Correctional Service Canada
Symbol of the Government of Canada

Common menu bar links

Women Offender Programs and Issues

Warning This Web page has been archived on the Web.

Understanding Violence By Women: A Review of the Literature

Studies Of Female Aggression And Violence

A. Violence in the home
B. Offence-based studies
C. Institutional studies

This section examines the literature on violence by women in three main groups: a) studies of violence in the home; b) offence-based studies; and c) institutional studies. With the exception of child abuse, most reviewers agree that relatively few studies have been published on female aggression or violence, reflecting the rarity of female violence, and the dominance of interest in explaining male violence.


A. Violence in the home

Family violence research is plagued by many of the problems identified for criminal violence. These include the isolation of the different disciplines such as medicine, psychology or social work working in the field; the limitations of individual explanations; underreporting because of the tradition of privacy about family life; the limitations of definitions, classification methods and data collection; and gender blindness (McCall and Shields 1986; Schwartz 1989; Ohlin and Tonry 1989; Dougherty 1993; Reiss and Roth 1993).

The term `family violence' includes consideration of physical, sexual and psychological abuse between any family members, but in practice has focussed primarily on men's physical violence against partners and children (Ohlin & Tonry 1989; Strauss 1987; Light 1987; MacLeod 1987; DeKeseredy & Hinch, 1991). A major area of recent dispute, nevertheless, has centred around the issue of the relative contributions of men and women to violence within the home, allowing some researchers to claim that women use violence in the home as much as men. For example, Strauss & Gelles (1990) claimed on the basis of American survey data that `in contrast with their behaviour outside the family, within the family women are about as violent as men' (p. 11) and a Manitoba study by Sommer, Barnes & Murray (1992) made similar claims. It is still necessary for others to assert that:

`Violence against often persistent and severe, occurs in the context of continuous intimidation and coercion, and is inextricably linked to attempts to dominate and control women (Dobash, Dobash, Wilson & Daly 1992, p. 71)

Much of this dispute centres around the methodology used, in particular the Conflict Tactics Scale (CTS) developed by Gelles (see Strauss & Gelles 1990), and the assumptions that violence by men and women can be regarded as the same phenomenon, that their power, responsibilities and status within the family are similar, and that their socialization patterns in relation to the use of aggression or violence are the same (Dobash et al., 1992; Dougherty, 1993; Campbell 1993). Thus Dobash et al., (1992) point out that the CTS in `measuring' incidents of violence perpetrated by husbands and wives fails to consider the `intentions, interpretations and the history of individuals' relationship' (p. 79) and cannot measure the continued repetitious violence associated with battering. Faith (1993) notes that the Manitoba study (Sommer et al., 1992) included women throwing things, but not necessarily at their partner, in their definition of violence.

The evidence on the basis of the most serious outcome of violence in the home, murder, clearly underlines the very different consequences of such violence for men and women (Cote 1991; Dobash et al., 1992; Wilson and Daly 1993):

`Men often kill wives after lengthy periods of prolonged physical violence accompanied by other forms of abuse and coercion; the roles in such cases are seldom if ever reversed. Men perpetrate familicidal massacres, killing spouse and children together; women do not. Men commonly hunt down and kill wives who have left them; women hardly ever behave similarly. Men kill wives as part of planned murder-suicides; analogous acts by women are almost unheard of. Men kill in response to revelations of wifely infidelity; women almost never respond similarly although their mates are often adulterous. The evidence is overwhelming that a large proportion of the spouse killings perpetrated by wives, but almost none of those perpetrated by husbands, are acts of self-defence.' (Dobash et al 1992 p. 81)

Dobash et al., (1992) also point out that part of the claim for the equivalence of male-female violence rests on the similar rates of spousal homicide in the US, a situation which is peculiar to that country alone as Wilson and Daly (1993) have demonstrated for Canada. It is also well recognized that the consequences of men's and women's violence within the home are also more severe for women (Stets & Strauss 1990; Reiss and Roth 1993). In Canada, women account for 80-90% of victims in assaults and sexual assaults between partners (Johnson, 1989). Even those who do argue that women are violent in the home conclude that `husbands as victims constitute an extremely small amount of incidents resulting in serious but not fatal injury.' (Strauss & Gelles 1990).

Faith (1993) notes that violence between same-sex partners has only recently been acknowledged, but that there is no evidence that `butches' are more aggressive than `femmes', and no association between lesbianism and violence (Lobell 1986; Brooks 1981 cited in Faith 1993).

As with all forms of violence, therefore, the dominance of male violence has meant that most research focusses on the characteristics of male batterers, on demonstrating their greater contribution to violence in the home, and on explanations which stress their power and control over women. Little has been written about women who are violent towards their partners except in the case of murder (Campbell 1993; Dougherty 1993). Stets & Strauss (199O) like Campbell (1993) suggests that men's violence against their partners is instrumental in maintaining control, while women's is an expression of their frustration or stress.

It is not only feminist writers, or those writing from the perspective of women as victims of abuse, who put forward such views. Reiss and Roth (1993) conclude that feminist theory which focuses on the unequal distribution of power within the family is one of the most fruitful ways of understanding family violence and for developing prevention and intervention programmes.

Child abuse by women

Child abuse has not gone unnoticed by researchers as a serious social problem requiring understanding and intervention. While the research is extensive, gender is rarely a conceptual marker (for an exception see McCall & Shields 1986).

Feminist studies of child abuse have focussed almost solely on male perpetrators, an approach which Gordon (1987) argues should be transcended. Dougherty (1993 p. 94) argues that child abuse by women has `virtually been ignored' by those working from a women-centred perspective leaving only `gender neutral psychiatric factors and defects in character structure' as explanations.

`Defending women against violence is so urgent that we fear women's loss of status as political, deserving `victims' if we acknowledge women's own aggressions. These complexities are at their greatest in the situation of mothers, because they are simultaneously victims and victimizers, dependent and depended upon, weak and powerful' (Gordon 1987 p. 69)

Nevertheless there are exception as Washbourne (1983) suggests:

`feminists have recognized that women are on occasion violent towards men and understand that violence as the result of societal and interpersonal pressures on women. Women's violence towards children needs to be recognized and discussed in the same context. Women's abuse of children stems directly from their own oppression in society and within the family'. (p. 291)

No single definition of child abuse has been widely accepted (Garbarino 1989; Lenton 1990) and there is no consistency in the use of terms which may include `neglectful', `maltreating', `aggressive', `abusive', `physically abusive' and even `potentially abusive'. Some researchers include spanking as abuse, others exclude it. Such inconsistencies create methodological and theoretical problems (Keller & Erne 1983).

The extent of women's violence towards children

Since women undertake the majority of child care they are more likely to be reported in abuse statistics. Thus in England and Wales Heidensohn reported (1992) that natural mothers were implicated in 33% and natural fathers in 29% of physical abuse cases from 1983-87. However, when account was taken of who the child was living with, the fathers were implicated in 61% of cases and the mothers in 36%. As Gordon argues:

`Given that men spend on the whole so much less time with children than women, what is remarkable is not that women are violent towards children but that men are responsible for nearly half of the child abuse.' (1987 p.69)

In the US, Gelles (1987) examined violence against children in a small sample of families known to agencies and the police for violence, and a group not so identified. Most of the violence involved spanking or slapping. Not surprisingly, he found mothers, the primary caretakers and disciplinarians, to be the most physically aggressive parent. Studies of child abuse treatment programmes also find an over-representation of women since they tend to be held during the day when fathers are unable to attend (Dougherty 1993). Furthermore:

`women are always implicated because even when men are the culprits, women are usually the primary caretakers who have been, by definition, in some ways unable to protect the children.' (Gordon 1987 p.69)

It is of interest, however, that while the numbers of single parent households headed by women has increased, rates of severe violence in the family are decreasing. Ethier, Palacio-Quinton & Jourdan-Ionescu (1992) argue that three-quarters of `child maltreatment' cases involve neglect rather than violence.

Nonetheless, some mothers are violent towards their children. For Gordon (1987) this means that `child abuse research becomes more interesting and challenging to feminists'. The question becomes `what motivates women's rage and abuse of power'?

Mothers who abuse their children are most often the subject of clinical studies which tend to be `gender-blind' in failing to consider the differences between men and women's experiences within the family (Milner & Crouch 1993). Individual-based studies have not identified a clear psychological profile of parents who maltreat their children (Lenton 1990) and Milner and Crouch (1993) conclude that less than 10% of all child abusers in the US have psychiatric disturbances. The characteristics of child abusers which have been identified in such studies include hyper-reaction to misbehaviour of children (Bradley & Peters 1991) a sense of general incompetence, loss of control over one's life, overall distress and depression, anxiety, loneliness, occasional thought disorders, and an inability to cope with everyday, not to mention extraordinary, demands' (Meier (1995) in Dougherty 1993 p. 102). Dougherty argues that:

`the exhibition of these specific types of behaviours and emotions by women is best understood as the manifestation of a generalized condition among women who conform to the normative expectations that the patriarchy sets for them, rather than as the manifestation of a defective character shared by abusive parents.' (p. 103)

Intergenerational explanations suggest that the majority of women who abuse children are themselves abused, they may recreate a disciplinary style which involves violence, inherit the same social realities such as poverty or isolation which leads to frustration and aggression, or react aggressively to a child's distress which recall's their own (Lenton 1990; Ney 1988). As the discussion in the previous section outlined, however, the path from victim to victimizer is neither straight nor clear. Oldershaw, Walters & Hall (1989 p. 258) cite data that reveal `aggressive, non-compliant tendencies' in abused children, no differences between abused and non-abused children, or find abused children to be more compliant than others.

Cultural explanations suggest that certain cultures may sanction violence towards children. Lenton (1990) reports that some parents often believe they are rightly `disciplining' their children, rather than `abusing' them. In terms of situational factors, DeKeseredy & Hinch (1991) report that Canadian sociologists have spent little time studying child abuse. Studies of the role of stress (Milner & Crouch 1993) socio-economic disadvantage (Ethier et al., 1992-3; Friedrich & Wheeler 1982) isolation and lack of support (Lovell & Hawkins 1988) however, have all provided clear results. Overall, in a thorough review of the social and structural theories of family violence McCall & Shields (1986) conclude that while they may not be able to explain violence, they `cast light on how intrafamily violence is possible'. (p. 99)

Child sexual abuse

Information on child sexual abuse by women is rare, whether based on crime reports, clinic samples or random population surveys although there are indications of an increasing interest in the issue. The Virginia Child Protection Newsletter (VCPN) (1989) reports that 4% of offenders in a Dallas incest treatment programme were female, as well as 4% of perpetrators reported by a random sample of women. Studies which do exist are usually based on very small samples eg. Johnson (1989) or single case studies eg. Higgs, Canavan & Meyer (1992). Knopp and Lackey (1987) collated information from 44 treatment programmes and provide a detailed account of some of the problems of measuring incidence. Finkelhor and Russell (1984) based on two incidence studies estimated that some 14% of sexual abuse of boys was committed by females, and 6% of girls. The definitions used in such studies, however, are often vague or overinclusive. Thus one study reported by Finkelhor and Russell (1984) classified women as perpetrators of sexual abuse if they allowed sexual abuse to happen to their children.

Mathews, Matthews and Speltz (1991) distinguish between women who have been coerced into abuse by males, those `taught' without coercion, and those with a long history of family sexual abuse.

There is some indication that female sex abusers are more likely to have been sexually abused themselves than male abusers. Knopp and Lackey (1987) found that between 93-100% of female perpetrators in treatment had been sexually abused. Johnson (1989) discusses some of the literature in her own study of 13 girls aged 4 to 13 in a programme for sexual perpetrators. She found that all those in her study had been sexually abused, mostly by family members, and came from families with histories of sexual, physical and substance abuse. It is evident, nevertheless, that many women who experienced sexual abuse in childhood do not become abusers. Higgs et al., (1992) suggest this is explained by women's socialization patterns which stress caring and nurturing.

Most reviewers, therefore, point to a constellation of factors associated with violence in the home, including stress, the social learning and family transmission of violence, social isolation, substance abuse, and poverty. A few stress the differential impacts of these factors on women.


B. Offence-based studies

In terms of violent crime, the few studies of women's violence have usually focussed on spectacular events such as murder or infanticide. Others forms of violence such as assaults and robbery, or violence by girls have been largely ignored.

Girls and adolescents

There has been `a virtual absence of studies concerned with the natural expression of aggression by girls' (Campbell 1982 p.138). The few studies of physical bullying in schools (Pulkkinen and Saastamoinen, 1986) suggest that, as with other forms of aggressive behaviour, it is twice as common among boys. Traditionally, delinquency studies have excluded girls. Among those which have compared boys and girls the much lower level of violent behaviour among girls than boys is evident (eg. Gomme 1985 in Simourd and Andrews, 1994; Kirsten 1990).

Campbell (1982, 1986, 1993) has explored the motivation for fights among working class girls. Fighting by adolescent girls she argues has tended to be interpreted as `chaotic activity reflecting individual pathology or inadequate socialization' rather than meaningful behaviour establishing their position in their social order (Campbell, 1982 p.138). In England and Wales she concluded that such fights, as with those among boys, helped to establish and maintain public self-respect and social status. In the US she reports that around 10% of gang members were estimated to be female (1993 p.125) and have always been a feature of gang life. Among girls in gangs in New York she found they were less concerned than boys with the use of aggression and criminal violence for monetary gain. Their use of aggression was to retain status and reputation, and to protect themselves from victimization. The gang offered acceptance and safety. Fear of loosing both, she argues, is transformed into belligerance - a shift from expressive to instrumental aggression:

`Self-control and the containment of anger do nothing to prevent victimization. So they come to see aggression as a way of controlling other people. It becomes instrumental.... The indisputable law on the street is fight or get beaten. The reasoning that leads women to this instrumental use of aggression is not confined to the gang. Wherever women face lives of brutal exploitation that destroys their faith in the value of trust and intimacy, they will be driven to it.' (p. 140)

Such experience from the US should not be seen as universal, however. Reporting on girls in street gangs in Paris, Lagree and Fai (1989) concluded that they occupied a subordinate position in relation to the young men in the gangs, reflecting the sexual divisions of roles in the working class society from which they came. They were not involved in fighting and in fact debarred from taking part.

Recent media attention in the US has focussed on the emergence of `girl gangs' associated with the drug trade and violence (Maher & Curtis, 1995). Chesney-Lind (1992, in Faith 1993) argues, as do Maher and Curtis that their existence has been largely promoted by the media, and that there is little evidence to support a new wave of violent females. While girls do engage in violent acts on the streets, they are more likely to be the targets than initiators.

Infanticide/killing children

As noted earlier the interpretation of gender differences is difficult in this area because women are more likely to have care of children, often as single mothers, and to spend more time with them. In the US Reiss and Roth (1993) report that infants and small children are more likely to be killed by their mothers than their fathers, in part as a result of the mother's greater caretaking role. Child deaths are also likely to result from combinations of circumstances and actors eg. an individual parent, both parents, boy-friends, step parents and grandparents, foster parents and babysitters (Greenland 1987). They may result from a single event or an extended history of battering or neglect. In very rare cases they may be identified with severe pathology (eg. the Munchausen by Proxy Syndrome, Schreier & Libow 1993).

A detailed study of deaths from child abuse and neglect in Canada, the United Kingdom and the USA, and of the many problems of research in this field, was undertaken by Greenland (1987). Of the 100 cases examined in Ontario, he found that slightly more women than men were responsible for a child's death, they tended to be younger than male perpetrators, and the child more likely to die as a result of neglect than abuse. Male perpetrators were more likely to have injured the child physically. In the United Kingdom, among 68 deaths, there was a higher frequency of male perpetrators.

Greenland stresses the variety of circumstances in such deaths and the importance of studying a total population rather than the most extreme cases. In both samples he attributed the largest proportion of deaths to the `battered child syndrome', followed by child neglect and homicide (ie. a single event not related to a history of abuse). In both countries he also identified baby-sitters and temporary carers as a specific group. Some of the factors associated with high risk children and their parents were also identified. He concluded that the proportion of deaths attributable to mental illness was rare, and that there is an indisputable link between child abuse and neglect deaths, and poverty and family stress in all three countries.

Morris and Wilcznski (1993) in their study of mothers who kill their children report that children under one year of age made up 12% of all deaths in England and Wales in 1989. Most of those children were killed by parents. An analysis of all such cases where the suspect was a parent between 1982 and 1989, a total of 493, indicated that almost half of the children were killed by their mothers. As they underline, this is in marked contrast to other types of homicide where women are usually well outnumbered by men.

What is also evident from the work of Wilcznicki and Morris as well as other writers (eg. Allen, 1987a & b) is the differential way in which such men and women were treated by the courts. Of those originally charged with murder, more than half of the fathers were sentenced to imprisonment, compared with under 10% of the mothers. The great majority of those mothers were subsequently convicted on a lesser charge and received probation or (psychiatric) hospital orders. This was generally on the grounds of diminished responsibility (that at the time of the crime they suffering from an abnormality of the mind). Of those cases where the initial charge was manslaughter, just over half the mothers received a sentence of imprisonment, compared with the majority of the fathers. Thus overall, the criminal justice system in England and Wales is less likely to convict mothers who kill their children for murder, and less likely to sentence them to prison. In the USA the authors suggest, such mothers are more likely to receive a sentence of imprisonment.

Those mothers who do receive a prison sentence tend to be seen as `bad' mothers in contrast to otherwise `good' mothers who were seen to be suffering from some form of personality disorder or depressive illness. Morris and Wilczynski conclude that this tendency to see women's violent behaviour as unnatural is not in the end helpful to women. Like Greenland (1987) they argue that the reasons mothers may kill their children are `many and varied', and `normal' women can kill their children when they are confronted by social and economic circumstances which are severe enough' (p. 215). The focus on the pathology of the mother diverts attention away from the poverty and isolation in which such mothers often live and, they argue, their lack of social and economic power in a society which regards all women as natural mothers.

Husain, Anasseril and Harris (1983) in a study of 23 homicidal women admitted for pre-trial psychiatric evaluation found those who had killed a child were much younger than other women. Korbin (1989) in a study of nine women imprisoned for killing their child suggests that the deaths followed a pattern of abuse of the child, that the women had provided warning signals to professionals, family members and neighbours after previous incidents, and had rationalized and minimized the abuse to themselves. Her work confirms that of other researchers in the field in highlighting the `plethora of adverse conditions and risk factors' in the life histories and current circumstances of the women, including their own histories of abuse. On the basis of other work in the field (eg. Daro 1987; Fontana & Alfaro 1987) she suggests that prediction of such fatal incidents may be impossible, but that intervention and education should be directed beyond individual families to community networks which can support them, and research, at the circumstances leading to such events.

Murder and manslaughter

As has already been indicated, the most infrequent violent behaviour by women has received the greatest amount of attention. Over ten years ago in a study of women as victims and perpetrators of homicide, Wilbanks (1982) noted the relative lack of good accounts of murder by women, and the reliance on mythology or sexual steriotypes about women which were generally used to explain such murders. Again, explanations of murder by men he argued, as Allen (1987) has also convincingly shown, have usually been in terms of poor socialization, or economic or financial factors, but for women, in terms of mental instability or pathology.

Wilbanks also noted the problems of interpreting `official' data on murder or manslaughter given that a charge or conviction itself reflects the interpretation of an event which might in other circumstances have been seen as an accident; of the limitations of clinical studies which `explain' individual motivation; and of the limitations of prison-based studies which examine the characteristics of prison populations of women who kill. All such studies, he suggested, have serious biases. The selective processes which result in a charge being laid, the type of charge, a finding of guilt and the type of sentence, all filter out cases which might provide a more balanced picture of women whose actions result in a death.

As has been indicated, it has long been recognized that when women commit murder it usually involves a close partner or relative (Rasche, 1990; d'Orban 1990). Since the early 1980's there has been a considerable increase in studies of homicide by women, most of them American (Husain, Anasseril & Harris 1983; Weisheit 1986; Browne 1986, 1987; Ewing 1986; Goetting 1987, 1988; Walker 1989; Daly and Wilson 1988, 1992; Wilson, Daly and Wright 1990; Block 1990; Jurik and Winn 1990; Bannister 1991; Silverman and Kennedy 1993). The great majority of these accounts, as the discussion of family violence suggests, situate killing by women in the context of their histories of abuse within the family, and as acts of self-defence (Browne & Williams 1989).

Few of these accounts consider murder of a partner which does not follow this pattern. As Dobash and Dobash argue (1994)

`There are of course some exceptions to this pattern...but such cases are relatively rare in comparison to the general pattern...and would not pose a challenge to the dominant statistically established pattern of domestic homicides which are committed in the context of male violence.' (p. 17).

In the USA there have also been a number of studies exploring the apparently high rate of homicides among black women (eg. McClain 1982-83; Mann 1990b) and the links with alcohol or drug use by both women convicted of murder and their partners (Mann 1990a; Goetting 1987; Brownstein et al 1994; Blount et al., 1994). The later suggest that substance abuse by both parties is more likely to be present in situations where women kill an abusive partner, than among abused women who do not.

Few accounts consider murder outside a domestic situation. A number of writers have reported that multiple or serial murder is almost always, although not exclusively, committed by men (Skrapec 1993; Gresswell & Hollin, 1994). Other accounts include discursive studies of notorious cases and their representation by the press and public, including the case of Aileen Warnos characterised as the first female serial killer (Birch 1993; Skrapec 1993).

Canadian studies of murder and manslaughter

In Canada there have been a number of studies of male and female conjugal homicides (eg. Daly & Wilson 1993; Wilson, Daly & Wright 1990; Cote 1991; Silverman and Kennedy, 1993) `documentary' studies of women who kill (eg. Priest, 1992; Vallee, 1986) biographical and autobiographical accounts (Walford, 1987; MacDonald & Gould, 1987) and empirical descriptive studies (Silverman and Kennedy, 1987, 1988, 1993; Nouwens, 1991; Moyer, 1992). Moyer, Nouwens and Silverman and Kennedy have also documented differences between rates of homicide by Aboriginal and non-aboriginal women, and the links between murder and alcohol or drug abuse.

Silverman and Kennedy (1993) using data compiled from police reports of murder for 1961-1990 reports that `the trend in rates of killing by women...has been very stable since the early 1970's' (p. 146) while that by men has risen considerably. Secondly, they confirm that most homicides by women take place within a domestic context, 75% of female offenders kill members of their own family, 40% a partner, and 22% a child. They tend to be relatively young, and two-thirds of them married or cohabiting. They almost all kill alone (89%) and only one victim (95%). Two-thirds are white (65%) but 28% Aboriginal.

This overrepresentation of Aboriginal women (and men) is explored further by Sharon Moyer (1992) in a comparative study of murder and manslaughter by Aboriginals and non-Aboriginals between 1962 and 1984. Using police reports of incidents she found that 30% of all female suspects were Aboriginal compared with 18% of male suspects. In addition, there had been an increase in the proportion of Aboriginal women suspects between 1962 and 1984 from 12% to 22%, while among non-Aboriginal women the rates increased little (from 10% to 12%). Police reports also indicated that Aboriginal women were less likely to kill someone in a domestic situation than non-Aboriginal women (57% compared with 72%). Among young offenders (or those under 18 years) there were similarly greater increases in the involvement of Aboriginal girls over the twenty year period than non-Aboriginals, although it is important to stress that the total number of female young offenders involved was very small (125).

While information on sentencing in Canada is very limited, Moyer suggests some important differences in how the courts dealt with Aboriginal and non-Aboriginal women. An analysis of court data for 1976 and 1980 indicated that non-Aboriginal women were more likely to be found unfit to stand trial than Aboriginal women, and 21% of the former were found not guilty by reason of insanity compared with 2% of Aboriginals. On the other hand, when both groups of women were sentenced, Aboriginal women were more likely to be convicted of manslaughter and given shorter sentences than non-Aboriginal women. The latter were more often given life sentences for first or second degree murder. Nevertheless, over half of the women in both groups received either a non-custodial sentence or a provincial sentence under two years.

Similar patterns of conviction and sentence were noted by Nouwens (1991) in a review of all women receiving federal sentences between 1971 and 1990 for murder or manslaughter. While this review excluded women who had not received a federal sentence, of the 293 women in the sample, 61% were Caucasian, 34% Aboriginal (including Metis and Inuit) and the remaining 5% other races. She found that 91% of Aboriginal women had served sentences for manslaughter rather than murder, compared with 64% of the non-Aboriginal women. Their sentences were also shorter, an average of 3.7 years compared with 5.7 for the non-Aboriginal women, and given that few had been convicted of first or second degree murder, very few received life sentences. The Aboriginal women also tended to be younger than the rest, and more likely to be serving their first federal term.

As Moyer (1992) has pointed out, however, such descriptive material cannot take account of the social contexts in which the events took place, nor the social and economic changes which have affected the lives of Aboriginal women over the past 20 or 30 years.

On the basis of a file review of 85 women serving federal sentences for murder or manslaughter in 1989, the variety of immediate circumstances is evident (Shaw 1992). Some had killed an abusive partner, or an acquaintance after sexual advances; some a lover or a member of a lover's family; some had killed someone in a very complex situation after days of drinking and drug use, and often had no recollection of the event; some had killed clients or acquaintances such a `Johns' or drug dealers; a few had killed a child, although not always their own; four women had been convicted of murder as a partner in a robbery.

Walford (1987) and Priest (1992) provide biographical accounts of women convicted of murder or manslaughter in Canada. They too underline the variety of circumstances and chance which are involved, illustrating that the act may be `accidental' or planned, the outcome of a long history of abuse, or a response to overwhelming frustration. Priest, however, still feels compelled to demonize and differentiate the `genuine killers' from the rest. Overall, much of the literature on women committing murder or manslaughter points, as Faith (1993) underlines, to the ordinariness of most women who do so, not to their extraordinary and abnormal characteristics.

Assaults and robbery

Pat Carlen (1988) is one the few writers who has not shied away from considering women's potential for violence, but does so in the context of their experiences of growing up as women. In her study of the criminal careers of 39 working class women in England and Wales she found that they:

`had between them committed the whole gamut of more serious crimes including `granny bashing', `baby bashing', `mugging', armed robbery, arson...'(p.17)

This is not surprising given that she had specifically selected the sample to enable her to look at the lives of long-term offenders. A number of them had been members of mixed hell-raising gangs as teenagers, where fighting was just one source of excitement. They had lived in predominantly poor areas where crime and fighting was a way of life. Twenty two of the 39 women had been placed in children's homes `in Care' for behaviour or family reasons rather than breaking the law. They had chosen to offend in response to finding themselves exploited because of their working-class origins, their unwillingness to conform to gender expectations, and in some cases their race. She argues that their experience in Care in particular had had very negative consequences.

Fourteen of the 39 women had convictions for violence against the person (although only three for robbery). Asked how they had learnt to fight, most said they had been forced to fight or threathen violence in order to survive family life, street life, Care, welfare surveillance, Borstal (youth custody), police or prison. They stressed they had only used violence as adults when they had felt threathened.

Warren & Rosenbaum (1986) in a follow-up study of Californian women committed for status and offence behaviour as adolescents found that all had offended as adults, including 35% arrested for a violent offences. On the other hand two thirds of those arrested for violence as juveniles had no further arrests for violence as adults. More recent concerns in the US have centred on increases in violence by male and female cocaine users. Maher & Curtis (1995) review some of this research in their study of women and crack cocaine in New York and cite Inciardi (1986) as claiming that the `new female criminal' had arrived. This is countered by other research (eg. Goldstein et al., 1991) indicating that among male users there is an increase in the use of violence, but among female users an increase in victimization. Mayer and Curtis confirm Goldstein's findings and conclude that there is little evidence of the violent female urban `Gansta':

`It is the everyday contexts in which these women work that provide the high propensity for violence, not the women themselves.' (p. 161).

Very little research has been completed on women charged with robbery. Immarigeon and Chesney-Lind (1992) report (although based on studies up to 1977) that women's role in robbery is less active than men's and they are not usually the initiators. A more recent study by Sommers and Baskin (1991) reported in Maher & Curtis (1995) suggests that drugs and a prior history of prostitution are important factors in robberies by women in the US.

In Canada, Loucks and Zamble (1994) in a selected study reported higher proportions of women serving a federal sentence for assault than men, and lower levels for robbery. Only two studies have considered robbery specifically. Girouard (1988) looked at the offence characteristics of all 13 women serving sentences for robbery in Quebec in 1985. He concluded that while vastly outnumbered by the numbers of men committed for robbery, these women's involvement was generally as serious as that of men, requiring various combinations of planning, intimidation and aggression. Over three-quarters acted with a male partner but did not necessarily play a peripheral role.

In the second, and broader study, Savard and Biron (1986) interviewed 19 women convicted of non-domestic violent crimes in Quebec. They excluded offences involving a family member or cases indicating mental disturbance. Eleven had been convicted for armed robbery, five manslaughter, two attempted murder and two kidnapping, but they stress that such violent offences were the exception in their patterns of offending behaviour. Even among those with considerable offence histories, the use of violence was sporadic and rare. Evidence of violence and sexual abuse in their childhood, running away from home and surviving on the street through drugs, theft and prostitution was much more common.


C. Institutional studies

A few studies based on populations of offenders in institutions have included consideration of violent women. They have included basic descriptive studies of female prison populations (Robertson, Bankier & Swartz 1987; Long, Sultan, Keifer and Schrum 1984; Shaw 1992, 1994); differences between non-violent and violent female offenders (Balthazar and Cook 1984); comparisons of male and female serious offender populations (Loucks and Zamble 1994); surveys of psychiatric diagnoses in male and female prison populations (Maden, Swinton & Gunn 1994) and of psychiatric diagnosis and violence (Brownstone and Swaminath 1989).

Balthazar and Cook (1984) could find no factors differentiating women sentenced for violent and non-violent offences. The most detailed accounts tend to be those based on psychiatric assessments, and using standardized psychological or psychiatric measures. In general they focus on current or past diagnosis of pathology and on individual demographic factors which are not placed in the context of the lives of the women. Many focus on the long-standing finding of higher levels of psychiatric disorders among women compared with men in institutional populations.

One of the most recent and detailed accounts by Maden et al (1994a 1994b) is based on samples of male and female prison populations in England and Wales. They recognize some of the problems of studies based on institutional populations including the differential processing of men and women through the courts which influences gender differences in prison populations, that cross-sectional studies are biased towards those serving long sentences, and that individual prison staff can influence levels of recorded diagnoses. They noted a higher prevalence of personality disorder, neurosis, mental handicap, drug abuse and self-harm among women than men, but stress that not all female prisoners were more disordered than male prisoners, nor that they all had a higher need for psychiatric treatment. Their only finding in relation to violent offending is its relationship to self-harm (suicide attempts as well as cutting).

A study of women admitted to Winnepeg Remand Centre (Robertson et al., 1987) fails to consider race as a variable when some two thirds of the female prison population in that province are likely to be Aboriginal (Comack, 1993). Their findings in relation to violent offences included an association with drug and alcohol abuse, a past history of violence, a family history of crime, alcoholism and physical abuse, low education and socio-economic status, and childhood disruption.

Brownstone and Swaminath (1989) undertook a retrospective study of female offenders admitted to psychiatric hospital for psychiatric assessment or under a Warrant of the Lieutenant Governor. Over half the sample had been charged with a violent offence. They found that women with longer criminal histories were not more violent than others, and no relationship between type of crime and specific diagnosis. They recognized their sample as unrepresentative, and suggested that violence may be too complex an event to predict on the basis of demographic variables. They suggest the need to examine the history of sexual and physical abuse, as well as family history and other factors in relationship to criminality.

Loucks and Zamble (1994) report the preliminary findings of a study comparing selected male and female federally sentenced offenders, convicted of `serious crimes'. They do not indicate how they selected their sample nor defined `serious', and note that the women included were not representative of female offenders as a whole. The main differences noted were in the much higher incidence of previous suicide attempts, depression and drug abuse among the women than the men.

On the basis of her study of 30 violent female offenders at the Prison For Women, Julie Darke (1987) outlined the prolonged and severe physical and sexual abuse they experienced as they grew up, much of it by family members or friends. Mostly, but not always, the assailants were men, and the abuse continued for periods of up to 15 years. The women's response included considerable anger and lack of trust in people, especially men. They tended to blame themselves for their past abuse, and felt out of control and powerless.

`Given the justifiable reasons for anger, it should come as no surprise that 73% of these women have problems with anger control and personal aggression. They have periods of explosions. The tendency of women on the street is to suppress anger until there is an event which triggers some kind of lashing out or explosion. Some may allow themselves to express their anger only under the influence of drugs and alcohol.'(Ibid. p.143)

In prison they appeared passive and withdrawn much of the time. The majority had histories of depression, attempted suicide and slashing, and substance abuse.

`To alleviate feelings of powerlessness, some women take on the facade of being aggressive and invulnerable. This facade is traditionally male in a lot of ways. It may be seen as the only alternative for some of these women breaking from a stereotypically passive and vulnerable female role.' (Ibid. p.144)

Violence in institutions

It is important to separate a conviction for a violent offence from the use of aggressive or violent behaviour in prison. It is recognized that offenders convicted for murder, for example, and those serving long sentences are less likely to be charged with discipline offences in prison than others (see discussion in Shaw 1991a; Faith 1993).

Most studies of institutional violence, whether general hospitals, psychiatric hospitals or correctional institutions, underline the fact that serious assaults are rare, and are usually committed by a small minority of individuals (Rice, Harris, Varney and Quinsey 1989; Zamble & Porporino 1988; Davies 1991). They tend to be young and male and in psychiatric hospitals, to have `low coping skills', be likely to self-injure or attempt suicide, and to be withdrawn or depressed. Rice et al., (1989) on the basis of 15 years study of institutional violence point out that the great majority of the literature has focussed on the individual characteristics of assaultive individuals. Only more recently have situational or structural explanations been considered.

In a recent review Davies (1991) similarly argues that violence in institutions is an interactive phenomenon. Overcrowding, provocation by staff or other inmates, staff inexperience or tolerance of violence are all situational factors which interact with individual and systemic factors. Both Rice et al., and Davies underline the fact that staff are often unaware of triggering an event, while patients claimed they were provoked. Assaults were often associated with staff demands for action, refusal of requests, or the imposition of sanctions. They concluded that assaults resulted from an interaction between environmental and internal factors and could not be explained by patient pathology.

On the basis of their findings, Rice and her colleagues developed a staff training programme to develop skills in handling patients and situations using non-restrictive, nonauthoritarian and nonprovocative ways of interacting with them, and the indentification of situational factors associated with assaults.

Violence by women in prison

As Axon notes (1989) many female offenders come from very abusive backgrounds in which they were either victimized and/or do not learn appropriate ways of handling violence. `Many women suffer from unrecognized and unexpressed grief and rage which may escape their control. In prison, living conditions are stressful and may trigger underlying fears and emotions.' (p. 53)

Although in some countries disciplinary charges in men's and women's prisons may be similar, or even higher for women, there is considerable evidence that women are charged for more trivial behaviour than men and much less serious violence (Lindquist 1980; Hattem 1984; Nesbitt & Argento 1984; Mandaraka-Sheppard 1986; Shaw 1991a; Kersten 1990). Higher rates of discipline charges are found in institutions with higher security, and in young offender institutions, and among young single women. Women are more likely than men to act alone rather than in a group. Women serving shorter sentences tend to be charged more than long-termers. Lindquist (1980) found that both men and women with a history of juvenile incarceration were more likely to be charged than others.

There is also evidence that, as in men's institutions, women sentenced for crimes against children are more likely to be the targets of attack by other inmates (Carlen 1985; Faith 1993).

Comparing institutions for male and female adolescents Kersten (1990) found major differences in the extent and type of aggressive and violent behaviour. Girls were more likely to self injure or attempt suicide, and less likely to fight than boys. There were also clear expectations about `feminine' and `masculine' behaviour which were penalized in the case of the girls.

In the only detailed comparative study of discipline in women's prisons, Mandaraka-Sheppard (1986) demonstrated that organizational practices including punishment methods, the quality of inmate/staff relations, staff age and experience, a perceived lack of autonomy, and lack of incentives to good behaviour were the main factors explaining discipline charges, and not age, offence histories or a history of violent behaviour.

Her study, based on prisons in England and Wales, highlights the importance of inmates' perceptions of events and the dynamics of interaction in the prison setting. Inmates saw boredom and provocation as the main factor leading to fights and arguments. The witholding of their rights, favouritism, unfairness or victimization were seen as the main source of confrontation with staff. Physical violence was significantly higher in high security prisons, individual rather than collective, and involved younger women who felt the institution to be a hostile environment.

`Inmates who responded defiantly, and whose behaviour took the form of physical violence, were more likely to interpret the actions of others in the prison environment (either staff or inmates) as hostile and threathening to their autonomy, not withstanding the degree of their rated potency and previous violent criminal record.'(Ibid. p.203)

But the comparison between institutions made it clear that not all institutions induced the same responses from such women. Moreover, Mandaraka-Sheppard stresses the adverse consequences of labelling women as manipulative, violent or dangerous which set up staff expectations, encouraged the hostile interpretation of actions and induced resistence:

`Physical violence was more likely to be a result of harsh institutional practices, which induced defiant responses on the part of inmates, which in turn shaped the form of overt behaviour resulting in more punishment'(Ibid. p.203)

A number of writers have noted the tendancy to medicalize women's violence in institutions, and the use of drugs to deal with women perceived to be violent (Mandaraka-Sheppard 1986; Sim 1990; Hattem 1992).

Self injury and suicide

Many studies of women indicate that violence is more likely to be inflicted on themselves, in terms of suicide attempts or slashing, than other people, and that this is particularly true of women in institutions. Kersten (1990) found a higher incidence of suicide and self-injury in institutions for female adolescents in Germany and Australia than boys' institutions. Loucks and Zamble (1994) report higher rates among federally sentenced women (48% a prior suicide attempt) than federally sentenced men (13% a prior suicide attempt). Darke (1987) reported a history of suicide and slashing among violent offenders at the Prison for Women. Similar findings are reported by Maden, Swinton and Gunn (1994b) among male and female prisoners in England and Wales (32% of the women reported prior self-injury or suicide attempts and 17% of the men). They also confirmed other findings that self-injury was also higher among women sentenced for a violent offence than others (Cookson 1977; Wilkins and Coid 1991).

Grossman (1992) considers the disproportionate incidence of suicide among Aboriginal women in the federal population, and stresses the importance of recognizing the impact of the institution as well as the history of violence and discrimination experienced by Aboriginal women. Jan Heney (1990) has stressed the high level of childhood sexual abuse and physical abuse among women at the Prison for Women who self-injure or attempt suicide.


This section has reviewed three areas of work on women's violence: violence in the home; offence-based studies; institutional studies. The literature illustrates the considerable disputes and differences in approach and methods used, and the lack of consensus over definitions and measurement.

Much of the literature relating to women's violence in the home is concerned with arguments over the extent of such violence and whether male and female actions can be equated. Most reviewers suggest a constellation of factors associated with such violence, including social learning and the family transmission of violence, stress, poverty, social isolation, and substance abuse. The differential experiences of women and men in the family in terms of their socialization, their use of anger and aggression, and the unequal distribution of power and status are seen as the most fruitful ways for understanding such violence and developing prevention and intervention programmes.

Offence-based studies provide very patchy information about women's violence reflecting the arbitrary nature of many categories. Most of the literature focusses on the least frequent offence of murder, and in terms of its links to family violence. The differential response of justice systems to men and women's violence is evident. What little work exists on adolescents, and on assault and robbery by women suggests considerable histories of abuse, family disruption, institutional care, substance abuse and street crime.

Institutional studies which consider women's violence often include demographic and descriptive surveys using standarized psychiatric and psychological measures. They do not take account of context of women's violence, or the problems of pathologizing women's behaviour. Disruptive family backgrounds and abuse histories, substance abuse, slashing and suicide attempts would appear to be common among women who use violence more than others.

Finally, studies of violence in institutions suggest that the differences between men and women's behaviour in institutions is in keeping with accounts of the differential socialization of men and women in the use of anger and aggression. The factors associated with the production of disciplinary events including management and organizational practices, would appear to be similar in both men's and women's institutions. While there are many problems in comparing violence and discipline in male and female institutions, size, management style, inmate-staff relations and staff training would all appear to be crucial factors. In women's institutions the differences in women's use and expression of anger and aggression, and the severe consequences of past experiences of abuse particularly among those likely to use violence or self-injure in prison, reinforce the crucial importance of staff training and alternative responses, apart from any programmes developed for the women themselves.