Factors influencing attitudes towards intimate partner violence in Bangladesh

Ullah, M. R & S. Parvin (2015) described that the socio-economic status of women influences on domestic violence in Dhaka city. They found that lack of higher education, productive activities, occupational status, dependency on husband income, and less access to mobility are considered as major factors for increased violence against women in the urban area of Bangladesh. The study showed that the lower position of women at different levels also responsible for the high rate of verbal and psychological violence in their married life. Moreover, their high dependency on men was still found in the urban area and less engagement was in productive economic activities.

Afsana, Rashid, and Thurston (2015) mentioned that the impact of domestic violence/intimate partner violence was not addressed as a distinct issue at the policy level. The national policy of Bangladesh did not clearly address the issue of violence against women. Though, the government of Bangladesh introduced to a one-stop crisis center in the hospital run by the Ministry of Women and Child affairs. The center usually offers medical treatment for victims and also arranged legal and social supports through NGO. It is necessary to have a comprehensive understanding of the policy makers and planners regarding the effect of domestic violence on women's health.

Hossain, A., (2016) explained that violence against women is a major threat in Rural Bangladesh. Despite the remarkable achievement of women development and empowerment, violence against women is still prevailing in society and deemed as a burning issue. Husband, in-laws and other family members domestically involve in violent activities against women. Existing social norms and values are supporting such violence in the society of Bangladesh.

Several empirical and qualitative studies were conducted on physical/domestic violence/ intimate partner violence against women by scholars from different disciplines. These previous studies mainly focus on the question that the effect of different factors on men and women's attitude and practice of violence against partners. This study is methodologically different from previous studies. No studies might be available on internally migrated men's attitudes towards IPV at urban slum and urban non-slum areas of Bangladesh. This study is striving to examine the influence of different factors on men's attitude towards their partner (wives).

2.2 Theoretical Background

Intimate partner violence is a common phenomenon that is explained in different disciplines. Social scientists, public health experts, and other disciplines' scholars have been striving to analyze and understand the characteristics, causes or driving factors, effect and consequences of intimate partner violence against women from different perspectives. In Psychology discipline, frustration-aggression theory, the social learning theory, and cognitive behavioral theory are popular to explain the causes of IPV and its risk factors. In addition, there are also some sociological theories striving to explain IPV, including feminist theory, family conflict theory, resource theory and its branches of relative resource theory and gendered resource theory, and dependency theory. In this study, feminist theory and the ecological model of intimate partner violence have been briefly discussed to understand the term theoretically.

The Feminist Theory:

Second wave feminist theories (the 1960s and 1970s) emphasize on the need for power and control on the societal arrangements of patriarchy and tolerance (not support for) of VAW that supports individual abusers in terms of this behavior as acceptable (Kelly, 2011). The feminist theorist highlights “power and control in relationships, social norms overlooking wife beating, and structural and economic forces that keep women trapped in abusive relationships” (Heise 2012, p. 47). According to contemporary feminist theorists, women's subordination is seen as inherent and part of overall exploitation and injustice. In order to their view, women are not only ignored due to gender identity or being female but also for their socioeconomic status. Some radical feminist theorists viewed that oppression of women is visible in all male-dominated societies whether it can be capitalist or socialist or communist societies. However, according to the socialist feminist theory, women are exploited and victimized of violence due to patriarchal structure and lack of value. They also claim that violence exists in society as a result of economic exploitation. Another model was developed by Pence and Paymar (1986) focused on a pattern of behavior considered to dominate and control one's partner, and is represented as a “power and control wheel”. The spokes of the wheel represent forms of abuse not connecting to actual assault. These “tactics” of domination and control comprise of verbal, psychological, financial and social abuse, threats and intimidation and use of “male privilege”.

The Ecological Model

The ecological model was proposed by Heise (2002) and used by WHO is measured as one of the most promising approaches for understanding the interaction of different risk factors associated with IPV. This framework incorporates research findings and theories from several disciplines, adding feminist theory, into an explanatory framework of the backgrounds of gender-based IPV. Through this ecological framework, IPV is easy to understand as a multidimensional phenomenon that is the outcomes of dynamic interaction among individual, relationship, community, and societal factors that influence a person' risk to perpetrate or become a target of violence (Kelly, 2011).

The ecological model highlights some risk factors according to the following four levels of influence (WHO, 2010; Kelly, 2011):

i. Individual: focus on biological and personal history factors that may increase the possibility of a person become a victim or perpetrator of violence.

Figure 2.1: The Ecological Model for IPV (WHO, 2010; Kelly, 2011)

ii. Relationship: Focus on such relationships with peers, intimate partners and family members that may influence the risk of a person to perpetrate or become a victim of violence. There is an asymmetrical power structure within the relationship.

iii. Community: Focus on such social relationship as neighborhoods, schools, and workplaces. Community poverty, unemployment, and alcohol outlets have also been measured as risk factors for the perpetration of violence or victimization by violence

iv. Societal: Broadly focus on factors that affect intimate partner violence at the individual, relational and community level because of cultural and religious beliefs and activities, gender inequity and inequality, social norms, social policy, and economic system create gaps and tensions among groups.

This model assists to examine the influence of different factors on men's (perpetrators) attitudes towards intimate partner violence in Bangladesh. Four different level factors are mentioned here: individual (socio-demographic characteristics, lifestyle, and health status); relational (domestic authority/decision-making power); community (exposure to mass media) and societal (gender-role beliefs, and attitudes toward IPV). intimate partner violence migrating

Chapter 3. Data & methods

3.1 Data: The data in this study are drawn from the 2013 Urban Health Survey conducted by National Institute of Population Research and Training (NIPORT), MEASURE Evaluation, University of North Carolina at Chapel Hill, USA, International Centre for Diarrheal Disease Research, Bangladesh (Icddr,b) and funded by United States Agency for International Development (USAID) and the United Kingdom's Department for International Development (DFID)

The study explores the experience and attitudes towards intimate partner among man and women in urban areas of Bangladesh.

3.2 Research Design

A cross-sectional study was conducted among internally migrated ever-married male (age 15-59) living in Slum and Non-Slum areas in Bangladesh.

Study Area: Slum and Non-Slum located in Urban areas of six divisions like Dhaka (5617), Chittagong (4094), Barisal (577), Sylhet (601), Khulna (1527) and Rajshahi (895) of Bangladesh.

Sample Technique, Size, and Frame: Cluster sampling has been used for selecting the area (Slum and non-slum).

The dependent variables for this study are - (1) man would be justified to beat his wife - she neglects child, (2) man would be justified to beat his wife - she argues husband, (3) man would be justified to beat his wife - she fails to provide food on time, (4) man would be justified to beat his wife - she visits to family without her husband permission, (5) man would be justified to beat his wife- she visits friend without her husband permission. Respondents answers “No” were coded -0 and those respondents said “Yes” were coded -1. Respondents answer as “no opinion” (total 679) were dropped from each variable in this research analysis.

Figure 3.1: Sample Frame

Definition of Variables:

Dependent/ Outcome Variables

Independent Variables

a. Socio-demographic characteristics: The independent variables are characterized by socio-demographic characteristics namely [1] age of respondents, [2] type of household, [3] highest grade of schooling, [4] religion, [5] currently working, [6] how is your health (health status) [7] domain. Age (in years) of respondents will be specified in four ten year categories: 15-24, 25-34, 35-44 and 45-59. Three categories will be measured for educational level: none, complete primary schooling, and more than secondary education. Type of household is classified as “non-mess” and “mess”. Employment status is specified as “yes” and “no”. Religion is represented by a dichotomous variable: Muslim versus non-Muslim. The response to health status categories was recoded as very healthy, somewhat healthy, somewhat unhealthy, or unhealthy. For this study, health-status is reclassified into a dichotomized variable: "excellent health" (two highest categories) and "poor health" (two lowest categories). The domain is categorized into slum and non-slum (recoded as Slum-1 and Non-Slum-0). Responses of other municipalities (1,989) are dropped from the survey. The response of “where you born” are categorized into city corporation, district town, other town, village, abroad. The variable “where you born” is recoded as “internally migrated men” (Coded-1) and “internally non-migrant men” (Coded-2).

b. Lifestyle: The two dichotomous variables are used for identifying the lifestyle of respondents: [1] Alcohol or drug use in the past month (Yes=1, No=0) and [2] Had Sexually transmitted disease (STD) in past six months (Yes=1, No=0). Men who habituated with drinking alcohol and taking recreational or addiction drugs are becoming more aggressive towards partners (Kulis et al. 2003 cited by Sambisa et al., 2010).

c. Decision-Making Power: Three main variables are measured to decision making the power of a man in the household such as [1] decision about healthcare, [2] decision about hh purchase, [3] decision about a visit to friends. The answers of respondents to this variable are coded into one of the three categories: (1) wife-dominated, (2) husband-dominated, and (3) joint decisions (4) someone else.

d. Role of Gender: The two variables which assessing men's perception of the acceptability of married women's working in outsides the home: [1] Not acceptable for a wife to work outside the home under specific circumstances and [2] Acceptable for a wife to work outside the home under specific circumstances. The respondents' answers are coded into dichotomous Yes=1 and No=0.

e. Exposure to Mass Media: This study will also include the variables of exposure to mass media such as [1] read paper and magazine, [2] listen to the radio, [3] watch TV. All variables answers are represented by a dichotomous variable- “yes” or “no”.

Hypothesis: Following hypothesis are designed based on objectives of research and problem statement:

· The socio-demographic factors are negatively associated with the justification of beating wives by men.

· Internally migrated men are more likely to justify beating their wives than internally non-migrant men living in a slum and non-slum areas.

· Lifestyle is positively associated with the justification of beating wives

· Decision-making power and gender role positively influence on men's justification of beating wives.

· Exposure to mass media is negatively associated with men's attitudes

Data Management and Analysis: Data are entered and analyzed by using STATA 13.0. Descriptive analyses are performed to calculate the prevalence estimates for all dependent and independent variables. Multivariate Logistic regression is carried out to study the association between men's attitude towards IPV (dependent variable) and socio-economic and demographic characteristics (independent variables) such as age, education, occupation, income, religion, household type and living place and other independent variables. The estimated coefficient is considered the value by which unit of the event (attitude towards violence) change when the independent variable is increased by one unit/step. And it is calculated by adjusting for all other independent variables in the multivariate model. A P value of 0.1, 0.05 and .01 was considered as the level of significance. Although, all results are weighted by population weights known from the sample design.

Chapter 4. Descriptive statistics & empirical analysis

4.1 Statistical Analysis

Three types of statistical analysis are carried out in this study. First of all, descriptive statistics (total observation, mean, standard deviation, minimum value, and maximum value) are calculated of each dependent and independent variables that examine the justifying IPV against wives. Secondly, the prevalence of each item is presented in the tables (4.2 & 4.3). Thirdly, multivariate logistic regression is calculated and presented the men's attitudes towards IPV against wives of urban slum and non-slum area in the table (4.4) and also presents the regression analysis outcomes separately for internally migrated men in the table (4.5) and another table (4.6) is presented for internally non-migrated men living both in urban slum and non-slum areas of Bangladesh.

4.2 Descriptive Statistics

The table-4.1 presents the summary statistics of all outcome variables and the major independent variables based on intimate partner violence. The table shows that the mean of justifying beating wife for visiting a friend without husband permission is high (0.33) compared to other justification of beating wife variables and the spread of data value is also higher (0.47) than the other relevant variables. The lowest mean and data value variation (0.18 and 0.38) of justifying beating a wife is for failure to provide food on time. In addition, the means of justifying beating wife for neglect child and visit family without husband permission are equal, but the spread of data value for neglect child variable is slightly higher than the visiting family without husband permission variable. In the part of socio-demographic characteristics, the means of age group (2.29) and education (3.1) and its data value variation are higher in comparison with other socio-demographic factors. For the type of household, the mean is comparatively higher than the religion variable. On the other hand, the mean of religion (0.78) is the lowest compared to other socio-demographic variables. It also shows that the means and data value variation of working status and health status are the same.

Table 4.1: Summary statistics of the dependent and independent variables

Among the lifestyle variables, the mean and data value variation of taking alcohol or drug use is higher than the mean of STD variable. In the household decision-making variables, the mean (1.67) of household decision for visiting family is higher than the means of household decision for health care and household purchase variables. The data spread of the household decision for household purchase and a visit to family variables are alike. Among the gender role-related variables, the mean of married women works outside if husband making not enough money variable (0.68) is higher in comparison to the mean of married women work outside if husband making enough money variables. There are media exposure related variables such as read paper/magazine, listening to the radio and watching TV. Among these variables, the means of reading paper/magazine (0.89) and watching TV (0.94) variables are relatively higher than the mean of listening radio variable, but the data value variation of listening radio variable is greater than other two media-related variables. The mean of the domain (slum and non-slum) is with 0.52 and the data value variation is at 0.50.

4.3 Prevalence of Factors and Men's Attitude towards IPV

The following table-4.2 describes the percentage of socio-demographic factors and attitudes towards intimate partner violence for five different reasons in urban slum and non-slum areas of Bangladesh. The result shows that 53% of men are living in the urban slums compared to the urban non-slums (47%) based on sample size. It is also found that most of them (72%) are internally migrated living in both urban slum and non-slum areas, whereas, 28% is recorded as internally not migrated. The large percentage of people is with age 16- 25 and 26-35 are in the sample. Almost 44% of men have completed secondary schools and the following 31% of men have finished their primary schooling. More than 90% of households are non-mess and a big percentage of men are religiously identified as Muslim. Moreover, above 90% of men are currently working and living with excellent health. The table (4.2) also shows that the prevalence of justifying IPV for all reasons excepting failure to provide food on time is higher among men living in urban slums compared men to urban non-slums. It is also found that the prevalence of men's attitude towards justifying beating wife is higher among internally non-migrant people compared to internal migrants. It is surprisingly seen that the prevalence of justifying IPV is decreasing with respondents' ages. Young men are more favor IPV than older men for all reasons. Education is strongly correlated to IPV. The prevalence of justifying IPV is less among higher educated men in comparison to men with secondary and below level of education. The prevalence of men living in mass households is more agreeable to IPV. The result also reveals that Non-Muslims are more favor IPV compared to Muslims. Furthermore, men who are currently working and have excellent health are also more justify beating wives for defined reasons.

Table 4.2: Prevalence of socio-demographic factors and men's attitude towards IPV

The table -4.3 shows the prevalence of lifestyle, household decision-making, gender role and exposure to mass media factors. 13% of men who consumed alcohol or used the drug in the past month and only 3% of men were affected by STD in the last six months. Majority of men (72%) don't support to married women work outside if husbands making enough money. On the other hand, approximately 62% of men support to married women work outside if the husband is not making enough money. Most of the respondents are reading a paper/magazine (87%) and watching TV (93%) compared to listening to the radio. In order to the table, most of the household decision regarding health care (86%), household purchase (64%) and a visit family (61%) are dominated by the husband. More than 30% of household decision about the household purchase and a visit to the family are taken by husband and wife jointly. The table - 4.3 also explains that the prevalence of justifying IPV for all reasons is high among men who consumed alcohol or used drug. Similarly, it is also high among men who affected STD in the last six month compared to men who have not affected by STD. It is also found that the prevalence of justifying IPV is high among men who accepted the statements that married women can work outside if the husband earns enough money and married women can work outside if the husband earns not enough money. Even the prevalence of favor IPV for all reasons more common among men who exposure to mass media. Moreover, the highest prevalence of justifying IPV for all reasons are seen among men who take household decision-making regarding health care, household purchase and visit family.

Table 4.3: Prevalence of other factors (lifestyle, decision-making power, gender role & exposure to mass media) and men's attitude towards IPV

4.4 Model of Specification

The empirical estimation focuses on the effect of different factors on men's attitude toward intimate partner violence in slum and non-slum areas of Bangladesh. The baseline model is given as below:

Y = в0 + в1*age_ group + в2*education + в3*household type + в4*religion + в5*working status+ в6*health status + в7*alcohol/drug use + в8*STD status + в9*decision_ health care + в10*decision_ household purchase + в11*decision_ visit family + в12*women working_ enough money + в13*women working_ not enough money + в14*reading paper + в15*listening radio + в16*watching TV+ в16*domain + ?.

Where Y signifies the outcome variables for measuring men's attitude towards intimate partner violence. The outcome variables include justify beating wife for neglect child, argues with husband, fails to provide food on time, visit family without husband permission, visit a friend without husband permission. Moreover, some socio-demographic variables such as, age group, education, household type, religion, working status, health status; human lifestyle variables like alcohol/drug use and STD status; decision-making variables for example, decision for health care, household purchase, visit to family; gender role variables for instance, married women working if husband earn enough money, married women working if husband earn not enough money, access to media variables such as reading paper, listening radio and watching TV are included as explanatory variables.

4.5 Empirical Results

In this study, we examine the influence of different socio-demographic, lifestyle, household decision-making power, gender role and mass media exposure factors on men's attitude toward intimate partner violence both in urban slum and urban non-slum areas of Bangladesh.

Men's Attitudes towards Intimate Partner Violence in Slum and Non-Slum Areas

The table-4.4 presents the results from adjusted logistic regression models calculating men's attitude towards IPV against their wives in a slum and non-slum areas of Bangladesh and reports estimated coefficients of variables. From the results, it is found that the respondents' age is one of the significant factors for justifying IPV. The estimated coefficient for the men's age group 46-55 is negative and statistically significant at all levels. Men with age of 46-55 are less likely to justify IPV for all reasons compared to the other age groups 16-25, 26-35 and 36-45. Moreover, men with an age of more than 56 are also less likely to favor IPV for all reasons, excepting fails to provide food on time.

The likelihood of justifying IPV against wives is less among men with higher secondary and higher education. Men with primary and secondary education are also less likely to favor wives beating for all reasons, excluding the reason for neglect child. Men of non-mess households are less likely to favor IPV for reasons- fails to provide food on time, visit family without husband permission and visit a friend without husband permission compared to men of mess households. Furthermore, Muslim men are more likely to justify beating wife or IPV for arguing with husband and visit friends without permission of him in comparison to Non-Muslim men. In addition, Muslim men are less likely to agree on IPV against women to failure to provide food on time.

No statistically significant results are found between men currently working and not working. Men with excellent health are negatively likely to justify beating wives for arguing with husband and visit friends without permission of husband compared with men with poor health status. However, other reasons for beating wives are not statistically significant at all levels. The association of men's lifestyle and some justifications of IPV is positive and statistically significant. Men who consumed alcohol or used the drug and have STD are significantly more likely to favor beating wives for all reasons, excepting the reason for failure to provide food on time in comparison with men who did not consume alcohol or use the drug and have STD.

Table 4.4: Table showing Logistic Regression outcomes for the relationship between the dependent and independent variables in Slum and Non-Slum Areas of Bangladesh

Standard errors in parentheses

*** p<0.01, ** p<0.05, * p<0.1

The above table -4.4 also shows that for a wife who argues with the husband and taking health care decision individually, men are more likely to justify beating them. Furthermore, wives who neglect child, but jointly taking health care decision with husband, men are more likely to justify beating. When health care decision is taken by someone else, are more likely to favor beating wives for all reasons. No statistically significant outcomes are found for a household decision on household purchase variable. When wives take a decision to visit family men are less likely to support beating wives for argues with husband and visit a friend without husband permission. When the household decision on visit family is taken by wife and husband jointly, men are less likely to justify beating for all reasons. Men who have accepted that married women should work outside if the husband earns enough money, are less likely to justify beating the wife for all reasons excluding reason for failure provide food on time. Men who have accepted that married women should work outside if the husband earns not enough mone...