EMPOWERMENT AND CRITICAL CONSCIOUSNESS: AN INVESTIGATION OF - BEATING IN KENYA By Alaina Marie Bur A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of Sociology Master of Arts 2018 ABSTRACT EMPOWERMENT AND CRITICAL CONSCIOUSNESS: AN INVESTIGATION OF - BEATING IN KENYA By Alaina Marie Bur serves as an ideal case study because it has strong ethnic and cultural diversity and a high proportion of men and women who ju stify the patriarchal practice of wife - beating. This investigation Survey (DHS) in Kenya to perform logistic regressions of factors shaping justification of wife - beatin g across five hypothetical scenarios that can occur between a husband and wife . Results reveal that there are significant differences - beating across ethnic groups , educational categories , and experience migrating . I draw on theory of culture to draw two conclusions from these findings . First, Kenyan ethnic groups are a source of culture that women draw on t o form their repertoire of patterned action and belief during periods of settled life . Second, educat ion and migration represent Kenyan women encounters with new gender ideology and experiences with resocialization during periods of unsettled life . These experiences offer opportunities for women to gain critical consciousness , or a reflexive vantage point from which they gender beliefs about wife - beating are, indeed, unjust . iii ACKNOWLEDGMENTS The author would like to thank committee members Jennifer Carrera, Wynne Wright, and Sand y Marquart - Pyatt , as well as Tom Dietz, Zhenmei Zhang, and Aaron Ponce for their valuable comments and suggestions throughout the revision process. An earlier version of this paper was presented at the annual conference held by Soci ology of Development S ection in Detroit, Michigan on October 5, 2017 and benefitted from many helpful comments. iv TABLE OF CONTENTS LIST OF TABLES v LIST OF FIGURES vi KEY TO ABBREVIATIONS vii INTRODUCTION 1 LITERATURE REVIEW 4 Resistance to Patriarchal Constraints 4 Patriarchy and Social Identity 4 Power for Resistance 5 Theories of Empowerment 5 7 Cultural Influences on the Justification of Wife - Beating 8 Ethnicity as a Unit of Culture in Kenya 10 Ethnicity, Education, and Migration as Cultural Tools 12 DA TA AND METHODS 13 Data 13 Measures 13 Dependent Variables 13 Independent Variables 14 Methods 15 RESULTS 16 Descriptive Statistics 16 Ethnicity an d Justification of Wife - Beating 19 Education 22 Migration 22 Control Variables 23 DISCUSSION 2 4 Hypothesis 1: Justification of wife - beating is uniform across all ethnic groups 2 4 Hypothesis 2: Education has no effect on the justification of wife - beating 25 Hypothesis 3: Migration has no effect on the justification of wife - beating 2 6 CONCLUSION 2 8 REFERENCES 31 v LIST OF TABLES Table 1 . Descriptive statistics of women in unions ages 15 - 49 in Kenya in 2014 (N=8534) 17 Table 2 . Percent of women who justify wife beating, by reason 19 Table 3 . Odds ratios of logistic regression models of five reasons for justifying wife - beating amon g women ages 15 - 49 living in unions in Kenya in 2014 20 vi LIST OF FIGURES Figure 1 . Percent of women why justify wife - beating by reason 18 vii KEY TO ABBREVIATIONS DHS Demographic and Health Survey IPV Intimate partner violenc e 1 INTRODUCTION In 2000, the U nited Nations as a Millennium Development Goal, bringing the concept to international attention (United Nations 2006) . W robbed of its original political intent, strategic value, and feminist conceptual roots as it has be co me mainstreamed and simplifie d in international development discourse (Batliwala 2007; Cornwall 2007) . Despite feminist scholars original intent that empowerment be understood within its local, cultural context (Kandiyoti 1988; Rowlands 1997) , general empowerment strategies continue to be applied as a universal panacea (Cornwall 2016) . Thus, u nde rstanding the role of cultural variability in stimulating empowerment remain empowerment initiatives. conscious social identity, ability to question her subordinate status, capacity to strategically control her own life, and ability to reshape society by participating on equal terms with men (Kabeer 2011:499) iquely shape (Kabeer 2011; Kandiyoti 1988) . B ecause women often internalize their subordinate socia l status, they can recognize unequal gender relations but not perceive such disparities as unjust, leading them to accept and perpetuate gender norms that do not support their well - being (Kabeer 1999) . Thus, women m ust recognize gender injustice before they reject their complicit role in its reproduction (Kabeer 2011; Stromquist and Gischman 2009) . I will investigate how the first aspect of empowerment, patriarchal norms. 2 Domestic violence is one kind of patriarchal relationship that spans a variety of contexts, to which, despite its detrimental effect on their own well - being, women sometimes acquiesce because they have internalized a lesser status (Linos, Khawaja, and Al - Nsour 2010; Uthman et al. 2009) . Wife - beating is a specific form of domestic violence most commonly practiced by men to control and discipline wives who transgress their ascribed gender responsibilities (Lawoko 2008; Sim ister 2010; Wekwete et al. 2014) . Studies have found that condoning intimate partner violence (IPV) is strongly associated with being a perpetrator or victim of IPV (Hanson et al. 1997; Stith et al. 2004) . These findings support the theory that the ability to choose conditions - being is dependent on the willingness to reject those unjust relations (Guiné 2014; Kabeer 2011; Stromquist and Gischman 2009) . S tudies of emp leverage resources such as micro - ability to control her own life and participate on equal terms with men (Allendorf 2007; Boateng et al. 2014; Lesser Blumberg 1984; Mishra and Sam 2016) . Less clear, however, is the link social identity , the socially - constructed beliefs a woman holds about her position, roles, and responsibilities, and critical consciousness , the reflexive perspective of (Freire 1993; Kabeer 2011) . justification of wi fe - beating is one way of measur ing critical consciousness (Linos et al. 2010; Mann and Takyi 2009; Mugoya, Witte, and Ernst 2015; Rani, Bonu, and Diop - Sidibe 2004) . Studies have found ethnicity to be a key social driver of a wom justification of wife - beating in a study of ten ethnic groups in Kenya ( Mugoya et al. 2015; Simister 2010) and five in Ghana (M ann and Takyi 2009) . Upon discovery of this association, 3 inheritance (Mann and Takyi 2009), polygamy, and dowry payments (Mugoya et al. 2015) create varied patriarchal structures of gender norms. Mugoya et al. (20 15) argue that ethnic groups in has also been found to be associated with the justificat ion of wife - beating in sub - Saharan Africa (Mann and Takyi 2009; Mugoya et al. 2015; Simister 2010) . Yet studies have yet to interrogate how interactions with - of - birth, such education and migration, affect perce ptions of the patriarchal practice of wife - beating. Because Kenya is very ethnically diverse (Mugoya et al. 2015) and because Kenyan men and women are among the most likely in sub - Saharan Africa to justify wife - beating (Lawoko 2008; Uthman et al. 2009) , I use Kenya as a case study for investigating the relationship between social identity, ethnicity, and the justification of wife - beating. The Kenya Demographic and - beating and, as this is a strong beliefs about patriarchal practices, this data set is particularly appropriate for this analysis (USAID 2013) . This paper examines how culturally - shaped social identity and periods of unsettled life influence their critical consciousness. I do so by examining whether Kenyan affect s their justification of wife - beating. I also investigate how interactions with competing cultural ideology through education or migration provide women with ideological vantage points from which they can critically examine whether the social interaction of wife - beating is, indeed, un just. 4 LITERATURE REVIEW Resistance to Patriarchal Constraints Patriarchy and Social Identity. (de Beauvoir 2009:330) . identity is bound to their incessant performance of gendered norms, regardless of whether they (Butler 2004:1) . Individuals are always socialized to perform gender in a given context . Sandra Harding argues m asculine and feminine are always categories within every class, race, and culture in the periences, desires, and interests differ within every class, race, (Harding 1987:7) . Yet, because societies construct gender as endowed with unequal roles and resources, Patriarchy is the structure that transmits an d perpetuates unequal gender relations. Kandiyoti (1988) posits that gender i patriarchal constraints that vary across space, time, class, and culture. This arrangement , termed is a set of strategies and gender scripts that are the means and (Kandiyoti 1988:285) . Patriarchy is both material and relational, as it relies on an unequal distribution of resources, yet is embedded within com control, and rights over resources, and allow for the flow of material power toward men (Agarwal 1997; Kabeer 2011; Malhotra and Mather 1997) . Because patriarchy is relational, a race, class, and ethnicity, determine the patriarchal constraints that shape her gender 5 consciousness, or her subject ive perceptions of and beliefs about her roles and responsibilities (Batliwala 1993; Collins 1991; Kabeer 1999, 2011) . Power for Resistance. that patriarchal structu res can be changed to achieve greater equality between men and women (Batliwala 2007; Cornwall 2016; Kabeer 1999) . Empowerment is defined which those who have been denied the ability to make strategic life choices acquire such an (Kabeer 1999:435) . Empowerment does not mean that women are always using their expanded power can exercise power even within the context of patriarchal constraint. They draw on resources, strategies, and scripts that ena ble them to bargain with, deceive, manipulate, or resist me n and women to improve their own well - being (Agarwal 1997; Kabeer 1999; Kandiyoti 1988) . Sometimes women even exercise power to perpetuate patriarchal constraints for other women. There are also times when women use their power to renegotiate their patriarchal bargain (Kibria 1990) . Renegot iation involves resistance, or (Stromquist and Gischman 2009:468) . However, for a woman to engage in resistance and transgression she must realize th e existence of those gendered constrain t s (Cornwall 2016; Kabeer 2011) . Theories of Empowerment. Rowlands (1997) recognizes power as a process whereby women gain understanding and consciousness, or power within , enabling them to expand their power with others through collective action, gain the power to challenge unjust gender norms and improve their own lives , and gain power over people and resources (Rowlands 1997:13) . One vein of empowerment theory focuses on a power over . This perspective is rooted 6 in a materialist understanding of power, which assumes that women need improved economic and human capital to resist patriarchal constraints (Parpart, Rai, and Staudt 2002; Wri ght and Annes 2016) . It emphasizes how women can leverage their power over economic resources, positions within institutions, and decision - making to gain power over (Agarwal 1997; Lesser Blumberg 1984; Mishra and Sam 2016) . However, t his approach is criticized for overlooking the dynami c social processes by which women activate their internal critical consciousness, or power within , that enables them to expand their power to work with others to improve their own lives (Parpart et al. 2002; Wright and Annes 2016) . An alternative understanding of the empowerment process draws on a Foucauldian understanding of power (Batliwala 1994; Cornwall 2016; Kabeer 1999) . This conceptualization recognizes that power is not only (Parpart et al. 2002:1964) . The approach emphasizes the centrality - percepti on, the importance of a conscious understanding of unequal gender relations , and the recognition of unjust gender norms and ideology as being a key starting point for empowerment (Batliwala 1993; Rowlands 1997) . A woman must re cognize unequal gende r relations and regard them as unjust before she is willing to leverage her assets to resist oppres sive gender norms and responsibilities (Kabeer 2011) . Critical consciousness is thought to be the mechanism that develops this power within (Batliwala 1994; Cornwall 2016; Freire 1993; Stromquist and Gischman 2009) . Patricia Hill Collins argues that a fundamental form of e (Collins 1991:111) . When a wom a sense of self is cons she receive s and internalize s a subordinate social identity (Batliwala 1993) . It is by participating in expanded 7 social relationships that a woman can gai n reflexive vantage point from which to evaluate (Kabeer 2011:503) and cultivate her critical consciousness. These theories focus on education and general consciousness - raising initiatives as a means of stimulating critical consciousness (Cornwal l 2016; Freire 1993; Guiné 2014) . with the tools to demand a better position in society (Stromquist and Gischman 2009) . However, because these approaches focus on how interaction with non - traditional ideology through formal education encourages critical consciousness, th ey have neglected to consider how interaction with new cultural ideology through other means shapes consciousness. sociological Swidler (1986) practices, art forms, and ceremonies, as well as informal cultural practices such as language, (Swidler 1986:273) . I understanding of culture as the means by which agents construct a repetoire, or - kit , . Women draw on these strategies to decide how to act to creatively solve problems. Culture provides actors with a patterned means of solving institutional problems, giving culture a causal role in shaping beliefs and action (Swidler 1986, 2001) . Swidler (1986) ac counts for cultural change She argues that a gents with settled lives reflect little on the meaning of their norms and it appears that social structure and cultural strategies are fused , as both co - constitute one another. However, those with unsettled lives are living in a moment of rapid social transformation, in - 8 provides agents with new blueprin ts of cultural strategies (Swidler 1986:278 79) . This is a period when individuals grapple with traditions, sets of cultural beliefs and practices that they take for g ranted because they have become so ingrained into daily life. They also reconsider their Periods of unsettled life produce moments when individuals, confronted with ne w ideology and patterns of action, can choose, based on their current concrete situation, to apply a new cultural framework to transform their personal lives and possibly even restructure their old cultural framework (Swidler 198 6) . (Kandiyoti 1988) , culture shapes the repertoires of strategies and an understanding of when and how to creatively apply those strategies. However, the introduction of a gender discourse creates a period of of critical consciousness. The next section explores what past studies have found about the relationship between cult beliefs of whether the patriarchal act of wife - beating is ever justifiable . Cultural Influences on the Justification of Wife - Beating has been found to be significantly associated with the justification of wife - beating across sub - Saharan African countries, where a high concentration of empowerment initiatives are focused (Mann and Takyi 2009; Mugoya et al. 2015; Rani et al. 2004; Simister 2010; Uthman et al. 2009) . Generally, women with fewer household resources to draw upon may be less able to leave an abusive relationship, and thus more likely to find reasons to accept the relational constraint of wife - 9 beating ( Simister 2010) . However, t he act of resist ing an abusive, patriarchal relationship is (Mann and Takyi 2009:333) . Thus, both material and ideological conditions should play a role in shaping whether a woman justifies or accepts the constraint of wife - beating. Increase in ag wife - beating across sub - Saharan African countries (Linos et al. 2010; Mann and Takyi 2009; Rani et al. 2004; Uthman et al. 2009) . This is likely because as a woman ages, she gains more varied and mature perspectives from which she can see the unequal power relations inherent in IPV (Stromquist and Gischman 2009) . Being in a polygynous union is also found to significantly - beating, probably because it is a patriarchal custom that frames a wife (Rani et al. 2004) . rejection of wife - beating, especially in sub - Saharan Africa (Linos et al. 2010; Mann and Takyi 2009; Mugoya et al. 2015; Rani et al. 2004; Simister 2010; Uthman et al. 2009) . Two complementary st, a a resourc e that later supplies her with economic capital through employment (Simister 2010) . Second, e ducation is regarded as providing men and women with ideology that is opposed to w ife - beating (Mann and Takyi 2009:333) , which supplies new gende r norms and democratic problem - solving strategies and opens up new cultural strategies of action (Rani et al. 2004; Stromquist and Gischman 2009) . Finally, ethnicity has been found to have a consistently significant influence on a - beating in several sub - Saharan African countries (Mann and Takyi 2009; Mugoya et al. 2015; Simister 2010) . It is postulate d that each ethnic group present s a 10 different set of patriarchal constraints because each has its own unique cultural customs such as a matrilineal or patrilineal a wife is free to remarry at will or must re - practiced; whether a wife can divorce her husband and, if she does, whether her dowry will be returned; and whether an ethnic group practice s female circumcision and child marriage (Mburugu and Adams 200 5; Mugoya et al. 2015; Simister 2010) . For example, the significance of ethnic identity in Ghana is attributed to the differences between the matrilineal Akan group and other patrilineal groups, as the Akan pass property down through the matrilineal lin e, affording women more power over resources (Mann and Takyi 2009) . Thus, each group provides women with a unique cultural toolkit from which they select a line of action that allow s them to navigate such custo ms. Ethnicity as a Unit of Culture in Kenya Kenya presents an optimal case for studying the effect of ethnicity on beliefs about wife - beating (Mugoya et al. 2015; Simister 2010) . Kenya is very ethnically diverse, as it is home to 42 ethnic groups, each with their own language, customs, and gender norms (Mugoya et al. 2015) . Ethnic groups are divided along linguistic lines (Mburugu and Adams 2005) and the region where one lives (Kenya National Bureau of Statistics 2014) , meaning each ethnic group has its own linguistic and spatial boundaries that allows for ethnicity to be treated as a unit of culture (Mugoya et al. 2015) . 42 ethnic groups fall under three linguistic categories : Bantu, Nilotic, an d Cushite. Only t the four largest ethnic groups being the Kikuyu (Bantu) , Luhya (Bantu) , Kalenjin (Nilotic) , and Luo (Nilotic) (Kenya National Bureau of Statistics 2014) . The largest ethnic group, the Kikuyu, mainly live in the highland regions in the 11 (Kanyinga 2009) . The Luhya and Luo ethnic groups live mainly in western Kenya near Lake Victoria. A large population of ethnic Somalis, as well as several small groups that are also Cushite - Somalia . The Kalenjin and other Nilotic language - speaking groups such as the Samburu, Turkana, and Maasai, mainly populate the semi - arid low lands of the Rift Valley, which cuts across the country from north to south (USAID 2013) and, aside from some Kalenjin sub - groups that live in the fertile highlands, have traditionally been semi - nomadic pastoralists (Mburugu and Adams 2005; Mugoya et al. 2015) . Groups that are located far from the capital of Nairobi and on less fertile land , such as the Samburu and Turkana , were neglected during the colonial era (Mburugu and Adams 2005) , and continued to be overlooked by the post - colonial state until power was devolved from the central government to the local count y government under the new 2010 Constitution Cornell 2016) . At present, those neglected groups retain a legacy of l ow economic and infrastructure development, low education rates, and overall less societal change than other regions of Kenya have undergone . Ethnic boundaries continue to be reinf orced through ethnic politics (Kanyinga 2009) . Ethnic identity has been found to play a significant role in determining how people will vote (Taylor 2017) , how the government distributes aid money in the country (Briggs 2014) and levels of local violence (Okumu et al. 2017) . Regardless of the socially - constructed nature of these ethnic groups, Kenyans uphold and experience the ma terial consequences of ethnicity as real. Thus, one may postulate that ethnicity play s a strong role in shaping whether women reject wife - beating 12 Ethnicity, Education, and Migration as Cultural Tools One study of Kenya suggests that justifying wife - for women who are under high cultural constraints that prevent them from leaving a n abusive marriage (Mugoya et al. 2015) . Upon discovery of the significance of ethnic identity and for perceptions of value, self - Still , neither study attempt s to untangle the relationship between ethnicity and education as competing sources of gender discourse. Nor do they consider how migration, as an opportunity to interact with the gender ideology of other ethnic groups (Kibria 1990) , might influence beliefs about wife - beating. Yet both conclude that it is the task of future research to probe the relati onship between ethnicity and education (Mann and Takyi 2009; Mugoya et al. 2015) . What, then, is the effect of ethnicity , education , and migration about the justifiability of wife - beating? If ethnicity is, indeed, a cultural source of customs that beliefs about wife - beating. Additionally, because education and migration beyo - of - birth unsettle - for - , both should be related to the justification of wife - beating. Given that I expect differences in justification of wife - beating across ethnicity and that education and migration will contribute to the experience of having an unsettled life , thus affecting b eliefs related to wife - beating, I investigate the following three hypotheses in the Kenyan context : Hypothesis 1 : The justification of wife - beating is uniform across all ethnic groups Hypothesis 2 : Education has no effect on the justification of wife - beati ng Hypothesis 3 : Migration has no effect on the justification of wife - beating 13 DATA AND METHODS Data This study uses data from the Kenya DHS conducted in 2014 (Kenya National Bureau of Statistics 2014) . This is a nationally representative, cross - sectional survey conducted in households. T he DHS ha s conducted six rounds of questionnaires in Kenya since 1989; this data draws on the latest round. The DHS includes a and long - form surveys. Data about perceptions of wife - - form survey. Thus, t his study uses data from t - form surveys of 14,741 women between the ages of 15 - 49 , and the household survey data that corresponds with each of the . The - form survey had a response rate of 9 6.2 %. I merged the two data sets using the househo ld ID number assigned to each woman and household. Because the justification of wife - beating takes place between husbands and wives, I restrict my sample to women who currently live in marital unions , resulting in a sample size of 7,901 women after droppin g missing data . Measures Dependent V ariables. I have six models total. For the first five models, I use five binary dependent variables to see how ethnicity, education, and migration effect the justification of wife - beating across five hypothetical conditions. Each variable is related to the five survey questions asked in the Kenya DHS about t he justification of wife - beating. In the survey, respondents respond to the question, his wife in the following situations: (1) If she goes out without telling him? (2) If she neglects the chil dren? (3) If she argues with him? (4) If she refuses to have sex with him? (5) If she burns the (USAID 2013) . I exclude 14 totaled 318 women, or about 3% of the observations . For each of the five outcome measures, I constructed binary variable s coded 1 if the woman responded that the situation justified wife - beating, and 0 otherwise. Following past studies of wife - beating, for my sixth model, I construct one binary variable that measures whether a woman believes that there is any reas on for wife - beating, coded 0 if a woman provides no justifications for wife - beating, or 1 if a woman justifies one or more reasons (Linos et al. 2010; Mann and Takyi 2009; Rani et al. 2004; Uthman et a l. 2009) . While the five former binary variables capture how context influences the conditional justification of wife - beating, the latter estimates the risk factor of any tolerant attitudes towards wife - beating (Mann and Takyi 2009) . Independent Variables. on ethnicity. Of the 42 ethnic groups in Kenya, 14 are large ethnic groups (Mugoya et al. 2015; USAID 2013, see page 32) . Mugoya et al. (2015) retain eight ethnic groups, including the Kikuyu, Kalenjin, Kamba, Kisii, Luhya, Luo, Embu/M eru, and Mijikenda/Swahili, Simister (2010) retains ten ethnic groups and drops each constitute their own unique culture with diverse languages, livelihoods, and places of residence, it is neither appropriate to treat them as one category of homogenous ethnic identities , nor category and the Taita/Taveta ethnic group, which is only 1.49% of the sample. I drop a total of 973 cases, or 9.3% of the total sample. I also collapse the Embu and Meru into one group, 15 following Mugoya et al. (2015), since these groups are considered to have remarkably similar customs. The twelve remaining ethnic group categories are at least 2% of the final sample. competing ideology, I measure the level of education (Linos et a l. 2010; Mugoya et al. 2015; Rani et al. 2004; Simister 2010; Uthman et al. 2009) . I also measure whether a woman has ever moved outside her place - of - in a polygynous union (Rani et al. 2004) ; their age in years (Mann and Takyi 2009) ; their number of children (Rani et al. 2004; Wekwete et al. 2014) ; and their household wealth index , which serves as a proxy of socio - economic status (Mann and Takyi 2009; Mugoya et al. 2015; Rani et al. 2004; Simister 2010; Uthman et al. 2009) . The household wealth index is a variable constructed by the DHS based on responses in the household survey to questions about household assets, and produces five wealth quintiles ranging from poorest to richest (Kenya National Bureau of Statistics 2014) . Methods I analyzed a ll data with STATA Version 15.1. The six dependent variables are all binary variables; therefore, I use logistic models for all six models (Hoffmann 2016) . Model 1 is a logistic regression of the effect of the ethnicity variable on whether it is e ver justifiable for a husband to beat his wife , holding education, whether a wife has ever moved, her household wealth index, age, number of children, and if she is in a polygynous union constant. Models 2 through 6 are all logistic regressions of the effe ct of the ethnicity variable on whether a defined reason for wife - beating is justifiable, holding all else constant. The ethnicity, education, and 16 migration variables are included in all models to see what their effect is on the justification of wife - beati ng in one or more situation (Model 1) and under specific situations (Models 2 through 6). The models were tested for multicollinearity and all VIF values were found to be within an acceptable range, all below 5.00 (Hoffmann 2016) . Model 1 is a logistic regression of whether a woman justifies any reason for wife - beating. This model does not measure why a woman justifies a given condition of wife - beating, but the underlying idea that wife - beating is ever acceptable. Models 2 through 6 investigate whether women of certain cultural backgrounds perceive the five constraints presented in the wife - beating question differently. For example, in one ethnic group, women may perceive their responsibility to care for their children as very important, though cooking the food properly is unimportant, while in another ethnic group the opposite is true. These models investigate how women apply - being in different situations, depending on her social background. The following logistic regression equation is used for all six model s (Hoffmann 2016) : = + I calculated o dds ratios for the six models using the following formula: = RESULTS Descriptive Statistics. Table 1 and Figure 1 display the descriptive statistics of the six dependent variables used in the six models. Forty - e ight percent of women in the sample justify at least one of these five reasons. The most common reason that a woman would justify wife - beating is if she neglects the children (38%), followed by if she goes out without telling her partner (26%), argues with her partner (25%), refuses sex with her partner (19%), and burns the 17 food (8%). Of the all the ethnic groups, the Kikuyu ethnic group is the largest , most politically and economically dominant group, thus I use it as the reference variable in all models. 18 Figure 1. Percent of women who justify wife - beating by reason Table 2 provide s - beating by providing a cross - tabulation of the percentage of women of each ethnicity who justify wife - beating by reason. A strikingly high number of women claim that they would justify at least one reason fo r wife - beating in the Maasai (75%), Turkana (67%), Kalenjin (66%), and Samburu (59%) ethnic groups. Wo men in the Maasai, Turkana, and Samburu also have the largest percent of women who would justify wife - beating across all five reasons . Furthermore, the Ma asai are the group with the largest percent of women who would justify all reasons for wife - beating, except burning the food. The Embu/Meru ethnic groups have the smallest proportion of women who justify at least one reason (33%) and have the smallest perc entage of women who justify wife - beating 19 under three of the specified circumstances. Based on these descriptive statistics, ethnic identity appears to be related to whether a woman justifies wife - beating in any and all conditions. Ethnicity and Justific ation of Wife - Beating. T able 3 displays the six logistic regression models. Across all models, women of at least five ethnic groups have statistically different odds of justifying wife - beating when compared to those of Kikuyu women, holding all else consta nt. Models 2 through 6, of the specific reasons for justifying wife - beating, show that every ethnic group but the Somalis have statistically different odds than the Ki kuyus of justifying wife - beating in at least one situation. Based on the data in Table 2, it is not surprising that Maasai and Turkana women have a significantly higher odd s of justifying wife - beating across all six models in at the p<.01 level . The Maasai and Turkana, two pastoralist Nilotic groups that fall under the classification of Plain Nilotes, live in marginalized regions and that have been neglected by the government . These are the only groups of 11 comparison groups that have consistently higher odds of justifying wife - beating across all six models. 20 21 Across models 2 through 6 , eight categories, the Kalenjin, Kamba, Kisii, Luo, Embu/Meru, Mijikenda/Swahili, Somali, and Samburu all have odds ratios which are statistically different than the Kikuyu at the p<.05 level, though their odds are not statistically different from the Kik justifying wife - beating than the Kikuyu at p<.001 under the condition that a woman has argued with her partner. However, Luo women do not have statistically different odds in co mparison to the Kikuyu in any other situation. Interestingly, Kamba women have 1.698 greater odds of justifying wife - beating than the have .631 lower odds of justifying wife - condition that a woman has burned the food. In the situations presented in model 2 , 4 , and 5 , Kamba women do not have statistically different odds than the Kikuyus do. This is the only ethnic group which, in one model, has statistically greater odds than the Kikuyus do, and statistically lower odds in another model. Furthermore, results from model 1 reveal that, though the Kamba have statistically different odds from the Kikuyu in two models, their odds of justifying at least one reason for wife - beating are not statistically different from the Kikuyu. It is also the case among the Kisii, Luo, and Samburu that women do not have statistically different odds of justifying one or more reaso ns for wife - beating in comparison to the Kikuyu. Yet, in the Models 2 through 5 of specific reasons for wife - beating , these three groups also have at least one reason for which they have either statistically higher or lower odds of justifying than the Kiku yu. This would indicate that women in these ethnic groups generally have the same odds of justifying wife - beating when 22 compared with the Kikuyu. Yet, the groups differ by which cases of wife - beating they have greater or lesser odds of justifying. Educatio n. Across all 6 models, in comparison to women with no education, women with a secondary or higher education have significantly lower odds of justifying wife - beating when holding all else constant. For example, in M odel 2 women with a secondary education h ave 29.6% lower odds of justifying wife - beating in cases when a wife neglects her children than women with no education at the p<. 01 level. Additionally, women with any education higher than the secondary level will have 70.3% lower odds of justifying wife - beating than women with no education at the p< .001 level. For Models 5 and 6 , women with a primary education also have significantly lower odds of justifying wife - beating than women who have no education. Unlike n has a consistent ly negative effect across all six models. In all six models, any increase in education decreases the odds of justifying wife - beating, though this effect is only consistent ly significant at the secondary and higher levels. Migration. The variable ever moved - of - birth has a consistently - beating across all six models. Yet, this relationship is only statistically significant in determining whether wife - beating is justifi able when a woman neglects the children (p< .05 ), goes out without telling her partner (p< .01 ), and whether she justifies at least one reason for wife - beating (p< .05 ). For example, in model 1 , women who have migrated from their place - of - birth have 14.7% lower odds of justifying wife - beating in one or more situation s when compared with women who have never migrated. While less significant than the effects of secondary and higher education, moving does also have a consistent effect acros s all six models. 23 Control Variables. Women in the richer or richest household wealth category have consistently lower odds of justifying one and any specific reason for wife - beating. For each year that a woman increases in age, she has significantly lower odds of justifying wife - beating when a wife neglects the children, argues with her partner, goes out without telling her partner, or justifies at least one reason for wife - beating . For each additional living child, women ha ve significantly higher odds of justifying wife - beating in all situations but when a wife burns the food . Being in a polygynous union also has a consistently justifying wife - beating in comparison to being in a monogamous union, though the effect is al so not significant in influencing whether a wife justifies wife - beating when she burns the food. 24 D ISCUSSION Hypothesis 1: Justification of wife - beating is uniform across all ethnic groups . The at least in comparison to the Kikuyu, do have significantly different odds of justifying wife - beating. Therefore, I reject the null hypothesis that t he justification of wife - beating is uniform. The results from M odel 1 - beating in one or more scenario supports Mann and Takyi (2009) outcome variable in Ghana. I find that in Kenya ethnicity also influences whether a woman will justify at least one reason for wife - beating. Furthermore, while Mugoya et al. (2015) also argued that ethnicity plays a major role in determining perceptions of wife - beating in Kenya, they did so by looking at the count of reasons a woman would justify wife - beating, effectively measuring the degree specific conditions und - beating. Models 2 through 6 parse whether women across all ethnic groups have greater odds of justifying one condition over another in comparison to the Kikuyu group. When compared with M odel 1 , it becomes apparent that ethnic groups do not just differ by whether they will justify one or more cases of wife - beating, but by which cases they have different odds of justifying. Among women who do justify wife - have a uniform effect, providing them with consistently higher odds of justifying all the cases of wife - beating in wife - beating in comparison to Kikuyu women do not vary in a consistent direction. Women in 25 some groups, such as the Kamba, are more l ikely than the Kikuyu to justify wife - beating in one scenario yet are less likely to justify wife - beating than the Kikuyu under another circumstance. This suggests that each ethnic culture offers to its female members a different repertoire of tools and worldviews for when it is appropriate to apply tools of acceptance or rejection of wife - beating , supporting the conceptualization of patriarchy as being a context - dependent (Kandiyoti 1988) , and the idea that culture shapes worldviews and action (Swidler 1986) . In both tterned action that instruct them to follow scripts of complicity in or rejection of wife - beating. These scripts are Nevertheless, some groups in Kenya, particularly the Maasai, Samburu, and Turkana, do ha ve consistently higher justification of wife - beating in comparison to the Kikuyu. These three groups mainly inhabit the semi - arid lowlands of the Rift Valley , were neglected during the colonial era and the period before the Kenyan government decentralized 2016; Kanyinga 2009; Mburugu and Adams 2005) , and have lowe r rates of education and lower levels of household wealth (Kenya National Bureau of Statistics 2014) . Therefore, because wome n in these groups have had fewer chances to cultivate critical consciousness through the exchange and compar ison gender ideologies to that of other cultures, it is unsurprising that women from the Turkana, Samburu, and Maasai have much higher odds of justifying wife - beating under any situation than other ethnic groups . Hypothesis 2: Education has no effect on the justification of wife - beating . The results across all six models indicate that both secondary and higher education levels, in comparison to - beating. Across all models, any 26 increase in level of education consistent - beating. Therefore, I reject the null hypothesis that education has no effect. H ypothesis 3: Migration has no effect on the justification of wife - beating . The results life, consistent ly decrease s - beating . However, the effect is only statistically significant in determining whether a woman justifies wife - beating when a wife neglects the children, goes out without telling her partner, or when wife - beating is justifi able in at least one situation. Yet I am still able to reject the null hypothesis that migration has no effect on the justification of wife - beating. These significant findings alone are not sufficient to argue that education and migration s critical consciousness and reflection on the acceptability of gender - based violence. However, when one considers these interactions to be moments of unsettled life nested within a relational theory of empowerment , they contribute to the strengthening of our The concept of critical consciousness suggests that women who gain a reflexive vantage point from which they can critically assess their place - of - ble to reject and resist unjust norms (Cornwall 2016; Kabeer 2011) . Swidler (1986) suggests that we only question and reject the traditions and common sense of our culture when our lives become unsettled by a confrontation with competing ideology . When a woman moves to a new place or attends school, she immerses herself in new cultural ideology that presents her with new worldviews and the learning of these conflicting tools and frames , which create the possibility for reflexive and critical periods of unsettled life , that I suggest a woman can cultivate her critical consciousness. 27 The results affirm that education and migration both empower women to be able to reject - beating. Increased greater odds of rejecting any and all reasons for wif e - beating. Thus, education is not just a (Simister 2010) , or a form of cultural capital that teaches a woman how to reject wife - beating and apply new strategies of action (Rani et al. 2004; Stromquist and Gischman 200 9) . These findings support the idea that education is also a tool that women can use to cultivate critical consciousness (Cornwall 2016; Freire 1993; Guiné 2014; Kabeer 1999) . The variable ever moved affirms and furthers what we know about how ideology forms critical consciousness. This variable does not account for if a woman is moving from a place with more traditional beliefs about wife - beating to a more progressive place, or vice versa, but - beating. It indicates that simply moving from one place to another, irrespective of what kind of new ideology women encounter - beating. I suggest that this allows them to develop a critical gender consciousness. Fina number of children, and whether she is in a polygynous union reaffirm and are supported by the literature (Mann and Takyi 2009; Mugoya et al. 2015; Rani et al. 2004; Uthman et al. 2009) . 28 C ONCLUSION willingness to resist the patriarchal constraint of wife - beating. In order to resist the idea that wife - beating is acceptable, c ritical consciousness , or the realization of subordinate status through reflexive thought , must precede the struggle against it. (1986) theory of culture suggests that ethnicity, a proxy of culture, plays a significant role in perception of self , and that interaction with new ideology of any kind during periods of unsettled life can disrupt and reshape her beliefs about justifiable gender norms. This - beating. I have found t hat culture relates to the rejection/acceptance of wife - beating in two key ways. First, culture in when they can apply not simply determine the structure of patriarchal interests and constraints that she faces but provides her with strategies of action and an understanding about when she can apply those strategies. This is eviden t in the findings from models that indicate wha t conditions justify wife - beating. THESE Results suggest her odds of justifying any given reason for wife - beating, but that each ethnic identity has its own repertoire of more and less jus Second, exposure to new cultural ideology allows women to reshape their understandings of gender by comparing and reflecting on their own culture, thereby cultivating critical consciousness. This study affirms the large body of literature that claims that increased levels of education affect a decrease in the probability of a woman justifying wife - beating (Linos et al. 2010; Mann and Takyi 2009; Mugoya et al. 2015; Rani et al. 2004; Simister 2010; Stromquist 29 and Gischman 2009; Uthman et al. 2009) . I support the idea that education is not just the means through which women increase t heir access to resources and skills, but also of gaining the critical I also employ a new variable, whether a woman has ever migrated from her place - of - birth, to the analysis of a wo - beating. The effect of this variable is marginal, but it has a consistently negative effect on the odds of a woman justifying wife - beating. This indicates that exposure to a new culture does not simply mean that a woman will be re - ready to be refashioned in the image of whatever new ideology confounds their current one (Swidler 1 986:277) regardless of whether the ideology is progressive or traditional, improves the odds that she will critically reflect on her subordinate position within society. In studyin g the relationship between ethnicity, social identity, and beliefs about wife - beating, this study has several limitations . First, the DHS questionnaire does not have a question that specifically measure s whether a woman has actively reflected and still cho se to believe that wife - beating is justifiable . This hides an important population of women who have developed critical consciousness but justify wife - beating as a way of maneuver ing within constraint s . These understand that the cost of resistance is high. Rejecting wife - beating might mean running away from an abusive husband, and simultaneously, from the only family, community, and resources one has. Future qualitat ive research may be able to parse why unsettled experiences do not always lead to empowered beliefs. 30 Furthermore, this paper did not address why some women who believe that wife - beating is not justifiable do not act in resistance to the practice. The Keny a DHS has an intimate partner violence module that measures actual experiences with wife - beating. Future research c ould use this data to investigate why empowered beliefs do not always translate to empowered action. Especially because intimate partner viol ence is such an invasive topic, future studies should thoroughly analyze these studies before qualitative work ensues. I t is theorize d that after developing critical consciousness , women choose which ideological framework of pattern ed actions to apply to their lives based on their material conditions (Kabeer 1999; Swidler 1986) . The choice to believe or act in resistance is interwoven This paper has controlle d for household wealth index but has apply to their lives. It also does not discuss religion, as past studies have not found religion to be a significant determinant of wife - beating in Kenya. Future research should probe how, in other contexts, the intersection of ethnic, class, and religious identity affects wife - beating perceptions. Overall, this study supports the idea that interaction with new discourses about ge nder ideology, whether more or less conservative, create the opportunity for women to think critically about the nature of gender relations. This is a hopeful perspective; education and consciousness - raising initiatives are not the only means by which wome n can develop this reflexive thinking tool. In a world of increased migration and access to competing discourses on gender relations, these findings suggest that women will have greater opportunities to question and resist gender norms that they find unjus t. However, we should take heed that women who have restricted means of traveling and accessing new information are not neglected from these opportunities. 31 REFERENCES 32 REFERENCES Feminist Economics 3(1):1 51. World Development 35(11):1975 88. Batliwa la, S. 1994. Boston Massachusetts Harvard University Harvard Center for Population and Development Studies 1994 Mar. Batliwala, Srilatha. 1993. Empowerment of Women in South Asia: Concepts and P ractices . Mumbai. - Development in Practice 17(4 5):557 65. de Beauvoir, Simone. 2009. The Second Sex . edited by C. Borde and S. M. Chevallier. New York: Vintage Books. Social Indicators Research 115(1):137 58. World Development 64:194 205. Butler, Judith. 2004. Undoing Gender . Routledge. Collins, Patricia Hill. 1991. Black Feminist Thought: Knowledge, Consciousness, and the Politics of Empowerment . Routledge. New York. Cornwall, Andrea. Development in Practice 17(4 5):471 84. Journal of International Development 28(3):342 59. African Affairs 115(459):246 73. Freire, P. 1993. Pedagogy of the Oppressed . 20th anniversary. New York: Continuum. from Dalit Women International Journal of Educational Development 39:173 80. Violence & Victims1 12(3):191 208. 33 Hardi 14 in Feminism and methodology: Social Science Issues . Bloomington: Indiana University Press. Hoffmann, John P.(John Patrick). 2016. Regression Models for Categorical, Count, and Related . Development and Change 30:435 64. Development and Change 42(2):499 528. Gender and Society 2(3): 274 90. Post - Journal of Contemporary African Studies 27(3):325 44. Kenya National Bureau of Statistics. 2014. Kenya Demographic and Health Survey 2014 . Nairobi. Gender and Society 4:9 24. Journal of Interpersonal Vio lence 23(8):1056 74. Sociological Theory 2:23 101. Linos, Natalia, Marwan Khawaja, and Mohannad Al - Support for Wife Beating: Findings from a Populat ion - Violence and Victims 25(3):409 19. Sociological Forum 12(4):599 630. Retrieved (ht tp://www.jstor.org/stable/684734). Impact of Resources and Socio - Cultural Processes on Attitudes Towards Intimate Partner Journal of Fami ly Violence 24:323 35. 24 in Handbook of World Families , edited by B. N. Adams and J. Trost. Thousand Oaks: Sage Publications. World Development 78:360 71. Victimization Factors That Impact Attitudes Toward Intimate Partner Violence Among Journal of Interpersonal Violence 30(16):2851 71. 34 Politics in Livestock Raids i Journal of Modern African Studies 55(3):479 509. Rethinking Empowerment: Gender and Development in a G lobal/Local World . New York: Routledge. Rani, Manju, Sekhar Bonu, and Nafissatou Diop - Attitudes towards Wife - Beating among Men and Women in Seven Sub - Saharan African African Journal of Reproductive H ealth 8(3):116 36. Rowlands, Jo. 1997. . Oxfam. Journal of Family Violence 25(3):247 57. Stith, Sandra M., Douglas B. Smith, Carrie E. Penn, David B. Ward, and Dari Tritt. 2004. - Aggression and Violent Behavior 10(1):65 98. - From Denouncing Gender Inequities to Undoing Gender in Education: Practices and Programmes Toward Change in International Review of Education 55(5):463 82. 86. Swidler, Ann. 2001. Talk of Love: How Culture Matters . University of Chircago Press. A Democratization 24(6):951 69. (http://www.unmillenniumproject.org/goals/). DHS Publication 1 7. Attitudes towards Intimate Partner Violence against Women: A Comparative Analysis of 17 Sub - BMC International Health and Human Rights 9(1):14. We Currently Married Women Aged 15 - 49 in Zimbabwe: Evidence from the 2010 - 11 Zimbabwe Demograp African Population Studies 28(3):1413 31. Value - Rural Sociology 81(4):545 71.