Intimate Partner Violence and Forced Migration during Pregnancy: Structural Constraints to Women’s Agency

Abstract

Little is known about migration during pregnancy related to intimate partner violence (IPV). In this paper, we examine issues of agency in relation to pregnant women’s migrations in a high HIV prevalence area of Kenya. We qualitatively explored forced migration among pregnant women, using data from in-depth interviews, focus groups, and IPV screening forms. To quantitatively examine migration during pregnancy, we analyzed data from a prospective study of 614 pregnant women. The qualitative data revealed that women had varied responses to violence in pregnancy, with some being able to leave the marital home voluntarily as a strategy to escape violence. Others were ‘sent packing’ from their marital homes when they dared to exercise autonomy, in some cases related to HIV status. Quantitative analyses revealed that pregnant women who migrated were more educated, less likely to be living with a partner, and had fewer children than other women. Migration among pregnant women in Kenya illustrates the complexity of understanding women’s agency in the context of IPV. The findings indicate that there is not a dichotomy between “victim” and “agent”, but rather a complex dynamic between and within pregnant women, who may sequentially or simultaneously experience aspects of victimhood and/or agentic response.

INTRODUCTION

In the last decade, concepts of women’s empowerment, autonomy and agency have been increasingly invoked as key leverage points for reducing gender inequalities that are reinforced through intimate partner violence (IPV). In WHO Guidelines for addressing IPV in the health sector (WHO, 2013), the concept of “shared decision making” between the woman and the health professional is stressed. The value attached to women’s agency can be seen also in the choice to avoid the term “victim”, for it suggests passivity, lack of agency, and lack of ability to fight back (Brosi & Rolling, 2010; Salazar & Casto, 2008). While it is crucial both theoretically and in practice to focus on women’s resilience, strength, and autonomy, there is the risk that health workers will expect too much from women who are experiencing violence: far from being a choice or a psychological state, exercising ‘agency’ requires control over resources that many women lack (Cole, 2007; Mardorossian, 2002). Examining how the structural context affects agency is vital if we are to understand how women and communities respond to, resist, or perpetuate IPV (Kabeer, 1999).

Migration is a key aspect of the structural context in much of sub-Saharan Africa. Women’s internal (within-country) migration has been increasing in the region (Camlin, Snow, & Hosegood, 2014), as in other developing regions of the world (Bilsborrow, 1992; Zlotnick, 2003). The global ‘feminization of internal migration’ has been driven mostly by women’s increasing labor force participation, and a recent intensification of women’s internal migration has been observed in Kenya (Beguy, Bocquier, & Zulu, 2010; Brockerhoff & Biddlecom, 1999). IPV is among the factors known to precipitate women’s migration (Camlin, Kwena, Dworkin, Cohen, & Bukusi, 2014), and female migrants may be vulnerable to sexual assault and violence during and after the migration process (Guruge, Khanlou, & Gastaldo, 2010; Idemudia, Williams, & Wyatt, 2013).

Although migration specifically related to nuptiality appears to be declining over time in Kenya, as it is elsewhere in Africa (Camlin, Kwena, et al., 2014), among Luo people in western Kenya women generally still migrate to the location of the husband’s household upon marriage. While Kenya’s new constitution has enshrined women’s individual property rights (FIDA & International Women's Human Rights Clinic, 2008; Ghai & Ghai, 2011), in practice, women’s ability to cultivate crops and have some control over food and income is generally conferred through marriage. In this setting, the expulsion from a marriage also involves being forced off of land, by a husband or relatives, and outside of community settings in which the husband’s family resides. A woman’s decision to leave the marital home is sometimes possible, but challenging, given women’s economic and social dependence on husband and his family.

Intimate partner violence (IPV) is closely linked to the HIV epidemic, as it not only renders women more vulnerable to HIV infection (Decker et al., 2009; Jewkes, Dunkle, Nduna, & Shai, 2010), but also may occur when women disclose their HIV-positive status. Thus, in a high HIV prevalence setting, HIV infection may serve as a trigger for violence against women and forced migration. Because they are often the first in their families to be tested for HIV, pregnant women are especially vulnerable to stigma, discrimination, and IPV related to HIV testing and disclosure (Medley, Garcia-Moreno, McGill, & Maman, 2004; Turan & Nyblade, 2013).

In this study we explore conceptions of women’s “agency” within the phenomenon of women’s migration and violence during pregnancy, in a setting with structural constraints including severe poverty, gender inequality, and stigma. We explore the questions, ‘What circumstances precipitate pregnant women’s forced migration in a high HIV prevalence area in rural Kenya?’ and ‘To what extent are women able to exercise “agency” and undertake migration as a strategy to escape a violent partner/family in this setting?’ First we present findings from qualitative data on pregnant women’s experiences of returning to their maternal homes, or other destinations, as a consequence or a response to IPV and/or family violence. Next, in order to explore how access to resources may influence pregnant women’s mobility in this setting, we present findings on the extent and correlates of migration by pregnant women from a prospective study that aimed to follow women from pregnancy to 6 weeks after the baby’s birth. We consider evidence for the continuum along which forced migrations can be considered to be a form of violence against women, while also examining circumstances under which women’s migrations may also be undertaken as an expression of agency— an action not always, or only, undertaken for survival, but also in pursuit of preservation of dignity and an expansion of life choices.

METHODS

Setting

The Nyanza Region of Kenya has the highest prevalence of physical violence against women in Kenya, with 57% of women aged 15–49 reporting having ever experienced physical violence since age 15, and 36% reporting physical violence in the past year (KNBS & ICF Macro, 2010). Sixty percent of ever-married women aged 15–49 in Nyanza report ever having experienced IPV (39% emotional, 51% physical, and 22% sexual). Nyanza also has Kenya’s highest HIV prevalence rate, (15% of persons aged 15–49) (NASCOP, July 2008). In this context of high HIV prevalence, stigma and discrimination related to HIV are common (Akullian et al., 2014; Nyblade et al., 2013).

Sources of Data

Quantitative and qualitative data for this paper were from two linked studies conducted in rural Nyanza, Kenya during the period 2007–2011. The first of these was the Maternity in Migori and AIDS Stigma (MAMAS) Study, a prospective mixed methods study in which we examined the effects of pregnant women’s fears and experiences of HIV-related stigma on their utilization of essential health services (Turan et al., 2012). Based on preliminary findings from MAMAS on the frequency of IPV experienced by pregnant women, our team also conducted the Gender-Based Violence (GBV) Study in the same area, in which we explored the social context of GBV and developed and piloted an intervention for prevention of violence against pregnant women. For the current secondary analyses on violence, migration and agency, we used 1) baseline and follow-up data from the MAMAS quantitative interviews with pregnant women recruited at their first antenatal care (ANC) visit (n=614); 2) formative data from the GBV Study, including qualitative data collection with pregnant women (n=29), male partners (n=32), and service providers (n=20); 3) post-intervention focus groups (2 groups, n=17) and in-depth interviews (n=25) from the GBV Study; and 4) Clinic IPV screening data (n=134) and reports of “cases” from workers involved in the intervention. The methods used in each of these studies have been described in detail elsewhere (Turan et al., 2012; Turan et al., 2013).

Measures and Definitions

Migration: In this paper, we define any move away from the pregnant woman’s residence at baseline as a migration. This included international migrations (studies were conducted in communities near the Tanzanian border) and internal migrations within Kenya (the most common type), whether voluntary or forced. Forced migrations encompassed events in which women were forced out of their homes by their partner or family, or fled their homes because of partner or family violence.

Violence against women: We define intimate partner violence (IPV) broadly according to the WHO definition, “Behavior by an intimate partner that causes physical, sexual or psychological harm, including acts of physical aggression, sexual coercion, psychological abuse, and controlling behaviors. This definition covers violence by both current and former spouses and other intimate partners” (WHO, 2013). Family violence is also relevant in the context of this analysis and includes acts of physical aggression, psychological abuse, and controlling behaviors by a member of the woman’s or her partner’s family.

Analytic Methods

Qualitative: Data were transcribed verbatim in the language in which they were conducted and, as necessary, translated from the local language (Kiswahili or Dhuluo) into English by professional transcriptionists. To ensure accurate translation, each transcript was reviewed by the researcher who conducted the interview or focus group discussion, and queries were resolved through discussions among the researchers. Transcripts and researcher notes were managed using the Dedoose qualitative software program, using a thematic analysis approach (Attride-Stirling, 2001), with a coding framework developed by three research team members. An initial coding framework was developed based on thematic broad codes. Two authors then applied the final list of broad codes to a single Dedoose database. A preliminary research report was created by printing out excerpts related to each code, elucidating fine codes grounded in the data, reviewing the text for any divergence of opinion, and summarising the views of participants alongside illustrative quotes. Pseudonyms are used in this manuscript.

Quantitative: Descriptive data were extracted from risk assessment forms used in the GBV Study to examine the frequency of women’s reports of various types/risks of violence, including forced migration. In order to examine the extent and correlates of women’s migration during the pregnancy/postpartum periods, we classified women in the follow-up register from MAMAS (n=614), as 1) followed up, 2) lost-to-follow-up due to migrating from the area, or 3) lost-to-follow-up for other reasons. We then examined bivariate associations of these follow-up categories with women’s baseline socio-demographic characteristics and HIV status, using chi-square and ANOVA methods. All quantitative analyses were completed using SPSS v20.

Ethics

Ethical approval was granted by University of California San Francisco (CHR 10-00389) and the Kenya Medical Research Institute (SSC 1688 and SSC 1264). All interviews and discussions were conducted after signed informed consent, and anonymity of participants was protected throughout the research process.

RESULTS

GBV Risk Assessment Results from the Antenatal Clinic

A total of 134 women were screened for GBV in the antenatal care (ANC) clinic during the months December 2010 through April 2011. Among 49 pregnant women who screened positive for IPV over the 5 months, 19 women (39%) reported that their partner or another person close to them had “sent them back to their maternal home”. Other types of violence reported by the women included physical (18%), sexual (17%), psychological (19%), and economic (11%) violence.

Qualitative findings on Forced Migration

Here, we explore the qualitative findings around forced migration and IPV. We first define the types of forced migration discussed by participants. We then examine the precipitating factors for forced migration that emerged from the stories of our participants. Lastly, we describe the experiences of women who undertake migration as a strategy to avoid further violence.

Forced migration as a form of violence

The idea of ‘sending a woman away’ was mentioned by a large number of participants as a form of violence or neglect experienced by women. Women described being ‘sent away’ as a severe form of punishment when men are angry:

Facilitator: What else do men do when angered by their women?

Sarah: He may decide to send you back to your parents’ home.

Wirimu: Some would just neglect you, not even give you money for food because he has been angered.

(Older Pregnant Women FGD)

Men in our study also described how a woman being ‘sent packing’ was a step enacted by male partners in the case of relationship conflict:

What can make them get injured is the bad relationship in the house. That is what will bring about injury […] and even the woman being sent packing. (Younger Male Partners FGD)

Multiple participants also used the Dhuluo word for ‘chased’ to describe how a woman might be forced to leave the marital home:

If the husband knows that they [pregnant women] have been HIV-tested they might be beaten. […] If the mother attempts to mention that ‘I was tested and you are supposed to go for the same’, the mother can end up being chased to go to their motherland. (IDI 6)

Being ‘chased’ from the home by a husband was described as a culmination from other types of violence and neglect, and as a severe form of punishment. This form of violence was enacted by male partners as well as extended family members, such as brothers in-law, and was often mentioned in the same sentence as physical violence. The consequences of being ‘sent packing’ are severe for women, as expulsion from the marital home is an act of shaming and stigmatization, resulting in loss of social status; it also results in women losing access to accommodation, food security, and a means of supporting children. Study participants viewed being ‘sent packing’ or ‘chased’ as one of the most severe forms of physical, psychological, and financial violence enacted against women.

Precipitating factors for forced migration

Several triggers for this type of forced migration were identified in the qualitative data. In many instances, sending a woman ‘packing’ was described as a social sanction against women for taking actions that were deemed inappropriate or unfeminine. For example, in several cases, when a woman was unable to bear children, the woman was forced to leave her marital home:

It is the woman who gets hurt. […] You find that in the house where […] a man and a woman have not gotten a child. All the blames go to the wife, that she is the one who is not giving birth— she gets a lot of problems and she might be sent packing back to her maternal home because she is not giving birth. (Younger Male Partners FGD)

In another case, a woman was rejected by her partner because she was pregnant with a girl rather than a boy. In two other cases, men suspected that pregnancies were not their own biological child and ‘chased’ women away. Such rationale aligns with other forms of violence, which are enacted partly as ways to sanction undesirable actions by women in patriarchal societies.

Another cause of being sent away was if a woman refused to have sex with her partner:

If as a woman you refuse to have sex with your husband, he may decide to send you away because he thinks that you could be having sex with other men elsewhere. Telling you to ‘go and stay with those husbands of yours’. (Older pregnant women FGD)

Taking up roles that were considered inappropriate for women was another reason that women could be ‘sent packing’. For example, if women took up new societal roles such as gaining education or applying for a job away from home, they would be at risk of being sent away, since such a step would be aligned with societal perceptions of promiscuity and haughtiness:

Facilitator: What happens when a woman does what other family members don’t approve of, for example if she is an educated woman, she goes ahead to apply for a job and gets it yet the family members don’t approve of it?

Elinah: Some relatives may not like it, because of jealousy— since the woman who gets a job will be doing better than them.

Facilitator: What would happen in such cases?

Elinah: That would even result in death.

Sandra: She can be bewitched.

Hope: She can be sent away or neglected.

(Older Pregnant Women FGD)

In a case of a woman using violence towards her male partner, the social sanctions included sending her away from the marital home, as well as a public shaming of the woman as ‘cursed’:

Facilitator: What happens if a woman uses violence against a man?

Rose: It is that you can be sent packing immediately. She is sent packing to her home.

Amondi: It is forbidden. That is forbidden.

Rose: That is a woman who is a taboo. A woman who beats up the husband like that that is a woman people don’t live with.

Mary: They say that it is a woman who is cursed. (Younger Pregnant Women FGD)

HIV testing and disclosure was also a frequently mentioned cause of forced migration. There were many reasons that were given as explanation for why a woman who has HIV was sent packing. One was that men may not want to ‘keep’ a sick wife:

Facilitator: What would happen if the man learns that the wife has HIV?

Betty: The husband may conclude that you have been unfaithful and that may cause chaos.

Rebbekah: He may decide to send you away, saying that he doesn’t want a sick wife.

(Older Pregnant Women FGD)

Relatives and neighbors of a man may encourage him to send his wife away because they perceive that having an HIV-positive wife relative reflects badly on the entire family or community.

Apuyo: The relatives … if they know that their daughter-in-law who is pregnant is HIV-infected then they will talk about it. And if the man hears that, … he would take steps to send his wife packing. But even if you send her away you are still HIV-infected!

Kevin: […] if the community members know that one is HIV-infected, they will think that ‘this is a corpse’. The advice they will give is ‘send her away, then marry another one’.

(Younger Male Partners FGD)

A fear of being ‘sent away’ can cause some women to avoid disclosing their HIV status to male partners:

In the ANC when she is pregnant, we’ll always advise them to go and disclose their status to their spouse. But some, most of them tell us it’s difficult to go and approach the husband and tell them their status, because they fear some are so cruel, some are harsh, so they cannot speak some things to them. They may be beaten, or may be sent away or divorced, so they always fear telling the husbands about their status. (Interview 11)

Because of the fear surrounding disclosure, many of the female participants said they would not disclose their HIV status to their partners. Indeed, many described hiding their HIV status from their partners until their male partners fell ill or tested HIV-positive on their own.

Forced migration as a form of resistance

Some women in our study migrated as a strategic response to avoid further violence.

In several cases, women decided to move to a new location in order to avoid the risk of more physical violence:

Domestic violence has destroyed homes in this village and some have ended up having their wives run away. They fight, and when things get tough, the woman takes off. (Older Male Partners FGD)

In one instance, a service provider described how a young, well-educated woman stayed with an abusive partner for three years before moving away:

I have a case of a young woman who decided to leave the man completely. She married, she wanted a very good life, she did everything for this man. The man was controlling her, beating her also, and then sleeping with other women. She tolerated that for three years, but eventually she just pulled out and said ‘enough is enough’. (IDI 10)

In most cases, women’s moves entailed brief stays at the maternal home, not necessarily a long-term change of residence, as maternal families often lacked the economic resources to support another person. For example, one service provider described how a woman’s maternal family’s poverty caused her to continually return to an abusive husband, even when physical abuse injured the woman and her baby:

There was a husband and a wife, and this lady kept on being beaten almost every day. And there is a time she was beaten when she was pregnant but she stayed, because the husband was from a well-off family and she was from a poor family, so the parents of the man kept encouraging her to stay. She would run to her mother’s home and stay there even for a week then come back… So she suffered physically and psychologically, and because of the economic dependence on the man she stayed on. (IDI 18)

The decision to leave one’s marital home was not always discussed as a particularly agentic action. Indeed, one service provider discussed moving as a reactionary step taken when a woman had run out of other options:

The majority suffer silently because of having no structured organizations that are dealing with this. So I would say that some are really suffering there, but nobody knows, they tell nobody they know about it, but they are there. This is not good because they are living in bitterness, resentment and all these things, and it can blow into a bigger thing, sometimes committing suicide, running away from home, going and not coming back. (IDI 1)

Not all women were able to use migration as a strategy to avoid further violence. Many women described being unable to migrate because they lacked the necessary financial resources to do so:

Not many people report these issues … for fear of losing the financial support. They’d rather remain in that abusive marriage because they don’t want to spoil the family name. They don’t want to lose the benefits. Poverty has played a big role. (Interview 24)

Pressure from families to return to violent relationships seemed to be a common occurrence. For example, mothers would tell their daughters that violence in marriage was normal, and that it was better to return to the marital home:

Some are being battered when they ran away from their matrimonial homes to their parental homes. The mother there will tell her that “Even if I’m just running because I’m beaten I would have not stayed here and given birth to you. So you just go back. It is normal. (Interview 22)

The following case described by a Community Referral Person (CRP) illustrates the ‘gray area’ between being forced out of the home as a form of violence, and using migration as a form of agency to escape violence.

A pregnant woman was beaten and sent away from the home by her husband. The husband suspected the woman of having an affair and forbade her from returning to the marital home. The next day, the woman came to the clinic to address the physical wounds from the beating. After the doctor dressed her wounds, he phoned a CRP to assist with the woman’s situation. The following morning, the woman was accompanied by the CRP to safely gather her possessions and return to her maternal home (about 3 hours away by public transport). She went to her maternal home. Although the woman did not want to return to her violent husband, her father was insisting that she return to the husband after the baby was born.

This quote highlights the potential limits of ‘agency’ in the context of poverty and lack of family support, since the woman’s agentic notion of returning home with her possessions was cut short by her own family. Her father’s insistence that she return home may have been rooted in subsistence concerns as well as a patriarchal sense that his daughter ‘belongs’ in the marital home, regardless of her husband’s use of violence.

Migration during Pregnancy and Postpartum

In this section, we examine how resource constraints may relate to pregnant women’s migration in this context, and the extent to which pregnant women are reported to migrate for violence-related reasons. Of 188 women selected for follow-up in the MAMAS Study who could not be located to participate in a postpartum follow-up interview, 43 (23%) had a migration-related reason for loss-to-follow-up. These included pregnant women who were reported by a proxy informant to have moved to another location (31), migrant fish traders (7), and women who were specifically reported to have moved due to having been ‘chased’ or ‘sent packing’ by a husband/family (5 women, 12% of those who migrated). Five of the women who migrated were reported to have moved across an international border (to Tanzania).

Table 1 shows associations of follow-up category (did not complete the post-partum interview due to migration; did not complete the interview due to other reasons; and completed the interview) with socio-demographic and other variables measured at baseline. Subsequent migration was significantly associated with higher education, lower likelihood of co-habitation with male partner, and a lower number of living children. Among other covariates, HIV status at the time of the first ANC visit during pregnancy was not associated with follow-up category.

Table 1.

Women’s Completion of the Postpartum Follow-up Interview by Baseline Socio-demographic and HIV-related Characteristics (N=598)

Variable	Did not complete interview due to migration reasons	Did not complete interview due to other reasons	Completed Interview	Statistical Test Results
	(n=43)	(n = 144)	(n = 411)
Age, n (%)
< 25	33 (76.7)	89 (61.8)	261 (63.5)	χ2 = 3.377, p = 0.185
≥ 25	10 (23.3)	55 (38.2)	150 (36.5)
Mean age, (Standard Deviation, Range)	22.7 (5.1, 18–40)	23.9 (5.3, 18–48)	23.8 (5.1, 18–40)	F-test = 0.963, p = 0.383
Education, n (%)
More than Primary	12 (27.9)	24 (16.7)	56 (13.6)	χ2 = 6.339, p= 0.042
None or Primary	31 (72.1)	120 (83.3)	355 (86.4)
Religion, n (%)
Roman Catholic	10 (23.3)	31 (21.5)	70 (17.0)	χ2 = 2.453, p= 0.653
Seventh Day Adventist	13 (30.2)	47 (32.6)	131 (31.9)
Other	20 (46.5)	66 (45.8)	210 (51.1)
Ethnicity, n (%)
Luo	39 (90.7)	139 (96.5)	382 (92.9)	χ2 = 2.978, p = 0.226
Other	4 (9.3)	5 (3.5)	29 (7.1)
Number of Living Children, n (%)
Zero or one	31 (72.1)	74 (51.4)	190 (46.2)	χ2 = 10.740, p = 0.005
More than one	12 (27.9)	70 (48.6)	221 (53.8)
Mean number of living children, (Standard deviation, Range)	1.4 (1.66, 0–7)	1.7 (1.574, 0–7)	1.8 (1.552, 0–9)	F-test = 2.155, p = 0.117
Household goods, n (%)
Electricity	1 (2.3)	6 (4.2)	12 (2.9)	χ2 = 0.648, p = 0.723
Television	4 (9.3)	15 (10.4)	40 (9.7)	χ2 = 0.073, p = 0.964
Mobile Phone	21 (48.8)	60 (41.7)	195 (47.4)	χ2 = 1.567, p = 0.457
Radio	31 (72.1)	94 (65.3)	323 (78.6)	χ2 = 10.251, p = 0.006
Woman’s job, n (%)
Housework	7 (16.3)	34 (23.6)	92 (22.4)
Selling things/ Fish trader	7 (16.3)	33 (22.9)	97 (23.7)	χ2 = 6.760, p = 0.344
Farming/Agriculture	17 (39.5)	49 (34.0)	159 (38.8)
Other	12 (27.9)	28 (19.4)	62 (15.1)
Marital Status, n (%)
Not currently married	9 (20.9)	22 (15.4)	51 (12.4)	χ2 = 2.817, p = 0.244
Currently married	34 (79.1)	121 (84.6)	360 (87.6)
Living with a male partner, n (%)
Yes	31 (72.1)	124 (86.7)	355 (86.4)
No	12 (27.9)	19 (13.3)	56 (13.6)	χ2 = 6.628, p = 0.036
Polygamous Household n (%)
Yes	11 (25.6)	45 (31.5)	98 (23.8)	χ2 = 3.223, p = 0.200
No	32 (74.4)	98 (68.5)	313 (76.2)
Male Partner’s Occupationa, n (%)
Selling things	2 (6.5)	12 (9.7)	39 (11.0)	χ2 = 11.608, p = 0.170
Farming/ Agriculture	6 (19.4)	50 (40.3)	115 (32.5)
Fishing	6 (19.4)	18 (14.5)	61 (17.2)
Manual Labor	4 (12.9)	22 (17.7)	58 (16.4)
Other type of job	13 (41.9)	22 (17.7)	81 (22.9)
HIV Status, n (%)b
HIV-positive	15 (34.9)	57 (39.6)	156 (37.7)	χ2 = 5.128, p = 0.529
HIV-negative	17 (39.5)	45 (31.3)	166 (40.1)
Refused HIV test	7 (16.3)	26 (18.1)	52 (12.6)
No HIV testing available	4 (9.3)	16 (11.1)	40 (9.7)

^aData on this variable were available for women who reported currently living with a male partner. n = 509

^bPregnant women were offered HIV testing during their antentatal clinic visit, after they had completed the baseline questionnaire.

DISCUSSION

The complex circumstances driving migration among pregnant women in Kenya defy simplistic assumptions about the nature of women’s agency in the context of IPV in resource-constrained settings. We found that among women screened at an antenatal clinic in rural Kenya, being sent back to the maternal home was a relatively common occurrence, more commonly reported than other forms of IPV. In many cases, we found that women were being ‘chased’ or ‘sent packing’ from the home by male partners and/or family when they defied traditional gender norms and exercised forms of agency. Forced migrations from marital homesteads can be seen as acts of violence towards women, and certainly demand a broader conceptualization of violence that includes forced migration. Yet, embedded within women’s narratives are accounts of migration as a decision, a strategy, and an action taken.

Thus, the context of intimate partner and family violence in this setting challenges the usual binary categorization of migrations as either “forced” or “voluntary”, as well as categorizations of violence “victim” or “agent”. Migration can be a “choice”, an expression of agency. But even when forced, migration is often the result of an act of resistance on the part of the woman. Here, we present our findings in light of the literature and consider their implications for a re-theorization of “agency” within contexts of HIV, IPV and family violence, female migration, and pronounced resource constraints.

Our findings bolster arguments suggesting that forced migration is a form of violence that should be examined within a broader structural context. The HIV epidemic has converged with other political-economic trends to bring about disruptions in traditional forms of securing women’s access to land, such as inheritance (Camlin, Kwena, et al., 2014). In prior qualitative research among female migrants in western Kenya (Camlin, Kwena, & Dworkin, 2013; Camlin, Kwena, et al., 2014), while many women reported being ‘chased’ or ‘sent packing’ to their maternal home, they did not linger, as they either reported wishing not to burden their parents or not feeling welcome. Rather, most women undertook subsequent independent migrations to other locations to establish a new household. In these instances, women’s migrations were undertaken as agentic strategy to preserve their health, and also achieve aspirations for autonomy and a better standard of living (Camlin, Kwena, et al., 2014).

Our findings equally suggest that forced migration can serve a form of violence enacted upon women in instances where she pushes the boundaries of her culturally defined gender role. For example, women were ‘sent packing’ for refusing to have sex, using violence towards men, or taking up new roles without permission. HIV testing without permission and even having an HIV-positive diagnosis are further examples of this resistance, since HIV diagnosis may result in perceptions of promiscuity and extramarital sexual relationships, an association bolstered by public health campaigns (Knopf, Agot, Sidle, Naanyu, & Morris, 2014).

It is crucial to examine the language used by participants in a critical, reflective light. Scholars in the IPV field have posited that narratives are a method of human sense-making and also a tool for representing and reconstituting experience (Boonzaier & van Schalkwyk, 2011). In this rural Kenyan setting, the terms used for forced migration (i.e. ‘sent packing’, ‘being chased’) were examples of patriarchal and “victim-centric” community responses to violence. In a sense, the few narrative representations of women ‘choosing to leave’ are interesting because they serve as an important counterpoint to the common conception that men control these migration outcomes for women. Yet, women’s stories presented in this paper were rarely characterized by the participants as being both empowered and oppressed. This is a feature of constructions about female survivors of violence globally, and scholars have noted the tendency for many societies to dichotomize women into either victimhood or agency (Mahoney, 1994). Indeed, a main goal for violence research should be “to simultaneously acknowledge the existence of victimization and agency among all battered women and how they may play themselves out differently depending on social and historical circumstances,” (Sokoloff & Dupont, 2005). This research has aimed to do just that within the specific context of rural Kenya.

Data from the present study shed some light on the debate surrounding the dialectic between domination and agency, and the argument that an emphasis on women's oppression and victimization obscures the extent to which women make choices, resist coercion, and exercise agency in their own lives (Abrams, 1995). Our data show that victimization and agency, far from being situated at the opposite ends of a dichotomy, are deeply intermingled in the experiences of abused women in this Kenyan setting. Women exercise agency when, having some resources, they leave a violent man and migrate, but also because the violence was often triggered by their acts of agency or resistance to their husband’s domination. To add to this complexity, these very acts of resistance--refusing sex, getting tested for HIV, fighting back, having a lover--may bring on her the blame of her family and community, including health workers and the police. Analyses in Western countries indicate that blame may become a problem when women ask others for help, since “a person must be completely innocent and helpless to qualify as a victim” and therefore be considered deserving of support and help (Cole, 2007).

It is important to consider the issues of agency and structural inequalities in order to understand women’s migration in the face of partner and/or family violence. Our quantitative follow-up data revealed that women who migrate during pregnancy in this setting are more likely to have more resources and fewer dependents. It is challenging, for example, to assess how voluntary is “voluntary migration” for women in a context of extreme poverty, gender inequality, and HIV-related stigma, which overly penalize women. Conversely, when women do use migration as a strategic method to escape IPV, it is important to consider the resources that permitted this strategy (e.g. social support, better education, more disposable income, fewer children), and realize that “agency” is often conditional on such resources. Indeed, women’s highly strategic acts of migration were constrained by the structural realities of poverty at the maternal home and potential patriarchal views held by parents. In highlighting women’s agency alongside the realities of intersecting structures of economy, class, and gender in Kenya, we aim to avoid the tendency to focus only on a woman’s ‘choice’ to leave and unwittingly further dichotomizing victimhood and agency (Dunn & Powell-Williams, 2007). Instead, our data were able to unpack in preliminary ways how agency is “fraught with fluidities” (Lloyd, Emery, & Klatt, 2009). In alignment with other (non-IPV) literature, we conclude that agency represents a woman’s capacity to act even as she is constrained by a distinct socio-cultural reality, and it often involves the creation of a new and unfamiliar path that is neither purely agentic or oppressed (Ahearn, 2001; McNay, 2013).

Limitations

This study had several limitations. That our qualitative stories were unable to fully deconstruct the nuances between and across categorizations of ‘victim’ and ‘agent’ may suggest that longitudinal, qualitative techniques would be useful in future studies. There were also simplistic notions of agency used within participant narratives, many of which we are unable to fully problematize. For example, perhaps participants in this study underutilized ‘agentic’ stories because they, themselves, were part of the patriarchal society; or, perhaps stories of women being ‘forced out’ were easier to discuss with a researcher who had obvious interest in addressing intimate partner violence? Our qualitative findings should be considered in light of such critiques, and suggest future considerations for violence research in resource-constrained, patriarchal settings.

We used secondary data that were not specifically collected to examine this issue, but rather collected to understand the broader social context, triggers, and effects of stigma, discrimination, and IPV on pregnant women in this setting. In our quantitative analyses, reasons for loss-to-follow-up were unknown for some women, which could have led to an underestimate of the extent of migration during pregnancy. In addition, the reasons for women’s migration were often not specified, so that it was often unclear if the migration was forced or voluntary. Since we did not collect baseline data on experiences of IPV in MAMAS, we do not know if the women who did migrate (and thus were lost to follow-up) were more or less likely to have experienced IPV at baseline. Finally, we did not collect follow-up data on long-term outcomes of women in the GBV Study. This was due to a combination of lack of resources for the small pilot study and the need to protect these women from any additional risk posed by the research.

Conclusions

These findings underscore the importance of accounting for contextual factors that enable “agency” in the context of IPV. We argue against narrow, individualist notions of women’s agency, echoing calls to include access to resources as a crucial part of women’s ‘empowerment’ (Kabeer, 1999). In rural Kenya, women exercise agency within the context of marriage systems, property rights, land use and inheritance practices, HIV-related stigma, and gender systems in which their choices are often constrained. Being forced out of the marital home appears to be a common occurrence in this setting, but severe resource limitations may mean that women have nowhere to go in these situations. In these contexts, women must use social resources, resilience and coping in order to survive a violent situation. These larger structural considerations do not cancel out the ability of women to take action to improve their lives, but they are crucial to engage in future efforts to prevent IPV. Efforts to reduce IPV and family violence need to support women’s access to resources that can foster agency and create the possibility for women to live free of violence.