MURDER: AND ASSAULT: M EXPLORA'FGRY COMPARESONE 0? MALE MENTAL“ ILL OFFENDERS Dissertafion for the flagree of Ph. 'D. MECHEGAN STATE UNIVERSITY RQNALD NORMAN LUEHRIG 1974 k ‘ £18314 R Y1“? %\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\ \\\\\\\\\\\\\\\\\\\\\\\\\\ Mggigan gt, ' 3 1293 10524 2535 _ ABSTRACT MURDER AND ASSAULT: AN EXPLORATORY COMPARISON 0? MALE MENTALLY ILL OFFENDERS By Ronald Norman Luehrig Much of the literature on aggressive behavior has focused on its more moderate forms which can be experimen- tally induced in the laboratory. The obvious ethical con- siderations and the potential physical dangers involved in extreme forms of violent behavior, murder and assault, necessitate their being studied in an ex post facto manner, usually involving either the demographic or case study. Al- though following the ex post facto format, this project at- tempts to discern empirically the relevant psychological dimensions of these difficult-to-study phenomena. Traditionally, murder and assault have been viewed as nearly coinciding points on a continuum of aggressive be- haviors. In general, these phenomena have been viewed as a single class of behavior, namely, extreme violence. This study challenged that view, because clinical ex- perience suggested that murder and assault are psychologi- cally different events. Primarily this study explored this suggestion with special emphasis on impulse control. Secon- darily it explored the possibility that the selection of the victim is not a matter of chance, but related to the person- ality characteristics of the offender. Only two previous studies are available in this area. q 75" 5, Ronald Norman Luehrig Both dealt with demographic data and concluded that murder and assault are basically similar events. Ten hypotheses were presented regarding the relationship between murder and assault focusing on impulse control, tol— erance of ambiguity, alienation, rigidity, social responsi- bility, self-concept, ego-strength, ascendance-submission, introversion—extroversion, and overt hostility. Two related questions were also posed without stating specific hypotheses: (1) What is the effect of the relationship to the victim-- familial vs. nonfamilial victim--on these ten dependent vari- ables; and (2) are there any interaction effects between the two independent variables? Subjects consisted of the us mentally ill offenders who had committed murder and assault and who met all criteria. Two "prison filler" subjects were included in the Assault- Pamilial group. Thus, there was a total of u? subjects selected from the #78 cases reviewed, and reasons for the high exclusion rate are discussed. The subjects fit into four experimental cells: Assault-Familial (n=7); Assault- Nonfamilial (n=10); Murder-Familial (n=15); and Murder- Nonfamilial (n=lS). They were tested in groups, and each subject proceeded at his own speed. All dependent variables were measured by available objective, paper—and-pencil tests. Data were analyzed in a 2x2 analysis of variance design. All individual comparisons were tested by means of t-tests, F- tests, or Z-tests, with alpha set at .05, using two-tailed tests. Only one specific hypothesis--murderers are more Ronald Norman Luehrig overcontrolled than assaulters-~was supported (p<:.05). Other findings of this exploratory project included: (1) Assaulters were better educated than murderers (p<:.025); (2) Assaults were committed more frequently by single men, murder by married men (p‘<.05); (3) Assault was committed more frequently by blacks, murder by whites (p<:.025); (u) Murderers showed a greater diversity in their ages than assaulters (p<:.05). These four findings contradict the available literature in that they point out significant demographic differences between murderers and assaulters. Further demographic comparisons revealed: (5) Blacks acted out against family non-members more frequently than against family members (p<:.025); (6) Catholics acted out against family members more than did Protestants (p<:.0002), but within their families Catholics injured rather than killed their victims (p<:.001); (7) Protestants acted out against family non-members more frequently than against fa- mily members (p<:.001); (8) Assault-Familial subjects had a higher I.Q. than Assault-Nonfamilial subjects (p<:.01). Analyses of the dependent variables indicated: (1) Those murderers who killed family members were less overcontrolled than those who killed family non-members (p<:.05); (2) Within the nonfamilial victim group, those individuals who murdered their victims were more overcontrolled than those who as- saulted theirs (p<:.01); (3) Within the murder group, non- familial murderers were more submissive than familial mur— derers (p<:.0$); (u) Within the nonfamilial group, those who murdered their victims were more submissive than those who Ronald Norman Luehrig assaulted theirs (p<.05); (5) Within the assault group, familial assaulters showed less overt hostility than nonfa- milial assaulters (p¢:.05); (6) Within the nonfamilial group, those who assaulted their victims showed greater overt hos- tility than those who murdered theirs (p<.05). Because it was not possible to randomly select the sub- jects in the four experimental groups, these findings cannot legitimately be generalized to a larger population of men- tally ill offenders and certainly not to the larger, non- mentally ill population of murderers and assaulters. The results are discussed in light of past and future research. This project, if nothing else, demonstrates that the choice of victim should prove of importance in understanding the personality dynamics of the extremely violent individual. Although not as dramatically as hoped, this project also demonstrates that, contrary to both popular and professional opinion, murder and assault may well be different psycholog- ical events that demand to be studied as separate phenomena if we ever hope to stem the rapidly rising murder and assault rates in our society. MURDER AND ASSAULT: AN EXPLORATORY COMPARISON OF MALE MENTALLY ILL OPPENDBRS By Ronald Norman Luehrig A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Psychology 197” © Copyright by RONALD NORMAN LUBHRIG 197'4 To LAUREL without whose inspiration and patience this would have been impossible... ii ACKNOWLEDGEMENTS I wish to express my gratitude and appreciation to Dr. Dozier Thornton for his patience, encouragement, and guidance as chairman of my committee. I would further like to thank professors Bertram Karon, Andrew Barclay, and William Mueller for their assistance as members of that com- mittee. In addition, my thanks got to Dr. A. A. Birzgalis, Medi- cal Superintendent (retired) of Ionia State Hospital and his staff for support and assistance in facilitating the collec- tion of this data. Mr. William Kime, Director of Research, Michigan Depart- ment of Corrections, deserves thanks for steering me through that Department in my attempt to find additional subjects. At the Michigan Reformatory I owe thanks to Mr. William Weideman, Director of Treatment, and one very special friend. At the State Prison of Sourthern Michigan my thanks go to Mr. William Grant, Administrative Assistant to the Warden, and his staff, especially all those helpful people in the records office. A special thank you must go to Mr. Lawrence Levey who Spent many weekend hours aiding me in testing the subjects in this study and helping in the grueling task of proof- reading. I would also like to thank Mr. Thomas Goretzka for iii his help in test administration. For permission to reproduce the published test materials in the test booklet, I thank Professor H. G. Cough and Mr. J. D. Black of the Consulting Psychologist Press, Inc. (Cali- fornia Psychological Inventory); Professor F. Nulty of the Edu- cational Testing Service (Minnesota T-S-E Inventory); Profes- sor S. Budner and Duke University Press (Tolerance of Ambiguity Scale); Houghton Mifflin Company (Ascendance-Submission Scale, royalties paid); Mr. J. Ricks, Jr., The Psychological Corpora- tion (MMPI, royalties paid); and Professor E. I. Megargee and the American Psychological Association (Overcontrolled Adjec- tive Check List and the 0-H Scale). I would also like to thank the Department of Psychology, Michigan State University, for reproducing the test booklet and the opportunity to do my graduate study under its tutelage. Special heartfelt thanks to my wife Laurel,for everything-- especially, this time, her editorial and typing skills. iv TABLE OF CONTENTS HISTORY The Literature Dealing with Homicide The Literature Dealing with Assault The Literature Comparing Homicide and Assault The Literature Dealing with Familial Homicide The Literature Dealing with Familial Assault STATEMENT OF THE PROBLEM SPECIFIC HYPOTHESES Related Questions METHODS AND PROCEDURES Definitions of the Independent Variables Murder Assault Familial Victim Nonfamilial Victim Apparatus - A Description of the Dependent Variables Demographic Information Form Impulse Control Tolerance of Ambiguity Rigidity Social Responsibility Alienation 13 19 39 us as so 53 an 5a 5a 55 56 56 57 57 57 61 63 SN 66 Self-Concept Ego-Strength Ascendance-Submission Introversion-Extroversion The MMPI and CPI The Overt Hostility Measures Procedure SUBJECTS, Demographic Data Summary of Demographic Data Control Subjects RESULTS Impulse Control Tolerance of Ambiguity Alienation Rigidity Responsibility Self-Concept Ego-Strength Ascendance-Submission Introversion-Extroversion Overt Hostility Review of Significant Findings Comparisons Needing Additional Research Summary of Results DISCUSSION CONCLUSIONS vi 66 68 69 70 71 71 72 71+ 79 an 85 86 86 93 9a 95 96 97 97 99 100 101 103 101: 105 107 111+ APPENDICES Appendix A: Field Research in Violent Behavior: The Perils and the Pitfalls Appendix B: Demographic Information Appendix C: "Overt Hostility" Measures Appendix D: Punting: Within Cell Variance LIST OF REFERENCES A Search for the Sources of vii 115 115 123 12” 130 1w Table Table Table Table Table Table Table Table Table Table Table Table Table Table Table Table Table Table Table 10. 11. 12. 13. IN. 15. 16. 17. 18. 19. LIST OF TABLES Classification of All 2n8 Murder and Assault Cases Found at Ionia State HoSpital Age at Time of Testing I.Q. Education Length of Institutionalization Age At Offense Race in Percents by Group and Totals for the Independent Variable Dimensions Religion in Percents by Group and Totals for the Independent Variable Dimensions Marital Status at Time of Offense in Percents by Group and Totals for the Independent Vari- ble Dimensions Total Prior Arrests Total Prior Arrests: Transformed Data Analysis of Variance Total Prior Arrests Summary: Prior Felony Arrests Analysis of Variance Arrests Summary: Prior Felony 0-H Adjective Check List: Original Data O-H Adjective Check List: Transformed Data Analysis of Variance Summary: 0-H Adjective Check List Self-Control Analysis of Variance Summary: Self-Control viii 77 79 80 80 80 81 81 82 83 87 87 87 88 88 89 89 89 90 90 Table Table Table Table Table Table Table Table Table Table Table Table Table Table Table Table Table Table Table Table Table Table Table Table 20. 21. 22. 23. 2“. 25. 25. 27. 28. 29. 30. 31. 32. 33. 3%. 35. 36. 37. 38. 39. #0. MI. “2. “3. Overcontrolled Hostility Analysis of Variance Summary: Overcontrolled Hostility Intercorrelation Matrix for the Four Measures of Impulse Control Tolerance of Ambiguity Analysis of Variance Summary: Tolerance of Ambiguity Alienation Analysis of Variance Summary: Alienation Rigidity Analysis of Variance Summary: Rigidity Responsibility ReSponsibility: Transformed Data Analysis of Variance Summary: Responsibility Self-Concept Analysis of Variance Summary: Self-Concept Ego-Strength Ego-Strength: Transformed Data Analysis of Variance Summary: Ego-Strength Ascendance-Submission Ascendance-Submission: Transformed Data Analysis of Variance Summary: Ascendance- Submission Introversion-Extroversion Analysis of Variance Summary: Introversion- Extroversion Overt Hostility Analysis of Variance Summary: Overt Hos- tility ix 91 91 92 93 9a 9a 9a 95 9s 96 96 96 97 97 98 98 98 99 100 100 101 101 102 102 Table Table Table Table Table Table Table Table Table Table Table Table Table A1. A2. A3. AN. D1. D2. D3. Du. D5. D6. D7. D8. D9. Reasons for Exclusion: Ionia State Hospital Results of State-Wide Subject Search Reasons for Exclusion: Reasons for Exclusion: Choice of Weapon: Analysis of Variance Arrests Analysis of Variance Check List Analysis of Variance Analysis of Variance Hostility Analysis of Variance tility Analysis of Variance tility Analysis of Variance tive Check List Analysis of Variance tility Summary: Summary: Summary: Summary: Summary: Summary: Summary: Summary: Ionia Reformatory Jackson Prison Percent Personal Weapon Prior Felony 0-H Adjective Self-Control Overcontrolled Overt Hos- Overt Hos- O-H Adjec- Overt Hos- 116 119 120 121 131 132 133 135 136 137 139 190 1H2 HISTORY Much of the literature on aggressive behavior has fo- cused on its more moderate forms which can be experimentally induced in the laboratory. The obvious ethical considera- tions and the potential physical dangers involved in the more extreme forms of aggression, homicide and assault, necessitate their being studied in an ex post facto manner, involving two basic research designs: the demographic and the case study. Although following the ex post facto format, this project at- tempts to discern empirically the relevant psychological di- mensions and relations of these difficult-to-study phenomena. Traditionally, murder and assault have been viewed as nearly coinciding points on a continuum of aggressive be- haviors. In general, these phenomena have been viewed as a single class of behavior, namely, extreme violence. This study proposes to challenge this view, because clinical experience suggests that murder and assault are psy- chologically different events. Primarily this study explores this suggestion, with special emphasis on impulse control. Secondarily it explores the possibility that the selection of the victim is not a matter of chance but is related to the personality characteristics of the offender. It should be noted that only two demographic studies have dealt Specifically with the comparison of murder and 1 assault. Nonetheless, a general review of the literature is presented to give the reader the flavor of the work being done in these areas. THE LITERATURE DEALING WITH HOMICIDEI In a fairly typical demographic investigation of homi- cide, Berg and Fox (1997) reported findings based on two hun- dred imprisoned murderers. They indicated that murderers were generally older than the rest of the prison population and performed less well on intelligence and achievement tests. Those men who killed men were younger than those who killed women; those men who killed women were more ego-involved and tended to behave in a more violent manner. The younger men generally murdered for gain or to avenge an insult; older men tended to slay women, primarily for reasons of infidelity or as a result of arguments over money. Harlan (1950), in a study employing five hundred sub- jects, attempted a more systematic classification of the in- terpersonal (attacker-victim) patterns found in cases of murder. The most striking finding was that homicide appears to be primarily an in-group phenomenon, according to race. He described six of the most common interpersonal patterns: black male kills black male; black male kills black female; black female kills black male; black female kills black fe- male; white male kills white male; white male kills white 1In this study no distinction will be made between the concepts ”murder" and "homicide"; they will be used inter- changeably throughout. female. Lanzkorn (1963) provided some background information on a different population of murderers in his survey of 150 patients who were committed to the Matteawan State Hospital over a five-year period (1956-1961) after they had been charged or indicted for murder. He described some of the basic demographic information for a group of hospitalized murderers, reporting that of these 150 patients the majority were male (n=118 vs. 32), white (86 vs. 68), and Catholic (92 vs. 58). Most were between the ages of twenty and thirty- nine (91 vs. 59), of average intelligence (90 vs. 60), and previously employed as unskilled workers (80 vs. 50). The last tally excludes the 20 patients who were housewives. He also looked at such additional vital statistics as marital status (89 were single, nu married, 32 widowed, and 25 divorced or separated), the method used to kill (stabbing, 55; shooting, “7; strangling or clubbing, l7; throwing the victim from a high place, 10; and drowning, burning, poisoning, gassing, or hitting with a car, 9), the relation of the victim to the murderer (95 members of the family, including in—laws, 62 family non-members [note: some subjects killed more than one person]). The psychiatric diagnoses included schizo- phrenia, paranoid type, 38; "psychosis with psychopathy, per- sonality instability with reactive features, with or without alcoholism," an; schizophrenia, mixed undifferentiated type, 18; schizophrenia, catatonic type, 12; psychosis with mental deficiency, 8; manic-depressive psychosis, depressive type, 7; u involutional psychosis with paranoia, 6; paranoid condition, u; involutional psychosis with melancholia, 2. In a study of ninety-two non-hospitalized murderers com- pared with their brothers nearest to them in age, Gillin (1952) found that the murderers left school earlier, had full-time jobs at an earlier age, and stayed on any one job for a shorter period of time. He also found that, if the murderers were married, their marriages were disharmonious, and their wives generally had more education than themselves. The mur~ derers felt that they were favored by one of their parents and generally showed a greater appreciation of their mothers. Unfortunately, these demographic studies provide little more than a general overview of the problem at hand. For example, Williams (1960) found that the murderers' perception of their mothers was bidimensional, rather than unidimensional, as indicated by Gillin (1952). As a result of his treatment of patients serving life sentences for murder, Williams indi- cated that early in infancy these patients had split their mental images of their mothers into two parts: the idealized, giving mother and the bad, persecutory mother. This dual perception of the mother was generally reinforced by an epi- sode in childhood. In contrast to the demographic research, the more fruit- ful, non-demographic studies suggest a great many causal factors of homicidal behavior. Banay (1952) proposed a means for establishing a classification system of the cultural and motivational factors involved in homicide. On the "sub- cultural" level, he suggested that the motivation is related to a dulled ethical perception which impairs the murderer's ability to control his "violent tendencies." On the "cul- tural" level, the inability to control "violent impulses" results from the conditioning by society--via the mass media, for example--to be less concerned with these violent out- bursts. On the “super-cultural" level, the controls are rigid and fragile because of excessive training in control, generally resulting from higher social status. These controls are adequate under usual conditions, but shatter when violent impulses become too strong. Close inspection of Banay's system shows that it is a general model rather than a de- tailed analysis of the dynamics involved. The two most im- portant features of his article are the emphasis which is placed on motivation and the presentation of the description of the murderer as an individual whose impulses, at least on one level ("super-cultural"), are rigidly controlled. This concept appears to be the forerunner of the "overcontrolled" concept. Tanay (1969) noted that in the cases he evaluated, all provided evidence of "a severe super-ego." Incidentally, Williams (1960) also presented a descrip- tion of the murderer which is equally suggestive of the idea of overcontrol; although, like Banay, he does not use this terminology. In one of the two cases presented there was a deep split between the sadistic murderous part of the patient's personality and the remainder of his personality, which was gentle, constructive, and inhibited. The patient showed no overt aggression. Even as a child, he never hit back when he was beaten by his father, and he was unable to kill puppies or mice when he was told to do so. Banay's emphasis on psychological motivation contrasts with Harlan's (1950) focus on motivation in the legal sense (the sex triangle, quarrels over financial loss, revenge for an insult, and the like) as part of his study of attacker- victim patterns in cases of murder. Banay (1956) continued to direct his attention to the importance of motivation in the psychological sense. Unfortunately, Banay's later work in this area (1959) focused on two concepts, "perversion of instincts," which is an outgrowth of the "sub-cultural" level of his classification system (Banay, 1952) and the ”epileptic model," which is his attempt at a neurological model. This model suggests that there is a massive discharge of neural energy as a consequence of frustration. Both of these concepts would seem to lead to blind alleys, because as stated they do not lend themselves to rigorous empirical investigation. When more clearly defined, however, these concepts can be of significant importance. The latter concept-~the "epileptic model"--calls to mind Megargee's concept of "overcontrolled aggression," while the former concept--"perversion of in- stincts"--suggests Megargee's concept of "undercontrolled ag- gression." In the same paper (Banay, 1959), however, he did intro- duce another important concept that was dealt with in greater detail by Hartley (1959), i.e., the perceptual distortion of reality. Hartley pointed out that when an individual per- ceives that his movement toward a goal is being blocked, whatever the cause, he will attempt to do something to defend himself against the blocking agent. This is very similar to the original version of the frustration-aggression hypothesis (Dollard, Doob, Miller, Mowrer, and Sears, 1939). It differs in emphasizing the subjectivity of perception, whereas the frustration-aggression notion deals with objective factors impeding the attainment of a goal. Palmer (1960) also found the notion of frustration resulting in aggression as an im- portant factor in murder. Specifically, the murderer has experienced a greater amount of frustration throughout his life and has had less Opportunity to express his feelings-- vent his aggression or anger--in socially acceptable ways, than his nearest age brother who did not commit murder. (Palmer did not report whether any socially unacceptable ex- pressions of anger occurred.) Is this description of events not suggestive of a greater degree of control over hostile impulse? Sattan, Menniger, Rosen, and Mayman (1960) also pre- sented an idea similar to perceptual distortion, but they discussed it in a different context. The subjects--unfor- tunately there were only four-~were described as suffering from a ”blurring of the boundaries between fantasy and re- ality, with periods of altered states of consciousness, blunted and shallow emotional reactions, and a violent and primitive fantasy life." They labelled this state "person- ality disorganization," which was a result of "extreme parental violence and early severe emotional deprivation." The family dynamics also had the effect of lowering the "mas- culine self-image." Their idea follows Banay's (1959) and Hartley's (1959) concepts of the distorted perception, but they extended it by adding the notions of blunted affect and lowered self-esteem, which were the result of distorted early interpersonal relations within the family. In their early attempts to ascertain the nature of the personality of imprisoned murderers, Romano and Paolella (1958) suggested that there are two distinct personality types. Basing their conclusions on their analysis of pat- tern and scatter on the Wechsler Bellevue, the Rorschach, and the TAT records of 20 imprisoned murderers, they found that one type is characterized by hyperthymia in emotional responsiveness, while the other type is characterized by a cold and detached schizoid affect. It must be pointed out that this is the only study in the literature reviewed that finds these two distinct emotional patterns within a popula— tion exclusively comprised of murderers. Megargee (1966) found a similar, but not identical, pattern in a mixed group of extremely violent individuals. He labelled the two pat- terns as undercontrolled and overcontrolled, respectively. Yet these findings are not totally inconsistent with the findings presented by Podolsky (1968) who focused on a parti- cular sub-group of murderers--those suffering from a manic- depressive psychosis. Romano and Paolella's description of the hyperthymic pattern sounds much like the manic phase of this form of psychosis. At first glance, Podolsky's descrip- tion of the murderer as committing his violent act when in a state of emotional excitement would seem to contradict the notion of blunted affect suggested above (Sattan et a1., 1960). But when the total symptom complex of the manic- depressive psychosis is taken into account, we see two fea- tures which are similar to those described by Sattan et al. First, there is a marked loss of self-esteem which is one of the key factors in the depressive phase of the cycle. Second, there is a distortion of certain emotional components of the personality, eSpecially the lowering of affective tone during the depressive phase of the cycle (Fenichel, 1958, Ch. 17). These aspects of the total description of the manic-depressive psychosis are neglected by Podolsky in his description of the murderers in his sample. The characteristic which he did in- clude--hyper-excitement--is nothing revolutionary, as evidenced by the idea that murder is an exploding of affects, suggested by Banay's concept (1952) of the murderer's inability to con- trol his "violent tendencies." In another basically demographic study, which in part dealt with the psychopathology found in cases of homicide, Neustatter (1965) attempted to classify the "state of mind" of the subjects at the time the crime was committed. He re- ported that in 31 of the 59 cases of murder there were no prior convictions for any offense. In 16 of these cases the subjects were found to be in a state of depression. In 12 cases of the original 59 there were convictions for "irrele- vant" offenses. Of the 19 cases with bad records, nine were 10 found to be psychopaths. Sixteen of the total number were psychotic, while 56 showed some psychiatric disorder. The killings occurring in the depressive states fell into three distinct groups: (a) killings resulting from delusional thinking, (b) "logical" killings, and (c) murders committed in an overwrought state of anger. The author concluded that many murderers are not fundamentally vicious and supported this conclusion with case examples. The emotional state of the murderer, as Specifically suggested by Sattan et al. (1960), was often the result of interpersonal interaction patterns within the murderer's family. A number of studies clearly demonstrate the impor- tance of these relationships. Blackman, Lamberti, and Weiss (Lamberti et a1., 1958; Weiss et a1., 1960; and Blackman et. a1., 1963) pointed out that there was a consistent life pat- tern in all cases studied. Murderers came from "cohesive family backgrounds, where conformity to the rules...was em- phasized.” But, because the demands were so rigid, their attempts at conformity failed, resulting in conflict, frus- trations, and feelings of being alone, isolated, angry, and worthless. Then, for a period of time, they functioned in a marginal way until something upset their tenuous balance, re- sulting in more tension and anger. In this state, the slightest insult or provocation set off the hostility and re- sulted in murder. This pattern was also found by G. B. Smith (1958), S. Smith (1965), Kahn (1960), Galvin and Macdonald (1959), and, to a more limited extent, by Banay (1956 8 1959). 11 Greene (1988) provided a case study that serves as a good illustration of these descriptions of familial relation- ships. Greene divided his case presentation into two areas: the circumstances, what actually happened, and the clinical picture. (What follows is a digest of Greene's report.) The Circumstances: This man spent the four days prior to the crime on a drinking spree with two of his buddies: 0n the afternoon of the fourth day, they decided to go to a nearby town to meet some girls. When they arrived--after more drinking--this man's date did not Show up, so he started to hitch-hike to his favorite sister's home in another town. He had had some difficulty getting a ride, but finally a middle—aged, male Sunday school teacher picked him up. Two miles down the road, the man shot the teacher in the temple, killing him instantly. Later, it was discovered that neither man had ever seen the other before, but the driver (teacher) had said, in the serious, reproving, remonstrating tone of a parent, "You are about drunk, aren't you? Why don't you quit drinking and make a man out of yourself?" After this state- ment, the murderer's mind "went blank,” and he shot before he knew what was happening. The remaining details of the capture of the murderer are irrelevant here, but it is interesting that he left such a clear trail that it looked as if he wanted to be captured. The Clinical Picture: A thorough clinical investigation of the case revealed the following factors. This man was profoundly disturbed, although neither psychotic nor legally 12 insane. He was diagnosed as having "constitutional psycho- pathic inferiority." His history revealed parental depri- vation, strict family setting, punative father, intense hostility towards the father who was seen as favoring an older brother, and the father's use of the same provoking statement-~st0p drinking and make a man out of yourself--on repeated‘occasions. Wolfgang (1957) described another pattern of homicide which he calls "victim precipitated criminal homicide." He pointed out that this form of homicide involves an intense personal interaction in which the victim's behavior plays an important role. In this form of murder the victim is, at the outset of the altercation, the attacker. The victim has a "direct, immediate, and positive contribution to his own death, manifested by his being the first to make a physical assault." Furthermore, in many cases the victim and the of- fender have many characteristics in common. In some cases this form of homicide results from the interpersonal inter- action of two similar personalities, and "it is probably only chance which results in one becoming victim...." The idea that the victim is a weak and passive individual, who seeks to withdraw from the interaction, and the attacker an aggres- sive individual who seeks out his victim, is a popular mis- conception. Society's attitude is usually positivelyfin favor of the victim and antagonistic toward the offender, who is viewed as violent and dangerous. Wolfgang's findings suggest that this 13 is a misconception. Megargee (1965a) described a similar situation for the criminal act of assault. Schultz's paper (1968) also reported similar findings, but goes a step fur- ther in suggesting that there are degrees of victim involve— ment, and the courts should take that fact into account when pronouncing sentence. THE LITERATURE DEALING WITH ASSAULT In an article with a demographic orientation, Pittman and Handy (1968) gave the following description of the "typi- cal pattern" of aggravated assault, based on a random sample of 281 assaultive acts committed during the year 1961 in St. Louis, Missouri. They indicated that assault usually occurs on the weekend between 6:00 p.m. Friday and 6:00 a.m. Monday, with the peak on Saturday, between 10:00 and 11:00 p.m. As- saults are also most likely to occur during the months of July and August. When the assaulter is female, the act is most likely to occur indoors, usually in a residence. When the as- saulter is male, the attack will usually occur on a public street. When the victim and the assaulter are related, the act will usually occur in the privacy of the home. These acts are most likely to occur in lower socio-economic, black neigh- borhoods. This last finding, of course, may be a result of the dif- ferential rate of arrests in the various areas of a given city. As Barron (1958) pointed out in his textbook on juvenile de- linquency, the records of delinquency are subject to the policy of the local police departments, and the methods of 1N handling misbehavior in different areas of the city. This policy, he suggested, is in part determined by the socio- economic level of the offender. Conceivably, this artifact of reporting could affect the statistics related to certain aspects of criminal assault, as it does in the case of juve- nile delinquency. For example, in most cities there are more police officers assigned to patrol the lower income and ghetto areas than to patrol the other areas of the city. Thus, it is more likely that an assaultive act will be no- ticed by, or be brought to the attention of, the police than in the higher income or suburban areas of a city. Pittman and Handy further reported that the weapon most frequently used by both men and women assaulters is the knife, while the gun is the second most frequently used wea- pon. In acts involving a white person as attacker, "personal force” is more likely to be involved than when the attacker is a black. Generally, the use of alcoholic beverages or drugs by either the attacker or the victim is rarely found to be associated with cases of assault. As in the case of homicide (Harlan, 1950), assault will more likely be an in-group phenomenon, where both the victim and the attacker are of the same race and generally of the same sex. Also, crimes of assault will usually have one vic- tim and one attacker, both will have been born outside the city in which the crime took place, and both will typically be from the same age group, most frequently between the ages of 20 and 35. The offender will usually be older, but the 15 victim is more likely to be married. Female attackers will most likely either be related to, or will have had recent intimate associations with, the victim. This, however, is not the case when the male is the attacker. In a study which goes beyond demography, Roebuck and Johnson (1962) focused on a very specific pOpulation of as- saulters, the black drinker and assaulter, which suggests that generalization to a larger population should be made with extreme care. Their findings can, however, provide re- levant dimensions to guide the study of the general pheno- menon of assault. They found that the black drinker—assaulter was, on the average, raised in.a home which had a rigid, fun- damentalist background, where the rules were effectively, but erratically, enforced by a domineering father. His pri- mary ties were close during childhood, and so continued throughout his adulthood. His general environment was less crimogenic than that of the larger control sample of crimi- nals. Although no direct evidence was gathered, Roebuck and Johnson concluded by hypothesizing that parental efforts at character develOpment were to some degree successful, so that this group could manifest hostility only after alcohol had weakened inhibitions. This speculative conclusion does not, however, coincide with Pittman and Handy's (1968) finding that alcohol was rarely associated with assaultive acts. In contrast, Easson and Steinhilber (1961) found that all of their assaultive patients came from families that 16 condoned acting out, while appearing to be socially accep- table, "normal" families. Over a period of twelve months they saw seven boys who had made murderous assaults, and one who had actually committed murder. Each of the seven boys who had made murderous assaults were found to be emo- tionally closely tied to their mothers, with whom they iden- tified in a hostile manner. The fathers of these boys were not available to them, due to death, brutal rejection of the boy, or the inability of the father to take a definite mas- culine role; and, therefore, identification with a male father-figure was impossible. Where the fathers were physi- cally present, there was evidence of defects of conscience in these fathers. For the most part these boys were maintained in a depen- dent position in the family. Their lives were characterized by lack of privacy, physical overcloseness, and in most cases there was overt sexual seduction on the part of the mother. A pattern of physical violence was set for these boys by either parental example or parental approval. Most boys had a collection of knives and guns, which they were permitted to retain even after their assaultive attacks and menacing behavior. In each case the boy was given reason to believe that his parents expected him to be physically violent, and no constructive attempts were made by the parents to deal with these assaultive acts. When these boys were stopped from their violent behavior, it was primarily because they had committed the act in some public place where the attacks could be stopped by an outsider. 17 From the study of these eight patients, these authors concluded that physical illness, sibling rivalry, parental seductiveness, or brutality, etc., may cause interfamilial turmoil; but only where there is parental approval is the child's conflict acted out in a violent fashion. In another study of youthful assaulters (average age was 18 1/2 years old), Moles (1963) concluded that, as com- pared to youthful thieves (a matched group), these assaul- tive subjects had an early life characterized by "parental deprivation, [which] when continued and arbitrary, [made] various later unstructured interpersonal situations appear frustrating." He based this conclusion on the following findings: Although both groups admitted to equally frequent assaults, and both groups had friends who admired "hard tough" boys, the assaultive boys were more impulsive in their attacks. Due to a generalized expectation of deprivation and weak internalized standards, the assaultive boys tended to interact with friends in a manner that allowed them to impul- sively gratify their own needs. In a case history report, Lunsky (1955) attempted to detail the dynamics of "murderous acting-out" (assault) in a schizophrenic individual. Murderous acting-out is described as a defense to master anxiety: overwhelming anxiety re- sulting from "abhorrent cravings" erupting into awareness, "cultural accretions" are lost, and "primitive patterns" be— gin to dominate the individual's behavior. He is then con- cerned only with survival. Delusions and hallucinations 18 create a "state of omnipotence." When this fails, intense motor activity is the result, and "direct aggression toward a representative of a past significant figure" culminates the process. This culmination is the "symbolic destruction of the significant object." A case history of a “5-year-old male is presented in support of the findings. In a unique approach to the study of violence, Toch (1966) was concerned with the detailed description of the interper- sonal interaction patterns between the victim and the attacker just prior to and during the altercation. He focused on two assaultive situations: assaults on police officers by citi- zens and assaults on one prison inmate by another. In brief, Toch found that in the case of the "games be- tween officers and citizens" the violence occurs most fre- quently (90% of the cases examined) after the following sequence. The sequence "begins with an order or request by the officer, which elicits a contemptuous response from the citizen.... The sequence repeats itself and ends a variable number of steps later--in some instances after a notification of arrest, in others without it..."--with the citizen as- saulting the officer. "The second most frequent sequence [27% of the cases] is one in which the violence is already manifest as the officer enters the stage." When the officer cannot prevent the shifting of the target of aggression to himself, his entry onto the scene results in the enactment of the basic sequence which results in an attack on himself. After determining these basic patterns, Toch attempted 19 to discover the roles played by each individual in the alter- cation. He focused on the "patterns among the incidents of violence in which g_gi!gn person has been involved." He found that there are basically two such patterns. "One in- cludes types of attitude or personal reaction. The chief source of patterning here lies in the limited range of situ- ations that a violent person defines as justifying or as re- quiring force. [The second pattern] of violence is that of violence-prone strategies." These he referred to as a "re- latively permanent [set of] personality traits and disposi- tions." Unfortunately, the literature in the area of assault is meager. (See Appendix A for possible reasons contributing to the scarcity of formal research in this area.) In most cases the research that is available does not give sufficient in- formation from which to derive reasonable hypotheses. Be- cause of this situation, hypotheses are based primarily on the material found in the area of homicide. THE LITERATURE COMPARING HOMICIDE WITH ASSAULT Most of the literature dealing with both murder and as- sault either combines the two events or confuses them. Teele's paper (1965) is a good example of how these two con- cepts are confused. In his report he used the terms "assaul- tive" and "homicidal" interchangeably. This confusion exists even in law. In the state of Michigan, for example, when an individual who is a patient at another state institution dem- onstrates assaultive and/or extremely aggressive behavior, he 20 can be sent to the Ionia State Hospital as "homicidal." In the statutes relevant to the Michigan Department of Mental Health (1968), dealing with the transfer of patients with "homicidal tendencies," the law states: 330.67 Transfer of patients with homi- cidal tendencies. Sec. 57. The medical superintendent of any hospital, home or institution may, with the approval of the state hospital commission, transfer any or all insane persons under treatment in any such in- stitutions who have been guilty of an act of homicide previous to admission to such institution, and whose presence is dangerous to others, likewise all insane persons who have committed any act of homicide while under treatment in any such institutions to the Ionia State Hos- pital. In case any atient under treat- ment in any of such Institutions develops unmistakable dangerous or homicidal ten- dencies, rendering his presence a source of danger to others, proceedings may be instituted as above. Both Lunsky (1955) and Easson and Steinhilber (1961) are also guilty of concept confusion. Both papers deal with as- sault, but the former describes "murderous acting-out" while the latter uses the phrase "murderous assault" to describe the behavior. The combination of the two events is far more common than the confusion of them. For example, in this author's Master's Thesis (Luehrig, 1968) the subjects were divided in- to three categories: violence against a person, violence against prOperty, and non-violence. 0f the three dependent variables, two--aggression anxiety and fantasy aggression-- showed no significant differences between the three groups, While one--the measure of "aggressive tendencies"--showed a 21 significant difference: the group that had committed violence against property showed greater aggressive tendencies than either the group that had acted violently against persons or the group that had acted in a non-violent fashion. This sig- nificant finding, however, contradicted the proposed hypo- thesis which was: If a population of prison inmates is subdivided into three sub-pOpulations (Violent Personal, Violent Non-personal, and Non-violent) on the basis of the na- ture of their offenses, then the Zaks and Walters aggression scale will dif- ferentiate these three groups with the Violent Personal having the highest score, the Violent Non-personal having the next highest score, and the Non-violent having the lowest score, i.e., a direct rela- tionship exists between measured and overt aggression. Now, after several years additional experience in working with violent individuals, it seems possible that the reason for both the non-significant results and the signi- ficant finding which were in contradiction with the stated hypothesis might well be the heterogeneous nature of the "Violent Personal" group. This group consisted of subjects who had committed such varied offenses as assault with intent to rob, armed and unarmed; felonious assault; manslaughter; negligent homicide; rape; robbery, armed; robbery, unarmed; Inurder; statutory rape; gross indecency; indecent liberties; and kidnapping. When the proportions of assault and homicide cases were calculated, 11% of the group was made up of as- sault cases of a non-sexual nature, 23% were homicide cases, and the remaining 66% was made up of the remaining eight 22 categories of crimes. Looking at this from another point of view, namely, fo- cusing only on the subjects who committed non-sexually moti- vated assault and homicide, revealed that approximately 33% of these were cases of homicide and approximately 67% were cases of assault. Now, if murder and assault are psycholog- ically different events, as will be suggested later in greater detail, then the mixing of various crimes into one category in all probability confounded the analysis. In addition it must be pointed out that, even though the findings seemed to be meager as a result of the mixing of types of crimes in what was then classified as the most ex- tremely aggressive group, the significant difference still remains; the moderately aggressive group (Violent Non-personal) showed a greater degree of "aggressive tendencies" than did the extremely aggressive group (Violent Personal). This finding certainly contradicts the usual assumptions about ag- gressive crimes--namely, that murder is the most violent of crimes committed by the most violent of persons--which was the basis for the hypothesis quoted above. Pokorny (1965) reflects another assumption regarding the nature of the crime of homicide. He wrote, "It has been stated that homicide offenders have, as a long—term charac- teristic, greater impulsivity, less ability to control emo- tions. It has been found that in many instances of homicide fierce arguments precede the murder, with progressive esca- lation in emotion and violence. It may therefore be largely 23 a matter of chance that an offense becomes a homicide or an assault.” Bromberg (1961) holds a similar view, saying of the murderer the "mechanisms of the ego are insufficient to curb direct expressions of aggressive impulses." Not only do these reports contradict the findings described above (Luehrig, 1968), but more recent research (Megargee, 196ua, 1968b, 1965a, 1966, 1967) has suggested that the more aggres- sive criminal acts may have been committed by individuals who are, in fact, "overcontrolled;" that is, they are generally less aggressive than individuals who commit less violent crimes. Statistical evidence reported in Federal Probation ("News from the Field," 1968) indicates that only three of 1,303 (0.2%) individuals paroled after serving a sentence for homicide committed another homicide. Ninety-one percent had a "favorable parole," and ninety-eight percent were free of any conviction during the first year on parole. Is this the behavior pattern expected of an impulsively aggressive individual? Megargee (1969a) expressed his concern over the state of research in the area of aggression because "...we find a great gap between the aggression described in‘our journals and that described in our newSpapers." He explained this in terms of the difficulty in, and ethical restriction on, research that evokes extremely aggressive responses, such as homicide. Be- cause of this situation, the experimental research reports deal only with the moderately aggressive reSponses, and thus 2“ their findings may not be applicable to the more extreme forms of aggression. "In fact," he said, "there is good reason to believe that when extremely assaultive or homicidal peOple are studied, the dynamics underlying the aggressive behavior may be found to be quite different from those using more mildly aggressive behavior." In his monograph, Megargee (1966) reviewed his research concerned with this issue and found that by-and-large, when an individual engages in mod- erately aggressive forms of behavior, his psychological dy- namics usually center around an undercontrolled aggressive pattern. On the other hand, those individuals who engage in the extremely aggressive forms of behavior will display dy- namics which are focused on an overcontrolled pattern, but not exclusively. He indicated (Megargee, 1968a) that "It would follow that a group of peOple who have committed Ex- tremely Assaultive acts such as homicide or assault with a deadly weapon would be likely to include some peOple of the Chronically Overcontrolled type and some of the Undercontrolled Aggressive type. A group of people who have engaged in Mod- erately Assaultive behavior, such as fist fights, should, on the other hand, consist almost exclusively of the Undercon- trolled Aggressive type." All his research in this area con- firms this formulation. Megargee combined assault and homicide in his "Extremely Assaultive" group and was forced to conclude that this group is made up of both the overcontrolled and the undercontrolled aggressive individuals. It seems possible that by mixing his 25 murderers with his assaulters, he ran into the same difficulty that Luehrig (1968) encountered; that is, when these cases are mixed, the results might reflect this mixture of psycho- dynamics, which in Megargee's terms could be expressed as finding both the undercontrolled and the overcontrolled ag- gressive individual in these extremely aggressive groups. Thus, again we see that it is possible that the combining of types of crime may affect the results obtained. Similarly, Easson and Steinhilber (1961) had a hetero- geneous sample. In their clinical, non-statistical study of murderous aggression by children and adolescents, they were forced to conclude not by providing the answer, but rather by asking the question: Why did one individual kill and the others not kill? From the above discussion of the confusion in terminol- ogy and the use of combined or mixed groups in researching the psychodynamics of extreme violence, assault and murder, it appears that it might prove to be sounder methodologically to study these two criminal types separately, even though both have acted out their aggressive impulses and injured an- other individual. The reasons for making this separation will, it is hoped, become even clearer in the remainder of this section. There are few studies that directly compare the acts of assault and homicide. Those that do, do not provide enough information concerning the personal and interpersonal factors to make them worthwhile, in that they leave the reader with 26 the same question with which he started reading the article. This can best be expressed by reporting the concluding com- ments made by Easson and Steinhilber (1961) in their study of murderous aggression in children. They raised the question this way: "All these 8 boys acted violently with parental permission and approval. Only 1 boy proceeded to actual mur- der. Why only he and not the other 7 committed murder is not clear and requires further intensive study." Nonetheless, according to the literature reviewed, no on has engaged in this type of intensive investigation. A few studies of direct comparison are available. In a demographic study, Pokorny (1965) compared assault with at- tempted suicide, and murder with suicide. At the outset, he speculated that "Just as Suicide has a related, larger cate- gory of Attempted Suicide, Homicide has a related and larger category of Aggravated Assault." He then raised the question "Is Aggravated Assault prOperly viewed as a bungled or unsuc- cessful Homicide, with the same motivations and characteris- tics? Or is it possible that, as in Suicide-Attempted Sui- cide, it is a different, though perhaps overlapping, class of human violence?" He concluded that "Homicide, and Aggravated Assault were similar in all aSpects studied [place, time, per- sons], which suggests that these are basically the same cate- gory of behavior." It must be pointed out that Pokorny asked his research question in terms of "motivations and characteristics," but did the actual research by looking at only the demographic variables: the location of the action, the time of the 27 action, and the persons involved. He did not look at the motivation in a psychological sense, nor did he look at the person in terms of the personality characteristics which would be the variables more relevant to his question. There- fore, his conclusion that the two events are "basically simi- lar" cannot be applied to the psychology of murder and assault, but only to the demography of these crimes. Pittman and Handy (1968) reported a similar study con- cerning the comparison of the acts of assault and homicide. They collected data on assault cases and then compared their findings with the findings of Wolfgang (1958) in his book Patterns of Criminal Homicide. This procedure leaves some room for questioning the methodological soundness of this study, since there is little evidence of the comparability of the two groups. They reported the following conclusions: A comparison of the findings in this study with those of Wolfgang [1958] re- veal more similarities than differences, even though the two studies represent different time periods, cities and police departments. Time. For both...occurrences were higfier on Saturday...for homicide between 8:00 p.m. and 2:00 a.m., and for aggra- vated assault, between 8:00 p.m. and 3:00 a.m. Location. Both crimes occurred more often in public streets than in any other location, with residences second. Summer months accounted for a higher percentage of crimes in both cases, but to a greater extent for homicide. In winter these crimes occurred indoors. Females committed both acts more often indoors. If the victim and offender are related, the crime most likely occurred in a residence.... Wea on. The weapon most often used differea between homicide and aggravated 28 assault; a pistol or revolver was most common in homicide, while a knife was most common in assault.... Alcohol Involvement. The ingestion of alcohol was more common in homicide than assault.... Situational Context. Verbal argu- ments preceded both crimes, but alcohol was involved in the arguments in homi- cide situations more often than in ag- gravated assault cases. Offender-Victim Relationshi . For both crimes, the victim and‘of ender were typically of the same age, sex, and race. There were most often only one victim and one offender. Prior Arrest Records. For both homi- cide and'assault, the majority of the victims had no prior arrest record, while the majority of offenders did. For homicide offenders, two-thirds had a prior record of crimes against a person, while for aggravated assault the number of of- fenders having this type of record...was negligible. [Thus,] the pattern of the two crimes is quite similar. Both acts, of course, are reflections of pOpulation sub- groupings which tend to externalize their aggression when confronted with conflict situations. Both these studies (Pittman and Mandy, 1968, and Pokorny, 1965) concluded that assault and homicide are basically simi- lar phenomena. Pokorny pointed out that "...it may be mainly a matter of chance that an assault becomes a homicide." Pittman and Handy suggested that "...the line dividing aggra- vated assault from homicide is so thin that a factor such as the speed of an ambulance carrying the victim to the hOSpital will determine whether the crime will be aggravated assault or homicide." This way of distinguishing between two events-- particularly two events that have such vastly different con- sequences in the eyes of the law-~seems rather simple-minded 29 and grossly unscientific. One must therefore look at these studies a little more carefully, both in terms of the find- ings reported and the procedures leading to those findings. When describing the method of data collection, Pokorny (1965) reported that "The aggravated assault cases were not studied individually; rather the findings...are taken from tables and summaries of the Police Department Annual Re- ports." He went on to indicate that the homicide data was "...a partial replication of the definitive study of Crimi— nal Homicide by Wolfgang...," again based on the tables and summaries in the Police Department Annual Report. This raises the question of accuracy, not from the point of view of what has been included, but from the point of view of what has been omitted. Another difficulty with both pieces of research is that both started with the assumption that the two phenomena are similar. To quote Pokorny, "...Homicide has a related, lar- ger category of Aggravated Assault." After stating this as- sumption, he went on to raise the question as to whether or not there is a difference. Pittman and Handy (1968) also based their research on a similar assumption, although only implicitly, since they felt the hypotheses set forth by Wolf- gang were directly applicable to the problem of aggravated as- sault. The fact that both studies begin with the assumption of similarity suggests that the methodology employed would tend to supply evidence for that assumption. This, in fact, is the case. Both studies reported that the demography of 30 the two crimes is similar. The question still remains, how- ever; is the psychology of these two crimes similar? Even within the findings of the work of Pittman and Handy one can begin to get answersto this question. Two in- teresting facts that they pointed out in their research sug— gest a difference between the two groups. First, alcohol is usually involved in the commission of murder, while this is not the case for assault. Second, the weapon chosen is a knife in the majority of the assault cases and a hand gun in most homicides. These differences apparently do not concern the authors. Yet these differences suggest that, despite the socio- logical similarities, there may be psychological differences. First, the presence of alcohol in homicide suggests a higher degree of inhibition requiring the effects of alcohol to re- lease the aggressive urges necessary to commit homicide. Second, the higher probability of the use of a knife in the case of assault suggests undercontrolled hostility. Since knives are both a common household utensil and frequently carried, particularly by males, it is possible that use of a knife is more likely to be an impulsive act. That is, as- sault seems more likely to be an undercontrolled form of ag- gression. Conversely, given the finding that the pistol is the weapon most frequently used in homicide and the fact that it is also a far less common implement, the use of the gun may be a less impulsive act. If this is the case, homi- cide is more likely to be committed by a less impulsive or, 31 in Megargee's terms, an overcontrolled person. Besides the possible differences in the degree of con- trol associated with the choice of weapon, there is another factor that may be related; namely, the degree of physical contact between the victim and the attacker. The use of the knife (as is most common in assault) requires close personal contact, while the use of a pistol (as is most common in homicide) does not. Is this a coincidence? Or does this suggest a difference in the psychology of the two groups of offenders? Other factors also raise further question as to whether these two events are similar. Banay (1959) and Hartley (1959)- reported evidence of perceptual distortion of reality in the case of homicide. Nowhere in the literature reviewed is this concept found to be relevant to assault. Frustration is a common factor in both events, but nowhere has there been a comparison of the two groups in terms of the degree of frus- tration, along the lines set forth by Palmer (1960). In comparison of two case histories (Greene, 1988, and Schultz, l960--found below in the section on familial assault) there is a suggestion of a difference in the degree of frus- tration and the source of that frustration. In Greene's re- port of a homicide one gets the impression that the subject was severely deprived and frustrated, with little compensa- tion; while in the case of assault cited by Schultz, there is evidence of at least some gratification of needs. A compari- son of this type is questionable for a host of reasons-~among 32 them are the degree of inclusion of relevant data in the re- ports, the nature of the data reported which is in part de- termined by the author's point of view, and the like. But this comparison is not intended aS a conclusion, rather as a bit of speculation suggesting that there is, quite possibly, some psychological difference between the two subjects, which was reflected in the acts that they committed. Even though the two studies directly comparing the two phenomena, murder and assault, Specifically conclude that they are similar events as far as demographic variables are concerned, a closer examination of the findings suggests that there are some psychological differences. There is, however, an even more compelling reason for this writer to be interested in a comparison between the mur- derer and the assaulter: his work at the Ionia State Hospi- tal, where mentally ill offenders are treated. There was a striking difference between the author's psychotherapy group that consisted, by virtue of the members' request, of all murderers, and another group that consisted of all assaulters. This latter group was not as homogeneous as the former since the assaults included assaults with sexual motivation as well as those without. It is the differences observed, both be- tween the two groups per se and the individuals who comprised these groups, that stimulated this piece of research. The overall impression of the differences between these two groups is that, unlike the average assaulter, the mur- derers were generally soft spoken, reserved, rather passive, 33 and very well controlled with regard to expression of emo— tions--especially the more negative emotions, primarily an- ger. For example, in therapy one young man who had killed his father related an incident which occurred while he was working on his hOSpital job: He was in the middle of a fairly involved task when another patient came in and inter- rupted him for what he felt was irrelevant horseplay. He continued to report, in a very flat voice, how this had "an- gered" him. There was no evidence of anger in his voice when he described the incident; even more interesting, the other patient, who was also a member of this therapy group, imme- diately replied that he was not able to detect any feeling of anger during the incident. On the contrary, the second patient saw the response of the first as simply a continua- tion of the horseplay. When the therapist checked with other persons who were present at the time of the incident, it was learned that the report of the second patient was, in fact, correct; the first patient had not displayed any signs of anger or annoyance. A contrasting example can be taken from the interaction between two members of the assaultive group. For a number of weeks one young man had been relating the incidents of his life to the group, in a circumstantial and guarded manner. Another patient, recognizing the first patient's resistance, kept probing in areas the first patient was trying to avoid. This probing was being done in a friendly, concerned manner with no evident hostile intent. Yet, the first patient found 3“ this annoying and, with no Specific provocation, exploded into a tirade of angry name-calling directed to both the patient doing the probing and the group in general. This patient's anger was unmistakable, since its eXpreSSion al- most resulted in a physical assault on the second patient by the first. There were other interesting differences between the two groups. First, although the two groups were approxi- mately equally intelligent, the murderers were much less verbal; they tended to be less open, less out-going, and-- interestingly enough--less alienated from their society than the assaulters. Two examples are relevant here. During one session the group of assaulters was, as it turned out later, testing the therapist in terms of what he was willing to ac- cept from the group. To do this they verbally assaulted the therapist for more than an hour. This interaction in the group was so loud and heated that the entire session was audible on the adjoining ward where another therapist had a group of his own. The other group was unable to function that day because the interaction was so disruptive, not only because of the noise (volume) but also because of the intensity of the emotional expression (anger). This sort of thing never occurred in the group of mur- derers. On the contrary, on one occasion one patient had be- come so angered by the therapist, as he reported later, that he had felt like "jumping on" the therapist as he left the therapy room. Yet there were no signs of this feeling. As 35 a matter of fact, they had walked back toward the therapist's office together through the tunnels that connect all buildings at the hOSpital, where there would have been ample Opportunity for the patient to "jump on" the therapist had he felt free to express his feelings of anger. Not only did he not attack the therapist, but he did not mention his feelings nor were there any visible signs of anger. It was not until several months later when this patient had again become angry in the group that he felt free enough to express his feelings on the first occasion to the therapist, and then in only a joking fashion. These observed differences can be demonstrated more clearly by comparing two individual cases from these therapy groups which are sociologically quite similar to each other and, in large part, to the findings of Pittman and Handy (196”). Both patients came from an upper-lower-class back- ground where financial conditions were meager, where housing was modest, and where education was usually terminated before high school was completed because the child was recruited to help support the family. Both were raised by their maternal grandmothers. In both cases the mother was sexually promis- cuous; the mother of the assaulter was a prostitute, while the mother of the murderer was "a loose woman." In both cases the grandfather on more than one occasion had assaulted the grandmother. The similarities extended even to the rela- tionship with siblings; both knew that they had them, but neither had had any real contact with them. 36 DeSpite the striking similarity between the backgrounds of these two individuals, there were some marked differences in their psychological makeup. The murderer had always been a quiet child and considered by all who knew him as a "good boy." The assaulter, on the other hand, had constantly gotten into trouble. Throughout his life, the murderer would try to avoid trouble, especially physical violence. The murderer had not been arrested before, while the assaulter had been ar- rested on numerous occasions for both violent and non-violent . offenses. When the murderer found himself in a situation that he could not deal with, he would either attempt to mud- dle through or, failing that, run away rather than attempt to cope with it in an appropriate fashion, while the assaulter would plunge headlong into it, usually with both fists swinging. There are other characteristics which distinguish between these two individuals. The murderer was more in tune with his society, while the assaulter was quite alienated from it; the murderer was more reSponsible socially than the assaulter and was better able to tolerate frustrating and ambiguous situations than the assaulter. Although both individuals had a poor self-image, the murderer's self-concept was somewhat better than that of the assaulter. (In these two cases the difference was slight, but when the remaining members of the two groups are taken into account, this difference can be more clearly drawn, with murderers by-and-large having a bet- ter self-concept than assaulters.) The murderer had greater ego-strength than the assaulter 37 in that he was more stable throughout his life; yet this stability seemed in part due to the greater degree of rigid- ity of his personality, while the assaulter seemed much less rigid. In their interpersonal relations, the murderer was much more submissive, the assaulter more assertive and ag- gressive, when interacting with both peers and authority figures. Similarly, the murderer was quite introverted, while the assaulter was more extroverted; the murderer was less likely to perceive or attend to aggressive cues in his environment, the assaulter was almost eager to do so. Fi- nally, the murderer displayed a great deal more control over all his impulses and emotions, especially anger, than did the assaulter. Both individuals in these cases were approximately the same age at the time they were in the hOSpital, late twenties or early thirties, and they had identical diagnoses: schizo- phrenia, parnoid type. Their respective groups were also very similar. They were both made up of patients who were young and who were diagnosed as psychotic, most of them schi- zophrenic, paranoid type or chronic undifferentiated type. Because of these similarities--the clinical diagnosis, the age, and the background of the two individual cases, and the similarities between the two groups-~the differences reported would appear to be more a matter of psychological differences than any artifact resulting from sampling, despite the fact that these patients were not selected randomly. These two cases and the general characteristics of the 38 groups from which they were drawn have been reported to demonstrate the possibility of psychological differences rather than to demonstrate the accuracy of the thesis being developed here. The individual cases were selected for two reasons. First, there is a great deal of similarity in back- ground in terms of both sociological factors and the demo- graphic factors associated with their crimes. Second, although they are by no means "typical" of their respective groups, they do clearly demonstrate that there are possibly more differences between the murderer and the assaulter than the literature suggests, especially when psychological vari- ables are taken into account. The above description of cases and groups is not in- tended as any empirical evidence of the differences between the two phenomena. It is intended, first of all, to explain some of the background for the present study; and, second, to further substantiate the position that there is quite pos- sibly, at least in terms of the psychology of these two events, a difference between murder and assault which is not found in the demographic studies now available. Thus, based upon the four areas reviewed in this section-- the confusion of the concepts "assaultive" versus "homicidal"; the research which combines these two criminal offenders into an extremely aggressive group; the studies that directly com- pare these two criminal events; and this author's experience working with individuals who have committed these offenses-- it would appear that the question raised by Easson and 39 Steinhilber (1961) as to why some kill and others do not might be answered, not by the introduction of chance factors (Pokorny, 1965, and Pittman and Handy, l9SU), but by looking at the psychological factors or characteristics of the indi- viduals who have engaged in these assaultive or homicidal acts. It seems that any study of these acts can best be done when the two behavioral phenomena--murder and assault--are kept separate. This is a guiding principle followed in this' research. THE LITERATURE DEALING WITH FAMILIAL HOMICIDE In a study concerned with the child who has killed one or both of his parents, Megargee (1967) again focused on the level of control as a key factor in the killing. Yet he sug- gested that there are possibly some differences in dynamics in the case of parenticide when he said, "...most children develOp some inhibitions against parenticide, if not against other forms of homicide, during the formative years. Conse- quently most matricides and patricides come after a certain amount of inner conflict and it is this conflict that makes them so interesting." However, the description of the case used to illustrate this point appears strikingly similar to the other descriptions of individuals who commit nonfamilial murder (Greene, 1998; G. B. Smith, 1958; S. Smith, 1965; Kahn, 1960; and Galvin and Macdonald, 1959). Two additional studies can shed some light on the nature of this internal conflict. Duncan and Duncan (1971), in a study of five cases in which adolescents killed one of their '40 parents, revealed an abrupt loss of control related to a change in the interpersonal relationship with the victim. This loss of control seemed to follow a sequence of events which became increasingly more unbearable and less amenable to the offender's control. They implied that the original control and the subsequent loss of control was a function of parental brutality. Sadoff's study (1971) of two cases of parricide found a cruel relationship between the victim and the offender, am- bivalent feelings on the part of the offender for the victim, and a high degree of predictable violence in the relationship. The offender appeared to be pushed to homicide through mis- treatment and violence. After the death of the victim, there is a sense of relief and a sense of freedom, rather than re- morse. According to Rasch (1967) homicide that occurs within the family, particularly the killing of one's spouse, does constitute a separate phenomenon. This he attributed to two factors: First, the motivations--here used in the psycho- logical rather than the legal sense--are quite different in the cases of familial homicide as compared with cases of non- familial homicide. Second, the patterns of the "homicide situation"--"the particular history of the lethal assault in its totality [which describes] the dynamics of the crime"-- are essentially different. Basically, Rasch took the same approach to the phenomenon of murder as Toch (1966) took to the phenomenon of assault. Rasch did not involve himself kl with the personality dynamics of the spouse killer, but he did describe four distinct patterns of the homicide situation. These include: "The killing of the lover by the abandoned partner," "The Spouse being killed by the abandoned partner," "The elemation [sic] of the partner jeopardizing marriage," and "Antisocial mateship." Cormier (1962) discussed the interpersonal dynamics of eight men who killed their wives in greater detail. He in- dicated "...we were struck by the great variety of personal- ity, of attainment, of social role and position [among these eight men. Yet] a repetitive pattern emerged." The key finding was an unusually strong, pathological tie between offender and victim, which began from their very first meeting. Cormier described this tie, this attraction, as an "interlocking" of needs; that is, "...they needed one another not only for their well being but for the reverse, because of their problems." As conflicts develOped to the point where others would separate, this interlocked couple could not; separation was not an alternative. Yet the final cri- sis--another in a long series of conflicts--leaves the offen- der in the position that he can no longer go on, but he cannot separate: "...murder is the only way to dissolve a union that other people terminate by more normal means." Cormier then traced the lives of the offenders over the next five or ten years, but this does not shed much light on the act of murder. Kurland, Mogenstern, and Sheets (1955), in a unique paper studying men who had killed their wives and women who N2 had been "homicidally assaulted by their husbands, but sur- vived," reported an interesting interactional pattern which could lead to a "homicidal potential" in the marital setting. They included such factors as: "a decompensating sado- masochistic relationship manifested by an increasing degree of acting out of hostile impulses..."; increased alcohol consumption; and increased expressions of "pathological jealousy, ideas of infidelity and persecution...." There was also evidence that the wife's increased anxiety led to increased attempts to control her husband by making him feel guilty, denial of his dependency needs, rejection of sexual advances, and repeated expression of her ambivalent feelings about their relationship. At the point when communications had totally deteriorated, the "homicidal assault" occurred. Olive (1967) investigated the factors involved in fili- cide. He compared a group of females who had killed their own children with two other groups: women who had killed their spouses or Spouse-surrogates, and women who had acted out in a non-violent anti-social manner. He found a signi- ficantly high rate of sexual motivation (related to incest) in the filicide group compared to the other groups. Although he did not focus directly on the question of familial versus nonfamilial victim, Olive's research does suggest that even within familial homicide there may be differences in dynamics. While Bach (1967) did not actually compare familial ho- micide with other forms of homicide, he implied that there is a difference between the person who kills his or her Spouse N3 and the person who kills someone other than a spouse when he described the factors leading to "intimate violence" as inter- action patterns that could only occur in the intimacy of mari- tal relationships. He includes such factors as: l. Conflict-evasion and aggression- phobia with passive-covert methods of hostility release and inability to ne- gotiate changes in the Open. In our sample, the supercoverts tended to kill the relatively more covert in a ratio of 3:1. 2. Non-transparency of expectations: deliberate miscuing as to where the part- ner stands. 3. Conceptual rigidity of and nar- rowness of thing-like fixed eXpectancies. u. Deep involvement emotionally in having the partner fulfill the expecta— tions. 5. Making up without change. Narrow bases for "making up": segmental sex is typical. 6. Symbiotic disparities in power— relationships between the controlling and the controlled, between the keeper and the kept. ... 7. Symbiotic diSparity in giving vs. taking, with the takers never getting enough and the givers worn out and un- appreciated. ... 8. Symbiotic disparity in outside social contacts, with the isolated stay- at-home getting easily angry at the freedom, fun and pOpularity of the so- cializing mate. 9. Flight blocking: "I'll never let you go" is a common enough warning - nor- mally an attempt to block impulsive exits and prevent the love-object from falling into competitive hands. ... 10. Last, but by no means least, the probability of violence increases to the degree that the murder situation [Rasch, 1967] facilitates or deters lethal hos— tilities. ... He did not, however, deal with any personality variables, Primarily because he is a transactionalist concerned with the patterns of interpersonal interaction in the here-and- nu now. Assuming that ways of interacting in interpersonal situationSéue at least correlated with personality charac- teristics, it may be speculated that if there are unique in- teraction patterns Operative in the killing Of one's spouse, then it may be possible that these patterns are correlated with unique personality characteristics. Another pattern of familial homicide was described in an article by Von Hentig (1957). He called this pattern "Premurderous kindness and post murder grief," and he indi— cated that it is a deliberate premeditated form of murder. Von Hentig described in some detail the measures that the killer takes to catch the victim unawares by disarming him "psychologically," which is more easily done in the marital situation than in any other relationship situation. The same pattern Of psychological disarmament is employed by the kil- ler after the crime, in order to defend himself from prose- cution: I could not have done it, look how good I have been to my spouse. The overall pattern goes something like this: The couple has a disagreement or a series of disagreements, after which one member, the future killer, makes up to the other, the future victim. Time is allowed to pass, during which the future offender is a good and loving spouse (a wolf in sheep's clothing) so as to give the impression to the as- sociates of the future victim that all is again all right be- tween the partners. After they are convinced that all is all right, the killer strikes and then, after the spouse is dead, puts on a display of intense grief over the death of the spouse. '45 THE LITERATURE DEALING WITH FAMILIAL ASSAULT A study by Schultz (1960) concerned with the wife as- saulter is based on only four subjects, one of whom is actu- ally not an assaulter, but a murderer. DeSpite these short- comings, it does provide a great deal of information. Schultz reported that in each case the assault was preceded by a per- ceived threat of dissolving the marriage. None of the sub- jects had a previous criminal record except for other assaults on their wives, nor had they used alcoholic intoxicants or drugs prior to the offense. From the description of the "typical case," the following information is obtained. In the subject's childhood, family living conditions were substandard, with chronic shortages of clothing and money. Having had to work long, hard hours, the parents would come home exhausted, with little time, energy, or patience for the subject's emotional needs or even his supervision. His parents showed concern only when he had done something wrong. This "concern" took the form of swift, harsh punishment. As brothers and sisters were added to the family, the subject was placed in the position of baby-sitter, meeting the needs of the other children while his own needs remained ungratified, a situation which was never eXpressed or discussed. Early in life it became clear to the subject that sub- mission to maternal authority and demands would bring some measure of affection, but equally early he found that any display of hostility brought on rejection and severe punish- ment. Therefore, he learned that control and inhibition of N6 hostility directed toward his mother must be maintained at all costs. Despite these conditions, and in large part due to them, the subject described his mother as a good, strict woman whom he loved dearly. When he was asked about the harsh punishment, he replied that he was an "evil child" and deserved everything that he received in the line of punish— ment. Yet he was able to eXpress the feeling that his mother was using him for her own gains rather than being a warm, loving figure. This "typical" subject married his first wife when he was still a teenager. This relationship was terminated when his wife's lover (a situation known to the subject) took his wife, his belongings, and his children and left. Eight years later he was again married, this time to a "hot-tempered" wo- man who was married twice before and who had killed her second husband. This marriage was characterized by constant verbal battles until the wife decided to visit friends, including an ex-sweetheart. This threat of losing his wife precipitated his earlier physical assaults on her, which she returned in kind. This Situation continued until the final altercation, which was again caused by the presence of a lover who threat- ened to take the wife from him. Focusing on family dynamics, Schultz reported that all the wives in this study were characterized as "masculine, out-Spoken, domineering women who had much in common with their husbands' mothers. The wives, in general, tended to exploit and profit from their husbands' passiveness and N7 dependency." Clearly, then, the description of the wife assaulter and his familial relationships is similar to the description of the murderer reported by Greene. There are striking dif- ferences between Schultz's description and that presented by Easson and Steinhilber (1961) dealing with nonfamilial as- sault, where not only are aggressive responses not prohi- bited, but encouraged. Because of these differences, it is felt that these two types of offenses, familial and non— familial assaults, should be kept separate. Although the literature dealing with homicide and as- sault having a member of the family as victim does not make any direct comparisons between killers of family members and non-members, there is enough information to suggest that it might be methodologically sound to separate the groups of subjects along this dimension, but not enough to develop specific hypotheses. Beyond the information in this section suggesting this procedure, earlier comments concerning keeping criminal classes separate seem to be applicable here. Therefore, this investigation will divide subjects along the cflimension of relationship with the victim: Was the victim a Inember of the family or was he not a member of the family? SUMMARY From the above review of the literature, it should be clear that there are two relevant dichotomous dimensions wflmich constitute the independent variables in this study: us Murder versus Assault and Familial versus Nonfamilial Victim. STATEMENT OF THE PROBLEM This study has two purposes. Its primary purpose is to explore systematically the possibility that murder and as- sault are psychologically different events, although reported as being demographically similar, since the intra- and inter- personal characteristics of the individuals who commit these offenses appear to be dissimilar. Seven intrapersonal char- acteristics--impulse control, tolerance of ambiguity, aliena- tion, rigidity, social responsibility, self-concept, and ego- strength--and two interpersonal characteristics--ascendance- submission and introversion-extroversion--have been selected as the dependent variables in this study. The secondary purpose of this investigation is to ex- plore the effects of the relationship of the victim to the offender. Recent investigations have suggested that the selection of the offender's victim in terms of the relation- ship to the offender (Familial or Nonfamilial) is related to the personality characteristics of the offender. In brief, this study is an attempt to determine whether or not there is a difference between murderers and assaulters along nine dependent variable dimensions. It is also an at- tempt to eXplore the effects of the relationship of the vic- tim to the offender (Familial or Nonfamilial) within the primary groups of murder and assault. N9 SPECIFIC HYPOTHESES Because of the exploratory nature of this investigation, not all of the questions raised are stated in the form of Specific hypotheses. Hypotheses are stated for the compari- son between the two dimensions of the major independent vari- able--murder vs. assault-~but not for the secondary indepen- dent variable--familial vs. nonfamilial victims. Therefore, this section is divided into two parts: Hypotheses and Related Questions. HYPOTHESES 1. If murderers are compared with assaulters regarding the degree to which they control their impulses, then mur- derers will show greater impulse control than assaulters. More specifically: (a) Murderers will have fewer prior arrests than assaulters; (b) Murderers will have a higher Overcontrol index on Megargee's (1966) Overcontrolled-Under- controlled Adjective Check List than assaulters; (c) Murderers will have a higher Self-Control score on Gough's (1956) California Psychological Inventory than assaulters; (d) Murderers will have a higher score on the 50 51 0-H Scale on the MMPI (Megargee et al., 1967) than assaulters. 2. If murderers are compared with assaulters with re- gard to their ability to tolerate ambiguity, then murderers will be better able to tolerate ambiguity, that is, will score lower on Budner's (1962) Intolerance of Ambiguity Scale, than assaulters. 3. If murderers are compared with assaulters with re- Spect to the degree to which they are generally alienated from society, then murderers will be less alienated, that is, will score lower on Dean's (1961) Alienation Scale, than as- saulters. u. If murderers are compared with assaulters regarding their rigidity in thinking and social behavior, then mur- derers will be more rigid, that is, will score lower on the Flexibility Scale of Gough's (1956) California Psychological Inventory, than assaulters. 5. If murderers are compared with assaulters with re— spect to the degree of social responsibility, then murderers will show more social reSponsibility, that is, will score higher on the Responsibility Scale of Gough's (1956) Cali- fornia Psychological Inventory, than assaulters. 6. If murderers are compared with assaulters in terms of their self-concept, then murderers will possess a better self-concept, that is, will score higher on Fitts' (1965) Tennessee Self Concept Scale, than assaulters. 7. If murderers are compared with assaulters regarding 52 ego-strength, then murderers will show greater ego-strength, that is, will score higher on Barron's (1953) MMPI Ego- Strength Scale, than assaulters. 8. If murderers are compared with assaulters with re- spect to the dimension of ascendancy and submission, then murderers will be more submissive, that is, will score lower on the Allports' (1928) Ascendance-Submission Scale, than assaulters. 9. If murderers are compared with assaulters regarding the extroversion-introversion dimension, then murderers will be more introverted, that is, will score lower on the Minne- sota T-S-E Inventory (Evans and McConnell, 1957), than as- saulters. 10. If murderers are compared with assaulters with re- gard to their reaction to this survey, murderers will diSplay less overt hostility, that is, will score lower on the overt hostility measures, than assaulters. More specifically: (a) Murderers will be less critical of the survey and the examiner than assaulters, that is, murderers will score lower on the Likert-type objec- tive scale of overt hostility, than assaulters; (b) Murderers will make fewer negative remarks in the additional comments section of the overt hos- tility scale, that is, they will score lower on the supplemental overt hostility score, than assaulters. 53 RELATED QUESTIONS Because the literature regarding the victim dimension (familial vs. nonfamilial) was so limited, specific hypo- theses are not offered. Yet two questions must be asked: 1. Are there any significant differences on the depen- dent variables when individuals who attacked family members (familial group) are compared with those who attacked family non-members (nonfamilial group)? 2. Are there any significant interaction effects on the dependent variables between the independent variable di- mensions of crime by victim? METHODS AND PROCEDURES DEFINITIONS OF THE INDEPENDENT VARIABLES Before describing the procedures employed in this study, it is important to define the dimensions of the independent variables. Murder Generally, murder is defined as an attack on one indi- vidual by another which results in the death of the victim. Specifically, for the purposes of this study, murder was operationally defined by the charges brought against the in- dividual by the State of Michigan and his admission to a state hospital, in this case Ionia State Hospital, as either incompetent to stand trial (P.A. 1966: 266, 27a) or not guilty by reason of insanity (P.A. 1966: 266, 27b). Only first and second degree murder were included. Manslaughter or negligent homicide were excluded. Also, within the cate- gories of first and second degree murder, certain specific situations were excluded: the charge of murder resulting from the commission of another felony; the charge of murder when the killer was hired and paid for the commission of that murder; and the charge of murder involving a sexual act, e.g., rape-murder. 5N 55 Assault Generally, assault is defined as an attack on one indi- vidual by another which results in the physical injury, but not the death, of the victim. Specifically, assault was operationally defined by the charges brought against the in- dividual by the State of Michigan and his admission to a state hospital, in this case Ionia State Hospital, as either incompetent to stand trial (P.A. 1966: 266, 27a) or not guilty by reason of insanity (P.A. 1966: 266, 27b). This in- cluded felonious assault, aggravated assault, assault with intent to do great bodily harm less than murder, assault with a deadly weapon, and simple assault. Excluded from this study were all forms of assault which have sexual intent as- sociated with that assault, such as assault with the intent to rape, rape, etc. Also excluded were crimes in which the intent of the offender was not clear, as in the case of as- sault with intent to commit murder and attempted murder. When records specifically indicated that there had been no injury to the victim the case was excluded. When the matter of injury, or degree of injury, was not mentioned in the re- cord, injury was assumed and the case was included in the study. These definitions of the independent variables of mur- der and assault in terms of charges brought against the individual by the State refer only to those charges that led to the current incarceration of that individual, without consideration Of any previous charge(s) and/or conviction(s) 56 which may have appeared on that individual's record. Familial Victim The victim was considered as a member of the family if he or she was directly related to the offender by birth or marriage. This included mother, father, sister, brother, son, daughter, husband, wife, common-law wife, common-law husband, mother-in-law, father-in-law, Sister-in-law, and brother-in-law. Grandmother, grandfather, second cousin, etc., were excluded unless that person acted as a parent sur- rogate, such as a grandmother who raised the offender most of his life. Also included were step-relations, adoptive parents, even a neighbor who was "like a father to" the Of- fender. This dimension is concerned with the functional relative as well as the natural relative. Where the victim was the girl friend or boy friend of the offender, the case was excluded because the nature of the relationship was difficult to determine. That is, it would be difficult to say if it were a common-law relation- ship (Familial) or a friend relationship (Nonfamilial), ex- cept where the relationship was clearly Specified in the records. Nonfamilial Victim The victim was considered as not a member of the family if the records indicated that he was not immediately related to the Offender by either birth or marriage or as a functional relative. This included stranger, friend, neighbor, shop- keeper (when the act was committed independently of another 57 felony), etc. In the original research design there was to be an ad- ditional dichotomous dimension: the mentally ill offender and the non-mentally ill offender. Due to difficulties in obtaining subjects, this dimension was eliminated from the study. APPARATUS - A DESCRIPTION OF THE DEPENDENT VARIABLES Demographic Information Form (See Appendix B) Demographic information was collected primarily to as- sign cases to the pools of potential subjects within each group or stratum. These forms were also used to collect in- formation related to the criteria for inclusion-exclusion from the pools of potential subjects. This demographic information was also collected to aid in finding additional relevant dimensions of the independent variables, and so that these subjects can be compared with the subjects in and the findings of other studies. Further, this type of information--Specifically, prior arrest records-- as suggested by Megargee (1966), was used as an index of the degree of control. This will be discussed further in the following sub-section. Impulse Control The impulse control measures were the primary dependent variable in this study. The measurement of impulse control closely followed the techniques used by Megargee and his as- sociates (Megargee, 1968a, 1968b, 1965a, 1966, and Megargee, Cook, and Mendelsohn, 1967). They developed the conceptual 58 dimension of overcontrolled-undercontrolled hostility. Fol- lowing their lead, impulse control was measured in several different ways: the number of prior arrests, the overcon- trolled index based on a modified adjective check list, the Self-Control Scale of the California Psychological Inventory (CPI), and the Overcontrolled Hostility (O-H) Scale for the MMPI. According to Megargee (1966), the greater the number of prior arrests, the greater the degree of undercontrolled hos- tility. Conversely, the fewer the prior arrests, the greater the degree of overcontrolled hostility. A question must be raised regarding the latter portion of this relationship-- the fewer the prior arrests, the greater the degree of over- controlled hostility--in that it may be ambiguous, since it is not clear if the fewer prior arrests indicate overcon- trolled hostility or normal behavior. (This was tested by finding the degree of cOrrelation between the number of prior arrests and the scores on the other impulse control measures. In fact, an intercorrelation matrix was generated for all four measures of impulse control; see Results, below.) In his monograph, Megargee (1966) reported a modifica- tion of Gough's Adjective Check List which was specifically designed to measure the degree of both undercontrolled and overcontrolled hostility. In Megargee's study this check list was administered to the counselors of the subjects in a.detention home. In this study, however, the adjective check list was used as a self-report measure; that is, the 59 subjects themselves checked the adjectives which they felt applied to themselves. To avoid response set, the adjec- tives were presented to the subjects in a random order. Ac- cording to Megargee, this scale is scored by "counting the number of adjectives of each type. Then an Overcontrolled index was created by subtracting the [number of] Undercon- trolled Aggressive adjectives [checked] from the [number of] Chronically Overcontrolled ones." Thus, a high positive score (high in terms of the absolute numerical value of the score) indicates undercontrolled aggression. Megargee (1966) reported no evidence of inter-rater re- liability, and no studies were found in which this specific adjective check list was used. Validity of the measure is primarily of the construct type. As Megargee stated, "Twenty of [the selected forty adjectives] were adjectives which seemed descriptive of the Chronically Overcontrolled person such as 'meek,' 'self-controlled,'.... Twenty of the others seemed descriptive of the Undercontrolled Aggressive type." Like the Adjective Check List, the Self-Control Scale of the CPI is an index which measures both overcontrolled and undercontrolled hostility, with high score indicating overcontrolled hostility. In the CPI Manual, Gough (196“) reported test-retest reliabilities of three independent samples of .68, .75, and .86. He also reported four validity studies: (a) a correlation of -.25 with raters' scores of "impulsivity" in medical students, (b) a correlation of -.23 with ratings of "impulsivity" and +.21 with Q-sort ratings of 60 "over-controls his impulses" in a group of military officers, (c) a correlation of +.3u with Q-sort ratings of "patient and self-controlled..." in college students, and (d) a clear and significant differentiation of the "most" and "least" impul- sive high school students. Thus the validity of the scale can be claimed. This scale was administered as part of the total CPI. The method of administration followed the stan- dard procedures for administration of the total inventory. Megargee and his associates were not totally satisfied with these methods of differentiating between overcontrolled and undercontrolled aggression, because they felt the scales were somewhat weak for the detection of overcontrolled hos- tility. Therefore they develOped an MMPI scale specifically for the detection of the overcontrolled individual (Megargee, Cook, and Mendelsohn, 1967). They made special note of the fact that this scale was not designed for the measurement of general aggressiveness, and so they labelled it the Overcon- trolled Hostility (O-H) Scale. This development was an im- portant contribution to the study of the degree of impulse control, since earlier scales emphasized the detection of the undercontrolled individual. Megargee (1967) reported a Kuder-Richardson reliability score of .56, which is consistent with the "median split- half reliability coefficient of .58 reported for the nine basic MMPI scales." In the same paper Megargee reported on the validity of the O-H measure. He reasoned that if this is in fact a mea- sure of overcontrolled hostility, "...then within the normal 61 ranges of scores (T less than 70) it should relate to other measures of less pathological control while at the upper end it should be able to discriminate between overcontrolled and undercontrolled assaultive criminals." His findings sup- ported this reasoning in that there was a high positive (sig- nificant) correlation between 0-H and "...L, K, Hy, Hy-S (inhibition of aggression), and E0 (Ego overcontrol), and a significant negative correlation between 0-H and the Byrne Repression-Sensitization (R-S) scale...." With regard to the scale's ability to discriminate between under- and over- controlled subjects, it did so, in three separate studies and at statistically significant levels. In a later study Megargee (1968) again demonstrated that the 0-H Scale does discriminate between overcontrolled sub- jects (a group of conscientious objectors) and two groups of "normal young adult males." Two other authors, Blackburn (1968) and Fisher (1970), replicated Megargee's validation study (1967), and both confirmed his findings. The O-H Scale was administered as part of the standard MMPI, not as a sep- arate scale, Since the items were taken from the standard MMPI. The scale was scored, however, with a separate scoring key. A high score on the 0-H Scale indicates overcontrolled aggression. Tolerance of Ambiguity Although it was h0ped that a standardized scale of frus- tration tolerance could be found, no such measure appeared to be available. Therefore, the investigation of the role of 62 frustration tolerance was deferred to future research. Since the current project consisted of administering questionnaires and standardized scales to groups of subjects, it was neces- sary to find a scale which would approximate the frustration tolerance dimension and still be appropriate to the project. Budner's (1962) scale of intolerance of ambiguity seemed most appropriate as a substitute. In his study, Budner defined intolerance of ambiguity as "the tendency to perceive (i.e., interpret) ambiguous situations as desirable." He defined an ambiguous situation as "...one which cannot be adequately structured or cate- gorized by the individual because of the lack of sufficient cues." It is possible to identify three such situations: "...Situations characterized by novelty, complexity, or in- solubility." From his definition of ambiguity and his defi- nition of intolerance of that ambiguity, it seems clear that the latter concept is not far removed from the concept of in- tolerance of frustration, in that intolerance of ambiguity seems to be a Specific form of intolerance of frustration. It must be pointed out, however, that it would be unwise to generalize from the findings concerning intolerance of ambi- guity to the more general intolerance of frustration. Budner reported that his scale for the measurement of tolerance-intolerance of ambiguity shows up favorably in terms of both its reliability and its validity. More speci- fically, he reported its reliability as .89, using Cronbach's alpha formula. He attributed this low figure to (a) the use 63 of the alpha formula rather than a split-half formula, and (b) the freedom of this scale from acquiescence and social desirability factors. In the same article, Budner reported (a later reliability study using the Split-half approach and finding a reliability score of .85. Budner also found Significant correlations between his scale and three other scales (Coulter, Walk, and Prinston's scales of tolerance-intolerance of ambiguity), suggesting the validity of the Budner scale. Further, when he compared the scores from his scale with four independent professional ratings of individuals' biographies, he found significant correlations with the ratings of three of the four judges. When scale scores were correlated with composite judgment scores, "a correlation of .88 was obtained (p<.05)." A cor- relation of .39 (p<.05) was obtained between peer ratings and the scale. All results support the validity of the scale. Budner's scale is a seven-point, sixteen-item Likert- type scale. For this scale a high score indicates intolerance of ambiguity. Rigidity Rigidity was measured by the Flexibility Scale of the Cali- fornia Psychological Inventory. According to the CPI Manual (Gough, 1968), the Flexibility Scale was "created by the tech- nique of internal consistency analysis." Gough acknowledged that this method contains the danger that "the examiner might very well be wrong in deciding how an item ought to be scored, or even in deciding that that item belongs in the preliminary 6“ scale." DeSpite the limitations, three separate samples had a test-retest reliability of .65, .59, and .73. Gough also reported significant negative correlations in three independent studies between the Flexibility Scale and (a) staff's rating of graduate students regarding "rigidity" (-.88), (b) ratings of medical students' rigidity (-.36), and (c) college student F-scale scores (-.58). Thus the validity of the Flexibility Scale can be claimed. The Flexibility Scale was not administered separately, but as part of the standard CPI. On this scale a low score indicates rigidity, while a high score indicates flexibility. Social Responsibility The Responsibility Scale (Re) of the California Psycho- logical Inventory was used as the measure of social reSponsi- bility. This scale was an outgrowth of Gough, McClosky, and Meehl's (1952) development of "A personality scale for social reSponsibility," later refined and revised and then incorpo- rated into the CPI (Gough, 1968). In the CPI Manual, Gough (1968) indicated that the Re- sponsibility Scale is an "empirically" developed scale and reported test-retest reliabilities of .73, .65, and .85 for three independent samples. He also reported four validity studies, in which the Re Scale showed: (a) a significant correlation (+.38) with ratings of "positive character inte- gration" in a group of graduate students; (b) a correlation of +.38 with ratings of "responsibility" of medical students; (c) clear and significant differentiation of high school 65 students who were rated "most" and "least" reSponsible; and (d) clear and significant differentiation of groups rated as "best citizens" and "discipline problems" among high school students. This evidence supports the validity of the scale. Vingol (1968) was not able to confirm the validity studies of Gough. One criticism made was that the technique of extreme groups used by Gough omits a large portion of a "population," and thus the correlations are "grossly inflated." Using the total range of self-rating and peer-rating scores of responsibility (no validity or inter-rater reliability measures were offered), Vingol found negligible correlation between the ratings and the Re scores. He also used Eysenck's personality inventory and "derived measures of self-acceptance" as a basis for his comparisons and found no significant corre- lations. He thus concluded: "No support was found for the validity of the Responsibility Scale." Because no evidence of reliability and/or validity was Offered by Vingol regarding the self-ratings, peer-ratings, and "derived measures of self-acceptance," his strong con- clusion must be carefully interpreted. The present study does raise some question regarding the validity of the Re Scale, but the evidence does not appear to be strong enough to invalidate the scale. This scale was not administered separately, but as part of the CPI. A high score on this scale indicates social re- sponsibility. 66 Alienation Alienation was measured by Dean's (1961) Alienation Scale. Although Dean has Specifically develOped three di- mensions of alienation--powerlessness, normlessness, and social isolation--this scale was used as a Single scale, without regard to these three dimensions. Dean reported a split-half reliability coefficient for the powerlessness subscale of .78; for the normlessness sub- scale, .73; and for the social isolation subscale, .88. The split-half reliability for the total scale was .78. Inter- correlation of the three subscales suggested that they "... belong to the same general concept." Thus the use of this Scale as a single scale seems legitimate. Alienation correlated significantly with the F-scale (.26, p<.05). In a further attempt to validate the scale, Dean predicted a negative correlation between social status determined by occupation (r=—.19, p<.01), education (r=-.21, p<.01), and income (r=-.23, p<.01), and a positive correlation with age (r=.l2, p<.01). Thus he felt he had demonstrated the validity of his scale. His validity evidence does not seem to be the strongest, but seems sufficient for the in- clusion of the scale in this study. The higher the score on this scale, the greater the alienation. Self-Concept Although there were several scales for the measurement of self-concept available, Fitts' (1965) Tennessee Self-Concept 67 Scale (Research Form) was the least complex from the point of view of the subject, and therefore was selected. (Further- more it provides considerable additional information that will be used in later research, but will not be dealt with in this study.) Greenberg and Frank (1965) found significant reSponse set on the subscales of the Self-Concept Scale. They sug- gested that the items be presented in random order rather than as originally indicated by Fitts. This investigation is not concerned with the various subscales, and therefore this issue is minimal. Ideally, this precaution should have been taken in any event, but permission to reproduce the scale was not granted (it had to be purchased and used as published), and this randomization could not be performed. In the Tennessee Self-Concept Scale Manual, Fitts (1965) reported a test-retest reliability of .92 for the total score. Unfortunately his measure of reliability was based on only one sample of 60 subjects. Nonetheless, this figure is exceptionally high, which counteracts to some degree the use of a small, single sample. Fitts Spent much more time researching the validity of the scale than the reliability. Briefly, he reported that the scale: (a) clearly and significantly (p<.001) differen- tiates between psychiatric patients and non-patients, (b) clearly differentiates between "people characterized as high in personality integration" and the norm group, (O) was able to differentiate between job applicants, paranoid 68 schizophrenics, depressive reactions, and emotionally un- stable personalities, (d) differentiated between delinquents and non-delinquents, (e) correlated significantly with many other measures (too numerous to detail here), and (f) mea- sured personality changes under specified conditions, e.g., psychotherapy, stress and failure, and tranquilizing medi- cation. The validity research is far more impressive than the reliability research and certainly suggests support of the validity of this measure. Havener and Izard (1962) pro- vided some support of the validity of the Fitts scale (an earlier form thereof) when they were mildly successful in differentiating paranoid schizOphrenics from non-paranoid schizophrenics (p<.08) using the Self-Concept Scale. The higher the score on this scale, the better the self- concept. Ego-Strength Barron (1953) develOped a scale for the measurement of ego-strength using standard MMPI items, for which he reported an odd-even reliability of .76 and a test-retest reliability of .72. He also reported that the Ego-Strength Scale suc- cessfully differentiated between "improved" and "unimproved" patients as rated by their psychotherapists (p<.01). When high and low ego-strength subjects were rated by professional raters in an adjective check list, with ego-strength level unknown to raters, they were able to successfully differen- tiate the two groups (p<.05). Like Fitts, Barron correlated his scale with many other 69 scales (again, too many to enumerate here) and found many significant correlations in appropriate directions. He pre- sented sufficient evidence to support the validity of the scale. Korman (1960) provided additional evidence of the vali- dity of Barron's Ego-Strength Scale, in reporting that high ego-strength subjects were able to resolve a discrimination conflict situation more rapidly that low ego-strength sub- jects, and the group differences increased as the discrimi- nation problem became more complex. Himelstein (1968) also supported the validity of the Ego-Strength Scale. He found that (a) the scale clearly differentiates between therapy and non-therapy groups (p<.01) and (b) that self-referred counselees scored higher on the Ego-Strength Scale than did the referred counselees. The Ego-Strength Scale was administered as part of the standard MMPI, but scored separately. The higher the score on this scale, the greater the ego-strength. Ascendance-Submission Gordon and Floyd Allport (1928) develOped a scale to measure ways peOple characteristically interact with one an- other, in terms of ascendance and submission. Their scale was used to measure this one aspect of interpersonal inter- action. The Allports reported a split-half reliability of .85. Repeated reliability measures fluctuate around .79. (Regarding validity, they reported positive correlations between this 70 scale and ratings of ascendance and submission ranging from .29 to .79. Warren (1965) complained that no attempts have been made to up-date and refine this scale, nor have there been attempts to further validate it. He reported that the scale does successfully differentiate between different occupations and does not seem to be affected by I.Q. He also reported reliability scores ranging between .85 and .96, but did not specify the form of reliability reported. Although weak, there is some evidence suggesting the validity of the scale. A high score on this scale indicates ascendance. Introversion-Extroversion Another aspect of interpersonal interaction was investi- gated by using Evans and McConnell's (1957) Minnesota T-S-E Inventory of introversion and extroversion. Although this scale focuses on three dimensions of introversion-extroversion (thinking, social, and emotional), the scale was used only as an overall measure of this variable. Evans and McConnell reported split-half reliability co- efficients of .91 (Thinking), .88 (Social), and .75 (Emo- tional). Test-retest reliabilities were also reported: .89 (Thinking), .88 (Social), and .88 (Emotional). They tested the validity of the scale in a number of dif- ferent studies. The T-S-E differentiated 13 different groups of college majors which were classified as to introversion- extroversion in the predicted direction (p<.01). Home eco- nomics majors selecting merchandizing as their vocational choice were shown to be more socially extroverted than those 71 intending to teach, a finding that correlated with the per- suasive scale of the Kuder; further comparison with the Kuder revealed thinking introversion was related to literary in- terests and thinking extroversion correlated with social service interests (p<.05 for both). Social extroverts and introverts were differentiated by the persuasive score on the Kuder, but not the social service score (p<.01). Numerous other studies showed the relationship of the T-S-E and ar- tistic interest, scholastic aptitude, and academic success, as well as differentiating known groups such as conscientious objectors, life insurance salesmen, teachers, etc. Evans and McConnell presented considerable evidence supporting the vali- dity of the T-S-E scale. A high score on this scale represents extroversion. The MMPI and CPI The MMPI and the CPI were administered in their entirety, in the standard manner. The relevant scales have been dis- cussed separately, and the remainder will provide additional data for future analysis. The Overt Hostility Measures In order to get some feedback from the subjects regarding their reactions to this survey, a twenty-item questionnaire with additional space for voluntary comments was developed. (See Appendix C.) The main concern of this scale, which fol- lows a five-point Likert format, was to tap the degree of hos- tility generated by the survey. This scale was administered last and was presented in terms of feedback and evaluation of the survey, rather than as a hostility measure. For 72 further details see Appendix C. PROCEDURE The prOcedure for selecting subjects will be discussed below. Once the subjects were selected, the scales were ad— ministered to as large a group as was available at any given time. Because of the number of scales to be administered, the administration took place on two separate occasions: in the morning and afternoon of the same day for each testing session. For some subjects an extra half-day was needed be- cause of the volume of test material and their own slow reading speed. Although subjects worked at their own pace, the test booklets were divided into two sections. The first contained the MMPI, the Tolerance of Ambiguity Scale, the Adjective Check List, and the Tennessee Self-Concept Scale. The second contained the CPI, the Introversion-Extroversion Scale, the Alienation Scale, the Ascendance-Submission Scale, and the Overt Hostility Measure. By using this division, the total time of the two sessions was kept approximately equal for most subjects with average reading speed, the second session being slightly longer than the first. Ordering the scales as indicated above, generally the longer scale given first and the shorter last, was an attempt in some small measure to counteract the effects of fatigue, boredom, and frustration in the testing situation. Two variations from this general approach were necessary. First, because permission to reproduce the Tennessee Self- 73 Concept Scale was not granted, this scale was administered as the first scale in the first session. This simplified administration--the distribution and collection--of the scales. Second, because of the nature of the A-S Reaction Study, this scale was administered just prior to the Overt Hostility Measure. In order to collate the demographic information with the test data and the first session with the second, each scale was identified with a code number for each subject. A separate list of subjects' names and correSponding code num- bers was made so as to get the right forms with the right subjects. Subjects were instructed not to place identifying information anywhere on the teSt booklets, thus guaranteeing the subjects' anonymity. It was originally proposed that the lists of subjects with the subject code numbers be destroyed after each testing session. But because of the difficulty in getting subjects and the long period of time over which sub- jects were gathered, a master list was kept to insure that no subject was used twice. SUBJECTS Subject selection attempted to follow a modified form of stratified random sampling (Goodaand Hatt, 1952; Sell- tiz, Jahoda, Deutch, and Cook, 1962). That is, the hetero- geneous pOpulation of the Hospital was stratified to include only murderers or assaulters. These groups were then further stratified so as to incorporate the relation of the victim to the offender. The resulting four homogeneous groups-- homogeneous with reSpect to the two independent variables-- were then to serve as a population from which a random sample was to be selected. All patients who were in residence at Ionia State Hospi- tal as of 10-1-68 and who had been charged with either murder or assault (as defined above) were listed, their case records reviewed, and the basic demographic information recorded on the Demographic Information Form (see Appendix B). In the vselection of the subjects, particular attention was paid to the criminal act (murder or assault) and the relation of the victim to the offender (familial or nonfamilial). This resulted in a two by two classification of mentally ill offenders. From within each of these classes (or strata)-- Murder-Familial (M-F), Murder-Nonfamilial (M-NF), Assault- Familial (A-F), and Assault-Nonfamilial (A-NF)--a random sample of fifteen subjects was to be selected. Because there 78 75 were too few subjects meeting the criteria, all available subjects who were willing to participate were included. Stra- tified random sampling became a luxury that could not be af- forded. Statistical procedures were modified accordingly. This difficulty in getting subjects was in part due to the change in the nature of Ionia State Hospital from an institu- tion for mentally ill Offenders to a civil psychiatric hOSpi- tal. (For additional factors, see Appendix A.) Subjects met the following criteria: 1. The subjects were male. 2. The subjects had a full-scale or a verbal I.Q. of 70 or better. This was an attempt to insure that they could comprehend the instructions and questions presented on the various scales. If the examiner felt during the testing that a subject was not able to comprehend the material be- cause of limited intelligence, a psychotic condition, poor reading skills, etc., the subject was excluded from the study. 3. The subjects were free Of any indications of organic brain damage. This was also an attempt to insure that the subjects could comprehend the instructions and ques- tions on the various scales. More importantly, however, this criterion was included to avoid any confounding of variables. Woods (1961) reported that certain distorted EEG patterns are highly correlated with violent acting-out. Williams (1969) pointed out that abnormal EEG patterns are more frequent in prisoners who were involved in crimes of violence as compared to prisoners involved in other offenses. Gunn and Bonn (1971) 76 found no such differences, but did find differing frequencies of violent behavior between temporal lobe cases (greater fre- quency of prior convictions) and idiOpathic cases (greater frequency of violent behaviors). 8. The subjects' crimes met the criteria set forth in the definitions of the independent variables. 5. If therevere some form of ambiguity concerning the relationship between the victim and the offender, that is, if this relationship was not specified in the case record, the individual was excluded from this study. Also, if the offender was currently incarcerated for attacking more than one person, and one person was a member of his family and the other was not, the individual was excluded. 6. The subjects' alleged offenses were committed within 15 years preceding the testing of the subject. Ori- ginally this time criteria was set at five years, but this excluded too many subjects; yet the average length of in- carceration for all subjects was 3.07 years. It was hOped that each of the four groups would have 15 subjects per group, randomly selected from each of the four population pools. Alas, this was not to be. Over a four-year period, starting with all 152 cases in residence at Ionia State Hospital as of 10-1-68, then adding two subsequent searches for subjects (June, 1969, and Novem- ber, 1972), a total of 288 cases of murder and assault were reviewed. These 288 cases were classified, and the frequencies are found in Table l. 77 Table 1. Classification of All 288 Murder and Assault Cases Found at Ionia State Hospital Familial Nonfamilial Unclassifiable Assault 15 76 5 96 Murder 61 81 10 152 76 157 15 Of these 288 cases, 85 were usable as subjects. (See Ap- pendix A, Table A1 for a summary of the reasons for exclusion of cases.) These 85 subjects were classified as follows: Assault-Familial, 5; Assault-Nonfamilial, 10; Murder-Familial, 15; Murder-Nonfamilial, 15. Starting in January, 1973, the search for usable mentally ill offender subjects was expanded to all other state hospitals in the lower peninsula of Michigan. It was hoped that the four groups could be expanded to 20 subjects each. This was not possible, as only 31 murder and assault cases were found, spread over four widely separated institutions. Given the 82.7% exclusion rate of mentally ill offenders at Ionia State Hospital, the necessary effort did not appear worthwhile. At the same rate of exclusion this effort would have produced approximately 5.5 subjects spread over four hospitals from Traverse City to Northville, or 1.37 subjects per hospital. Incidentally, there was only one additional Assault-Familial case found in these 31 cases. These were the subjects most needed. (See Appendix A, Table A2.) The second Step in the expanded search was directed at the Michigan Department of Corrections. Given the findings 78 of McKie (1971) and Pouget and Renaud (1971) that there are no differences between mentally ill murderers and non- mentally ill murderers, it seemed reasonable to explore this source, provided that there were no Significant differences on the demographic and dependent variables between hospital subjects and "prison-filler" subjects within cells of the 2x2 matrix. Because assaulters were the most difficult to get, this effort concentrated on assaulters only. If enough subjects were obtained to fill the assault groups to 20 subjects each, then five additional murderers would have been added to each of those two groups. All 57 cases of men convicted of assault in residence at the Ionia Reformatory were reviewed. Of these 57 cases, six were potential subjects (meeting the prescribed criteria) for the Assault-Nonfamilial group. There were no potential Assault-Familial subjects. (See Appendix A, Table A3.) Be- cause Of this, and the marked differences between the 57 Re- formatory cases and the 288 Hospital cases, particularly with regard to age, the search was shifted to Jackson Prison, where the population was older and more nearly similar to the hOSpi- tal population. In July, 1973, there were 173 men in residence at Jackson Prison who were convicted Of assault. Of these 173 cases, 21 met the criteria for inclusion as subjects. Of these 21, 2 were actually used as "filler" subjects and were included in the Assault-Familial group. (See Appendix A, Table A8.) When 79 the two "prison-filler" subjects were compared with the five hospital Assault-Familial subjects on age at time of testing, 1.0., and all measures of the dependent variables, no statis- tically significant differences were found. The murder groups were not expanded. Thus the subjects in this study consisted of: 7 men, including the 2 "prison-fillers," in the Assault-Familial group; 10 in the Assault-Nonfamilial group; 15 in the Murder- Familial group; and 15 in the Murder-Nonfamilial group. DEMOGRAPHIC DATA The basic demographic information collected for each subject was analyzed and is summarized in the following tables (Tables 2 - 9). Table 2 shows that the four groups were essentially si- milar in age at the time of testing. There was a greater diversity of age in the murder groups (8:12.20) than in the assault groups (s=8.08, p<.05). Table 2. Age at Time of Testing (Y/s) Familial Nonfamilial sum Assault 38.1/ 7.31 32.1/ 7.63 38.5/ 8.08 Murder 38.6/12.l8 36.5/12.20 37.6/12.20 sum 38.8/10.88 38.8/10.70 Table 3 reveals that individuals who assaulted members of their own family had the highest I.Q., while those who as- saulted family non-members had the lowest I.Q. of the four groups (99.3 vs. 87.9, p< .01). Overall, murderers and 80 assaulters had similar I.Q. levels (95.7 vs. 92.6, n.s.). Those who acted out against family members and those who ac- ted out against family non-members also had similar I.Q. lev- els (97.7 vs. 91.8, n.s.). Table 3. I.Q. (Y/s) Familial Nonfamilial sum Assault 99.3/11.97 87.9/10.98 92.6/12.70 Murder 97.0/12.59 98.8/18.36 95.7/13.57 sum 97.7/12.65 91.8/13.89 Table 8. Education (R/s) Familial Nonfamilial sum Assault lO.8/2.l9 10.7/1.95 10.6/2.10 Murder 9.1/1.80 9.5/2.81 9.3/2.13 sum 9.5/2.02 10.0/2.30 Table 8 indicates that assaulters were better educated than murderers (10.6 vs. 9.3 years, p< .025). As Shown in Table 5, there was essentially no difference between those individuals who acted out against family mem- bers and those individuals who acted out against family non- members with reSpect to the length of time spent in the institution (3.8 vs. 2.8 years, n.s.). Table 5. Length of Institutionalization (X/S) Familial Nonfamilial sum Assault 3.21/2.93 2.50/1.67 2.79/2.31 Murder 3.52/2.78 2.93/2.59 3.23/2.71 sum 3.83/2.82 2.76/2.28 81 In Table 6 we see that all four groups were essentially the same age at the time the alleged offense was committed. There was a greater diversity of age at the time of the al- leged offense for murderers than assaulters (3:12.09 vs. s: 7.13, p< .05). Table 6. Age at Offense (Y/S) Familial Nonfamilial sum Assault 38.8/ 5.58 30.0/ 7.63 31.9/ 7.13 Murder 38.5/12.55 33.5/ll.58 38.0/12.09 sum 38.5/10.68 32.1/16.2l Table 7 shows that whites were more frequently involved in murder than they were in assault (73.33% vs. 81.18%, p< .025). It also shows that blacks were more frequently in- volved in acting out against family non-members than they were against family members (52.0% vs. 22.0%, p<.025). Table 7. Race in Percents by Group 8 Total for the Indepen- dent Variable Dimensions Familial Nonfamilial total B* 28.57 B 80.00 58.82 Assault W* 71.83 W 20.00 81.18 B 20.00 B 66.67 26.67 Murder W 80.00 W 33.33 73.33 B 22.73 B 52.00 total W 77.27 W 88.00 *8 represents blacks, W represents whites. groups were represented in this sample. No other racial 82 Table 8 shows that: (a) there were more Protestants in all groups taken individually, except the Assault-Familial group, and over-all they constituted 75.00% of the sample; (b) Catholics acted out against family members more frequently than family non-members (80.91% vs. 8.00%, p<.0002); but (c) within their own families, Catholics injured rather than mur- dered the victim (57.18% vs. 33.33%, p<.05); and (d) more Protestants acted out against family non-members than against family members (88.00% vs. 85.85%, p<.001). Table 8. Religion in Percents by Group 8 Totals for the Inde- pendent Variable Dimensions Familial Nonfamilial total C* 57.18 C 10.00 29.81 Assault P* 28.57 P 80.00 58.82 N* 18.29 N 10.00 11.77 C 33.33 C 6.67 20.00 Murder P 53.38 P 86.66 70.00 N 13.33 N 6.67 10.00 C 80.91 C 8.00 total P 85.85 P 88.00 N 13.68 N 8.00 *C represents Catholic, P represents Protestant, N indicates information was not available. In this sample (see Table 9) murder was committed as fre- quently by married as single men (80.00% vs. 85.67%, n.s.). Yet married men committed more murders than assaults (86.67% vs. 23.52%, p4(.05). There was, however, no significant 83 difference with regard to victim choice, familial versus nonfamilial victim (50.00% vs. 28.00%, n.s.). Table 9. Marital Status at Time of Offense in Percents by Group 8 Totals for the Independent Variable Di- mensions Familial Nonfamilial total Single 28.57 50.00 81.30 Separated 18.29 10.00 11.77 Divorced 28.57 10.00 17.53 Assault Widowed - - - Married 28.57 20.00 23.52 NOA. -' 10000 5.88 Single 33.33 86.67 80.00 Separated - - - Divorced - 20.00 10.00 Murder Widowed 6.67 - 3.33 Married 60.00* 33.33 86.67 NOAO " "' - Single 31.80 88.00 Separated 8.55 8.00 Divorced 9.10 16.00 total Widowed 8.55 - Married 50.00 28.00 NOAO " “000 *Note: 8 of the 15 men (53.33%) who murdered some family member, killed their wives. an SUMMARY OF DEMOGRAPHIC DATA The significant findings in the demographic data for the offense dimension include: 1. Assaulters were better educated than mur- derers (p< .025). 2. Assaults were committed more frequently by Single men, murder by married men (p< .05). 3. Assault was committed more frequently by blacks, murder by whites (p< .025). 8. Murderers showed a greater diversity in their age at the time of testing than assaulters (p< .05). Incidentally, this evidence contradicts both Pokorny (1965) and Pittman and Handy's (1968) conclusions that mur- der and assault are demographically similar events. The significant findings in the demographic data for the victim dimension include: 1. Blacks acted out against family non-members more frequently than against family members (p<.025). 2. Catholics acted out against family members more than did Protestants (p<.0002), but within their own families Catholics injured rather than killed their victims (p< .001). 3. Protestants acted out against family non- members more frequently than against family members (p<:.001). 8. Assault-Familial subjects had a higher I.Q. than Assault-Nonfamilial subjects (p< .01). 85 CONTROL SUBJECTS It was hOped that two groups of control subjects could be employed: (a) a "normal" control group, which proved al- most impossible to find; and (b) a non-violent prison control group, which was ruled out because of the difficulty in get- ting prisoners to participate in research. (When attempting to get the "prison-filler" subjects 81% of the potential sub- jects were unwilling or unable to participate. The total ex- clusion rate in the Jackson Prison group was 97.7%. This was due to the Operational definition of assault and the frequent plea-bargaining found. (See Appendix A, Table A8.) Thus, the control groups which were to be matched on the above- discussed variables were excluded from this study. RESULTS The data were analyzed by means of a 2x2 analysis of variance, using procedures fOr unequal cell frequencies. This provided a means of testing the Specific hypotheses and the two related queStions simultaneously. All specific com- parisons were tested by means of t-tests, F-tests, or Z-tests, with the alpha level set at .05, using two-tailed tests. IMPULSE CONTROL The first hypothesis suggested that murderers are more overcontrolled than assaulters. Impulse control was assessed by four measures. Prior arrests were divided into total prior arrests and number of prior felony arrests. In Table 10 the means and standard deviations of total prior arrests are Summarized. Because of the heterogeneity of variance, the scores were transformed using Bartlett's formula (square root of X + .5). A summary of the transformed data can be found in Table 11. The data transformation did not stabilize the variance. The analysis of variance therefore used the original scores and that summary can be found in Table 12. The original data will be reported in the summary of all analyses of variance tables, even in those cases when Bart- lett's transformations were used in an attempt to stabilize 86 87 the variance. As Edwards (1960) pointed out, "the F-test for the means in the analysis of variance is little influenced by heterogeneity of variance." Table 10. Total Prior Arrests (Y/s) Familial Nonfamilial sum Assault 2.86/3.398 3.50/ 3.500 3.28/ 3.872 Murder 3.67/6.208 7.60/18.388 5.63/1l.250 sum 3.81/5.878 5.96/ll.536 Table 11. Total Prior Arrests: Transformed Data (R/s) Familial Nonfamilial sum Assault 1.598/0.916 1.838/0.789 1.737/0.887 Murder 1.686/l.208 2.228/l.776 1.935/2.389 sum l.629/1.12l 2.069/1.876 Table 12. Analysis of Variance Summary: Total Prior Arrests Source df SS MS F p Victim 1 76.15 76.15 0.883 - Offense l 62.80 62.80 0.691 - VxO l 81.59 81.59 0.861 - within 83 3888.29 90.33 86 8068.83 The hypothesis that murderers had fewer prior arrests was not supported (5.63 vs. 3.28, n.s.). The slightly ele- vated mean for murderers is attributable to four individuals of the 30 murder subjects who had a total of 108 prior ar- rests (mean=27). Individuals who acted out against family 88 non-members had no more prior arrests than those who acted out against family members (5.96 vs. 3.81, n.s.). Again the mean for the murder group was distorted by the same four sub- jects. Because of the distortion by the four extreme cases, this hypothesis was re-tested using prior arrests for fel- onies only. A summary of the means and standard deviation can be found in Table 13. Table 13. Prior Felony Arrests (R/s) Familial Nonfamilial sum Assault 1.18/1.856 2.30/1.791 1.82/l.757 Murder 1.20/2.007 2.00/2.851 l.60/2.897 sum 1.18/1.889 2.12/2.887 Again the hypothesis was not supported (see Table 18). Assaulters had no more prior felony arrests than murderers (1.82 vs. 1.60, n.s.). Individuals who acted out against family non-members had no more prior felony arrests than in- dividuals who acted out against family members (2.12 vs. 1.18, n.s.). Table 18. Analysis of Variance Summary: Prior Felony Arrests Source df SS MS F p Victim l 10.30 10.30 1.933 - Offense l 0.58 0.58 0.101 - VxO 1 0.01 0.01 0.002 - within 83 229.36 5.33 86 280.21 89 A summary of the means and standard deviations of the second measure of impulse control, the O-H Adjective Check List, is presented in Table 15. It was found that murderers were more overcontrolled than assaulters (16.70 vs. 18.88, p<;.05). Again there was heterogeneity of variance, and again Bartlett's transformation did not stabilize the vari- ance. A summary of the transformed data can be found in Table 16. A summary of the analysis of variance for the ori- ginal data is presented in Table 17. Table 15. O-H Adjective Check List: Original Data (R/s) Familial Nonfamilial sum Assault 15.00/6.187 18.80/2.27l 18.88/8.337 Murder 16.87/8.333 19.87/3.181 16.70/8.086 sum 16.00/5.088 l7.60/3.655 Table 16. O-H Adjective Check List: Transformed Data (R/s) Familial Nonfamilial sum Assault 3.826/0.928 3.901/0.287 3.870/0.637 Murder 8.079/0.181 8.858/0.359 8.267/0.510 sum 3.999/0.713 8.233/0.827 Table 17. Analysis of Variance Summary: O-H Adjective Check List Source df SS MS F Ap Victim 1 29.96 29.96 1.711 - Offense 1 103.22 103.22 5.895 .05 VxO l 37.72 37.72 2.158 - within 83 753.06 17.51 86 923.96 90 Overcontrolled hostility, as measured by the Adjective Check List, did not differentiate the familial from the non- familial group. Within the murder group, those individuals who murdered family members were less overcontrolled than those who murdered family non-members (16.87 vs. 19.87, p<;.05). Within the group of individuals who acted out against family non-members, those who murdered their victims were more overcontrolled than those individuals who assaul- ted theirs (19.87 vs. 18.80, p< .01). The Self-Control Scale of the CPI was the third impulse control measure. The means and standard deviations are sum- marized in Table 18, the analysis of variance in Table 19. Table 18. Self-Control (R/s) Familial Nonfamilial sum Assault 88.18/13.983 88.00/18.758 85.71/18.587 Murder 50.67/11.881 89.00/11.117 89.83/11.515 sum 89.86/12.617 87.00/12.933 Table 19. Analysis of Variance Summary: Self-Control Source df SS MS F 4p Victim l 95.97 95.97 0.550 - Offense 1 188.86 188.86 1.060 - VxO l -8.86 -8.86 0.026 - within 83 7508.19 178.52 86 7780.56 This measure of impulse control revealed no significant dif- ferences either between each of the two dimensions of the two 91 independent variables or among the four groups. The final measure of impulse control was the Overcon- trolled Hostility Scale. The data are summarized in Table 20 and the summary of the analysis of variance is presented in Table 21. Table 20. Overcontrolled Hostility (R/s) Familial Nonfamilial sum Assault 59.57/ 9.280 57.80/11.078 58.29/10.813 Murder 61.07/10.579 59.80/11.951 60.83/ll.303 sum 60.59/10.l96 58.88/1l.667 Table 21. Analysis of Variance Summary: Overcontrolled Hos- tility Source df SS MS F _p Victim 1 35.88 35.88 0.273 - Offense 1 89.66 89.66 0.378 - VxO 1 -8.82 -8.82 0.038 - within 83 5685.88 131.29 86 5726.88 Again, no significant differences were found. Interestingly, only 10 subjects (21.3%) attained an overcontrolled hostility score which exceeded the critical value of 70. The percentage reaching that value was relatively small in each group: A-F, 18.3%; A-NF, 20%; M-F, 26.7%; and M-NF, 20%. To determine the degree of interrelatedness of the four measures of impulse control, an intercorrelation matrix was generated for the four measures. (See Table 22.) It should 92 be pointed out that prior felony arrests were used rather than total prior arrests because of the extreme cases and heterogeneity of variance in the total prior arrests which distorted the mean values. Table 22. Intercorrelation Matrix for the Four Measures of Impulse Control Adjective Self— Overcontrolled Check List Control Hostility Prior Felony Arrests .08 -.03 -.03 Adjective Check List .26* .09 Self-Control .69** *p<.05; **p< .01 There was a high correlation between the Overcontrolled Hostility Scale and the Self-Control Scale (r=.69, p< .01). The O-H Adjective Check List showed a significant correlation with the Self-Control Scale of the CPI (r=.26, p