Intimate partners’ violence in Southern Ethiopia: Examining the prevalence and risk factors in the Sidama Zone

Nigatu Regassa
1
Hawassa University, Institute of Environment, Gender and Development; E-mail negyon @yahoo.com: P.O.Box 679,
Hawassa, SNNPR, Ethiopia; Phone 251-0911808662
The high level of intimate partner violence (IPV) against women in many population groups in Ethiopia and the risk factors associated with the practice is not well understood among scholars and decision makers. This study examined the prevalence and risk factors associated with intimate partner violence in Sidama, a populous zone in Southern Ethiopia. A combination of simple random and multistage sampling techniques were used to select 1094 households, comprising women and men participants, for the field study. Quantitative and qualitative data were obtained using structured questionnaire and focus group discussions. Household, women and husband characteristics were used as explanatory variables while intimate partner violence served as the dependent variable. The study revealed that the prevalence of intimate partners’ violence is high in the study population (ranging from 14.7 to 61.2%) with physical violence (beating,
causing physical damage and slapping) accounting for the largest share of the overall abusive acts. The predicted probability, using logistic regression, shows that literate women living with alcoholic husbands, women engaged in gainful income generating activities and women living in food insecure households were more susceptible to intimate partner violence. The study concluded that while the main determinants are generally embedded in the socio-cultural practices and attitudes of the community, there are certain individual and household level variables which significantly affect its likely occurrence.
Key words: Intimate Partner, Violence, Sidama Zone, Southern Ethiopia.
Background
According to the World Health Organization (WHO), violence is “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). This definition actually capitalizes on physical violence, which may have far reaching consequences on the victims.
A common form of violence, domestic violence, also known as domestic abuse, spousal abuse, family violence and intimate partner violence (IPV), is generally understood as a pattern of abusive behavior by one or both partners in an intimate relationship such as marriage, dating, family, friendship or cohabitation (Garcia-Moreno, 2006). Domestic violence manifests largely as violence against women (VAW) and is often performed by a husband or an intimate male partner.
IPV occurs in many countries, irrespective of social, economic, religious or cultural group. In 48 population-based surveys from around the world, between 10% and 69% of women reported being physically assaulted by an intimate male partner at some point in their lives.(Heise, Ellsberg, Gottemoeller, 1999). The percentage of women who had been assaulted by a partner in the previous 12 months varied from 3% or less among women in Australia, Canada and the United States to 27% of ever-partnered women (that is, women who have ever had an ongoing sexual partnership) in León, Nicaragua, 38% of currently married women in the Republic of Korea, and 52% of currently married Palestinian women in the West Bank and Gaza Strip ( Heise, Ellsberg, Gottemoeller, 1999).
A study by WHO estimated that the most common act of violence experienced by women was being slapped by their partner, ranging from 9% in Japan to 52% in provincial Peru. Furthermore, this was followed by being struck with a fist, for which these two settings again represented the  extremes (2% and 42%, respectively). The percentage of everpartnered women experiencing severe physical violence, according to the WHO report, ranged from 4% of women in Japan to 49% in provincial Peru, with most countries falling between 13% and 26%. The exceptions were  urban Bangladesh, Ethiopia, Namibia, and Samoa where more women reported severe violence in the past 12 months (WHO,
2001). Another study indicated that one in three women are physically assaulted, sexually coerced or otherwise abused in their lifetime, in most cases by an intimate partner ( IFPP, 2007).
The prevalence of intimate partner violence toward women varies greatly across settings, according to a 15-site study conducted in 10 countries: Bangladesh, Brazil, Ethiopia, Japan, Namibia, Peru, Samoa, Serbia and Montenegro, Thailand and Tanzania (Garcia-Moreno  et al, 2006). The proportion of ever-partnered women who reported ever having experienced either sexual or physical violence, or both, ranged from 15% in a Japanese city to 71% in  an Ethiopian 1.  Nigatu Regassa, PhD, is Vice President for Business and Development, Hawassa University, Ethiopia.102 Inkanyiso, Jnl Hum & Soc Sci 2011, 3(2) province. Sexual violence tended to be less prevalent than physical violence, but in most sites, between 30% and 56% of
ever-partnered women had experienced both kinds of  partner violence. Similarly, the lowest prevalence  of sexual violence in the past 12 months was reported in Japan, Serbia and Montenegro (1%), while the highest current prevalence (44%) was in rural Ethiopia ( Garcia-Moreno  et al., 2006). Studies in Africa have reported that about half of all evermarried women in Zambia, 46% in Uganda, 60% in Tanzanian, 42% in Kenya, and a high of 81% in Nigeria have experienced some form of violence in their lives (Kishor and Johnson 2004).
The above few national level studies indicate that Ethiopia ranks as one of the topmost with regard to domestic violence. To the best of the author’s knowledge, there are no population based attempts to estimate the prevalence of intimate violence on a cultural or ethnic group or small population level. Most of our knowledge of domestic violence is dependent on the national level reports by the Demographic and Health Surveys of 2000, 2005 and 2010, which normally do not present the statistics for each ethnic and cultural group. Also, most studies estimate or analyze the prevalence of violence and its risk factors solely based on the affirmative responses of the women respondents on lifetime occurrence or absence of an abusive event. This study has tried to overcome some of these weaknesses and aimed at estimating the
magnitude of occurrence of intimate partner violence – violence perpetrated against a woman by her husband – in one of the most populous zones (one ethnic group) of Southern Ethiopia, the Sidama. The present study is limited to answer three important questions: What is the magnitude of intimate partner violence in Sidama zone? What most common abusive acts occur at household level? What are the most important risk factors for the likelihood of violence to occur?
Data source and methodology of the studyThe study setting The study is conducted in Sidama Zone, one of the most populous district of Southern Ethiopia (SNNPR). Sidama is located in the Southern Nations,  ationalities and Peoples Region (SNNPR). The Sidama zone is bordered in the south by the Oromia Region except for a short stretch in the middle where it shares a border with Gedeo, on the west by the Bilate River which separates it from Wolayita, and on the north and east by the Oromia Region. The administrative center
for Sidama is Hawassa town. According to the recent census (CSA, 2007), the total population of the zone was 2 954 136, of which 1 491 248 are men and 1 462 888 women; with an area of 6 538.17 square kilometers. Sidama has a population density of 451.83. While 162 632 or 5.51% are urban inhabitants, a further 5 438 or 0.18% are pastoralists. The total households enumerated in 2007 was 592 539, which resulted in an average household size of 4.99 persons. (CSA 2007).
A substantial area of the Sidama land produces coffee, which is the major cash crop in the region, and the bulk of the population of the area are known to depend heavily on ‘enset’ (enset ventricosum) as a major source of survival. It is the single most important root crop grown in the study.
Sampling
The 1 094 households were selected from two agro-climatic zones; areas representing the high land and low land of Sidama zone of Southern Ethiopia using appropriate probability sampling. The sample size determination formula adopted for this study is given by Woodward, 1992, where Z is the upper  points of standard normal distribution with =0.05 significance level, which is =1.96, d is the degree of precisions (0.04), p is the proportion of wives facing violence (which is taken as 0.5 or 50%). The estimated sample size, using the above formula yields 600, weighted by 1.5 to get a total size of 900, with a 20% contingency added.
Then, probability sampling in a form of simple random and two-stage sampling method was employed for selecting the required size from the study areas. Since the two sub-districts (the low and high land) were decided in advance, the first stage of the sampling started by selecting five kebeles (small villages) from the list of the two sub-districts using simple random sampling. At the second stage, a random sample of households was selected from the available list, giving a total of 1 094 households. The selection of two focus group discussion participants (one for male and one for female groups) followed purposive sampling techniques.
n
Z
2
p( ) 1 – p
d
2
--------------------------
1
1

+ ----
Z
2
p( ) 1 – p
d
2
-------------------------- – 1
 
 
 
= -------------------------------------------------------
2
α
α Z103
Inkanyiso, Jnl Hum & Soc Sci 2011, 3(2)
Data collection
The data for this study were generated through a well structured interview schedule. Prior to the data collection, the checklists/schedules underwent intensive review and pre-testing on 15 subjects from all categories of respondents, and the final instrument was duplicated once some adjustments were made following the pre-test. During the interview, the enumerators went through all the items with the wife and husband separately.
The most important data forming the main response variable, intimate partner violence, was cautiously framed using standard and universally accepted tools. An attempt was made to develop measurement of domestic violence consistent with WHO guidelines. The information on the dependent variable was collected using five sets of questions asked for both the husband and wife independently. The nature of the questions call for responses on the occurrence/absence of selected violence related events (such as beating,  insulting, chasing, causing physical assault and slapping) during a reference period of twelve months prior to the survey date.
Based on the recommendation of WHO (2001), safety and ethical considerations in the collection of violence data were observed. For instance, an attempt was made to conduct the interviewing of the women in a private setting, administering the questionnaire on only one woman per household to keep the secrecy of the responses given by the women. In addition to the quantitative data, two focus group discussions (one for males and one for females) were held in both selected areas.
Data processing and analysis
The data processing and analysis started with computing the percentages agreed and disagreed by the couple on the violence indicators mentioned above. The dependent variable was framed by taking the proportion of couple who agreed on all the five violence indicators.
Both bivariate and multivariate analyses were employed to identify the risk factors for intimate partner violence. In the bivariate analysis, Pearson's chi-square test of independence was used to test the existence of significant association between categories of violence status and selected risk factors. The fact that the chi-square bivariate analysis indicated effects or associations of variables without controlling the confounding effects, the net effects of each independent variable was further examined using multivariate (logistic regression) analysis. The logistic regression model is given by the formula:
Where: P is the probability that the event y occurs, at p(y=1);
p/(1-p) is the “odds ratio”;
In this case, p would be the probability of facing violence by women, where as 1-p would be probability of not facing violence, a the constant term, and  B the logistic coefficient and  exp(B) the factor by which odds change when the independent variable is increased by one unit.
The analyses rest on two partner violence outcomes; whether a woman faced violence by her partner or not. Since the interest is in identifying women at risk of facing violence, the dependent variables are coded as 1 if the woman faced violence and 0 if not. A woman was counted as facing violence if she gave affirmative responses to all five violence indicators (beating, slapping, insulting, physical or sexual abuse, chasing by partner) and the husband agreed on their occurrence during the reference period.
The odds ratio, which is determined from the logistic regression coefficients, tells us the increased  or decreased chance of violence occurrence given a set level of the independent variable while controlling for the effects of the other variables in the model. Estimates of odds greater than 1.0 indicate that the risk of violence is greater than that for the reference category. Estimates less than 1.0 indicate that the risk of violence is less than that for the reference category of each variable.
Results
Table 1 presents the background characteristics of respondents. The age distribution of the women showed that a large proportion of them (47.7%) were in early adulthood (age 25-34) followed by those in the age group 15-24 (39.4%). With regard to the number of children born by women, larger proportions of women gave birth to 1-3 (46.3%)  and 4-7 children (35.5%). Protestant Christians accounted for the highest proportion in the religious distribution (73.5%) followed by Catholic and Muslim (10.8 and 9.4% respectively), while the remaining religious groups contributed a small proportion of the respondents. As expected, the large majority of the respondents (60.6%) came from households with p ⁄ ( ) 1 – p = exp( ) a Bx c + +104 Inkanyiso, Jnl Hum & Soc Sci 2011, 3(2) 4-7 members and still 22.1 percent had more than 7 members. The average household size for the study population was 5.87. The analysis shows that 15.3 percent of the women were engaged in polygamous marriage arrangements (i.e the husbands were reported to have two or more wives during the survey), which is slightly above the national average (i.e. the DHS 2005 reported 11 percent for the country).
The distribution of the respondents by educational status revealed that the majority of the women respondents were illiterate (56.3%) followed by elementary school level (27.9%), while the remaining respondents accounted for a smaller proportion of the respondents. The majority (47.1%) of the  espondents were farmers, 39.9 percent were selfemployed while the remaining employment categories  contributed a smaller proportion of the respondents. Land ownership by households in the study area was quite small and fragmented. It is seen from Table 1 that about 95 percent of the households owned a land less than one hectare in size.
Table 1 Percentage distribution of respondents by selected background characteristics, Sidama Zone, 2011 (n= 1094) Characteristics Number (n= 1 094) Percent
Age of women
15-24
25-34
35-49
431
489
174
39.4
44.7
15.9
Number of children born
None
1-3 children
4-6 children
7-10 children
50
506
388
150
4.6
46.3
35.5
13.7
Religion
Orthodox Christian
Catholic
Protestant
Muslim
Traditional
28
118
804
103
25
16
2.6
10.8
73.5
9.4
2.3
1.5
Household size
0-3 persons
4-7 persons
greater than 7 persons
189
663
242
17.3
60.6
22.1
Marital form
Polygamous
Monogamous
166
928
15.2
84.8
Educational status
Illiterate
Elementary (1-6)
Junior secondary (7-8)
Secondary (9-12)
College diploma
Others
616
305
69
44
14
46
56.3
27.9
6.3
4.0
1.3
4.2
Usual occupation
Self employment
Civil servant
Farmer
Petty trader
Others
437
13
515
60
69
39.9
1.2
47.1
5.5
6.3105
Inkanyiso, Jnl Hum & Soc Sci 2011, 3(2)
Table 2 presents the percentage distribution of respondents (both husband and wife) by reported violence acts during the past 12 months prior to the survey date. Husband insulting the wife is the highest rate reported and agreed by 61.2 percent of the couple, followed by beating (34 %); slapping (32.9%); chasing the wife out from home for short or longer period (25.4 %); and the least agreed act of violence is causing serious physical damage to the wife (14.7 %).
Looking at the summary statistics done on Table 3, one finds couples significantly disagreeing on most of the violent acts i.e. either the husband or the wife denied the occurrence of the event during the reference period. The percentage of the respondents agreeing on none of the violence indicators accounts for 12.2 percent. In a more strict sense, there are still quite a large proportion of the respondents (32.7%) agreeing on the occurrence of all five abusive acts during the reference period. The 32.7 percent prevalence is taken to frame the dependent variable in the logistic regression analysis.
Once the prevalence of intimate partner violence is estimated, the second objective of this small scale study is to examine the risk factors for violence to take place. Table 4 presents the association between categories of violence status and selected background variables using the Pearson’s Chi-square test of independence. The result suggests that six of the ten variables have significant association with intimate partner violence at different p values.
Because the chi-square bivariate analysis indicates effects or associations of an independent variable and that of the outcome variable without controlling the confounding effects, the net effects of each independent variable were further examined using multivariate analysis. The logistic regression analysis results given in Table 5 included 10 independent variables to regress against the dependent variable of interest. Of the total variables entered into the model, five were significantly associated with the dependent variable, namely literacy status of the women, participation in income generating activities, household size, alcohol intake and food security status at different p value.
Land size owned by the household
Landless
< 0.5 hectare
0.5-1 hectare
> 1 hectare
31
459
579
25
2.8
42.0
52.9
2.3
Table 2 Percentage distribution of husband and wife by reported agreement on each type
of violence indicators, Sidama Zone, 2011 (n = 1094)
Type of abusive acts
% agreed by the
partners
% not agreed by the
partners
Beating of wife  34.0 66.0
Wife slapped by husband 32.9 67.1
Wife insulted by husband 61.2 38.8
Wife chased from home 25.4 74.6
Faces serious physical damage by the husband 14.7 85.3
Table 3 Summary on the percentage distribution of partners’ agreement by type of violence
indicators, Sidama Zone, 2011. (n = 1094)
Violence indicators  Number Percent
Not agreed by any of the indicators 134 12.2
Agree on one violence indicator 73 6.7
Agree on two of the violence indicators 129 11.8
Agree on three violence indicators 138 12.6
Agree on four violence indicators 262 23.9
Agree on all five violence indicators 358 32.7
Total 1094 100.0106
Inkanyiso, Jnl Hum & Soc Sci 2011, 3(2)
Table 4 Results of bivariate analysis for associations between intimate partners’ violence and selected explanatory
variables, Sidama Zone, 2011 (n = 1094).
Variable Category
Violence status
Chi-square test
χ
2
Faced violence
Not faced
violence Total
Age of the women
age 15-24 14.1 25.3 39.4
6.14***
(.046)
age 25-34 14.6 30.1 44.7
age 35-49 4.0 11.9 15.9
32.7 67.3 100.0
Household size
0-3 persons 7.1 10.1 17.3
8.463*
(.015)
4-7 persons 19.3 41.3 60.6
greater than 7 persons 6.3 15.8 22.1
32.7 67.3 100.0
Participation in decision
making
Yes 25.9 48.2 74.0
6.949
***
No 6.9 19.1 26.0
(.008)
32.7 67.3 100.0
Age of household head
age 15-24 4.8 8.7 13.4
2.904
(.407)
age 25-50 23.1 46.6 69.7
Above age 50 .7 2.7 3.4
Not reported 4.1 9.3 13.4
32.7 67.3 100.0
Litracy status of the
women
Literate 13.2 21.9 35.1
6.13*
(.013)
Illiterate 19.6 45.3 64.9
32.7 67.3 100.0
Participation in income
generation
Yes 7.4 18.0 25.4
2.179
No 25.3 49.3 74.6
(.140)
 32.7 67.3 100.0
Religion
Orthodox Christian .6 1.9 2.6
14.668*
(.012)
Catholic 4.4 6.4 10.8
Protestant 23.2 50.3 73.5
Muslim 2.6 6.9 9.4
Traditional 1.4 .9 2.3
Others .5 .9 1.5
32.7 67.3 100.0
Educational status of the
husband
Illiterate 9.4 19.9 29.3
1.045
(.959)
Elementary (1-6) 12.7 26.3 39.0
Junior secondary (7-8) 5.5 12.0 17.5
secondary (9-12) 3.3 5.9 9.1
college diploma .5 1.0 1.6
Illiterate 1.3 2.2 3.5
32.7 67.3 100.0
Marital form
Polygamous 4.7 10.5 15.2
.356
Monogamous 28.1 56.8 84.8
(.551)
32.7 67.3 100.0
Age difference between
husband and wife
Less than five years 26.1 55.6 81.7
1.193
(.275)
5-10 years 6.6 11.7 18.3
32.7 67.3 100.0
Alcohol taking
Yes 3.0 10.3 13.3
7.840*
No 29.7 56.9 86.7
(.005)
32.7 67.3 100.0
NB: *=p<0.05, **=p<0.01, ***=p<0.001107
Inkanyiso, Jnl Hum & Soc Sci 2011, 3(2)
Table 5 Results of logistic regression for associations between intimate partners’ violence and selected explanatory
variables, Sidama Zone, 2011 (n= 1094) .
 Variables  B S.E. Sig. Exp(B)
Age of women
Age 15-24 (RC) - - .- -
Age 25-34 -.350 .224 .118 .705
Age 35-49 -.332 .211 .116 .718
Religion
Orthodox Christian(RC) - - - -
Catholic .461 .706 .514 1.586
Protestant -.426 .578 .461 .653
Muslim .215 .548 .695 1.240
Traditional .247 .589 .674 1.281
Others -1.278 .694 .066 .279
Marital form
Polygamous (RC)
Monogamous .150 .191 .434 1.162
Literacy status
Literate (RC) - - -- -
Illiterate  -.280 .144 .050 .756*
Participation in income generating activities
Yes (RC) - - - -
No  .274 .152 .017 1.315*
Household Size
0-3 persons (RC) -- - - -
4-7 persons -.577 .231 .012 .561*
greater than 7 persons -.171 .178 .033 .843*
Educational status of the husband
Illiterate (RC) - - - -
Elementary (1-6) .154 .369 .676 1.167
Junior secondary (7-8) .185 .366 .613 1.204
secondary (9-12) .350 .388 .367 1.419
college diploma .104 .414 .802 1.110
Others  .371 .633 .557 1.449
Alcohol intake by the partner
 Yes (RC) -- - - -
No -.585 .223 .009 .795*
Age difference between husband and wife
No difference (RC) - - - -
Up to 10 years difference .195 .175 .267 1.215
Food security status
Insecure (RC) - - - -
Secured  -.590 .169 .000 .554***
Constant .802 .680 .238 2.230
RC-Reference category, -2log likelihood = 1321.68; Number of cases = 1094 * Significant at .05; ** Significant at .01; ***
Significant at .001
Source: Field survey, 2011108
Inkanyiso, Jnl Hum & Soc Sci 2011, 3(2)
It is seen that illiterate women were 24.4 percent  times less likely to face violence compared to the  literate ones (OR=.756 and p = 0.05). Women who were not participating in gainful income generating activities were 1.32 times more likely to face violence compared to the reference category (OR=.1315 and p = 0.017). Households with size of 4-7 and 7+ were 43.9 and 15.7 percent less likely to be exposed to violence compared to the reference category. Women living with non-alcohol drinking husbands were 20.5 less likely to face violence compared to the reference category (OR=.795 and p = 0.009). Finally, food secured households (which is used as a proxy of economic status) were 44.6 percent less likely to face violence compared to the food insecure households (OR=0.554 and p = 0.000).
Discussion
The study examined the magnitude of IPV based on a representative sample of 1094 households taken from the Sidama Zone. It is understood from the analysis that quite a good proportion of women had faced the full scale of domestic violence by their partner during the past twelve months. Among the list of violence indicators used to collect the data, being insulted and beaten by the husband accounted for the highest rate, 61.2 and 34 percents respectively. Slapping was very commonly practiced in the study area accounting for about 32.9 percent. If one looks at the different rates given for beating (beating by stick, leg, stone, or any other material) slapping and physical damages reported and agreed by the couple (see table 2), this gives a clear understanding of how women in the study were living in abusive environments. It is also noteworthy that many of the physical abusive acts generally ranged from moderate to severe categories per the WHO’s definition. According to WHO, the severity of a physically violent act is ranked according to its likelihood of
causing physical injuries: Being slapped, pushed or shoved is defined as moderate physical violence: Being hit with a fist, kicked, dragged, threatened with a weapon, or having a weapon used against one is defined as severe physical violence (WHO, 2002).
There were various reasons/pretexts used by husbands for beating or physically abusing their wife in the study areas. The informal discussions held with some FGD participants suggested that most commonly cited reasons for husbands were issues related to child care and food while other reasons can also be used in other times such as, arguing back or responding back to the man, not obeying rules; nagging the man for money, visiting relatives or going somewhere without the husband knowledge and consent, infidelity and the like. But above all, there was a general understanding among the community members that “a husband should discipline his wife through any mechanisms he feels appropriate”. This thinking seems to be imbedded in the social, cultural and religious practices in most localities in the study area.
The study found out that the factors associated with IPV were related to either the women’s characteristics or household socio-economic status, or husband’s characteristics. The two most women’s characteristics predicting the likelihood of violence occurrence were literacy and participation in income generating activities. Contrary to the literature, education (measured in terms of literacy status) negatively affected the chance of facing  intimate partner violence among the study population. Women who were illiterate were 24.4 percent less likely to experience intimate partner violence compared to their counterparts, literate women, suggesting that modest increase in educational status of women does not bring significant changes in the reduction of violence. From a feminist perspective, this finding could be valid. Women with more education tend to be more assertive, and thus will not settle for less in their gender relationships. Their male partners consequently feel that their power and control are being threatened, leading to violence. This has been corroborated by similar studies in Ghana by Ami (2008). Garciá-Moreno (2002),  in a crosscultural study, also noted that women with very poor education were less likely to be abused, but those with higher and enough education to challenge the status quo are at the greatest risk of IPV. In contrast, other studies conducted in lowincome countries by Morales and Reichenheim (2002)  and Koenig  et al. (2003) found that women who have low
educational attainment were more likely to experience violence.
This study also documented that women who take active part in income generating activities outside their home are more susceptible to partners’ violence compared to those not participating. It is understood that women whose income is a substantial contribution to their household behave more assertively and gain better bargaining power which threatens men’s sense of control and superiority due to possible shift in the balance of power relationship. It can also be argued that income generating women are assumed to enjoy better freedom of movement, and hence, become far from the usual control of their husbands which again threatens the power relationship in the family. This result is consistent with the finding of Roger (2007) who found wives working outside tbe home are 1.04 times more likely to be abused by their husbands than wives not currently working. Similarly, a study in Bangladesh supported this hypothesis (Lisa et al, 2004).
The study has also documented that food security status and IPV were strongly associated (p value of 0.000). This variable is used as a proxy to measure the household income or economic status. It is measured by using standard and universally accepted tools developed by the Food and Nutrition Technical Assistance Project (FANTA).The project has developed a set of questions (Household Food Insecurity Access Scale Generic Questions) that have been used in several 109 Inkanyiso, Jnl Hum & Soc Sci 2011, 3(2)
countries and appear to distinguish the food secure from the insecure households across different cultural contexts. The set of questions are used to assign households along a continuum of severity, from secure to severely insecure (Coats et’al, 2007). The logistic regression result revealed that women living in food secure households are by far less likely to be exposed to violence by partners compared to those living in food insecure households. Studies from a wide range of settings show that, while physical violence against partners cuts across all socioeconomic groups, women living in poverty are disproportionately affected (Hindin 2003; Jewkes 2002). The low income in itself may not directly breed violence.
However, it may operate through different mechanisms such as instilling fear in men and promote  opelessness, stress, frustration and a sense of inadequacy among men for having failed to live up to their culturally expected role as providers.
In situations, where men lack the resources associated with their assumed dominant role of a male breadwinner, they are more likely to express their frustration through violence. In a similar fashion, it has been observed that if women have little access to resources, or if they have greater access to resources than their partners, thereby eroding the traditional position of men, they are placed at a higher risk of victimization by their male partners (Hindin 2003; Jewkes 2002; Brinkerhoff et al. 1992).
In relation to the husbands’ characteristics, alcohol has become an important risk marker for partner violence that appears especially consistent across different settings (Walton et al 2005; Sharps et al. 2003; Moreno, 1999; McCauley, 1995). Many of the studies that examined alcohol use or excessive drinking as a risk factor for partner violence found a significant association, with correlation coefficients ranging from r = 0.21 to r = 0.57. Population based surveys from Brazil, Cambodia, Canada, Chile, Colombia, Costa Rica, El Salvador, India, Indonesia, Nicaragua, South Africa, Spain and Venezuela also found a link between a woman’s risk of suffering violence and her partner’s drinking habits (Ellsberg, 2000). Many researchers believe that alcohol operates as a situational factor, increasing the likelihood of violence by reducing inhibitions, clouding judgment and impairing an individual’s ability to interpret cues. Consistent with this finding, a national level analysis reported a positive correlation between men’s problem drinking and domestic  violence, a relationship that has been found in high-income countries and other social classes as well (Jewkes et al. 2002; Koenig et al, 2003).
The data collected for this study have overcome some of the inherent limitations of most national level surveys. Most previous studies (Abramsky et al. 2011; Michelle et al. 2011) for instance, did not collect the violence related information from the husband, and the analysis solely depend on women’s responses; the instruments usually miss some important abusive acts; a women is counted as experiencing violence if she gives affirmative responses to one or few of the violence indicators. This study on the contrary has tried to overcome many of these weaknesses during the course of data collection and analysis.
There are also some limitations worth mentioning. The study provides a single/snapshot look at the study population, hence it may not be a sound base for studying the underlying causes and mechanisms related to domestic violence. Thus, until longitudinal data collection is possible, this study can only be used to understand the risk factors (not causal relationship) of intimate partners’ violence. In addition, the timing of episode of violence cannot be established through such cross-sectional study, i.e women were asked to report if they have ever experienced violence in a reference period prior to the survey date. Finally, it is worth mentioning that the current study did not compare violence by rural-urban residence since the data were collected from rural districts, and comparison in violence acts was not made between husband and wife as most of abusive acts occurs on women in the study areas. Despite these weaknesses, it is believed that this small piece of work contributes to our understanding of the depth and width of intimate partners violence  in Sidama zone, Southern Ethiopia.
Conclusion and policy implications
The study has shown that IPV is very high in the study area, and that there are certain risk factors aggravating its occurrence. While the main determinants are generally embedded in the socio-cultural practices and attitudes of the community, there are certain individual and household level variables which significantly affect its likely occurrence. The first and foremost call of this study is the prevention of the occurrence of domestic violence by all machineries at both national and local levels. At the national level, priorities include improving the status of women, establishing appropriate norms, policies and laws on abuse, and creating a social environment that is conducive to non-violent relationships. The most common reforms involve criminalizing physical, sexual and psychological abuse by intimate partners. There is a need for public education on IPV and its criminality using intensive behavioral change strategies.
At local level, in addition to strictly implementing laws at grassroots levels where abusive acts are presumed to be higher, efforts should be made to customize the experiences of more developed countries on human right (more specifically women’s right and protection) such as by introducing special domestic violence courts or sessions in the court system, training police and court officials and prosecution lawyers, and setting in special and free of cost consultation/ 110 Inkanyiso, Jnl Hum & Soc Sci 2011, 3(2) advises to victim women at local areas. Equally important is provision of peer education at household and community level, use of communication campaigns, integrating the issue of violence into other events and other community friendly strategies can be used. Such attempts to raise awareness and behavioral change endeavor, however, should also take into account men, informal community leaders and young community members.
Acknowledgement
This research is part of the post-doctoral research fellowship, fully sponsored by the third phase Norwegian government supported fund. The author would like to express his heart felt gratitude to Hawassa University-NORAD Project for the financial support.
References
Abramsky Tanya, Charlotte H Watts, Claudia Garcia-Moreno, Karen Devries, Ligia Kiss, Mary Ellsberg,Henrica AFM Jansen,
Lori Heise. 2011. What factors are associated with recent intimate partner violence? findings from the WHO multi-country study on women’s health and domestic violence BMC Public Health 2011, 11:109
Ami R. Moore. 2008.Types of Violence against Women and Factors Influencing Intimate Partner Violence in Togo (West Africa).
J Fam Vol 23,777–783 Central Statistical Authority (CSA) 2007. Summary and Statistical Report of the 2007 Population and Housing Census Results.
Addis Ababa, Ethiopia Coates, J., Swindale, A. & Blinsky, P. 2007. Household Food Insecurity Access Scale (HFIAS) for Measurement of Household
Food Access: Indicator Guide (v.3). Washington D.C: Food and Nutrition Technical Assistance Project, Academy for Educational Development.
Brinkerhoff, M. B., Grandin, E., & Lupri, E. 1992. Religions involvement and spousal abuse: the Canadian case. Journal for the Scientific Study of Religion, 31, 15–31. doi:10.2307/1386829 Ellsberg MC. 2000. Candies in hell: women’s experience of violence in Nicaragua. Social Science and Medicine, 51,1595–1610.
Garcia-Moreno C. 2006. Prevalence of intimate partner violence: findings from the WHO multi-country study on women's health and domestic violence, Lancet, 368(9543),1260–1269.
Garciá-Moreno, C. 2002. "Violence Against Women: Consolidating a Public Health Agenda". In G George and A. Ostlin (eds),
Engendering International Health: the Challenge of Equity, Cambridge (MA): MIT Press, 111-141
Heise, L.L, Ellsberg, M, Gottemoeller M. 1999. Ending violence against women. Baltimore, MD, Johns Hopkins University
School of Public Health, Center for Communications Programs,(Population Reports, Series L, No. 11).
Hindin, M. J. 2003. Understanding women’s attitudes toward wife beating in Zimbabwe.  Bulletin of the World Health Organization, 81, 501–508.
International Family Planning Perspectives (IPPF).2007. Levels of Intimate Partner Violence Volume 33, Number 1,
Jewkes, R. 2002. Intimate partner violence: Causes  and prevention.  Lancet, 359, 1423–1429. doi:10.1016/S0140-6736(02)08357-5.
Kishor, S., & Johnson, K. 2004. Profiling domestic violence – A multi-country study. Calverton, Maryland: ORC Macro
Koenig, M. A., Lutalo, T., Zhao, F., Nalugoda, F., Wabwire-Mangen, F., Kiwanuka, N., et al. (2003. Domestic violence in rural
Uganda: Evidence from a community-based study. Bulletin of theWorld Health Organization, 18, 53–60.
Lisa M Bates, Sidney Ruth Schuler, Farazana Islam and Md. Thairul Islam. 2004. Socioeconomic factors and processes associated
with domestic violence in rural Bangladesh. J. IFPP, Vol.30 ,4.
McCauley J. 1995.The "battering syndrome": prevalence and clinical characteristics of domestic violence in primary health care internal medicine practices. Annals of Internal Medicine, 123,737–746 Michelle J. Hindina, Linda S. Adairb. 2002.Who’s at risk? Factors associated with intimate partner. Violence in the Philippines.
Social Science & Medicine 55 (2002), 1385–1399
Moraes, C. L., & Reichenheim, M. E. 2002. Domestic violence during pregnancy in Rio de Janeiro, Brazil. International Journal of Gynecology and Obstetrics, 79, 269–277. doi:10.1016/S0020- 7292(02)00250-3
Moreno Mart´n F. 1999. La violencia en la pareja. [Intimate partner violence.]  Revista Panamericana de Salud Pública, 1999,5,245–258.
Roger Y. Klomegah 2007. Intímate Partner Violence (IPV) in Zambia: An Examinatíon of Risk
Factors and Gender Perceptíons. Journal of Comparative Family Studies ,558-569.
Sharps, P., J.C. Campbell, D. Campbell, F. Gary, and D. Webster. 2003. Risky Mix: Drinking, Drug Use, and Homicide. NIJ Journal 250, 8–13,WHO 2002. World report on violence and health. Geneva: World Health Organization.
Woodward. 1992 Formula for sample size, power and minimum Detectable Relative Risk in Medical Studies. The Statistician,Vol 41(2),185-196
World Health Organization (WHO). 2001. Putting women first: ethical and safety recommendations for research on domesticviolence against women. Geneva, World Health Organization, (document WHO/FCH/GWH/01.01).
Walton-Moss BJ, Manganello J, Frye V, Campbell JC. J Community Health. 2005
http://www.inkanyiso.uzulu.ac.za/journals/InkanyisoVol3issue2.pdf