Available online at www.sciencedirect.

com

The Social Science Journal 44 (2007) 698720

Domestic violence against women: A eld study in Turkey

Faruk Kocack a , Aziz Kutlar b, , Feray Erselcan b
a

Department of Sociology, Cumhuriyet University, Faculty of Science and Literature, 58140 Sivas, Turkey b Department of Economics, Cumhuriyet University, Faculty of Economic and Administrative Sciences, 58140 Sivas, Turkey

Abstract Factors affecting domestic violence against women in four Turkish cities (Adyaman, Sivas, Denizli and Krklareli) having different socioeconomic structures, are analyzed in this study. These factors consist of social, cultural, economic and psychological factors. In contrast to what we expected based on earlier literature, family income level has a positive relationship with violence. Logistic regression analysis also revealed that being a university graduate and having a personal income decreases the prevalence of violence as expected. However, working women and women with children are more prone to domestic violence. Again, there is a strong association between the neighborhood where the family lives and the incidence of violence. The extent of male dominance, as measured by the question How are decisions taken in the family is also associated with domestic violence: woman is less likely to be abused in households where decisions are taken collectively. Likewise, families where women have to get permission from the husband to carry out certain activities, have an increased incidence of suffering from husbands violence. Psychological factors, like being abused or having witnessed violence as a child, are also signicantly correlated with domestic violence. 2007 Elsevier Inc. All rights reserved.

1. Introduction Violence against women has gained worldwide interest among researchers in both developed and developing economies in recent years. Domestic violence or wife abuse, including physical abuse towards adult and adolescent women, by male intimate partners, is one of the most comCorresponding author. Tel.: +90 346 2191010x2184; fax: +90 346 235 06 00. E-mail addresses: kocacik@cumhuriyet.edu.tr (F. Kocack), akutlar2001@yahoo.com (A. Kutlar), ferayerselcan@yahoo.com.tr (F. Erselcan).
0362-3319/$ see front matter 2007 Elsevier Inc. All rights reserved. doi:10.1016/j.soscij.2007.10.016

F. Kocack et al. / The Social Science Journal 44 (2007) 698720

699

mon forms of gender-based violence. A great number of detailed studies have been conducted particularly in the West, conceptualizing violent relationships and wife abuse (Brownridge & Halli, 2002; Ellsberg, Pe a, Herrera, Liljestrand, & Winkvist, 2000; Ferraro & Johnson, 1983; n Kirkwood, 1992; Landenburger, 1989). Similar research has been carried out in non-Western societies as well (Amoakohene, 2004; Avotri & Walters, 2001; Casimiro, 2002; Haj-Yahia, 2002). It is reported that in the USA, 28% of the women are subject to domestic violence by their partners at least once in their lifetime. It is also recorded that in the developing countries more than half of the female population are beaten by their husbands or partners. In India, this rate is 45%, in Philippines 47.2% and in Kenya 52% (UNESCO, 2000). Similar percentages are given also in Nasir and Hyder (2003) review of the literature on violence against pregnant women. Twenty-eight percent of all women in the developed countries and between 18% and 67% of those in developing countries are reporting at least one incident of physical abuse. Heise, Pitanguy, and Germain (1995) review as well, highlight a rate of women between 16% and 39% having been abused physically and/or psychologically by their spouses. Another review (Heise, Ellsberg, & Gottmoeller, 2002) including 50 population-based studies in 36 countries, demonstrates that between 10% and 60% of ever-married or partnered women had experienced physical violence by their partners at least once in their lifetime. Other studies also draw attention to childhood experiences of domestic violence increasing the risk of further victimization as an adolescent or an adult (Romito, Saurel-Cubizolles, & Crisma, 2001). There are very few studies concerning domestic violence in Middle-Eastern countries where religion plays an important role in shaping society, the results of which indicate that in countries such as Egypt, Palestine, Israel and Tunisia at least one out of three women is beaten by her husband (Douki, Nacef, Belhadj, Bouasker, & Ghachem, 2003; Haj-Yahia, 2002). The issue of violence against women which gained worldwide signicance during 1970s, began to be discussed in Turkey during the mid-1980s. The rst massive reaction to this kind of violence came with the protest march entitled Say No to Battering on 17th of May 1987, which was followed by Career Womens Fest on 4th of October 1987. Thus, it was only after the 1980s that domestic violence began to be treated as a sociological matter (Tlc, 1997, p. 119), which explains the limited number of studies about Turkey. A market research rm, PIAR (1988), conducted the earliest study which concluded that 75% of the women were physically abused by their husbands. In an interview with 140 married women, who applied for counseling to the Istanbul University Medical Center, Y ksel (1990) found that 57% of these women had a personal history of abuse. Esmer u (1991), having interviewed 116 couples in Istanbul, stated that 54% of husbands admitted that they have battered their wives. The results of another nationwide PIAR (1992) survey with a sample of 1,181 women showed that a 22% reported physical abuse by their husbands. Later on, there have been other studies such as those carried out by Icli (1994), the Foundation for Womens Solidarity (1997), Ilkkaracan (1998), Yldrm (1998), Ergin and Bilgel (2001), Mayda and Akkus (2004), Sahin and Sahin (2003), B t n, S zen, and Tok (2003) and Balci uu o and Ayranc (2005). Research has highlighted that, similar to other women in other parts of the world, many women in Turkey have also been the victims of physical assault as well

700

F. Kocack et al. / The Social Science Journal 44 (2007) 698720

as other forms of violence by their husbands. In a study carried out by the Foundation for Womens Solidarity, 46.8% of interviewee reported being subjected to mild forms of violence from time to time by their husbands; 34.6% said their husbands used violence of medium severity, while 15.6% reported being frequently subjected to violence. A survey of middle and upper-income families in 1996 found that 23% of women said their husbands were violent towards them when initially questioned, but that this gure rose to 71% when they were asked questions about specic types of violence (T.C. Basbakanlk, 2001). Mor Cat, the Purple Roof Foundation, also has conducted a survey comprising 1,259 women between 1990 and 1996 where it was reported that 88.2% were living in an environment of violence and 68% were hit by their husbands (Purple Roof Foundation, 1997, p. 3435). The results of other surveys demonstrate other aspects and various types of domestic violence that women experience, the perpetuators being mainly their male partners, but may include the other male members of the family (Amnesty International, 2004). Concerning the reasons and consequences of domestic violence, research which was conducted in 1994 on behalf of Family Research Institute of the Prime Ministry gives further evidence about the situation in Turkey (T.C. Basbakanlk, 1995a). Covering ve regions, 3112 women and 1318 men in 4287 homes were surveyed, and a positive association was found between the socioeconomic level of the family and domestic violence. The percentage of women who reported that they had been subject to verbal aggression by their partners at times of conicts and disputes was 52.47%. While 34.04% of the men admitted they have beaten their wives during conicts, 29.59% of the women declared they have lived such an experience. Among the reasons of perpetrating violence were both economic problems and psychological factors. Children were also affected negatively by domestic violence. A study by KA-MER in 23 cities covering 2007 people demonstrated that 64% of the interviewee saw wife beating as a right behavior. Mens tolerance rate of wife beating is higher than womens. The percentage of women who think their behavior deserve beating was around 35.1 (T.C. Basbakanlk, 1995b). Patriarchal norms still predominate in the Turkish society, particularly inside the family; having strong impact on relations between husbands and wives. This popular local saying is a reection of norms that sanction domestic violence: After all, hes your husband; he can both love you and beat you. A eld study carried out by G lcur (1999) during 19931994, on u domestic violence and family life in Ankara, as part of a larger research and documentation project by Women for Womens Human Rights (WWHR), highlighted that the victimized women nd it difcult to access legal and institutional mechanisms for support, because they do not believe that applying to these institutions would be of any help. For example, police often encourage abused women to go back home and resolve this private problem within the family. A study of 599 women in the southeast found that 57 percent of the women questioned had experienced physical violence, but of this group only 1.2 percent had notied the police and 0.2 percent led a complaint (Ilkkaracan, 2000, p. 241). Furthermore, internalized social norms which sanction domestic violence lead the woman to believe that she somehow deserved it (G lcur, 1999). u In Turkey, many acts of violence involve traditional practices, including crimes of honor. Some women who have apparently committed suicide have in fact been killed or forced to kill themselves by family members. Although many women conform to the expectations of

F. Kocack et al. / The Social Science Journal 44 (2007) 698720

701

their family and community and are not subjected to physical violence, they remain constantly under threat. The example of other women who have been ostracized, beaten or killed serves as sufcient warning to restrict their behavior and limit their life choices (AI, 2004). The UN Development Fund for Women has found that cultural factors associated with higher levels of family and community violence include sexual double standards, rigid gender roles, lack of access to education, womens isolation and lack of support, community attitudes that tolerate physical punishment of women and children, and acceptance of violence as an appropriate means of resolving conict (UN Development Fund for Women, 2003).

2. Theoretical framework The World Health Organization (WHO) Report on Violence and Health denes violence as the intentional use of physical force or power, threatened or actual, against oneself, another person, or against a group or community, that either results in or has a high likelihood of resulting in injury, death, psychological harm, mal-development or deprivation (WHO, 2002, p. 5). It is important to note the element of intentionality being emphasized in this denition, whatever the outcome of the violent act might be. Potter (1999) also makes a reference to the intentionality by dening violence as the violation of a characters physical or emotional well-being. According to the typology of the World Report on Violence and Health (WHO, 2002), violent acts can be physical, sexual, psychological, and involving deprivation or neglect. There are certain constraints which limit our knowledge of violence. First of all, data on women who have not attended shelters or other services as victims of violence, is relatively scarce. The majority of research takes into account only the battered women who have applied for such services (Sorenson & Saftlas, 1994; Strube, 1988). The lack of cross-cultural research is another weakness. Violence is inuenced by culture and therefore depending on cultural differences among societies perceptions of violence might differ (Counts, Brown, & Campbell, 1992; Heise et al., 1999; Levinson, 1989). However, there have been very few attempts to compare these issues in different cultures. This might be due to the lack of uniformity in how data on violence are collected, which makes it difcult to make comparisons among different communities.1 There have been several approaches to explain violence. The ecological model, which tries to explain the multifaceted nature of violence was rst introduced in the late 1970s (Garbarino & Crouter, 1978). Later on, several authors have used an ecological model to conceptualize gender-based violence (Dutton, 1995; Heise, 1998). The model explores the

Societal

Community

Relationship

Individual

Fig. 1. Ecological model for understanding violence (WHO, 2002).

702

F. Kocack et al. / The Social Science Journal 44 (2007) 698720

relationship between individual and contextual factors, and highlights the multiple causes of violence and the interaction of risk factors operating within the family and broader community. These factors are conceived as concentric circles as shown in Fig. 1, the inner circle of which focuses on the characteristics of the individual; and the second circle refers to the intimate relationship where abuse takes place. The social and institutional context is represented in the third circle and the outer (fourth) circle refers to the dominant cultural views and attitudes within the society, such as laws, social and economic policies, and cultural norms (Heise, 1998; Hotaling & Sugarman, 1986; WHO, 2002).

3. Aim of the study In this study we investigate factors affecting domestic violence in Turkish society, through the use of a questionnaire in which individual and interpersonal relations as well as other environmental factors are incorporated. We have chosen the variables by taking into account the denitions of the ecological model. In this context, the study aimed at nding out which factors increased the risk for violent victimization and perpetration. The variables which fall into the rst level of the ecological model are expected to give information on the characteristics of the individual, including his/her personal history. Most of these are psychological factors; such as having a childhood history of domestic violence, violence towards children or demographic factors (like educational attainment). In the second circle we expect to be informed about the interaction with the partner the relationship itself. The variable which relates to this circle is the distribution of decision making power inside the family though it also has interrelations with the last circle when it comes to reect the extent of male dominance (patriarchy) as a cultural factor. The social factors forming the family and environmental structure are represented by variables such as the structure of the sample family, the type of neighborhood and the number of children, and they fall into the third circle of interactions with others. Poverty and unemployment also fall into this circle. So we included variables such as income, occupation and employment status, etc. Information providing insights regarding dominant cultural views and attitudes in the last circle is sought by the question are there any activities that woman should ask her husband for permission? However, just like this variable having interrelations with the second circle, the in-house democracy in the second circle is also interrelated with the last circle as explained above. Accordingly, our main assumption is that violence is the outcome of socio-cultural, economic and psychological factors. We have further assumed that domestic violence against woman would decrease with higher levels of family income and womans educational attainment, and better types of neighborhood where the family lives. We would also expect lesser violence if the woman works and also if she has a personal income, increasing her economic independence, thus leading her to feel more empowered. On the other hand, the employment status of the husband was also assumed to be related to violence. Again, we have assumed that if the woman has been exposed to violence in the past, either by being abused or having witnessed domestic violence as a child, she would be

F. Kocack et al. / The Social Science Journal 44 (2007) 698720

703

more prone to her partners violence as an adult. We also hypothesized that dominant cultural attitudes (patriarchal relations) are risk factors increasing the prevalence of wife abuse.

4. Data and method The research was conducted in the centers of Sivas, Adyaman, Denizli and Krklareli cities of Turkey. These four cities do not necessarily represent their corresponding regions, since within the same region there are cities having different characteristics and varying development levels. Therefore, instead of seeking regional representation, having totally different characteristics has become the main criteria in the selection of these cities, since we aimed at ending with a Turkish average. It would denitely be better if we had the chance to extend our survey to cover some other cities with multiple characteristics. However we had to limit ourselves with four cities where we could conduct the survey with less difculty and more condence. Denizli is a highly industrialized and developed city of the Western Anatolia where family type enterprises are very common. Sivas, located in the central Anatolia, can be characterized as one of the cities whose population is migrating out signicantly because of economic reasons. Krklareli, which is the closest city to Europe, comprises different cultural elements together. Located in the Southeast, Adyaman was selected due to its differing characteristics from neighboring cities demographically and from a developmental point of view: It is less urbanized than Van and Diyarbakr; but social life is more enriched than cities like Hakkari, Srnak and A r. g The quarters in central administrative districts of each province were categorized as central and periphery. The central quarters are the more advanced parts of the city, and characterized by their closeness to the shopping center, with high blocs of ats as dwellings. The quarters in the periphery on the other hand, are those poorer settlements in the outer city, far away from the shopping facilities with one to three-story houses (the suburbs). We used cluster sampling technique and the female samples were selected on the basis of sufcient statistical signicance level. In this method, we identied the sampling frame as all the houses or blocks in the chosen central and periphery quarters, and from this population, specic clusters (blocks, in this case) were chosen through simple random sampling. Once a cluster was chosen for inclusion in the sample, all members of the cluster are surveyed (in our example, all women within the chosen cluster, or block, were surveyed). Questionnaires were completed by 200 households in 10 quarters in Sivas, 137 households in 4 quarters in Krklareli, 306 households in 8 quarters in Adyaman and 66 households in 4 quarters in Denizli.2 On the other hand, our sample size3 is representative enough of these provinces population structure, based on data from the Census of 2000, carried out by the State Institute of Statistics (DIE, 2002). The survey was carried out by female interviewers including teachers, during the period MarchMay 2002, using face-to-face interview and questionnaire techniques.4 The research teams received special training about the subject in advance, taking into account the anticipated anxieties of respondents. Obtained data were rst evaluated statistically using SPSS 10 and logistic regression analysis was carried out by using a binary logistic model developed from discrete choice theory.5

704

F. Kocack et al. / The Social Science Journal 44 (2007) 698720

5. Statistical analysis of the sample prole Excluding the cases for which no interviews could be obtained, the number of respondents totaled to 695. Around 30 questions were asked in the questionnaire. All the people given questionnaires were women. 54.4% of our overall sample was above the age of 35, which means that most of the participants in the survey were mature. It is also understood from the answers that more than 90% of the responders were married and 93.2% have at least one child. Only 9.5% have a university degree and more than half of our sample had a very low level of education, of which 16% (which may even be higher, if we include those who did not want to answer this question) have not received any formal education. Not all but most, around 74% of the women, were housewives. Although most (81%) of the respondents had nuclear families, which consists of wife, husband and children, 16% of the sampled women were living in extended families, in other words, together with in-laws. In 65% of the families, there is only one person who works for a pay, which indicates that the total age dependency ratio6 is quite high. Questions about family prole were designed so as to provide a picture of family life, including the distribution of decision making power and the extent of male dominance inside the family. On the other hand, to complete this picture, we have also asked questions concerning the type of neighborhood, in order to get more information about the social environment in which the family lives. The statistical information about the sample prole is given in Table 1 , which is designed so as to reect any separate characteristics of the respondents in four provinces. We understand that the sample proles of Denizli and Sivas, just as the sample proles of Krklareli and Adyaman have similarities with each other concerning the age of the respondents, looking at their respective median categories. However, with respect to family incomes, though there is a certain amount of deviation, the households in Adyaman mostly belong to the lowest income category, whereas the sample prole from Denizli seem to constitute the highest income group. Households in Adyaman are also more crowded, having higher numbers of children. With respect to differences in womens education levels, Krklareli seems to be the province with the highest level of education and personal income. Especially when we look at the income groups higher than 600 YTL, around 20% of women in Krklareli and in Denizli belong to these categories while in Adyaman and Sivas only around 7% have such higher levels of income. It is also evident that Denizli has the highest number of working women in our sample.

6. Incidence and nature of violent acts We used different items to assess violence, including the act itself, whether it is physical (battering, wounding, keeping, etc.) psychological (threat or curse) or sexual (sexual abuse); as well as how it is performed (the means of violence use of physical power, a weapon or insult and curse). We also thought it would be useful to document the duration and frequency of violent acts. Almost all of the specic questions about domestic violence were answered. Women were also asked whether they were abused or have witnessed domestic violence as a child.

F. Kocack et al. / The Social Science Journal 44 (2007) 698720 Table 1 Sample prole* Province Age 1525 2125 2630 3135 36 and up Median category Education Illiterate Primary Secondary University Median category Occupation/Employment status (W) Unemployed (Housewife) Worker Gov. Ofcial Other Employment status (Man) Irregular/Unemployed Worker Gov. Ofcial Farmer Tradesman Other Marital status Single Engaged Married Widowed Sivas (N = 199) % 2.5 15.6 21.1 14.1 46.7 (3135) 10.5 53.7 30.0 5.8 Primary 89.7 .5 2.1 7.7 5.1 21.0 21.7 1.9 24.2 26.1 1.5 1.0 92.0 5.5 Adyaman (N = 293) % 2.1 6.9 12.4 21.0 57.7 36 and up 31.6 36.8 23.2 8.4 Primary 81.1 5.6 3.2 10.2 12.5 15.8 28.3 4.6 21.1 17.8 2.4 .3 92.8 4.5 51.3 19.4 22.0 3.9 2.2 1.3 Less than 200 77.4 20.1 1.1 1.4 1 person 45.4 20.4 21.6 Denizli (N = 66) % 10.6 9.1 13.6 22.7 43.9 (3135) 10.6 48.5 25.8 15.2 Primary 39.3 3.3 34.4 23.0 10.9 30.9 21.8 1.8 27.3 7.3 10.6 3.0 83.3 3.0 43.2 24.3 10.8 10.8 5.4 5.4 201400 44.6 49.2 6.2 2 people 9.5 27.0 27.0

705

Krklareli (N = 137) % .7 6.0 16.4 11.2 65.7 36 and up 3.7 35.1 44.0 17.2 Secondary 46.7 21.2 32.1 5.0 26.1 38.7 1.7 18.5 10.1 5.1 .7 87.6 6.6 36.8 23.1 20.5 9.4 .9 9.4 201400 46.5 44.9 7.9 .8 2 people 13.1 26.2 25.4

Personal income Yeni T rk Liras (YTL) u Less than 200 27.7 201400 42.0 401600 23.5 601800 5.0 8011000 1.7 1000+ Median category 201400 Working family members 1 person 2 people 3 people 4 and more Median category Family income (YTL) Less than 200 201400 401600 78.8 18.0 1.6 1.6 1 person 24.2 40.4 22.7

706 Table 1 (Continued ) 601800 8011000 1000+ Median category Dwelling Rent Owned Belongs to a relative Household members 13 people 46 people More than 7 people Median category Neighborhood Low level Middle level High level Median category Children No 12 34 56 7+ Median category Type of family Nucleus Extended Torn out

F. Kocack et al. / The Social Science Journal 44 (2007) 698720

6.1 2.5 4.0 201400 33.7 54.3 12.1 28.6 60.8 10.5 46 people 24.1 34.7 41.2 Middle level 6.0 52.3 34.7 6.5 .5 12 child. 72.7 22.7 4.5 66.3 28.1 5.5

4.8 4.8 3.0 201400 27.8 67.0 5.2 13.0 58.4 28.7 46 people 40.2 41.3 18.5 Middle level 5.2 26.8 36.2 27.5 3.8 34 child. 81.7 15.9 2.4 62.2 30.2 7.6

1.6 1.6 33.3 401600 48.5 43.9 7.6 28.8 71.2 46 people 47.0 37.9 15.2 Middle level 4.7 75.0 20.3 12 child. 92.3 7.7 66.7 31.8 1.5

11.5 4.6 19.2 401600 20.6 70.6 8.8 54.7 43.1 2.2 13 people 15.6 53.3 31.1 Middle level 12.9 64.4 18.2 3.8 .8 12 child. 87.5 10.3 2.2 69.1 29.4 1.5

Dom. violence in childhood Never Sometimes Yes, so many times

*Medians are given only when it is relevant, i.e., when the categories represent some kind of progression from low to high. Non-responses are excluded in the computation of percentages.

27.5% (191 out of 695) of the women clearly reported that they suffer from either physical or psychological violence within their own family. Twenty-ve women did not want to answer this question. Most (89%) of the perpetrators of violence are found to be the partners (husbands). More than half of the women have been experiencing violence for four or more years, whereas physical violence (battering) is most common, but psychological abuse is also quite often exercised. Women suffer from more than one type of violence as can be seen in Fig. 2. Table 2 helps us to compare violent acts in different provinces. Besides a number of similarities, it is interesting to note the variations between the cities especially with respect to specic characteristics of the violent act itself. There are differences between cities, for example, concerning the frequency and type of violence, how it is perpetrated, and as well as the perception of the reason for violence.

F. Kocack et al. / The Social Science Journal 44 (2007) 698720 Table 2 Nature of domestic violence*
Province Duration of violence Couple of months One year Two years Three years Four years or more Median category Type of violence Battering Sexual abuse Wounding Keeping Threat or curse Other Frequency of violence Everyday Few times a week Few times a month Few times a year Median category Perpetrator of violence Husband Father Son or brother Another family member Means of violence Physical power Weapon Insult or curse Other If physical power, where is it perpetrated? Out of sight Anywhere Other Womans perception of the reason for violence Failure to fulll domestic duties Failure to fulll husbands sexual will Woman wants to work Behaviors contrary to traditional perceptions of honor No reason at all Mans own social problems Jealousy Difference of view Resisting to get married against own will Mans getting married against own will Going out without permission Other Sivas (N = 53) % 11.9 4.8 9.5 11.9 61.9 Four years or more 66.0 2.0 20.0 12.0 18.0 16.0 28.0 38.0 Few times a month 96.1 3.9 52.0 46.0 2.0 57.9 31.6 10.5 23.5 7.8 11.8 9.8 19.6 15.7 5.9 5.9 Adyaman (N = 63) % 18.5 9.3 11.1 9.3 51.9 Four years or more 45.6 1.8 7.0 38.6 7.0 7.1 21.4 46.4 25.0 Few times a month 89.7 6.9 3.4 65.5 32.8 1.7 73.9 26.1 39.3 3.6 8.9 3.6 28.6 14.3 1.8 Denizli (N = 57) % 1.8 7.3 14.5 9.1 67.3 Four years or more 56.1 7.0 5.3 29.8 1.8 23.6 43.6 32.7 Few times a month 91.2 3.5 1.8 3.5 70.2 29.8 73.2 14.3 12.5 5.3 3.5 19.3 10.5 1.8 1.8 1.8 7.0 49.1

707

Krklareli (N = 18) % 5.9 5.9 88.2 Four years or more 38.9 5.6 22.2 33.3 5.9 35.3 29.4 29.4 Few times a month 83.3 5.6 5.6 5.6 33.3 5.6 55.6 5.6 100.0 5.6 33.3 5.6 38.9 11.1 5.6

*Medians are given when relevant. Non-responses are excluded in the computation of percentages.

708

F. Kocack et al. / The Social Science Journal 44 (2007) 698720

Fig. 2. Different types of abuse faced by the sampled women. Battering: 80, Curse: 62, Sexual abuse: 5. Battering + Curse: 43. Battering + Sexual abuse: 1.

The question about the victimized womans perception of the reason for domestic violence was designed so as to get an idea about which activities she was prohibited from carrying out, or which duties she thinks she failed to do. Forty-three women out of 191 (22.5%), believed that their husbands exercise violence for no reason at all (the percentages vary from 19.3 in Denizli to 38.9 in Krklareli), while 20% believed it is because she failed to do her domestic duties. According to a total of 25 women (13%) husbands own social problems lead him to get nervous easily at times of conict. For 7.3% of abused women, the main reason for the outcome of conict and domestic violence is the woman wanting to work outside home for pay. As can be followed from Table 3, tolerance of violence was quite high. The percentage of abused women who said they would show no reaction when they face with violence is 46%. Looking at the answers given to the question about when the woman should ask her husband for permission, most of the women thought when she goes out (usually for a visit). The activity in the second row the woman has to be allowed by her husband is shopping and this activity is reported by 16.2% of women suffering from husbands violence. Twenty-one percent have not identied any activity to take permission from the husband; meaning that they do not think it is necessary to take permission for anything. The decision making power and authority belongs mainly to men (68.6%) among the abused womens families, whereas women who reported each one in the family has the right to say in making the decisions constitute only a 22%. Among the psychological factors reecting the individuals characteristics that make her more prone to violence is witnessing or experiencing domestic violence (having been abused by parents) during childhood. Looking at our data, we recognize that women who did not have such an experience account for 64% within our original interviewed group, whereas this rate falls to 54.3% among the women who report partner violence. Similarly, those who declared that they had sometimes been subject to violence when they were child account for 29% in the overall sample, but increase their rate to 36% in the abused group.

F. Kocack et al. / The Social Science Journal 44 (2007) 698720 Table 3 Partners and family relations All women % (N = 695) When is it necessary to get permission from the husband? No answer 30.6 (213) Visiting parents 1.9 (13) Going out 36.0 (250) Going shopping 9.5 (66) What the woman wears 1.3 (9) Other 20.7 (144) Reaction if faced with violence No answer No reaction Verbal react. (insults) Would hit Would throw sth. Would shout Other 30.9 (215) 28.3 (197) 8.8 (61) 2.0 (14) 2.2 (15) 17.6 (122) 10.2 (71)

709

Women abused % (N = 191) 20.9 (40) 2.6 (5) 41.9 (80) 16.2 (31) 1.0 (2) 17.3 (33) 3.1 (6) 46.1 (88) 13.1 (25) 3.1 (6) 1.6 (3) 25.1 (48) 7.9 (15) 1.0 (2) 68.6 (131) 4.2 (8) 22.0 (42) 4.2 (8) 3.7 (7) 44.0 (84) 37.7 (72) 7.3 (14) 3.1 (6) 1.6 (3) 2.6 (5)

Distribution of decision making power in the family No answer 5.8 (40) Man makes the decision 31.1 (216) Woman makes the decision 3.6 (25) Both share equally in making the decision 56.4 (392) Other 3.2 (22) Do you ever behave violently against your child? No answer No Yes, when he/she is naughty. Yes, when he/she gives way to a damage. Yes, if he/she gets bad friends Yes, if he/she gets bad habits Other 6.0 (42) 53.2 (370) 25.8 (179) 4.0 (28) 5.8 (40) 1.7 (12) 3.5 (24)

Whether the woman ever behaves violently towards her own child, is also one of those psychological factors. The percentage of those who answered as they do not is 44% and of those who declared they would batter the child when he/she deserves is 52% within the group of abused women. These percentages are 53.2% and 40.8% for the overall sample. This tendency can be interpreted as woman who is a victim of partners violence herself, reects violence to her child on a legitimate ground. This was expected and we should perceive it as a negative consequence of violence on womans psychological health, as well as on her children; which may lead to their further victimization as adults. 7. Logistic regression results To assess the effects of social, economic and psychological factors (mentioned in Table 1) taken together, on the incidence of domestic violence, we undertook logistic analyses. Data analysis was carried out using SPSS 10. The prevalence of whether or not the woman had ever experienced domestic violence by an intimate partner and/or by a male family member at any

710

F. Kocack et al. / The Social Science Journal 44 (2007) 698720

time is the dependent variable.7 The approach was to t a multiple logistic regression model with a large pool of candidate explanatory variables. First, the model with all explanatory variables was tted and then certain explanatory variables were removed, which were distorting the model. Finally, we maintained the following variables in our analysis: age, education, marital status, number of children, employment status and personal income of the woman; employment status of the husband and the household income; type of family, ownership of dwelling and the type of neighborhood as well. We have also included variables connected to male dominance inside the family as cultural factors, and whether the woman has ever been abused as a child, and violence against her own child as psychological factors. Table 4 reects the results of the binary logistic regression analysis, based on answers given by 670 of the interviewed women. A total of 46 variables including two dummy variables are used as independent variables. The abbreviation rc in the table stands for the reference category. Looking at the odds ratios, we can observe that young womens prevalence of being abused is more than the other age groups. Likewise, widows are less prone to domestic violence when compared to unmarried women. The odds ratios are lower than 1, indicating a fall in the incidence of abuse as compared to the reference category in each question. On the other hand, university graduates are facing violence at a lesser incidence. As compared to women who did not get any formal education (rc for this variable), university graduates are less likely to experience violence inside family. The results show that if a woman works outside for pay (signicant only for the category of working women other than workers or government ofcials), this would increase her exposure to domestic violence. Such working women are found to face a signicantly higher risk of marital violence than (almost three times) women who do not work. On the other hand, womans personal earnings are also associated with her being abused. As compared to women reporting a personal income of less than 200 YTL, those who get slightly higher, between 400 and 600 YTL are less likely to be abused, since the odds ratio is less than 1. The signicance level is around 10 percent and relevant only for one category. In contrast to our assumptions, we found that family income is positively associated with partners violence. As the income category rises, the level of positive signicance also rises; in other words the signicance level falls below to 1% from levels around 10%. The odds ratios are above 1 for the higher income categories of 400600 YTL, and over 1000 YTL (which are also signicant at 1% condence level), indicating that the incidence of violence is much greater as compared to poor families; the category of less than 200 YTL being the reference category. Women who have reported their husbands are tradesmen were almost twice as likely to have reported major aggression, than women who reported that their husbands are unemployed. We have also found odds ratios being lower for socially better neighborhoods, reecting a lower incidence of partners violence as the category of neighborhood rises. Partners violence is also related with the number of children in the family. As the number of children in the family increases, the odds ratios and also violence against woman increases. In families with no child, the incidence of violence is lower. The odds ratios for the women who were ever abused in childhood increase over 1 as compared to women who did not have a personal history of abuse as a child. If the woman is abused as a child, she is more likely to be abused as a wife.

Table 4 Logistic regression analysis of domestic violence against women

Variables in the equation B Sig. Odds ratios EXP(B) 95.0% C.I. for EXP(B) Lower Age 1520 (rc) 2025 2630 3135 >35 Illiterate (rc) Primary Secondary University No (HW) (rc) Worker Gov. Ofcial Other Unempld. (rc) Worker Gov. Ofcial Farmer Tradesman Other Single (rc) Engaged Married Widowed <200 (rc) 201400 401600 601800 8011000 >1000 <200 (rc) 201400 1.366 1.789 1.227 1.211 .243 .229 1.195 .097 .363 1.157 .049 .426 .428 .775 .228 1.330 .409 1.593 .004 .669 .211 .932 1.347 .584 .053* .007*** .068* .062* .428 .541 .076* .813 .455 .046* .897 .260 .592 .025** .550 .387 .534 .064* .992 .099* .785 .336 .128 .093* 1.0 .255 .167 .293 .298 1.0 1.275 .795 .303 1.0 1.101 1.437 3.180 1.0 .952 1.531 1.534 2.170 1.256 1.0 .265 .664 .203 1.0 .996 .512 .810 2.539 .260 1.0 1.793 .064 .045 .079 .084 .700 .382 .081 .494 .555 1.020 .453 .729 .321 1.101 .595 .013 .183 .038 .521 .232 .178 .380 .046 .908 Upper 1.018 .618 1.094 1.060 2.322 1.657 1.135 2.456 3.719 9.914 2.002 3.214 7.326 4.277 2.652 5.392 2.407 1.100 1.907 1.133 3.694 16.963 1.475 3.540

F. Kocack et al. / The Social Science Journal 44 (2007) 698720

Education

Working out for a pay?

Occupation (M)

Marital status

Personal income

Family income (YTL)

711

712

401600 601800 8011000 >1000 Dwelling Rent (rc) Owned Relatives 13 (rc) 46 79 Low (rc) Middle High No (rc) 12 34 56 78 Nucleus (rc) Extended Torn out No (rc) Sometimes Frequently No (rc) Yes Man (rc) Woman Equally shared No (rc) Yes d.f. 46 Sig. .000***

1.123 .018 1.173 2.480 .367 .337 .207 .164 1.012 .744 1.290 .887 .748 1.881 .082 .414 .522 1.099 .788 .381 2.218 .380

.005*** .980 .135 .000*** .151 .436 .420 .481 .000*** .019** .011** .096* .216 .041** .788 .604 .034** .019** .003*** .495 .000*** .103 2 Log likelihood 565.375

3.073 1.019 3.233 11.939 1.0 .693 .714 1.0 .813 .849 1.0 .363 .475 1.0 3.634 2.427 2.113 6.559 1,0 1.086 1.512 1.0 1.686 3.002 1.0 2.199 1.0 .683 .109 1.0 1.463 Cox & Snell R Square .419

1.392 .248 .693 3.538 .420 .306 .491 .538 .208 .255 1.351 .855 .646 1.084 .597 .317 1.040 1.198 1.297 .228 .068 .926

6.788 4.184 15.076 40.287 1.143 1.665 1.345 1.339 .634 .883 9.773 6.892 6.913 39.698 1.973 7.220 2.733 7.521 3.730 2.044 .174 2.310 Nagelkerke R Square .558

HH members

F. Kocack et al. / The Social Science Journal 44 (2007) 698720

Neighborhood

Children

Family type

Dom. violence in childhood

Any activity to get mans permission Decision making power

Any violent behaviour against own child?

Chi-square Model: rc: reference category.

363.442

p 0.10. p 0.05. p 0.01.

F. Kocack et al. / The Social Science Journal 44 (2007) 698720

713

The extent of male dominance, as measured by the question How are decisions taken distribution of decision making power inside the family, is also associated with domestic violence. Data conrms a negative association between violence and in-house democracy: woman is less likely to be abused in households where decisions are taken collectively than in households where man has the absolute power in making the decisions. Another question which would give us information about patriarchy inside the family, was by asking which activities the woman is prohibited from carrying out: whether there are any activities to get permission from the husband. When compared to women who reported no activity to get permission, those who reported any of the alternative activities have an increased incidence of suffering from husbands violence. With respect to factors other than these, data do not support any signicant relationship with domestic violence against women.

8. Conclusion and discussion In this paper, we investigated the relationship between certain economic, social and psychological factors and domestic violence against women, based on a sample survey of four Turkish cities with varying socioeconomic indicators. The survey provides useful information regarding economic, social and psychological conditions under which domestic violence occurs in the selected urban centers of Turkey. The general characteristics of violence described in the survey data are consistent both with international and previous Turkish research, suggesting that experiences are common to women all around the world. However, there are some interesting results as well, which might be a reection of the countrys characteristics and different culture. Both the Mediterranean culture and the Islamic tradition have impacts on the culture (M ft ler-Bac, 1999). On the other hand, u u Turkey is also a rapidly transforming society. The results of our study, we believe, reect certain situations that are very specic and particular to this transformation process. More than 96% of surveyed women directly answered the questions about violence and it is obvious from the answers that 27.5% of these women have experienced such violence. Some women might have failed to disclose their experiences of violence, and actual prevalence of domestic violence maybe higher than the gures presented in this study. Since women tend to understand violence as something that requires you to go to hospital (AI, 2003); or as mentioned above, being abused by the husband is usually treated as a private problem in the Turkish society. We can conclude that almost all the factors that fall into the inner circle of ecological factors used in our model (the age, education, domestic violence in childhood) are associated with violent behavior by the male partner. The prevalence of domestic violence is found to decrease as the age of the respondents increases over 20. This is because over time a marital relationship can achieve a degree of stability and so reduce the husbands tendency to resort to violence. After long years of marriage, a woman might also learn to cope with domestic violence. This outcome is in line with ndings from other research which conclude younger women reporting more violence (Romito, Turan, & De Marchi, 2005). Our results concerning womans educational attainment coincide with other authors ndings both from the world (McCall & Shields, 1986;

714

F. Kocack et al. / The Social Science Journal 44 (2007) 698720

Straus, Gelles, & Steinmetz, 1980) and from Turkey (Balci & Ayranc, 2005; Ilkkaracan, 1998; Mayda and Akkus, 2004; Sahin & Sahin, 2003; Balci & Ayranc, 2005). The women who had a personal history of violence in childhood, when they grow up and get married, would perceive violence as a normal behavior in punishment and conict situations (Romito, Crisma, & Saurel-Cubizolles, 2003; Romito et al., 2005). This would increase those womens tolerance towards violence which may also impact on their self-esteem, reducing their ability to leave potentially violent relationships. To our knowledge, there are not many studies about Turkish women who grew up in violent homes, learning behaviors in childhood which place them at risk. Second, at the level of an intimate relationship, we can follow that if there is in-house democracy when taking decisions, a woman is less likely to be assaulted by her partner. Our results coincide with Sahin and Sahin (2003) who have also found that abused women were not involved in family decisions. In addition, the women who reported that it is necessary to ask the husband for permission to carry out certain activities have an increased incidence of being abused. Among these activities shopping is particularly highlighted, indicating the prevalence of male control over economic resources. These ndings are useful tools in understanding dominant cultural views and attitudes prevailing in society. In this context we nd it useful to take note that failure to fulll domestic duties is perceived as one of the important causes of being subjected to violence. Interestingly, a high percentage of women could not perceive any reason at all for being beaten; and together with those who thought the mans own problems caused the violent act. We can conclude that majority did not see any connection of violence with their own behaviors or failures, thus they do not blame themselves for being abused. However, they seem not to blame their husbands either, stating there are other problems of his own, which may well be out of his control, such as psychological distress connected with social and economic problems. The signicant negative association found between womans state of working and domestic violence also suggest that gender roles are rigidly dened and enforced. In a society where patriarchal relations are still dominant, men usually would not allow their wives to work outside. The simple explanation for that comes from the traditional roles of men as breadwinners and women as homemakers. Taking the traditional view, a working woman, who neglects her in-house duties assigned to her under her traditional role as the homemaker, would lead to conicts in a patriarchal family. Under the Turkish legal system, a decade ago (until the 1990s), the woman could not work outside without the consent of her husband. Gainful work outside would actually empower woman by increasing her economic value, her bargaining power and decision making authority inside the family (Sen, 1989; Horton, 1996). When the woman wants to work outside however, the man perceives this as an intervention into his territory. Due to this, there is room for conict inside the family, which might be followed by a violent act by man. A man behaves aggressively toward a woman when he can no longer dominate her. Violence is most probably used to resolve crises of male identity which are provoked by challenges to patriarchal control (Moore, 1994). Focusing on our nding about women who have personal incomes being less likely to face domestic violence, we see that it conrms research in other settings which has also found domestic violence to be less likely for the women who were economically independent (Panda & Agarwal, 2005; Rao, 1997; Sahin & Sahin, 2003; Schuler, Hashemi, Riley, & Akhter, 1996).

F. Kocack et al. / The Social Science Journal 44 (2007) 698720

715

On the other hand, we have also concluded that if a woman works outside for pay, she is more likely to be abused by her husband. Although these two outcomes seem to contradict, there is more to discuss in this matter. One should keep in mind that being paid does not necessarily mean retaining signicant control over the income. Quite a number of studies from developing countries (for example in Agarwal, 1994; Berik, 1987; Hafeez, 1989; Standing, 1991; Zohir & Paul-Majumder, 1996) provide evidence of women handing over a large proportion (or even all) of their pay to other family members. This undermine the simple assumption that earning a wage will automatically empower women, increase their ability to control their earnings or their role in household decision making (Elson, 1999). Our results might be interpreted as follows: a woman, although she works, probably does not have the ability to control her earnings, so she would not feel empowered and her role in household decision making actually does not increase. As far as the other group of women is concerned, (i.e., who have personal incomes and are less likely to face partners violence), we understand that they do have control over their money. Actually, the mentioned personal income might be other than the wage earnings, such as personal assets or rental earnings from an immovable property (land or a house) over which she has more control. In such a case, a womans overall sense of empowerment would increase and her tolerance to violence would be reduced. However, employment alone does not give enough shelter from violence. This outcome is also consistent with the ndings of a recent study in India, by Panda and Agarwal (2005), in which women owning immovable property are found to face a signicantly lower risk of marital violence than women without any property. Our results concerning the family income need a closer look. As in the households whose incomes range between 200 and 600 YTL, and above 1000 YTL, violence inside the family increases as compared to the poorest families. The rst group of households get a very low level of income (around the formal prevailing minimum wage), which is quite understandable that violence takes place, since there would be room for conict concerning household expenditures. On the other hand, women whose family incomes are above 1000 YTL are also found more likely to be assaulted by their partners. One cannot easily make an assessment of the reasons behind this. We can discuss the matter by developing two separate arguments based on explanations about the level of income. First, it might be the case that the woman also works and contributes to the family income. Recalling our nding that the likelihood of being abused is much higher if the woman works, we can conclude that these two ndings are self-explanatory. As a second argument, we can again refer to one of our other ndings related to husbands occupation. Actually this income group mostly consists of those families (husbands) who work as artisans and tradesmen. We have found that they are more likely to abuse their wives than any other occupation group. This group can be characterized as earning a higher level of monthly income compared to a worker or an ordinary government ofcial, and as having a rather lower level of education. One more explanation for Why tradesmen and not another group? could be found going back to the dates we conducted this research: An economic crisis prevailed in the last years (2000 and 2001) in Turkey. Tradesmen who have small or medium sized enterprises were affected most by the downturn and faced risks of closing down their enterprises. This might well be a source of conict in their homes, reecting the distress in their work. The most interesting conclusion here is that in families under extremely poor conditions (the lowest range of income) violence is lower. This might be due to the fact that expectations

716

F. Kocack et al. / The Social Science Journal 44 (2007) 698720

are also at minimum. The family is most probably dependent on the natal family or relatives. Turkish families usually support their children, especially if they are in need, even if they are married. This attitude is quite typical for a Turkish traditional family. However this point needs to be further discussed and we need more studies to deepen our understanding on whether it is due to social and cultural differences. Domestic violence is also found related to the number of children in the family. Having many children increases the risk of battering, by increasing stress or by providing more reasons for disagreement. This outcome has been the conrmation of the results by many other Turkish studies (Balci & Ayranc, 2005; Sahin & Sahin, 2003). Concerning our conclusions about violence being less in the districts where families from middle and upper social groups live, but more in the families with higher levels of income, these two outcomes seem to contradict with each other. However, choosing a district to live might depend on factors other than income. Education level, social and cultural background might be more effective in this respect. In order to understand the reasons, we have to look further into the internal rural to urban migration process which dates back to 1950s.8 For example, in this process, a chain of migration from a particular locale in rural Turkey to a specic urban area is quite common (Hemmasi & Prorok, 2002). However, it is beyond the scope of this study to make further comments on the social and cultural transformation in the Turkish society at this point. We believe that the strength of this study is that it is one of the very few on this issue from this part of the world with some interesting results that need further attention. This study also differs from many others conducted in Turkey with respect to the method and approach it employs, as well as its sample covering women from different regions of Turkey.

Notes 1. There are many difculties in collecting data on violence, for example such data generally come from different organizations (i.e., the police stations and hospitals) that operate independently of one another. Even with the surveys there is difculty, because they are limited by the extent to which a person recalls events, by the manner in which questions are asked and it also depends on when, where and how well the interview is conducted. In addition, it is also difcult to develop measures relevant and specic to subpopulation groups and different cultural contexts (WHO, 2002). 2. To estimate the necessary sample size with a certain degree of precision for a simple random sample, we used the following formula,n = p(1p) here n: the necessary (S.E.)2 sample size, p: the estimated proportion of victimization of domestic violence, and S.E.: the standard error of the sample proportion. Applying this formula, taking a proportion of victimization of at least 20% (based on prior studies) and assuming a standard error of 5%, would yield the following: n = .20 (.80)/.052 or n = 64+ respondents. 3. The sample size is calculated based on data for women eligible for our survey (between the ages 1549). 4. Though we have not intended to collect any qualitative data in particular, and our main aim was to collect answers to our questions as previously prepared, at some point during the

F. Kocack et al. / The Social Science Journal 44 (2007) 698720

717

5.

6.

7.

8.

interview, women had to give some details about their particular situation and these were taken as special notes by the interviewers. These unintentionally collected qualitative data, together with explanations given under answering options of other than above in certain questions, were also used later, to illustrate the statistical results of our study. In the logistic model, which is known as limited dependant variable model in econometrics or discrete choice model in social sciences, the dependent variable is limited by two different values, 1 or 0. The model was rst used in biology in 1930s and then in economics in 1950s (Cramer, 2001). There has been signicant theoretical work on the model, carried out by a number of authors, such as Hosmer and Lemeshow (2000), Pampel (2000) and Menard (2002). It has been used under a number of different applications such as binary, ordered, multinomial and censored models (Borooah, 2001). The studies by Maddala (1983), Manski and Fadden (1990), and Cramer (2001) are among those applications including the discrete choice theory. The logistic regression analysis, which is a version of this model, has been employed recently in other studies on violence by various researchers, such as Romito et al. (2003) and Jewkes et al. (2002). Age dependency ratio is calculated using this Formula: ADR = (P014 + P65+ )/ (P1564 ) 100 where P stands for population and indices for age intervals. P014 shows population between zero to fourteen years of age; P65+, 65 and above; and P1564 stands for population between fteen to sixty-four years of age. The odds ratio is dened as the rate of womans victimisation of violence over nonvictimisation. The victimisation is taken as Pi = 1, and non-victimisation as (1 Pi ) = 0. Accordingly, the log odds ratio is, ln(Pi /(1 Pi )), which is the dependent variable. For further information about migration in Turkey, see Hemmasi and Prorok (2002); Erman (2001); Gedik (1996); Ilcan (1994); and Kiray (1990).

References
Agarwal, B. (1994). A eld of ones own - gender and land rights in South Asia Cambridge. Cambridge University Press. AI-Amnesty International. (2003, July). Interview with a womens activist in Istanbul, (as cited in Turkey: Women confronting family violence). Retrieved November 2005, from http://www.amnesty.org/ library/index/engeur440132004. AI-Amnesty International. (2004, June). Turkey: Women confronting family violence. AI Index: EUR 44/013/2004). Retrieved November 2005, from http://www.amnesty.org/library/index/engeur440132004. Amoakohene, M. I. (2004). Violence against women in Ghana: A look at womens perceptions and review of policy and social responses. Social Science and Medicine, 59, 23732385. Avotri, J. Y., & Walters, V. (2001). We women worry a lot about our husbands: Ghanaian women talking about their health and their relationships with men. Journal of Gender Studies, 10, 197212. Balci, Y. G., & Ayranc, U. (2005). Physical violence against women: Evaluation of women assaulted by spouses. Journal of Clinical Forensic Medicine, 12, 258263. Berik, G. (1987). Women carpet weavers in rural Turkey: Patterns of employment earnings and status. Geneva: ILO. Borooah, V. K. (2001). Logit and Probit. Sage University paper 138. Thousand Oaks: SAGE Publications. Brownridge, D. A., & Halli, S. S. (2002). Understanding male partner violence against cohabiting and married women: An empirical investigation with a synthesized model. Journal of Family Violence, 17, 341361.

718

F. Kocack et al. / The Social Science Journal 44 (2007) 698720

B t n, C., S zen, S., & Tok, M. (2003). Evaluation of violence against women resulting in death. In Conference uu o Proceedings. Istanbul: Institute of Forensic Medicine. Casimiro, C. (2002). Social perceptions of marital violence. Analise Social, 37, 603630. Counts, D. A., Brown, J., & Campbell, J. (1992). Sanctions and sanctuary: Cultural perspectives on the beating of wives. Boulder, CO: Westview Press. Cramer, J. S. (2001). An introduction to logit model for economist. London: Timberlake Cons. Ltd. DIE (SIS State Institute of Statistics). (2002). T rkiye Istatististik Yll , 2002. (Statistical Yearbook of Turkey, u g 2002). Ankara: SIS, Turkish Republic, Prime Ministry. Douki, S., Nacef, F., Belhadj, A., Bouasker, A., & Ghachem, R. (2003). Violence against women in Arab and Islamic countries. Arch Womens Mental Health, 6, 165171. Dutton, D. B. (1995). The domestic assault of women. Vancouver: University of British Columbia Press. Ellsberg, M. C., Pe a, R., Herrera, A., Liljestrand, J., & Winkvist, A. (2000). Candies in hell: womens experiences n of violence in Nicaragua. Social Science and Medicine, 51, 15951610. Elson, D. (1999). Labour markets as gendered institutions: Equality, efciency & empowerment issues. World Development, 27, 611627. Ergin, N., & Bilgel, N. (2001). A survey to establish the situation reviolence against women in central Bursa- Uludag University Midwifery Department student survey in Bursa. Journal of Nursing, Turkish Nursing Association, 51, 12. Erman, T. (2001). Rural migrants and patriarchy in Turkish cities. International Journal of Urban and Regional Research, 25(1), 118133. Esmer, Y. (1991). Alglama ve anlatmda esler aras farkllklar (Perception and communication differences among couples). In N. Arat (Ed.), Kadn ve cinsellik (Women and sexuality). Istanbul: Say Yaynlar. Ferraro, K. J., & Johnson, J. M. (1983). How women experience battering: The process of victimization. Social Problems, 30, 326339. Foundation for Womens Solidarity (Kadn Dayansma Vakf). (1997). Orta ve ust sosyo-ekonomik d zeydeki ailel u erde kadna y nelik siddet (Violence against women in families of middle and upper socioeconomic status). o Ankara: Kadn Dayansma Vakf. Garbarino, J., & Crouter, A. (1978). Dening the community context for parentchild relations: the correlates of child maltreatment. Child Development, 4, 604616. Gedik, A. (1996). The effects of in and out migration on urban growth in Turkey (196585) and a comparison with the developed countries. Papers in Regional Science, 71(4), 405419. G lcur, L. (1999). A study on domestic violence and sexual abuse in Ankara, Turkey. Women for Womens Human u Rights Reports No.4. Istanbul, Turkey. Hafeez, S. (1989). Women in industry Islamabad: Womens Division, Government of Pakistan. Haj-Yahia, M. M. (2002). Beliefs of Jordanian women about wife beating. Psychology of Women Quarterly, 26, 282291. Heise, L. (1998). Violence against women: An integrated, ecological framework. Violence Against Women, 4, 262290. Heise, L., Ellsberg, M., & Gottemoeller, M. (1999). Ending violence against women. Population Reports, Series L No. 11. Baltimore, MD: Johns Hopkins University. Heise, L., Ellsberg, M., & Gottmoeller, M. (2002). A global overview of gender-based violence. International Journal of Gynecology and Obstetrics, 78, 514. Heise, L., Pitanguy, J., & Germain, A. (1995). Violence against women: The hidden health burden. World Bank Discussion Paper, 255. Hemmasi, M., & Prorok, C. V. (2002). Womens migration and quality of life in Turkey. Geoforum, 33, 399411. Horton, S. (1996). Women and industrialisation in Asia: An overview. In Susan Horton (Ed.), Women and industrialisation in Asia. London: Routledge. Hosmer, D. W., & Lemeshow, S. (2000). Applied logistic regression. New York: John Wiley and Sons Inc. Hotaling, G. T., & Sugarman, D. B. (1986). An analysis of risk markers in husband to wife violence: The current state of knowledge. Violence and Victims, 1, 101125. Icli, T. (1994). Kriminoloji (Criminology). Ankara: Bizim B ro Basmevi. u

F. Kocack et al. / The Social Science Journal 44 (2007) 698720

719

Ilcan, S. M. (1994). Peasant struggles and social change: migration, households and gender in a rural Turkish society. International Migration Review, 28(3), 554577. Ilkkaracan, P. (1998). Exploring the context of womens sexuality in eastern Turkey. Reproductive Health Matters, 6, 6674. Ilkkaracan, P. (Ed.). (2000). Women and sexuality in Muslim societies. Istanbul: Women for Womens Human Rights. Jewkes, R., Levin, J., & Penn-Kekana, L. (2002). Risk factors for domestic violence: ndings from a South African cross-sectional study. Social Science & Medicine, 55(9), 16031617. Kiray, M. B. (1990). The family of the migrant worker. In F. Ozbay (Ed.), Women, Family and Social Change in Turkey (pp. 7087). UNESCO Supported Series on Womens Studies in Asia and the Pacic. Kirkwood, C. (1992). Leaving abusive relationships. London: Sage. Landenburger, K. (1989). A process of entrapment in and recovery from an abusive relationship. Issues in Mental Health Nursing, 10, 209227. Levinson, D. (1989). Violence in cross cultural perspective. Newbury Park, CA: Sage. Maddala, G. S. (1983). Limiteddependent and qualitative variables in econometrics. Cambridge: Cambridge University Press. Manski, C. F., & Fadden, D. (1990). Structural analysis of discrete data with econometric application. Massachusetts: MIT Press. Mayda, A. S., & Akkus, D. (2004). Domestic violence against 116 Turkish housewives: A eld study. Women & Health, 40(3), 95108. McCall, G. J., & Shields, N. M. (1986). Social and structural factors in family violence. In M. Lystad (Ed.), Violence in the home: interdisciplinary perspectives. New York: Brunner/Mazel. Menard, S. (2002). Applied logistic regression analysis. Thousand Oaks, CA: Sage. Moore. (1994). A passion for difference: Essays in anthropology and gender. London: Polity Press. M ft ler-Bac, M. (1999). Turkish womens predicament. Womens Studies International Forum, 22(3), 303315. u u Nasir, K., & Hyder, A. A. (2003). Violence against pregnant women in developing countries: Review of evidence. European Journal of Public Health, 13, 105107. Pampel, Fred C. (2000). Logistic regression: A primer. Thousand Oaks, CA: Sage. Panda, P., & Agarwal, B. (2005). Marital violence, human development and womens property status in India. World Development, 5, 823850. PIAR-Gallup. (1988). Kadnlarn sorunlar ve beklentileri (Problems and expectations of women). Research report for the Directorate General on the Status and Problems of Women. Ankara: Turkish Prime Ministry. PIAR-Gallup. (1992). T rk kadnnn g ndemi aratrmas (Research on status of Turkish women). Research report u u s for the Directorate General on the Status and Problems of Women. Ankara: Turkish Prime Ministry. Potter, W. J. (1999). On media violence. Thousand Oaks, CA: Sage. Purple Roof Foundation. (1997). My future in my hands, (Gelece im Elimde). g Istanbul: Mor Cat Kadn S na g g Vakf (Purple Roof Foundation). Rao, V. (1997). Wife-beating in rural south India: a qualitative and econometric analysis. Social Science and Medicine, 44(8), 11691180. Romito, P., Saurel-Cubizolles, M. J., & Crisma, M. (2001). The relationship between parents violence against daughters and violence by other perpetrators: an Italian study. Violence Against Women, 7, 14291463. Romito, P., Crisma, M., & Saurel-Cubizolles, M. J. (2003). Adult outcomes in women who experienced parental violence during childhood. Child Abuse and Neglect, 27, 11271144. Romito, P., Turan, J. M., & De Marchi, M. (2005). The impact of current and past interpersonal violence on womens mental health. Social Science & Medicine, 60(8), 17171727. Sahin, H. A., & Sahin, H. G. (2003). An unaddressed issue: domestic violence and unplanned pregnancies among pregnant women in Turkey. European Journal of Contraception & Reproductive Health Care, 8(2), 9398. Schuler, S. R., Hashemi, S. M., Riley, A. P., & Akhter, S. (1996). Credit programmes, patriarchy and mens violence against women in rural Bangladesh. Social Science and Medicine, 43, 17291742. Sen, A. K. (1989). Womens survival as a development problem. Bulletin of American Academy of Arts and Sciences, 18(2) Sorenson, S. B., & Saftlas, A. F. (1994). Violence and womens health. The role of epidemiology. Annals of Epidemiology, 4, 140145.

720

F. Kocack et al. / The Social Science Journal 44 (2007) 698720

Standing, H. (1991). Dependency and autonomy: Womens employment and the family in Calcutta. London: Routledge. Straus, M. A., Gelles, R. J., & Steinmetz, S. K. (1980). Behind closed doors: violence in the American family. New York: Anchor Press. Strube, M. (1988). The decision to leave an abusive relationship: Empirical evidence and theoretical issues. Psychological Bulletin, 104, 236250. T.C. Basbakanlk, Aile Arastrma Kurumu. (1995a). (Family Research Institute of the Prime Ministry) Aile i i c siddetin sebep ve sonu lar (Causes and consequences of Domestic Violence) (Publication No. 86). Ankara: c T.R. Prime Ministry. T.C. Basbakanlk, Kadnn Stat s ve Sorunlar Genel M d rl gu. (1995b). (General Directorate of Womens uu u u u Status and Problems of the Prime Ministry) Eylem Platformu ve Pekin Deklarasyonu D rd nc D nya Kadn o u u u Konferans, 415 Eyl l 1995 (Forum for Action and Beijing Declaration, The Fourth World Conference on u Women, 415 Sept. 1995). Ankara: T.R. Prime Ministry. T.C. Basbakanlk, Kadnn Stat s ve Sorunlar Genel M d rl gu. (2001). (General Directorate of Womens Status uu u u u and Problems of the Prime Ministry) T rkiyede Kadn (Women in Turkey). Ankara: T.R. Prime Ministry. u c Tlc R. (1997). Aile I i Siddet: Bir Sosyolojik Yaklam (Domestic Violence: A Sociological Approach) (Publication s No. 285). Ankara: TODA IE. UN Development Fund for Women. (2003). Not a Minute More: Ending Violence Against Women, New York. Retrieved March 2005, from http://www.unifem.org/index.php?f page pid=207. United Nations Educational, Scientic, and Cultural Organization. (2000). Domestic violence against women. Florence Innocenti Research Centre. WHO. (2002). World report on violence and health. Geneva: World Health Organization. Yldrm, A. (1998). Sradan siddet. (Ordinary Violence). Istanbul: Boyat Kitaplar. Y ksel, S. (1990). Es daya ve daya a kars dayansma kampanyas (Spouse abuse and the campaign against spouse u g g abuse). In S. Tekeli (Ed.), Kadn bak a sndan 1980ler T rkiyesinde kadn (From womens perspectives: s c u Women in the 1980s in Turkey). Istanbul: im Yaynlar. Iletis Zohir, S. C., & Paul-Majumder, P. (1996). Garment workers in Bangladesh: Economic, social and health conditions, Research Monograph No. 18. Dhaka: Bangladesh Institute of Development Studies.