Abstract
This article investigates how ‘gender-based violence’ is taken up as a subject of research, and more specifically, how gender violence is understood and researched in the ‘developing world’. Based on analysis of the domestic violence module included in India’s National Family Health Survey-3 (NFHS-3, 2005–2006), I argue that current international trends in survey research on gender violence pay insufficient attention to place-based contexts and the working of structural forces. The domestic violence module used in the NFHS-3 and other demographic health surveys (DHS) derives from a measurement scale developed in the US in the 1970s, and deploys particular assumptions about what constitutes domestic violence. This quantitative survey measure universalises a conception of domestic violence as a personal problem, rather than a systemic social issue. As a travelling knowledge paradigm, the survey elides histories of development intervention in India that have exacerbated and produced women’s vulnerability to violence, and ultimately offers more development as the solution to the problem of gender violence. The mobility of this travelling knowledge paradigm derives from a particular form of biopolitics. By prioritising scope and speed, these surveys erase the complexities of gender violence, thereby significantly hindering not only efforts to understand this violence, but also those aimed at preventing it.
Introduction
Gender violence is inescapably related to material considerations: to control of the reproductive body and control of fertility, to uneven distribution of labour and resources, to exploitative production relations, to the articulation of caste with class, and to the logics of an uneven spread of capitalism.
‘Culture’ or ‘tradition’ cannot justify violence against women, and communities need to challenge norms that view violence as acceptable or as a private matter.
These quotations represent divergent representations of gendered violence emerging from critical feminist and postcolonial analyses, on the one hand, and development-driven research and interventions, on the other. They serve here to illustrate the central argument of this article—that developmental framings, produced by the United States Agency for International Development (USAID) and affiliated agencies, deploy a behavioural definition of domestic violence that fails to account for the complex material and social histories and realities of violence and vulnerability. This behavioural assessment of domestic violence emerges from a biopolitics of development, which aims to tame this complex social phenomenon into a countable problem. This knowledge paradigm also obscures development itself—that is, the history and present of technical interventions and restructuring intended to increase economic growth and alter social relations. Without attention to the ways in which these interventions have exacerbated women’s vulnerability to violence, the problem of domestic violence accounted for in development research can be presented as one solvable by more development.
Development research on domestic violence is relatively new. Starting in the late 1990s, development and global health actors working in India took up domestic violence in their nationwide public-health survey, the National Family Health Survey (NFHS). Domestic violence is understood in the most recently published round of this survey and data (the 2005–2006 NFHS-3) as primarily a behavioural problem. The characterisation of domestic violence as discrete acts of violence exerted by one individual against another depoliticises the issue by ignoring structural factors, power inequalities and social network dynamics, which all contribute to the relational production and maintenance of domestic violence as the exertion of power and control over an intimate partner. Understanding the divide between this development–research approach and feminist conceptions of gendered violence, such as Sangari’s quotation above, requires a deeper consideration of how exactly this survey instrument defines and measures domestic violence.
I argue that this survey mechanism is a travelling knowledge paradigm, which universalises global North ideas about domestic violence. The move to universalise divorces the understanding of domestic violence from place-based and structural processes. The mobility of the paradigm reveals a particular biopolitics of development, through which powerful institutions of governance attempt to contain and count a complex problem. Ultimately, this approach simplifies the problem of domestic violence in India, and the global South, presenting the problem in inter-personal terms—as a set of actions or behaviours committed by one individual against another. The scale used in NFHS-3 to measure domestic violence has been highly contested by feminist scholars and domestic-violence activists in the US for decades. Yet, its adoption into global demographic health surveys, including India’s NFHS-3, has been met with relative silence. My analysis of this survey mechanism as a travelling knowledge paradigm reveals the significance of ‘how’ domestic violence is taken up in this development/global-health research for determining ‘which’ understandings of domestic violence are normalised. I interrogate dominant research methodologies employed by global health and development agencies and argue that they limit our understanding of domestic violence in India and the global South. Resisting the circular logic of this travelling knowledge paradigm, I argue for a ‘relational’ understanding of violence, which requires that it be ‘measured’ in place, as processual, and as it articulates with systems of power and inequality.
My argument proceeds in two parts. First, I show that this particular quantitative survey instrument measures domestic violence in a static manner, as a set of actions taken by one individual against another. The measurement tool assumes that specific actions or behaviours have universal meaning (across space and time), and through this emphasis upon countable incidents, or ‘discrete acts of violence’, reduces a dynamic and relational social process to an interpersonal problem. I trace the travel of this survey instrument from the US, through a global mechanism of development and global-health research, to India’s NFHS-3. Second, I explore the political effects of this movement, as the NFHS-3 is taken up by development research to argue that the solution to domestic violence is more development. I demonstrate that the characteristics that allow for this particular model of domestic-violence research to travel from the US to India also account for its adoption into a developmental approach to gender violence. By removing domestic violence from its temporal and spatial contexts, the NFHS-3 effectively elides the role of prior development interventions in contributing to Indian women’s vulnerability to domestic violence, and instead proposes ‘more’ development as the solution. Both parts of this argument build from discourse analysis of the domestic-violence measurement tool, the NHFS-3 survey questionnaires and data reports, and related texts produced by agencies affiliated to NFHS-3, such as USAID.
The failure of the NFHS-3 to capture the nuances and place-based realities of domestic violence across India stands in stark contrast to the complex, yet grounded, understandings of gendered violence produced by Indian feminist scholars and women’s movements and NGOs. Sangari’s quotation, with which this article opens, is just one example of a much larger mobilisation in response to gendered violence in India. Since the turn of the 20th century, women’s and social movements across the country have galvanised campaigns seeking state and societal response to various forms of violence against women, most notably dowry murder, sati and rape (Gandhi and Shah, 1992; Gangoli, 2007; Kumar, 1993; Ray, 1999; Sinha, 2006). Indian feminists have also challenged anti-violence movements to pay more heed to everyday, more ‘mundane’ forms of violence, such as domestic violence, marital rape and sexual abuse of children (Ghosh, 2004). At the same time, these and other forms of gendered violence have been taken up in a range of scholarly analyses, with some qualitative studies focused on violence itself (Bhattacharya, 2004; Coomaraswamy and Perera-Rajasingham, 2008) and many others on the relationships between violence against women and colonial and post-colonial projects of imperialism and nation-building (Jayawardena and Alwis, 1996; Mani, 1998; Narayan, 1997; Sinha, 2006; Sunder Rajan, 2003). A number of provincial and regional surveys have also been undertaken across India, yielding important information about the complex manifestations of violence against women at the intersections of gender, class, caste, religion and education. This scholarly and activist work has combined to simultaneously politicise violence against women and to demonstrate the complexity of gender violence across time and space. Gender violence in India arises at the intersections of oppressive social, political and economic structures, and has clear linkages to histories of colonialism and development intervention (Kapadia, 2002; Sangari, 2008). This article brings these complex analyses of violence to bear in a critique of ‘gender violence as a problem of development’.
Travelling Knowledge Paradigms
All women would probably agree what constitutes a slap, but what constitutes a violent act or what is understood as violence may vary among women and across cultures. (Kishor and Johnson, 2004, pp. 5–6)
This quotation appears in a report written by affiliates of MeasureDHS, 1 the creator of the demographic and health surveys that use the Conflict Tactics Scale (CTS) to measure domestic violence. Murray Straus created the CTS in the 1970s for use in nationwide surveys of violence in US families conducted in 1975 and 1985. Straus contends that the accuracy of the CTS extends largely from its use of a list of pre-set acts of violence, to each of which the respondents answer with a ‘yes’ or a ‘no’ according to their experience in the last 12 months. The logic of measuring domestic violence through this list of discrete acts of violence is that, by separating specific actions, the contaminating influence of the subjects’ differential interpretations of broader concepts, like ‘violence’, is removed (Straus and Gelles, 1990). This emphasis on the comparability of discrete acts of violence forms the crux of the justification for using the CTS model, and the quotation above appears in the MeasureDHS report’s text and is also highlighted as an excerpted line in its otherwise empty left margin on page six. The contention is that certain acts (such as a slap) are understood as violence in all languages and places.
In this section, I trace the travel of the CTS model from the US to India and discuss its ‘global’ reach, documenting the survey mechanism’s reception, adaptation and continuity at three primary nodes along the arc of its mobility, that is (a) within the US; (b) as a component of demographic and health surveys across the ‘developing world’; and (c) as adopted into India’s own demographic and health survey, the NFHS-3. I argue that the mobility of the CTS model derives from particular notions of expertise and from the appeal of numbers as highly comparable across place because they are viewed as objective facts, uncontaminated by place-based or subjective biases. Further, debates surrounding the meaningfulness of the CTS measure (at its site of origin in the US) demonstrate the role of survey research in conceptualising ‘domestic violence’, especially given the multiple meanings of domestic violence circulating in the US. Lastly, in exploring the arrival of the CTS in India, I argue that the NFHS-3 is severely limited because it frames domestic violence as an interpersonal problem, erasing social and economic processes and specific geo-histories.
Debating the Conflict Tactics Scale in US Research on Domestic Violence
Prior to the 1970s, much of the research on domestic violence in the US involved localised surveys and qualitative projects. At the time, researchers interested in measuring the extent of the problem often relied on a combination of crime statistics and statistics derived from the records of domestic-violence shelters and perpetrator-treatment programmes. Partly in response to a growing realisation that many women did not seek help from shelters or report violence to the police, researchers in the 1970s began to develop new tools for measuring violence. During this period, sociologist Murray Straus created the CTS for use in the National Family Violence Survey (1975), which was ‘the earliest attempt to measure the incidence of violence in a large and representative sample of American families’ (Straus and Gelles, 1990, p. 3). The CTS model includes 18 measures of discrete actions considered to fall into three modes of handling conflict in intimate relationships: reasoning, verbal aggression and physical violence. The instrument (used both in interview-style surveying and mail-in surveys) provides a list of behaviours as potential actions taken by either partner in a relationship in response to an interpersonal argument. A prelude to this list of behaviours also explicitly frames them as ways of resolving conflict or couples ‘trying to settle their differences’ (see Table 1, left column, for a reproduction of the original CTS instrument). 2
Comparison of CTS and NFHS-3 Questions about Violence1
The CTS and modified versions of it continue to be used to measure the prevalence of domestic violence in the US, as well as in at least 32 other countries across the world (Straus and Mickey, 2012). However, for more than 30 years researchers and anti-violence activists have critiqued this model, making it the centre of a heated debate about whether or not women’s use of violence against male partners is on par with men’s use of violence against female partners (DeKeseredy, 2011; DeKeseredy and Schwartz, 1999; Straus and Gelles, 1990; Tjaden, 2006). I interrogate an issue at the core of many challenges to the CTS: Straus’s contention that pre-determined behaviours measured as ‘a limited set of violent acts’ (a slap, a kick, a punch) provide the most valid data on ‘partner violence’, and that if/when contexts are measured they ought to be measured ‘separately’ from the occurrence and frequency of these incidents of violent behaviour (Straus, 1990, pp. 52–57). 3 As other scholars have demonstrated (DeKeseredy, 1993, 2011; Johnson, 2008), this emphasis on discrete actions, separated from their contexts, sheds light on only one type of intimate partner violence, and arguably this is not the form of interpersonal violence typically depicted in popular understandings of ‘domestic violence’ in the US.
Straus (1990, p. 52) argues that ‘the acts in the CTS have been determined to be almost universally meaningful in in-depth interviews’, although he does not cite the studies to support this claim. The validity of this claim will be examined further below, with regard to the CTS’s travel across geographic and linguistic borders. However, even accepting some degree of shared understanding of the CTS violent acts among English-speaking respondents in the US, the absence of context surrounding the use of a ‘kick’ or ‘slap’ by an intimate partner creates substantial ambiguity about the significance of that action, and whether or not it constitutes ‘domestic violence’. As Lindhorst and Tajima (2008, p. 364) argue,
…just knowing that one partner has hit another does not necessarily mean that IPV [intimate partner violence] has occurred. If IPV is conceptualized as encompassing (a) a pattern of behaviors that (b) yields adverse effects perceived by the victim (e.g., injury, harm, fear, intimidation, etc.) and that is (c) motivated by the perpetrator’s need for power, then measuring the physical act alone is insufficient to accurately measuring the construct.
In order to better capture these aspects of domestic violence, Lindhorst and Tajima (2008) suggest inserting what they call ‘contextual measures’ into survey research, which means situating acts of violence relative to the immediate surroundings, to subjective meaning-making around the acts, as well as in relation to contexts of history, culture and oppression. The absence of such contextual measures in domestic-violence survey research has political effects: it obscures the evolution of an abusive situation over time, the operation of a dynamic of power and control, and the significance of histories and structures of dominance and oppression. Dominance and oppression can both produce intimate partner violence and subject marginalised communities (including both men and women) to a whole spectrum of violence (such as state violence, structural violence and war).
One glaring example of the limitation and confusion produced by measuring only discrete physical acts of violence is highlighted in the controversy over results from some CTS studies that suggest that women use violence about as often as men in intimate-partner conflicts. Straus argues that his scale proves this to be a reality, at least in the US, and contends that critics of his model simply refuse to accept this reality (Straus, 1990). Quite a number of scholars, including feminist critics, have demonstrated that in fact this result is produced by the lack of context when using the CTS as the sole measure of domestic violence (Bograd, 1984; Dobash et al., 1992; Morse, 1995; Rhodes, 1992). When Straus’s scale is used with context measures or more qualitative questioning, it has been found that women’s use of violent action often arises from long-term victimisation by their intimate partner; it thus falls into the category of what advocates and scholars call ‘violent resistance’ (DeKeseredy, 2011; Johnson, 2008). Proponents of the CTS model argue that quantitative measurement of distinct behaviours allows for comparison across different experiences and avoids the more messy ground of subjective meanings of ‘violence’ broadly writ. Yet their own interpretations of the CTS data construct the meaning of ‘domestic violence’ based upon a partial view of its lived reality. The expectation is that the CTS can and does travel across boundaries of difference within the US. Yet significant evidence from other studies suggests that victims’ and families’ intersectional identities strongly influence both vulnerability to domestic violence and how violence plays out in relationship dynamics, including very different meanings being attached to the same action (Connelly et al., 2005; Lindhorst and Tajima, 2008; Sokoloff and Pratt, 2005). It is important to note here that meaning is of course still being attached to CTS violent acts, but, instead of measuring the subjects’ meaning-making, with the CTS, meaning is attached by researchers and analysts, who assume their definitions of violence translate across a wide range of different women’s experiences. At the same time, they also assume that the reporting of certain behaviours by intimate partners (regardless of the dynamics of their relationships with not only one another but with others inside and outside the home) demonstrates the presence or absence of domestic violence.
The debate surrounding this model demonstrates the significance of how domestic violence is measured for theorising the social phenomenon itself. A number of scholars—inserting themselves into the middle ground of this debate surrounding the CTS—argue that Straus and his critics are actually attempting to measure different types of domestic violence. Johnson (2008) proposes a typology of domestic violence with three primary categories: intimate terrorism, violent resistance and situational couple violence. 4 Those who subscribe to the Straus/CTS school of measurement are really measuring situational couple violence—the kind that arises from familial conflict and is perpetrated by both partners relatively equally. On the other hand, feminists and advocacy-based researchers understand domestic violence more in terms of what Johnson (2008, p. 3) terms ‘intimate terrorism’, which entails expansive and long-term control exerted by one partner over another through physical, emotional and sexual violence. In heterosexual relationships, men commit the bulk of intimate terrorism, and Johnson (2008, pp. 1–3) also contends that it is this form of violence that the American public typically associates with the term ‘domestic violence’. The third aspect of Johnson’s typology, violent resistance, captures the violence committed by victims of intimate terrorism (primarily women) against individuals who have been using violence to control them. It is a way of describing a woman’s violence when ‘fighting back’ against a partner who has been controlling and abusive over the long term (Johnson, 2008, p. 48). Thus, even in the survey model’s site of origin—the US—substantial ambiguity persists with regard to what exactly the CTS data can tell us about domestic violence. And even before the instrument crosses a geographic boundary, evidence both from studies employing the CTS and those not using it demonstrates the possibility—even likelihood—that measuring primarily discrete acts or behaviours actually strips away the dynamics of power that characterise domestic violence as a process.
A ‘Global’ Measurement of Domestic Violence? Demographic and Health Surveys and the Modified CTS
A degree of uncertainty remains as to whether or not the CTS actually measures what is commonly labelled ‘domestic violence’ in the US. Why, then, has the model been exported through US-led development interventions to dozens of places across the globe, including India? I argue that the mobility of this survey mechanism is tied to its allure as a standardised, numerical measure, and therefore to the ordered way in which it presents domestic violence as an individualised, technical problem. The survey travels, then, as a calculative, rational technology for a particular form of biopolitics. Following Foucault (2008) and the expansion of his ideas of biopower and biopolitics by feminist and critical social scientists (Braun, 2007; Li, 2009; Mohanty, 2011), I employ this term to denote technologies of modern, liberal governance that seek to control and discipline individual bodies and entire populations under the assumption that responsibility for the sustenance of life belongs to those who govern it. In this instance—a survey mechanism used by governments, development institutions and global-health agencies— biopower manifests through a particular discursive construction of domestic violence that renders the bodies and behaviours of both victims and perpetrators in need of corrective interventions. As such, this survey is another instance of apolitical development measurement that reduces violence to a technical concern that needs more development, while simultaneously obscuring broader political and economic relations and interventions producing violence.
The mobility of the CTS model seems to derive more from the appeal of its abstraction of domestic violence as an interpersonal problem than from any consensus as to its effective representation of domestic violence as a complex social phenomenon. When employing the CTS, US researchers rarely use the scale alone and often combine it with additional measurements particular to the case study. However, when the model travels, it typically travels in a standardised form from which contextual measurements are absent. More complex surveys thus may be created using the CTS, by adapting it and adding to it in order to capture more nuances in subjects’ experiences and interpretations of violence. The problem, then, of exporting the CTS on its own is the assumption that it can stand alone as a useful measurement of domestic violence. And in fact, the continued use of the CTS itself (with or without added measurements) seems to be in the role of providing a basic, quantifiable measure of intimate violence. By producing single numbers—in the form of percentages of reported violence in a place, or for a certain group (Canadian women, for example)—the CTS offers a representation of the ‘problem’ of domestic violence that is legible to governments and accessible to the public as a statistical sound bite. The move to ‘count’ domestic violence in this manner is political— it tames a complex problem into a technical, governable issue. As a form of biopolitics, the simplification and quantification of domestic violence as an interpersonal problem also authorises the entry of powerful institutions (of the state, of international development) into homes and communities.
By the late 1990s, domestic violence had gained traction in international development and global health circles as a significant threat to women’s health and rights and to national development. At this time, the creators of a commonly used form of demographic and health surveys—MeasureDHS—decided to include domestic violence in its surveys (Kishor and Johnson, 2004, p. xv). MeasureDHS is an affiliate of USAID, which funds its operations and the deployment of demographic and health surveys across the ‘developing world’. In taking this model global, MeasureDHS selected most, but not all, of the CTS ‘violent acts’. The MeasureDHS version re-phrases some of the terms used on the list of violent acts and it also rewrote the remarks preceding the list of violent acts. In this new DHS form, the framework of familial ‘conflict’ is not overtly deployed. Instead, women are asked about violent acts in the context of their marital relationship—simply, whether their husbands have ever done certain things to them. 5 The primary aspect of the CTS, which is retained in the DHS module, is the choice to measure domestic violence by counting specific acts, as well as the character of many of these acts (such as slapping, kicking, choking, beating) and their frequency (within 12 months prior to the survey date). The coding and analysis of DHS by the agencies who conduct the survey also retain some aspects of the CTS model, such as coding for ‘severe’ and ‘less severe’ violence based upon predetermined ideas of which acts qualify as severe.
Explanations for the adoption of Straus’s CTS for measuring domestic violence in the ‘global’ demographic and health surveys appear in reports by MeasureDHS and affiliated agencies that use its survey mechanism. The common refrain in these reports echoes Straus’s own justification for the mechanism’s validity—that is, it measures ‘several discrete acts of violence’, pre-defining the actions or behaviours that might constitute a woman’s experience of violence, and explicitly not asking her any questions with broader terms, such as ‘violence’ or ‘physical mistreatment’. As Kishor and Johnson (2004, p. 5) explain in their MeasureDHS comparative report, ‘By asking separately about specific acts of violence, the violence measure is not affected by different understandings between women of what constitutes violence. A woman has to say whether she has, for example, ever been “slapped”, not whether she has ever experienced “violence”.’ Herein lies the crux of argument that the simplification accomplished through the CTS model ‘makes the data comparable across “cultures” and places’. Proponents argue that in order to count ‘violence’, we must be assured that we are counting the same thing; thus the survey pre-defines violence as specific acts and behaviours to determine prevalence. Yet, the model normalises domestic violence as interpersonal violence defined through a series of questions originally meant to gauge the use of violence in the resolution of ‘conflicts’. The relevance of this definition in different place-based or community settings is not questioned. Rather, the survey mechanism travels by virtue of its ability to produce ‘numbers’, numbers that can then be easily compared to produce a ‘global’, or God’s eye, view of the problem. A problem thus defined becomes amenable to the technical solutions of government, and, as will become clear in the second part of this article, this God’s eye view of domestic violence produces the imperative of more development intervention by obscuring the geo-historical and structural forms of violence at work in the colonised world.
Not only is measuring the prevalence of domestic violence ‘anywhere’ prioritised, but the ability to compare rates of violence in this manner is used to reinforce the assertion that the CTS model increases the validity of domestic-violence measurement. Here, the MeasureDHS reports employ language that directly contrasts the ‘validity’ and ‘effectiveness’ of this model with alternate models for researching violence that focuses more on ‘meaningfulness’ and the particularities of violence in a given situation or place. While the limitations of quantitative survey measurement are noted, the desire for numbers to represent large-scale prevalence is sufficient justification for de-emphasising situational and place-based contexts in research on domestic violence. This choice of measurement affects how domestic violence is conceptualised by a host of actors and institutions. However, the grounds for using the CTS model in a global-survey measurement of domestic violence can also be questioned based on its own internal logic.
The choice to use the CTS in MeasureDHS surveys presumes, to some extent, that an act of violence pre-exists (or exists aside from) a subject’s understanding of its meaning (in that situation, place, relationship, etc.) and that it can have the same meaning across cultural, linguistic, class, race, age and other differences. In fact, the acts of violence included in the CTS are understood as such through specific, place-based histories of activism and scholarship on domestic violence in the United States. Thus, while there may be some level at which a ‘punch’ is understood on similar terms across place, we understand very little about ‘violence’ if we only measure how many times a woman has been punched by her husband in the past 12 months. For example, ‘punch’ may well mean something different to different women, and it may be of less significance for some women than a pattern of emotional abuse and isolation from loved ones. Additionally, we will understand very little about violence ‘in place’ if we assume that standardised violent acts can thoroughly or properly account for the specific tactics of abuse used in specific places, cultural or linguistic groups, class groups, etc. For example, India’s NFHS-3 was translated into 17 different languages. Can we know that the translation of ‘kick’ into each of these languages provides a sufficiently comparable action? Are women responding to oral questionnaires given in different languages truly reporting ‘the same’ experiences of domestic violence? Or, perhaps, in some social groups—often higher on the wealth and/or education scale, where the physical markers of violence (such as those resulting from a hard punch) draw negative attention from family or neighbours—tactics involving food or sleep deprivation may be more common, but are not being measured by the NFHS-3 mechanism.
Finally, this assumption of the universality of the CTS measures makes a significant leap from a global North setting to the deployment of DHS surveys conducted across the formerly colonised, currently ‘developing’ world. I emphasise this last point not to suggest that colonisation or development ‘cause’ domestic violence; rather, understanding domestic violence as a complex social phenomenon and understanding women’s differential vulnerability to it requires a deeper consideration of historical and contemporary structures and processes of dominance and oppression. This is true both in the global North and in the global South. However, in the case of the latter, the linkages between colonisation and contemporary assemblages of development and global health require serious consideration.
India’s National Family Health Survey-3
In tracing the travels of the CTS from the US to the ‘developing world’, and specifically to India, I demonstrate how a narrow concept of domestic violence, relying on an interpersonal understanding of domestic violence, obscures what is in actuality a dynamic social process. When domestic violence is understood simply as a series of violent acts that men use against their wives, the ways in which a dynamic of domestic violence is embedded in and produced relationally with various forms of oppression and dominance is lost. In the case of spousal abuse in a heterosexual marriage, a dynamic of domestic violence refers to a pattern of power and control enacted by utilising any number of tactics and tools, often drawing upon the privileges and powers afforded to some men through patriarchy, class or caste status, and hierarchical familial and kinship structures. These structures and institutions of power are also themselves dynamic—they shift over time and in relation to one another. For example, Indian feminist scholars have demonstrated the ways in which regional patriarchies have shifted and reformed in relation to colonialism, the perceived threat of ‘modernity’, and economic development initiatives such as Structural Adjustment Programmes (Arora, 1999; Kapadia, 2002; Sangari, 2008). Such shifts also often affect particular women’s vulnerability to intimate violence as well as the forms and practices of violence to which women are subjected. These nuances are more than merely ‘added information’ about violence. Rather, these factors can significantly shape what kinds of actions and relationship dynamics are experienced as violent, and can offer vital insight into the kinds of social change that might actually prevent violence.
While the NFHS-3 and other surveys using the CTS model might intend to establish a ‘baseline’ and then measure the domestic violence module again at regular intervals, this cannot provide a complete picture over time. In those future surveys, it will not be the same women who are interviewed about their experiences, so again we get only a set of snapshots of certain moments in time. And while the survey does include a significant question about a woman having experienced a violent act in her lifetime, it still cannot capture a dynamic of domestic violence, nor that one violent incident may in fact not actually be evidence of an abusive relationship. In the case of the NFHS-3, this limited temporal frame is also one within which it is very difficult to see the changes brought through development interventions. For example, present conditions may have been significantly affected by past interventions—such as the scaling down of social service provision resulting from past liberalisation or structural adjustment measures. Thus, without capturing the historical underpinnings of present conditions and structures, the impacts of development are invisible and the current situation (of high rates of domestic violence) is instead assumed to result from a ‘lack’ of change (in ‘culture’, men’s behaviour, etc.). The significance of this erasure and related assumptions are explored in the next section.
The limitations of this domestic-violence measurement are further compounded by both the NFHS-3’s primary reliance on the CTS and the meagre attempts to customise this measurement for use in India. The use of the CTS in the NFHS-3 carries forward the notion that we can or ought to try to understand domestic violence as a set of ‘distinct’ actions or behaviours. Even if we then attempt to compare the determined data or ‘rates’ with other, separately measured variables, the concept of domestic violence remains individuated—contained in interpersonal relationships and households and separated from social networks, institutions and structural processes. The types of changes and additions made in order to adapt the standard demographic health survey to India themselves reflect an assumption that domestic violence is primarily an interpersonal problem, and thus they retain the logic of the CTS as a travelling knowledge paradigm.
First of all, in the NFHS-3 the CTS measures still ‘stand alone’ as a set of questions about pre-defined, discrete acts of violence. The NFHS-3 also includes a handful of questions about household dynamics and women’s agency, but these are not asked about in relation to violence or abuse. Therefore, as the survey stands it cannot sufficiently capture the dynamic, relational processes through which domestic violence arises and is either perpetuated or disrupted. This travelling knowledge paradigm—that is, the particular definition of domestic violence born in the US and exported through the use of the CTS in demographic and health surveys—remains intact in the NFHS-3. A joint report authored by the International Institute for Population Sciences (IIPS) in Mumbai and Macro International (MI) in the US suggests that the choice to frame the NFHS-3’s violence questions based upon the CTS model was indeed due to a desire to focus the measurements upon ‘specific acts of violence’:
Asking about the experience of specific acts of violence, rather than about the experience of violence in general, has the advantage of removing from the measure of violence the effect of variations in the understanding and interpretation of what constitutes violence. A woman has to say whether she has, for example, ever been slapped, not whether she has ever experienced any violence. Most women would probably agree on what constitutes a slap, but what constitutes a violent act or is understood as violence may vary among women, as it does across cultures. In fact, summary terms such as ‘abuse’ or ‘violence’ were avoided during the NFHS-3 training of interviewers, and not used at all in the title, design or implementation of the module. (IIPS and MI, 2007, p. 495)
The second aspect of my critique brings a critical eye to the kinds of questions included in the NFHS-3 as augmentations to the CTS measurements. I argue that these additions for India—questions about women’s attitudes, household decision-making abilities, and ‘help-seeking’ behaviours—focus primarily on the interpersonal/marital relationship. The survey systematically excludes measurement of structural factors and changes, such as shifts in labour relations, market regulation, and state provision of social services, or even trends in familial and kinship relationships in terms of living arrangements or social-network support. The effect is the perpetuation of the individualised notion of domestic violence established through the use of the CTS. For example, while measuring a woman’s involvement in household decision-making may shed some light on power relationships within her family, pairing this data with a measure of how many times she reported experiencing certain acts (pre-defined as violence) tells us nothing of the dynamics of those experiences or the characteristics of a possible relationship between decision-making and vulnerability to violence. There is evidence to suggest, for example, that women’s increased status or power within the household (that is, through income generation and/or increased mobility outside the home) may actually increase their vulnerability to domestic violence (Karim, 2008).
Development as an Absent Presence in India’s NFHS-3
In the course of its worldly travels, the CTS has become an instrument of development and global health research. This is no accident, no coincidental meeting in the halls of Washington DC en route to Delhi, Dhaka or Kampala. The same characteristics that grant this model its mobility also make the CTS an ideal tool of development. As Nikolas S. Rose (1999, p. 197) argues, ‘numbers have achieved an unmistakable political power within technologies of government’. The ‘clean’, segregated and quantified measurement of domestic violence offered in the CTS suits the needs of a development apparatus that continually frames technical problems for which it might offer technical solutions (see Ferguson, 1990). My analysis of the NFHS-3 demonstrates that the concept of domestic violence deployed in the survey instrument and in related reports and texts is indeed a ‘problem’ framed as solvable through economic and social development.
The presence of development in the NFHS-3 seems clear enough: it is a survey created and funded primarily by development agencies and implemented in India with the ‘technical assistance’ of USAID-affiliated agencies. The Indian state acts through the Ministry of Health and Family Welfare, which ‘coordinates’ international funding and ‘technical’ expertise with local implementation (through the Mumbai-based IIPS). All these institutions have their feet firmly planted in development, specifically through their self-described investment in a set of projects and interventions seeking improvement in public health, individual and family health metrics, family planning, and nutrition, all of which they aim to measure and then to improve through varied means (from economic growth to direct health and social programmes).
NFHS-3 is conceived and implemented squarely within the institutions of development in India. The survey stops short, however, of any serious consideration of development practices within the survey mechanism itself. Similarly, discussions of NFHS-3 results mostly avoid the subject of how previous development interventions might have shaped the data collected therein. As such, the survey actually elides measurement or analysis of the possible effects of prior development interventions on how domestic violence plays out for diverse communities in India. Significant evidence exists to demonstrate that a number of development interventions in India, such as structural adjustment, have actually increased women’s vulnerability to poverty and gendered exploitation (Arora, 1999), including their vulnerability to various forms of gender violence (Coomaraswamy and Perera-Rajasingham, 2008; Kapadia, 2002). Yet, precisely because these histories are invisible in the survey measures, development can continue to be framed as a solution to domestic violence. Development’s curative capacity in this case also relies upon the discursive construction and targeted measurement of domestic violence as a problem associated with ‘tradition’ and with the conditions of underdevelopment (that is, poverty). In fact, the NFHS-3 reaffirms familiar and misleading tropes of the primary causes of violence (such as ‘culture’). In effect, meaning and causality (not being measured in the survey itself) are inserted by the surveyors in a manner fitting with their own objectives and funding imperatives (that is, economic growth and social-development programmes).
I demonstrate below how the NFHS-3 measurement of domestic violence, together with the travelling knowledge paradigm it relies upon, becomes a technology of apolitical development practice (Escobar, 1995; Ferguson, 1990). I reveal some ways in which this particular universalised notion of domestic violence serves to bolster a development imperative. In turn, this abstraction from place and from histories of colonisation and modernisation reinforces problematic assumptions about which ‘contexts’ are relevant for understanding domestic violence, and reifies ‘culture’ as the starting and ending point for explaining the persistence of gender violence in India.
Erasing Development’s Past to Make Way for its Future
I have argued that the CTS travels well in the DHS and India’s NFHS-3 for a reason. Put simply, it quantifies domestic violence in terms legible and useful to powerful institutions and governments, whose claim to power relies in part upon their ability to manage populations and demonstrate some commitment to solving social ‘problems’. This choice to employ the CTS is also a choice to intentionally measure domestic violence separate from its contexts. 6 This move allows a depoliticisation of development. When socio-cultural and political-economic processes and their effects (that is, development discourses, policies and programmes) are deemed ‘separate from’ domestic violence, then we cannot understand their relational production ‘together’. For example, we can look at reported domestic-violence rates compared with women’s employment (each measured in NFHS-3, but in separate sections). We might look at the north-central state of Uttar Pradesh, which had a reported 33 per cent of married women employed at the time of the survey and a reported rate of domestic violence of 45 per cent. 7 The southern state of Tamil Nadu, on the other hand, has a significantly higher number of employed married women—48.4 per cent—but a similar rate of domestic violence at 44.1 per cent. We might wonder from this slight variation in reported violence if higher employment rates translate into some, if meagre, reduction in violence. Yet, if we look at Rajasthan, women’s employment is seen at 56.6 per cent, but violence is reported by women at 50.2 per cent. Looking at these variations by state, we might not see a statistically significant pattern emerge to indicate that there is some clear relationship between women’s employment and their likelihood of experiencing violence. And indeed, whatever relationship may exist is likely to vary by region. The problem, however, is also that this method of measurement (employing the CTS) intentionally ‘avoids’ asking respondents questions that seek to understand such relationships in the context of the individual’s life experiences and perspective. Married women’s employment is also fixed in time in the NFHS-3, without measured consideration of the shifts in such employment that might be caused, for example, by the informalisation of women’s labour (Bhatt, 2006; Kantor, 2009).
It is precisely within these relationships between experiences of domestic violence (or, importantly, the absence of such experience) and social, political and economic processes that we might see some effects of development intervention. 8 Women’s employment and un/paid labour, for example, has received tremendous attention in the past 40 years within development literature, policy and programming in India (John, 2013; John et al., 2006; Kabeer, 1994). Most recently, in India and across South Asia, development discourses and programming have honed in on women’s ‘entrepreneurial capacity’, and many champion microfinance as the last best hope for ‘empowering’ women, not just economically, but within social and political structures as well (Kabeer, 2005). A number of studies, primarily in Bangladesh, have also explored potential relationships between women’s participation in microcredit schemes and their vulnerability to domestic violence. The Grameen Bank model, which intentionally excludes men from microloans, has been found to exacerbate domestic violence for wives who receive the loans for income-generation activities (Karim, 2008). And looking at women’s employment more broadly, Agarwal and Panda (2005, 2007) suggest that while no causal linkage can be made between women’s employment and reduction in domestic violence, there is such a linkage for women’s ownership of land. This body of research reveals that development programmes (for example, microcredit) that aim to empower women through fuelling their income-generation and labour capacity may not always have this intended effect when it comes to women’s safety and freedom from violence in intimate relationships. Yet, in the NFHS-3 we have no means of even scratching the surface of such development effects. There are no questions about the kinds of development schemes women or men have participated in, and no attempt is made to measure any kind of change in structural conditions that could contribute to the presence or absence of domestic violence in a woman’s life. We cannot see development or labour-market changes, only whether or not a married woman reports having paid employment.
It might be suggested that such in-depth analysis of domestic violence is not the role of this type of a survey. I agree that the NFHS-3, employing the CTS model, is an insufficient measurement tool for documenting or understanding a complex issue like domestic violence. However, my critique does not end but begins from this point. As demonstrated in the first part, the travelling knowledge paradigm framing domestic violence in the NFHS-3 presents it as an inter-personal problem, artificially separated from historical and place-based contexts. Furthermore, the NFHS-3 makes no attempt to measure structural change or conditions resulting from social and economic programmes of state and non-state development entities. This move actually makes it possible to suggest that the ‘solution’ to current rates of domestic violence in India is ‘more development’. USAID describes itself as
…positioned to play an important role within the U.S. Government’s efforts to prevent and respond to gender-based violence. The Agency’s development assistance program encompasses the entire range of activities that help make women and girls less vulnerable to violence and its consequences, through poverty reduction, access to safe migration, peace and reconciliation processes, safe and supportive education systems, the provision of appropriate health services, and promotion of the rule of law as well as equal rights and economic and political opportunities for women. (USAID, 2006)
Here USAID development projects are presented as a holistic form of anti-violence. The list of programmes, however, does not mention the kinds of economic restructuring interventions so often at the forefront of development-funding packages. And in this same report, when economic transitions are described as part of the upheaval that fuels ‘cultures of violence’ in ‘some parts of the world’, the role of US- and foreign-led economic and market interventions in controlling the conditions of those economic transitions is also omitted. Without measured consideration of how particular development interventions might have affected women’s vulnerability to violence, ‘development’ itself is presented as the solution to the problem of gendered violence ‘in the developing world’. This assumption is made without the slightest consideration of widespread domestic violence and sexual assault in most so-called ‘developed’ nations. Furthermore, in this and other USAID publications on gender and development (USAID, 2006, 2009), the imperative to end gender violence is explicitly, and repeatedly, stated in terms of its negative impact on ‘other’ development indicators. That is, domestic violence is a deterrent to development. So not only is it a problem that development can solve, but one that must be solved in order for development to be successful.
The Problem and Predominance of ‘Culture’ in Explaining Domestic Violence in India
The fact that similar forms of violence take place inside and outside the family, that the family both absorbs and radiates violence, indicates that the public and the domestic are jointly formed, and that they are not amenable to a tradition-versus-modernity reading. (Sangari, 2008, p. 11)
Through the construction of domestic violence as an inter-personal problem, compounded by the absence of any significant measurement of structural processes, the NFHS-3 not only lends itself to the allure of a development-based solution to domestic violence, but also opens the door for culturalist explanations. Rather than incorporating time-place specific measurements of cultural factors as one aspect of measuring context, the NFHS-3 draws on the DHS model (for all ‘developing countries’) and includes a few questions about household decision-making, women’s ‘help-seeking behaviour’, and ‘attitudes’ towards domestic violence. These then become the basis for statements such as ‘Traditional gender norms, particularly those concerning wife beating, remain strongly entrenched’. 9 This invocation of tradition is one that understands cultural and social norms as static and assumes that a variety of justifications for, or normalisations of, domestic violence are ‘hold-overs’ of an archaic past, as yet unbroken by the liberating potential of modernity. Recent work on violence in South Asia draws our attention to the recent ‘adoption, expansion and spread of what are perceived as age-old, medieval or feudal customs’ (Sangari, 2008, p. 22; emphasis in the original). However, this work notes the dynamic nature of such practices, which have not continuously operated ‘untouched’ by time, but have re-emerged, changed, and spread in recent times. And while Sangari (2008) also describes gender violence in relation to shifting terrains of culture, she emphasises the interconnections of various forms of violence, including structural violence. She suggests, ‘Perhaps it is not India or South Asia that has a particularly pre-modern state or civil society but that alibis for violence shuttle between tradition and modernity in ways that can obscure the range of interconnections’ (Sangari, 2008, p. 25).
Both the measurement of whether or not men and women find domestic violence ‘justified’ in given situations and the measurement of women’s ‘help-seeking behaviour’ lend themselves to this culture-tradition trope. In the first case, the social factor that is measured (that is, ‘attitudes’ toward violence) maintains the definition of domestic violence as an inter-personal problem (as established by use of the CTS). From this, a ‘traditional norm’ is aggregated from the responses of women and men—significant percentages of whom report at least some ‘acceptable’ situations for the use of violence. The survey presents no evidence to confirm that these ‘attitudes’ in fact derive from long-standing norms, but the claim is made that they do. In the second case, women are asked only whether or not they have ever told anyone about the violence they have experienced, and if so, whom. Nowhere is it suggested or directly queried if women have been offered help by family, friends, neighbours or others. In this framing, not only are women responsibilised as the creators of their own liberation, but they are positioned as victims within social and cultural spaces that do not offer help or any real alternative to enduring the violence.
A wealth of scholarship and movement-based evidence demonstrates that regional patriarchies and cultural constructs influence women’s vulnerability to domestic violence in India. However, the model and interpretation of the NFHS-3 does not mirror the complexity of those analyses. Rather, it pre-defines domestic violence as a set of actions or behaviours and fails to account for the potential effects of prior development projects, especially those in the economic sphere. And through those measurements, intended to widen the frame for understanding women’s experiences of domestic violence (that is, attitudes and help-seeking), the NFHS-3 actually falls into a culturalist approach to explaining gender violence. In this I evoke Sangari’s critique that ‘such culturalism works as a code for tradition and religion, conflates religion and patriarchies with “culture”, and turns acts of violence into religion-driven third world pathologies or customary/sacred traditions’, and her proposal that ‘moving beyond culturalism entails looking at patterns, structures, conjunctures and constellations of gendered violence within each country in ways that are context-specific but not culturalist’ (Sangari, 2008, p. 2).
Conclusion
I have argued here that framing domestic-violence research through global health and development paradigms both limits understanding of domestic violence and depoliticises development. The story of the CTS as a travelling knowledge paradigm reveals the operation of biopower in this particular approach to researching domestic violence. Adoption of the CTS into global DHS surveys, and thus the NFHS-3, takes the model as a universal mechanism capable of producing results that are comparable across place and time. This supposed universality allows for its mobility, and makes the paradigm appealing to powerful institutions and states. Counting violent acts to produce statistical measures of the prevalence of domestic violence tames a complex problem to one (literally) manageable for governance—it becomes a social problem amenable to technical solutions. Part of the complexity that is erased, however, is the effect of development interventions and structural changes on vulnerability to and experiences of intimate gender violence. The usage of the CTS in DHS surveys and the NFHS-3 thereby allows USAID, the Indian state, and various partner institutions to claim that more development (in health, education, economic growth, etc.) will ‘solve’ this ‘traditional’ problem of domestic violence. The contestation of the CTS in the US, and the significant evidence suggesting that it does not measure what is commonly called ‘domestic violence’ in the place of its origin, seemingly has had no effect on the growing international usage of the CTS. In fact, authors of USAID, MeasureDHS, and NFHS-3 reports repeatedly explain the validity of the CTS as based in its intentional ignorance of the contexts surrounding violent acts, so as not to taint them with subjective meanings.
The CTS and its travel to India and across the ‘developing world’ draws critical attention to which ideas move through these types of travelling knowledge paradigms and what exactly makes those ideas so mobile. Yet this is not merely a concern with how we talk about or measure domestic violence. The institutions conducting this research produce policies and programme initiatives that impact the daily lives of women and men. This is not merely a quibble over who gets to define ‘domestic violence’ in India or in any one place. Rather, what is at stake here is our collective ability to understand various actors’ and institutions’ responsibilities for (non)violence in their homes, neighbourhoods, societies, nations and global communities. Prevention of gendered violence requires such an understanding.
This article moves beyond mere deconstruction of the CTS and NFHS-3 and their limitations for advancing understanding of the complexities of domestic violence in India (and really anywhere in the global South or global North). My aim in tracing the CTS as a travelling knowledge paradigm is, constructively, to argue for a feminist reclamation of knowledge-making around domestic violence as a ‘development’ issue. It is not enough that domestic violence has ‘made it’ on to the agenda of development writ large, or that public health officials at various scales are finally taking seriously the realities and effects of domestic violence for families and communities. These have been goals for feminists and women’s movements for decades, but that work is not yet finished. Feminist research on domestic violence, much of it deriving from ‘on-the-ground’ work with survivors and advocacy organisations, has already advanced our understanding of domestic violence as a social phenomenon, based in power and emerging at the intersections of gender, caste, class, religion, age and sexuality. But we need more in-depth, qualitative research on the relationships between domestic violence and ‘development’. This will entail a deeper interrogation of structural factors influencing vulnerability to gender violence. The goal must also be to connect these processual and structural aspects with everyday experiences and historical and spatial contexts. By insisting upon more than a mere ‘counting’ of incidents of violence, feminists can reclaim the grounds for substantive social change in response to intimate gender violence.
Footnotes
Acknowledgements
I am grateful for the advice and insights provided by Mary John regarding my research and analysis of the NFHS-3. In particular, our conversations prompted me to consider more deeply the significance of how domestic violence becomes a subject of research, especially development research, and therefore to draw attention to the travel of this measurement mechanism from the US to India.