yenriquez@ucss.edu.pe
http://www.dx.doi.org/10.14718/ACP.2020.23.1.13
Articles
Yordanis Enríquez-Canto 1
Yonathan Josué Ortiz-Montalvo 2
Katherine Jenny Ortiz-Romaní 3
Giovani Martín Díaz-Gervasi 4
1 Universidad Católica Sedes Sapientiae, Lima, Perú. Facultad de Ciencias de la Salud, Esq. Constelaciones y Sol de Oro s/n, Urb., Sol de Oro, Los Olivos, Lima, Perú. C. P.: 15302. Tel.: +51 (01) 533-0079,
yenriquez@ucss.edu.pe
0000-0003-0449-5015
2 Universidad Católica Sedes Sapientiae, Lima, Perú.
0000-0002-7678-8303
3 Universidad Católica Sedes Sapientiae, Lima, Perú.
0000-0003-0331-4446
4 Universidad Católica Sedes Sapientiae, Lima, Perú
0000-0002-2505-2689
Recibido, marzo 27/2019
Concepto de evaluación, mayo 10/2019
Aceptado, mayo 30/2019
Como citar este articulo / How to cite this article: Enríquez-Canto, Y., Ortiz-Montalvo, Y.J., Ortiz-Romaní, K.J. & Díaz-Gervasi, G.M. (2020). Ecological analysis of intimate partner sexual violence in Peruvian women. Acta Colombiana de Psicología, 23(1), 272-286. doi: http://doi.org/10.14718/ACP.2020.23.1.13
Abstract
The ecological model explains intimate partner sexual violence by the interaction of multiple factors that coexist at different levels. The aim of this paper was to analyze the interaction between those levels to estimate the increase or decrease of risk in recent sexual violence by the current partner in 21 414 women at childbearing age (M=30.3 years; SD= 9.2), of whom 59.4% were in cohabitation and 56% had independent work. A two-stage sampling by conglomerates and stratified by areas of Peru's departments was used. To this end, a secondary data analysis from the Demographic and Family Health Survey (2017) was carried out taking into account the modules of violence against women. Four models were determined using the Poisson regression and calculating prevalence ratios and the area under the curve. The results showed that 6.3% of the sample experienced intimate partner sexual violence. Among the individual factors that increase the risk there is a history of domestic violence (PR=1.19 IC:1.07-1.32). In the Microsystem, physical violence (PR = 11.04 IC: 8.49-14.36), economic threats (PR = 2.58 IC:2.24-2.94) and the partner's frequency of drunkenness were found (PR=1.98 IC:1.53- 2.56). On the contrary, communication (PR=0.45 IC:0.40 -0.51) and belonging to higher wealth quintiles (PR=0.59 IC:0.45-0.76) are protective factors of sexual violence. Finally, it is concluded that the simultaneous and reciprocal influence between ecological levels may not be equal regarding the predictive weight of the factors.
Key words: sexual violence, intimate partner violence, ecological model, risk factors.
Resumen
Según el modelo ecológico, la violencia sexual de pareja se explica por la interacción de múltiples factores que coexisten en diversos niveles. Teniendo esto en cuenta, en el presente estudio se buscó medir la interacción existente entre los niveles según el aumento o disminución del riesgo de violencia sexual reciente por parte de la pareja actual en 21 414 mujeres en edad fértil —59.4 % conviviente, 56 % con trabajo independiente y M = 30.3 años (DE = 9.2)—, por medio de un muestreo bietápico por conglomerados y estratificado por áreas departamentales del Perú. Para ello, se realizó un análisis de datos secundario a partir de lo obtenido en los módulos de violencia contra la mujer de la Encuesta Demográfica y de Salud Familiar (ENDES, 2017), con lo cual se determinaron cuatro modelos a partir de la regresión de Poisson y del cálculo de las razones de prevalencia y del área bajo la curva. Como resultado, se encontró que el 6.3 % experimentó violencia sexual de pareja; que entre los factores individuales que aumentan el riesgo se incluyen, principalmente, los antecedentes de violencia familiar (PR = 1.19; IC: 1.07-1.32); que en el microsistema se encuentran la violencia física (PR = 11.04 IC:8.49-14.36), las amenazas de tipo económico (PR = 2.58 IC:2.24-2.94) y la frecuencia de embriaguez del compañero (PR = 1.98 IC:1.53-2.56); mientras que la comunicación (PR = 0.45 IC:0.40-0.51) y pertenecer a quintiles superiores de riqueza (PR = 0.59 IC:0.45-0.76) resultaron ser factores protectores de la violencia sexual; y, por último, que la influencia simultánea y recíproca entre los niveles ecológicos no resultó paritaria respecto al peso predictivo de los factores.
Palabras clave: violencia sexual, violencia de pareja, modelo ecológico, factores de riesgo.
Introduction
Partner violence is a worldwide phenomenon that constitutes a violation of human rights with a significant impact on women's health (WHO, 2013). The World Health Organization (WHO) defines it as any behavior on the part of an intimate partner within a relationship that causes psychological, physical or sexual harm (WHO, 2002). It is also one of the types of violence most frequently suffered by women and may include one or more acts of physical or sexual violence (WHO, 2012; WHO, 2013). Intimate partner sexual violence is reported when the woman is forced to have sex or humiliating or degrading sexual acts (WHO, 2013). The same may be accompanied by other forms of abuse or occur by itself having an impact on health with adverse results in the short and long term (WHO, 2013; Krebs, Breiding, Browne & Warner, 2011). These include emotional distress, suicidal thoughts and attempts, addiction to substances such as alcohol and increased risk of acquiring sexually transmitted infections (Devries et al., 2011; Garcia-Moro, Jansen, Ellsberg, Heise, & Watts, 2006).
Internationally, approximately three out of ten women have suffered physical or sexual violence from their partner (WHO, 2013). In the Americas region, this prevalence reaches 29.8%, the highest after the regions of Africa, the Middle East and Southeast Asia (37%) (WHO, 2013), while in Peru this prevalence in the course of the life of women older than 15 years is 22.5% (WHO, 2012). The Household Demographic and Family Health Survey (ENDES, for its Spanish acronym) of 2017 reported that intimate partner sexual violence in the last twelve months had been more frequent in women with an average age of 35 years, separated (19.8%), with a higher educational level than the partner (7.3%) and belonging to the lower wealth quintile (7.6%) (INEI, 2017b).
Data on the prevalence of sexual violence need to be deepened. Peruvian studies have been conducted that investigate the factors associated with intimate partner violence. However, these are rare and, often, sexual violence is overlooked. Some of them focus on physical violence (Flake, 2005; Blitchtein & Reyes, 2012), using indices that show the intensity of violence without distinguishing between types of violence (Castro, Cerellino & Rivera, 2017), also considering sexual violence in specific populations, for example, pregnant women (Barrios et al., 2015) or residents in very low income communities (Nóblega, 2012). Therefore, the factors associated with recent sexual violence would not be sufficiently explored.
The prevalence of sexual violence is explained by Lory Heise's (1998) ecological theory approaching it from its multi causality following the vision of the ecological model of human development of Urie Bronfenbrenner (1977). This model considers human development as the result of the constant interaction between the human being who is in permanent activity and various changing environmental factors within a permanent process of accommodation (Bronfenbrenner, 1977). In fact, the ecological term derives from taking into account the human being within different contexts in their natural environment. Heise (1998), on the other hand, proposes the ecological theory as a heuristic tool to understand violence against women from levels that are self-contained, where the interaction of personal, social and cultural factors occurs, establishing predictive variables in each one (Heise, 1998). Likewise, the nested model allows us to take into account the relationships between the predictors of partner violence, through a continuum that goes from the social to the personal history (WHO, 2002), helping to make distinctions between these factors. Heise's ecological model was adapted to the Peruvian context (see figure 1) considering factors present in the exosystem, microsystem and personal history. However, factors related to the macrosystem were not considered because the variables studied at this level of the ecological model were not collected in the database analyzed from the ENDES Demographic Survey. In this sense, factors that are related to the likelihood of experiencing sexual violence were included (Puente-Martínez, Ubillos-Landa, Echeburúa, & Páez-Rovira, 2016; Devries et al., 2011, García-Moreno et al., 2006) (See Figure 1).
Figure 1. Classification of factors according to Heise's adapted ecological model (1998).
The exosystem covers social and community elements that encompass the microsystem (Heise, 1998). Studies suggest that belonging to low socioeconomic levels would increase the risk of experiencing sexual violence by the partner. The inverse relationship between intimate partner violence and the wealth quintile was reported by Dominican women (Bott, Guedes, Goodwin & Mendoza, 2013) and Mexican women (Casique, 2010). This association could be explained by considering poverty as a factor that fuels conflict in relationships (WHO, 2002; Postmus, Plummer, McMahon, Murshid & Kim, 2012) and as an element that favors acceptance (Casique, 2010). Nevertheless, other studies (Castillo, 2015; Abramsky et al., 2011) point out an ambiguity in the relationship between sexual violence and the belonging of women to a particular socioeconomic level.
On the other hand, microsystem factors typical of the context where violence occurs are connected to a higher probability of sexual violence. A predictor in Latin America is the frequency with which the partner gets drunk (Rey, 2017; Martínez, Landa, Echeburúa, & Rovira, 2016; Jaén, Rivera, Amorin, & Rivera, 2015; Casique, 2010), which happens also in Peru (Alarcón & Ortiz, 2017; Castro, Cerellino & Rivera, 2017; Blitchtein & Reyes, 2012; Flake, 2005). Around 50% of Peruvians experienced physical violence when their partner was drunk and 69.8% of them were from rural areas (INEI, 2017b). Likewise, the physical violence exerted by the partner has been reported as another predictor. Data from Nicaragua (Ellsberg, Winkvist, Peña & Stenlund, 2001), Uganda (Karamagi, Tumwine, Tylleskar & Heggenhougen, 2006), Spain (Rodriguez, Puig & Sobrino, 2014) and the United States (Krebs et al., 2011) corroborate it. In Peru (INEI, 2017b) almost 31% of the prevalence of physical violence against women showed that the most common types of aggressions were shoving or throwing objects (26%), being slapped (18.2%) or being hit with fists (15.3%). Regarding this association, the literature points out an overlap of types of violence (Krebs et al., 2011; Rodriguez, Puig & Sobrino, 2014) experiencing it within the same relationship, or by different partners (García-Moreno et al., 2006).
Receiving economic threats is also related to an increased risk of experiencing sexual violence (Postmus et al., 2011). The prevalence of economic violence, which may include threats, ranges from 48% in Chile (Barría & Macchiavello, 2012) to 29% in Mexico (Casique, 2010). In Peru (INEI, 2017b) 15.9% of women reported economic threats from their partner. These are used to try to isolate women to create dependence and maintain coercive control (Stark, 2007). Women who are forced into economic dependency are at greater risk of sexual violence (Postmus et al., 2011). In Peru, there is no detailed information on the prevalence of economic abuse.
The biological factors and experiences contributing to interpersonal relationships belong to the personal history level. Latin American studies indicate that a history of child abuse or having witnessed the father hitting the mother are important predictors of it (Rey, 2017; Martínez et al., 2016; Jaén et al., 2015; Casique, 2010). Findings in Peru (Alarcón & Ortiz, 2017; Barrios et al 2015; Blitchtein & Reyes, 2012; Flake, 2005) also confirm this by corroborating the cycle of family generational violence (Aldarondo, Kantor & Jasinski, 2002; Halford, Sanders & Behrens, 2000). In the same way, women's employment status as a risk factor has been reported (Postmus et al., 2011). In European countries, it has been pointed out that women in employment are more independent and have greater resources to end a violent relationship (Devries et al., 2011; García-Moreno et al., 2006). However, studies in Latin America indicate that employed women are at greater risk of experiencing violence than those who only perform work at home (Bott et al., 2013). This association has been confirmed by Peruvian studies (Flake, 2005; Castro, Cerellino & Rivera, 2017).
Intimate partner sexual violence is caused by the interaction of multiple factors that coexist at different levels (Heise, 1998; WHO, 2002). This interaction is simultaneous and reciprocal (Heise, 1998; WHO, 2002; Flake, 2005; Puente-Martínez et al., 2016). For the empirical research conducted from this perspective, it is insufficient to isolate the effect of risk factors, so, the main effects caused by their interaction must also be investigated. This vision raises two questions: what factors have the greatest explanatory power in sexual violence? And what group of factors remains as a relevant predictor?
Interactions between the factors present in the levels can modulate the predictive magnitude of intimate partner violence (WHO, 2002). However, most Peruvian studies on the subject do not examine the interaction within the ecological model, limiting themselves to determining the predictive effect of risk factors (Castro, Cerellino & Rivera, 2017; Alarcón & Ortiz, 2017; Barrios et al., 2015; Blitchtein & Reyes, 2012; Nóblega, 2012). Therefore, the action of the different ecological levels in establishing the predictive effect of the risk factors of recent sexual violence is still unclear. Considering these elements, it was proposed to evaluate the influence of the factors that intervene in the increase or decrease in the risk of recent sexual violence by the current partner in Peruvian women, considering the interaction between ecological levels of violence.
Method
Participants
A representative sample of 21. 414 Peruvians women aged 15 to 49 years was selected. Those who reported having a partner and responded to the violence module were included. The average age was 30.3 years. In relation to the residence geographical area, 10.88% of the respondents resided in the Lima metropolitan region, and 30.72% in the rest of the coastal region. Similarly, 32.51% lived in the sierra region and 25.89% inhabited the jungle region. Of the total, 59.4% were cohabiting with their partner, 56% had an independent job, 62.6% had a level of education equal to the partner's, 56.7% developed an independent job, 27.1% were in the lowest wealth quintile, while 9.7% were in the top quintile.
Design
The study was cross- sectional descriptive (Montero & León, 2007) with secondary analysis of the ENDES database (2017), which was representative at the national level and conducted by the National Institute of Statistics and Informatics (INEIa, for its Spanish acronym). Sampling was carried out in two stages by conglomerates (primary sampling units) composed of approximately 140 households and stratified by urban and rural areas (secondary sampling units) of Peru's 24 departments and two provinces (INEI, 2017a). The analysis units were women aged 15 to 49 years. Considering the complexity of the sampling to determine the estimation errors, several variables were used, which can be consulted in Appendix B of the ENDES report (INEI, 2017a) which indicates the estimator used and the reference population. Consider, as an example for calculating the sampling errors of the total population, the "currently in union" variable that is a proportion with a base population of women between the ages of 15 to 49 years. It yielded an estimated value of 0.566 (56.6%) with a standard error of 0.005 (0.5%) and a 95% confidence interval (CI: 0.555-0.577).
Instruments
The Demographic and Family Health Survey (ENDES) of 2017 in its section 10 consists of 48 questions regarding acts of physical and sexual violence throughout the life of the participant and in the last 12 months. The approximate response time for this section is 30 minutes. Likewise, it obtains information on the consequences of the episodes of violence and the subsequent attendance to health services. In the measurement of violence, the one-question threshold approach was used to determine, firstly, whether the respondent had experienced violence. The participants who answered affirmatively were asked about the characteristics of the violence (eg., frequency). The ENDES aimed to examine the prevalence of violence and its correlates, and follows the model and methodology of the Demographic and Health Surveys (Kishor & Kiersten, 2006). In this sense, in Peru, the questionnaire was applied for the first time in the year 2000. Subsequently, it has been used annually with a continuous survey design for a period of five years (2004-2008) and after three years (2009-2011, 2012-2014, 2015-2017). Likewise, the instrument reflects the comparative measurement methodology of the WHO Multi-Country Study on Women's Health and Domestic Violence (WHO, 2005), making it an adequate tool to measure the prevalence of violence and its contextual characteristics. Despite its repeated and widespread use, psychometric data are not available.
Nine questions from section 10 of the ENDES related to sexual partner violence were used. The main variable, intimate partner sexual violence, defined by WHO (2013) as forced sexual relations and other forms of sexual coercion within an intimate relationship was constructed from two questions: Has your (last) spouse (partner) ever used the physical strength to force you to have sex even if you did not want to? And In the last 12 months, did he force you to perform sexual acts that you do not approve? A Cronbach's alpha = 0.72 was obtained. It was categorized into yes and no. The remaining seven questions are detailed below following Heise's theoretical model (1998) in three groups of factors: personal history (individual), microsystem and exosystem. Macrosystem factors were not considered because the ENDES Demographic Survey did not collect variables classified at this level of the ecological model.
Predictive factors
Among the individual factors, the following were considered: age of the woman (in years); the marital status categorized as married, cohabiting, separated. The widowed and divorced categories were excluded from the inferential analysis since they constituted less than 1% of the study sample; type of work, categorized as dependent, independent; the history of physical violence from the father towards the mother, measured through the questions in section 10: Did your dad ever hit your mom? Since you were 15 years old, did any other person, apart from your spouse / partner ever hit, kicked, slapped or physically abused you? These questions were categorized into yes and no.
In the microsystem, the following factors were considered: physical violence by the current couple, that was measured through the question in section 10: In the last 12 months, has your (last) husband (partner) ever: shaken you, slapped you, hit you with fist, kicked, tried to strangle, threatened or assaulted you with a knife, gun or other type of weapon? A Cronbach's alpha = 0.81 was obtained and categorized into yes and no. The partner has threatened to take away the financial aid was categorized into a yes and no measure through the question: Has he threaten you to withdraw the economic aid? Likewise, communication with the current partner was measured through the responses about whether he is affectionate, spends his free time with her, consults her opinion on different household topics, and respects her wishes and rights. A Cronbach's alpha=0.77 was obtained, which was categorized into yes and no. Frequency of drunkenness was measured with the question: Does your spouse (partner) get drunk too often, sometimes or never? categorized into: Does not get drunk, sometimes gets drunk, gets drunk often. Equally, the difference with respect to the educational level, defined as the maximum level in the educational formation reached with respect to the man, was categorized as: Both with the same level, woman with higher level, man with higher level.
In the exosystem, the quintiles of wealth, defined in terms of wealth of the households surveyed were considered. Five quintiles were determined with respect to assets, instead of income or consumption. It was categorized into: Bottom quintile, second quintile, middle quintile, fourth quintile, top quintile (Enríquez-Canto, Y., Ortiz-Romaní, K., & Ortiz-Montalvo, Y., 2017). Tolerance to violence was measured through the answers to the question: In your opinion, do you agree that a man can beat his wife when she leaves without telling him anything, neglects the children, argues with him, refuses to have sexual intercourse or burns dinner? A Cronbach's alpha of 0.80 was obtained and it was categorized into yes and no.
Procedure
The ENDES' modules were applied in each of Peru's Departments between the months of February and December 2017, after contacting the women usually resident in the randomly selected dwellings, belonging to a conglomerate previously determined at random. To this end, a trained surveyor collected the information through face-to-face interviews conducted at the respondents' homes, assisted by a personal computer. At the beginning of the interview the informed consent to participation was read and verbal assent was obtained. The data collection was done according to the availability of the participants with instruction to interrupt the interview in the absence of privacy.
Ethical considerations
In order to maximize the security of the data collected and the non-disclosure of personal information, INEI conducted specialized training for interviewers, interviewing only one woman per household and maintaining complete privacy during the interview.
Data analysis
With the statistical program Stata SE 14, means and standard deviations of the quantitative variables were calculated first, as well as frequencies and percentages for the qualitative variables. Before carrying out the descriptive bivariate analysis using the Chi-square and Student t tests, the Kolmogorov-Smirnov test showed that the numerical variables were adjusted to a normal distribution in age (p>0.05) in the comparison groups. Subsequently, the homogeneity of variances was verified (Levene's test). Once these requirements were verified the Student t test was performed (for women's age). The tendency to collinearity was evaluated by the inflation factor of variance. In the multivariate analysis Poisson regression was used with the robust variance estimation calculating the prevalence ratio (PR) (Espelt A, Marí Dell'Olmo, Penelo & Bosque-Prous, 2017) and 95% confidence intervals, considering a level of statistical significance less than or equal to 0.05.
Four models were determined whose coefficients represent the increase or decrease in the probability of recent sexual violence associated with the change of a unit (or category) in an independent variable. The first model determines the probability of recent sexual violence considering the personal history factors. In the second, the effects of the microsystem factors were measured. The third model measured the probability of sexual violence considering the factors of the exosystem. The fourth model considered the factors of the three ecological levels to understand the effect of their interaction on sexual violence. Finally, the goodness of fit was estimated with the Akaike information criterion (AIC). The area under the curve (AUC) of each estimated model was calculated.
Results
Table 1 describes the factors connected to recent sexual violence in Peru. Of the sample, 6.3% suffered sexual violence by their partner in the last twelve months. 29.5% of the participants suffered physical violence, 43.6% had a history of violence from the father towards the mother, almost 15% were financially threatened, and 73.1% of their partners got drunk a few times.
Table 1. Recent sexual violence by current partner and sociodemographic data
Statistical association was reported between the women's marital status and sexual violence by their partner (p=0.001). Almost 20% of the separated women suffered sexual violence in the last twelve months. Likewise, there was an association between women's age and sexual violence (p=0.001). Women who experienced violence had a higher average age (M=34.80 SD: ± 8.08) than those who did not report it (M=32.14 DS: ±7.88). Similarly, there was an association between having experienced sexual violence and a history of violence by the father towards the mother (p=0.001): 8% of women who experienced sexual violence evidenced a history of violence from their fathers towards their mothers. There is a significant association between sexual violence and communication with the partner (p=0.001): 42.5% of women who reported sexual violence expressed an absence of communication with their partner. Likewise, there was an association between physical violence towards women and sexual violence by the partner (p=0.001): 19% of women who experienced physical violence also suffered sexual violence. There is an association between the frequency of drunkenness and recent intimate partner sexual violence (p=0.001): 32% of women who experienced sexual violence reported that their partner got drunk often. The rest of the factors studied can be seen in Table 2.
Table 2. Bivariate analysis between factors and recent sexual violence by the current partner
In relation to multiple regression, the first model estimated the probability ofrecent sexual violence considering personal factors (see Table 3). Age is positively associated with violence (PR=1.03 CI: 1.02-1.04) (p=0.001). Women with a history of violence from the father towards the mother are more likely to experience sexual violence by their partner (PR=1.53 CI: 1.36-1.72) (p=0.001) with respect to those who lack this background. On the other hand, those who indicated aggression on the part of the father are more at risk of recent intimate partner sexual violence (PR=1.61 CI: 1.32-1.96) (p = 0.001) than those who do not report it. In relation to the AIC, the model reported a value of 7637.
Table 3. Predictors of recent sexual violence by the current partner
The second model evaluated the relationships between the factors of the microsystem and the probability of experiencing violence (see Table 3). Partner's physical violence is the strongest predictor of the model. Women who have experienced physical violence by their partner are more likely to also suffer sexual violence (PR=12.8 IC: 10.15-16.24) (p = 0.001) than those who have not experienced it. Women who receive economic threats (PR=2.6 CI: 2.29-2.95) (p=0.001) are more likely to experience sexual violence than those who are not threatened. Also, the respondents who reported that their partner gets drunk often have twice the risk of recent sexual violence (PR=2.01 CI: 1.61-2.51) (p=0.001) with respect to those whose partner gets drunk only a few times. Women who report having communication with their partner are less likely to experience sexual violence than those who report absence of communication (PR=0.44 IC: 0.39-0.49) (p=0.001). In the AIC the model reported a value of 5656.
The third model considered the relationship between exosystem factors and sexual violence in the last twelve months. There is an inverse relationship between belonging to the quintile and violence. Women belonging to the upper quintile are less likely to suffer sexual violence (PR=0.44 CI: 0.34-0.56) (p=0.001) compared to those in the lower quintile. This model reported the highest AIC value 9957 (see Table 3).
On the fourth model the joint interaction of factors at each level of sexual violence was assessed. Figure 2 shows the interconnection between ecological levels where the physical violence of the couple remains a strong predictor followed by economic threats (see figure 2). The effect of family violence antecedents is attenuated: experiencing violence from father to mother decreases from 1.53 (p=0.001) to 1.19 (p=0.001). Likewise, the protective effect of the top wealth quintile decreases from 0.44 (p=0.001) to 0.59 (p=0.001). This model reported the lowest AIC value 4380.
Figure 2. Factors associated with recent sexual violence according to levels of the adapted Ecological model.
Note: Values are regression coefficients of Model 4 *p <0.05 **p <0.001.
Finally, the discriminant capacity of each model was estimated with the AUC. The results were: for model one, 0.61 (CI: 0.59-0.63); for model two, 0.89 (CI: 0.88-0.90); for model three, 0.58 (CI: 0.56-0.60); and for model four it was 0.90 (IC: 0.89-0.91).
Discussion
From the perspective of the ecological model, sexual violence by partners is explained by the interaction of multiple factors that coexist at different levels (Heise, 1998; WHO, 2002). Empirical research on violence should investigate the effects caused by the interaction of ecological levels. Taking this into account, the objective was to analyze the interaction between these levels to determine the increase or decrease in the risk of recent sexual violence by the current partner in Peruvian women.
Among the personal factors, it was found that women who experienced episodes of family violence are more likely to experience recent intimate partner sexual violence. The family violence background has been documented as a factor associated with the risk of sexual violence in Colombia (Rey, 2017, Martinez et al., 2016), Mexico (Jaén et al., 2015; Casique, 2010) and Peru (Alarcón & Ortiz, 2017; Blitchtein & Reyes, 2012; Flake, 2005). In the study, four out of every ten Peruvians experienced episodes of family physical violence. A history of violence in childhood is a predictor of violent behavior, which is consistent with the cycle of generational violence transmitted in families (Aldarondo, Kantor & Jasinski, 2002; Halford, Sanders & Behrens, 2000). From the theory of social learning (Bandura, 1977), episodes of violence between parents would reinforce the predisposition to violence in couple relationships, leading to a greater probability of experiencing it (Halford, Sanders & Behrens, 2000).
In the microsystem, the role of physical violence in recent intimate partner sexual violence constitutes a strong predictor of this behavior. Women who suffered physical violence from their partner are more likely to be sexually assaulted. International studies report similar findings in Spain (Rodriguez, Puig & Sobrino, 2014), the United States (Krebs et al., 2011), Uganda (Karamagi et al., 2006) and Nicaragua (Ellsberg et al., 2001). Despite cultural diversity, physical violence is often accompanied by sexual violence. In this regard, the literature points out an overlap of types of intimate partner violence (Krebs et al, 2011; Rodriguez, Puig & Sobrino, 2014). In fact, it is possible to experience different types of violence within the same relationship or throughout a woman's life and in different contexts (García-Moreno et al., 2006). In addition to the above, an episode of partner violence could involve a combination of threats, acts of physical aggression and sexual violence (Krebs et al., 2011). On the other hand, Kamaragi et al. (2006) point out that sexual violence can be part of the punishment that a partner inflicts on a woman, and add that when sometimes she refuses to having sexual activity, she is first beaten and then forced to having sex. This information is confirmed by the Latin American comparative study carried out by Bott et al. (2013).
Another interesting finding reveals that women who receive economic threats are more likely to experience sexual violence. Postmus et al. (2011) propose that the relationship between economic violence, which may include threats, and other forms of partner violence, has not been sufficiently explored empirically. According to Stark's (2007) theory of coercive control, an aggressor tries to establish positions of power through a variety of tactics such as threats and acts of sexual violence that allow to maintain them. In this way, women who are forced into economic dependency are at greater risk of suffering sexual violence and not abandoning the relationship. In Latin America, a Chilean study reported a 48% prevalence in the use of economic violence (Barría & Macchiavello, 2012), compared to 29.3% among Mexican women (Casique, 2010). On the other hand, the findings of the present study show that 15% of the sample experienced economic threats. These differences could be explained by the little space dedicated in the ENDES to questions that inquire about this type of violence.
The results confirm that women whose partners get drunk frequently, are more likely to experience recent intimate partner sexual violence. Other Peruvian studies report similar findings (Alarcón & Ortiz, 2017; Castro, Cerellino & Rivera, 2017; Blitchtein & Reyes, 2012; Flake, 2005), and also internationally (Devries et al 2014; Abramsky et al., 2011). In Peru, frequent consumption of alcohol is associated with the image of masculinity and the social role attributed to it (Castro, Cerellino & Rivera, 2017). In this regard, the study data reveal that seven out of ten Peruvians report that their partner gets drunk a few times. These data are worrisome since the consumption of alcoholic beverages has traditionally been related to violence and aggressive behaviors (Puente-Martínez et al., 2016). However, alcohol consumption in an episode of abuse does not necessarily mean that it is the cause of violence. According to Abramsky et al (2011) it should be considered that intimate partner violence frequently occurs without alcohol consumption. The role of alcohol in partner violence can be explained by the disinhibitory effect it has on behavior (Alarcón & Ortiz, 2017; Abramsky et al., 2011).
Among the protective factors, results indicate the importance of communication in the couple. Women who report communication with their partner are less likely to experience sexual violence. Another Peruvian study confirms the intuitiveness of this finding (Blitchtein & Reyes, 2012). It should be considered that sexual or physical violence could be part or result of the deterioration of the relationship, explained by Heru, Stuart & Recupero (2007), or as an outcome of the interaction of numerous factors, among which are communication skills. Therefore, clear and direct communication is seen as a strength of the couple (Heru, Stuart & Recupero, 2007) and as an important component in the non-violent resolution of problems (Espinoza, 2018, Devries et al., 2011).
On the other hand, women located in the upper wealth quintile are less likely to suffer sexual violence from their partner, being this a protective factor. In the Dominican Republic, an inverse relationship was reported between intimate partner violence and the wealth quintile to which the respondent belongs (Bott, Goodwin & Mendoza, 2013). However, Bott et al. (2013) highlight in their report on violence against women in Latin America and the Caribbean the existence of ambiguous results about this association, adding that the likelihood of violence is not always lower in the richest or most educated women. For a better understanding of these differences, the wealth quintile should be considered as an indicator of family well-being and not necessarily of women's economic autonomy. Therefore, it could have a lower predictive force with respect to their income. In this sense, the association between sexual violence by the partner and women's income would not be linear and would be mediated by other contextual factors (Ismayilova, 2015). For example, perhaps in Peru women's employment could represent a challenge to the cultural expectation that men should be the only family providers (Nóblega, 2012).
The statistical models show the interconnection of ecological levels where the effects of some factors decrease and others remain unchanged. Heise's ecological model (1998) proposes that the factors are reinforced and modified reciprocally (WHO, 2002). The predictive effect of the family violence background is attenuated in the joint model where levels overlap. This decrease could be due to the interaction with other factors of the micro-system such as the couple's communication and the exosystem as the wealth quintile. On the other hand, the factors of the microsystem do not modify its effect. What could explain this phenomenon is that these operate with some autonomy with respect to the other levels analyzed. However, the simultaneous action of the levels and their mutual involvement in the understanding of violence is postulated (WHO, 2002; Flake, 2005; Puente-Martínez et al., 2016). Nonetheless, considering the differences described, this action would not be equal in terms of the predictive weight of the factors in the ecological levels.
Finally, the analysis of the AUC suggests differences in the discriminant capacity of the models. That is, models two and four have a good discriminant capacity for recent sexual violence. An explanation is linked to the choice of predictive variables from Heise's ecological model (1998), which considers the interaction between personal, couple and social factors that impact it. In this sense, model four, which contains all the factors, is the one with the greatest discriminant capacity, supporting the thesis of Bronfenbrenner (1977) that considers the entire ecology of the individual.
Among the limitations, it was impossible to obtain information on other predictors of sexual partner violence because it is a secondary data analysis. For example, at the macrosystem level, variables such as the isolation of women and the family were not explored, which limits a complex consideration of the ecological model. There was only information from the woman about her partner. The recall bias may have led to an underestimation of the associations between recent sexual violence and past episodes of violence. However, the analysis highlights the fact that it had a representative national sample of Peruvian women and that it outlines a profile of victims of sexual violence. Finally, a robust analysis was carried out that allows us to consider the specificity of statistical models in intimate partner sexual violence.
In conclusion, from the ecological perspective, the personal factors that increase the risk of experiencing sexual violence are: having witnessed violence from the father towards the mother and having been a victim of physical violence by the father. In the microsystem, the factors are: physical violence, economic threats and the frequency of drunkenness by the partner. Among them, as a strong predictor, physical violence stands out. On the contrary, adequate communication and belonging to higher wealth quintiles act as protective factors against recent intimate partner sexual violence. Secondly, the interconnection between the ecological levels is evidenced through the change of the effect of the factors used in the models. The ecological approach heuristically explains violence through the simultaneous and reciprocal interaction between levels. However, it should be added that this may not be equal since there are differences in the predictive power of the factors.
The findings suggest favoring activities of primary prevention of sexual violence, focusing on actions at the microsystem level. Interventions are needed to strengthen the communication skills inside couples to enhance their protective influence. On the other hand, it is necessary to carry out primary prevention activities with couples who have several risk factors, for example: cohabitants, members of lower quintiles, residents of rural areas or those who have reported physical violence by the partner. Finally, more research has to be done on the overlap between economic violence and sexual violence. Further studies are necessary to understand the unequal interaction of ecological levels in order to explain in depth the multi-causal phenomenon of recent intimate partner sexual violence in the Peruvian context.
References
Abramsky, T., Watts, C. H., Garcia-Moreno, C., Devries, K., Kiss, L., ... Heise, L. (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, 11(1), 1471-2458. doi: https://doi.org/10.1186/1471-2458-11-109
Alarcón, L., & Ortiz, J. (2017). Qué factores se asocian a la violencia psicológica doméstica? Análisis del último decenio de la ENDES. Casus Revista de investigación y casos en salud, 2(1), 11-27. Recuperado de http://casus.ucss.edu.pe/index.php/casus/article/view/28/12
Aldarondo, E., Kantor, G. K., & Jasinski, J. L. (2002). A risk marker analysis of wife assault in Latino families. Violence Against Women, 8(4), 429-454. doi: https://doi.org/10.1177/107780120200800403
Bandura, A. (1977). Social Learning Theory. Englewood Cliffs: Prentice Hall.
Barría-Muñoz, J. R., & Macchiavello-Rodríguez, A. (2012). Anatomía de los hombres que ejercen violencia hacia sus parejas: primer levantamiento de datos para el diseño de un perfil. Psicologia.com [Internet], 16(19), 1-29. Recuperado de https://www.academia.edu/27419754/Anatomia_de_los_hombres_que_ejercen_violencia_hacia_sus_parejas
Barrios, Y. V., Gelaye, B., Zhong, Q., Nicolaidis, C., Rondon, M. B., Garcia, P. J., & Williams, M. A. (2015). Association of childhood physical and sexual abuse with intimate partner violence, poor general health and depressive symptoms among pregnant women. PloS one, 10(1), e0116609. doi: https://doi.org/10.1371/journal.pone.0116609
Blitchtein, D., & Reyes, E. (2012). Factores asociados a violencia física reciente de pareja hacia la mujer en el Perú, 2004-2007. Revista Peruana de Medicina Experimental y Salud Pública, 29(1), 35-43. Recuperado de http://www.scielo.org.pe/pdf/rins/v29n1/a06v29n1.pdf
Bott, S., Guedes, A., Goodwin, M., & Mendoza, J. A. (2013). Violencia contra la mujer en América Latina y el Caribe: análisis comparativos de datos de 12 países. Washington: OPS.
Bronfenbrenner, U. (1977). Toward an experimental ecology of human development. TheAmerican Psychologist, 32(7), 513-531. doi: http://dx.doi.org/10.1037/0003-066X.32.7.513
Casique, I. (2010). Factores de empoderamiento y protección de las mujeres contra la violencia. Revista mexicana de sociología, 72(1), 37-71. Recuperado de http://www.scielo.org.mx/pdf/rms/v72n1/v72n1a2.pdf
Castillo, M. (2011). Violencia de pareja en el Paraguay según la Encuesta Nacional de Demografía y Salud Sexual y Reproductiva 2008. Revista Latinoamericana de Población, 5(9), 27-48. Recuperado de https://www.redalyc.org/pdf/3238/323827305002.pdf
Castro, R. J., Cerellino, L. P., & Rivera, R. (2017). Risk factors of violence against women in Peru. Journal of family violence, 32(8), 807-815. doi: https://doi.org/10.1007/s10896-017-9929-0.
Devries, K., Watts, C., Yoshihama, M., Kiss, L., Schraiber, L. B., . Garcia-Moreno, C. (2011). Violence against women is strongly associated with suicide attempts: evidence from the WHO multi-country study on women's health and domestic violence against women. Social science & medicine, 73(1), 79-86. doi: https://doi.org/10.1016/j.socscimed.2011.05.006
Ellsberg, M. C., Winkvist, A., Peña, R., & Stenlund, H. (2001). Women's strategic responses to violence in Nicaragua. Journal of Epidemiology & Community Health, 55(8), 547-555. doi: http://dx.doi.org/10.1136/jech.55.8.547
Enríquez-Canto, Y., Ortiz-Romaní, K., & Ortiz-Montalvo, Y. (2017). Análisis de los determinantes próximos e impacto de la ocupación en la fertilidad de mujeres peruanas. Revista Panamericana de SaludPública, 41(1), 1-10. Recuperado de https://www.scielosp.org/scielo.php?pid=S1020-49892017000100208&script=sci_arttext&tlng=pt
Espelt A., Marí-Dell'Olmo M., Penelo E., Bosque-Prous M. (2017). Estimación de la Razón de Prevalencia con distintos modelos de Regresión: Ejemplo de un estudio internacional en investigación de las adicciones. Adicciones, 29(2), 105-112. doi: https://doi.org/10.20882/adicciones.823
Espinoza-Bravo, S. P. (2018). Violencia sexual en la pareja: una práctica invisibilizada (tesis de licenciatura), Universidad Academia de Humanismo Cristiano, Santiago de Chile, Chile. Recuperado de http://bibliotecadigital.academia.cl/handle/123456789/4601
Flake, D. F. (2005). Individual, family, and community risk markers for domestic violence in Peru. Violence against women, 11(3), 353-373. doi: https://doi.org/10.1177/1077801204272129
Garcia-Moreno, C., Jansen, H., Ellsberg, M., Heise, L., & Watts, C. (2006). Prevalence of intimate partner violence: findings from the WHO multi-country study on women's health anddomestic violence. Lancet, 568(9543), 1260-1269. doi: https://doi.org/10.1016/S0140-6736(06)69523-8
Halford, W. K., Sanders, M. R, & Behrens, B. C. (2000). Repeating the errors of our parents? Family-of-origin spouse violence and observed conflict management in engaged couples. Family Process, 59(2), 219-235. doi: https://doi.org/10.1111/j.1545-5300.2000.39206.x
Heise, L. L. (1998). Violence against women: An integrated, ecological framework. Violence Against Women, 4(3), 262-290. doi: https://doi.org/10.1177/1077801298004003002
Heru, A. M., Stuart, G. L., & Recupero, P. R. (2007). Family functioning in suicidal inpatients with intimate partner violence. Primary care companion to the Journal of clinical psychiatry, 9(6), 413-418. Recuperado de https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2139922/
Instituto Nacional de Estadística e Informática [INEGI]. (2017a). Encuesta Demográfica y de Salud familiar ENDES-2017. Nacional y Departamental. Recuperado de https://www.inei.gob.pe/media/MenuRecursivo/publicaciones_digitales/Est/Lib1525/index.html
Instituto Nacional de Estadística e Informática [INEGI]. (2017b). Perú: Indicadores de violencia familiar y sexual, 2000-2017. Recuperado de https://www.inei.gob.pe/media/MenuRecursivo/publicaciones_digitales/Est/Lib1465/libro.pdf
Ismayilova, L. (2015). Spousal violence in 5 transitional countries: A population-based multilevel analysis of individual and contextual factors. American journal of public health, 105(11), e12-e22. doi: https://doi.org/10.2105/AJPH.2015.302779
Jaén, C., Rivera, S., Amorin, E., & Rivera, L. (2015). Violencia de Pareja en Mujeres: Prevalencia y Factores Asociados. Acta de Investigación Psicológica, 5(3), 2224-2240. doi: http://dx.doi.org/10.1016/s2007-4719(16)30012-6
Karamagi, C. A., Tumwine, J. K., Tylleskar, T., & Heggenhougen, K. (2006). Intimate partner violence against women in eastern Uganda: implications for HIV prevention. BMC public health, 6(1), 284. doi: https://doi.org/10.1186/1471-2458-6-284
Kishor, S., & Kiersten, J. (2006). Perfil de la violencia doméstica: un estudio en varios países. Calverton: Macro International Inc. Recuperado de https://dhsprogram.com/pubs/pdf/OD38/OD38.pdf
Krebs, C., Breiding, M. J., Browne, A., & Warner, T. (2011). The association between different types of intimate partner violence experienced by women. Journal of Family Violence, 26(6), 487-500. doi: https://doi.org/10.1007/s10896-011-9383-3
Martínez, A., Landa, S., Echeburúa, E., & Rovira, D. (2016). Factores de riesgo asociados a la violencia sufrida por la mujer en la pareja: una revisión de meta- análisis y estudios recientes. Anales de Psicología, 52(1), 295-306. doi: http://dx.doi.org/10.6018/analesps.32.1.189161
Montero, I., & León, O. G. (2007). A guide for naming research studies in Psychology. International Journal of Clinical and Health Psychology, 7(3), 847-862. Recuperado de http://www.aepc.es/ijchp/GNEIP07_es.pdf
Nóblega, M. N. (2012). Risk and protective factors for physical and emotional intimate partner violence against women in a community of Lima, Peru. Journal of interpersonal violence, 27(18), 3644-3659. doi: https://doi.org/10.1177/0886260512447522
Organización Mundial de la Salud [OMS]. (2002). Informe mundial sobre la violencia y la salud. Washington, D. C.: OMS. Recuperado de https://www.who.int/violence_injury_prevention/violence/world_report/es/summary_es.pdf
Organización Mundial de la Salud [OMS]. (2005). Estudio multipaís de la OMS sobre salud de la mujer y la violencia doméstica: resumen del informe. Ginebra: OMS. Recuperado de http://www.who.int/iris/handle/10665/43390
Organización Mundial de la Salud [OMS]. (2012). Understanding and addressing violence against women: Intimate partner violence (Paper No. WHO/RHR/12.36). Ginebra: OMS. Recuperado de https://apps.who.int/iris/handle/10665/77432
Organización Mundial de la Salud [OMS]. (2013). Global and regional estimates of violence against women: prevalence and health effects of intimate partner violence and non-partner sexual violence. Washington, D. C.: OMS.
Postmus, J. L., Plummer, S. B., McMahon, S., Murshid, N. S., & Kim, M. S. (2012). Understanding economic abuse in the lives of survivors. Journal of Interpersonal Violence, 27(3), 411-430. doi: https://doi.org/10.1177/0886260511421669
Puente-Martínez, A., Ubillos-Landa, S., Echeburúa, E., & Páez-Rovira, D. (2016). Factores de riesgo asociados a la violencia sufrida por la mujer en la pareja: una revisión de meta-análisis y estudios recientes. Anales de psicología, 52(1), 295-306. doi: http://dx.doi.org/10.6018/analesps.32.1.189161
Rey-Anacona, C. A. (2017). Diferencias por sexo y variables asociadas con las agresiones sexuales en el noviazgo en universitarios. Psychologia. Avances de la disciplina, 11 (1), 25-37. Recuperado de https://www.redalyc.org/pdf/2972/297251403002.pdf
Rodriguez-Menés, J., Puig, D., & Sobrino, C. (2014). Poly- and distinct-victimization in histories of violence against women. Journal of Family Violence, 29(8), 849-858. doi: http://dx.doi.org/10.1007/s10896-014-9638-x
Stark, E. (2007). Coercive control: How men entrap women in personal life. Nueva York, EE. UU.: Oxford University Press.