Printer Friendly

Towards a typology of homicides on the basis of personality.

This paper presents the results obtained by examination of 112 homicide convicts in a penitentiary institution. The following psychodiagnostic instruments were applied: standardized interview, MMPI, and S-R scales of morality and aggressiveness. A series of demographic, criminological, victimological, sociopathological, and psychiatric data, classified into twenty-six variables, was gathered by reviewing police files. On the basis of typical MMPI profiles, the subjects were classified into four groups: psychoses, hypersensitive - aggressive personalities, psychopaths, and |normal' subjects. Factor analysis was carried out using the personality variables and others relating to criminality etc. Six factors were interpretable, each describing more closely the types suggested. Discriminant function analysis proved this typological classification to be justified. The role of aggression and psychopathological factors in homicide etiology is discussed.

In spite of widely held beliefs that homicides are committed by aggressive, anti-social individuals who easily fit into the category of |psychopath', there is considerable evidence that a significant number of murderers are not aggressive and are |normal' in a psychopathological sense (McGurk 1978; Megargee 1966). Trying to find out why there are so many non-aggressive individuals among those who commit such an aggressive act as homicide, Megargee (1966) suggested that murderers could be divided into two types: |over-controlled' and |under-controlled' (by this, of course, meaning the control of aggression). Taking this idea further, Blackburn (1971), who was studying a population of murderers in a psychiatric prison hospital, put forward a classification of murderers into four types: |paranoid-aggressive', |depressive', |over-controlled repressors', and |psychopathic'. Objections were raised that Blackburn's results were the consequence of dealing with a selected population. In response to that, McGurk (1978) examined the |normal' homicides, that is, the prison population of those convicted for homicide, and obtained results that were almost identical to Blackburn's typology. In both studies the typology was based on MMPI profiles, which makes them comparable to our results, presented in this paper. Still, it can be remarked that previous studies have neglected the influence of non-test variables, and that they have been performed on only a small number of subjects (forty-five and forty subjects respectively in the two studies cited).

In the research described in this article, we have tried to overcome these imperfections and also to obtain an answer to the following questions: Is it possible to find common personality characteristics among homicide convicts which would help us to understand and, possibly, prevent this crime? Can we discover and empirically prove the existence of a group of features, habits, and traits that represent manifestations of a general personality structure among murderers, through which we could justify scientifically their typological classification? Or is it possible that every murderer is a unique personality, and our assumption about a |typical' killer, or |types' of killers, just another example of the usual scientist's article of faith that the universe contains regularities that are there for us to discover?


Being of the opinion that the usual method of research into the phenomenon of homicide (i.e. the study of legal or psychiatric files) would not give us the answers we sought, we decided to investigate the prison population. This we were able to do thanks to the kindness of the management and personnel of the penitentiary institution in S. Mitrovica (Yugoslavia). This was a state prison for an area inhabited by about two million people. At the time of our investigation, there were about 1,000 male and about 200 female convicts in it.

The sample consisted of 112 males convicted of homicide - the total population of male murderers in the prison. We had to discard nineteen non-valid psychodiagnostic protocols, so our final sample consisted of ninety-three subjects. (Non-valid protocols mostly resulted from the fact that the subjects did not understand test materials, due to their low educational or intellectual levels.)

A standardized interview was carried out with all the subjects, through which we gathered demographic, criminological, victimological, socio-pathological, and psychiatric data. The subjects' criminal record files were used to guide the data obtained by interview. After the interview, we asked the subjects to complete group personality tests: the MMPI (Dahlstrom et al. 1975) and S-R scales (Ignjatovic el al. 1986) of aggression (|VAPO') and morality (|SEG'). S-R (Situation-Reaction) scales were specially developed tests based on the idea of Endler et al. (1962). The VAPO scale consists of a set of aggressive situations and possible reactions to those situations. The task of a subject is to choose his typical reaction (out of six possible solutions) to each aggressive situation. For the SEG scale, situations represent circumstances which call for moral judgement.

Results and Discussion

MMPI profiles

As previous investigitors had found (Blackburn 1971; McGurk 1978), so we also noticed a great deal of uniformity among the MMPI profiles obtained, which made us believe that typological classification would be possible. On the basis of uniform MMPI profiles, the subjects were grouped into four groups: |psychotic' profiles (five subjects); |hypersensitive-aggressive' profiles (forty-six subjects); |psychopathic' profiles (sixteen subjects); and |normal' profiles (twenty-six subjects).

The first group showed psychotic MMPI profiles (a characteristic elevation of the Pa and Sc scales). The validity of such results was confirmed by data obtained from police and medical files: reports of delusions, hallucinations, and usually some form of bizarre homicidal act (e.g. cooking one's own child in a pot). Even though there were only five subjects in this group, its existence was a great surprise to us, because, according to our legal rules, one would not expect to find any psychotic subjects in a regular prison, since psychotic prisoners are supposed to be placed in psychiatric prison hospital institutions.

This type of profile was not observed by other authors. McGurk (1978), however, found one type of profile that is in between our |psychotic' and |hypersensitive-aggressive' profiles and identified it as |paranoid-aggressive'; but he interpreted it as a psychotic-paranoid profile. However, one should notice that this |group' of McGurk's consisted of only three subjects.

The largest group (49 per cent of the sample) consisted of those with a |hypersensitive-aggressive' profile. Here the profile is characterized by increases in the Pa scale (sometimes with a small peak on a D scale), but without any elevation of the Sc scale. Such a profile was interpreted as the result of hypersensitive personality characteristics giving rise to particular difficulties in interpersonal communication, denoting a person who is easily offended, intolerant of frustration, prone to impulsive-aggressive outbursts, possibly introverted and dysphoric. Such a person will be always dissatisfied and disappointed with people, and is very likely to show paranoid interpretations of social situations. Extremely rigid, such people are very often fanatically oriented in one direction. Uncooperative, unable to adjust socially, even though often conscious of the faults they are making in interpersonal communication, they are not able to correct themselves. Despite some similarities between our |psychotic' and |hypersensitive-aggressive' profiles, the clinical data confirmed that subjects with the latter profile were not psychotic.

In the third group, we had subjects with a typical psychopathic profile (i.e. a peak on the Pd scale). The usual interpretation of this profile refers to emotionally unstable, immature, and impulsive persons. Such people tend to be outstandingly egocentric, irresponsible persons, with a strong tendency to reject moral norms. They also have a tendency to reject any authority, which leads them into constant professional and interpersonal conflicts, followed by aggressive outbursts. This type is particularly characterized by poor aggression control. Rebellious and antisocial behaviour is often accompanied by overestimation of self. There is also an inclination towards alcohol abuse.

Similar or identical profiles, interpreted as psychopathic or sociopathic, have been found by most other authors. Despite the usual stereotype of a psychopathic killer as a |typical criminal', in our sample we registered a relatively small number (17 per cent) of psychopathic personalities.

The fourth group, comprising almost one-third of our sample (28 per cent), consisted of subjects with |normal' MMPI profiles. Its existence convincingly proved to us that a large number of persons convicted for homicide do not belong to any psychopathological group. In comparison to the previous groups, where the single profiles were almost identical in their configuration, this group showed more variety. Besides completely |non-pathological' profiles (where scores on all scales were below a t-score of 70 and the profile did not have a specific configuration), we also found in this group several profiles that could be characterized as depressive (n = 4) or neurotic (n = 5). Still, analysing the complete psychodiagnostic material available to us, we reached the conclusion that these characteristics could be treated as reactive to the penal situation, and not as elements of a permanent personality structure. Thus, they were not of etiological importance for the act of homicide. For this reason, these subjects were also placed in the |normal' group.

Factor analysis

Despite the relatively high uniformity of classified MMPI profiles, they alone are not sufficient for drawing conclusions about the possibility of making a typology of homicides. For discussion on the justification of assessment of |type' or |class', we need multivariate analysis. Therefore, we included psychodiagnostic conclusions on the personality (as above) as separate variables, and along with twenty-six other (demographic, criminological, victimological, sociopathological, and psychiatric) variables conducted factor and discriminant function analyses. The psychotic subjects, because there were so few of them, were not included in this step; the total sample therefore consisted of 88 subjects.

The method of principal component analysis gave ten factors whose eigenvalue was higher than 1. There were six interpretable factors (cumulative percentage of variance = 47.54), which we shall review independently.

The first factor clearly points up the characteristics of the |hypersensitive-aggressive' type. High loadings on this factor linked to both tested (VAPO) aggression and manifest aggression, as well as auto-aggressive behaviour. At the same time, there were negative correlations with results on the morality (SEG) scale, pointing out the low level of aggression control. On the other hand, the homicidal act of this type most often qualified as unpremeditated (i.e. affective). This shows that their aggressive charge was obviously extremely high and poorly controlled, but its discharge was mostly perceived as |rational' (being provoked). Still, it is precisely this |reason' (and this is very often the paranoid interpretation of social situations) which produces such a stigma for this type of homicide convicts.

(*) Professor of Psychology, University of Novi Sad. (**) Professor of Psychiatry, University of Novi Sad. All correspondence should be addressed to: Prof. Miklos Biro, University of Novi Sad, Department of Psychology, 21000 Novi Sad, Stevana Musica b.b., Vojvodina.

The variables of low educational level, alcoholism, and acute alcoholic intoxication at the moment of the homicidal act add to the characteristic description of this group. Data on the existence of crime, alcoholism, and particularly psychiatric disorders in the family direct us to the hypothesis that the existence of hereditary or developmental load can be treated as possible dispositional elements for the formation of this type of personality.

The second and third factors can be considered together, since they are both marked by psychopathic personality variables (the second factor is saturated mostly with criminological variables which naturally intercorrelate highly). The picture we get by looking at these factors agrees entirely with our usual perception of an antisocial personality, i.e. the |typical criminal': previous crime record, act most often premeditated, an unaggressive and unknown victim, manifest aggressive behaviour, strikingly undeveloped superego (low SEG score), history of juvenile delinquency, low educational level, unemployed status, poor housing, and poor financial condition. Here also we meet with variables that can be interpreted as predisposing factors: alcoholism and crime in the family.


     Factor I (% of variance = 11.01)

|Hypersensitive-aggressive' personality    0.61505
Aggression scale (VAPO) score              0.58703
Manifest aggressive behaviour              0.52855
Low education level                        0.50187
Crime in the family                        0.47737
Premeditation                             -0.46931
Alcoholism in family                       0.44571
Morality scale (SEG) score                -0.42902
Psychiatric heredity                       0.40719
Victim unknown                             0.39204
Alcohol addiction                          0.37981
Self-destructive behaviour                 0.36564
Acute alcohol intoxication                 0.31864

     Factor II (% of variance = 9.81)

Premeditation                  0.54678
Previous crime record          0.52732
|Psychopathic' personality     0.49179
Aggressive victim             -0.48897
Juvenile delinquency           0.40564
Morality scale (SEG) score    -0.39652
Victim unknown                 0.37652
Unemployment                   0.32438
Poor housing conditions        0.30121

     Factor III (% of variance = 7.89)

|Psychopathic' personality        0.51632
Aggressive behaviour              0.43448
Alcoholism in family              0.42314
Morality scale (SEG) score       -0.41413
Crime in the family               0.39463
Low educational level             0.36574
Aggression scale (VAPO) score     0.33681
Acute alcohol intoxication        0.31006

An incidental piece of information might be interesting here: the |broken home' variable does not show any significant connection either to psychopathic personality or to the degree of manifested aggression, while alcoholism in the family correlates highly with psychopathic personality and particularly with aggression (correlation with VAPO score is 0.43). This result suggests the hypothesis that parental absence does not have a significant etiological role in the genesis of aggressive behaviour, whereas parental alcoholism (and very probably parental aggression) does have a big influence.

The fourth factor gathers variables which are closely connected to alcoholism, while personality variables do not have any significant loading with it. This is undoubtedly due to the fact that alcoholism is relatively evenly distributed among all groups of subjects (the total proportion of alcoholics in the sample was 43 per cent). Another interesting fact is that alcoholism (and acute alcohol intoxication) of the murderers very often coincided with alcohol intoxication of the victims also.

     Factor IV (% of variance = 6.77)

Alcohol addiction             0.45788
Acute alcohol intoxication    0.42631
Alcoholism in family          0.33428
Unemployment                  0.31657
Self-destructive behaviour    0.31132
Poor housing conditions       0.30879
Victim alcohol-intoxicated    0.30323

Factors V and VI can be commented upon together, since they reflect the existence of two subgroups of |normal' murderers. The basic difference between them lies in the degree and type of aggression expressed. Even though the group with a |normal' personality profile has, on the whole, the lowest scores on the aggression (VAPO) scale (M being even lower than M of the standardization sample, in contrast to the group of |psychopaths' and |hypersensitives', whose M scores on the VAPO scale were significantly higher statistically), factor analysis shows a certain amount of differentiation within the group. The subgroup indicated by factor V is characterized by a strong superego (SEG score) - i.e. high aggression control - but not by absence of aggression, which is the case with the subgroup indicated by factor VI.

     Factor V (% of variance = 6.05)

Morality scale (SEG) score               0.41361
|Normal' personality                     0.40115
Victim alcohol-intoxicated               0.37443
Aggressive while alcohol-intoxicated     0.37418
Low educational level                   -0.33657
Alcoholism in family                     0.33109
Previous crime record                   -0.32118
Aggressive behaviour                    -0.31009
Acute alcohol intoxication               0.30989

     Factor VI (% of variance = 6.01)

Premeditation                    -0.39879
|Normal' personality              0.36332
Victim = spouse                   0.32336
Aggression scale (VAPO) score    -0.32174
Juvenile delinquency             -0.32006
Previous crime record            -0.30896
Victim aggressive                 0.30007

Factor V reminds one of Megargee's (1966) hypothesis on |over-controlled' murderers or Blackburn's (1971) |over-controlled repressors'. Reviewing the subjects' files, we observe that they were educated persons, without previous recorded criminal behaviour, who, when sober, were very calm and highly socialized, but who, under the influence of alcohol, became |wild'. Their homicides most often developed according to the stereotype of the bar fight. It is very probable that alcohol functioned as a means to decrease tension: when aggression accumulated up to the point when it could no longer be controlled, drunkenness became a suitable alibi for its discharge.

Factor VI describes the non-criminal, non-aggressive group of killers. Their act was most often described as |manslaughter with diminished responsibility' and was usually categorized as an affective act or a homicidal act provoked by the victim's aggression. One high loading on this factor was the victim being the spouse of the murderer. Reviewing their files, we found situations that represented typical |acting out'. By definition, acting out is an inadequate (aggressive) reaction to an insignificant provocation, which symbolically represents a series of similar previous situations when the killer had repressed his aggression; for example, a husband strangles his wife because she refuses to sew on a button for him. But, besides these, we came across situations which were very close to murder in self-defence against an aggressive victim.

The reader will certainly have noticed by now that we have used factor analytic results to describe our types more closely, which is not quite correct in a methodological sense. Of course, factor analysis shows the connection among variables and not among people. For the purpose of classification, it would be preferable to use cluster analysis. However, since we had already performed a classification using MMPI profiles, and since the variable of type of personality appeared as the |leader' in most of the factors, we consider this sufficient justification to interpret these factors as classification attributes. But, in order to check whether the classification including these broad characteristics makes sense, we applied discriminant function analysis.

Discriminant function analysis

The three variables of personality type (based on the MMPI profiles) were the criterion for grouping. All other variables (the same as in factor analysis) were included.

The discriminant function analysis gave results which, in fact, were nothing more than slight modifications of the factor analysis. Therefore, we present in Table 1 only the probability of classification in particular groups (based on a multivariate procedure).


As can be seen from the table, our typology has a very solid base. The proof can be found in the fact that 72 per cent of cases classified on the MMPI profile fitted well into the hypothetical classification performed using the multivariate procedure (which included a whole series of extraclassifying attributes). The most complex, and consequently, the most heterogenous group was the |hypersensitives', while the |normal' group most strikingly differed from the whole sample. This shows that our groups do not behave as distinctive categories, and that there are transitory forms as well.


In reply to an implicit question raised in the introduction, we can say that almost one-third of our sample of murderers did not show any element of psychopathology.

The results obtained enabled the classification of the homicide convicts into four types: psychotic, hypersensitive, psychopathic, and normal. Within the |normal' group it was possible to differentiate a subgroup of |over-controlled' persons.

With respect to manifestation of aggression and its possible etiologic role in homicide, it can be concluded that manifestations of aggression exist as an undoubtedly significant factor in the case of two types: psychopaths and hypersensitives. Here they appear as a permanent and pronounced part of the personality structure, which, along with decreased control mechanisms and situational components, results in homicide. On the other hand, in the |normal' group, aggression is either not expressed or is strongly repressed. Therefore, for this group we can quite confidently state that their aggressive act is of a reactive nature, i.e. it is not in accordance with their basic personality structure.


Blackburn, R. (1971), |Personality Types among Abnormal Homicides', British Journal of Criminology, 11: 14-31. Dahlstrom, W. G., Welsh, G. S., and Dahlstrom, L. E. (1975), An MMPI Handbook. Minneapolis: University of Minnesota Press. Endler, N. S., Hunt, J., and Rosenstein, A. J. (1962), |An S-R Inventory of Anxiousness', Psychological Monographs, 96/4. Ionjatovic, I., Momirovic, K., Dzamonja, Z., Wolf, B., and Sipka, P. (1986), S-R Skale. Belgrade: SDPS. McGurk, B. J. (1978), |Personality Types among "Normal" Homicides', British Journal of Criminology 18: 146-61. Megargee, E. I. (1966), |Undercontrolled and Overcontrolled Personality Types in Extreme Anti-social Aggresion', Psychological Monograplas, 80/611.
COPYRIGHT 1992 Oxford University Press
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1992 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Title Annotation:Yugoslavia
Author:Biro, Miklos; Vockovic, Nikola; Djuric, Veljko
Publication:British Journal of Criminology
Date:Jun 22, 1992
Previous Article:Alcohol use and offence type in young offenders.
Next Article:From Prohibition to Regulation: Book-Making, Anti-Gambling, and the Law.

Terms of use | Privacy policy | Copyright © 2019 Farlex, Inc. | Feedback | For webmasters