During the 1990s, the French police and gendarmerie recorded an annual average of almost 2,000 cases of homicide, attempted homicide and fatal blows. What is the social context of this phenomenon in our society ? The present study originates in the realization that it was difficult to answer such a question in the current state of our knowledge. Research on the subject has been very scarce in France. The earliest studies were linked to the development of research on juvenile delinquency from the 1960s on, mainly around the Vaucresson research centre (e.g. Henry and Laurent, 1974, pp. 71-87). In 1976 a thesis by J.-C. Chesnais on Violent deaths in France since 1826 established the major historical, geographical and social characteristics of homicides. No further data on homicides were produced until J.-M. Bessette published his doctoral thesis in 1984, and much later some articles based on that research (mainly Bessette, 1994), but even then these relied on empirical materials dating back to the 1960s and 1970s. Finally, B. Michel’s (1991) thesis on murder provided an interesting review of the legal, historical and social science literature, but by way of empirical evidence relied only on a few dozen French and Swiss newspaper articles that also dated back to the 1970s. In fact, in the French intellectual world, the social scientist must turn mainly to the historical school for extensive and informative work in this area, both in an empirical and a methodological perspective.

In the United States the situation is quite different. In the 1950s, the books published by Albert Morris (1955) and by Marvin Wolfgang (1958), a follower of Thorsten Sellin, soon became “classics” that inspired many studies that were usually based on long-term analyses of series of criminal cases processed by the police. Since then, there have been hundreds of empirical studies that are the staple every year of numerous American journals of sociology and criminology. As the end result, what was initially a mere research topic has become a fully-fledged sub-field of its own, as attested by the launching in 1997 of the journal Homicide Studies. A mere journal article like this one cannot begin to inventory that scientific literature. However, some of the major issues can be mentioned here. First, and most detailed, is the search for the factors through which behaviour can be analysed, both from the point of view of the individuals (perpetrators and their victims) and of the urban backgrounds in which they are to be found with particularly high frequency. Second, the analysis of the relationship between perpetrators and victims has been the object of fewer studies, but these are all the more interesting because they usually provide a foundation for attempts to construct typologies of homicides. Third, the investigation of the circumstances and the context (whether situational, material, relational, etc.) of criminal activity is the subject of a smaller number of studies. Finally, the important variability of homicide rates between and within countries is a common issue that elicits continued historical, anthropological and sociological studies.

For this article, we accept the general definition of homicides as acts that result from interactions between perpetrators and victims, in given situations and given historical, social and cultural contexts (Mucchielli, 2002, p. 148). The article addresses one aspect of that complex reality by presenting the demographic and social characteristics of the perpetrators and their victims on the basis of a local sample. The general sociological implications are discussed.

The population under study is in no way representative of the French population. However, it may be considered highly representative of the penal population with regard to the three demographic indicators (age, sex and nationality) mentioned in the court statistics [2].

In a more general way, the critical assessment of the source materials raises two series of questions: what they reveal and what they do not.

The critical assessment of what the sources reveal focuses usually on what is called “reconstruction”. This is “a ‘discourse on’, or reconstruction of, the raw material — whether persons or facts — that [the institution] has contributed so as to make it usable within the framework drawn by the institutional logic and its social function” (Robert, Lambert and Faugeron, 1976, p. 2). The sociologist will not observe facts, but an institutional narrative of facts, a reconstruction by the police and the court. In fact, the final information laid out by the public prosecutor upon completion of any criminal investigation is a two-way narrative — the narrative of the criminal facts, but also that of the criminal investigation itself, in which the more believable elements of the investigation, as well as those that are less so, can be highlighted (Macchi, 2001, pp. 181-182).

Such criticism should not be taken lightly, but it should not entail a rejection for the purpose of a sociological study of behaviour. It is a fact that judicial errors may happen, which constitute precisely an erroneous reconstruction of the facts. On completion of the study, however, it appears that the phenomenon is extremely rare — out of the 102 cases under study, only one seemed partly questionable. Moreover, the defendants seldom question the facts with which they are charged. In two out of every three cases, the perpetrator did not try to conceal the crime and was arrested — or even surrendered — without putting up any resistance (Mucchielli, 2004). In virtually all cases, the perpetrators soon confess their crime, and their defence usually will consist of denying the intent to cause death rather than the crime itself.

Finally, even though judicial reconstruction is necessary, the resulting picture of the facts is not an imaginary construct that misrepresents social reality. Except for extremely rare cases in which the magistrate and the investigators responsible for the prosecution charged the wrong person, the elements in the files that are most useful for analysing what happened do not seem to be fundamentally misrepresented or distorted or transformed by the institutional process.

Actually the data sources raise a problem not so much by what they reveal, but rather what they do not. The main biases are not in the content of the judicial narrative but outside of it — in the topics on which it is silent and those that it ignores. One of the glaring omissions is evidence on the victim. In their pioneering studies, von Hentig (1948), Morris (1955) and Fattah (1971) had already stressed that point. Homicide files often provide only the first and last name of the victim, the date of birth and the date of death. It is usually through the perpetrator and his human environment (his family, neighbours, colleagues) that pieces of information can be gathered concerning the victim, in virtue of the fact that the perpetrator and his victim knew each other in almost 80 % of the cases.

Do women and men commit the same crimes ? The fact that there are so many more men in the sample makes it difficult to interpret some of the relations presented in Table 1. However, it appears that men are more often involved in murders (as opposed to attempted murders and fatal blows) committed in the street, as a result of a poorly motivated conflict (originating in circumstances of the moment, not long brewed), between a perpetrator and a victim who did not know one another. To the extent that small numbers allow any conclusion, it may be broadly assumed that women commit crimes of a rather different type. In addition to infanticide, women are more often involved than men in crimes against a spouse, in the marital home, with a view to punishing him (for his treacherousness, his cowardice, desertion, etc.).

Table 1
Main variables differentiating the perpetrators of homicide by sex

The contrast between crimes committed by men and by women can thus be accounted for, in a first analysis, by different modes of interaction and life style (public places, bars, night life are more characteristic of men), and this is more broadly applicable to all physical assaults other than sexual ones (Pottier, Robert and Zauberman, 2002, p. 56). When a crime is committed not in some public place but within the family circle, the share of women perpetrators increases markedly, though not enough to reach parity with men. Beyond differences in behaviour and life style, the contrast also reflects social stereotypes and gender identities that encourage men’s resort to physical violence. And the fact that, in criminal matters, the distribution of perpetrators by sex has not evolved over time — despite the deep transformation in the social status of both sexes in western societies — bears testimony to the strength of the stereotypes and the identities in question.

The age structure of the population studied in the Yvelines is similar to that of all the persons sentenced for homicide at the national level (Figure 1). In France, the youngest perpetrator sentenced in 1990 was 15 years old at the time of the crime, and the oldest one was 62. Those, however, are exceptional cases. The age distribution is similar to that observed in other European countries, for example Switzerland (Massonnet, Wagner and Kuhn, 1990, p. 88) or the Netherlands (Smit, Bijleveld and Van der Zee, 2001, pp. 304-305). Homicides committed by young people under age appear to be extremely rare; in the sample, crimes are committed by men aged 18 to 50 in almost nine cases out of ten, and by men aged 18 to 30 in almost four cases out of ten.

Compared with the overall population, however, the perpetrators of homicides are rather young. Comparing them to the age distribution of the total population of the département in the 1990 census indicates that the criminals are definitely more numerous below age 35, and less over that age. The 20-24 and 30-34 age groups are markedly over-represented in the sample. Homicide is thus definitely a matter for young adults.

Table 2
Persons sentenced for homicide, by nationality

Table 3
Age distribution of French and foreign nationals in the Yvelines département according to the 1990 census, and in the study population

Foreigners in the population under study are also over-represented among employed and unemployed blue-collar workers in this post-industrial area in the southwest of the Paris region. One third of the foreigners sentenced for homicide are blue-collar workers (as against 12 % of the French), and 21 % are unemployed (as against under 10 % of the French). These are very large differences (a ratio of 1 to 3 for the former and of 1 to 2 for the latter). The population in question is subjected to more precarious prospects [9], and is largely concentrated in the socially supported housing schemes situated close to the former industrial sites [10].

Finally, the statistical analysis shows that (i) foreigners do not commit racist crimes (though they are sometimes victims thereof), and (ii) in similar situations of conflict, foreigners react under the influence of panic more often than French nationals. The latter trait may reflect a greater vulnerability in every respect, not only on economic but also on legal, relational and emotional levels. Quite often, these persons are living in conditions that are characterized by great uncertainty. At this stage, however, these considerations are no more than hypotheses, although they are corroborated by other analyses on the delinquency of foreigners (Mucchielli, 2003).

Table 4
Social origin of the perpetrators of homicides

In 18 % of cases, it is impossible to know the father’s occupation with any certainty. This is the case mostly when the parents are unknown or dead and the children were fostered out by the DDASS (Direction départementale de l’action sanitaire et sociale) [11] from an early age, usually after they were abandoned. In most cases, however, the incidental mention of an occupation, of the standard of living or the housing conditions in a court document, a hearing or a police report makes it possible to infer that the family also belongs to the working classes. Reading the whole of the file may even suggest that the father was usually an agricultural labourer or a small craftsman [12].

Taken together, those results thus stand out sharply — about four out of five individuals in the sample belong to the working classes. Inasmuch as they have themselves largely reproduced that general social position, as shown below, it can be stated that one characteristic of the social history of the families of homicide perpetrators is the reproduction of a lower social position. It could also be termed “inherited poverty” (Chambaz and Herpin, 1995, p. 124).

Table 5
Socio-occupational categories of the population of the Yvelines département in 1990 and of the perpetrators of homicides

Considering their level of schooling and the period studied (the years 1987-1996), which was characterized by a very high rate of unemployment particularly among blue-collar workers, it comes as no surprise that only 39 % of the perpetrators of homicides had a job and were economically active. However, it is more surprising that the rest are much less often unemployed (a bare 10 % of all perpetrators) than simply economically inactive (about four cases out of ten). After examining the files, it appears moreover that among the economically inactive a high proportion of persons do not have a permanent home (over 1 out of every 3 cases); the proportion would probably be twice as high if some had not been able to rely on family solidarity [14]. The other situations are comparatively unimportant (two retired persons, six students, a few housewives and one young man drafted in military service).

Among the 48 perpetrators of homicides who are in the labour force, the vast majority belong to the working classes and to the category of lowest wages [15]. Blue- and white-collar workers account for almost four fifths of the total number (79 %). This includes individuals with a CAP or a baccalauréat, as well as some with no diploma. There are also six self-employed craftsmen, tradesmen or businessmen, who were actually all in small enterprises (two restaurant owners, one fruit and vegetables market vendor, one electrician, one owner of a small scrap yard, and one scrap dealer). These occupations require hardly any qualification and are not very lucrative. At the other extreme, only one administrative manager and one of the restaurant owners (who was in fact running a medium-sized business) obviously belong to a higher social stratum. (To them we should perhaps add a retired businessman.) Finally, the group includes two individuals with middle-level occupations — a teacher in a secondary school and a computer technician in a bank, and perhaps the manager of a small supermarket. On the whole, about 90 % of the perpetrators of homicides who are employed belong to the working classes, with many of them at the lowest salary levels. And if those who are economically inactive are included, the proportion exceeds 95 %.

Let us take another look at the opposition between the economically active and the economically inactive. Does this opposition make sense, when we know that inactivity is often temporary, and that in fact many young men alternate employment and inactivity ? The computations make two elements clear.

Thus, it appears that the “economic activity” variable is of no trivial importance. Moreover, it is closely linked to marital status and general health.

Table 6
Life circumstances of homicide perpetrators during childhood

We note the frequency, seldom mentioned in the specialized scientific literature, of situations where the individuals were not raised by either parent, or had only distant and occasional contacts with their parents. One characteristic of the population under study is the large proportion of particularly unstructured family situations. In close to 20 % of cases, the perpetrators were raised for several years by a third party (their grandparents, for example), by foster families selected by the DDASS, or by educators in homes for the youth. This is an important variable because statistical analysis makes it possible to identify certain consequences of the family histories on the psychological profile of the perpetrators. Emotional deprivation, immaturity and a tendency toward depression are strongly linked to the fact of not having been raised by their parents. The analysis also shows that the perpetrators of homicide who experienced that type of childhood are more likely to have committed the crime after a one-to-one fight for a reason resulting from immediate circumstances (as opposed to a bitter, long-standing conflict). In total, the family factor constitutes a personal handicap and a considerable social risk.

Confirming another firmly established fact (Mucchielli, 2001b), family conflict is one more factor that strongly contributes to making individuals vulnerable, even though it is usually hidden behind apparently stable forms of family relationships. The nature of the perpetrators’ relations with their parents is tentatively delineated below, as far as the information collected made it possible. In 18 % of cases, it proved impossible. And even in the other cases, the data presented in Table 7 are only rough approximations that are probably well below the real figures. Violence within the family is typically under-reported (and is even sometimes not perceived as such by the victims). Nevertheless, certain tendencies are clearly noticeable. In cases where we know the nature of the relations between parents and children, they are usually conflictual, and in more than one case out of two (34 out of 59), the conflict involved physical violence. In at least 18 cases the father (or stepfather) used violence against the mother, and also in at least 18 cases, he used violence against the children; in at least 4 cases the mother (or stepmother) used violence against the children, with those situations sometimes occurring concurrently.

Table 7
Parent-child relationships in the family of origin of homicide perpetrators

We attempted to measure the impact of those conflictual relationships on the other variables. It can be established that being raised in a conflict-ridden family atmosphere, apparently free from physical violence, is on the one hand very often linked with a highly emotional psychological profile, and on the other hand quite often associated with economic inactivity and occupational instability. Moreover, being raised in a conflict-ridden family atmosphere that involves physical violence is very frequently linked to a psychological profile characterized by immaturity, emotional deficiency and, possibly, psychotic tendencies (the three cases recorded in this category are also the only three in the sample). This element is also present in the history of the only three cases where the conflict between the perpetrator and the victim had lasted for more than 10 years, as though having undergone violence made a conflict-ridden, long-term relationship — usually in married life or married-like situations — tolerable until the day when it resulted in an act that was more violent than usual.

Here, the sample can be roughly divided into two groups. On one side, half the perpetrators are “solitaries”, in that they have never lived in a couple (32 %), are separated or divorced with no children (9 %) or have children that they never see (9 %). On the other side, we find couples with children who live in their home (25 %) or away from home (3 %), then childless couples (11.5 %) or separated or divorced individuals who still see their children (10.5%). If the latter category were to be added to the first group, one might say that a clear majority (about 60 %) of the individuals in the sample had no married life at the time of the crime [18].

Table 8
Sex of the victims

The age distribution of the victims corresponds roughly to that of the perpetrators, though there are fewer minors and more individuals aged over 50 among the victims. Close to 85 % of the victims are aged between 20 and 50.

Finally, a look at the occupation of the victims confirms the observation of a close social proximity of perpetrators and victims (Table 9). The economically inactive and the blue- and white-collar workers are clearly over-represented once again (though there are more white-collar than blue-collar workers as a result of the large proportion of women among the victims). The proportion of victims who belong to higher social strata is slightly higher, but the number is too low (4 cases) to formulate a hypothesis at this stage of the research.

Table 9
Socio-occupational category of the victims

Table 10
Nationality of the victims

How about France ? On close re-examination of the 58 cases of murder and the 75 perpetrators involved, it appears that the intra-racial character is also observed in France. Out of the 75 perpetrators, 51 are of French origin, and 24 are either foreign nationals or French nationals of foreign origin. Among the murderers of French origin, 39 (or 76 %) had murdered other persons of French origin, and among the 24 foreigners or French nationals of foreign origin, 20 (or 83 %) had murdered other foreigners or French nationals of foreign origin. In total, only in one fifth of cases (16 out of 75) were French nationals of French origin involved with foreigners or French nationals of foreign origin. The qualitative analysis of these files brings out a few specific characteristics that lead to certain hypotheses, even though the reasoning is based on a very small number of cases. Among the four inter-racial murders in which the victim was of French origin, one was a case of marital conflict, one was a group fight and two were settling a score between delinquents. The twelve inter-racial murders in which the victim was a foreign national or a French national of foreign origin also include a few fights and marital crimes; they include no settling a score, but three crimes with a strong racist connotation.

In fact, at this level of generalization, the victims’ profile proves to have only one special characteristic if compared to that of perpetrators — there are more women involved. In terms of age, occupation and nationality, the perpetrators and the victims form a homogenous population. This comes as no surprise, considering the weight of marital and family crimes, the fact that perpetrators and victims live geographically very close to one another, and the high degree of concentration of the poor in certain urban areas.

Though the debate is not quite settled in American statistical criminology [24], the relation between homicides and the lack of economic resources remains one of the most solid results of research. The puzzle remains as to why, as a rule, the poor kill so frequently, while the rich seem not to. Wolfgang thought that, among the upper classes, homicide was quite clearly different from what it was among the working classes — much more often linked to mental pathology or rational calculation (premeditation) on the perpetrator’s part. This has been generally confirmed by subsequent studies, however scarce. The best known was based on a selection of 119 criminal cases involving a perpetrator who belonged to the upper or middle classes, and featured specific homicide scenarios. Green and Wakefield (1979, p. 175 ff.) believed that about one quarter of the murderers who belonged to the upper classes committed murder for financial reasons (insurance, property, inheritance); this is a higher proportion than the murders related to theft among the poor. Another part of the perpetrators seemed to have acted because they were severely depressed (most of them in fact later committed suicide or attempted to). Conversely, the most important type of homicide among the poor, i.e. quarrels that unexpectedly break out between young men and then get out of hand, usually with drinking as a factor that relaxes inhibitions, is virtually nonexistent among the higher classes. As noted by Wolfgang (1967, p. 7), murder among the upper classes seems to be more related to the marital and family private sphere, with greed as the main motivation (hence the much greater recurrence of premeditation) or with a depressive condition as an associated factor. If we had to propose a general hypothesis at the close of this study and of the review of the American literature, we would suggest the following: in the vast majority of cases, individuals are the more likely to kill the less they have to lose in social life — no high status occupation or reputation to preserve, no personal or family projects, no real prospects for the future.

Such macro-sociological profile of murderers, in which the lack of economic resources and the fragility of marital and family ties are combined, is very similar to the profile of other “outcasts” in French society (the “disqualified”, “disaffiliated”, “dis-inserted”), as emphasized in works by Castel (1995), Paugam (1991, 1993) and Gaulejac and Taboada Léonetti (1994). Over time the same studies have started taking into account the disruptions of social links that occurred early in individuals’ childhood and left their mark for the rest of their lives on both the social and the psychological plane (Marpsat, Firdion and Meron, 2000; Paugam and Clemencon, 2002) [25]. This is a lead that we had decided to explore from the beginning of this research, in 1998, because we believe that the biographical approach is important, both from a statistical point of view, with event history analysis, and from a qualitative point of view, with life histories (Bertaux and Léomant, 1987).

At a first level of analysis, the perpetrator’s loneliness does appear to be a factor correlated with homicide. In the sample, about 60 % of the individuals had no marital life at the time of the event, whether they had never lived as a couple or were separated or divorced, and whether they were childless or had children they saw frequently or not. As to the economic factor, we have seen that it is even more decisive. But such accumulation of handicaps or situations of disaffiliation are encountered too frequently in the general population to be a sufficient macro-sociological characteristic of homicide perpetrators. In our opinion, one weakness of American studies on homicide (as well as of many statistical studies in macro-sociology, for that matter) is that their perspective is too “horizontal” — in other words, that they are concerned exclusively with the current situation of individuals, and neglect their biographic itinerary or, to put it in simpler words, their history. At least three additional general results, which seem impossible to ignore both empirically and in theory, came out of our study. First, we have seen that in 60 % of the cases where their nature was known, the relations between the murderer and his or her parents during childhood were conflict-ridden (the conflict involving physical violence endured by the murderer-to-be in at least half the cases). Second, we have recorded the fact that less than one murderer out of ten had a “normal” school career (i.e. one conforming to the norm promoted by the institution), that over two thirds left school without any qualifications, in a situation of complete failure. Third, we have noted that in about one case out of five, the murderers had a particularly unstable and emotionally deprived childhood, as they were raised by third parties (such as grandparents), by foster families selected for them by the DDASS, or educators in institutions. Last, we have suggested that those deficiencies and handicaps, which left their mark throughout the murderers’ life histories, also had a major impact on their psychological state. In particular they are the determining factor of the very high proportion of individuals characterized by immaturity at the affective level, but also by anxiety and a tendency to be highly emotional and depressive. And this psychological state resulting from the family background and the social history of the individual can also be considered as one of the structural data discussed here. Murderers are not only individuals, as suggested above, who find it easy to commit murder because they have nothing to lose in the present and nothing to expect for the future. They are also individuals, in many instances, who never got much in the past from (family and social) life. It seems as though individuals value human life less to the extent that nothing had ever suggested that their own life could be worth anything. The past, the present and the future are truly the three segments of a time chain that must be reconstructed. Such an approach also applies on a more general level in the sociology of deviance [26] and certainly in other areas of sociological research.

These are the general conclusions that emerge from this study at the macro-sociological level. This level of analysis is undoubtedly insufficient and many more developments should be pursued at a micro-sociological level. In particular, the relationship between perpetrators and victims should be studied, as well as the context and actual circumstances in which the crime was committed. Such developments would be needed to construct an operational typology and encompass the complexity of the real world. Although the macro level is insufficient, however, it remains indispensable. In their writings on homicide, some criminologists refuse to take into consideration those massive research findings and limit themselves to the study of the perpetrators’ apparent motives and the circumstances in which the crime was committed (cf. for example Cusson, 1998, pp. 22-35). Such neglect of the social contexts is very questionable. From a statistical and macro-sociological point of view, homicides are not committed randomly anywhere, in any social environment or at the end of any life itinerary. Homicide is a social phenomenon that undoubtedly reminds society that it cannot complete its process of policing mores as long as pockets of poverty subsist, where men and women are concentrated whose life histories incite to value life cheaply, be it that of others or their own.