People's Equality Network - January 1995
Domestic Violence - Recent Statistics in Victoria
Domestic violence is a disturbing cultural problem, and a highly emotive issue for everyone. In most public presentations of the subject, it is portrayed primarily in gender terms: that of male violence against women. One major report - the National Strategy on Violence Against Women - has gone so far as to depict domestic violence as 'a specific women-only problem', and to call for the term 'domestic violence' to be replaced exclusively with the term 'violence against women'. Such an attitude is likely to be counter-productive, because it fails to address - in effect explicitly refuses to address - the equally real problem of violence by women.
Violence by women is a problem which our society has consistently failed to address, usually on the grounds that it is and always has been statistically insignificant, especially by comparison with men's violence against women. A recent report on domestic violence published by VicHealth? (Victorian Health Promotion Foundation) states that the relative injury ratios as a result of inter-partner physical assaults are approximately 5:1 female to male. Such a ratio would indicate that although physical violence by women would seem to be far less common than men's, it would have to be considered 'statistically significant', and realistic allowances made for it within the development of any meaningful strategy on domestic violence.
Even this 5:1 ratio, however, mispresents the problem, because the methodology used to derive it has significant flaws: it does not allow for known sociological and other factors, and detailed analysis was applied solely to female injury-cases. For comparable injuries, the female:male injury ratios reported in the study are less than 2 to 1, a figure consistent with certain US studies; and the report explicitly states that injuries to males "heavily outweigh" those to females under the definitions used for 'probable' and 'suggestive' cases of domestic violence. Other available evidence strongly suggests that women initiate physical assault far more often than men: violence by women is by no means 'statistically insignificant', and may be as significant a problem as violence by males. This fact may be unpalatable to many, and may even be described as 'anti-feminist', but it is a fact which must now be addressed.
Violence by men against women is recognised as a significant problem; violence by women is not. Present 'solutions' such as rigid criminalisation have demonstrably failed to solve the problems of male violence; they are likely to be even less effective, and even less acceptable, for tackling female domestic violence - especially female violence against children. A new approach is required which addresses domestic violence not as a gendered problem, but as a human problem with gendered overtones: a problem for which we are all, collectively, responsible.
Data source and content
The data used in this report were derived primarily from the VicHealth?/Victorian Injury Surveillance System/Monash University Accident Research Centre report "Domestic Violence", prepared by Virginia Routley and Jenny Sherrard, and published in 'Hazard', edition no. 21, December 1994. This report emphasises particularly the statistics relating to sex-comparisons. The VISS study does not give any information to distinguish between inter-gender (heterosexual) and intra-gender (homosexual) domestic violence: although the former appears to be assumed throughout, the latter - particularly female-to-female - are also known to be significant social problems.
The VISS study provides factual data on incidents resulting in hospital attendance, extracted from medical records of five major Victorian hospitals over approximately two years. The definition of domestic violence used in the study is "direct partner-inflicted violence, resulting in injury, to those aged 15 years and over, occurring both within and outside the home"; it does not include violence against children. The primary model used for analysis was that developed by Stark and Flitcraft in the USA, which had, however, only been based on studies of female injury. Stark and Flitcraft's model is gender-specific, and concerned only with 'wife abuse'; it interprets domestic violence as an inherent by-product of a male-dominated 'patriarchy' (Stark, Flitcraft and Frazier, 'Medicine and Patriarchal Violence', in "International Journal of Health Services 9", 1979). In the VISS study, however, an attempt has been made to extend this model to adult domestic violence in general: incidents involving injuries to both females and males have been included.
The study's authors state that "On the VISS database there were estimated to be 288 positive, 397 probable and 313 suggested cases of domestic violence, representing 2% of all adult injury cases." Of the 288 'positive' cases - those in which domestic violence was explicitly involved - 239 were female, and 49 were male. The ratio of injuries in this category was therefore 4.9:1 female:male (rounded upward in the text to 5:1). Male injuries tended to be significantly more severe: 19 females and 12 males required hospital admission, a ratio of approximately 1.6:1. This figure of 31 admission cases represents approximately 0.05% of the total of all cases of injuries, from all causes, represented in the study's database. (These figures are derived from Table 4 in the study; the text states that 43 cases were admitted, 72 percent of them female, but this appears to be derived from a simple transcription error, taking all 31 cases to be female, and then adding on the 12 male cases separately. The study contains a number of similar inconsistencies between the text and the tables; where possible, the data from the tables have been used in this report.) Female injuries tended to be bruising (40%), inflammations or pain (17%), and laceration (16%); male injuries tended to be lacerations or abrasions (54%), bruising (14%) and puncture wounds (11%). Knives were used by women against men (15 cases) more than twice as often as the reverse (7 cases).
The study contains conflicting data concerning 'probable' domestic violence incidents. Table 6 in the study's report in 'Hazard' indicates an estimated 309 female cases and 93 male, giving a total of 402 cases and a ratio of approximately 3.3:1. The text, however, refers to a total estimated number of 397 cases, 344 female (265 from extrapolation of a sample of female descriptions, 77 from cases where the word 'domestic' had occurred, and 2 where the word 'claimed' had occurred) and 53 male, giving an apparent ratio of 6.5:1.
However, the methodology used to estimate the male figure for 'probable' cases is somewhat unusual, and is significantly different to that used to estimate the female figure. The statistics from the database itself, according to Stark and Flitcraft's definition of 'probable' as "where the injuries in the person are the result of an assault but the injuries were not sustained in a street assault, mugging or robbery", indicate that injuries to males "heavily outweigh" those to females. Yet unlike the detailed analysis of medical records from which the female figure was derived, no analysis was undertaken to establish the estimated figure of 53 males in the 'probable' category: this figure is based solely on applying the 5:1 ratio derived from the 'positive'-category cases to the figure extrapolated for the sample group of female reports (53 being one-fifth of 265). This estimate of 53 male cases would therefore appear to be unreliable; and without equivalent analysis and allowance for sociological factors, it may be incorrect by as much as several orders of magnitude.
The estimated figure of 313 'suggestive' cases - "those cases whose explanations did not account for or were inconsistent with the injury sustained" was derived from a similarly inconsistent methodology. A detailed analysis of the types of injuries and age-groups was used to derive a figure of 261 cases from the 2531 women in the database who had injuries "typical of domestic violence" (and incidentally indicating that an over-estimate of cases by a factor of almost ten times could occur by mistakenly assuming that all 'typical' cases were in fact derived from domestic incidents). However, the study's authors stated that "it was impractical to apply the same methodology to men, lacerations being their most common domestic violence [injury] and overall injury"; this was apparently considered sufficient reason to do no analysis on male injuries. Instead, the 5:1 ratio derived from the 'positive' category was again applied, giving 52 estimated cases. Once again, this latter figure cannot be considered reliable, and may not be in any way valid.
The study's figures for suicide and intentional self-harm also show a sex disparity, of 1254 female cases compared to 1056 male. Approximately 17% of the female cases were attributed by the study to domestic violence or related issues, compared to approximately 8.5% of the male cases, a ratio of 2:1. No figures were provided of comparisons between attempted and completed suicides: other published figures, for example, indicate an opposite ratio of up to 1:7 female:male for suicide attempts resulting in death (Institute of Family Studies, 1988), but no meaningful comparisons can be made on the basis of the data available in the VISS study.
In total, an estimated 2% of all injuries presenting at the hospital emergency departments were attributed to 'positive', 'probable' and 'suggestive' cases of domestic violence. This is roughly one-tenth, or less, of the figures derived from some female-only studies, such Stark and Flitcraft's, in the USA. Despite the reservations of the VISS study's authors, it is possible that this 2% figure is more realistic than higher figures given in the American studies, since it is known that some of the American studies have been severely misrepresented in the literature: for example, a 22% figure for the proportion of domestic violence injuries in the emergency-room case-load, routinely quoted as typical, was derived from a single study of one highly atypical department in an urban black area in Detroit ('Domestic Violence Victims in the Emergency Department', in "Journal of the American Medical Association", 1984); some commentators on that study have assumed that all injured parties were female, when in fact 38% of the cases were male - an injury ratio of approximately 1.6:1 female:male.
Clinical and other follow-up checks in the VISS study were applied to a selected group of 'positive' female cases and a 'control' group, also all female. The follow-ups indicated a high degree of repeat incidents: domestic violence tended to occur around specific people, rather than in the population at large, although approximately 20% of the 'control' group had medical histories implying domestic violence at some stage of their lives. No equivalent follow-up checks or analysis were applied to any male cases.
The VISS study notes that deaths reported from the Coroner's Facilitation System showed a far greater disparity between males and females: in the two years represented in the study, two males were reported as having been killed by their partners, and 28 females: a ratio of 1:14. This difference may, however, be in part an artefact of the notation used by the Coroner's system: the definition of 'domestic violence' used by the Coroner's office was not specified, and during in those two years, significantly more than two males were reported in newspapers as having been killed in Victoria by their partners under conditions which would match the definition of 'domestic violence' used in the VISS study. Despite this, there should still be no doubt that there is known to be a significant disparity: more females are killed each year by their partners than is the reverse. From the figures given in the VISS study, the average risk for females in Victoria for death resulting from adult domestic violence would appear to be approximately 4 per million per annum, and for males less than 1 per million - although the overall male risk of death by assault is similar to that for females, at approximately 18 per million, and death by injury almost three times higher than that for females, at approximately 250 per million.
The report in 'Hazard' was a summary derived from a larger 140-page report available only to 'persons with a professional interest in this area', and was not available to us. Without access to the original source data, a full analysis is not possible: this report is therefore necessary limited to the sometimes conflicting data and statements in the shorter study-report provided to us.
The primary concerns of the following analysis are psychological, educational and political rather than clinical, in the sense that adequate preventive measures are preferable to increased medical requirements after the event. From all these perspectives, accurate data on sex-differences in domestic violence are important: serious strategic errors are likely to be made if the data provided are significantly incomplete, erroneous or misleading.
The VISS study is more useful than most, in that a serious attempt is made to gather data on both females and males, rather than females only. If data are gathered only on females, from a perspective which assumes that only females are affected, the risk of circularity - circular reasoning - is extremely high. Such studies may be politically expedient, but from a clinical point of view create more problems than they solve. The VISS study, for example, states, in its Summary, that "injuries in females presenting to emergency departments were 10 times more likely than those in males to be due to domestic violence": the statement is basically valid, yet misleading, because males are more than twice as likely as females to present at hospital with injuries of all kinds. The statement is useful for indicating the need for improved diagnostic and referral procedures for women; but it has the side-effect of minimising and trivialising the scale of domestic-violence injuries to males, discouraging the development of improved diagnostic and referral procedures for men. Similar problems occur throughout the VISS study.
The methodology used for the analysis of female cases of injuries appears to be adequate: despite its inherent gender bias, the model of 'positive', 'probable' and 'suggestive' appears to be sound, and is backed by specific analysis in each category. The only specific methodological doubt relates to the prevalence of violence during pregnancy, which was highlighted in the study, but which, as Straus, Gelles and Steinmetz in the USA have demonstrated, is more likely to be a side-effect of prevalence in the 20-29-year age-group.
The methodology used for analysis of male injuries appears to be far less sound, is not backed by any analysis, and in each case would tend to underestimate rather than overestimate the proportion of male injuries attributable to domestic violence. This may be in part the result of using Stark and Flitcraft's 'female-only' model, but several important gender-specific factors have not been allowed for in either the data-gathering or the analysis in the VISS study.
First, in deriving estimates of 'probable' and 'suggestive' categories of male injury, no account appears to have been taken of the considerable sociological pressures against men describing any injury as being derived from domestic assault, compared to the considerable legal, cultural and other support for women to do so in their case. For example, in a recent incident (not reported in the VISS study), a police sergeant mocked a man for "being a bit short on testosterone" for not hitting back at an assailant, but then stated the man would have been charged with assault if he had hit back - a 'double-bind' situation which would be most unlikely to be applied to a woman. These differences in social pressures would appear to be reflected in the disparity between the serious injury ratio (at about 9:5 female:male), the nominal injury ratio (at around 9:2 female:male) and the ratio of reporting of incidents to police (at about 9:1 female:male).
Second, no account appears to have been taken of the cultural expectations on men to accept considerably higher degrees of physical damage than women, and to only present for treatment where such damage cannot be left to heal on its own. This is consistent with the VISS data, in that in the 'positive' category far more women (96 cases, or 40%) than men (7 cases, or 14%) presented with bruising, compared to the far smaller proportion of women (26%, or 62 cases) than men (70%, or 35 cases) presenting with open wounds. These last data indicate a female:male ratio of 1.8:1, similar to the severe-injury ratio of 1.6:1 - or approximately 3 times lower than the nominal ratio of 5:1.
Third, the use of a direct extrapolation of the 5:1 ratio to derive the male figures for the 'probable' and 'suggestive' categories, in place of any analysis, cannot be considered as anything better than guesswork - especially against the direct evidence that according to the definitions used, injuries to males would appear to "heavily outweigh" those to females in both these categories. In the case of the 'probable' category, the justification for rejecting the need for further analysis on the basis that the actual data was "inconsistent with the male to female ratio for the positive domestic violence cases" appears to be based on circular reasoning and inadequate allowance for the sociological factors outlined above. In the case of the 'suggestive' category, the statement that "it was impractical to apply the same methodology to men, lacerations being both their most common domestic violence and overall injury" does not seem sufficient excuse to substitute a guesswork figure in place of any analysis whatsoever - especially as this figure is presented as having the same factual weight as for the female cases. The absence of follow-ups or controls on any male cases has further compounded the problem.
The study does give sufficient information to be able to establish the prevalence of domestic violence against women in Victoria. This appears to account for approximately 5% of all injuries presenting to hospital by women, and approximately 1.5% of the total injury caseload. The 'control' figure of 20 percent of women with a history of at least one incident of any kind of physical domestic violence in their adult lives - i.e. a lifetime risk - is consistent with several other studies in Australia, and also in Canada, Great Britain and the USA. This is a significant problem for our society: the recommendations made in the report for further data-gathering and, especially, more effective referral procedures would appear to be justified. These figures do not, however, support claims - such as those made by activist groups such as Men Against Sexual Assault, or the National Committee on Violence Against Women - of a domestic-violence problem of 'epidemic' proportions, regularly affecting 'one in three women', or 'one out of ten Australian homes each day': such claims would now appear to be little more than political propaganda, and do not help to clarify any of the issues involved.
No information is given on intra-gender violence against women (i.e. in lesbian relationships), although this is now acknowledged as a serious problem in the lesbian community. Certain political groups have been keen to promote the notion that domestic violence is always and only male abuse, but this is a dangerous fallacy which is increasingly creating difficulty for those working with abusers and victims alike. Because of political problems, accurate data have not been easy to collect, but Martin Hiraga of the (US) National Gay and Lesbian Taskforce has stated that "all the available evidence indicates that [lesbian domestic violence] occurs no less and no more than in heterosexual relationships". 'Anita', a spokeswoman for the Domestic Violence and Incest Resource Centre in Melbourne, which now runs programs for victims of lesbian domestic violence, stated that their work is greatly complicated by "a myth that violence against women is committed exclusively by men". She also expressed concern about the minimising of non-physical assault as 'ordinary relationship problems': "a relationship can be extraordinarily abusive without the violent partner laying a finger on her partner. This type of abuse can be the hardest to put a stop to because it is so hard to explain" (references from Latrobe University women-student's magazine "Rebellious", November 1993, confirmed by telephone interview). Realistic, rather than heavily-politicised, attitudes to female violence should help to improve these problems, but accurate data are required; reliable yet politically acceptable methodologies to gather them have yet to be developed.
The VISS study does not give sufficient information to be able to establish the prevalence of domestic violence against males; and hence makes it difficult to derive accurate sex differences. No 'control' group of males was assessed for long-term risk, and no lifetime-risk figure, for comparison with the women's figure of 20%, appear to be available elsewhere; hence no reliable sex-difference ratio for lifetime risk can be derived. However, sufficient data is provided to refute the common claim - again, from groups such as the National Committee on Violence Against Women, and Men Against Sexual Assault - that female violence against males is 'statistically insignificant'. The VISS study's figures, combined with the flaws in the study's methodology, suggest that the 5:1 female:male ratio is a maximum; for comparable injuries, the study's figures indicate a female:male injury ratio of less than 2:1 - consistent with the Detroit figure of 1.6:1, and a similar ratio from the Australian Bureau of Statistics - and possibly considerably less once sociological differences are taken into account. Other evidence (most notably Straus and Gelles' US National Family Violence Survey, 1975 and 1985) even suggests a ratio beyond parity, with females indicating that in their own experience they were significantly more likely to assault their partners than was the reverse - though they would probably cause somewhat less damage in doing so.
To place domestic violence in a wider perspective, the VISS study indicates that, averaged over the two-year lifetime of the study, and assuming a client population base of approximately one million for the five major hospitals involved, some 2.5% of the total population each year received injuries requiring hospital attention. Roughly 2% of this (0.05% of the population base) received injuries attributed in the study to domestic violence. Of these, 15 (or 0.0015% of the population base) received injuries requiring admission to hospital: rather more than one-third of these were male. Out of the same population base, about 2 women (or 0.0002%) were murdered by male partners; 50 other women, and 150 men, died of other injuries (primarily road accidents and industrial injuries).
These figures also need to be understood within a wider context of family violence: the definition of 'domestic violence' as "direct partner-inflicted violence" between adults artificially restricts the scope of the term, and may easily lead to misperceptions of the overall problems. For example, the group most at risk of murder is not adult women, or even young males, but babies and infants: a baby stands a far greater risk - at about 20 per million (1989 figures) - of being murdered by its mother than the mother has of being murdered by her partner. Adult-initiated family-violence injuries to children are also known to occur at significantly higher rates than between adults, with women predominating as the abusers - particularly against younger children. Although this is outside the immediate context of the VISS study, there appears to be a lack of political will in our society to face the implications of this aspect of family violence, or even to encourage accurate data-gathering on the problem.
The figure of 0.05% for inter-partner domestic-violence injuries is roughly consistent with the Australian Bureau of Statistics' figure of 0.6% of households suffering 'significant' abuse per annum (twelve times higher, but including child abuse, and adult abuse not requiring treatment, or treated at a general practitioner's office), but is not consistent with the Office of Status of Women's claimed figure (for females only) of 30%: the latter is known to have been derived from inadequate methodology (extrapolating the annual risk figure of 0.6 percent as a 50-year lifetime risk of 30 percent, then depicting this lifetime risk-figure as an ongoing daily incidence), and should probably be discounted as erroneous.
In general, data collected on female injuries from adult domestic violence, such as by the VISS study, appear to be adequate to indicate with some precision the true scale of the problem - at the physical level, at least. Further data would be useful, especially on non-physical violence; however, because of the methodological problems indicated above, further studies on women only - such as that recently announced by the Office of the Status of Women - would appear to be of doubtful value. The only studies which are likely to prove useful in the future are those which, like those developed by Straus and Gelles and by Steinmetz in the USA, do apply exactly equivalent methodologies to both female and male experience of domestic violence.
Research on male injuries from adult domestic violence is still far from adequate; neither collection of statistics nor analysis have been carried out to the same degree or with the same thoroughness as for females. The recommendations in the VISS report for increased and improved data-gathering must be applied particularly to gathering accurate data - rather than unsupportable 'guesstimates' - on males.
The VISS study indicates that adult domestic violence continues to be a problem that does warrant serious concern in our society, though it is significantly less serious than has been suggested by some groups in recent times.
Adult domestic violence is a problem that does, on the surface, appear to affect more women than men: but the disparity between men and women as injured parties is shown by the VISS study to be very considerably less than has been suggested in the past, and may not be statistically significant. More data are required, particularly on men's experience of domestic violence; but even in its present state the VISS study clearly indicates that the argument that domestic violence is 'a specific women-only problem' can no longer be supported. Any strategies intended to tackle adult domestic violence must now take male and female violence equally into account.
"The needs, concerns, feelings and fears of men and of women are of exactly equal value and importance."
A report by People's Equality Network, PO Box 777, Kew, VIC 3101. (c) PEN 1995.
For more information on the content and context of this report,
contact Tom Graves on 0407 058 542 (b.h.) , or (03) 5423 9205 (24hrs/fax).
For general information about People's Equality Network,
contact Ray O'Sullivan on (03) 625 0535 (pager), or (03) 428 3006 (fax).