The family is typically a source of refuge—a place where individuals seek love, comfort, and security. However, for many women, this refuge becomes a place of mortal danger, with some of the most tragic forms of violence being perpetrated against them at home. In fact, violence in intimate relationships is the most common form of gender-based violence experienced by women across the world (Duvvury et al. 2004). Violence in the domestic sphere is referred to as domestic violence and takes place when one partner is dominated and controlled through physical or psychological means. Domestic violence comes in many forms, such as physical and sexual abuse, threats and intimidation, emotional and social abuse, and financial deprivation. In most cases, the violence is perpetrated by those who are, or who have been, in positions of trust, intimacy, and power, and more often by men against women. Although women can also be violent, their actions account for a small percentage of domestic violence. Recognizing domestic violence as a worldwide problem, in 1999 The United Nations General Assembly designated November 25 as the International Day for the Elimination of Violence Against Women.

Despite advancements made by women in almost all areas of social and economic development, the global dimensions of domestic violence are alarming. Domestic violence occurs in countries worldwide and traverses ethnic, religious, and income groups, and that the only variation is in the patterns and trends that exist in countries and regions. A 2005 World Health Organization (WHO) study based on data from over 24,000 women in 10 countries representing diverse cultural, geographical, and urban/rural settings found that the proportion of women who had ever suffered physical or sexual violence, or both, by a male intimate partner ranged from 15% in Japan to 71% in Ethiopia province, with most countries falling between 29% and 62% (Garcia-Moreno et al. 2005). Even a country as advanced as the United States is not spared from this epidemic that has devastating physical, emotional, financial, and social effects on women, children, families, and their communities. Data from a US population-based study showed that 29% of women had ever experienced physical violence, sexual violence, and abuse of power and control (Coker et al. 2002).

Previous studies have examined the determinants of violence within the family (Farmer and Tiefenthaler 1997; Tauchen and Witte 1995; Tauchen et al. 1991) in one of the pioneering studies in this area. In their model, the man punishes the woman with violence for behavior of which he does not approve, and the woman adjusts her behavior in an attempt to minimize this loss. Under the Tauchen et al. (1991) model, the impact of income on level of violence was analyzed. Both spouses have the ability to make transfers to the other, and each has a threat point level of utility that must be maintained. In this model, when a woman is held to her reservation utility, an increase in the man’s income enables him to ‘buy’ more violence by increasing transfers to her; an increase in her income compels him to reduce the level of violence. When both partners enjoy a surplus from the relationship however, the effect of increased income on domestic violence is ambiguous, depending among on other things, the change in his marginal utility of violence with respect to her consumption.

Tauchen and Witte (1995) investigated how police treatment in cases of domestic violence affects subsequent violence, and found a significant relationship between violence and the man’s employment. Farmer and Tiefenthaler (1997) reported that a woman’s income and other financial support received from outside the marriage (family, welfare, shelters, divorce settlements, etc.) will decrease the level of violence in intact families because they increase the woman’s threat point. In their study, the marriage is modeled as a strategic relationship in which each partner maximizes utility subject to the constraint that the other remains in the marriage. The presence of children was the key marital capital variable tested in this study (Farmer and Tiefenthaler 1997). In one data set, they found that the presence of children has a negative and marginally significant effect on the violence. In another data set, they found that the presence of young children (under 6 years old) has a negative but insignificant effect, while the presence of older children has a significant positive effect on level of violence. A similar significant positive association was found between duration of the relationship and level of violence. Other recent investigations focused on the types, prevalence, and risk factors of intimate partner violence (Ammar 2006; Cunradi et al. 2008; Miller 2006; Stith et al. 2008; Zohreh and Nouzar 2008). For example, Stith et al. (2008) carried out a meta-analysis to investigate the relationship between marital satisfaction/discord and intimate partner violence in heterosexual relationships, while Zohreh and Nouzar (2008) examined the prevalence of the types of domestic violence and its associated risk factors among pregnant women in Iran.

Similar to other countries, domestic violence is a serious but often hidden problem in Malaysia. A national survey conducted by Women’s Aid Organisation (WAO) between 1990 and 1992, estimated 1.8 million or 39% of women over the age of 15 were beaten by their husbands or boyfriends in 1989 (Abdullah et al. 1995). The incidence of under-reporting in Malaysia is also significant—of the estimated 1.8 million women, only 909 reported the violence to the police in 1989. Various factors contribute to the reason why domestic violence is an under-reported crime in Malaysia. These include the reluctance of battered women to make their problem public and seek assistance due to shame, fear of retaliation from their husbands, lack of family support, unawareness that physical abuse is a crime, and a belief that the police and the legal system cannot help them. Under-reporting of domestic violence is also attributed to differences in women’s definitions and perceived causes of domestic violence in different settings and cultures (Tehee and Willis Esqueda 2007). However, it is important to note that Malaysia has made some progress in terms of having specific legislation addressing the issue of domestic violence. The Domestic Violence Act (DVA) was passed by the Malaysian Parliament in 1994 and implemented in June, 1996. Under the DVA, domestic violence is not defined as a specific crime punishable by new penalties; however, it is attached to the Panel Code under definitions and procedures for hurt, criminal force, and assault. The DVA defines domestic violence as an act by a person against his or her spouse, his or her former spouse, a child, an incapacitated adult, and any other member of the family.

This paper investigates the extent of domestic violence within a multi-ethnic society in Malaysia. Specifically, it examines the relationship between the level of abuse in violent households and the demographic as well as socio-economic characteristics of the survivors of domestic violence and the perpetrators of the abuse. Marital capital factors, such as the number of children and duration of the marriage or cohabitation which may affect domestic violence, are also included in this study.

Methodology and Data

This study is based on secondary data obtained from Women’s Aid Organisation (WAO) Malaysia, which among other services, provides temporary shelter and counselling to survivors of domestic violence and women in crisis at the WAO Refuge. The data generated from WAO case files of women who sought shelter services during 2002 to 2005 were originally culled from open-ended interviews conducted by WAO social workers at the point when the women arrived at the shelter.

The data contain selected socioeconomic information on the survivors of domestic violence and the perpetrators of the violence, details on the nature and type of violence, as well as the frequency of the domestic violence incidences. In this data set, five types of abuses are characterized as acts of domestic violence namely, physical, psychological, social, financial, and sexual. Physical abuse is listed as a range of violent acts including beating with hands and kicking, beating with objects, throwing objects, hitting against the wall, assault with a knife, and scalding or burning. Psychological abuse includes humiliating the survivor and controlling what the survivor can and cannot do; social abuse includes isolating the survivor from family and friends; financial abuse occurs when the perpetrator has some or total control over the survivor’s money and other economic resources, and/or preventing the survivor from obtaining employment; and sexual abuse refers to all forms of coerced sexual activity by the perpetrator on the woman. One filtering condition imposed on the data was to include observations relating only to survivors when the perpetrators of the abuse were their male, romantic counterparts to whom they were either married to or lived with. Observations relating to women who had sought shelter at the WAO Refuge in the past (i.e., repeat visitors) were omitted to eliminate the possibility of double counting.

The number of case files in the period 2002 to 2005 used for this study is 164, as shown in Table 1. The dependent variable in this analysis is the level of violence in the household measured by the frequency of its occurrence as shown in Table 2, while the independent variables include demographic and socio-economic characteristics. Descriptive statistics, cross-tabulations, chi-square tests and multinomial logistic regression analysis were used to examine patterns of domestic violence, as well as the relationship between level of violence and the independent variables.

Table 1 Distribution of domestic violence case files by year
Full size table
Table 2 Frequency distribution of domestic violence
Full size table

The data in Table 2 suggest that occurrence of abuse is quite high with 26% of the survivors being abused nearly daily, 37% abused up to three times a week, and 26% are victims of an unpredictable frequency of abuse.

In terms of the demographic and socioeconomic characteristics of the survivors and perpetrators, it can be observed from Table 3 that generally the perpetrators are older and earn a higher monthly income than the survivors. Almost half of the survivors are in their 30’s while 33% are aged 29 years and younger. Similarly, slightly more than half of the perpetrators are in their 30’s (52.5%). However, the proportion of perpetrators in the oldest age group (40 years and older) is about 33% compared to only 18% of the survivors. The data show that nearly 8% of the perpetrators are aged 50 years or older while none of the survivors is in this age group. In terms of ethnic breakdown, the majority of the domestic violence victims are in the Indian ethnic group followed by the Malays, and Chinese. It can be observed that ethnic variation in the proportion of survivors and perpetrators within the same ethnic group is very small. Indian survivors and perpetrators constitute 61% and 58% of the total cases, respectively, while the proportion of Malay survivors and perpetrators is 22% and 24%, respectively. There is no variation in the proportion of survivors and perpetrators among Chinese (17% for both survivors and perpetrators). This suggests very few cases of inter-ethnic marriages in women who had gone for shelter at the WAO.

Table 3 Percentage distribution of demographic and socio-economic variables
Full size table

A third of the survivors completed 11 years of schooling, 25% of them completed 9 years of schooling and 19% had 6 years or less of school education. However, unlike the survivors, the largest number of perpetrators had 6 years or less of schooling (32.0%), 24% completed 9 years and about 21% of them completed 11 years of schooling. There is no difference in the proportion of survivors and perpetrators with at least 12 years of education (23.0%). Table 3 also shows that more than half of the survivors (53.5%) did not have any income, while the proportion of perpetrators with no income is only 17%. The substantial difference in income between the survivors and perpetrators is also observed in the monthly income bracket of RM2000 and above (19% of survivors and 44% of perpetrators). The proportion of survivors with monthly income of RM1000 or less is quite similar to that of the perpetrators (15.3% and 16.4%, respectively).

The marital capital variables used in this study are the number of children and length of marriage or cohabitation as shown in Table 4. The analysis takes into account the number of children born by the victim and her perpetrator. Ten percent of the survivors had no children while 49% had one to two children. Five percent of the remaining 41% of the survivors with three or more children actually had at least six children. In terms of the length of marriage or cohabitation, the data show that only 6% of the survivors were married or cohabitating for less than a year and slightly more than 10% were with their partners for more than 15 years. For the purpose of analysis, the length of marriage or cohabitation is grouped into three categories with 38% of the survivors reported being together with their partners for 5 years or less, 29% for six to 10 years and the remaining 33% for at least 11 years (Table 4).

Table 4 Distribution of the number of children and length of marriage/cohabitation
Full size table

Results

Chi-square tests were employed to examine the association between frequency of violence and selected demographic and socioeconomic variables of both the survivors and perpetrators and the results are presented in Table 5. For this analysis, the income of survivor was regrouped into two categories, namely women with no income and those with income to avoid having cells with very small number of observations. Similarly, the educational attainment of both the survivor and perpetrator and income of perpetrator were reduced to two categories. Three factors are found to be statistically associated with frequency of violence, age of perpetrator (Chi-square = 17.62, df = 6, p = .007), income status of survivor (Chi-square = 8.25, df = 3, p = .04), and number of children (Chi-square = 12.62, df = 6, p = .05). Older perpetrators appear more likely to abuse nearly daily (37% of perpetrators aged 40 years and older) compared to their younger counterparts (17% of perpetrators aged 29 years and younger). Meanwhile younger perpetrators are more likely to abuse once or twice a month (30% of perpetrators aged 29 years and younger) compared to their older counterparts (4% of perpetrators aged 40 years and older).

Table 5 Pearson chi-square tests between frequency of violence and selected variables
Full size table

The cross tabulation of the frequency of violence and the income status of survivors indicates that 37% of survivors who have a monthly income were abused nearly daily compared to 17% of survivors who did not have a monthly income. However, the proportion of survivors who were abused between one to three times per week is higher among women with no income compared to those with income (43% and 31%, respectively). The proportion of survivors who were abused once or twice a month is similar for both groups (11%).

The other significant association is found between frequency of domestic violence and the number of children. A much higher proportion of women with three or more children were abused nearly daily compared to those with one or two children (33% and 26%, respectively), while none of survivors who do not have children were abused nearly daily. More than half (56%) of survivors who did not have children were abused one to three times a week compared to 41% of survivors who have three or more children. The data suggest that for survivors who have children, the level of abuse is likely to increase as the number of children increases, although there are a number of cells with expected values less than five.

Having identified the significant associations between the frequency of violence and age of perpetrator, income status of survivor, and number of children, the data were further analyzed to examine the direction and strength of these factors. A multinomial logistic regression was carried out incorporating these three independent variables in the model (see Table 6). The reference category for the dependent variable is unpredictable occurrence of abuse, while for the independent variables the respective reference category is age of perpetrator 40 years and older, survivor with monthly income, and having at least three children.

Table 6 Multinomial logistic regression of frequency of violence
Full size table

It can be observed that the three determining factors affect the level of violence quite differently. The column Exp (B) provides the odds of women being abused for a given frequency violence level as opposed to unpredictable frequency. Age of perpetrator and income status of survivor are found to have a significant influence on the odds of women being abused nearly daily. The result shows that the odds of a survivor whose partner is aged 30 to 39 years being abused daily (as opposed to unpredictable frequency) is 0.375 that of a survivor whose partner is at least 40 years of age, while the odds of a survivor with no income being abused daily (as opposed to unpredictable frequency) is 0.311 that of a survivor with income. Examining the occurrence of violence between one to three times a week (as opposed to an unpredictable frequency), the only significant factor is the number of children where the odds of a survivor with one to two children experiencing this level of violence is 0.441 that of a survivor having at least three children. For violence that occurs once or twice a month (as opposed to unpredictable frequency) the odds of a survivor whose husband is younger than 30 years being abused is 15 times that of a survivor whose husband is at least 40 years old.

Discussion

Three significant determinants of the level of domestic violence emerged from this study: age of the perpetrator, income status of the survivor, and the number of children. The data indicate that older men are more likely to inflict daily abuse than younger men, while younger men are more likely to inflict abuse between one to three times per week. One reason for this could be that as a man ages, he gets more used to asserting his dominance in an aggressive and violent manner, or he becomes more conditioned to act in a violent manner. Violent behavior is habit forming; and as men age, they may act with increasing impunity. Older men are also likely to see themselves more of an authoritative figure than younger men. And this could have some explanation value as to why older men are more likely to inflict more abuse. This finding indirectly supports the study by Ammar (2006) that younger women are more likely to be subjected to abuse than older women.

In terms of the marital capital factors, length of marriage or cohabitation does not seem to have a significant effect on the level of violence while for survivors who have children, the level of abuse is likely to increase as the number of children increases. One reason for this could be that having more children appears to increase the survivors’ marital utility relative to the threat point. This suggests that women see more benefits to remaining in the marriage for their children as the number of children increases. This may be because of their increased financial needs, or the relatively greater attachment of the children to their fathers.

Studies which have empirically evaluated the effects of a woman’s income on the likelihood that she leaves a violent relationship found that the fewer resources a woman has, the less power she has; therefore, she is less likely to leave the abusive relationship. Feminist resource theory within the sociological literature also predicts that as women’s income increases and their alternatives to violent marriages improve, there will be less wife abuse. However, the result of this study shows that women with income are more likely to be abused more frequently compared to those without income and thus, does not support the prediction of a negative relationship between a woman’s income and the level of violence. This could be due to data limitation in which the analysis only took account whether a woman has an income or not without looking at the various income levels. In, the data set does not contain information on the type and severity of violence which could probably be influenced differently from just examining the level of violence.

Evidence from previous studies suggests that women are more likely to leave abusive relationships if they have better economic opportunities outside the household. These results support the sociological model which views domestic violence as a result of the inequality of women. Stopping the problem requires curing the individuals rather than society. The use of services, such as domestic violence shelters, telephone help lines, police intervention, and the court system, should also be studied in determining the level of domestic violence. These variables increase a woman’s threat point; and she may use these services to signal to the perpetrator that the violence must decrease or she will leave the relationship (Farmer and Tiefenthaler 1997). Such a study would help explain why many women use services only to return home or drop charges after a short period of time.

The value of economic studies on domestic violence lies in the potential to promote social policy and ultimately reduce violence against women. Public intervention into domestic violence is now mostly based on considerations of justice and fairness. A better understanding of the determinants of domestic violence can help provide the basis for action within an additional policy framework. Adding the economic perspective increases the range of ways in which policy interventions can be articulated, measured, and evaluated (Walby 2004), while policy implications arising from developments here can be significant, for instance, in addressing spending priorities. Eradicating domestic violence from society is vital to achieve an equitable and democratic society in countries like Malaysia. The task of eliminating domestic violence, especially violence against women, is not only the role of women, but everyone. The process of achieving a violence-free society requires the understanding of the dynamics of forms of violence and the social, cultural, and legal aspects that underpin them.