Violence against Women: A literature review commissioned by the National Group to Address Violence Against Women
2. The extent of the problem
"Whilst clear categories and definitions are important for statistical and research purposes, we must never forget that these are abstract analytic concepts developed for a specific purpose - to count the extent of violence. They do not reflect experiential reality, which is always more complex...."
Liz Kelly, Domestic Violence: Enough is Enough conference, London, October 2000
What, why and how to measure the true extent of violence against women are questions which have stimulated much debate among practitioners, policy makers and researchers (Desai and Salzman, 2001; Dobash and Dobash, 1998; Hester, Kelly and Radford, 1996). In relation to domestic abuse, at both national and local level, much effort has gone into trying to identify how a common approach to data collection might be achieved. Attempts to evaluate new service developments and public education initiatives are hampered by the lack of available 'benchmarks'. Differences in data systems and data collection, uncertainty about how to overcome 'double counting', and concerns about the legal aspects of data sharing, are some of the issues currently under debate.
2.1 Recorded crime
Statistical Bulletins produced by the Scottish Executive are the main source of official information on violence against women in Scotland. Information about recorded crime is collected quarterly and published annually. The data is 'offence based' rather than 'incident based' - several offences may be involved in one incident, there may be more than one offender and there may be more than one victim. The figures therefore provide a record of the levels of crime occurring in Scotland, but not the numbers of individuals affected by it (Scottish Executive, 2003). Since what is being counted is 'offences' rather than individuals, gender disaggregation is not possible, with one notable exception.
'Crimes of indecency' is the category which includes figures relating to sexual violence. Prior to 2001, this was broken down into sub headings of 'sexual assault', 'lewd and indecent behaviour' and 'other'. From 2001, the sexual assault category was split into 'rape and attempted rape' and 'indecent assault'. The purpose of this was to allow easier counting of offences covered by the police Statutory Performance Indicator of serious violent crime. An interesting by-product, given that in Scotland the legal definition of rape is gender specific, is that there will now be the possibility of analysing reporting trends in this one area of sexual violence against women.
For the year from 1 st January to 31 st December 2002, there were 913 crimes of rape or attempted rape in Scotland. It is noted that this represents an increase of 21% on 2001, and is the highest level of rapes and attempted rapes ever recorded in Scotland (Scottish Executive, 2003). No particular reason is offered for this Scottish-wide phenomenon. However an increase of 37% in crimes of indecency recorded by two Scottish forces, in Lothian and Borders and in Tayside, is attributed variously to increased reporting of 'historical' abuse, a rise in indecent assault reports as a result of proactive work with children and young people, and a more proactive approach to working with women's support organisations to encourage reporting by their service users (Scottish Executive, 2003).
Incidents of violence against women may involve a range of crimes and offences including serious assault, petty assault, sexual offences and breach of the peace. Women's Aid groups and other women's support organisations have been aware of an increasing use of mobile phones by abusive men as a means of harassing and intimidating women. An increase of 23% in offences involving 'threats or extortion' as a result of this increase in threatening phone calls or text messages is noted by Lothian and Borders Police. Again, because the focus is on the incident rather than the people affected by it, it is not possible to see how far this might relate to violence against women.
Specific statistics on domestic abuse have been collected by the Scottish police forces since 1999. These give more detail about recorded incidents of domestic abuse. 36,010 incidents were recorded in 2002, the majority of which (59%) did not go on to be recorded as a crime or an offence. 90% of the incidents involved a female victim and a male perpetrator (Scottish Executive, 2003). A particular concern raised in the most recent report is the level of repeat victimisation - where the information was available, about half of the cases involved repeat victimisation (Scottish Executive, 2003).
The data from different police forces raises some questions about differences in police recording. Some forces do not record a crime or offence if no further action is taken after the initial report, for example if the victim does not wish to pursue the matter, while other forces will record. As a result, Central Scotland shows only 24% of incidents leading to the recording of a crime or offence, while Grampian shows 98% of incidents being recorded as a crime or offence. The report notes that "these recording practices are under continuing review with the intention of achieving consistency across Scotland" (Scottish Executive, 2003: 27)
2.2 Agency statistics
There is, then, some information available on the incidence of violence against women as encountered by the criminal justice system. Clearly this does not provide a full picture of the extent of the problem. If the numbers of workers seeking training on how to respond to women who have experienced violence is anything to go by, there are many more women seeking help, support or information from agencies outwith the criminal justice system. Women present to housing departments, social work departments and health professionals looking for a range of services. However, this is not reflected in the statistical information available from these agencies. In some instances, the options available to staff when recording the reason for a referral do not include 'violence'. Where it is recorded, it may not be in a format which enables collation. Health professionals, for example, will record the information that a patient has disclosed violence or abuse, but this will be in a narrative form in the patient's record. Even where these records are held on a computer system, it is not possible to extract only information about 'violence as a reason for referral' without an appropriate coding system.
Recognising the need to address this difficulty with data collection, the Scottish Executive undertook a three day snapshot in December 2003 to gauge the extent to which women are presenting to agencies as a result of domestic abuse. Participating agencies across Scotland included Women's Aid, accident and emergency departments, primary care teams, social work teams and homelessness units. A report on the outcomes is expected shortly.
Women's voluntary organisations are the other main source of data about the numbers of women experiencing violence. The 40 affiliated local groups in the Scottish Women's Aid network received 72,029 requests for information and support between April 2002 and March 2003, a 10.3% increase on the previous year. In addition, 5,873 women requested refuge, this need for emergency accommodation implying a recent history of abuse by a partner (Scottish Women's Aid, 2003). The Women's Support Project, a Glasgow based organisation working with women affected by any form of violence or abuse, received 1550 requests for support between April 2002 and March 2003 (Women's Support Project, 2003). Statistics will be available shortly from Rape Crisis Scotland, the national office for the network of Rape Crisis Centres in Scotland. A new database is due to come on stream to allow routine collation of statistics from all local groups.
The under-reporting of violence against women to any agency is well documented. A survey of women in Edinburgh found that although over half of the respondents had experienced some form of violence, only 21% of those who had experienced physical or sexual violence had approached a support agency for help (Henderson/CEC 5, 1997). Sources other than agency statistics must be examined in order to assess the true extent of violence against women.
2.3 Crime surveys
Survey results on the prevalence of violence against women may vary depending on a range of factors including the definitions used, the methodology used, and the context of the survey (Johnson, 1998; Walby and Myhill, 2001). However, it is estimated that between one in two and one in ten women will experience some form of violence at some point in their lives, with between 0.4% and 10% of women experiencing violence in any 12 month period (see, for example, Budd, Mattinson and Myhill, 2000; Henderson/CEC, 1997; Macpherson, 2002; Mirrlees-Black, 1999; Mooney, 1993; Myhill and Allen, 2002; Statistics Canada 1993, cited in Johnson, 1998; World Health Organisation, 2002).
National crime surveys have provided some extension of the data available from official statistics. The Scottish Crime Survey (SCS) has run independently of the British Crime Survey (BCS) since 1993, and was repeated in 1996 and 2000. The SCS 2000 survey included a self-completion questionnaire which asked about domestic violence. 6% of women responding had experienced either threats or force from a partner during 1999; 19% of women had experienced threats or force from a partner at some point in their lives (Macpherson, 2002). Similar self-completion modules on rape and sexual assault were included in the BCS in 1998 and 2000. A report based on findings from both of these surveys noted that 0.9% of women had experienced some form of sexual victimisation in the previous 12 months; 9.7% in their lifetime. Partners were the perpetrators in 32% of cases (Myhill and Allen, 2002).
Walby and Myhill note that reporting of violence against women in these generic crime surveys increases over time and as survey methodology is refined and developed, and consider whether the methodology applied by some of the national surveys on violence against women might improve reporting rates still further (Walby and Myhill, 2001). They identify some problems with generic crime surveys, including limited time available to build a rapport with survey participants, or to ask "nuanced questions" 6 about women's experiences of violence. They also consider whether, in generic surveys, less priority might be given to selection and training of interviewers. Finally, they raise the question of how far women see what has happened to them as 'a crime' and whether questions of definition influence how far women report their experiences (Walby and Myhill, 2001). They are not convinced that this is a major problem and cite the 1996 BCS survey on domestic violence, in which "significant numbers of people did in fact report domestic violence even when they said they did not consider it a crime" (Mirrlees-Black, 1999, cited in Walby and Myhill, 2001: 508).
Crime surveys tend to focus on a single aspect of violence against women, and by the nature of their perspective - violence as crime -are inclined not to consider in depth areas of women's experience which are harder to frame as 'crime'. They may not address the particular issues related to violence against women from marginalised groups, including black and minority ethnic women, women with disabilities, lesbian women, and women working in prostitution. Women from these groups may be subject to higher levels of some types of violence than women in the general population, and the violence they experience may impact on their lives in different ways (Barnard et al,
2001; Farley, 1998; Henderson/CEC, 1997; Johal, 2003; Kelly, 2000; Saxton et al, 2001; Siddiqui, 2003).
The authors of a report which addresses the intersection of domestic violence and 'minoritisation' 7 make the point that whilst it is important to acknowledge that domestic abuse happens across all ethnic and socio-economic groups, by doing so there is a danger of making the specific experiences of 'minoritised' groups invisible (Batsleer et al, 2002). Similar difficulties have been noted in relation to women with disabilities (MacLeod, 1995, Nosek and Howland, 1998)). For both of these groups of women, the standard obstacles to reporting which may be experienced by many women - fear of the consequences, fear of not being believed, lack of access to information about services which might assist - are compounded by additional barriers, such as lack of information in appropriate languages or formats, lack of cultural awareness within agencies and greater dependency on the abuser, who may be the main carer or interpreter for the woman.
The experiences of older women may not be considered. Although older women are deemed to be at less risk of violence than younger women and girls (Statistics Canada, 1993; VAWS, 1996; Henderson/CEC 1997), they consistently express more anxiety about their safety than younger women (Henderson/CEC, 1997; Home Office, 2003; SCS, 2002). There has been little specific research on the extent of violence against older women, and the possibility that older women are less likely to report violence cannot be discounted. The Scottish Executive has commissioned research into older women's experiences of domestic violence which was completed in May 2003 and is due to be published in Spring 2004.
Poverty may also make women more vulnerable to violence. It limits choices and forces women into types of employment which carry more risks of violence, for example prostitution, or work in the service industries. It may also put women in the position of having to take on shift work or work which is far from home, reliant on public transport, and they may be more at risk for this reason (
, 2001; Byrne et al, 1999; Statistics Canada, 1993).
2.4 Counting the gaps
There can be difficulty in extracting from a general crime survey the significance of the event in a woman's life and a sense of the interconnectedness of the issues. The prevalence of experiences of stalking and sexual harassment, the involvement of women in pornography, prostitution and organised or ritual abuse, and the extent to which women experience violence in more than one context and at more than one point in their lives has been largely unexplored until relatively recently. Some areas of women's experience remain invisible in any attempt at 'counting'. Those studies which have been done indicate that significant numbers of women have been affected by these issues.
Stalking, defined as 'persistent and unwanted behaviour', affected 17% of women at some point in their lives in a study of stalking and harassment in Scotland (Morris, Anderson and Murray, 2002). When asked about experiences they would categorise as 'stalking', only 10% of women reported a lifetime experience. This is closer to the findings from the U.S. national survey on violence against women, which asked respondents about stalking which 'involved a high level of fear'. In this study, 8% of women reported a lifetime experience of behaviour which fitted the description they were given (Tjaden and Thoennes, 1998).
The trafficking of women for sexual exploitation has gained increasing attention over the last few years, as awareness has increased of global trafficking in persons. The hidden nature of trafficking makes it difficult to assess accurately the numbers of women involved. Using a range of data sources, including a survey of police forces and a review of data on immigration patterns, organised crime and prosecutions for prostitution related offences, a recent study on the extent and nature of trafficking of women in the UK estimated that between 142 and 1420 might be trafficked into and within the UK per annum (Kelly and Regan, 2000).
There are similar difficulties with estimating the numbers of women working in prostitution. Routes Out Of Prostitution (ROOP), a Glasgow based project working to address prostitution as an issue of violence against women, estimates that over 1000 women in Glasgow are involved in prostitution (ROOP, 2003). This, however, is in the context of acknowledging that the 'visible' face of prostitution is on the streets, and that the numbers of women working in indoor prostitution are largely unknown and difficult to monitor (Kelly and Regan, 2000).
Women working in prostitution report much higher levels of violence than other women in the population - 82% of respondents in a San Francisco study had experienced physical assault while working in prostitution and 68% had experienced sexual assault 8 (Farley and Barkan, 1998). A study of women working in Edinburgh, Glasgow and Leeds found that two thirds of the women interviewed had experienced violence from clients (Barnard et al, 2001).
The San Francisco study also found that 57% of the respondents had experienced sexual assault in childhood. This is at the higher end of estimates of the prevalence of experiences of child sexual abuse in the general population, which range from one in four (Creighton and Russell, 1995) to one in eight (Baker and Duncan, 1985). Although Kelly et al (1991) found that 59% of the young women they surveyed had experienced some form of sexual abuse before they reached 18, their definition of sexual abuse included the spectrum of behaviours from flashing to rape. Narrowing the definition to exclude flashing and other 'less serious' abuses, the prevalence rate drops to 1 in 20 young women. This highlights some of the difficulties inherent in comparing studies. Few prevalence studies have been done in this area, and those that have use a wide range of definitions, methodologies, and sample profiles. The study by Kelly et al was designed to address some of these issues, and to identify what adaptations to methodology might help to provide a clearer picture of childhood experiences of abuse. The authors recommend taking a broad approach to 'defining', maintaining that narrowing the definition used in a survey too much excludes the possibility of gathering valuable qualitative data, for example in relation to the seriousness of the abusive experience for the individual at the time, and also in relation to the long term consequences (Kelly, Regan and Burton, 1991).
The use of narrow definitions of violence against women excludes more than qualitative data. It is difficult to count something which has not been named; but until it is counted, how do we know it exists? For survivors of ritual abuse, the narrow focus in most surveys on the type of assault, e.g. 'rape' or 'physical assault', may exclude the possibility of naming the context in which the assault takes place, and thereby help to perpetuate the belief that such abuse does not exist (Matthew, 2001, 2002; Scott, 2001). A small scale survey which sought to assess agency awareness of ritual abuse asked respondents how many survivors of ritual abuse were known to have made contact with their agency between August 2002 and August 2003. 25 agencies responded; between them they reported contact with 96 survivors of ritual abuse (TRASH, publication pending).
It is clear that although single issue studies can be helpful in assessing the need for a specific service or policy development - for example, the need for development in housing policy to meet the needs of women fleeing domestic abuse - they do not allow the linkages between different types of gendered violence against women at different times in the lifecycle to be fully explored. In order to assess the full extent of 'violence against women' in a society, a different, broader approach may be required.
2.5 National surveys
The Canadian national survey in 1993 asked questions about a wide range of experiences, and framed the questions around the theme of women's safety. The results were significantly higher than those found in other surveys. In 1993, Canadian police recorded 15,200 sexual assaults against women; the General Social Survey (GSS) 9 recorded 316,000 sexual assaults against women. Based on responses from 12,300 women, the Violence Against Women Survey (VAWS) estimated that there had been 572,000 sexual assaults against women in Canada that year. The pattern for assaults against women by their partners was similar - 46,800 'assaults against wives' were recorded by the police, 107,500 assaults were recorded by the GSS, and 201,000 assaults were estimated by the Violence Against Women Survey. Citing these figures in 1998, Johnson notes:
"The specialized survey of violence against women captures almost twice as many incidents as the traditional crime-victim survey, four times as many cases of wife assault as are reported to the police, and about 38 times as many cases of sexual assault as police statistics." (Johnson, 1998: 39-40)
She attributes this increase in large part to a meticulous design process, in which every effort was made to create an approach which would encourage women to discuss their experiences of violence as fully and as safely as possible with the interviewers. The design team consulted with academic researchers, frontline practitioners, policy makers and women survivors of male violence. A random sample telephone survey was deemed to be the best approach, for a population with 99% access to a phone. The questionnaire design involved a rolling programme of consultation through focus groups, moving from general discussion through detailed analysis and testing of the content of the questionnaire. Consideration was given to how the questionnaire would be administered, including the selection and training of interviewers. It was held to be important that the interviewers be women, removing one perceived barrier to disclosure. Potential interviewers went through standard job screening to assess their suitability for the post and then underwent a second interview with a clinical psychologist to assess their ability to deal with the particular stressors associated with a large scale survey about violence and abuse. Interviewers were provided with eight days training which focussed on issues such as responding to disclosure and distress and ensuring that interviewees were in a safe place to participate in the interview. During the initial stages of phone contact, interviewers provided women with a 'toll-free' number to call if the interview was interrupted, or to discuss anything raised for them by the interview (Johnson, 1998).
The results of the Canadian survey are significant, not just because of the numbers of women who reported violent experiences, but because of the links made between different experiences of violence at different times in women's lives, and in what is revealed about the context, meaning and impact of violence in women's lives. Of the 12,300 women surveyed, 51% had experienced at least one physical or sexual assault since age 16 and almost 60% had been assaulted more than once. 25% of the women surveyed had been physically and/or sexually assaulted by partners or ex-partners - 20% of the women who were assaulted by a partner were sexually assaulted. In all, 45% of the women surveyed had been assaulted by someone known to them, compared with 23% who had been assaulted by strangers. 38% of women had been sexually assaulted by a man known to them (Status of Women Canada, 2002).
Since 1983, Canada has had no specific offences of rape, attempted rape or indecent assault. Instead, there are three levels of 'sexual assault' ranging in severity from "unwanted sexual touching" to "sexual violence resulting in serious physical injury sustained by the victim" (Status of Women Canada, 2002: 20). Over 90% of all incidents reported to the police are recorded as level 1 assaults - minor physical injuries or no injuries to the victim, with a maximum possible sentence of 10 years. The 1993 survey found that only 6% of all sexual assault incidents had been reported to the police, and only 4% of incidents involving 'unwanted sexual touching' (Status of Women Canada, 2002). The survey explored reasons for low reporting, which were similar to those reported in UK surveys - the incident was considered too minor (44%), women didn't think the police could do anything (12%), or they wanted to keep it private (12%). Interestingly, given how many sexual assaults were committed by men known to women, fear of the perpetrator, and not wanting the perpetrator arrested or jailed, were given as reasons for not reporting by only 3% of the women.
In their review of the development of national surveys on violence against women, Walby and Myhill attempt to define 'state of the art methodology'. They identify several issues to be considered, including the context of the survey; interviewing practices; training for interviewers and ensuring a 'good fit' between interviewer and interviewee; the sampling frame; the mode of enquiry, and situating the event in relation to others (Walby and Myhill, 2001). They conclude that dedicated surveys on violence against women are likely to be more effective than general crime surveys which include questions about violence against women, since they pay greater attention to maintaining the safety of interviewees, building rapport and focussing on the meaning and impact of the violence as much as on the act itself.
In their 2002 report on assessing violence against women, Status of Women Canada 10 identify a set of indicators to be used as benchmarks against which to measure progress on tackling violence against women. These are designed to parallel the Economic Gender Equality Indicators released in 1997 by the F/P/T 11 Ministers Responsible for the Status of Women. The violence against women indicators follow six themes:
- Severity and prevalence of violence against women
- Impact of violence against women
- Risk factors associated with violence against women
- Institutional and community based responses
- Victims' use of services
- Public attitudes and perceptions
The authors stress that these indicators are limited by the quality of the data available, and note that as data collection methods improve, so must the indicators expand and adapt over time (Status of Women Canada, 2002). However, it is undoubtedly the case that the baseline data gathered in Canada through the national surveys of violence against women provide an invaluable benchmark against which to measure the effectiveness of policy development and legislative change (Hague, Kelly and Mullender, 2001).
2.6 Extent of the problem - conclusions
In summary, a review of the research on the prevalence and incidence of violence against women tells us that:
- violence against women is widespread, affecting women of any age, class, race, religion, sexuality, or ability
- women are most at risk from men they know
- factors which may increase women's vulnerability to some types of violence include age, disability, and poverty
- when asked, significant numbers of women describe patterns of abusive behaviour and repeat victimisation, rather than discrete assaults
- women experience violence at different points in their lives, and significant numbers of women experience more than one type of violence
- surveys which address violence against women in all its forms may yield more information than 'single issue' surveys about the meaning and impact of violence in women's lives
- few studies have been designed specifically to record the experiences of marginalised groups of women