Jump to ContentJump to Main Navigation
Domestic violence and sexualityWhat’s love got to do with it?$

Catherine Donovan and Marianne Hester

Print publication date: 2014

Print ISBN-13: 9781447307433

Published to Policy Press Scholarship Online: September 2014

DOI: 10.1332/policypress/9781447307433.001.0001

Show Summary Details
Page of

PRINTED FROM POLICY PRESS SCHOLARSHIP ONLINE (www.policypress.universitypressscholarship.com). (c) Copyright Policy Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in PPSO for personal use.date: 20 September 2021

How we did the research

How we did the research

the COHSAR research approach

(p.35) TWO How we did the research
Domestic violence and sexuality

Catherine Donovan

Marianne Hester

Policy Press

Abstract and Keywords

This chapter outlines the development of the COHSAR survey questionnaire and interview schedule. It explains the feminist epistemological approach used to design the survey and to explore how processes of gendering and power might operate in similar or different ways in abusive female and male same sex or heterosexual relationships. Combined with interviews this allowed exploration of the intersection of gender, sexuality, ‘race’, ethnicity, age, disability, class, income and education. A national community survey was used, as a representative approach was not possible in the UK. The survey, asked those in same sex relationships about experiences and perpetration of violence and abuse as well as impacts and motives for using them. The interview schedule was based around an exploration of a best and a worst relationship experience in same sex and/or heterosexual relationships. The demographic profiles of the final data set of 746 survey and 67 interview respondents are given.

Keywords:   feminist epistemology, COHSAR research tool, intersectionality, demographic profile, national community survey, interviews

It was important that we adopted a research approach and developed methods that could deal with issues of gender, power and sexuality, let alone other differences. We used a feminist epistemological approach as this would help us to construct research instruments (survey, interviews) geared to exploring how processes of gendering and power might operate in similar or different ways in abusive female and male same sex and heterosexual relationships. Following our analysis of existing research (see Chapter One), the survey instrument also needed to provide data regarding a range of domestically violent and abusive behaviour while taking into account both context and impact, and to include questions about experiences of abuse from partners and use of such behaviour against partners. Some of these issues were further explored in the interviews, which also looked at links between love and violence. In this chapter we explain the rationale for our research approach, discuss the survey and interview methods that resulted, and outline the basis for our analysis. The overall approach is called COHSAR (COmparing Heterosexual and Same sex Abuse in Relationships)

Framing the research

Methods are situated historically (Hester et al, 2010). They not only reflect the socio-economic context and concerns of different eras, but develop and change over time as different interest groups and needs emerge (Savage and Burrows, 2007). Different research traditions have consequently developed in different parts of the world and at different times, and different methods have also achieved differential degrees of credibility and impact. The often critical stance of feminists regarding quantitative methods have to be seen in this light. Quantitative methods, and survey approaches in particular, have tended to be used for policy research, often in ways that have not been sensitive to gender or women’s experiences and concerns (Skinner et al, 2005). The (western) feminist academic project has been to uncover the positioning of ‘women’ and gendered beings in the social world. In (p.36) that respect the feminist project has always been about ‘deconstruction’ in some way and about the uncovering of meaning. It is therefore not surprising that feminists have often argued that qualitative approaches are more appropriate to the feminist project as these provide rich data that allow the application of a variety of textual and deconstructive techniques, allowing us to uncover how gender is construed, perceived and so on in a myriad of contexts. We would, however, agree with Ann Oakley (2000) for the need to move beyond and indeed to close the ‘paradigm war’ that has existed between quantitative and qualitative approaches, as it stops us using the most appropriate methods for the job or producing the best research possible. As Oakley argues:

Qualitative research is not more authentically female or feminine than ‘quantitative’ research…It is not necessarily more ethical either…There is no such thing as ‘simply’ recording or publishing data. There must always be selection; the critical issue is whether this is made according to the kinds of open and systematic criteria which other people can inspect, or not…The more appropriate goal is…the continuation of systematic enquiry.

(Oakley, 2000, 296–97)

The key issue is therefore to obtain ‘a more critical, and ethical, approach to all kinds of methods’ (Oakley, 2000, 302). Echoing Oakley we would suggest that drawing on the lessons from feminist approaches is an important means of achieving this (Hester et al, 2010).

To think epistemologically is to ask questions about what can be known, and the interrelationship between knowledge, experience and ‘reality’ (Skinner et al, 2005). How knowledge, experience and reality might interrelate, however, is an area of contestation within feminist debates (Ramazanoğlu with Holland, 2002). There can be different epistemologies and they lead to different knowledge about the social world. For us, the importance of adopting a feminist epistemological approach is that such approaches have a questioning of power, gender and sexuality as a central focus. The relationship between gender and power is, of course, something that is not straightforward, and different approaches have been adopted, for instance, ‘standpoint’ or ‘postmodern’ approaches with distinctions between how to understand reality and who the knowers are. ‘Standpoint’ approaches seek to understand the experience of oppression from the positioning as subordinated, where women or lesbians and gay men may all be constituted as the ‘ruled’ (Smith, 1988). Postmodern approaches instead attempt to understand the many ‘realities’ and subjugated knowledges discoursively produced (p.37) at different times and locations, tending also to a rejection of ‘static’ identity positions such as ‘women’, ‘lesbians’ and ‘gay men’. In these approaches, moreover, conceptualisations of power and control have been increasingly neutralised in many respects. For instance the Foucauldian project has created understandings of power and control as fluid concepts created through interactions between individuals, but largely without structured power. The identification and understanding of structural inequalities, and in particular what Stark (2007) calls ‘sexual inequality’ (male–female inequality) has also largely been lost from the application of ‘power and control’ models, at least in the US. This is what Stark takes issue with and why he is reluctant to use the model of power and control. He sees the use of ‘coercive control’ as a way of putting sexual inequality back into the equation.

Our research approach draws to some extent from both standpoint and postmodern perspectives, but (as outlined in Chapter One) with an emphasis on the material and on the constructions and experiences related to structural inequalities and oppressions. We build on the knowledge of situated power from standpoint approaches and the recognition of difference from postmodern approaches. This allows us to take into account the intersection of inequalities and difference such as those associated with gender, sexuality, ‘race’, ethnicity, age, disability, class, income and education (Crenshaw, 1989; Bograd, 2005; and see Chapter One). Understanding how gender and sexuality intersect with regard to how individuals may use, experience and embody domestic violence and abuse (DVA) is crucial to our project of comparing similarities and differences across abusive LGBTQ or heterosexual relationships. That provides only a very partial picture, however. Incorporating intersections and mutual shaping with regard to our participants’ experiences and locations as they relate to ethnicity, age, disability, class, income and education is also crucial and further enhances our understanding of DVA use, experience, embodiment and help-seeking. Linked to this, we see knowledge about DVA – what it is, what it does – as rooted in the accounts of victims/survivors, and to a lesser extent in the accounts of perpetrators of DVA and witnesses (Hester et al, 2007). The use of ‘experience’ in feminist research is of course an area of contestation (Ramazanoğlu with Holland, 2002). We do not see experience as providing ‘truth’, rather, accounts of experiences are ‘stories’ that may vary in their telling over time and to different audiences. Nonetheless, how individuals report experience does help us to begin to understand similarities and differences across heterosexual and LGBTQ lives, to develop our understanding of the questions that need to be asked, and thus to construct better research (p.38) instruments that reflect situated knowledge. As Ramazanoğlu with Holland (2002) point out ‘[d]espite the problematic status of accounts of experience, they provide knowledge that otherwise does not exist’ (p 127).

The COHSAR survey

We wanted to apply a feminist epistemological approach in the development of a questionnaire survey that would reach a much wider sample of individuals in same sex relationships, and which could be used to compare data on DVA among individuals identifying as LGBTQ with those identifying as heterosexual. We therefore set out to develop a questionnaire that would not only draw on existing surveys of DVA, such as the Crime Survey for England and Wales (CSEW), but would incorporate questions that might reflect to a greater extent ‘how we know’ about such violence and abuse in same sex as well as in heterosexual relationships. In other words, to reflect what previous research on ‘experience’ of DVA tells us about the possible features and dynamics of such abuse, while at the same time allowing new knowledge to emerge. It can, however, be difficult using questionnaires to capture ‘patterns over time’, to identify what individuals may experience as ‘coercive control’ or ‘situational’ violence, let alone to take into account different contexts for the abuse. In what follows we discuss some of the ways in which we attempted to address these issues in constructing the COHSAR questionnaire.

Much of the debate about the use of surveys in researching DVA has focused on measures and approaches used in heterosexual surveys. Similar measures have, however, increasingly been used for same sex DVA surveys, and also need to be examined if comparison is to be made across heterosexual and same sex relationships (Hester et al, 2010). Consequently, many of the critiques in relation to heterosexual surveys are also of relevance to the development of a survey approach for SSDVA.

Obtaining the COHSAR survey sample

Ideally, the survey should have used a representative sample, but this was not possible in the UK, as there was no national dataset in existence with information about sexual identity and contact details or location of the individuals concerned. Indeed, the difficulty in obtaining representative samples within LGBTQ communities has generally been seen by researchers of same sex DVA as ‘the greatest (p.39) challenge facing researchers in this area…and a true random sampling strategy probably is…impossible to achieve’ (Murray and Mobley, 2009, 377). Although sexual identity was added as a question in the UK Integrated Household Survey from 2009, this was not available when we carried out the COHSAR survey. Instead, we decided to carry out a large UK-wide ‘community’ survey with a convenience sample as the most wide reaching and ethical alternative to a representative sample (McCarry et al, 2008). To maximise the sample we developed an extensive network of contacts (over 220) with LGBTQ and DVA organisations across Scotland, Northern Ireland, Wales, North-East England, North-West England, Central England, South-West England and South-East England including London using internet searches, LGBTQ literature, national helplines, the media and personal contacts.

A total sample of 800 responses was obtained, from which 54 cases were removed because: their sexuality was unknown; or they had not had a same sex relationship; or they identified as heterosexual and had never had a same sex relationship (some individuals who define as ‘heterosexual’ may have had a same sex relationship in the past). This resulted in a final data set of 746 individuals who had been or were in a same sex relationship.

The questionnaire included sections on: personal demographic information; decision making and conflict resolution in own relationship; own experience of negative emotional/physical/sexual behaviours including impact; own use of negative emotional/physical/ sexual against partner including motives, help-seeking; and a final section asking whether respondents had experienced DVA plus other questions eliciting views and opinions (for a detailed outline of the contents and sampling, see McCarry et al, 2008).

To name or not to name something as DVA

In developing the questionnaire we were immediately faced with an important question – whether or not to name as ‘domestic violence and abuse’ the phenomenon we were ostensibly studying (Hester and Donovan, 2009). Should we be up-front in stating that this was a questionnaire about DVA? Or develop a questionnaire about something less obviously defined such as ‘problems in relationships’? Previous research has indicated the difficulties (even greater than in heterosexual relationships) involved in naming as ‘domestic violence and abuse’ harmful or abusive behaviours or experiences within LGBTQ relationships (for example, Giorgio, 2002). We also acknowledged the difficulties some individuals might have in perceiving anything other (p.40) than physical violence as ‘domestic violence and abuse’. Ethics are an important feature in feminist research (Skinner et al, 2005) and the ethics involved in using a covert or an overt approach therefore also had to be considered.

We decided to carry out an extensive consultation exercise, with representatives from a range of LGBTQ groups, to test which approach to use and why. Two alternative cover sheets were produced, both introducing the research as ‘Same sex relationships: when things go wrong’. One used the following sentence as part of the more detailed description for the research: ‘Recently, in the UK, there has been a growing concern to make services more relevant, appropriate and accessible to people in same sex relationships who might need help or advice because of domestic abuse’, while the other, using mostly the same sentence, omitted the last part: ‘because of domestic abuse’. The majority of those consulted said they would prefer the latter, framing the questionnaire in terms of relationships generally, rather than explicitly stating a focus on DVA:

‘Personally, I think you should go covert as people who may be suffering domestic abuse might be put off filling it in. In addition, people who think that they aren’t suffering domestic abuse but whose partners are exhibiting some of the behaviours listed might be more ‘honest’ about their answers if the questionnaire is not marketed as a domestic abuse questionnaire.’

(Critical reviewer)

It was also apparent from the interviews, carried out once the questionnaire survey had been completed, that this approach helped to elicit a wider range of responses. For instance, one lesbian interviewee talked about controlling experiences she had had, not being able to get her partner to leave the house, being continually questioned about everything she did and about whom she was with. She had wondered if this was adequate as the basis for saying on the questionnaire that she had experienced domestic abuse:

‘And…when I was filling out the questionnaire…I did think, “Well actually, is this really going to count,”…but it does fall into it, I think.’

(Kay, a white lesbian, aged 35–9 at interview, 32 years old when DVA relationship began)

Individuals such as Kay might not as readily have responded to a survey explicitly about DVA. With regard to ethics, taking a more covert (p.41) approach thus seemed justifiable in that it allowed a wider range of individuals to talk about potentially abusive relationship experiences.

Following extensive piloting, the survey was thus described to potential respondents as examining ‘when things go wrong’ in same sex relationships and the term ‘domestic violence and abuse’ was deliberately not used in the survey until the last page (Hester and Donovan, 2009). The sample, however, still had a larger proportion of respondents who had experienced potentially abusive behaviours from a partner than the more general health surveys carried out within LGBTQ communities (Hunt and Fish, 2008; Guasp, 2012; Henderson, 2003; and see Chapter Four).

Comparison with previous surveys

To achieve comparison between male and female same sex relationships, and with heterosexual relationship experiences, the survey replicated some of the CSEW self-report module on domestic violence. Areas for replication included time periods and violence/abuse types. Relevant US studies were also drawn on for development of same sex specific questions relating to types of abuse, as well as items on decision-making and conflict resolution (Renzetti, 1992; Turell, 2000), which allowed the questionnaire to move beyond the ‘heteronormative’ approach of the CSEW and to allow comparison across sexuality and gender (McCarry et al, 2008; Hester and Donovan, 2009).

While the CSEW refers to both intimate partner and non-partner violence (Povey et al, 2008), we wanted to focus only on intimate partner violence. From our previous research (for example, Radford and Hester, 2006) we were aware that questions regarding experiences of DVA needed to be both detailed and nuanced. Consequently a wide range of questions pertaining to respondents’ experience of emotional abuse (27 items), physical abuse (13 items), and sexual abuse (nine items) both within the last 12 months and earlier, were asked. In each case respondents were asked whether they had ‘never’, ‘sometimes’ or ‘often’ experienced the behaviour in question. Many of the questions on emotional behaviour reflected those used in the CSEW including questions about being isolated and about financial control. We also included questions that were directly targeted towards the same sex community about being ‘outed’ and having sexuality used as forms of abuse. We replicated questions from surveys with gay men that particularly related to HIV related abuse, for example, withdrawing medicines, while recognising that this kind of abusive behaviour could be used in relation to any health condition where medication is used. (p.42) In addition we also asked questions about other identity abuses, for example, whether respondents had ever had their ‘race’, social class, disability and so on used against them or used their partner’s identities in these regards against them.

The CSEW questions on rape and sexual assault accommodate the 2003 Sexual Offences Act definitions of rape. According to the Act, while both a woman and a man can be a victim of rape, it remains that only a man can commit rape. In order to incorporate the experiences of women who felt they had been raped in a lesbian relationship, however, our questions needed to be open enough so that women could define their own experiences and not be excluded because of legally prescribed gendered definitions. Also, while both heterosexual and LGBTQ individuals may participate in sado-masochistic sexual activities, there has been much more debate about issues of consensual and non-consensual behaviour in this regard and in relation to HIV and sexual experiences more widely in LGBTQ communities. Questions about breaches of requests for safer sex and safe words were thus included, and were deemed important in discussions with the LGBTQ community during the pilot phase. This is a key area that differentiates our questionnaire from the CSEW and again, moves beyond the heteronormative.

In addition, respondents were asked to identify whether their responses related to the behaviour of a current partner, to a previous partner, or to both. To determine validity and reliability of the items relating to potentially abusive experiences both separate and combined scales were developed. Three separate scales relating to emotional, physical and sexual abuse behaviours were created, as well as a combined scale including the three items. All were found to be reliable (Cronbach’s Alpha 0.865, 0.895, 0.807, 0.915 respectively). In Chapter Four we outline the findings from the survey and discuss some of the issues and meanings by also incorporating findings from the interviews.

In order to check whether respondents who had answered affirmatively to experiencing any of the potentially abusive behaviours might also consider that their experiences constituted ‘domestic violence and abuse’, we included a further question towards the end of the questionnaire that explicitly asked if the respondent had ever experienced domestic violence or abuse in a same sex relationship.

(p.43) Incorporating impact

As indicated in our discussion of the previous studies on SSDVA, we would argue strongly that surveys looking at prevalence of seemingly DVA behaviours should also look at the impact of these behaviours (Hester et al, 2010). Without including questions about impact it is not possible to differentiate between individual behaviours that may create the kinds of changes and feelings in victims/survivors associated with experiencing the power over and control of DVA, and those that do not. Other surveys have tended to use narrow measures of such impact, via severity (using number of times something happened) and whether or not there is physical injury. For instance in research based on the Conflict Tactics Scale (CTS), severity is determined by frequency of incidents and rating tactics used. It is difficult to assess whether or not this is commensurate with the actual impact on the individuals concerned, however. We wanted to go beyond these approaches and thus constructed a much more detailed and sophisticated set of questions. It is generally assumed that higher levels of abuse should be associated with a greater impact upon respondents (Walby and Allen, 2004). We were able to test whether the relationship between the frequency of incidents of abuse and its impact on respondents’ lives reflected this assumption. As detailed further below and in Chapter Four, we found strong association between frequency of incidents and greater impact, which was especially marked where individuals experienced combinations of physical, sexual and emotionally abusive behaviours.

In order to explore the impact that abusive behaviours may have had on the respondent, each of the three sub-sections on potentially abusive behaviours (physical, emotional and sexual) also included questions about impact related to those behaviours. To move beyond the limitations of previous surveys we included a wide range of questions that might allow us to understand the nature and severity of impact, and which we devised from previous research. The result was a multi-response survey item listing 26 possible outcomes and inviting respondents to tick all that applied in relation to emotional, physical and sexual abuse separately. The questions included physical and psychological impacts, effects on relationship quality and partner interactions, as well as questions regarding acts that may be seen as self-defence or retaliation. There was also the possibility of answering that there was no impact. The result was three scales with a high degree of Alpha reliability measuring the impacts of emotional, physical and sexual abuse (Cronbach’s Alpha 0.933, 0.959, 0.951 respectively). (p.44) With regard to the assumption that higher levels of abuse should be associated with a greater impact upon respondents, this was also reflected in the relationship between the frequency of incidents of abuse and its impact on respondents’ lives. Overall the empirical (Spearman’s rank) correlation between scores on the impact scales and potential abuse scales relating to the previous 12 months supported this assertion with strong correlations evident between impact and emotional behaviour (0.503, p<0.001), physical behaviour (0.463, p<0.001) and sexual behaviour (0.432, p<0.001) (Hester et al, 2010). We discuss the findings related to the intersection of the abuse and impact scales in Chapter Four.

In relation to those respondents who had disclosed that they had used some of these abusive behaviours against their partner/s, we asked respondents to explain ‘why’ they had abused their partners and they were given a choice of 21 closed responses from which to choose (they could opt for as many as applicable). It should also be noted that this question was only relevant to those who had identified that they had used potentially emotionally, physically or sexually abusive behaviours against any of their ex/partners. This question was important in differentiating between behaviour carried out by partners with the intention of harming or controlling their ex/partners and those behaviours used in self-defence for example. This is a significant point of departure between our survey and some other surveys that are unable to differentiate between mutual abuse, aggressive abuse and actions carried out in self-defence. The consequence of not doing so leads in other surveys to data that may misrepresent some actions as actively DVA when instead it constitutes defensive behaviour. The findings regarding use of potentially abusive behaviour against partners are discussed in Chapter Four.

The COHSAR survey participants

The COHSAR survey included a wide range of demographic questions. Murray and Mobley (2009) recommend that where representative sampling is not feasible, researchers should report in great detail the demographic characteristics of the sample, and report at least:

age, gender, ethnic background, self-reported sexual orientation, income level, education level, employment status, current relationship status, the length of the relationship and the partner’s demographic characteristics.

(Murray and Mobley, 2009, 378)

(p.45) In what follows, we outline all such details for our sample, apart from employment status and partner’s demographic characteristics. In addition, we report on respondents’ parenting of children and about disability, as these are issues that have previously been found to make individuals vulnerable to DVA (Radford and Hester, 2006; Thiara et al, 2011).

The ages of respondents to the COHSAR survey ranged from 16 years to individuals in their late 60s, although most were in their 20s and 30s. This is a wider age group than in most other surveys on DVA in the LGBTQ community in the UK (see for example, Henderson, 2003; Hunt and Fish, 2008). The average age of our respondents was 35.37. Female respondents tended to be a bit older than the men (for women mean age 35.77, median 37; for men mean age 34.48, median 32). The age distribution for the transgender individuals, just four people, was older, with one in the 20–24 age group, and the rest aged between 40 and 59 years.

When age is looked at in relation to sexuality, there were slight differences in age by self-identified sexuality groups. For instance older women were more likely to call themselves lesbian and younger women were more likely to use the term ‘queer’. Overall the oldest individuals were lesbians (mean age 37.11, median 37). The second oldest group were gay men (mean age 34.90, median 37), followed by homosexuals – who were mainly men (mean age 34.62, median 32), then gay women (mean age 34.48, median 32), and queer – who were mostly women (mean age 32.50, median 32). The youngest were individuals identifying as bisexuals (mean age 30.95, median 27).

Gender was apparent for 736 of the 746 individuals in the sample. Nearly two thirds identified as women (61 per cent), and more than a third as men (38 per cent). There were also four individuals identifying as transgender (0.5 per cent), and one individual identifying as queer (0.1 per cent). Women were most likely to identify as ‘lesbian’ and over two-thirds of women (70 per cent) defined themselves in this way. The second largest category among women was ‘gay woman’ (16 per cent). Men mainly identified as ‘gay man’, with more than three-quarters of the men identifying in this way (76 per cent). The second largest category among the male respondents was homosexual (18 per cent). Other self-identifications used were bisexual and queer. More women than men defined themselves as bisexual (10 per cent compared to 4 per cent of men) or as queer (3 per cent compared to 1 per cent of men). Very few women (1 per cent) identified as homosexual. One individual identified as queer in relation to both gender and sexuality. The four transgendered individuals identified themselves in four (p.46) separate ways: as bisexual, gay woman, lesbian and queer. Of the ten individuals whose gender was unknown, half identified as homosexual (50 per cent), and the rest as bisexual (30 per cent) or queer (20 per cent).

The question about ethnicity used mostly the same categorisation as the 2001 Census, and our findings echoed those in the general population where that was the case. Most respondents identified as white (95 per cent compared to 92 per cent in Census). The proportions identifying as mixed or Chinese were similar to those in the Census, but we had considerably smaller proportions of Asian or black respondents, possibly because the COHSAR survey did not subdivide the categories of Asian or black, which was the case in the Census. In particular the ‘other’ category, which was more than four times as large in our survey than in the Census, may have contained what the census termed ‘other Asian’ and ‘black other’. Individuals identifying as homosexual were most likely to be white (98 per cent), with bisexuals most likely to be represented among individuals of mixed, Asian, black or other ethnic backgrounds (92 per cent).

The income level for the COHSAR respondents was slightly higher than the population generally at the time of the research. The average (mean) income for all the respondents was £ 22,432.43, rising to £ 23,569.67 if only those aged 20 and over were taken into account (with a median, ’midpoint’, income of £ 25,500 – compared to general population median for full-time men of £ 25,000 in 2005, and for full-time women £ 19,400 (National Statistics, 2006)). Even so, one in five earned less than £ 10,000, and nearly half earned less than £ 20,000. As may be expected, there was a tendency for income to increase with age until 60 years and decrease again thereafter at retirement age. Those earning over £ 40,000 were clustered between 30–55 years. The income distribution also reflected the wider income inequality between men and women in the UK population at that time. The largest group of men were earning £ 21–30,000, compared to only £ 11–20,000 for the largest group of women. These gendered ‘norms’ become especially important in discussion about the apparent vulnerability to DVA for lower income groups, and this is explored further in Chapter Four.

The educational attainment of the survey respondents was generally much higher than that of the UK population. Half of the respondents (51 per cent) were educated to at least degree level compared to 27 per cent in England and Wales generally (2011 census), and nearly one in five of our respondents (19 per cent) had at least an ‘A’ level qualification. Just over a quarter of respondents (28 per cent) had (p.47) attained GCSE, NVQ or vocational level qualifications. Very few respondents (3 per cent or less) had no qualifications, compared to 23 per cent in England and Wales (2011 Census).

The vast majority of respondents (87 per cent) had been in a same sex relationship during the past 12 months, with more than two-thirds currently in such a relationship (71 per cent). For about one in seven it was their first same sex relationship (15 per cent). This was similar for both women and men, but there were significant differences between men and women in terms of length of relationships (Chisquare=15.503, p = 0.03). Men predominated in shorter relationships, lasting up to one year, but also in relationships lasting two to five years or more than 20 years. Women were generally more likely to have longer relationships, lasting between one and 20 years. This also reflected what happened regarding staying or leaving a DVA relationship: men were more likely to leave a DVA relationship more rapidly than were women.

Out of 713 individuals who answered the question ‘Do you have a disability?’, more than one in ten said they did (11 per cent). By comparison, in the 2001 Census, a greater number of respondents answered the broader question asking whether they had a long-term illness, health problems or disability that limited their ability to work or their daily activities (18 per cent). Slightly more of the women (12 per cent) than the men in our survey (10 per cent) said they had a disability. Half of those individuals indentifying as transgendered said they had a disability (50 per cent, although the numbers are very small). If sexuality is taken into consideration, the largest proportion of individuals with a stated disability was among those identifying as queer (22 per cent, although small numbers). One in six bisexuals (17 per cent) and just over one in ten lesbians (11 per cent) had a disability, while less than one in ten gay women, gay men, or homosexuals said that they had a disability.

One in six of the survey respondents (16 per cent) parented children. The majority of parents – more than two-thirds (71 per cent) – had all or some of these children living with them. This included most of the school age and teenage children, and a few of the adult children. Not surprisingly, women were almost three times as likely as the men to be parents. One in five women parented children (22 per cent) compared to less than one in ten men (7 per cent). Individuals identifying as transgendered were most likely to be parents, although the numbers are very small (33 per cent). This gender pattern also meant that women identifying as lesbians were most likely to parent children, and nearly a quarter said they were parents (24 per cent). Other, largely female, (p.48) groups such as bisexuals (19 per cent), gay women (14 per cent) and queers (10 per cent), were more likely to parent than those comprising largely men, that is, homosexuals (9 per cent) and gay men (8 per cent).

The COHSAR interviews

Given the lack of detailed data on SSDVA, especially in the UK, the obvious approach would be to develop knowledge of intimate relationships that might be abusive, via in-depth interviews with a range of LGBTQ individuals and heterosexuals. We did indeed adopt an interview approach, which included a follow-up sample from the survey and further participants. As explained earlier, in a departure from most other research in this area, our research was not labelled as being about DVA. Instead we encouraged participation from anybody with experiences of relationships that had ‘gone wrong’. In this way we intended to forestall pre-conceptions on the part of participants about what kinds of relationship experiences ‘counted’. In addition, other researchers have been concerned that, in framing their research in terms of an exploration of DVA in heterosexual, same sex or LGBTQ relationships, they have ‘primed’ participants to give a particular account of DVA reflecting dominant binaries of perpetrator and victim with particular constructions of what perpetrators and victims ‘are like’ (for example, see Ristock, 2002a). The expected account coincides with the public story of DVA, that is, that it should be heteronormative, based in a long-term monogamous relationship and predominantly experienced through physical violence (for example, Aguinaldo, 2004). Because we did not approach the research in this way, it is possible that we were able to facilitate a range of different DVA stories to be told (notwithstanding the caveats outlined above regarding participants intersecting identities and access to resources) in relationships which were short (sometimes only a matter a months) and long (up to 20 years), predominantly emotionally abusive (as typically the accounts of women in same sex relationships were) in relationships living apart as well as together, with children and without; and with accounts from victims/survivors that not only challenge their construction as passive victims but overturn them insofar as they understood themselves to be the (emotionally) stronger person in the relationship (see Chapter Six).

The interview schedule reinforced our intention to provide spaces for participants to reflect on how love can be understood in relationships that go wrong as well as those that go right. The development of the interview schedule was informed by both the survey and four focus groups (8 heterosexual women, 6 heterosexual men, 3 lesbians, and 2 (p.49) gay men). Questions were based around an exploration of two accounts: a best and a worst relationship experience and from beginning to end or current situation. For some respondents their best relationship was also their worst and for others they had only had one relationship so we were led by respondents in how they wanted to respond. Questions then asked about how they had met the partner in their best relationship, whether they loved that partner and how they knew that they did, whether their partner loved them and how they knew that, what the best and the worst things were about their relationship, how they organised their relationship (including how household tasks were distributed, decisions made, bills paid, holidays organised, interior design decided on in shared homes, parenting and so on), how they resolved/accommodated disagreements and differences and, if appropriate, how the relationship had ended. They were then asked the same set of questions about a worst relationship experience. At the end we asked more general questions about respondents’ views of love and DVA in same sex and heterosexual relationships, whether they had experienced domestic violence and abuse and how they defined this. For those respondents who were recruited to the interviews through the survey it is entirely possible that they might have expected to talk about abuse given the tenor of the survey questions; that not every participant did so suggests that our recruitment strategy was successful in its aims to be inclusive of a range of relationship experiences. It is perhaps another measure of the success of our approach that several participants were ambivalent about whether their experience could be named DVA, especially if it had not involved physical violence, but also precisely because it did not seem to fit what they understood to constitute DVA from the public story. On the other hand it is also of some concern that some participants apparently normalised experiences that were by any standards emotionally abusive and were concerned to explain away the behaviours of their ex-partners with reference to a mental health or other crisis the ex-partner had been experiencing at the time.

Analysis of the interview data was thematic. That is, interview transcripts were read and re-read to identify and code themes that emerged from the data in relation to the key research questions: what kinds of differences and similarities occur in the abusive experiences of lesbians, gay men, heterosexual women and heterosexual men; what, if any, narratives of love are drawn on to make sense of abusive experiences and whether these differ across gender and sexuality. The analysis was underpinned by a reading of the sociological literature on love and intimacy to aid the identification of behaviours that have, in (p.50) that literature, been gendered, for example, caring relationship practices, disclosing intimacy, emotion work, decision-making, setting the terms of relationships (Duncombe and Marsden, 1993, 1995, 1996; Jamieson, 1998; Hochschild, 2003; Cancian; 1990; Giddens, 1992). This was to both interrogate how these relationships practices are understood and lived in same sex relationships as well as to consider whether and how they are enacted differently or in similar ways when the relationship is abusive and regardless of gender and sexuality. NVivo7 was used to aid coding of the themes. Codes were tested and collapsed to identify three key features of abusive relationships: types of abuse; relationship practices and spheres of power and control; and narratives of love. A separate reading of whole transcripts was also carried out in relation to individuals who took part in the questionnaire survey, which highlighted that individuals were more likely to focus on a previous relationship (before the previous 12 months) when talking about abusive experiences (see also Lie and Gentlewarrier, 1991; Turell 2000).

As discussed above, we do not claim that the interview accounts given were the ‘truth’. They were in any case dependent on respondents’ memories and their tendency to tell relationship stories that were necessarily selective and put together with hindsight and from their own perspective. Participants, however, also made active choices to take part in the research, about which relationships they wished to give accounts of and about what to tell about those relationships. As Gabb (2008, 52) argues: ’the selection of a particular story, the details that are emphasised and the manner in which the story is told all signify and represent choices made by the participant’. Thus we can suggest that while memories and hindsight might provide particular lenses on relationship experiences that make those experiences unstable as (absolutely) factual or truthful, it might also be the case that particular experiences because of their shocking or harmful impacts (or alternatively their kind, loving impacts), remain ‘alive’ in people’s memories and act as ‘critical moments’ that can be selected as authentically representative of a relationship experience. Given the topic we were exploring we also encountered respondents for whom the interview acted as a cathartic experience insofar as in the telling of a relationship story in the context of our study they understood for the first time that they had experienced abuse which resulted in some distress (see also Kelly, 1988; Ristock, 2002b).

Thus, the interview data collected was the result not just of participants’ accounts but also the result of an interaction with us as interviewers asking particular questions and so shaping the stories that were told (Duncombe and Marsden, 1996). Nonetheless, we present (p.51) them as stories that have validity in that they provide accounts that are ‘a necessary element of knowledge of gendered lives and actual power relations’ (Ramazanoğlu with Holland, 2002, 127) and thus provide insights into the ways in which relationships can be understood. One of our key intentions with this research project was to enable new stories to be told (Plummer, 1995) about DVA in the UK context. While, as we have said in Chapter One, there have been some pioneers raising their heads above the parapet to signify that DVA is an issue in lesbian relationships (Hall, 1992; Taylor and Chandler, 1995), it is also the case that there has been silence about this issue across LGBTQ communities. Because of this we approached participants as willing audiences for their abusive relationship stories and conveyed our willingness to believe and honour their accounts as authentic and our intention to use the research and endeavour to make changes in wider society so that others might benefit. In this way we further sought to enact the feminist principles underpinning the work (for example, Ristock, 2002c; Harrison et al, 2001).

In total, 68 interviews were conducted with 20 lesbians (including one who identified as a trans woman), 19 gay men, 14 heterosexual women, nine heterosexual men, three bisexual women and three queer women. Of the lesbians, bisexual, queer and heterosexual women, 19 gave accounts of female same sex DVA relationships and 13 gave accounts of heterosexual DVA relationships. Just over a half of the gay men and a third of heterosexual men said that they had experienced abusive behaviours in their same or opposite sex intimate relationships.

Respondents’ ages ranged from 19 to 64 years of age. Lesbians, bisexuals and queer women’s ages ranged from 19 to 54 years, gay men from 20 to 64 years, heterosexual women from 20 to 59 years and heterosexual men from 20 to 59 years. Most were between the ages of 20 and 59 years old. The overwhelming majority of participants identified as white (with one identifying as white French) or white British. One lesbian identified as black British and one identified as African. Although the survey sample indicates that we were able to recruit a profile that reflects that of the UK population in terms of ‘race’ and ethnicity, we were singularly unsuccessful through the survey at recruiting black and minority ethnic people to the interviews (only one lesbian volunteered through this route). We then pursued several different routes to recruiting participants from these groups, including placing adverts in mainstream newspapers read by particular groups; contacting websites that targeted members of these groups and snowballing through personal and professional networks. At the end of this process we were only successful in recruiting one African lesbian. (p.52) Our status as a research group that was all white will undoubtedly have had an impact on our recruitment in this area. The need remains for work to be done in the UK with LGBTQ people from black and minority ethnic groups on DVA (see Hester et al, 2012 for some initial work in this area).

Five respondents identified as having a disability. In general, the sample was educated with incomes at or above the average. Incomes ranged from under £ 10,000 to over £ 60,000. In parallel with the profile of those who completed the survey, most men earned £ 21– 30,000 with the highest earner being a heterosexual man while most women earned £ 11–20,000. Most respondents were educated to degree or above with women slightly more likely to have higher educational qualifications. However, women were also more likely to only have GCSE or A levels. It is interesting to note that while women as a group were as likely to be educated as men their incomes did not reflect this.

It is important to reflect on the knowledge that can be ascertained about DVA from a sample that is for the most part, white, and, at least at the time of the interview, relatively well resourced in terms of education and income. Clearly the accounts of DVA will not be inclusive of the experiences of those who are not white, nor who are less well resourced. That the men were, on the whole more resourced than the women will no doubt have provided more choices about their response to their experiences, for example that they might have found it easier to leave an abusive relationship (see also Aguinaldo, 2004). Yet a complicating other factor in this approach to acknowledging the positionality of respondents who said they had experienced DVA is that of the age they had been when they had done so. Only one person we interviewed was still in the relationship they described as abusive. The African lesbian, Zoe, explained how she believed that she had made changes in her relationship such that it was no longer abusive. Everybody else who described abusive relationships was talking about experiences in the past, some of which had occurred several decades in the past. As will be discussed more fully in the following chapters, age is an important factor in understanding experiences of abuse, particularly but not exclusively in same sex relationships. Acknowledging how age and resources (or ‘social capital’), defined here in terms of education and income, might intersect in ways such that the former dictates the extent of the latter must be done to make sense of their accounts. It might be that their youthful age and corresponding lack of relationship experience had more salience in shaping experiences of abuse than resources, for example.

(p.53) Other differences in gender might have also influenced participants’ ability to respond to an abusive relationship. Typically, the women, regardless of sexuality had lived with their abusive partner and had longer abusive relationships while the gay men typically did not live with a partner (only three out of nine interview accounts of abusive relationships were in cohabiting relationships) and had shorter abusive relationships. Living together (nine of those in female same sex relationships), being married (eight of those giving accounts of abuse in heterosexual relationships), and parenting children, (seven of those who were married and seven of those giving accounts of female same sex relationships) will have been factors making it more difficult to leave an abusive relationship. Here, it is possible that structured dependency, more associated with heterofemininity yet also evidenced in these lesbian, and female bisexual and queer relationships played a crucial role in these women remaining in abusive relationships for longer than men.


The COHSAR survey has provided the basis for a new generation of research and created much interest among researchers of DVA in the UK, US, Sweden and Australia. In the UK the COHSAR survey has been used as the basis for an important national study of teen relationship violence (Barter et al, 2009), and to inform cross sectional surveys regarding DVA of men accessing GP surgeries and sexual health clinics (PROVIDE, http://www.bris.ac.uk/social-community-medicine/projects/provide/).


  • We used a feminist epistemological approach as this enabled us to construct survey and interview schedules geared to exploring how processes of gendering and power might operate in similar or different ways in abusive same sex or heterosexual relationships.

  • Our research approach emphasised the constructions and experiences related to structural inequalities and oppressions. This allowed us to take into account the intersecting of potential inequalities or differences such as those associated with gender, sexuality, ’race’, ethnicity, age, disability, class, income and education.

  • (p.54) We developed a questionnaire that drew on existing national surveys of DVA, such as the Crime Survey for England and Wales, and incorporated further questions that might reflect to a greater extent than previous surveys ‘how we know’ about such violence and abuse in same sex as well as in heterosexual relationships. In addition, in order to move beyond the heteronormative, we drew on existing North American surveys that include questions specific to those in same sex relationships, such as the use of sexuality, access to medication and/or agreements around safer sex and sadomasochism to control a partner.

  • Ideally, the survey should have used a representative sample, but this was not possible in the UK, as there was no national dataset in existence with information about sexual identity and contact details or location of the individuals concerned. Instead, we carried out a large UK-wide ‘community’ survey as the most wide reaching and ethical alternative.

  • The survey questionnaire included sections on: personal demographic information; decision making and conflict resolution in own relationship; own experience of negative emotional/ physical/sexual behaviours including impact; own use of negative emotional/physical/sexual against partner including why s/he did this, help-seeking; and a final section asking whether respondent had experienced domestic abuse plus other questions eliciting views and opinions.

  • The survey, about ‘problems in relationships’, resulted in a final data set of 746 respondents who had been or were in a same sex relationship. The sample had a larger proportion of respondents who had experienced potentially abusive behaviours from a partner than the more general health surveys carried out within LGBTQ communities in the UK.

  • To explore the impact that abusive behaviours may have had on the respondent, the sub-sections on potentially physical, emotional and sexual abusive behaviours also included questions about impact related to the experience of those behaviours. This included physical and psychological impacts, effects on relationship quality and partner interactions, and whether acts may be seen as self-defence or retaliation.

  • (p.55) The ages of survey respondents ranged from 16 years to late 60s. Nearly two thirds identified as women, and more than a third as men, with a further four identifying as transgender, and one as queer. More than two-thirds of women identified as ‘lesbian’ and more than three-quarters of men identified as gay.

  • Sixty-eight interviews were conducted with 20 lesbians (including one who identified as a trans woman), 19 gay men, 14 heterosexual women, nine heterosexual men, three bisexual women and three queer women. The sample was majority white women, most of whom had experienced domestic abuse. All but one of the lesbian, gay, bisexual and queer respondents also took part in the survey. All the heterosexual interviewees were recruited separately.

  • The interview schedule provided an opportunity for respondents to talk about how love can be understood in relationships that go wrong as well as those that go right. The interview schedule was based around an exploration of a best and a worst relationship experience.