical, ethnic as veil as cosmic Influences. Both Durkheim's and Morselli’s statistical data were later re-examined and re-evaluated by Halbwachs (4l), a student of Durkheim. Seme of the findings of Halbwachs Included the difference In both committed and attempted suicide rate In men and wanen. In addition to differences In rate, he found that the type of method used differed also in the two sexes. It was at this time period that different streams of scientific thought were being propounded. The scientific approach with its ac curate description of phenomena and Its emphasis on facts and scien tific methodology had already attracted many followers. The ap proach of the environmentalists, such as Durkheim, who looked for facts and explanations In the environment, exercised Its Influence. There was also the beginning influence of the dynamlclsts, such as Freud, who concentrated on the inner life of person for facts and explanations. Much of the published suicide literature In the early 1900- 1920 still remained of a purely descriptive nature. W. E. O’Hara (76) for Instance, described a case of an attempted suicide who, although receiving medical treatment, died of self-inflicted wounds. G. R. East (25) also, gives only a case description of a suicide attempt by a 56 year old woman. let there were variations in de scriptive approaches. Examples of another type of descriptive report other than the preceding, are those which give suicide statistics. For instance, In 1903, Benham ( 9) writes about suicides In public asylums* He gives statistics for the number of patients who commit ted suicide as well as for the proportion of male and female patients who were considered to be suicidal .ly Inclined. Hoffman (48) too, has this actuarial approach. He, however, tries to ccme up with certain conclusions. He gives suicide statistics for various cities and states In the U.S. He also gives figures for such variables as age, sex, urban, rural, time of day, season, month, method and motive. His theory was that suicide is not a specific disease but "a pheno menon of wrongful conduct arising out of a more or less distorted state of mind." He tried to show that Davenport, who believed that "suicide Is not Inherited, but only a tendency to an impulse that leads to suicide, 1 1 was right. Hoffman believes that most people are able to control this self-destructive Impulse but that those who are "feebly Inhibited" will commit suicide. He concluded that suicide runs in families whose members inherit this "feebly inhibited" ten dency. Stekel ( 9 0) has a more theoretical approach. He tries to ex plain the Increase in the number of suicides among persons between 15 and 25 years of age. He believes that the younger person is more susceptible to the commission of suicide because of his tendency to overreact with his already abundant emotional reactions. The reason or cause of suicide In the young however, is the same as in the adult. Stekel believes that suicide: ....Is a punishment that the departing one inflicts on himself. The principle of retaliation seems to play the chief role. Ho one kills himself who did not wish to kill another or who, at 30 least, does not wish the death of another. The punishment exe cuted on themselves Is at the same time also the punishment of the person supposed to be at the bottom of their sufferings. (p. 3**0) By 1920, most articles which made use of descriptive or actu arial statistics tried to draw conclusions regarding causative fac tors or tried to show correlations, between suicide and other vari ables. Lowrey ( 63) wrote an article in 1920 in which he described k6 different cases of attempted suicide and their psychiatric diagnoses. He discusses the precipitating causes of the attempts as he obtained them from the patients and their relatives. A much more extensive study was done by Anita Muhl in 1927 (75) • She investigated over 500 cases of suicide in San Diego. She came to the conclusion that for the majority of the cases she had investigated, the suicide act was the result of failure to adjust to everyday problems. Many of the cases also seemed to exhibit signs of regressive tendencies. Depres sion often seemed to be a precursor. This latter symptom was also found by Buckle (l8) in attempted suicides. With these, alcohol and ill health played a role as well In precipitating the suicidal behavior. Pollack (79) Is another author who describes cases of committed and attempted suicide and then expounds a general theory of suicide. This approach of relating fact to theory had not yet become ccranom procedure. Many studies were still of the speculative type. Gordon (39) for Instance, postulates without the aid of data that suicide is due to a lack of proper adjustment. He believes that per sons who have an adaptive type of approach to life are incapable of 31 committing suicide. It is those persons who do not possess adaptive hehavioral traits who go on to commit suicide whenever they encounter a situation to which they are unable to adjust. He believes that these persons are psychotic and he maintains that no neurotic person will commit suicide. Psychoanalysts especially are apt to follow a speculative type of approach. Instances of this can be found In such writings as those of Freud (35) , Glover ( 38) and Kilpatrick (56). Freud, who Initiated the psychoanalytic approach towards sui cide, discussed this problem in his well known paper an depression, Mfnimiry and tfeinnohnU a . . According to Freud, the tendency towards suicide in a person is due to his unconscious sadistic impulses arising from strong love and hate feelings which he has for someone else but which he directs against himself. Glover (38) expounds Freud’s theory. To him suicide Is due to a disorder of the superego as well as the ego. In order to understand the disorder one should either Investigate: • •••the process of Instinctual defuslan, the mechanism of intro- Jection and the resultant exploitation of defused aggressive tendencies, or, ..••the pre-ideal-ego formations for the libidinal fixations and ego development at more primitive levels, or, alternatively the reactivated primitive development In regression. He further states that: The regression not only affords an aggressive super-ego more energy, but there Is also a backward movement of the ego. The libido regression leads to a reactivation of oral sadistic ten- 32 dencies and the ego reactions are of the nature of an increased tendency to introject external stimuli and. take an attitude of animistic omnipotence to them. The external stimuli are attacked only after their introjection and this leads to destruction. Suicide Is an extreme overdetermination of the vicissitudes of both the ego and the Instincts. There seems to be under usual conditions an even balance between the direction of destructive Impulses against objects and the direction of them against the ego. When the regression Is of sufficient depth to produce a confusion of the distinction between the ego and the external world in the sense of objects, the possibility of suicide is present. Suicidal states reduplicate destructive mechanisms of an antithetical nature: the introjected object attacking the ego (super-ego) and the ego attacking the introjected object. (P* 89) Kilpatrick (56), another psychoanalyst, has an opposite point of view from that of Gordon. She believes that it is only the se verely neurotic person who commits suicide rather than the psychotic person. She looks upon suicide as a conflict between self-hate and neurotic pride. Suicide will occur when in the mind of the neurotic, the act will restore the pride. Mennlnger ( 68) has probably been one of the most prominent psychoanalytic writers on the topic of suicide since Freud. He feels that suicide is the result of the interplay of three forces, the wish to kill, the wish to be killed, and the wish to die. The psychoanalytic approach clearly shows the influence of those persons who are mainly concerned about the dynamics of suicide as it affects the individual. The majority of the theoretically oriented investigators however, either look for a combination of in ternal and external causes, or concentrate more on environmental fac tors. The more experimental type of approach, as we understand it 33 today, did. not show its influence until after the second world war. Kallman and Anastasio (5^, 55) did a study on eight dizygotic twins where one of the partners had committed suicide. The other partner was used as a control* They investigated whether or not differences in psychiatric, personal or social histoxy could he found which would provide a motive for the commission of suicide. They were unable however, to find any difference between the twins that could be construed as a reason for the commission of suicide by one part ner. Farberow in 1950 ( 28) Investigated response differences to various psychological tests of different types of suicidal mental hospital patients. His purpose was to determine whether or not he could find specific personality patterns which would be indicative of suicidal patients. This study is of special interest not only for its experimental approach, but also for the fact that MMPI profile patterns of the different suicidal and non-suicidal groups showed that the profile pattern of the threatened suicide group was signi ficantly different from those of the other groups. That is, differ ences showed up not only between threatened suicides and non-suicidal persons, but also between the former and the other suicidal groups. Rosen et al (83) replicated part of Farberow's study also using the MCPI. They corroborated Farberow' s finding that the pro file pattern of the threatened suicide group was the only one that was significantly different from the other groups. Simon and Hales ( 89) investigated fifty MAPI profiles of 3^ suicidal patients with various psychiatric diagnoses. They found a consistent rise in the "D" and "Pt" scales of the M4PI. Schmidt and Van Arsdol (85)* using a sociological approach, ttipH** a comparative analysis of completed and attempted suicides in Seattle for the years 19^8 and 1952. Their interest focused on com parative differences among types of suicides. They determined that differences existed in such variables as sex, age, race ratio, marl- tal status, occupation, social status and suicide methods. Because of the difficulties involved in obtaining a sufficient number of suicidal subjects for samples which would satisfy statis tical and design prerequisites, the number of experimental studies is of necessity limited. However, if the trend towards greater sci entific rigor that is evident in the suicide literature is any indi cation of things to come, we can be confident that the possibility of Increased understanding of this complex soclo-psychological phenomenon continues to be advanced. CHAPTER IH METHOD AND PROCEDURES Previously some of the differences in a normative versus a clinical approach in exploratory research were discussed. The MMPI was selected for use in this study as representing a normative testing approach. Thw M-Tnrwanfha Mnitfphaslc Personality Inventory Since its inception in 19^0 by Hathaway and McKinley, the MMPI has become one of the most widely used personality inventories. It is familiar to most psychologists and persons working in the field of mental health. A few characteristics of the MMPI are of special importance for this study. The items which mak** up the MMPI cover 26 different behavior areas. The respondent is asked to answer each question with "true", "false1 1 , or "cannot say". In the development of the test, these items were given to approximately TOO persons who were thought to be a representative sample of the general population. They were also given to a total of 800 hospital patients. This latter population was made up of carefully selected homogeneous groups of different types of hospital patients. The responses of these patients to the M1PI were then compared with those of the subjects in the normal sazqple. Scales were derived which differentiated each of the homogeneous clinical groups from the normal group. The criterion for scale construction, or item inclusion in the scale, was the agreement between scores on the scale and the diagnosis of the subjects in the homogeneous groups by the neuropsychiatric staff members. All scale scores were transformed into McCall's scores which made direct com parison of the scales possible. A total of nine clinical scales and four validity scales were originally derived. The various standard scores are usually plotted on a profile chart. The clinician common ly evaluates the pattern of the total profile rather than any individ ual score. From this pattern, he then derives a description of the subject which follows established psychiatric classifications. In this study, the MMPI will be used in the so-called "shotgun" approach of the measurement of suicide. The emphasis will center on the indi vidual item rather than on the total profile. The merits of concen tration an the Individual Items rather than concentration on a pro file in exploratory research, has been discussed previously. In addition to the MMPI, a special test was constructed with two requirements in mind. The first one was that the test should attempt to measure suicide only. The second requirement was that the construction of the test should follow a clinical type of ap proach. This would then allow the comparison of the identification of suicide by means of a normative approach, through the MAPI, and a clinical approach, through the specially constructed test. This latter test was called the Suicide Potential Personality Inventory. 3T The Development of a Suicide Potential Personality Inventory* Since it vas planned, that the SPPI was going to be adminis tered in conjunction with the MMPI, it was constructed in such a manner that it would have the same type of format, namely a self- report type inventory. Basically, there were two ways in which items for such a test could be collected. One method would be to ask professional persons working with suicidal persons to name outstanding characteristics which they thought would be indicative of suicidal behavior. Another method would be to Investigate past research and collect all refer ences to specific suicidal behavior. It was decided that this last method would be more objective and would allow for a wider coverage of types of suicidal behaviors. In order to accomplish this task, the critical incidents technique (33) was used. The Critical Incidents Technique It is assumed that experts are able to give exa^lee of what they consider to be efficient or inefficient performances in their particular specialties. Different experts in any one field can be asked to list Incidents of efficient and Inefficient performance. These incidents are then grouped into series of one hundred. Each series is then compared with the previous ones in order to determine how many new Incidents have been reported. When less than five new *The Suicide Potential Personality Inventory will hereafter be referred to as the SPPI. 38 incidents have been reported, it is thought that the range of total incidents has been tapped. Collection of additional incidents is thereupon terminated. The next step in the procedure involves analyzing the inci dents and classifying them according to similarity into a progression of more inclusive categories. These steps can be illustrated as follows: 1. Critical incidents may be reported involving "identifica tion with a dead person", "a wish to kill undesirable features within oneself", and "fear of losing one’s mind." 2. These incidents can be classified into a more inclusive critical behavior, such as, psychotic behavior. 3» This critical behavior in turn can be combined with other critical behaviors such as depressive behavior, neurotic behavior, etc., into the more inclusive critical requirement, viz., psychi atric disorders. J+. This in turn may be combined with other critical require ments such as medical disorders, into a general major behavior area such as psycho-physical health. The first goal in the development of the SPPI was to conduct a critical incident study of same of the suicide literature, until the criterion of the critical incident technique for the collection of causes was satisfied. This criterion was reached in the third set of one hundred. To accomplish this, a total of 66 studies were investigated. (2, 3, 5, 6, 7, 8, 9, IQ, 11, 12, 14, 15, 18, 19, 20, 21, 23, 25, 28, 30, 32, 37, 38, 39, 40, 42, 45, 46, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 6o, 61, 63, 69, 70, 72, 75, 76, 77, 79, 81, 82, 83, 84, 85, 86, 87, 89, 90, 91, 92, 93, 94, 95, 100, 101) These studies were selected from the bibliography of suicide In the Cry for Help ( 29). The selection was made on the basis of what studies appeared to be good representations of different -types of re search. The various causes selected are listed In table 2. After combining identical causes, the total number of 333 causes collected reduced to 224. It was desirable to have as few overlapping items as feasible in both the SPPI and M1P1. For this reason, the critical Incidents were sorted into more inclusive critical behaviors which were, as much as possible, similar to the 26 areas tinder which the MMPI items had originally been grouped. This procedure resulted in a total of 16 areas, see table 3* It was found that roughly 11 areas from the MMPI could be duplicated in the SPPI. The main difference centered around the items listed under each area. Many critical incident technique items could be considered similar although not necessarily identical to corresponding M4FI items. For practical and clinical reasons it had. been decided to de velop a SFPI that consisted of 50 items representing 50 different causes, selected from the 224. Each cause was chosen with certain conditions in mind. First, the causes selected should be duplicated as little as possible by MMPI items. Secondly, the items should be 40 TABLE II CAUSES OF SUICIDE OBTAINED FROM THE SUICIDE LITERATURE BY MEANS OF THE CRITICAL INCIDENT TECHNIQUE 1. Depressive state 32. Selfishness 2. Loneliness 33. Socially distant 3. Social Isolation 34. Fear of homosexuality 4. Impulsive 35. Tendency to seek seclusion 5. Marital friction 36. Better off dead 6. Matrimonial conflicts 37. Strong dependency demands 7. Impulsivity 38. Frequent and severe baro 8. Friendships not easily made metric pressure 9. Weak object relationships 39. Severe dejection 10. Transfer to other halls In 4o. Fear of insanity hospital 4l. Doubts and dreads pertaining 11. Mobile to the hereafter 22. Self contempt 42. Age 13- Debillltated constitution 43. Dread of losing mind 14. Isolation 44. Depression 15. Alienated froom self 45. Immigrant 16. High socio-economic status 46. Spring 17. External stresses 47. Family history of alcoholism 18. Changing night and day nurses 48. Chronic alcoholism 19. Social difficulties 49. Previous attempted suicide 20. Loneliness 50. Youth (factor in decreasing 21. Strong maternal fixation the suicide rate) 22. Social Isolation 51. Agitation 23. Finnneinl worry 52. Delusions 24. Immediate crises in accident 53. Worried 25. Picking at various parts of 54. No prospect of happiness the body 55. Psychopathic personality 2 6. Severe depression 56. Sense of guilt 2I* Restless movements 57. Relaxed vigilance by nursing 2 8. Unotional instability personnel 29. Spite feelings 58. Assaultive tendencies 30. Hate 59. Inability to ccme to terms 31. Cumulative crises In loss of with feelings of hate and property aggression 41 TABLE IX (Continued) CAUSES OF SUICIDE OBTAINED FROM THE SUICIDE LITERATURE BY MEANS OF THE CRITICAL INCIDENT TECHNIQUE 60. Nationality 85. Socially isolated 61. Individual1 s emotionality con stitutionally colored by ab 86. Recent committed suicide of intimate friend normal aggressiveness 87. Family history 62. Occupation 88. Heredity 63. Severly neurotic 89. Weak male identification 64. Neurotic 90. Threatened suicide 65. Cumulative crises in accident 91. Loss of sexual desire or 66. Fear of torture impotence 67. Insomnia 92. Social mobility 68* Business cycles 93. Inability to concentrate 69. Poor heterosexual adjustment 94. Evasion 70. Widowhood 95. Cumulative crisis in habits 71. Inaccessible agitated depres sion 96. Premises by depressed pa tient with request for 72. Self depreciation greater liberty or privacy 73. Social influences 97. Alcohol abuse 74. Insomnia with great concern about it 9 8. Failure of adjustment at some level 75. Social instability 99. Transients 76. Immediate crises in loss of 100. Surly property 1 01. Irritable attitude 77- Insecurity 102. Latent aggression 78. Overwhelming sense of suicide 103. No prospect of security purpose 104. Integration in the social 79.. Impatient group 80. Inability to form satisfactory IO5. Immature 81. relationships 106. Future living conditions Egotism 107. Neglect caused by others 82. Discouragement during convales 108. Worthlessness 83. cence 109. To gain sympathy Immediate crises in loss of job HO. Permanent emotional ties 8 i * - . Personal quarrels (factor in decreasing the suicide rate) 42 TABLE II (Continued) CAUSES OF SUICIDE OBTAINED FROM THE SUICIDE LITERATURE BY MEANS OF THE CRITICAL INCIDENT TECHNIQUE 111. Social suffering 135. Psychiatric abnormality in 1 1 2. Previous attempted suicide family history 113- Physical pain 136. Disharmony in home 114. Certain professions, e.g. 137. Repeated attempted suicide domestic service 138. Cumulative crises in loss of 115. Syphillis money 1 16. Rural occupation (factor 139. High density of population in decreasing the sui l4o. Chronic physical illness cide rate) l4l. Social isolation, e.g. living 117* Financial status alone 1 18. Fatigue 142. Fearful facial expression 119. Sudden fall in standard 143. Manic depression of living 144. Despondency 130. General hygiene 145. Environment 121. Physical illness 146. Loss of appetite 122. Residence in large town 147. Childlessness 123. Divorced state 148. Low density population (fac 124. Disease tor in decreasing the suicide 125. Deliberate planning rate) 126. Decreased menses in women 1^9. Self depreciation 127. Cumulative crises in sick 150. Bankruptcy ness 151. Decreased sexual interest (in 1 28. Severe hypochondriasis both men and women) 129. Family history of suicide 152. Age 130. Somatic inconveniences as 153. Sleep sociated with depression 154. Different methods 131. Tendency to take sedatives 155. Economic insecurity 1 3 2. Loss of appetite 156. Poverty 133. Economic crises 157. Single male 134. Cumulative crises in loss 158. Concealment of motive of Job 159. Broken homes 43 TABLE II (Continued.) CAUSES OF SUICIDE OBTAINED FROM THE SUICIDE LITERATURE BJf MEANS OF THE CRITICAL INCIDENT TECHNIQUE 1 60. Married (factor In decreasing 181. Semi-professional the suicide rate) 182. Weak and deficient psycho- 161. Insomnia sexual development 162. Large number of children 183. Immediate crises In sick (factor in decreasing the ness suicide rate) 184. Insomnia 163. Previous threatened suicide 185. Sex 164. Family history of committed 1 8 6. Constipation suicide (recent case or 187. Job Instability case of same age) 188. Alcoholism 165. 'Lonely* occupation 189. Psychopathic state 166. Loss of weight 190. Early morning 167. Family 191. Delusions 168. Severe hypochondriasis 192. Losing control over oneself 169. Insomnia 193. Dread of discharge from 1 7 0. Premeditation over a con security of hospital siderable period 194. SontnpA-hhl 0 p^T-nnnnl 1 -hy 171. Divorced 195. Markedly regressive incli 172. Somatic illness nation 173. Sex frustration, e.g. fri 196. Depression gidity 197. Type of personality associ 174. Social disorganization ated with type of depres 175. Professional sion 176. Employment In Industry 1 98. Type of religion 177. Loss of sleep 199. Fear of losing control, or 178. Financial dependency of hurting others, or one 179- Disappointment and indig self nation over frustrated 200. Feelings of anger attempted suicide 2 01. Abnormality of personality 180. Physical pain 202. Hallucinations or delusions 44 TABLE II (Continued) CAUSES OF SUICIDE OBTAINED FROM THE SUICIDE LITERATURE HT MEANS OF THE CRITICAL INCIDENT TECHNIQUE 203. Feelings of hate and aggres 227. Alcohol sion directed towards self 2 2 8. Complete immobility 204. Feelings of being unwanted 229. Abrupt apparent Improvement 205. Time of year 230. FftlUng prftBRiiw 206. Fear of Insomnia 231. Worry over sleeplessness 207- Loneliness 232. Broken homes 208. Time of day, early morning 233. Previous attempted suicide 209- Alcoholics 234. Wish to obtain reunion with 210. Rising temperature God or loved one 211. Mobility 235. Fear of surgical operation 212. Excessive drinking 236. Under influence of alcohol 213. Depression with guilt feel 237. Feelings of being unloved ings 2 3 8. Previous attempted suicide 214. Slow hesitating gait 239. Migration from South to 215. Episodic alcoholism stormy North 216. Schizophrenia 240. Protestant religious affili 217. Falling temperature (factor ation In decreasing the suicide 241. Psychoses rate) 242. Increasing age 218. Conversion reaction 243. Sudden Improvement in a de 219. Alcoholism pressed, hopeless and per 220. Hallucinations haps delusional patient 221. Depressive state 244. Fear of Insanity 222. Alcoholism 245. Depression 223. Wish to kill undesirable features within self 246. Roman Catholic religious affiliation (factor in de 224. Age creasing the suicide rate) 225- Feans of own aggressive drives 247. Asked to be excused or omit ted from part or all of his 226. Frequent severe tempera usual program of activities ture changes 248. Dread of being a burden to others ^5 TABLE II (Continued) CAUSES OF SUICIDE OBTAINED FROM THE SUICIDE LITERATURE BT MEANS OF THE CRITICAL INCIDENT TECHNIQUE 249. Fear of losing self control and doing damage 250. Previous attempted suicide 251. Frustration 252. Loss of interest in every thing 253. Self disgust 25J + . Threat of retribution 255. Alcohol 256. Autumn 257. Feelings of depression 258. Dietary measures 259. Pathological maladjust ment 260. Solitary personality 261. Emotionally unstable 262. Fear of punishment 263. Improving suddenly from a depressed state 264. Increasing age 265. Aged 266. Depression 267. 111 health 268. Immediate crises In loss of money 269. Insomnia 270. Family ties 271. Family history of psychl- atric abnormality 272. Hypochondriasis 273* Social disorganization (di vorce and illegitimacy) 274. Threatened suicide 275* Sense of guilt 276. Alcohol 277. Agitated 278. Social suffering 279. Self blame 280. High Income 281. Favorable economic circum stances (factor in decreas ing the suicide rate) 282. Impulsivity 283. Divorce 28^. Loss of ability to feel pleas ure (observed or expressed) 285. Solitary personality 286* Attitude of relatives 287* Homosexuality 288. Education 289« Demonstration of omnipotence 290. Immaturity 291* Irresponsible 292. Aggressive attitude towards himself and others 293. Immediate crisis in habits 29^. Loss of ability to feel pain . (observed or expressed) 46 TABLE II (Continued) CAUSES OF SUICIDE OBTAINED FROM THE SUICIDE LITERATURE MEANS OF THE CRITICAL INCIDENT TECHNIQUE 295. Associations 315. Self accusation associated 296. Egocentric with tension and agitation 297. Socially disorganized neigh 316. Frustration of instinctive borhood demands 298. Lowest social class 317. linpulsivity 299. Social fear 318. Loss of person depended on 300. Living alone 319. Identification with a dead 301. Rigid personality person 302. Difficulties throughout life 320. Unemployment 303. Frustration of instinctive 321. Feelings of spite dflmimda 322. Fears 304. Sensitiveness 323. Hedonic personality factors 305* Erratic 324. Occupation 306. Reaction to irksome rigid 325. Wish to be killed or to or too lax supervision atone and safeguarding 326. Dreads or doubts pertaining 307. Unemployment to the hereafter 308. Menopausal trouble 327. Mental illness 309. Experimentation 328. Concern about punishment 310. Manipulation of the environ 329. Previous attempted suicide ment 330. Rising pressure (factor in 311. Conflict between neurotic decreasing the suicide pride and sell hate rate) 312. Hopelessness 331. Sense of guilt and concern 313. Tense about punishment 314. Unemployment 332. Fear of losing mind 333. Worry over physical health ^7 TABLE IH CRITICAL BEHAVIORS BY FREQUENCY OF CAUSES Critical Behavior Total number of causes 1. General affect ^3 2. Social attitudes 30 3. Education and. occupation 2k 4. Physical symptoms 22 5* Morale 16 6. Overt behavioral expression 1^ 7. Delusions and. hallucinations 13 8. Affect, depressive and manic 13 9. Suicidal history 12 10. Marital status 11 11. Sexual attitudes 8 12. Climate 7 13. Alcoholism k l l j - . Time 3 15. Religion 2 16. Age 2 formulated in such a manner that they did not duplicate information such as age, occupation, marital status, etc., which was already ob tained on the hospital admission form. Thirdly, the causes selected should be those which seemed most promising for differentiating var ious suicidal and non-suicidal persons as evaluated by two independent judges who were acquainted with the topic of suicide.* The number of causes selected from each area is shown in table The items for the SPPI were formulated from the 50 causes selected from the sixteen areas. Each item was constructed in such a manner that the response would be either "true’ 1 , "false", or "cannot say". A major requirement was that no references to suicidal behavior were apparent. After the initial formulation of the actual items, the wording of each item was discussed with an independent judge** and reformulated if necessary. Next, the items were worded in such a manner that a normal non-suicidal group would answer, roughly, half of the items as true and half as false. This was done in order to avoid a difficulty found by Whipple ( 9 8)* that is, when items were positively phrased but keyed false, more items were ans wered Incorrectly than when the question had been keyed true. Sim ilarly, when the questions were phrased negatively, more items keyed true were answered Incorrectly than items keyed false. Five filler items were included in order to avoid possible unsuspected cumulative *R. E. Litman, M.D., chief psychiatrist at the Suicide Preven tion Center at Los Angeles, and the author. **Special recognition is due to S. Elsenthal, Ph.D., who so generously contributed his skill and time. TABLE XV NUMBER OF CAUSES SELECTED FROM EACH CRITICAL BEHAVIOR AREA Area Total No. No. of of Causes Causes Selected 1. General affect ^3 11 2. Social attitudes 30 7 3. Education and occupation 2 b 2 b. Physical symptoms 22 i f 5* Morale 16 b 6. Overt behavioral expression lb b 7. Delusions and hallucinations 13 5 8. Affect, depressive and manic 13 - 9. Suicidal history 12 - 10. Marital status 11 3 11. Sexual attitudes 8 3 12. Climate 7 1 13. Alcoholism b 3 11+. Time 3 2 15. Religion 2 1 1 6. Age 2 - effects, and the total number of 55 items randomized. These items were then given to nine non-psychiatric non-suicidal subjects. After they had filled out the inventory, each one was asked for comments on each of the items. Whenever items appeared ambiguous in wording or did not seem to elicit the desired response, a revision in wording was made. This revised form was given to six normal non-suicidal sub jects, all professional persons, and their comments were elicited. The final step in the development of the SPPI consisted of same minor adjustments and resulted in the following format (see table 5)» The Population The MiPI was administered to six hundred male veteran neuro psychiatric patients. Two hundred and twenty of these received the SPPI as well. The time interval between date of admission and date of test administration was on the average less than two weeks. Wo time interval was longer than thirty days. The clinical files of these six hundred patients had been carefully perused to determine the following information (see table 6). All subjects were patients in two local neuropsychiatric Vet eran Administration Hospitals. In order to keep the sample as homo geneous as possible, females and non-caucasians as well as Caucasians of Mexican descent were excluded. Although the original plan was to administer both the M>CPI and the SPPI to all incoming patients, this proved to be impossible. As a result only about one third of those who received the M1PT also received the SPPI. One important requirement was that an adequate 51 TABLE V SUICIDE POTENTIAL PERSONALITY INVENTORY ITEMS 1. I feel worse during the spring or fall than at other times of the year. 2. There were no heavy drinkers in my family. 3* I have not yet decided where I will make ny home. 4. Weakness in myself or others makes me uncomfortable. 5* nearest relatives understand the troubles that I am going through. 6. I wish I could be with someone I once loved very much. 7< I do not have serious financial difficulties. *8. I like sunny weather. 9. I do not became emotionally upset when I am sick in bed. 10. I cannot foretell changes in the weather. 11. Some of the people whom I like and admired have died. 12. I feel that the people who supervise me tend to be too strict. 13* I feel much better now than I have felt in some time. 14. I have the feeling that if someone would pinch me I would not feel it. 15. I am seldom sick. *16. I like watching some television programs. 17- % sexual frustrations have not worsened lately. 18. % future happiness looks promising. 19* X would rather follow than lead others. 20. Recently I have difficulty sleeping. 21. I have had much pain when I was 111. 22. I seldom engage in social activities. 2 3. f t f y * interest in sex has not declined recently. 24. Ify future looks secure. *Filler items 52 TABLE V (Continued) SUICIDE POTENTIAL PERSONALITY INVENTORY ITEMS 25* I think that I am to blame for almost all my troubles. *2.6. I enjoy listening to music. 27. When I am ill the doctor frequently prescribes sedatives for me. 28. Sometimes I am angry for a whole day. 29. I think that I have had more difficulties throughout life than most people. 30. Sometimes I am really very much afraid. 31. I often become very impatient. 3 2. I sometimes fear that I will lose control over myself. 3 3. I often feel that I am unwanted. 3to Lately things have happened to me that are enough to discourage anyone. *3 5. I like flowers. 36. Lately I have not felt like participating in ay usual activities. 37* I am not actively religious. 3 8. I go on occasional drinking sprees. 39. Within the last two years I changed ay jobs at least twice, to. I have no more family ties. h-l. I did not grow up in a broken home. k2. I always feel poorest in early morning than at any other time of the day. 1 4 - 3. There often has been disharmony in our family. 1+4. Making sexual adjustments is not easy for me. 4 9. I have someone whose welfare I very much care for. K6. I never have great fears about the hereafter. 47. Lately I feel quite restless and fidgety, to. I never feel that I am completely worthless. 49. I frequently have a drink in the morning. *50. I enjoy having a vacation. 51. I feel that I am more tenderhearted than most people. 5 2. In the last 5 years I have moved at least once every year. 53. I have no habits that are leading me into repeated difficulties. 54. I lik e to have most things done very precisely. 55* I am not Interested anymore in the things that I used to enjoy doing. *Filler items 53 TABLE VI INFORMATION OBTAINED FOR EACH SUBJECT 1. Name 2. Admission date 3. Test date k. Diagnosis 5. Birth date 6. Age 7. Occupation 8. Education 9. Religion 10. Marital status 11. Race 12. Type of suicide behavior 13- Date of suicide behavior 5^ clinical file be available for those patients whose tests would be used. This file was necessary in order to obtain background infor mation and suicidal or non-suicidal history about the patient. The subjects1 test protocols were collected over a two year time period, roughly 1961-1963. They represent all incoming patients for whom both the MMPI and a clinical file containing the information neces sary for this investigation was available. The main analysis consisted of determining whether or not the MMPI and the SPPI were able to differentiate between suicidal and non-suicidal groups of persons. The two inventories were analyzed separately. The MffI AnelvniH Suicide Variable As discussed previously, one of the prerequisites for the se lection of any subject was the requirement that a MMPI* had been ad ministered to him. After the 600 subjects had been collected, their clinical folders were examined in order to separate the subjects into two main groups. These groups consisted of those patients who never had shown suicidal behavior and those who had shown suicidal behav ior. In addition, the suicidal group was subdivided into various *This l-MPI consisted of the shortened version with a total of 3T3 items. This version still allows the use of all nine clinical scales as well as the four validity scales. 55 types of suicidal behavior (see table 1 for type of suicidal behav ior). It was found that a sufficient number of suicidal subjects could be obtained for analysis for only the threatened (TS), at tempted (AS), and threatened and attempted (TSt-AS) categories. As a result, the other suicidal categories were eliminated for use in this investigation. The final division of the 600 subjects therefore consisted of four groups. These were the non-suicidal (KPC non S), the threatened suicidal (TS), the attempted suicidal (AS), and the threatened and attempted (TS+-AS) suicidal group. The purpose of this division was to determine whether or not the M4P1 would differentiate the four groups from each other as well as the suicides from the non- suicides. Since a relatively large number of subjects was obtained for each group (NPC non S = 283* TS = 154, AS = 72 and T&t-AS = 83), each of the four groups was divided in half for the purpose of conducting both an original and a replication study. The division was accom plished by means of the alternation method. The analysis of the MMPIs of the four groups in the original study was identical to the analysis of the MMPIs of the four groups in the replication study. For each of the 373 items of the *MPI, the "true" and "false" frequency was determined.* A chi-square analysis was then '"Test protocols with more than twelve "don't know" responses were eliminated. The "don't know" responses which remained in some of the protocols were not considered in any individual item analysis. For some items therefore, the total N is somewhat less than the total for that group. Since this recurrence is quite Infrequent however, this deficiency is not thought to have affected the ob tained results. done for each item in order to ascertain which of the 373 items dis criminated between the four groups. This can be shown by- means of the following paradigm, see figure 1. Three levels of significance were used, the .03, the .01 and the .001. After this analysis was completed, each of the four groups was then compared with every other group on every item, again using a chi-square analysis, see figure 2. In all, the following comparisons were made, see table 7> The significance levels in these analyses were again the . 03, the .01 and the .001. There were two reasons why various levels of significance were used. First, it was thought that sample size might influence the number of items reaching significance at each level. Since larger samples tend to include more cases from the tails of a sampling distribution, it might be that with such samples more items will be significant at more stringent levels than would be the case for the same comparison with smaller samples. Secondly, there might be a possibility that certain pairs of groups are more different from each other than other pairs. This too might result in a shift of the number of items reaching significance at certain levels. For Instance, the same number of items may be obtained in two comparisons. In one comparison the number of items at the . 03, the .01, and the .001 levels might be 66, 36, and 16 respectively, while in the second com parison the number of items might be 66, 13, and l < - respectively. If only the .0 3 level of significance had been used such differences would not be apparent. 57 item response NPC non S TS__________AS__________TS*AS True False FIGURE I PARADIGM FOR THE 2X4 MMPI TEEM ANALYSES item resixtnse_____ group A group B True False FIGURE II PARADIGM FOR THE 2X2 HEM ANALYSES 59 TABLE VII COMPARISONS OF GROUPS IN NMFI ANALYSIS FOR *S* VARIABLE NPC non S versus TS NPC non S versus AS NPC non S versus TS4-AS TS versuB AS TS versus TS4-AS AS versus TS4-AS 60 These analyses investigated two important things. First, it would show whether or not items from the *®IP! could differentiate "between the non-suicidal group and any of the three types of suicidal groups. Secondly, it would show whether or not items from the MMPI could discriminate between the three suicidal groups. The reason that a 2 x ^ analysis (see figure l) was done in addition to the 2 x 2 analysis (see figure 2) was to determine whether or not possi ble differences would transcend a simultaneous analysis (2 x of all four groups. After these various analyses were completed another aspect of the overall study was investigated. This involved the influence of sample size on the total number of items found to be significant, as well as the number of such items found at each level of significance used. In order to accomplish this, the four groups in the original study were combined with their similar groups in the replication study. The same 2 x ^ and 2 x 2 chi square analyses as described above (see figures 1 and 2) were done on these combined groups. As was discussed previously, it is not completely certain that the separation of suicidal behavior into such categories as for in stance threatened, attempted and threatened and attempted suicide is meaningful. Instead, Just a global classification of "suicidal" might be sufficient. This aspect was investigated by dividing all 600 subjects into those who were suicidal and those who were non- suicidal and analyzing the MMFIa of the two groups. It will be re called that the division of suicidal and non-suicidal was achieved 61 through Information obtained from the clinical files of the subjects. The analysis of the MIFIs of the two groups, for each item, was done by means of a 2 x 2 chi square analysis (see figure 2). Control Variables The third stage of this investigation concerned itself with the so-called control variables of age, marital status, occupation, education, religion, number of Veteran Administration Hospital (VAH) admissions, as well as the experimental variable for which the MMPI had been developed, psychiatric diagnosis. The problem investigated here, consisted of determining whether or not MMPI chi square compari sons of subcategories of each of the variables above would yield any items which would discriminate between such sub categories in each variable. Such findings would be Important in the analysis of the suicide variable. For instance, if it would be found that the educa tional level of a person affects his MMPI responses, the analysis of the suicide variable could be done more profitably by taxing this fact into account. This could be accomplished by such means as separating the various suicidal and non-suicidal groups into those with elemen tary, those with high school and those with college education. The resulting suicidal subgroups would then be compared with the resulting nan-suicidal counterparts. In other words, Instead of using a simple randomized design, it would be more meaningful to use a stratification or treatments by levels design. It is of course possible that more than one variable will be found to contribute to a subject's MMPI re sponses. In such an event, additional analyses such as the calcula 62 tion of multiple correlations would become useful in more meaningfully interpreting any findings in the suicide variable analysis. Each of the variables was divided into the following subcategories, see table 8. For each analysis all 600 subjects were used. Information ob tained from the clinical folder of the patient determined the sub category of the seven variables to which he belonged. For each vari able, every subcategoiy was then compared with every other one by means of a 2 x 2 chi square item differentiation analysis, see the paradigm in figure 2. This resulted in the following chi square item differentiation comparisons for each variable, see table 9. In any comparison of two samples, the above seven variables of course operate simultaneously. It ms decided to investigate the in fluence of complete control of the variables within comparison samples versus no control. Two samples were used for the complete control and two for the no control method. One sample in each method consisted of non-sui cidal persons and the other sample in each method consisted of suicidal persons. As will be recalled, the suicidal and non- sulcidal division had previously been obtained. In order to achieve complete control of all variables, the same procedure was followed for the suicidal sample and the non-sui cidal sample. The initial step consisted of sorting all persons in the sample for type of diagnosis. The persons in the subcategory contain ing the largest frequency were then used for sorting on the next vari able, viz., education. Again the subjects in the educational 63 TABLE VIII SUBCATEGORIES OF VARIABLES ANALYZED IN THE M*PI ANALYSIS 1* Diagnosis neurotic psychotic personality disorder other 2. Religion Protestant Catholic Hebrew 3. Education elementary school high school college 4. Occupation professional clerical operator laborer 5. Marital Status single married divorced married/separated 6. Age up to 39 40-49 50 7. Number of V.A.H. Admissions 1 2 3 or 4 5 4- 6k TABLE IX COMPARISONS OP NON-SUICIDAL VARIABLES ANALYZED IN THE IMP! AND THE SFPI 1. Diagnosis neurotic versus neurotic versus neurotic versus psychotic versus psychotic versus personality disorder versus 2. Religion psychotic personality disorder other personality disorder other other Protestant Catholic Protestant versus versus versus Catholic Hebrew Hebrew 3. Education elementary school versus elementary school versus high school versus I j - . Occupation high school college college professional versus clerical professional versus operator professional versus laborer clerical versus operator clerical versus laborer operator versus laborer 65 TABLE IX (Continued) COMPARISONS OF NON-SUICIDAL VARIABLES ANALYZED IN THE MMPI AND THE SFPI Age J . OX-aoUS single versus married single versus divorced single versus married/separated married versus divorced married versus married/separated up to 39 versus 1*0 - 1*9 up to 39 versus 50* 1*0 - 1+9 versus 50* of V.A.H. Admissions 1 versus 2 1 versus 3 or 1 * 1 versus 5+* 2 versus 3 or 1 * 2 versus 5+ 3 or ^ versus 5*- 66 subcategory with the largest frequency were selected for sorting on the subcategories of the next variable, and so on. The only variables that were not used were religion and type of suicidal behavior. The two "no control" groups were selected at randan from the suicidal and non-suicidal populations which remained af oer the complete control groups had been drawn. It was decided to run all possible compari sons on these four groups, see table 10. Chance Significances Another aspect that was Investigated In this study after the analysis of the "complete control" and "no control" sample compari sons, consisted of determining how many items at the . 05, the .01 and the .001 levels of confidence would attain chance significance in samples of different size and selected by different sampling tech niques. The two sampling techniques were replacement and no replace ment sampling. Sampling without replacement was accomplished by selecting subjects at random from the population of 600. No subject, however, was replaced In the population after he had been selected; thus, no subject could be selected more than once. This was In con trast to the sampling with replacement method. Here, each subject was replaced In the population after having been selected. Each sub ject, therefore, could be selected more than once In any one sample. In order to determine the effect of sample size on the number of Items reaching chance significance at the various levels of con fidence, samples of size 30, ^0, 50, 100, 150 and 200 were selected for each of the two methods of sampling. All subjects were selected TABLE X COMPARISONS OF "NO CONTROL" SAMPLES WITH "COMPLETE CONTROL" SAMPLES 1. "complete control" versus "complete control suicidal non-suicidal 2. "complete control" versus "no control" suicidal suicidal 3. "complete control" versus "no control" suicidal non-suicidal h. "complete control" versus "no control" non-suicidal suicidal 5. "complete control" versus "no control" non-suicidal non-suicidal 6. "no control" versus "no control" suicidal non-suicidal at random. This was accomplished by means of a table of random num bers (80). In the sampling without replacement method the following number of comparisons were made for each of the samples, see table 11. In the sampling with replacement method, the following number of com parisons were made for each of the samples, see table 12. The taPPI ArmJjyfliH The analysis of the SPPI proceeded along similar lines as the MMPI analysis. The SPPI was administered to a total sample of 220 patients. Like the MMPI, these subjects were divided into those who were non-suicidal and those who were suicidal. Again, this separa tion was achieved through the examination of the clinical files of the patients. A 2 x 2 chi square analysis was done to determine how many items discriminated the two groups. After this analysis had been performed, the suicidal group was subdivided into a threatened suicidal (TS), an attempted suicidal (AS), and a threatened and attempted (T£H-AS) group. A 2 x 4 chi square analysis was done using these three suicidal groups and the non- suicidal group. In addition to the 2 x 4 analysis, each of the four groups was compared with the other by means of 2 x 2 chi square anal yses. After this procedure, all 220 subjects were sorted according to subcategories of age, education, occupation, marital status, re ligion, and number of V.A.H. admissions instead of the non-suicidal and suicidal division. As with the ttiPI, the groups, now subdivided into subcategories of the above variables, were then compared with 69 TABLE XI NUMBER OF COMPARISONS FOR THE ESTABLISHMENT OF CHANCE SIGNIFICANCES FOR SAMPLES OF DIFFERENT SIZE: WITHOUT REPLACEMENT SAMPLING size of sample number of comparisons 30 27 1 + 0 7 50 21 100 1 1+ 150 6 200 3 70 TABLE X U NUMBER OF COMPARISONS FOR THE ESTABLISHMENT OF CHANCE SIGNIFICANCES FOR SAMPLES OF DIFFERENT SIZE: WITH REPLACEMENT SAMPLING size of sample number of ccxqparlsons 30 13 1*0 13 50 13 100 11 150 6 200 3 71 each other (see table 9)• CHAPTER IV RESUHTS Due to the nature of the data the results will he presented in table form. Chance Findin^R in Sampling Without Re~m Table 13 shows a summary of the number of MMPI items which reached significance at the . 05, the .01, and the .001 levels respec tively in samples of size 30* 40, 50, 100, 150, and 200. These sam ples were obtained by means of random sampling without replacement. It will be recalled that in this method, none of the variables were controlled in the comparison of the samples and that the findings therefore only Indicate chance significances. This procedure there fore, establishes by objective means the number of MMPI items that will reach significance, regardless of the size of the item inter correlations. The number of comparisons for samples of size 30, 40, 50, 100, 150, and 200 were 27* 7* 21* l1 * - * 6* and 3 respectively for the three levels of significance. For the .05 level the mean number of items which reached chance significance were 9* 10, 18, 10, 13, and 22 respectively for samples of size 30, * 4 - 0 , 50, 100, 150, and 200. The range of items reaching chance significance at the .0 5 level for these samples sizes was 3 - 22, 4 - 19* 0 - 46, 5 - 18* 12 - 16, and 13 - 2 9. 73 TABLE XIII UMBER OF SIGNIFICANT M S AT THE . 05, . 0 1AND .001 IWELS OF SIGNIFICANCE, SAMPLES OF DEMENT SIZE OHTAUED THROUGH SAMPLING WITHOUT REPLACEMENT Comparison 1 2 3 k 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 2^ 25 26 27 x range sign, sample level size 30 5 7 8 5 7 5 7 13 22 8 5 8 3 8 7 9 8 15 16 i f 3 16 12 11 12 5 lA 9 3 22 to 11 19 12 8 i f 9 8 10 i f 19 50 16 1 5 0 33 Ik 9 7 20 16 11 13 20 lk 38 10 t o 13 19 to 13 I l f 18 0 to 100 11 9 15 10 5 10 8 9 11 12 5 18 13 10 10 5 18 150 12 12 Ik 16 12 I l f 13 12 16 200 29 13 23 22 13 29 30 2 0 0 0 1 0 1 1 5 0 1 2 0 1 0 1 2 1 i f 0 0 3 2 3 1 0 3 1 0 5 to 1 3 3 1 0 2 1 2 0 3 50 3 k 0 7 1 2 0 2 3 2 1 If 3 9 2 8 2 3 10 2 3 3 0 10 100 1 1 1 0 1 2 1 1 1 2 0 3 1 1 1 0 3 150 2 1 0 i f 0 0 1 0 i f 200 7 2 1* If 0 7 30 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 1 0 1 0 0 .1 0 1 to 0 0 0 0 0 0 1 .1 0 1 .001 50 0 0 0 1 0 0 0 0 0 0 1 1 1 1 0 0 0 1 0 0 0 .3 0 1 100 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 150 0 0 0 0 0 0 0 0 0 200 2 0 1 1 0 2 7*4- At the .01 level of significance the mean number of items found for sample sizes 30, * + 0, 50, 100, 150, and 200 were 1, 2, 3, 1, 1, and h respectively. The range of items obtained for these sample sizes were 0 - 5, 0 - 3, 0 - 10, 0-3, 0 - * 4 - , and 2 - 7. The mean number of items reaching chance significance at the .001 level was .1, .1, .3* 0, 0, and.l respectively for samples of size 30, * 4 - 0 , 50, 100, 150, and 200. The range of items for these sample sizes were 0 - 1, 0 - 1, 0 - 1, 0 - 0, 0 - 0, and 0 - 2. Chance Findings in Sampling With Replacement Table 1 * 4 - also shows a summary of the number of items which reached chance significance at the .05, the .01, and the .001 levels in samples of size 30, * 4 - 0 , 50, 100, 150, and 200. In contrast to table 13 however, the samples in table 1 * 4 - were obtained by means of random sampling with replacement. The results are reported for sam ples of size 30, * 4 - 0 , 50, 100, 150, and 200 respectively. At the .05 level the mean number of items reaching chance significance was 12, 1 * 4 - , 12, 91 16, and 16. The range of items for each sample size was 3 - 31, 1 - 33, 3 - 21, * 4 - - 16, 2 - 25, and 5 - 26. At the .01 level, the mean number of items obtained was 1, 2, 2, 1, 2, and 2 while the range of the items was 0 - 6, 0 - 5, ° - 6, 0 - 2, 0 - 5, and 1 - * 4 - . At the .001 level the mean number of items was .1, . 3, 0, 0, .2, and 0 and the range was 0 - 1, 0 - 2, 0 - 0, 0 - 0, 0 - 1, and 0 - 0. TABLE XIV NUMBER OF SIGNIFICANT M S AT THE .05, .01 AD .001 LEVELS OF SIGNIFICANCE, SAME® OF DfflfflP SIZE OBTAINED THROUGH SAMPLING WITH REFIACEMEHP Comparison 1 2 3 k 5 6 7 8 9 10 11 12 13 l*t 15 16 1? 18 19 20 21 22 23 2 \ 25 26 2? x range sign, sample level size 3 0 u 3 1 2 2 2 1 1 0 1 2 2 2 3 I t 3 9 I t 7 1 2 3 3 1 to 5 3 3 1 9 1 1 1 5 6 1 6 2 l t 1 3 1 3 9 2 1 l i t 1 3 3 5 0 I t 8 2 0 1 1 l i t 3 l i t l i t 1 7 2 1 8 1 1 l i t 1 2 3 2 1 1 0 0 I t 7 I t 1 3 1 6 1 6 9 i t 7 1 0 i t 9 i t 1 6 1 5 0 1 8 2 5 2 0 2 l i t 1 9 1 6 2 2 5 2 0 0 1 8 2 6 5 1 6 5 2 6 3 0 1 6 3 3 l 1 2 1 0 0 0 0 0 1 0 6 t o 0 5 0 0 l 2 0 3 5 1 2 0 I t 2 0 5 5 0 0 1 2 3 2 0 1 I t 2 6 1 1 1 2 0 6 1 0 0 0 1 0 l 1 2 1 2 1 0 0 1 0 2 1 5 0 2 5 I t 0 3 1 2 0 5 2 0 0 2 1 1 2 0 i t 3 0 0 0 0 0 1 0 0 0 0 0 0 0 0 . 1 0 1 t o 0 1 0 0 0 0 0 0 2 0 0 0 0 . 3 0 2 . 0 0 1 5 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 5 0 0 1 0 0 0 0 . 2 0 1 2 0 0 0 0 0 0 0 0 76 Analysis of the Diagnosis. Religion, Education* Occupation. Marital Status. Age, and Number of VAH Admissions Variables The total number of significant items for each comparison of the various subcategories of diagnosis, religion, education, occupa tion, marital status, age, and number of VAH admissions are listed in table 15. Columns one and two of this table show the number of sub jects in each comparison group. The next six columns, 3, ^, 5, 6, 7, and 8 are arranged in pairs. They show the expected and obtained number of items reaching the .05, the .01, and the .001 levels of significance. The expected number of significant items is that number of items which reached significance due to chance processes. This number was found from the findings summarized in table 13• Two quan tities are given in each of the "expected" columns. These quantities were determined by means of the largest sample in each comparison. This sample was used to establish the mean number of items in table 15 that was associated with this sample. This mean was taken as the number of items expected to occur by chance. The reason for choosing the sample with the largest martber of subjects in each comparison was because this would give the greatest number of "expected" items. Any conclusions as to the overall significance of the findings would therefore err on the conservative side. The larger of the two quantities in the "expected" columns indicates the greatest number of items which were found to have oc curred by chance for samples of that size and significance level. Since the smaller quantity represents the mean number of significant items found by chance, the two figures in the "expected" columns TABLE XV EXPECTED AND OBTAINED NUMBERS OF SIGNIFICANT MMPI ITEMS FOR COMPARISONS OF DIFFERENT GROUPS AT THE .05, THE .01 AND THE .001 LEVELS OF SIGNIFICANCE 1 nl 2 *2 3 Exp. .05 4 out. .05 5 Exp. .01 6 Obt. .01 7 Exp. .001 8 Obt. .001 9 Sign. Diagnosis Neurotic vs. Psychotic 193 235 22-29 171 4-7 62 1-2 28 S Neurotic vs. Pers. Dis. 193 98 22-29 30 4-7 b 1-2 1 S Neurotic vs. Other 193 66 22-29 8k 4-7 23 1-2 4 S Psychotic vs. Pers. Dis. 235 98 22-29 57 4-7 lb 1-2 3 S Psychotic vs. Other 235 66 22-29 59 b-7 13 1-2 2 s Pers. Dis. vs. Other 98 66 10-18 33 1-3 9 0 1 s Religion Protestant vs. Catholic 361 170 22-29 14 4-7 2 1-2 - - Catholic vs. Hebrew 170 38 13-16 13 1-b 2 0 - - Protestant vs. Hebrew 361 38 22-29 NC* b-7 N C * 1-2 N C * N C * Education Elementary vs. H.S. 117 330 22-29 ^3 b-7 7 1-2 1 s Elementary vs. College 117 121 10-18 80 1-3 27 0 7 s H.S. vs. College 121 330 22-29 48 b-7 14 1-2 3 s *Not calculated— not significant "by deduction TABLE XV (Continued) EXPECTED AND OBTAINED NUMBERS OF SIGNIFICANT MMPI ITEMS FOR COMPARISONS OF DIFFERENT GROUPS AT THE .05, THE .01 AND THE .001 LEVELS OF SIGNIFICANCE 1 2 3 4 5 6 7 8 9 nl n2 Exp. Obt. Exp. Obt. Exp. Obt. Sign. .05 .05 .01 .01 .001 .001 Occupation Professional vs. Clerical 104 n o 10-18 28 1-3 3 0 - S Professional vs. Operators 104 192 22-29 5T 4-7 15 1-2 3 S Professional vs. Laborers 104 174 13-16 76 1-4 21 0 5 s Clerical vs. Operators 110 192 22-29 3^ 4-7 10 1-2 2 s Clerical vs. Laborers 110 174 13-16 46 1-4 8 0 1 s Operators vs. Laborers 192 174 22-29 23 4-7 4 1-2 - mm Marital Status Single vs. Married 145 240 22-29 101 4-7 27 1-2 13 s Single vs. Divorced 145 145 13-16 62 1-4 17 0 3 s Single vs. Married/Separated 1^5 56 13-16 58 1-4 14 0 4 s Married vs. Divorced 240 145 22-29 57 4-7 19 1-2 6 s Married vs. Married/Separated 240 56 22-29 26 4-7 5 1-2 1 9 • Divorced vs. Married/Separated 145 5 6 13-16 13 1-4 2 0 - mm *Not calculated— not significant by deduction 00 TABLE XV (Continued) EXPECTED AND OBTAINED NUMBERS OF SIGNIFICANT MMPI ITEMS FOR COMPARISONS OF DIFFERENT GROUPS AT THE .05, THE .01 AND THE .001 LEVELS OF SIGNIFICANCE 1 nl 2 n2 3 Exp. .05 4 Obt. .05 5 Exp. .01 6 Obt. .01 7 Exp. .001 8 Obt. .001 9 Sign. Age 0-39 vs. 40-49 314 211 22-29 56 4-7 17 1-2 5 S 0-39 vs. 50* - 314 72 22-29 119 4-7 36 1-2 13 s 40-49 vs. 504 * 211 72 22-29 )+9 4-7 13 1-2 - s Number of Admissions 1 vs. 2 338 117 22-29 28 • 4-7 4 1-2 2 s 1 vs. 3 or 4 338 80 22-29 138 4-7 43 1-2 14 s 1 vs. 5+ 338 52 22-29 96 4-7 30 1-2 11 s 2 vs. 3 or 4 117 80 10-18 40 1-3 9 0 2 s 2 vs. 5+ 117 52 10-18 38 1-3 8 0 2 S 3 or 4 vs. 5+ 80 52 10-18 11 1-3 1 0 *Not calculated— not significant by deduction \ o represent the upper limits for chance expectancy. As a result, the conclusions in column namely those that indicate whether or not the obtained results are considered to have occurred by chance, re present the most conservative estimate possible regarding acceptance of overall significance. The results in table 15 show that all but two comparisons are either significantly different or not significantly different. Two comparisons, namely those of married versus married but separated and divorced versus married but separated are ambiguous. The number of items obtained for these two comparisons fall just within the upper chance range. All the results of the other comparisons are either well above or well below chance level. The only comparisons where the results obtained were below chance are all three of the religious subcategoiy comparisons, the operator versus laborer and the 3 or 4 admissions versus 5 plus admissions comparisons. A H the other re sults were well above chance occurrence. Analysis of the Suicidal Variable The results of the comparisons of the various suicidal and non-suicidal groups are listed in table 16. Except for column 10, the explanation of the various columns in table 16 is identical to the one for table 15* Column 10 indicates the number of overlapping items between the comparisons in the original study and their coun terparts in the replication study. The results shew that all of the comparisons of the NPC non S group on the one hand and the various suicidal groups on the other hand are significant. The comparisons TABLE XVI EXPECTED AND OBTAINED NUMBER OF SIGNIFICANT MMPI ITEMS FOR COMPARISONS OF DIFFERENT SUICIDAL AND NON-SUICIDAL GROUPS AT THE ,05, THE .01 AND THE .001 LEVELS OF SIGNIFICANCE 1 nl 2 “ 2 3 Exp. .05 4 Obt. .05 5 Exp. .0 1 6 Obt. .0 1 7 Exp. .001 8 Obt. .0 0 1 9 Sign. 10 Original NPC vs. TS vs. AS vs. TSfAS * 70 24 7 S 30 Replication NPC vs. TS vs. AS vs. TS+-AS - - - 79 M 23 - 14 S 30 Combined all NPC all AS vs. all TS vs vs. all TS+AS » 126 46 36 s co H TABLE XVI (Continued) EXPECTED AND OBTAINED NUMBER OF SIGNIFICANT MMPI ITEMS FOR COMPARISONS OF DIFFERENT SUICIDAL AND NON-SUICIDAL GROUPS AT THE .05, THE .01 AND THE .001 LEVELS OF SIGNIFICANCE 1 nl 2 n2 3 Exp. .05 1 * Obt. ...05 5 Exp. .01 6 Obt. .01 7 Exp. .001 8 Obt. .001 9 Sign. 10 Original NPC vs. TS 142 77 13-16 60 1-4 25 0 5 S 15 NPC vs. AS 142 36 13-16 30 1-4 7 0 0 S 9 NPC vs. TSt-AS lk2 42 13-16 67 1-4 24 0 7 S 23 TS vs. AS 77 36 10-18 1 * 1-3 1 0 0 - TS vs. TS*AS 77 42 10-18 10 1-3 1 0 0 - - AS vs. TS+AS 36 42 10-19 10 2-3 2 1 0 - - Replication NPC vs. TS l4l 77 13-16 63 1-4 21 0 8 s 15 NPC vs. AS l4l 36 13-16 62 1-4 22 0 5 s 9 NPC vs. TS+-AS l4l 41 13-16 60 1-4 21 0 7 s 23 TS vs. AS 77 36 10-18 8 1-3 1 0 0 - TS vs. TS+AS 77 1 * 1 10-18 5 1-3 1 0 1 - * AS vs. TS+AS 36 1 * 1 10-19 7 2-3 1 1 0 - - CD ro TABLE XVI (Continued) EXPECTED AND OBTAINED NUMBER OF SIGNIFICANT MMPI ITEMS FOR COMPARISONS OF DIFFERENT SUICIDAL AND NON-SUICIDAL GROUPS AT THE .05, THE .01 AND THE .001 LEVELS OF SIGNIFICANCE 1 2 3 4 5 6 7 8 9 10 nl n2 Exp. Obt. Exp. Obt. Exp. Obt. Sign. J- C. .05 .05 .01 .01 .001 .001 Combined all NPC vs. all TS 283 154 22-29 98 4-7 54 1-2 24 S - all NPC vs. all AS 283 72 22-29 74 4-7 42 1-2 12 S - all NPC vs. f i - T 1 TS+AS 283 83 22-29 120 4-7 68 1-2 27 S - all TS vs. all AS 154 72 13-16 6 1-4 0 0 0 - - all TS vs. all TS+AS 154 83 13-16 12 1-4 0 0 0 mm - all AS vs. all TS+AS 72 83 10-18 12 1-3 2 0 0 CD U J 84 of the suicidal groups themselves however, did not result in any greater than chance findings. The number of overlapping items in the original and replication study were 15, 9, and. 23 respectively for the comparisons of the NPC non S versus the TS, the NPC non S versus the AS, and the NPC non S versus the TEH-AS groups. Significant Comparisons for MMPI Items Table 18, see appendix, lists for each MMPI item the specific comparisons which reached significance at the .05 level or better. The results of this table give an indication of the complexity of individual MMPI items. For some items, no comparisons reached sig nificance. For others, various subcategoiy comparisons of one vari able contributed more than comparisons of subcategories of other variables. For still other items, comparisons of many variables con tributed something. For instance, for items 119, 151, 196, and 211 no comparison reached significance. In other words, responses to these items were completely at random. For other items however, such as 1, 5, 9, 13, 2 0, 2 1, 3^, ^5, 51, 137, 138, 153, 170, 179, 226, 243, 249, 265, 280, 286, 294, and 298 several comparisons between subcategories of many different variables reached signifi cance. These seem to be items which influence many variables. Some items on the other hand influence one variable more than others. Items such as 2, 3, 23, 31, 32, 39, 86, 9 h , 104, 129, 139, 1^2, 158, 159, 208, 216, 305, 318, 328, 339, 359, and 366 appear to influence the suicide variable more than they do other variables. The code that will be used in this summary table 18 to 85 represent the various comparisons is given in table 17, see appendix. The MMPI items which were found to differentiate the non- suicidal from the suicidal patients are shown in table 19 together with the obtained chi square value. This table shows that a total of 155 MMPI items reached significance at the .05 level or better. At the .01 level this number was k $ ) and. at the .001 level it was 51* The results of the "complete control” versus "no control" sample comparisons are summarized in tables 20, 21, 22, 23, 2k, and 25- These tables show that the largest number of significant items (l6) was obtained from the "complete variable control" - NPC non S versus "complete variable control" - combined S group comparison. However, the comparisons of the "no variable control" - UPC non S versus "complete variable control" - combined S group resulted in 15 significant items. Only one MMPI item, number 25, however, over lapped these two series. The other comparisons resulted in eight or less significant items. SPPI Analysis In table 26 a summary of the results of the various SPPI anal yses is shown. Columns one and two indicate the sample size for each comparison. Columns 3, 5, and. 7 show the number of significant items which are expected to occur by chance. This quantity represents the upper bound number of expected items and has been interpolated from M£PI chance findings in table 13* Columns k, 6, and 8 give the obtained number of items which attained significance. Column 9 In dicates whether or not this number is considered to have occurred by 86 TABLE XIX LIST OF MCI ITEMS AND THEIR OBTAINED CHI SQUARE VALUES WHICH ATTAINED THE .05, THE .01 AND THE .001 LEVELS OF SIGNIFICANCE All NPC : n =283 non S versus All S n =309 LEVELS OF SIGNIFICANCE Item nos. .05 .0 1 .0 0 1 2 24.65 3 1 6 .1 1 8 12.49 9 5 .2 2 11 1 2 .0 6 13 14.77 14 1 2 .2 8 IT 6.45 20 9.09 21 8.33 23 15.90 24 7.96 28 5 .8 1 29 9 .0 8 30 8.55 31 1 0 .8 2 32 16 .1 4 34 14.57 36 7.47 38 1 0 .2 7 39 12.91 41 3 0 .6 1 42 7.64 43 1 6 .8 6 45 14.61 87 TABLE X3X (Continued) LISE OF MMPI ITEMS AND THEIR OBTAINED CHI SQUARE VALUES WHICH ATTAINED THE .05, THE .01 AND THE .001 LEVELS OF SIGNIFICANCE All NPC non S versus All S n =283 n =309 LEVELS OF SIGNIFICANCE item nos. .05 .01 .001 46 25.62 47 13.12 51 7.13 53 5-05 54 7.27 57 4.56 61 11.35 62 9.00 65 10.09 67 7.80 73 4.15 76 11.28 77 5.96 79 4.58 84 6.01 86 22.67 88 21.60 90 14.02 93 4.90 94 24.89 96 5.79 97 15.05 100 4.38 104 36.05 88 TABLE XIX (Continued.) LIST OF MMPI HEMS AND THEIR OBTAINED CHI SQUARE VALUES WHICH ATTAINED THE .05, THE .01 AND THE .001 LEVELS OF SIGNIFICANCE All NPC non S versus All S n=283 n = 309 LEVELS OF SIGNIFICANCE item nos. .05 .01 .001 106 7.61 107 14.96 111 4.25 114 4.35 119 3.93 124 4.33 125 5.92 126 5.77 129 7.79 130 4.90 133 3.8 6 134 16.81 135 10.65 137 7.60 138 8.67 139 52.02 142 31.16 145 8.55 152 13.81 153 13.04 155 4 .6 0 156 7.74 158 4 .7 8 159 17.39 160 33.02 163 11.06 89 TABLE XIX (Continued.) LIST OF MMPI ITEMS AND THEIR OBTAINED CHI SQUARE VALUES WHICH ATTAINED THE .05, THE .01 AND THE .001 LEVELS OF SIGNIFICANCE All NFC non S versus All S n = 283 n = 309 LEVELS OF SIGNIFICANCE Item nos. .05 .01 .001 168 13.79 169 5.03 170 **•39 175 6 .8 7 176 3.88 178 16.18 182 8.10 186 11.87 187 9 .0 1 189 7 .8 6 191 7.70 192 4.07 195 4.26 198 7.1*5 205 7.32 206 3.88 209 6.68 215 17.29 216 17.60 217 15.79 220 k.20 225 k.zQ 226 4 .5 6 227 14-.11 228 6 .5 3 90 TABLE XXK (Continued) LIST OF M4PI ITEMS AND THEIR OBTAINED CHI SQUARE VALUES WHICH ATTAINED THE .05, THE .01 AND THE .001 LEVELS OF SIGNIFICANCE All NPC non S n =283 versus All S n = 309 LEVELS OF SIGNIFICANCE item nos. .05 .01 .001 230 7.92 232 15.76 236 23.80 238 9.75 242 4.1*3 251 8.92 252 6.60 255 6.09 257 4.20 259 10.62 261 10.30 264 6.92 2 66 5.05 270 4.16 276 4.88 277 9.35 278 9.69 282 4.28 28k 6.06 287 4.99 288 4.66 290 8.12 292 5.47 294 8.23 91 TABLE XIX (Continued) LIST OF MMPI ITEMS AND THEIR OBTAINED CHI SQUARE VALUES WHICH ATTAINED THE .05, THE .01 AND THE .001 LEVELS OF SIGNIFICANCE All NPC non S versus All S n =283 n =309 LEVELS OF SIGNIFICANCE Item nos. .05 .01 .001 295 7.58 301 10.06 305 2I 4-.07 308 13.^1 310 8.99 3H 10.5^ 312 8.83 313 5.16 316 17.28 318 20.65 321 5.27 322 k.k6 328 17.07 333 ^•93 335 1U.5 6 336 12.86 337 13.80 339 21.1 + 3 3^0 6.08 3^2 5.81 3^5 8.70 31+6 9.27 353 6.07 356 7.06 359 12.03 361 ^.65 366 29.18 367 1+.97 370 10.39 372 8.71 373 k.03 n =155 n =55 n=U9 n =51 92 TABLE XX SIGNIFICANT ITEMS AT THE .05, THE .01 AND THE .001 LEVELS OF SIGNIFICANCE FOR "NO VARIABLE CONTROL" — NPC GROUP (n=13) VERSUS "NO VARIABLE CONTROL"— COMBINED S GROUP (n=8) Item nos. SIGNIFICANCE LEVELS .05 .01 .001 25 7.50 175 4.85 226 7.29 249 4.59 271 6.33 n=5 n=3 n=2 n=0 93 TABLE XXI SIGNIFICANT ITEMS AT THE .05, THE .01 AND THE .001 LEVELS OF SIGNIFICANCE FOR "COMPLETE VARIABLE CONTROL" — NPC GROUP (n=13) VERSUS "COMPLETE VARIABLE CONTROL"— COMBINED S GROUP (n=8) Item nos. SIGNIFICANCE LEVELS .05 .01 .001 25 6.68 28 5.15 61 5.86 69 5.H 78 7.78 115 5.56 136 4.32 148 4.32 179 4.32 215 4.06 254 5.86 260 4.27 266 4.85 291 5.11 316 7.07 323 7.78 n=l6 n=l2 n=L n=0 9^ TABLE XXII SIGNIFICANT ITEMS AT THE .05, THE .01 AND THE .001 LEVELS OF SIGNIFICANCE FOR "COMPLETE VARIABLE CONTROL"— NPC GROUP (n=13) VERSUS "NO VARIABLE CONTROL"— NPC GROUP (n=13) item nos. LEVELS OF SIGNIFICANCE .05 .01 .001 6l 3.87 152 k . 51 185 5.U2 2kb 7.96 273 3.96 n=5 n=4- n=l n=0 95 TABLE XXHI SIGNIFICANT ITEMS AT THE .05, THE .01 AND THE .001 LEVELS OF SIGNIFICANCE FOR "NO VARIABLE CONTROL"— NPC GROUP (n=13) VERSUS "COMPLETE VARIABLE CONTROL" — COMBINED S GROUP (n=8) item nos. LEVELS • 05 OF SIGNIFICANCE .01 .001 25 10.22 28 5.15 32 l*.32 136 5.86 ll*9 5-11 179 5-86 233 6.68 237 5-86 2l*3 l*-32 25I * 5.86 259 4.32 260 7.07 316 1+.27 328 4.32 352 U.85 n=15 n=12 n=3 n=0 96 TABLE XXIV SIGNIFICANT ITEMS AT THE .05, THE .01 AND THE .001 LEVELS OF SIGNIFICANCE FOR "COMPLETE VARIABLE CONTROL" — COMBINED S GROUP (n=8) VERSUS "NO VARIABLE CONTROL"— COMBINED S GROUP (n=8) item nos. LEVELS OF SIGNIFICANCE .05 .01 .001 172 4.06 179 4.06 226 4.o6 229 4.06 254 6.25 260 4.65 n=6 n=6 n=0 n=0 97 TABLE XXV SIGNIFICANT ITEMS AT THE .05, THE .01 AND THE .001 LEVELS OF SIGNIFICANCE FOR "COMPLETE VARIABLE CONTROL"— NPC GROUP (n=13) VERSUS "NO VARIABLE CONTROL"— COMBINED S GROUP (n=8) item nos. LEVELS OF SIGNIFICANCE .05 .01 .001 25 4.38 3 6 6.68 120 4.32 153 10.21 154 4.85 214 4.85 224 4.32 361 4.32 n=8 n=6 n=l n=l tab:* xxvi EXPECTED AND OBTAINED NUMBER OF SIGNIFICANT SPFI ITSMS FOR COMPARISONS OF SUICIDAL AND NON-SUICIDAL VARIABLES AT THE .05, THE .01 AND THE .001 IEVELS OF SIGETPICANCE 1 nl 2 n2 3 Exp. .05 k Obt. .05 5 Exp. .01 6 Obt. .01 7 Exp. .001 8 Obt. .001 9 Sign. Suicide Status NFC vs. TS 83 55 1-2 k 1 10 0 1 S NFC vs. AS 83 33 1-2 8 1 2 0 - S NFC vs. TS+AS 83 k2 1-2 17 1 7 0 2 S TS vs. AS 55 33 2-6 2 1 0 0 0 - TS vs. TS+AS 55 k2 2-6 5 1 0 0 0 - AS vs. TS+AS 33 k2 2-6 1 1 0 0 0 - *01 NFC vs. ftTT S 83 130 1-2 11 1 6 0 2 S Education Elementary vs. H.S. 50 107 1-2 5 1 2 0 0 S Elementary vs. College 50 k2 2-6 1 1 0 0 0 - H.S. vs. College 107 1+2 1-2 2 1 1 0 0 - Age 0-39 vs. UO-^9 105 80 1-2 0 1 0 0 0 - 0-39 vs. 50 plus 105 28 1-2 5 1 2 0 1 s 1+0—i+9 vs. 50 plus 80 28 1-2 3 1 0 0 2 s VO CD TABLE XXVI (Continued) EXPECTED AND OBTAINED NUMBER OF SIGNIFICANT SFPI ITEMS FOR COMPARISONS OF SUICIDAL AND NON-SUICIDAL VARIABLES AT THE .05, THE .01 AND THE .001 LEVELS OF SIGNIFICANCE 1 nl 2 “2 3 Exp. .05 1 + Obt. .05 5 Exp. .01 6 Obt. .01 7 Exp. .001 8 Obt. .001 9 Sign. Marital Status Single vs. Married 53 85 1-2 5 1 1 0 0 S Single vs. Divorced 53 1+7 2-6 8 1 0 0 0 S Single vs. Married tut Separated 53 21 2-6 5 1 1 0 0 - Married vs. Divorced 85 1+7 1-2 3 1 1 0 0 - Married vs. Married but Separated 85 21 1-2 1 + 1 1 0 0 Divorced vs. Married but Separated k ' J 21 2-6 1 1 1 0 0 - Religion Protestant vs. Catholic 126 63 1-2 2 1 1 0 0 - Catholic vs. Hebrew 126 15 1-2 7 1 1 0 1 S Protestant vs. Hebrew 63 15 2-6 1 + 1 3 0 0 1 Occupation Professional vs. Clerical 33 32 1-3 3 1 0 0 0 - Professional vs. Operators 33 71 2-6 3 1 0 0 0 - Professional vs. laborers 33 67 2-6 1 + 1 0 0 0 - Clerical vs. Operators 32 71 2-6 1 1 1 0 0 - Clerical vs. Laborers 32 67 2-6 2 1 0 0 0 - Operators vs. laborers 71 67 2-6 1 1 0 0 0 - TABLE XXVI (Continued) EXPECTED AND OBTAINED NUMBER OF SIGNIFICANT SEPI ITEMS FOR COMPARISONS OF SUICIDAL AND NON-SUICIDAL VARIABLES AT THE .05, THE .01 AND THE .001 LEVELS OF SIGNIFICANCE 1 nl 2 ”2 3 Exp. .05 4 Obt. .05 5 Exp. .01 6 Obt. .01 7 Exp. .001 8 Obt. .001 9 Sign. Number of Admissions 1 vs. 2 74 54 2-6 6 1 0 0 0 1 vs. 3 74 30 2-6 2 1 0 0 0 - 1 vs. 4 72 17 2-6 3 1 2 0 0 - 1 vs. 5 74 30 2-6 4 1 2 0 0 - 2 vs. 3 54 30 2-6 0 1 0 0 0 - 2 vs. 4 54 17 2-6 5 1 2 0 0 - 2 vs. 5 54 30 2-6 0 1 0 0 0 - 3 vs. 4 30 17 1-2 2 1 0 0 0 - 3 vs. 5 30 30 1-2 0 1 0 0 0 - 4 vs. 5 17 30 1-3 3 1 1 0 0 wm 101 chance. As with the MMPI this column represents the most conservative decision possible. The results show that the only significant re sults in the comparisons of the non-suicldal variables were those of the elementaiy versus the high school, the up to 39 years of age ver sus the 50+ years of age, the *K) - k9 versus the 50+, the single ver sus the married, the single versus the divorced and the Catholic ver sus the Hebrew groups. In two of the comparisons, namely those of the married versus the married but separated and the Protestant ver sus the Hebrew groups, the results fell within chance significance limits. All other results were nonsignificant. In the comparisons of the suicide variable the results were similar to those of the MMPI. Only the comparisons between the NPC non S group and the var ious suicidal groups were significant. The comparisons among the suicidal groups themselves were well below chance significance. Table 27 shows the SPFI items which differentiate suicidal from non-suicidal patients. Thirteen items were found at the .05 level or better, 6 at the .01, and 2 at the .001 level of signifi cance. 102 TABLE XXVII Liar OF SPPI ITEMS AND THEIR OBTAINED CHI SQUARE VALUES WHICH ATTAINED THE .05, THE .01 AND THE .001 LEVELS OF SIGNIFICANCE NPC non S versus All S n=83 n=130 LEVELS OF SIGNIFICANCE Item nos. .05 .01 .001 18 8.6o 20 6.67 24 9.58 25 4.27 27 4.95 30 7.04 32 4.42 36 8.75 38 9-11 39 4.75 45 7.88 48 11.43 49 13.91 n=13 n=5 n=6 n=2 CHAPTER V DISCUSSION t No Replacement Sampling Among the results of the three levels of significance, the most outstanding were those found at the .05 level. As was discussed be fore, it is usually assumed that all 373 MMPI items are independent of one another and that the number of items that can be expected to reach chance significance is nineteen at the .05 level regardless of the sample size used. The results for this level of significance show a considerable deviation from this generally accepted assumption. An investigation of the mean number of items found to be significant by chance for samples of different size shows that there seems to be a trend that the greater number of mean items are obtained for the sam ples with the larger number of subjects. Whereas the mean number of items is nine for a sample of size 30, the mean number is twenty-two for samples of size 200, with an overall average of lU. According to this average then, the usual estimated number of 19 is on the high side.* The mean number of items found however, is very misleading. The range of the number of items reaching chance significance is a *It is of course apparent that the number of comparisons that were made are somewhat limited and that these conclusions should be subject to further verification. much more Illuminating statistic. It shows that the number of chance findings fluctuates widely, with the largest range being from 0 to k-6. These results make it clear that the assumption of accepting a finding of more than 19 items as a nonchance occurrence is often misleading. Setting 19 as a cutoff leads easily to an underestimation of the num ber of items necessaxy to reach nonchance significance, and although it can also lead to an overestimation, the error seems to lean towards the former. At the .05 level of significance, the safest assumption to make in order to estimate the number of chance findings is the largest number of items obtained in each of the differently sized sam ples. The decision whether or not this number is an overestimation should be male by other means. For Instance, the calculation of a correlation coefficient between the significant items should give an indication of the item homogeneity. A factor analysis of the items would show whether or not it would be possible to obtain psychologi cally meaningful factors. At the .01 level of significance, a somewhat different pattern from that found for the .05 level was obtained. The usual assumed number of expected chance findings of four items is an overestimation when compared with the mean number of items actually obtained. How ever, it might obviously be an underestimation if It is compared with the upper limit of items obtained. In order to make the decision whether or not obtained results can be accepted as a nonchance occur rence, the greatest number of items obtained by chance for a particu lar sample size should be taken, rather than the mean number of these 105 Items. In contrast with the findings at the .05 level of significance, the mean number of chance results at the .0 1 level appears fairly sta ble, only varying from one to four. The range of the number of items found in the different sample sizes is again considerable, from zero to ten. The picture for the .001 level of significance is different than that for both the .05 and .01 levels. The usual assumption that one significant item will be obtained by chance was found to occur in this investigation. The range of number of items obtained was from zero to two. Sampling with Replacement The results for this type of sampling are somewhat in contrast with those for sampling without replacement. In general, the mean number of items found to be significant were only slightly higher for samples of larger size than for samples of smaller size, 12 items for samples of size 30j and 16 items for samples of size 200. Fluctua tions in number of items obtained are also smaller. The upper limit appears more stable. The same trend appears to exist in all three levels of significance. As was discussed previously, these results were to be expected. These findings seem to point out that whenever the population is 600 or smaller the investigator should be aware of the fact that the method of sampling will affect his results. As a consequence his conclusions regarding the number of chance findings should be made in light of his sampling method. When the results of sampling with and without replacement are 106 compared with each other, It appears that for the size of the popula tion used in this investigation, the former method gives more stable results. At the .05 level of significance, the usually expected num ber of significant items, 19? is a definite underestimation of the number that might be found by chance if the upper range limits were used. Evaluating the results of this investigation, it seems that the use of the .05 level of significance, especiaUy in conjunction with no replacement sampling often leads to false conclusions. The most clear cut solution which would allow the investigator to make correct decisions se.ems to be the use of more stringent levels of significance. This conclusion seems reasonable in view of the fact that the results of this investigation show that it is possible to find that ten per cent of the items can be significant by chance at the .05 level of significance. At the same time, only about two per cent at the .01 level and less than one per cent at the .001 level may be found. Es pecially when a test is used as a "shotgun type" measurement device, the task of obtaining more than ten per cent significant items at the .05 level is much more difficult than obtaining only about two per cent significant items at the .0 1 level or one per cent at the .001 level. The results of this analysis seem to support the many research ers who have advocated the use of more stringent levels for determina tion of significance. The ttira The lack of ambiguity, whether or not the reaNfcte were due to chance occurrences, was unexpected and surprising. Of the 51 107 comparisons of the eight main variables and their subcategories that were made in this part of the analysis, see tables 7 and 9> there were only two comparisons where the results were ambiguous. In the other l j - 9 * the results were clear cut. Thirty-seven proved to be significant and twelve nonsignificant. All major variables except religion proved to influence MMPI responses. The results for each variable, as listed in tables 15 and 16, will be discussed separately. Diagnosis As was expected, items of the MMPI did differentiate groups of patients when they had been categorized according to various psychi atric classification categories. In this study, the patients had been grouped under four main headings. These were neurotics, psychotics, personality disorders, and others. These "others" consisted mostly of cases with brain syndromes but actually Included anything not in the first three categories. It has generally been recognized that the MMPI differentiates normal from non-normal individuals. There have been arguments whether or not it will also differentiate different types of non-normality. These results show that the MMPI does Indeed differentiate among non-normals when these are grouped under major psychiatric classifications. Considering the number of items found to be significant for each of the various cocaparisons, it appears that the results show marked agreement with clinical conceptions of the various types of mental illness. For instance, broadly speaking, clinically a neurotic would be considered as being more like someone with a personality disorder than like a psychotic, etc. The findings 108 here support such contentions. They also show that the greatest dif ference lies between neurotic and psychotic patients. Religion Comparisons on this variable showed that type of religion does not influence MMPI responses. This is, of course, not unexpected. The various religious classifications were based on religious prefer ence or parental background, but not on active participation. It is debatable whether or not the results would have been different if the subjects had been grouped for active participation under type of re ligion. It seems more likely that differences would show up between nonreHgious and actively religious persons who are grouped according to type of religion. Education As has been discussed earlier, a few investigators have found that MMPI responses are Influenced by the level of intelligence of the subject. In lieu of IQ scores, one could consider the educational achievement of a person as a coarse measure of his general intelli gence. Regardless, grouping subjects according to educational achieve ment does influence responses to the M4PI. Again the results follow logical expectations. As one would expect, the greatest difference in number of items is between persons with elementary and those with col lege education. Smaller differences although still greater than chance expectations, are found between the other two comparisons, elementary versus high school and high school versus college 109 education. Differences in response seem to indicate that groups with different educational backgrounds either use different verbalizations or different behavioral responses to similar situational and psycho logical stresses. Occupation It should be mentioned that the occupational classifications are not so precise as, for instance, the age and marital status clas sifications. In seme instances the person had been classified accord ing to the Job classification they had given themselves. That is, in a number of instances it could not be established what the actual du ties of the person had been during his last employment. It may be, therefore, that the Job classification given was the type that the person would have liked to have, or that it was the maximum position that he ever had obtained. Even so, the results are somewhat unexpected. It is true that it can be argued that Job attainment is an indication of the overall, intelligence of a person, but this relationship is certainly not so clear as the one for intelligence and educational achievement. Yet grouping subjects according to different occupational classifications does show differences in MAPI responses. Except for the operator ver sus laborer comparison, where the results are somewhat doubtful, all other comparisons proved to exceed chance expectancies. Here too, all results fall in a logical direction. Again we find that the greater differences in number of items are where one would expect them to be. The disparity between professionals and laborers for instance, is 110 greater than between professionals and operators and this in turn Is less than between professionals and clericals. The Interpretation of the findings here is difficult in terms other than general intelli gence where one would expect differences to lead to dissimilar re actions to stress situations. The answer to this problem may be found in future research studies involving this variable. Marital Status The results of the comparisons within this classification showed that all but the divorced versus the married but separated, and the married versus the married but separated comparisons were sig nificant. The number of items which differentiate these last two com parisons fell just within the upper range of chance findings. The am biguity whether or not these groups are different from one another seems to indicate that the difference between being divorced and being married but separated involves more than simply a difference in legal status. What is somewhat surprising is the fact that the MMPI should show differences at all between persons grouped according to marital status. The results show that the greater number of significant items were obtained between those group comparisons where one would expect the greatest difference to occur. A greater number of items differen tiated the single and married groups than the single and divorced or single and married but separated groups. Why were differences found at all in this category? The ex planation would seem to lie in the MMPI itself. The MtPX items were originally collected as indicators of adjustment in 26 behavior areas. Since the items were then grouped into scales which would measure or classify people according to various psychiatric diagnoses we have fallen into the habit of thinking that the MMPI measures psychiatric syndromes only. However, since the items originally intended to measure adjustment in various behavior areas, some of these items would play a role as indicators of marital adjustment. This would explain the differences in marital adjustment between such groups as married versus divorced, but what about differences in MMPI response between single and married groups? Can one infer from these findings that either the single or married group is less well adjusted than the other as measured by the MMPI? Such an Inference is, of course, high ly questionable in a general nonhospital population. In the popula tion investigated here, however, there could be a more selective fac tor operating. It is unlikely that the average age of the single group is different from the other groups since they all are veterans. If they are of the same age range, every person had the same oppor tunity for marriage. The reason that the single group did not marry may very well be a matter of adjustment and this could show in H4PI responses. An inspection of the results shows that the single group seems indeed the most maladjusted group— even more maladjusted than the divorced and separated groups. Presumably, the married and also the divorced and separated groups have had a period in which their adjustment at least allowed them to attempt to establish permanent interpersonal relationships. The adjustment of the single group never allowed such involvement. They would seem to be more chronically 112 maladjusted. A g g . This variable too shows its influence on JMFI responses. The greatest number of significant items was obtained between the up to 39 year age group and the 50 plus year group. The differences between the other two comparisons, up to 39 versus ^0-^9 and ^0-^9 versus 50 plus, were less pronounced, but they are nevertheless greater than chance expectancy. The explanation of the obtained differences again seems to lie in the adjustment factor. Presumably, the older veterans have had a greater opportunity to work out their adjustment problems. Those who did not succeed would be the more chronically maladjusted. They remained or returned to the hospital. Those that were able to adjust left or did not return. This fact would tend to make this old er group more homogeneous. The younger veterans, on the other hand, are more likely to be less homogeneous as a group insofar as the seri ousness of their adjustment is concerned. Natural selection has not had a chance to take place yet. The fact that the 50 plus age group is also different from the 40-49 year old group seems to indicate that middle age especially is a critical time as far as future prospects for adjustment are con cerned. One can conclude that if the patient is unable to achieve a satisfactory adjustment level during this age period, his chances of needing permanent hospital care seem to Increase considerably and his prognostic outlook would seem pessimistic. 113 Number < ~ > f Admissions Although the results were not altogether unexpected, they are nevertheless interesting. It is clear that a definite qualitative difference exists between patients with one hospitalization and those with two. Another definite change however seems to occur when a per son has 3 or ^ hospitalizations. The results show that the change be tween one versus two and between one versus three or four hospitaliza tions is quite striking. The three or four hospitalizations group seems to consist of already quite seriously and probably chronically ill persons. This seems to be borne out when this group is compared with those who have five or more hospital, admissions. A comparison between these two groups showed that there is no difference between them. It should be pointed out that it can be argued that the results of the analysis of this variable is a function of the number of times that a person has responded to the MMPI. Presumably persons with more admissions have had a greater probability of having had multiple M4FI administrations than someone with only one or two admissions. This however was not borne out in the study by Layton (59) discussed ear lier, who Investigated short tern changes. He found that the basic profile pattern did not change significantly within the time period of his investigation. Further research needs to be undertaken to deter mine how multiple administrations of the MMPI influence the test over long time periods. 114 Grenerftl Conclusion What do these various results of the analyses of the different variables indicate? The main implication seems to be that responses to the MAPI are not as straightforward or uncomplicated as has been thought. The current practice of scoring proves to be much too ele mentary. This should come as no surprise, of course. In such areas as aptitude, performance and achievement measurement, the percentage of correct inferences from test results are much higher than In the area of clinical measurement. In the former areas it has become cus tomary to use weighted or complex scoring formulas, often different formulas for different groups of people. There Is no reason to assume that this approach is not necessary in the field of clinical measure ment. On the contrary, this area has been overdue for more objective scoring systems. Many psychologists have pointed this out. Several studies have shown the advantages of applying statistical techniques to the scoring and interpretation of the tftfPI* The results of this Investigation support this point of view. They also question the habit of describing persons in terms of psychiatric syndromes. The influence of such variables as education, occupation, marital status, age and number of hospital admissions in addition to the influence of diagnosis seems to indicate that collecting items according to adjustment to be havior areas may be a more fruitful approach in understanding the in dividual and his problems than the collection of items according to de gree of agreement with psychiatric diagnoses. It may be argued that the MMPI is able to supply the clinician 115 with information regarding an individual‘s personality and adjustment from the current method of analysis. If this is the case, such con clusions are only reached via the 9 psychiatric classification scales since this is the present fashion of summarizing MMPI responses. Now that electronic computers are relatively widely available, it has be came possible to consider seriously the advantages inherent in more detailed analyses. Additional research will undoubtedly clarify whether or not this approach is an improvement over the present custom of MMPI analysis. Suicidal Behavior It will be recalled that the classification of the suicidal status of a person was made on the basis of infomation obtained from his clinical record. If any suicidal behavior had occurred at any time during the life of a person he was labeled suicidal, and was there fore included with the suicidal sample. Another alternative selection method would have been to include only those subjects in the suicide sample who had a MMPI admission within a certain time period before or after any signs of overt suicidal behavior. These two possible selection methods lead to two different measurements of so-called suicidal!ty. There are several possible ways in which the occurrence of suicidal behavior could be postulated. One way Is to consider suicidal behavior as time bound, as something that comes and goes. It is something that can happen to anyone. One either has it or does not have it, but while it occurs, emotional dis turbances are increased. Another way would be that suicidal behavior 116 comes and goes with concomitant Increased emotional disturbance, but It does not happen to anyone. It only happens to persons with certain personality characteristics. It could be that such personality charac teristics are of a specific suicidal nature such as a suicide trait. It could be that such personality characteristics are quantitatively similar to the personality characteristics of non-suicidal persons but that these characteristics are qualitatively different. The third way in which the occurrence of suicidal behavior could be postulated is similar to the second, but without the concomitant increase in emotion al disturbance. Here, suicide behavior is just another method of re sponse, one that can occur whenever certain specific stimuli are pre sent. This study investigated the possibility that suicidal behavior is qualitatively different from non-suicidal behavior. As a result, the first method of selecting suicidal subjects was followed. If dif ferences are found between non-suicidal and suicidal groups with this particular method of assigning subjects, then one can assume that this difference is the result of a relatively enduring personality pattern. The results indicated that in both the original and the repli cation study, a significant number of MMPI responses of the NFC non S group were different from those of the various suicidal groups. In view of this fact then, one can postulate that whatever differentiates these two groups seems to be of a stable and enduring nature. The re sults also showed that no differences are apparent among the three sui cidal groups themselves. This finding does of course not preclude the possibility that suicidal behavior Is concomitant with increased 117 emotional disturbance during the time of overt occurrence. It may be that when suicide behavior Is measured shortly before or after Its oc currence, that either a somewhat different overall personality pattern may be found than the one currently obtaine I, or that the same pattern is apparent but with certain specific increased emotional disturbances. Either of these possibilities could of course occur in different pro portions for the three types of suicidal groups. They do not have to be the same for each group. In other words, when suicidal behavior is measured by means of the JMPI within a short time before or after its occurrence, the test responses could be different for the various types of suicidal behavior. The results obtained in this investigation only show that whatever occurs at the time of overt suicidal behavior, it occurs in addition to an already existing, seemingly enduring, dif ference in personality pattern between suicidal and non-suicidal per sons. Regardless of how one perceives the occurrence of suicidal be havior, these MMPI results seem to show that such a personality make up, whether enduring or not, is potentially measurable. It is also interesting to note from the findings in this study that there is no substitute for using samples of large size, see table l6. The results show that the larger the sample, the more certain one can be that significant results were not obtained due to chance occur rences. This certainty seems to show itself in increases in the num ber of items found at the .0 1 and .001 levels rather than in increases of items at the .05 level. For instance, in the comparison of "all 118 non-suicidals" with a N of 283 versus "all suicides" with a N of 309, we find that of the overall total of k2 per cent, approximately I1 * per cent of the 1MP1 items are significant at the .001 level and another 13 per cent at the .0 1 level. This increase In significant findings at more stringent levels can clearly be seen when the results of the orig inal NPC (n = 1^2) versus the original TS (n = 77), the replication analysis (n-j_ = lUl, n = 7?) and the all NPC (n = 283) versus the all TS (n = 154) comparisons are examined. For the above three sets of comparisons, the results are 16, 17 and 26 per cent for the original., replication and "all NPC" versus "all TS" respectively at the .05 level of significance. They are J, 6 and Ih- per cent at the .01 level and 1, 2 and 6 per cent at the . 001. The effect of sample size shows that while the increase at the .05 level is from 16 to 22 per cent, the in crease at the more stringent .0 1 level is from 7 to 1^ per cent and at the .001 level from 1 to 6 per cent. It is clear that the habit of using «rn«n sized samples often leads to uncertain results at less stringent levels of significance. In view of the finding that other variables besides psychiatric diagnosis and suicide affect M4F1 responses, it is not possible to de rive a suicide scale which can be scored in the usual MMPI fashion. Any scale of this type will need differential weighting for its items. First, the contributions of other variables that influence the re sponse to the same item must be determined, as well as the contribution of that item to all suicide items combined. Only then is it possible to begin to develop a suicide scale which applies optimum scoring 119 formulas. This in turn would allow the clinician to make the most ac curate prediction possible. The ability to make use of all available test information is of course of prime Importance in the prediction of infrequently occurring events such as suicide. The SPPX It will be recalled that this test was specifically constructed to measure suicide. Ideally, therefore, only suicidal behavior should influence responses to this test. What do the results show? First of all, some non-suicidal variables do influence SFPI responses. Table 26 shows that, surprisingly, the religious variable does indeed influence SPPI responses. Catholics seem to respond somewhat differently than JewB. Differences between the responses of Protestants and Jews are not quite so clear nor so pronounced. No differences were obtained in the Cathollc-Protestant comparison. Marital status too seems to influence SPPI response. In this variable, the single group can be differentiated from both the married and divorced groups. Otherwise, no additional comparisons of the mari tal status subcategories gave any significant results. Age was another variable which contributed to dissimilarity of SPPI response. Differences were apparent between the 50 plus age group on the one hand and the other two age groups on the other hand. Among the various educational achievement comparisons, only the elementary versus high school analysis showed any results. The interpretation of these findings should wait until repli cation studies have confirmed the results. The significance of the 120 majority of the findings which were obtained from analyzing the vari ous control variables was based on results at the .05 level. As was pointed out in the discussion of the MMPI analysis, conclusions based on such a low level of significance are uncertain and open to doubt, especially in cases where cross validation is lacking. When these re sults have been verified, they may be explained through an examination of the significant items. For Instance, when the items that differen tiate the Jews on the one hand and the Protestants and Catholics on the other hand are examined, an interesting pattern seems to exist. Among these significant items are three which deal with alcoholism and one with the availability of assistance from friends. It has been stated that alcoholism among Jews is considerably less than among other religious groups. Similarly, the social structure of the Jewish community may be of such a nature that help of friends is fairly read ily available for those who need it. This could be in contrast to the non-Jewish community structure such as Protestant and Catholic groups. In these latter there may be little opportunity to build lasting close knit ties. As a result the opportunity to receive help from friends would be lesB. It may be therefore, that the differences obtained in the analysis of the religious variable do not reflect differences in religion per se but rather differences in social and ethnic group structure. In the comparisons of the non-suicidal and the various suicidal groups we find much the same pattern as was obtained with the MMPI analysis. Here too It appears that there are only differences between 121 non-suicidal on the one hand and suicidal groups on the other hand, but not among the suicidal groups themselves. The comparison of the NPC non S versus the TS+AS group gives the most pronounced results. This was true also In the JMPI analysis when the "all NPC non S" group was compared with the "all TS+AS" group. This fact is of same interest. It could suggest that this TS+AS group may be somewhat more disturbed than the others. However, the comparisons of this group with the other two suicidal groups indicate that no differences are apparent among the three suicidal groups. Any conclusion therefore regarding greater dis turbances among this group than among the other suicidal groups is ten uous. Unfortunately, no comitted suicide group was available for com parison with the other three suicidal groups. Statistics of the pro portion of persons in each group who go on to commit suicide are not available either. It may be that, although differences are not ap parent among the three suicidal groups investigated, the TS+AS patients have a higher committed suicide rate. It could also be that suicidal behavior may indeed show increased emotional disturbances when measured at the time of overt suicidal behavior. In addition, It could suggest that these increased disturbances are different for the various sui cidal groups. If these possibilities would be corroborated, then the proposed suicidal classification system would indeed prove to be a log ics!. extension of the usual division of threatened, attempted and com mitted suicides. The comparison of the "all NPC non S" versus the "all suicide" group was, as with the MMPI, a logical step since no significant 122 differences were found among the suicidal group comparisons. The re sults of the ’ ’ all UPC non SM versus "all suicide" comparison shows that twenty-six per cent of the items were found to be significant at the .05 level, twelve per cent at the .01, and an additional four per cent at the .001 level. The groups used in this study were fairly large, 83 versus 130, and the findings are therefore presumably fairly stable. When the results of the SPPI analyses are compared with those of the MMPI, it is apparent that the former are not so productive in producing discriminating items between suicidal and non-suicidal groups as the latter. The number of items found to be significant in the MCPI warn forty-two per cent but only twenty-two per cent was found in the SPPI. This would appear to indicate that in this study a shot gun type of testing approach is better than a theoretically based testing approach. There is, however, an explanation why such a large discrepancy was found in favor of the MMPI. It has been pointed out before that when the behavior areas of the MIFI are compared with the areas found in the critical incident study, most areas are represented in both. Many of the findings in the critical incident study had, for various reasons, already been incorporated in the MMPI. It is there fore not surprising that a greater proportion of significant items should be found in the MMPI. When cross validation studies support the findings in both the M4PI and the SPPI, It would seem that a theoretically based testing approach does have advantages. The results do indicate that variables other than suicide have considerably less Influence In the SFPI than In the M4PI. It Is debatable however, whether It would be more expedient to develop a specific suicide test such as the SPPI, or whether It would be more useful to spend the time in establishing special suicide scoring formulas for the MtPI. This investigator believes that the latter course of action would, be more productive, especially In view of the fact that the MMPI is a widely used test. CHAPTER VI SUMMARY AND CONCLUSIONS Life Is the art of drawing sufficient conclusions from insufficient premises. Note-Books. Lord, What is Man? Samuel Butler It is unfortunate that the suicidal person draws insufficient conclusions from insufficient premises and that this leads him to act out self-destructively. Although many persons from many disciplines have investigated why this would be, answers and satisfactory solu tions have yet to be found for this perplexing phenomenon. The major ity of the studies have been of descriptive and often speculative na ture. Experimental and systematic researches in this area have been scarce. The difficulty of obtaining subject material which satisfies scientific demands is of course readily apparent. In many ways there fore, research in this area is of an exploratory nature. It is with this type of research that the investigator should be especially aware of the fact that many methodological issues may arise when he approach es his problem. This investigation was designed to explore such methodological issues together with the main goals of furthering understanding of suicide. The goals were: (l) the discrimination of non-suicidal from suicidal neuropsychiatric hospital patients by means of two 125 personality inventories; (2) the discrimination of different types of suicidal neuropsychiatric hospital patients, such as, threatened sui cides, attempted suicides, and a camhination of threatened and at tempted suicides, by means of the same tests; (3) the determination of specific test items which may be used in the description of suicid al personality characteristics. The methodological problems which were analyzed concurrently, involved: (l) a normative versus a clinical approach; (2) replace ment versus no replacement sampling; (3) the number of chance findings at the . 05, the .01 and the .001 levels of significance for samples of different size; (M the influence of so-called control variables; and (5) the definition of suicide. A total of 600 neuropsychiatric veterans was used in this study. The MMPI had been administered to all. In addition to the MMPI, 220 had also received a specially constructed Suicide Potential Personality Inventory. The MMPI was chosen as representing a "normal" testing approach and the SPPI was developed to represent a "clinical" testing approach. The subjects were then sorted into various groups. The basic design consisted of analyzing two groups at a time. For each group the "true" and "false" frequency for each test item was determined. Differences in group frequency for each item were then evaluated for significance by means of 2x2 chi squares and the number of significant items obtained at the . 05, the .01 and the .001 levels was determined. The 600 patients were sorted for eight main variables, one at a 126 time. Each main variable was further subdivided into several cate gories. The main variables were: (l) diagnosis; ( 2) religion; (3) education; (U) occupation; (5) marital status; (6) age; ( 7) number of V.A.H. admissions; and (8) suicide status. For the last variable an original as well as a replication study was done. In order to investigate chance significances, samples were drawn at random from the population of 600. As before, two samples were compared at a time. This process was repeated a number of times for each of six different sample sizes. This procedure was followed for two methods of sampling. These were with replacement and without replacement sampling. The results showed that the mean number' of chance findings is less than the usual estimated number at the . 05, the .01 and the .001 levels of significance. The range of the number of chance findings is larger however. Using the usual estimated number of expected chance findings will lead to a faulty acceptance of the results as significant approximately fifty per cent of the time. The number as well as the range of chance significance is less stable for sampling without re placement than for sampling with replacement. Sample size influences the results mostly in sampling without replacement. The analysis of the non-suicidal variables shows that all vari ables except religion influence JMPI responses. All except occupation and number of hospital admissions also affect SFPI responses. Grouping the subjects for suicidallty results in significant differences between the non-suicide group on the one hand and the suicide group (made up of 127 threatened suicides, attempted suicides and threatened + attempted sui cides) on the other hand, but not between the three subcategories of the suicidal group. The finding that five other variables, namely edu cation, occupation, marital status, age, and number of V.A.H. admis sions besides diagnosis and suicidal status, affect MAPI responses, suggests that the current MMPI scoring methods are unsatisfactory for optimum prediction purposes. The main results can be summarized as follows: (1) The usual estimation of the number of expected chance find ings leads to faulty conclueiens approximately fifty per cent of the time; (2) Sople size affects the number of significant findings; (3) The method of sampling affects the variability of the num ber of findings; (1) Suicidality, education, occupation, marital status, age, and number of V.A.H. admissions all affect MMPI responses far above chance occurrences; (5) Threatened suicide, attempted suicide and threatened + at tempted suicide cannot be distinguished from each other on the basis of responses to items; (6) Suicide patients as a group can be distinguished from non- suicide patients. It is obvious from the present Investigation that many diffi culties confront the researcher who desires to Identify and predict suicidal behavior. This study however, also shows that using 128 appropriate controls and clear methodological procedures Improves con siderably the possibilities for future success. KTiiVRBTTfJfTKfi REFERENCES 1. Alf, E. F., Jr., MacEwan, Charlotte, Conaway, Jo. Configural scoring and prediction. Seattle, Washington: University of Washington, 1956. 2. Ballard, J. W. Suicide. Delaware st. med. J.. 1938> 10, 7^-77* 3. Batchelor, I. R. C. Alcoholism and attempted suicide. J. ment. Sci.. 195^, 100. h51-h6l. 1 + - . Batchelor, I. R. C. Attempted suicide (One of emergencies in gen eral practice). Brit, med. J., 1955* 1, 595-597* 5. Batchelor, I. R. C., and Napier, Margaret B. Attempted suicide in old age. Brit. med. J.. 1953> 2, II86-H 9 0. 6. Batchelor, I. R. C. Psychopathic states and attempted suicide. Brit, med. J.. 195^ h 13^2-13^7* 7. Batchelor, I. R. C. Repeated suicidal attempts. Brit. . 1 . med. Psvchol.. 195^, 21, 158-163. 8. Bell, C. R., Bond, H. et al. Suicide: social, forensic, and psy chological aspects. Mi lit. Surg.. 1932, 71. 399-^-03* 9. Benham, H. A. Seme remarks on suicides in public asylums. J. went. Sci.. 1903, kkl-k53. 10. Bennett, A. E. Prevention of suicide. Calif. Med.. 1 9 5 81, 396- kOl, 11. Bennett, A. E. The physician's responsibility in the prevention of suicide. Pis, nerv. Svst.. 1951 ** 15. 207-210. 131 12. Bergler, E. Problems of suicide. Psychiat. quart. SuttpI. , 19^*6, 20 . 261-275 • 1 3. Block, J. On the number of significant findings to be expected by- chance. Psychometrika. i960, 2 5, 369-380* 14. Brend, W. A. The mental condition preceding suicide. Practition er. 1919, 103. tol-JOi. 1 5. Bromberg, W., and Schilder, P. The attitude of psychoneurotics towards death. Psychoanal. Rev.. 1936, 23. 1-25* 16. Brower, D. The relation between intelligence and MMPI scores. J, soc. Psychol.. I9I J - 7, 2243-2^5* 1 7. Brozek, J., and Keys, A. Personality changes with age: an item analysis of the IMP1. Amer. Psychologist. 1951, §j 397• 18. Buckle, D. Attempted suicide. Med. . 1 . Aust.. 19^2, 1, 15^-155* 19. Carr, J. L. Suicide. Int. rec. Med.. 1957? 170. 1613-1620. 20. Chrichton^Miller, H. The psychology of suicide. Brit, med. J.. 1931, 2, 239- 2^1. 21. Cleaver, J. H. Prevention of suicide. Med. Her.. St. Joseph, 1930, j + 2, Ml 9 5. 22. Dixon, W. J., and Massey, F. J., Jr. Introduction to statistical analysis. New York: McGraw-Hill Book Co., Inc., 1957* 2 3. Drewey, P. H., Jr. Seme aspects of suicide. Virginia med. Mon., 19te, 62, 252-256. 2b, Durkheim, Qnile. Suicide: A study of sociology. Translated by John Spaulding and George Simpson. Glencoe, 111.: The Free Press, 1951* 2 5. East, G. R. A case of melancholia with attempted suicide. Brit. med. J.. 1911, 1, 870. 2 6. Estes, W. K. and Associates. Modern Learning Theory. New York: Appleton-Centuxy-Crofts, Inc., 195**-• 27- Farberow, N. L. Suicide: The gamble with death. Paper pre sented at the Los Angeles County Psychological Association, March, 1962. 2 8. Farberow, N. L. Personality patterns of suicidal mental, hospital patients. Genet, nsychol. Monogr.. 1950, ^2, 3-79* 29- Farberow, N. L., and Schneidman, E. S. The cry for help (bibli ography on suicide). New York: McGraw-Hill Book Co., Inc., 1961. 30. Farrar, C. B. Suicide. J. clin. exp. Psvchonathol.. 1951* 12. 79-88. 31. Feller, W. An introduction to probability theory and its appli cations, vol. I, 2nd ed. New York: John Wiley and Sons, Inc., 1961. 32. Finn, M. E. Study in suicidal attempts. J. nerv. ment. Pis.. 1955* 121. 172-176. 3 3. Flanagan, J. C. The critical incident technique. Psychol. 1 - - 195^, 2 1 , 327-358. 3k. Freedman, Ronald. Incomplete matching in ex post facto studies. Amer. . 1 . Soclol.. 1950, 2 1 , ^85-^8 7. 3 5. Freud, Sigmund. Mourning and melancholia, (in collected papers, vol. * * ■ ) London: Hogarth Press, 1956. 133 36. Friche, B. G. A coded profile method for predicting achievement. Educ. nsvchol. Measurement. 1957, 1£, 98-104. 37. Friedlander, K. On the * longing to die'. Int. . 1 . Psycho-anal.. 1940, 2 1, 416-426. 3 8. Glover, E. Same observations on suicidal mechanisms. Psychoanal. Rev.. 1928, 25j 89-90. 39. Gordon, R. G. Certain personality problems in relation to mental illness with special reference to suicide and homicide. Brit. . 1 . med. Psychol., 1929* 2j 60-6 6. 40. Greenbauh, B. V. Communication of suicidal thoughts. Can ad. •psvchlat. assoc. J.. 1957# 2, 150-153. 41. Holbuachs, Maurice. Lea causes du suicide. Paris: Libraire Felix Alcan, 193°. 42. Harbin, R. M. Suicide. Med, rec. N. Y., 1906, 70. 462. 43. Hathaway, S. R., and McKinley, J. C. Manual, revised 1951# Min nesota multiphasic personality inventory. New York: The Psy chological Corporation. 44. Hathaway, S. R., and Meehl, P. E. An atlas for the clinical use of the M4PI. Minneapolis: University of Minnesota Press, 1951* 45. Hendin, H. Attempted suicide: A psychiatric and statistical study. Psvchiat. Quart.. 1950, 24, 39-46. 46. Hendin, H. Suicide. Psychiat. Quart.. 1956, 30. 267-282. 47. Hilgard, E. R. Theories of learning. 2nd ed. New York: Apple ton-Century-Crofts, Inc., 1956. 48. Hofftnan, F. L. Suicide. Ref, hs^dh- mftd. Sci.. New York, 1917# 8, 13-18. 4-9* Holzberg, J. D., Cahen, Eleanor R., Wilk, E. K. Suicide: A psy chological. study of self destruction. J. nro.l. Techn.. 1951* 12, 339-351 *. 50. Ireland, W. W. On the causes of the increase of suicide. J. meat. Sci.. 1899* 4-51-4-66. 51. Jackson, D. D. Suicide. Sci. Amer.. 1954-* 191. 88-9 6. 52. Jamieson, G. R. Suicide and mental disease: A clinical analysis of one hundred cases. Arch, neurol. Psychiat.. Chicago, 1936, 36, 1-12. 5 3. Jamieson, G. R. f and Wall, J. H. Same psychiatric aspects of sui cide. P«veh-tf t - h- OnftT*. . 1933, 2, 211-229. 54-. Kallman, F. J., and Anastasio, Mary M. Twin studies on the psy chopathology of suicide. J. Hered.. 194-6, 21* 171-180• 55. Kallman, F. J., and Anastasio, Mary M. Twin studies on the psy chopathology of suicide. J. nerv. ment. Pis., 194-7* 105. 40-55. 5 6. Kilpatrick, Elizabeth. A psychoanalytic understanding of suicide. Amer. . 1 . Psy choanal.. 1948, 8* 13-23. 57* Kiorboo, E. Suicide and attempted suicide among old people. J. Geront., 1951* 6, 233-236. 58. Kupper, W. H. A study of suicidal soldiers in the European thea ter after V-E day. J. nerv. ment. Pis., 19^7* 105. 299-303* 59* Layton, Wilbur L. The variability of individuals' scores upon successive testings on the M'iPl. Educ. •psychol. Measurement. 1954, 16* 634-640. 135 60. Lewis, N. D. C. Studies on suicide. I. Preliminary survey of some significant aspects of suicide. Psychos^ - R«v-. 1933* 2 0. 241-273. 61. Lewis, N. D. C. Studies on suicide. II. Same comments on the biological aspects of suicide. Psychoanal. Rev.. 1934, 21. 146-153. 62. Lindquist, E. F. Design and analysis of experiments in psycholo gy and education. Boston: Houghton Mifflin Co., 1956. 63. Lowrey, L. G. An analysis of suicidal attempts. J. nerv. ment. Die., 1920, 5 2, 475-481. 64. Lubin, A., and Osbum, H. G. A theory of pattern analysis for the prediction of a quantitative criterion. Psychcmetrika. 1957, 2 2. 63-73. 65. McEvqy, Th. L. A comparison of suicidal and nonsuicidal patients by means of the Thematic Apperception Test. Ph.D. Dissertation, University of California at Los Angeles, June, 1963. 66. McQuitty, L. L. Isolating predictor patterns associated with ma jor criterion patterns. Bduc. psychol Measurement. 1957, 17. 3-42. 67. McQuitty, L. L. A pattern analysis of descriptions of "best” and "poorest" mechanics compared with factor analytic results. Psy chol. Monogr. 446. 1957, H, 1-24. 68. Menninger, Karl A. Man against himself. New York: Harcourt, Brace and Co., 1938. 69. Menninger, K. A. Psychoanalytic aspects of suicide. Arch. Psychlat.. Chicago, 1931, 2£, 1369-1372. 70. Menninger, K. A. Psychoanalytic aspects of suicide. Int. . 1 . Psv- choanal.. 1933, 2k> 376-390. 71. Michael, W. B. Development of statistical methods especially use ful in test construction and evaluation. Rev, educ. Research. 1959, 22, 120-122. 72. Mills, C. A. Suicides and homicides in their relation to weather changes. Aner. . 1 . Psvchiat.. 193^, 91. 669-677- 73* Moore, Charles, Rev. A full inquiry into the subject of suicide. To which are added (as being closely connected with the subject) two treatises on duelling and gaining. London: J. F. and C. Rlvington, 1790. Jk. Morselli, Henry. Suicide: An essay on comparative moral statis tics. London: C. Kogan Paul and Co., l88l. 75- Muhl, Anita M. America*s greatest suicide problem. A study of over 500 cases in San Diego. Psvchoanal. Rev.. 1927, 1^-. 317- 325- 7 6. O'Hara, W. E. A case of suicide, Lancet. 190*4-, 2, 1178. 7 7. Oliven, J. F. Suicide prevention as a public health problem. Amer. . 1. trubl. Health. 195*4-, 1*4-19-1*4-25. 7 8. Osgood, Ch. E., and Suci, G. J. A measure of relation determined both by mean difference and profile information. Psychol. Bull.■ 1952, UQj 251-262. 7 9. Pollack, B. A study of the problem of suicide. Psychlat. Quart.. 1938, 12, 306-330. 137 80. Rand. Corporation. A million random digits with 100,000 normal deviates. Glencoe, 111.: Free Press, 1955* 81. Read, C. S. The problem of suicide. Brit, med. J.. 1936, 1, 631- 634. 82. Robins, E., Schmidt, E. H., O'Neal, Patricia. Same Interrela tions of social factors and clinical diagnosis in attempted sui cide: A study of 109 patients. Amer. . 1. Psvchiat.. 1957, 114. 221-231. 8 3. Rosen, A., Hales, W. M., Simon, W. Classification of ’ ’ suicidal" patients. J. consult. Psychol.. 1954, 18. 359-362. 84. Schmideberg, Melitta. A note on suicide. Psychoanal. Rev*. 1948, 35. 181-182. 8 5. Schmidt, C. F., and Van Arsdol, Jr., M.D. Completed and attempted suicides: A comparative analysis. Amer. soc. Rev.. 1955, 20. 273-283. 8 6. Shapiro, L. B. Suicide: Psychology and familial tendency; Report of a family of suicides with history and discussion. J. nerv. ment. Pis.. 1935, 81, 547-553. 8 7. Shields, E. A. Depression, then suicide. Amer. . 1 . Nurs«. 1946, 46. 677-679. 8 8. Schneidman, E. S. Orientations toward death: A vital aspect of the study of lives. In The study of lives, R. W. White (ed.). New York: Atherton Press, 1963* 8 9. Simon, Werner, and Hales, W. M. Note on a suicide key in the M4PI. Amer. . 1. Psvchiat.. 1949, 106, 222-223* 90. Stekel, S. A. On suicide. Amer. , 1. Urol.. 1918, 337-350. 91. Stengel, E. The risk of suicide in states of depression. Med. £r*, 1955, 234. 182-184. 92. Stengel, E., and Cook, Nancy G. Recent research Into suicide and attempted suicide. J. forensic Med.. 1954, 1, 252-259* 93* Strecker, H. P. Body-length - leg ratio in the general popula tion and in mental hospital patients and its possible signifi cance in suicide. J. mynt. Sci.. 1936, 82. 38-42. 94. Sullivan, W. C. Cases of suicidal Impulse in conditions of cere bral automatism. J. ment. Sci.. 1899, 4£, 338-343. 95* Sullivan, W. C. Relation of alcoholism to suicide in England with special reference to recent statistics. J. ment. Sci.. 1900, 46, 260-281. 9 6. Webster, H. A note on profile similarity. Psychol. Bull.. 1952, 42, 538-539* 9 7. Wexner, Lois B. Relationship of intelligence and the nine scales of the MMF1. J. soc. Psychol.. 1954, 40. 173-176. 9 8. Whipple, J. W. A study of the extent to which positive or nega tive phrasing affects answers in a true-false test. J. educ. Reg.*, 1957, 53* 59-63* 9 9. Winfield, D. L. The relationship between IQ scores and M4FI scores. J. soc. Psychol., 1953, 3 8, 299-300. 100. Zilboorg, G. Differential diagnostic types of suicide. Arch. neurol. Psvchiat.. Chicago, 1936, 35, 270-291* 101. Zilboorg, G. The study of suicide. Med. Rec.. N. Y.. 1938, 147. 139 ( 3) in: Open Forum, Februaiy 2; 9f Uj 13* APPENDIX TABLE XVII CODE FOR COMPARISONS SHOWN IN TABLE XVIII code meaning NR No replacement. This Is used in conjunction with the different size samples, viz., NR30, NR^O, NR50, NRlOO, NR150, and NR200. WR With replacement. This is used in conjunction with the different size samples, viz., WR30, WRl* 0, WR50, WR100, WR150, and WR200. Dxl-2 neurotic vs. psychotic Dad-3 neurotic vs. personality disorder Dxl-^ neurotic vs. other Dx2-3 psychotic vs. personality disorder Dx2-k psychotic vs. other Dx3-^ personality disorder vs. other Rel-2 Protestant vs. Catholic Rel-3 Protestant vs. Hebrew Re2-3 Catholic vs. Hebrew Edl-2 elementary vs. high school Edl-3 elementary vs. college Ed2-3 high school vs. college Ocl-2 professional vs. clerical Ocl-3 professional vs. operator Ocl-4 professional vs. laborer 0c2 -3 clerical vs. operator 0c2-4 clerical vs. laborer 0c3-U operator vs. laborer Ms1-2 slgle vs. married Msl-3 single vs. divorced Ms 1-4- single vs. married but separated Ms2-3 married vs. divorced Me2-k married vs. married but separated Ms3“^ - divorced vs. married but separated 142 TABLE XVII (Continued) CODE FOR COMPARISONS SHfWN IN TABLE XVIII code meaning Agl-2 up to 39 vs• 40 - 49 Agl-3 up to 39 vs. 50 and up Ag2-3 40 - 49 vs. 50 and up #1-2 1 vs. 2 admissions #1-3 1 vs. 3 on 4 admissions #1-4 1 vs. 5 or more admissions #2-3 2 vs. 3 or 4 admissions #2-4 2 vs. 5 or more admissions #3-4 3 or 4 vs. 5 or more admissions Sjl-2 original NFC non S vs. original TS S-jl-3 original NPC non S vs. original AS S-j_l-4 original NPC non S vs. original TS+AS S-j_2-3 original TS vs. original AS S^-4 original TS vs. original TS+AS S- j _ 3-4 original AS vs. original TS+AS Sgl-2 replication NFC non S vs. replication TS S2I-3 replication NFC non S vs. replication AS 521-4 replication NPC non S vs. replication TS+AS 522-3 replication TS vs. replication AS 522-4 replication TS vs. replication TS+AS 523-4 replication AS vs. replication TS+AS SC1-2 combined NFC non S vs. combined TS SC 1-3 combined NFC non S vs. combined AS SC1-4 combined NFC non S vs. combined TS+AS SC2-3 combined TS vs. combined AS SC2-4 combined TS vs. combined TS+AS SC3-4 combined AS vs. combined TS+AS ST1-2 all NFC vs. all S combined TABLE XVIII TABLE OF MMPI ITEM NIMBERS WITH THOSE COMPARISONS WHICH WEI FOUND TO BE SIGNIFICANT FOR THAI III l f t 3 m item mber 1# 2 3 f t 5 6 J 8 9 10 11 12 13 1 f t 15 16 I T 1 8 19 1 130 2 IJO 3 NRftO f t 1 3 0 5 NRftO WR 30 WR 30 130 NR 50 NRI 50 — - Edl-3 Ocl -3 0c2-3 Oc 3- f t Ms 2-3 Ms2-ft — S 1-2 SC1-2 ST1-2 NR100 NR100 NR200 NR200 WRftO — M-2 0x2-3 #2-3 j f e - f t — S?l -2 Spl-2 SL1-3 S 0l - f t SCl-2 SC1-3 SCl-ft NR200 WR30 WRftO — M-2 0c2-ft Msl4 — - Si 1-2 S 1I-3 S,l4 SC1-2 SC1-3 SCl-ft STl-2 NR 30 NR 30 NR 30 — — 0c3-ft Msl-2 Mb1-3 — - M-2 0x2-3 Edl-2 Edl-3 Agl -2 Agl -3 # 1-3 #2*3 Msl-3 6 WRftO WRftO WR50 WRI 50 — - Ms2-3 — S 3- ft SCl-ft SC24 SC3-ft T NRIOO NR150 NR200 NR200 WRftO WRftO — - 1^34 Agl -2 #14 | 2- f t .... 8 130 WRftO WR 50 — M -2 #1-3 #14 — Sil -2 Spl-ft SC1-2 SCl-ft STl -2 9 WR 30 WR100 WR100 — M-2 M-3 M-ft Rel-2 Agl-2 Agl -3 Ag2-3 #1*3 il-ft Msl -3 10 130 NRftO WR100 #1-2 #1.3 .... SCl-ft STl-2 11 NR 100 NRIOO WR30 WRftO WR 100 12 WR 50 WH 200 ■■■■ He 2-3 13 130 SR30 NRIOO — dxl-2 l f t NR 50 NR150 WRftO WRftO WR 1Q 0 15 150 WR 30 WR30 WRftO WR 5 0 16 NR 200 NR 200 WRI 50 — Ed 2-3 IT NRIOO — - Re2-3 Oc2-3 -— 18 NRIOO — - Edl-3 Ed2-3 — ■ 19 NR 30 NR 30 NRIOO — - Ocl-ft 20 NR 30 130 NR 50 150 150 21 WRftO WRftO WRftO — M-ft 22 NR50 NR150 WR150 — - #1.3 23 1100 WR100 — Re2-3 #1-3 2 f t NRIOO NRIOO NR200 — Agl -3 25 WR 50 — - M-3 0x34 Ocl -3 26 NRftO WR 30 . . . . . SC 1-3 2T 130 WR 30 WR 30 . . . . . 28 150 150 150 NRIOO NRIOO 29 130 150 NRIOO NR200 — 30 150 IJO NR150 .— Agl-3 — ' M-2 Agl -2 Agl -3 Msl -2 -— 8,1-3 Sil4 SCl -3 SCl-ft SPl-2 S ] _ 2- f t Sj34 M 4 Ocl -3 Ocl-ft 0c2-ft Msl-2 Msl -3 — - S.l -2 Spl4 SC1-2 SCl-ft STl -2 WRlOO WR 100 — ■ •*— &I.3 Spl -3 SC 1-2 S 0I -3 “ 1‘ 2 — Dx3-ft — • SC14 Ocl -2 Ocl -3 Ocl4 Ms 3- f t — - 8,1-2 Spl -3 Sp2-3 S g i - 2 s ^ - 3 s c i - 2 s c i 4 a r i - 2 S o l - 2 S p l - 3 A g 2 - 3 — — NB50 NR150 NR 150 — - M-ft Edl -2 Agl -3 Msl -2 Msl-ft M-ft Oc2-3 Agl -2 Agl -3 Mel -2 Ms2-3 - — Sjl4 SC14 #2-3 ■— Sil-2 8,14 Spl -3 Sgl4 SCI-2 SCl-3 SCl-ft STl-2 — S 2l - f t SCI-2 SCl-ft 811-2 Oc2-3 — ■ S 1l -3 Sj_2-3 S^^ SCl -2 STl -2 S g i - 2 S214 s c i - 2 s c i - f t a r i - 2 Ms2-ft —— S 2l -3 — Ed 2-3 Ocl -2 Ocl -3 Ocl-ft — - SCl-2 STl-2 M -2 M -3 — S.l -2 SCl-2 SCl-ft STl-2 Msl-2 Msl -3 -- S 2I -2 Spl -3 Spl-ft SCl-ft STl-2 *The columns represent the number of significant outcomes for each MMPI item. A blank space has been left after outcomes representing chance significance and after outcomes of the control variables. TABU XVHI (Continued) TABU OF MK M BUKEBERS HD THOSE COMPARISONS WHICH M E FOUND 10 BE SKJDTCAHT FOR TEAT ITEM # I 1 2 3 \ 5 6 T 8 9 1 0 1 1 1 2 1 3 I * 1 5 1 6 1 7 I S 1 9 ltd 2 0 2 1 2 2 2 3 2 4 2 5 2 6 rnmbtr 3 1 moo • a a a #1 - 3 tlA #24 • a a a 8 , 1 - 2 Sil - 3 S .14 Spl - 3 S p l 4 3 , 2 - 3 S 9 24 S C l - 3 S C 14 S C 2 - 3 STl - 2 3 2 B1 3 0 N R I O O • a a a # 1 - 3 Sil - 3 Sii 4 Spl - 3 SC l - 2 S C l - 3 sci 4 Sll - 2 i 3 3 KR 5 0 • a a a M - 2 M 4 ■ ■ •■ 5 4 S C 24 S C 34 3 ^ WB 3 0 WR 3 0 WR 5 0 M - 3 Ell - 2 Edl - 3 0c l - 2 0 cl - 3 0cl 4 #14 Msl 4 Ms 24 a a a a S1 - 2 S .14 %34 SCl - 2 SC 14 m - 2 1 X 3 5 1 0 1 2 0 0 NR 2 0 0 ■■■■ M 4 M 4 Edl - 3 #14 #24 #34 Msl - 2 Msl 4 ■ a a a 3 6 WB 5 0 • a a a M - 2 M - 3 Rel - 2 Re 2 - 3 ¥■3 #L - 2 #14 a a a a 8^-2 Sil 4 8 C 14 S T l - 2 3 T WJ3 0 •M ia Agl - 3 Ag 2 - 3 ■■■a 6,14 3 8 130 NR 3 0 130 NR 5 0 N R I ( 0 V R I ( 0 NRIOO WR 1 0 0 a a a a M 4 M 4 0x 34 0 c 24 Agl - 3 Ag 2 - 3 . . . . Spl - 2 SCl - 2 S C 14 m - 2 i 3 9 WBAO m W Rto WR 2 0 0 a a a a 0*34 0c 24 Agl - 3 Ag 2 - 3 # 1 - 3 # 2 - 3 Msl - 2 a a a # S C 14 e S o 24 S T l - 2 S g l - 2 S g l - 3 sew 1 ( 0 HR 3 0 NR 3 0 N R I O O ■•■ ■ M 4 Msl - 3 1 ( 1 JR 5 0 WR 5 0 M - 2 M - 3 M 4 #L 4 #24 a a a a 8, 1 - 2 S i l - 3 8,14 Sol - 3 S g l 4 S C l - 2 SCl - 3 SC 14 m - 2 i ( 2 HR 3 0 HR 5 0 N R I O O NRIOO N RIOO WR 3 0 WR 5 0 a a a a E d 2 - 3 Agl - 2 fa Msl - 3 1634 c S i l - 2 Sil - 3 SCl - 2 STl - 2 ^ 3 SR 3 0 130 NR 5 0 NR 2 0 0 a a a a M - 2 0c 24 #1 - 3 #14 a a a a S i l - 2 S i l - 3 8 il 4 S g l 4 S C l - 2 S C l - 3 S C 14 STl - 2 4 4 1 0 1 2 0 0 wmoo N R I O O • a a a Edl - 3 #L 4 Msl 4 a a a a ^ 5 SR 3 0 ■ • • • M - 2 M 4 M - 3 0x 34 Edl - 2 Edl - 3 0 c 2 - 3 0c 24 Agl - 3 a a a a Sgl - 2 S g l 4 SCl - 2 sci 4 m - 2 1 ( 6 130 ■ •■ a M - 2 M 4 0c l 4 0c 24 Msl - 3 a a a a S i l - 2 Sil - 3 Sil 4 S „ l - 2 SCl - 2 S C l - 3 S C 14 STl - 2 1 ( 7 HR 5 0 WR 1 5 0 WR 1 5 0 WRI 5 0 a a a a #1 - 3 # 2 - 3 . . . . S g l - 2 %l4 S u l - 2 S C l - 3 S C 14 m - 2 1 ( 6 W 1 2 0 0 W B 2 0 0 • • • ■ M - 2 Edl - 2 #1 - 3 a a a a k9 M tJO NR 5 0 1 0 1 2 0 0 NRIOO • • • a 0cl 4 # 1 - 2 #1 * 3 #14 Msl - 2 a a a a 5 0 HB 2 0 0 RR 2 0 0 NR 3 0 who a a a a Agl - 3 a a a a 5 1 HH 3 0 m 150 V R l ( 0 WR 5 0 WR 5 0 NRIOO a a a a M - 2 M 4 0x 34 Edl - 2 Edl - 3 Agl - 2 Agl - 3 Ag 2 * 3 #1 - 2 #1 - 3 #14 1 0 1 1 ( 0 #24 #34 • a a a S,l 4 S C 14 STl - 2 5 2 NR 5 0 NR 2 0 0 NR 5 0 WRJO WR 1 0 0 NR 1 5 0 a a a a M 4 #1 - 3 #34 a a a a * 5 3 B3 0 1 0 1 1 ( 0 m aaaa M - 2 *14 aaaa m - 2 5 l ( 130 moo WRw ■ • ■ • Re 2 - 3 Edl - 2 #1 - 3 # 2 - 3 ■aaa Sil - 2 s , i 4 S C l - 2 s c i 4 STl - 2 5 5 WR 3 0 WR 3 0 WR 3 0 NRIOO aaaa Edl - 2 Edl - 3 0c l - 3 #L 4 aaaa t r Wf*f» U BI ZVni (Continued) ^ TAU (V Mffl Z O M K M 1 F I S V H H T B O S I C dC A SIddS rats m m » i mmm m m m 1 2 3 * 5 6 ? 8 9 1 0 1 1 ■ 1 2 1 3 1 5 1 6 I T 1 8 * 2 0 2 1 2 2 2 3 f t 2 5 2 6 Sg l - 2 SgU S C l - 2 8 Cl 4 3 1 1 - 2 h i i n u u i u ™ m a m w .-i « * m u t —M a. i n turn ari./' I . ...... S 1l - 3 S j l - 2 S C l - 2 8 C 14 STl - 2 I T 5 T * 8 no ■•■• A d -3 no D o A O no A O •■M A O A O A O H R 5 0 A O >VM Q 3 0 A O A 5 0 030 A O A o 1 3 0 A o 1200 1 3 0 A O A o 1 3 0 m o o 1 2 0 0 A O •«■■ m B Q O A O m 1 3 0 —— A D O 1150 0 1 5 0 B Q O 150 1 5 0 130 A O A O A0 A50 1200 130 A o 1 3 0 A > A o n o o 1 * 5 0 A O A o m o o A5 0 •M a D o A o A o 1 3 0 A O A o A O A O A o 1 3 0 n o o n o o AO A O A O m n o o -•Ml I S O ■ u i ran m W 5 0 n o o S3 A O A o A O 114 *34 M M 8^4 AO m . •M M P mp3 S R L4 # 2 - 3 O x l 4 U l-3 O c l - 3 o§l 4 0 c 2 - 3 0 c24 •■■■ S j i 4 H x l - 4 1 ^ 2 4 *34 A g l - 3 A g 2 - 3 * L - 3 * 1 -3 8^4 1200 <H> 0 x 1 4 0 x 2 - 3 0 ( 2 4 # 1 - 3 *14 *24 ■MB N N Agl4 # 1 * 3 H4 *1-3 ■M M S 1 14 S C 1 4 S T l - 2 AO AO ■M W O c l - 2 O o 2 - 3 MM AO AO AO 11 50 ■M M W14 — 12 00 AO 1 15 0 W K L 5 0 1200 ■■M 0 c34 ,MM S i l - 3 M14 11-2 • m 1 1 5 0 AO MM 0 x 1 4 MM *14 S C 1 4 3 1 1 - 2 H l-2 Wl -3 0 d - 3 O o 2 - 3 *14 * 1-3 *14 ■■■■ •ww *14 0 e l 4 0 o 3 4 *14 * 1-3 *14 ■■M 1 5 0 AO noo ■•Ml 1 2 - 3 — %34 ■«M AO #4 #L4 *2-3 M M AO AO AO — —— 1 4 1 - 3 #14 MM S C 1 4 S T l - 2 noo 1 0 0 5 0 11 50 « • M r l 0 o 2 - 3 0 b34 ■*M ■— 0 x 1 - 2 H-3 *14 *14 D x l - 2 O x l - 3 0 x14 MM S C 1 4 S C l - 3 S C 1 4 a n - 2 AO AO HO* Od4 0 x 2 - 3 O c 2 * 3 MM *1.3 S o l - 2 W B 1 5 0 2 R 1 5 0 MM * 2 - 3 O s l - 3 0 e l 4 *14 *14 £ ■M W AO AO AO M M 1114 m m M M m noo noo mwnr AOO — m *24 0d4 0 c34 A g l - 2 Ao M M M 0x14 m O c l - 3 A i l - 3 #24 ■■■■ m im m OeH O o l 4 9 ° AO M M Jil4 M M *14 80 14 # 2 - 3 114 M M SOU Ml «4 I . — ■ 0 x 1 * 2 n>3 0 cl4 *14 M M IABLE XVIII (Continued) TABLE OF Mffl HEM NUMBERS N M THOSE COMPARISONS WHICH WERE F011D TO BE SIGNIFICANT FOR THAT ITEM m 4 1 2 3 4 5 6 7 8 9 1 0 11 12 13 14 15 16 17 18 19 number 2 0 21 22 2 3 24 2 5 26 86 130 NRJ0 NR 5 0 150 WR40 NR40 HR 50 NR100 NR100 NRI00 WR150 NR150 NR 15 0 .PP. Dxl-4 pppp 3,1-2 Spl-2 Sgl-3 sPl4 SCl-2 SCl-3 sci-4 STl-2 c L 87 NR40 NR40 WR 5 0 ■■■• Dxl -2 0x2-4 Bel -2 Ocl -2 0cl-4 Msl-4 .... Sgl-2 33-4 SCl-2 SC2-4 88 nr4o NR50 NR 50 150 NRIOO NRIOO m»mm lM-2 #1-3 #14 .... S,l-2 3,14 Sgl-2 s,i*4 SCl -2 sci-4 STl -2 89 NR 30 130 NR40 NR 50 NR 5 0 NR 200 WR40 WRJO ■■■■ aa-2 Edl-3 E32-3 0c 2-4 L 90 NR 30 .... 0x 1-2 Dxl-4 #1-4 Msl -2 «■■■ Sgl-2 s 2i-3 SCl-2 SCl-3 SC1-4 8E 1-2 91 150 150 NR 5 0 NR 100 NRIOO WR150 WB150 ■ a.a Re2-3 #14 pppp 92 NR 5 0 NR100 NRIOO NRIOO WRlfO nr4o 150 HR 150 . . . . Dxl-2 #1-2 .... 93 NR40 150 NR50 WR30 WR 3 0 NR 5 0 WE50 NR 50 .... p p p p Sgl-2 SCl -2 STl -2 94 150 NR 5 0 NR50 nr4o WR200 appp 0x 1-2 0x14 0c24 #34 Msl4 p pp p 3,1-2 SCI-4 8,14 STl -2 Sgl-2 Sgl-3 Sgl-4 SCl-2 SCl-3 95 130 NR30 NR30 NR40 WR40 nr4o NR50 WR50 ■ PPP Oxl -2 Msl-3 Ms2-3 .... 9 6 NR200 WR30 NRIOO NRIOO NR200 WR200 .p.. Ocl -2 0cl-3 0c 2-4 Oc 3-4 Agl -2 Agl-3 Ag2-3 #3-4 pppp s c i - 4 SCl -2 9 7 NR 5 0 NR50 NRIOO NRIOO NRIOO WR100 WR150 WH150 H M D Agl-3 Ag2-3 p pp p Sgl-2 S 2l-3 Sgl-4 SCl-2 s c i - 4 SCl-2 9 8 NR 30 NR30 NR30 — - Ocl-3 Msl-2 Ms2-3 .app Sxl4 99 130 NR100 NRIOO WR40 WR100 WR200 .... Agl -2 Agl-3 10 0 NR 30 NR40 WR150 PPPH «■■■ s 2 l- 3 s c i 4 SCl -2 101 NR100 NR100 WR 3 0 WR 150 NR150 PPPa 0x2-4 0d4 0c 3 4 Msl-2 Msl-3 Msl-4 p.p. S ^ -2 102 150 NR150 app. Oxl-2 Msl -2 #1-3 3,14 8^24 Si 34 S,l-2 SCl-2 SC 3 4 103 E 3 0 NR40 ■■■■ Ebd -2 #1-2 #14 .... X i 104 .... Dxl-4 #1-3 #1-4 Msl4 Ms2-4 .... Sjl -2 Sjl -3 3,14 Sgl-2 s 2 l - 3 s 2 i- 4 SCl-2 SCl-3 s c i - 4 SCl-2 105 WR200 Msl -2 X 106 NR50 NR 50 WH 30 WR150 • M a .... Sgl-4 SCl -2 SC14 SCl-2 107 WR50 — Dxl -2 Dxl-3 Dxl-4 Ocl-3 u p.p. 11-2 3,1-3 s .14 SCl-2 sci- 3 SCl-4 SCl-2 108 NR40 150 .... #1-3 #2-3 — - 1 1 J 1 109 NR 30 130 WH30 NR 30 WR 30 .... 0x2-3 13x2-4 Oc3-4 Agl-3 Ag2-3 ■ape no NR 100 NR100 ■ pan 0x24 0c2-4 #1-3 — TABUS xvni (Continued) TA3HE OF M4PI ITEM HUMBEES WITH THOSE COMPAEISOWS WHICH WERE FOUND TO BE SIGNIFICANT FOR THAI M m ltm numbei 1 1 2 3 I t 5 6 1 8 9 10 11 12 13 20 l i t 21 15 22 16 23 IT 18 19 24 2 5 26 m NR 30 150 NR 50 • p p p Rel-2 Edl-3 Ed2-3 Ocl-3 0cl4 Oc 2-3 0c 2- l t STl -2 112 NR 30 NR30 NR30 WR100 p p p p Edl-3 Ed2-3 p p p p S 2l4 SC14 113 NR150 NRIJO p p p p Msl-4 Ms24 p p p p l i l t 130 p p p p #1*3 #14 #2-3 #24 p p p p SCl-2 STl -2 115 130 NR30 130 130 WR 30 WR 50 WR200 p p p p Re2-3 #14 #24 Ms2-3 — Sii4 s^-it SCl-4 116 130 p p p p #14 Msl-3 Ms2-3 pppp 111 UR50 WRto WR1D0 WR100 p p p p Msl-2 pppp 118 NR50 150 WR30 WR30 WR3 0 WR 30 WR200 p p p p Agl-3 p p p p 119 p p p p p p p p STl-2 120 130 NR 5 0 150 150 IJO WR 50 WR50 p p p p Msl-2 pppp Sgl-4 121 NR50 150 ■ ■HP M -2 Edl-3 0d4 0c24 #1-3 Msl -2 Msl-3 Msl-4 -p.- 122 NR50 NRJO 150 HR 200 WRltO Wife) Wife) WR 150 p p p p Edl -2 Edl-3 0cl4 Agl-2 p p p p Sg2-3 Sg2-lt 123 NRltO NRltO NRIOO NRIOO WR150 WR 150 WR150 pppp M -2 M -3 Edl-3 0cl4 #1-3 Msl -2 . . . . 12t NR30 150 150 Wife p p p p Agl-3 f f ll -2 125 NR30 130 NRltO • p p p M-2 #14 ppp- Sil -2 Sgl-3 Sg2-3 SCl-3 STl -2 126 NRltO p p p p M -2 Edl-3 Ed2-3 Msl -2 Ms2*3 p p p p Sgl-3 sci -3 STl -2 121 130 Wife) Wife) .... Edl-2 Agl-3 Ag2-3 p p p p Sgl-2 128 NR30 NR 30 150 150 WR50 WR 50 p p p p Ed2-3 Msl4 p p p p Sgl-2 129 130 NRIOO NR 200 130 WH40 Wife) p p p p M -2 0c24 #1-3 Msl-2 pppp Sgl-3 Spl4 S?2-3 sci -3 SCI-4 SC2-3 STl-2 130 NRltO p p p p M 4 Dx24 Edl-2 0cl -2 #14 #24 p p p p S ^ SCl-3 STl -2 £ £ £ 131 150 NR50 NRIOO WR30 WR5 0 WR50 150 WR100 WR150 p p p p Edl-2 Edl-3 Ocl-3 0cl-4 . . . . Sjl*2 132 NR30 NR200 WR 30 WR50 WR 50 . . . . Edl -2 0cl-3 Agl-2 #2-3 Msl4 Ms2-4 pppp 133 NRIOO WRitO Wife p p p p 0c24 #1-3 #2-3 #34 p p p p SC14 STl-2 $ WR200 « M a Ocl-3 0cl4 p p p p S]l4 Sgl -2 Sgl-3 SCl-2 SC14 SCl-2 135 WR200 p p p p pp«a Sgl -2 Sgl-3 Sgl4 SCl -2 SCl-3 sci4 STl -2 1W TABLE XVIH (Continued) TABLE OP MMPI HH WEES WITH THOSE COMPARISONS WHICH WERE JO® 10 EE SIONfflCANT JOE THAT HEM MMPI ^ 2 3 5 6 T 8 9 10 11 12 13 l i t 15 16 IT 1 8 19 item 2 0 21 22 23 24 25 26 number i— > O N 130 150 130 mmmm ■aaa Sgl-3 137 NR 50 150 150 mmmm Dxl -2 Dx2-3 m Ag2-3 #1-3 #1-1 Msl-3 Msl-4 Ms2-3 Ms2-1 aaaa S 1-2 138 SC2-3 SC3-1 STl -2 1 1100 1100 NRIOO WE 30 WEJO 150 WE 50 aaaa Dxl -2 Dxl-3 Dxl-4 Ocl -2 Agl-3 Ag2-3 Msl -3 aaaa S„3-l SCl-2 SCI-4 STl -2 139 150 NR 100 NRIOO WEJO WR150 WE200 aaaa #1*3 aaaa S ^ -2 Sil-3 Sil4 Sgl-2 Sgl-3 SC1-1 SC2-1 Sgl-4 Sri -2 S224 lto Msl-4 aaaa l l t l 130 130 130 150 I 50 WE 30 WRltO WElO aaaa aaaa S 1-2 142 M 30 150 1150 WR200 aaaa Dxl -2 aaaa s^l-4 Sol -2 S21-3 Sjl-1 SCl -2 SCl -3 sci-4 SCl -2 l l * 3 130 150 150 WH 50 aaaa Dxl-2 aaaa u m •■■■ Dxl -2 Dxl-4 Dx3-4 0el -2 Ocl-4 #1*3 Msl-3 aaaa 8,14 SC1-1 SC2-1 1 1 5 130 150 WR200 ■aaa 0cl-3 0cl-4 Agl -2 Agl-3 .... sjl-l SCl-3 SC1-1 STl-2 146 130 NEltO 150 WElO WE100 aaaa Dx2-3 0cl-4 0c3-l Agl -2 #1-3 # 2-3 Msl-2 aaaa Spl-3 14T 150 150 NRIOO WE30 WE 30 WE 30 WE 30 WR 3 0 WE 150 WR150 .... Edl -2 Edl-3 0cl-4 L aaaa 118 130 NElO 150 150 NR150 NE150 1150 WElO WRID0 aaaa aaaa 149 WE40 ■aaa Dxl-2 Msl-2 aaaa Sgl-4 S.3-4 150 WR40 aaaa Edl-2 Edl-3 aaaa Ms34 2 aaaa Hgi-4 b U X - i i Wi-t Sgl-2 Sgl -3 S 22-3 151 152 WRI 50 aaaa Dxl-2 Dxl-3 5*2-3 #14 Msl-2 Msl4 ■aaa S,l -2 8,14 SCl-2 s c i 4 STl-2 15 3 NE 30 NR 3 0 NRltO WR 30 WR 30 aaaa Dxl-2 5x2-4 Agl-3 Ag2-3 #1-3 #14 #24 Msl-2 SCl -3 SCI-4 151 * NR100 NR 100 WRltO aaaa Dxl-4 5x2-4 5x3-4 #14 #24 aaaa 155 NR50 NE 5 0 aaaa Dxl-4 Dx2-4 Dx3-4 0c3-4 ■aaa Sil-3 s154 Sl24 SCl -3 s c i - 4 SC2-4 156 NE50 IJO WBltO WR40 WR 150 WR200 aaaa Agl-3 Ag2-3 #1-3 #14 #2-3 #24 Ms2-3 157 NE 5 0 150 aaaa Dxl -2 Agl-3 #14 aaaa 158 WR50 150 WR50 WR150 aaaa Dxl-2 5x2-3 5x2-4 Rel -2 Re2-3 .... Sil -2 S,l4 159 130 130 NE100 NR100 WR30 aaaa Dxl -2 Dxl-3 Dxl4 Edl-3 aaaa S,l-2 SJ4 S.3-4 160 130 aaaa Dxl -2 Dxl-3 Dxl-4 Dx2-3 Dx2-4 Ocl-4 0c2-4 Agl-2 Agl-3 X Sjl -2 Sjl-3 Sil -2 V-3 s 1 i - 4 8 1l4 SCl-4 STl -2 s c i 4 SC2-3 SC34 SCl-2 SCl-3 ■ ■ ( 0 CO Sgl-3 Sgl-4 SCl -2 TABLE XVni (Continued) TABU! OF MI I® NUMBERS WITH THOSE COMPARISONS WHICH WEB FOP TO BE SIGNIFICANT FOR THAI ITEM MMPI item numbei 1 « 2 3 I t 5 6 7 8 9 10 11 12 13 20 Ut 21 15 22 16 IT 23 2* t 18 19 25 26 161 NR 50 NR 150 #1*3 #2-3 aaaa 162 WR 30 130 130 WRltO WRltO WR50 aaaa #1-2 aaaa 163 130 NRltO NRltO aaaa Oxl-2 0x24 Agl-2 Agl-3 Ag2*3 #14 . . . . s. 1-3 SCl-2 SCl-3 SC14 SCl -2 1ft NRIOO 130 150 WR150 aaaa #1-3 aaaa £ 165 130 130 NRltO NR150 WR 30 WRltO 150 WR100 WR100 aaaa Msl -2 aaaa S^-lt 166 NR 5 0 NRIOO 150 WR100 «■■■ Ocl-3 aaaa 167 NRltO 150 130 aaaa Re2-3 Edl-3 Ed2-3 0cl -2 0cl-3 0cl4 aaaa 168 NR 30 130 150 WR50 aaaa 0x14 aaaa Sil4 S o l - 2 S?l4 S C l- 2 sci4 STl-2 169 NRIJO NR150 WR 30 WR 30 a a aa #1-3 #2-3 b & aaaa S j . 1-2 SCl-2 SIl-2 170 VR30 WR150 . . . . Oxl -2 0x2-3 Edl-3 Ed2-3 0cl4 0c24 ff -2 Msl -2 Msl4 aaaa Ssl-3 STl -2 171 130 NRltO 150 150 150 WR 30 WB 50 WR 50 WRJO WR50 WR200 aaaa aaap 172 150 WR 30 130 ■MR Oxl-3 0x2-3 aaaa &Ll4 SC14 173 WR200 ■ O M R Edl -2 Edl-3 Ed2-3 Ocl -2 Ocl -3 0cl4 Agl-3 aaaa SC24 1 7 1* NR50 aaaa Dxl-4 0x24 0x34 aaaa Sil4 175 150 150 150 NR150 NR 150 . . . . Edl-3 Ed2-3 0d3 #14 #24 aaaa Sll-3 S 2I-3 SCl-3 SC34 STl-2 176 150 150 150 NRIOO WRltO WRltO a a a a 0cl-3 0c34 a a a a STl -2 177 WRJO WR50 WR 50 WR150 a a aa A g l - 2 #1-3 #2-3 aaaa 8,1-3 3,14 178 ■130 150 WR 50 WR50 aaaa 0x1-3 A g l - 3 #14 #24 J. aaaa sJl-3 s,i4 SCl-2 SCl-3 sci4 STl -2 179 150 150 NRI 50 WRltO WR 50 WR5 0 WR 100 WR100 aaaa Dxl4 0x24 0x34 Rel -2 Edl -2 Agl-3 #14 #34 Msl-3 Msl-lt aaaa SCl -2 180 130 IJO 150 150 WRJOO WR100 WR150 aaaa Edl-3 Ed2-3 0cl-3 0cl4 aaaa 181 130 150 NR150 WRltO WR50 aaaa Edl -2 Edl-3 Agl-3 Ag2-3 aaaa 182 130 150 . . . . 0xl4 0x24 #1-3 #2-3 a a a a S. 1-2 Spl-3 SCl-3 SC14 STl-2 183 130 NR150 a a a a Msl-3 a a a a Si 24 SCl-2 1 £ 1ft NR 150 ■■■■ 0x1-2 Msl-2 Msl-3 X a a a a 105 130 NRlJO WR50 a a a a 0x1-2 0x24 0x34 Agl-3 #1-2 #14 aaaa Sj.2-3 150 TABLE XVHI (Continued) TABLE OF MMPI I0M1MHERS WITH THOSE COMPARISONS WHICH IRE POUND TO BE SIGNIPICANI FOR THAI M m 1 2 3 I t 5 6 T 8 9 10 11 12 13 l i t 15 16 IT 18.. 19 Item 20 21 22 23 21 * 25 ■ 26 number 186 NR30 130 NR 3 0 NR40 WR 30 WR 30 WRltO WRltO aaaa Edl-3 Ed2-3 0cl -2 #1-3 #2-3 ■ aaa S.l-2 Spl-4 SCl -2 SCl-3 SC14 SPl -2 i £ 187 130 m o o m o o aaaa Dxl -2 #1-3 ■■■■ Sjl-2 SCl-2 SC14 STl -2 188 VRI*0 .... Dxl-2 Dx24 Edl -2 Ocl -2 M - 2 Agl-3 #14 #24 Msl-2 Ms2-3 Ms24 aaaa 189 130 WRltO aaaa Dxl-3 Agl-3 Ag2-3 #14 #24 ■a«a SjM SCl-2 STl-2 190 130 NR50 NR50 150 ■aaa Dxl -2 0x2*4 #1-3 Msl -2 Msl4 aaaa s 22-3 SCl-3 SC34 191 MR30 NE200 NR200 ■aaa Dxl-lt #1-3 aaaa S,l4 SCl-3 SC14 STl -2 192 NRto 150 NRJ0 m o o NR 100 NR 100 aaaa M-3 0x3*4 Agl-3 #1-3 #14 #2-3 #24 mmmm S, 1-2 ^1-3 SCl -3 193 NR 30 130 NR50 WR 30 WR 30 WR 30 WRIJO aaaa Oc2-3 0c2-4 ■aaa L ■ 194 NR30 ■aaa M -2 0x2-3 #1-3 #14 Msl-3 aaaa ^2*3 195 ■■■■ Dxl-3 0ol-lt ■ aaa S 2l-2 SCl -2 196 ■■an aaaa 197 NR50 N B200 NR200 mmmm aaaa 198 r i h Dxl -2 M 4 Edl-3 aaaa Sgl-2 Sgl-3 SCl-3 sci-lt SCl-2 199 m o o NRIOO WRltO WRltO aaaa 200 150 NR50 150 WR 30 aaaa Dxl-2 Dxl4 aaaa 201 NR3 0 NRltO ■■■a Dxl-3 Dxl-lt aaaa 202 m o NR200 WR30 aaaa ■aaa #1-3 203 NRltO NRIOO NR200 aaaa Dxl -2 0c2-3 Msl-2 Msl-3 ■■■a 2 0 1 * m NR 200 NRltO WR50 ■aaa Ed2-3 0c24 Ms24 aaaa 205 ■ nan Agl-3 aaaa SCl -2 STl -2 206 WR50 M 5 0 WR150 aaaa Dxl -2 Msl-3 Msl4 aaaa SC14 STl-2 20J M 50 WRltO aaaa Dxl-3 0x2-3 #1-3 ■aaa 208 NR 3 0 150 NR50 WR5 0 aaaa Agl-3 Ag2*3 aaaa Sjl -2 Sjl-3 Sj24 Si34 SCl-2 SCl-3 1 209 NR 5 0 150 WR150 aaaa .... SPl-2 1 , 210 NH50 NR 50 WR150 . . . . . 151 TABLE XVHI (Continued) T A B E S O F M O T IM N 1K E ® W TH T H O S E C G K P A R IS O H S MCE W E F O U N D 101 S M F M C F O R T H A I M uto Item number 1 2 3 I t 5 6 7 8 9 10 11 12 13 20 l i t 21 15 22 16 23 17 2l t 18 25 19 26 211 .... .... 212 NR50 NR100 NRIOO NRIOO WRltO WRltO WRltO WR 50 aaaa Dxl -2 0c2-3 Agl-2 Agl-3 Msl -2 aaaa Sgl-2 213 NRltO aaaa Bxl -2 Edl -2 Edl-3 Ocl -2 Ocl-3 Qcl4 Msl-2 Baaa 2l J t NR 50 aaaa 0c24 # 1-2 #1-3 ■aaa 215 WR 30 .... 0x2-3 0x24 Agl-2 Ag2-3 #1-3 #2-3 #34 Msl-3 Msl4 Ms24 aaaa SjM S 1l4 Sgl -2 Sgl-lt SCl -2 sci4 311-2 216 130 NRIOO NRIOO NRIOO NR100 NRIOO NR200 NR200 130 aaaa Agl-3 Ms2-3 Ms24 aaaa S.l-2 s 1-3 S 1i4 S44 SCl-2 SC14 STl-2 X 1 £ 217 130 RR50 NR 50 aaaa Oxl -2 Oxl-3 Mel-2 aaaa S,l4 s2 l-3 SCl -2 SCl-3 SC14 311-2 218 NR150 ■aaa 0x24 Ocl -2 Ocl-3 Qcl4 Agl-3 aaaa X £ 219 NR 30 NR50 NR 50 WRltO ■aaa 0x34 Ocl-3 0cl4 0c2-3 0c24 Agl-3 Msl-3 Ms2-3 Mb34 aaaa 220 NR 5 0 .... .... Sgl-2 sci -2 221 130 NR50 NRIOO NR200 wrjo WRI 50 aaaa 0cl -2 Oc2-3 0c34 ■aaa 222 130 NR 50 NR 200 NR200 WRltO WRltO aaaa aaa# SCl -2 223 150 NR 5 0 150 NR150 WR30 NR100 WR 150 aaaa Edl-3 0d-3 0c2-3 0c24 ■ aaa 22* t NSI 50 ■ M B Edl-3 E32-3 0cl -2 m i 0*14 m m i Hsl -2 Hs2*3 M»g4 ■aaa 225 130 NR30 NR30 NR 30 NR 50 150 WR150 aaaa Edl -2 Edl-3 aaaa Sgl-2 SC14 STl-2 226 NR 30 NR30 130 NRltO NR 50 NR 50 WRltO WRto WRJO WR 50 WRJO aaaa 0xl4 0x24 0x34 Edl-3 Agl-3 Ag2-3 #1-3 Msl-2 Msl-3 Msl4 ■■■a Spl-lt SC14 STl-2 227 150 NRJ 00 NR100 WR150 .... #1-3 aaaa Sol-3 SCl-3 SC2-3 STl -2 £ 228 130 NR 5 0 W O .... aaaa Sil-2 SCl -2 STl-2 229 H30 NR30 NRltO WR100 . . . Oxl -2 Msl-2 aaaa 230 m 150 150 aaaa Oxl -2 Agl-2 Agl-3 #1-3 #2-3 aaaa S 1l-2 SCl-2 sci4 SCl-2 231 150 150 WRJO aaaa Edl -2 Edl -3 Agl-3 Ag2-3 Msl -2 ■aaa 232 NRltO NRltO NRIOO VRIJO WR150 WRIJO 0x2-3 Ms2-3 ■aaa s.14 Sjl-2 s«i4 SCl-2 SC14 STl -2 233 23^ NR150 NR 30 NR 50 'Agl-3 NR150 NR150 Sgl-2 WR100 Ed2-3 0cl-3 #1- 2 . aaaa X £ £ 235 NR 30 NR30 NR 30 150 NRIOO WRltO WRltO WR100 WR150 .... Msl-3 Ms2*3 ■aaa 152 TABLE xvm (Continued) TABLE OF Ml 1M NUMBEHS WITH THOSE COMPARISONS WHICH HERE FOOT) TO BE SIGNIFICANT JOE THAI ITEM m item numbei 1 l 2 3 4 5 6 T 8 9 1 0 11 12 13 2 0 14 2 1 15 22 16 23 IT 18 24 25 19 26 236 130 130 150 150 130 150 WR150 aaaa Oxl-2 Od-3 Dxl-4 #1-3 aaaa sci-4 314 STl -2 Sgl-2 Sgl-3 S 2l-4 SCl -2 SCl-3 237 NR100 NR100 ■■■■ Edl-3 Ocl -2 ¥-3 Ag2-3 aaaa 238 NHJ40 1100 WR1O0 1100 aaaa Dxl-2 Oxl-4 #14 #2-4 aaaa S 2l-3 Sgl-4 SCl-3 SCI-4 SCl -2 239 ■■■a ¥ ■ 3 Ag2-3 #1*3 Msl -2 Msl-3 Ms2-3 Mb24 aaaa Sgl-2 S 2I -4 sci-4 SC2-4 240 150 WR 5 0 WR100 aaaa Edl -2 Edl-3 0c2-4 Oc3-4 #14 aaaa 241 140 150 NR100 130 130 130 150 aaaa Edl -2 Edl-3 Msl -2 aaaa 242 140 WR40 WR40 WR100 WB10O aaaa Dxl -2 Dxl-3 Bxl4 Ocl-4 ¥ - 2 Msl-2 .... s 1-2 SCl-2 sci-4 SCl -2 243 130 130 130 130 WR40 aaaa Oxl -2 0x24 Edl -2 0c3-4 ¥ ■ 2 ¥-3 Msl-2 Msl-3 .... 244 130 150 150 1100 WR150 .... 0x34 aaaa 245 140 WR 100 ■■■• Agl-3 #14 .... Sgl4 246 150 WE50 WR100 .... 0x1-3 0x2-3 0x34 #1-3 aaaa 247 WR40 WR150 WHIJO aaaa aaaa 3 O J NR100 1100 ■ ■•a aaaa 249 130 1200 we4o WE40 150 150 WE 100 aaaa Edl-3 Ed2-3 0cl -3 Ocl-4 Agl-3 ¥-3 #14 #2-4 Msl -2 Ms2-3 aaaa Sjl4 Sj24 250 130 WB200 .... Ed2-3 Agl-2 ¥-3 #2-3 Msl-2 Ms2-4 aaaa 251 150 1150 aaaa #1-3 #14 #24 aaaa Sgl-3 sci -3 SC14 STl -2 252 150 150 150 WRIOQ WH200 aaaa #14 #2-4 Ms2-3 Ms2-4 aaaa Sgl4 STl -2 253 130 WRIOQ aaaa M -2 Od-3 0cl4 aaaa 254 130 130 150 1150 NH 150 WRI 5 0 WR200 WR200 aaaa aaaa 255 130 aaaa Agl-3 Ag2-3 aaaa Sgl-2 Sgi4 STl-2 256 130 1100 aaaa Ocl-4 aaaa 257 150 1100 .... 0c2-3 aaaa S 1-2 SCl -2 STl -2 258 150 150 aaaa Bui-2 0x1-3 Msl-2 Ms2-3 aaaa 259 130 140 150 150 150 130 WR 150 aaaa 0c2-3 0c34 ¥-3 ¥-3 aaaa Sgl-2 Sgl-3 SCl-2 SCl -3 STl-2 C M WR30 WE40 WR40 WR100 WR 150 WR150 Dxl-2 0x2-3 Edl-3 Ed2-3 Ocl-2 0d-3 Od-4 0c2-4 — 153 TAKLE XVni (Continued) TABLE OF WtPI HEM NUMBERS WITH THOSE COMPARISONS WHICH HERE FOUND TO BE SIGNIFICANT FOR THAI ITEM t a item number 1 2 3 I t 5 6 7 8 9 10 11 12 13 2 0 l i t 21 15 22 16 23 17 18 24 25 19 26 261 1 3 0 ®50 NR50 W R50 W R IO O W R 150 W R 150 aaaa M -2 Edl -2 Ed2-3 #24 Msl -3 Msl4 aaaa s 1-4 SCl -3 SC14 STl-2 262 1 5 0 NR50 NR200 N R 200 1 3 0 W R 30 W R 30 W R 30 W R 30 W R 30 WRI 5 0 aaaa Edl -3 Ed2-3 aaaa 1 263 NR30 N R 30 NR200 aaaa M -2 Msl4 Ms3-4 aaaa 8,1-4 SCI-4 2 (k ffl30 NR30 1 3 0 NR 5 0 NRltO aaaa Hel -2 Edl-3 Ocl-3 0cl4 0c2-3 aaaa s . 1 4 SC14 STl-2 265 NR50 NR50 N R 50 N R 50 W R IO O WRIOO W R IO O .... M 4 0 x 2 4 Edl-3 Ed2-3 h #14 Msl-2 Ms2-3 Ms24 — - S 2l-2 266 1 3 0 N R 150 NR150 NRI50 NR150 W R 30 W R 30 W R200 aaaa #1-3 # 3 4 .... s1i 4 Spl-3 SCl-3 SCI-4 STl -2 267 NR30 ®50 NR50 NRIOO NRIOO NRIOO NR100 W R ltO aaaa Ed2-3 0cl4 0c2-4 aaaa c 268 ® 30 .... M -2 aaaa 269 NR50 W R 50 .... M -2 Agl-2 Agl-3 Msl-2 aaaa 270 aaaa Ocl-3 Ocl-4 Msl -2 aaaa Sgl -3 SCl-3 STl -2 271 NRto NR 50 NRJO N R 50 aaaa aaaa 272 W R IO O aaaa M -2 Agl-3 #1-3 #1-4 aaaa 273 N R 30 NR50 N R 50 N R 150 W R 50 WRJO aaaa #1-3 # 1 4 #2-3 # 2 4 Msl-3 aaaa 27^ ® 50 NR150 W R ltO aaaa M -2 0x2-3 0x2-4 Agl-2 Agl-3 Msl-4 aaaa 275 NR50 NR50 N RIO O NR150 NR1J0 aaaa M -2 0c2-4 Msl-3 Ms 2-3 Ms2-4 aaaa 276 1 3 0 1 3 0 1 3 0 NRIOO NR200 NR200 W R JO W R IO O W R IO O aaaa #1-2 #1-3 # 1 4 aaaa Sjl-3 SCl -2 277 1 5 0 1 5 0 IJO N R 200 .... Agl-3 Ag2-3 #1-3 Mb24 aaaa S]l4 SCl-2 sc i-4 STl-2 278 1 5 0 NRJO NR150 NR150 W R30 W R30 W R ltO WR150 aaaa 0cl4 Agl-2 #1-3 .... s 21-3 SCl-2 SCl -3 STl -2 279 ® 30 WR30 W R 30 W R JO aaaa Ed2-3 aaaa 280 1 3 0 W R 30 W R IO O aaaa M -2 0x2-3 M 4 Edl-3 Ed2-3 Ocl-3 0cl4 0c24 #1-3 # 1 4 aaaa 281 1 3 0 NRl+O NR150 NR150 aaaa M -2 Edl-2 Edl-3 aaaa 282 NRJO NRIOO aaaa 0 x 2 4 Edl-3 Ed2-3 Agl-3 Ag2-3 #1-3 aaaa S2] . - 2 s c i-4 STl -2 283 W fltO WRJO W R JO aaaa aaaa 2& 1 5 0 NRJO NRJO NR50. N R 150 NR150 W R ltO aaaa aaaa Sgl-4 SCI-4 BPl -2 285 1IR30 N RltO W R JO aaaa M 4 aaaa 0J 1 C O $ TABLE! XVIII (Continued) TABLE OF MMPI HI NUMBERS WITH THOSE COMPARISONS WHICH WERE FOUSD TO BE SIGNIFICANT FOR THAI HIM MCI 4 4 * j m m 1 2 3 I t 5 6 7 8 9 ' 10 11 12 13 14 15 16 17 18 19 number 20 21 22 23 24 25 26 286 I 30 NR 30 HR 50 WR 5 0 WRIOO ■■■■ M-2 M-3 Edl-3 Ed2-3 0c24 0c34 #1-3 #2-3 Msl -2 Msl-3 ■ ■ ■ ■ Sgl-3 sci -3 28? ■ ■ ■ « * M -2 M 4 0c2-3 0c 2- l t ■•■■ S.14 8,24 Sgl-3 SCl-3 sci4 SC2-3 SC24 STl -2 £ 288 NRI 50 WRIOO ■■■« #1-3 #2-3 ■■■■ SCl-3 Scl4 SEL-2 289 130 130 WR 3 0 WR3 0 WR 15 0 WR 150 Agl-3 ■■■■ 290 I 3 0 NR30 NRltO NR5 0 NR50 WR 50 ■ ■■ ■ M -2 M-3 M 4 Ocl-lt 0c24 Msl-2 Ms34 ■ ■■ ■ S 1-2 Sl4 SCl-2 sci-4 SCl -2 1 £ 291 ■■■■ M -2 Ocl-lt Msl-2 Msl4 Ms24 Sgl-2 292 WRl^O ■■■■ Agl-2 #24 ■■■■ sci -3 SPl -2 293 NR5 0 150 NRIJO WRltO WRltO WRIOO ■ ■■M M -2 M-3 Msl-2 Msl-3 Msl4 ■■■■ 294 NR150 WR5 0 WR 200 WR200 ■■■• Ocl -2 Agl-2 #1-3 #14 #24 Msl-3 Msl4 Ms2-3 Ms24 ■ ■ ■ ■ s,l4 sci-4 SC24 SIl-2 295 130 NR 50 WRltO WRltO ■ ■■ ■ M-2 Edl-3 Ocl-3 Ocl-lt Msl-3 Msl-4 ■■«■ SCl-3 sci-lt STl-2 j. 296 WR30 WR30 WRIOO ■■■« #1-3 #2-3 ■■•■ Sgl-2 297 Msl-2 Msl-3 «■■■ #14 298 NRltO NR 5 0 NRIJO WR 30 130. WR 3 0 WRIOO ■■■■ M-2 Ed2-3 Ocl-lt Agl-2 Msl-2 Msl-3 Msl4 Ms2-3 ■■■■ 299 150 NR5 0 NR 50 150 WR 3 0 WRltO WR 50 WR50 WR 5 0 WRIOO . . . . Edl-2 Edl-3 ■ ■ ■ ■ 8.14 300 NRIOO NRIOO WRltO . . . . . . . . X 301 1 3 0' WRIOO WRIOO . . . . M -2 M-3' M 4 #1-2 #1-3 ■■■■ •Sil-2 Sjl4 SCl-2 STl-2 302 ■ M M M-2 M-3 Msl-2 . . . . 303 NR50 NR50 NR150 WR 30 WR40 . . . . Ms2-3 Ms34 ■ ■ K M #24 304 130 NR30 150 NR50 WR150 WR 150 ■■■■ Ocl-lt #14 . . . . S C l- 3 SC14 305 WRltO WRltO WR40 wr4 o WRltO WR50 •■■■ M 4 M 4 M 4 #1-3 #14 Ms2-3 SCl -2 ■•■■ SCl-3 S,l-2 sci-4 s.14 STl -2 Sgl -2 Sgl-3 I p 306 150 NR200 130 WR 30 WR 30 ■■■■ M-3 Eel-2 0c2-3 307 130 150 150 150 NRIOO WR 30 WR 30 WR30 WR 30 WR40 WR40 WR 50 . . . . M-3 M - 3 Ms2-3 . . . . 308 NRltO 150 WR 50 WR 200 ■■■■ M-4 Agl-2 Agl-3 #1-3 #2-3 Ms2-3 Ms24 ■■■« S. 1-2 SJ-2 Sol4 SCl-2 sci-4 Sffl-2 309 NRIOO NR150 NR150 NR150 WR30 • ■■■■ M-3 1 £ £ 310 130 150 NRIJO NR150 WR50 WR 5 0 WR 5 0 ■■■* M-3 M 4 Edl -2 Edl-3 Ocl -3 ■ ■ ■■ Si !-2 SCl-2 SCl-2 155 TABLE XVIII (Continued) TABLE OP Ml ITEM NUMBERS WITH THOSE COMPARISONS WHICH WERE FOUND TO BE SIGNIFICANT FOR THAI HEM w in ^ 2 3 I f 5 6 7 8 9 10 11 12 13 I l f 15 16 17 18 19 Item 20 21 22 m OJ 2\ LT\ OJ 26 number 311 NR 30 NR 30 WRltO WRltO — - M 4 312 NRIOO p p p p Agl -2 ¥'3 . . . . s . l - 3 313 NRI+O HR 50 HR50 HR150 WR 30 WK.0 0 3 I f c HR50 1150 WR30 130 WR50 WR 50 315 150 NRIOO p p p p Edl-3 Ocl -2 0cl4 316 WRltO WRltO WR150 p p p p M 4 O J r H -§f 317 HSltO HR 50 150 WRltO WRltO WR 150 318 NRltO IJR50 150 HR 50 WR 30 WRltO 319 H30 WR200 — p p p p 320 HR50 WR 30 WRltO WR200 • p p p Agl -2 321 WRltO WRltO pppp 0x14 Ms2-3 p p p p 322 NR 50 HR50 150 I 50 p p p p M -2 323 NR 50 NRIOO NR 100 . . . . Oxl -2 M 4 32^ ■pan 0x2-3 Edl-3 0c24 #1-2 Msl-2 325 NR50 NRIOO WRltO p p p p ¥■3 ¥-3 326 HR150 1200 WR 30 p p p p # 1-2 #1-3 327 130 150 WR 50 WR 200 p p p p M-3 328 HR30 130 150 150 WR 50 p p p p 329 RltO NRltO p-pp Rel -2 Edl -2 Msl-3 330 #30 NRIOO WRltO WR 5 0 WR100 WR100 331 150 150 150 I 50 WR 200 pppp 332 I 50 p p p p — - 333 IJO WRltO WR 2QQ pppp ¥-3 ¥*3 33^ I 50 WE 50 WR 50 p p p p M -2 Ocl-4 335 130 150 150 150 150 p p p p M 4 M 4 Edl -2 0cl -2 0c24 ¥ - 2 ¥-3 STl -2 Sj_ i4 sci-3 SC14 STl -2 p p p p Ed2-3 0c24 pppp S,l -2 SCl -2 OT 1-2 WR 150 . . . . #1-2 #24 X pppp SC14 #1-3 #14 Msl-2 Msl4 Ms2*3 Ms34 p p p p ¥-3 p p p p Spl -2 s 2i-3 S,l4 SCl -2 sci-3 WR 150 WRI 5 0 C Re2-3 u p p p p Sil4 Si34 WR 50 p p p p M -2 M-3 pppp Sjl -2 Sgl-2 ¥-3 p p p p s2i 4 SCl-lt aci-2 M 4 M-3 0cl-3 0c34 Msl -2 Msl4 Ms2-3 Edl-2 Edl-3 ¥-3 Msl -2 Ms2-3 p p p p r o a Msl-3 p p p p Msl-4 pppp p p p p p p p p s2i-it M-3 M 4 M 4 O J H -§p #1-3 #14 p p p p O J Ms2-3 PPP p p p p p M 4 M 4 0x3-4 ¥-3 #1-3 #14 M-2 Rel-2 Edl-3 0cl -2 #14 #24 Msl4 p p p p SCl-2 STl-2 #1-3 #2-3 p p p p M-3 #1-3 p p p p Sjl-2 S ^ Sgl-2 ¥■3 Ag2-3 -• Sgl -2 Sgl4 SCl-2 SCl4 Sgl-3 SCl-3 sci4 an-2 sc i4 SC34 Sgi-3 Sgi-ij sci-2 sci-3 s c i4 ari-2 —— Sji4 3 ^ 4 sc i4 SC3-I| STl-2 8 1 4 s 1-3 8 14 8 24 SCl-3 SC 14 sii-2 2 2 j j l g - 3 — " — ■ S g l - 3 s c i - 3 SCl-2 SCl-3 STW 156 TAHIE m n (Continued) M E 0? Ml M MBERS W IT H T H O S E C 0M PA EIS0N S W H X C K W E R E FO O D ) T O I B E i FO R TH A I I I I f 1 2 3 it 5 S 7 8 9 10 E 12 13 H l ; 16 IT 18 IS “■ 20 21 22 23 26 25 26 336 Wife) . . . . 0x 1-2 Msl -3 a a a a si4 SCl -2 SCl-3 SC14 O J « ■ — 1 S3 337 KR 30 130 WR30 WR30 WR 30 WR 30 130 150 m o o aaaa 338 150 150 150 m o o m o o HR200 WR 5 0 WR 50 WR5 0 WRIOO 339 MR30 m o 150 150 150 m o o m oo m o o # 1*2 Agl-3 3 * to m o o m o o WRltO Wife) Wife) WR200 — — Oxl-2 #1-3 % HR 5 0 150 150 HR50 m oo m o o 130 WR 30 Wife) WRltO 3^ 2 WR30 WR100 WR150 WR 200 ■ ■ ■a #1-3 #14 Msl-2 aaaa SCl-3 3^3 m 5 o HR150 1200 WR 30 ■ M M 0X1-2 0x2-3 Msl -2 Msl-3 Msl4 34 150 150 150 150 150 150 m o o m o o NRIOO WR 30 3^5 150 RR50 HR50 1100 150 aaaa Dxl4 1X2-3 0x24 ¥-3 34 MRltO IJO .... 0x2-3 #1-3 mmmm Sgl-3 SCl-2 SCl-3 SCl-2 3^T WR150 ■■■a 0x2-3 0x34 #1-2 #1-3 — a. WRltO 3 4 MR30 RR30 130 130 m o o m o o m o o m o o WR150 3^9 WR50 ■■■* #14 aaa* #1-3 350 WR30 WR200 Dxl-2 0X2-3 # 2-3 aaaa 351 WRljO WRltO ...a Msl4 Ms24 Ms34 aaaa SC34 352 130 WRIOO . a. a Oxl4 0x24 ¥-3 Msl -2 aaaa Sgl-2 353 130 SR 50 .... Oxl4 aaaa S ^ -2 SC14 SCl -2 35^ 1100 150 WR5 0 ««■■■ Msl -2 aaaa 355 150 ■ Ba B Agl-3 SX24 SC14 356 aaaa 0x2-3 0cl4 #14 aaaa S]l-3 SCl -3 STl -2 35T 15s 150 WR30 0x2-3 aaaa #1-3 358 NR30 150 150 aaaa Oxl-2 0x24 aaaa SCl-2 359 150 WR30 Wife) WRltO WRltO WR150 WR 200 aaaa 0X14 .... 360 130 130 150 150 150 WR 50 aaaa #1-3 Ms24 aaaa Dxl-2 0x14 Edl-3 0c24 #1-3 — * S,l-2 SCl-2 SCl-3 311-2 VR100 WRIOO ■— #1-3 -— #1-3 - * Sjl-2 Sjl4 S 1-2 S 1-3 Sgl4 SCl-2 #14 Msl -2 -- Sgl-2 Sgl -3 SCl-2 ffll-2 ®50 WR 50 — - M -2 11-3 #14 — SC L -2 Ms24 — - Sjl-3 WR 30 -— Rel-2 Msl-2 Ms 2-3 — Ag2-3 #1-3 Ms2-3 *— SCl -2 SC14 STl-2 W R 1 5 0 — — Edl-3 Ed2-3 0 cl4 0c24 S1-2 S1-3 8^4 SCl -2 SC14 STl -2 Sgl- 3 Sg34 aci-3 SC34 157 M E XVIII (Continued) TABLE OF MMPI ITEM UMBERS TOR THOSE COMPARISONS WHICH HERE FOUND TO BE SIGN3JIGAIIT FOR THAI ITEM * 1 2 3 ^ 5 6 T 8 9 10 11 1 2 13 1 1 1 15 16 I T 18 19 item 20 a % 23 ^ ^ 26 number 361 130 WR50 ■■«■ M-2 M 4 Msl-2 ■■■■ sci4 M -2 362 150 WR40 WR5 0 ■■■■ Edl-3 0c2-3 ■■■■ 363 150 NR100 . . . . M 4 Edl-3 Agl-3 ■«■■ S 2l-3 3ft NR 5 0 NR100 WR100 . . . . M -2 0cl4 Agl-3 Msl4 Ms24 Ms34 ■■■■ 365 150 NR200 NR 200 WftO ■■■«■ Edl-3 Ocl -3 Agl-3 # 1-2 ■ ■■■ 366 WR 30 WR30 WRliO wfto WRltf WR1*0 — M 4 M 4 at34 #1-3 #14 Ms2-3 SCl-2 SCl -3 SLl-2 sci4 S,l4 Spl -2 Spl -3 s.14 M-2 1 36? NR 50 ■ ■HP 0c 3ft ■ ■■■ 321-3 SCl-3 ari -2 368 150 150 150 NR 50 NR 5 0 150 150 ■■■■ Msl-3 Ms2-3 mmmm 369 1JR100 NR100 NR100 WR 30 WR100 WR150 WR 150 WR200 ■•«• Ocl-2 0c 24 ■ ■■■ 370 HR 30 . . . . #1-3 . . . . S 2l-2 SCl-2 sci4 811-2 371 ■ P M Agl-3 Ag2-3 #1-2 #14 . . . . #1-3 372 HR50 NR 50 WR 50 . . . . M -2 M 4 ■■■• S.l-3 SCl -2 SCl-3 STl-2 373 RR30 NR50 NRI 5 0 NR20Q WR 3 0 WR 30 ■n«p Ed2-3 Agl-2 Agl-3 Msl-2 ■■■■ ari-2 \

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Prediction Of Overt Behaviors In Hospitalized Psychiatric Patients

PDF

The Use Of Multiple Objective Measures In Survey Research As A Means For Predicting Respondent Future Behavior

PDF

A Study Of The Relationships Between Factors Found In Cattell'S Sixteen Personality Factor Questionnaire And Factors Found In The Guilford Personality Inventories

PDF

A Factor Analysis Of The Symbolic-Evaluation Abilities

PDF

On The Relationship Between Anxiety And Aggression In Nine-Year-Old Boys

PDF

Factorial Stability As A Function Of Analytical Rotational Method, Type Of Simple Structure, And Size Of Sample

PDF

First Steps In An Attempt To Construct An Objective Test Of Character Structure

PDF

A Comparison Of Two Methods Of Adapting Self-Instructional Materials To Individual Differences Among Learners

PDF

A Factor Analysis Of The Semantic-Evaluation Abilities

PDF

Empathy And Social Perception

PDF

Factor Analysis Of Certain Aptitude And Personality Variables

PDF

Intellect After Lobotomy In Schizophrenia: A Factor-Analytic Study

PDF

A Rotational Approach To Psychological Invariance

PDF

Self-Worth, Future Goals, Mood And Time Perception

PDF

A Multidimensional Scaling Of Mood Expressions

PDF

The Effect Of Subject Sophistication On Ratio And Discrimination Scales

PDF

A Factor Analysis Of The Figural-Evaluation Abilities

PDF

A Factor-Analytic Study Of Problem-Solving Abilities

PDF

Effects Of Group Behavior Therapy Imagery On Basketball Performance

PDF

Defense Choice And Identification

Asset Metadata

Creator Devries, Alcon Gysbertus (author)

Core Title Methodological Problems In The Identification Of Suicidal Behavior By Means Of Two Personality Inventories

Degree Doctor of Philosophy

Degree Program Psychology

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag OAI-PMH Harvest,Psychology, clinical

Language English

Contributor Digitized by ProQuest (provenance)

Advisor Guilford, Joy P. (committee chair), Farberow, Norman L. (committee member), Ruch, Floyd L. (committee member)

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c18-320708

Unique identifier UC11359022

Identifier 6406237.pdf (filename),usctheses-c18-320708 (legacy record id)

Legacy Identifier 6406237.pdf

Dmrecord 320708

Document Type Dissertation

Rights Devries, Alcon Gysbertus

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the au...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus, Los Angeles, California 90089, USA