Women's disengagement from legal proceedings for intimate partner violence: Sociodemographic and psychological variables

We conducted interviews with an initial sample of 806 women who voluntarily accepted to participate in the study. Only 2% of the women we contacted refused to participate. We analyzed data from a final sample of 345 women, since most of the legal processes had not finished by the end of this study. Their age ranged from 17 to 72 ( M = 37.28, SD = 11.07) and they were all involved in ongoing legal proceedings for IPV in Andalusia, southern Spain. Of the 345 women, 153 were users of the Andalusian Victims Assistance Service (SAVA in its Spanish acronym) of the IPV Courts in Seville and 64 in Granada. This is a free public service belonging to the Justice Department of the Andalusian Regional Government. SAVA professional teams consist of a psychologist, a social worker, and a lawyer. These professionals carry out their work in the courts where they accompany different kinds of victims, including victims of violence against women, during the legal process. They do not provide systematic and intensive psychological treatment, just legal and emotional support during court proceedings. The remaining 128 women were users of Sheltered Housing ( n = 16), Municipal Centers for Information for Women (CMIM) in the province of Seville ( n = 89), and victims help foundations, Genus ( n = 11), Anabella (n = 3), and Sevilla Acoge (n = 9). We collected data from users of non-SAVA services in order to have data from women who did not have assessment and accompaniment in the legal process, and to include finalized proceedings which were less likely in the SAVA services where most of the cases had just started and would not have finished by the end of the study. In turn, the provinces of Seville and Granada were chosen in an attempt to include the significant variability in the ratio of disengagement between different provinces in Andalusia ( Consejería de Igualdad, Salud y Políticas Sociales, 2014 ), with the highest ratio in Seville (21%) and the lowest in Granada (1%).

The 345 cases were classified in two groups depending on whether they had disengaged or not from the judicial process. In 214 of the cases analyzed (62%), the subjects continued with the judicial process until the judge's ruling and in 131 (38%) they dropped charges after commencing proceedings. This last percentage is not at all representative of the level of disengagement occuring in Spain. We included a higher proportion of cases of disengagement to balance the percentages for the comparison between the two groups of women.

Data was collected using an extensive questionnaire drawn up on the basis of a preliminary study ( Cala et al., 2012 ) that included an exhaustive revision of the scientific literature on the topic and interviews with a wide range of victims and professionals from the judicial, police, and women's support areas about why they thought women withdrew from the judicial procedure. Data collection stage of this previous study concluded due to information saturation. To improve the content validity of the questionnaire, all the questions mentioned by the bibliography and/or the interviewed experts as a possible cause were included as a different question (collinearity among questions were considered in the statistical analysis). Then the questionnaire was reevaluated by a group of professionals and proved successively with different women until no items generated questions.

The Appendix includes the basic format of the questions analyzed in this study, taken from the complete questionnaire. However, those questions had to be adapted for most of the women, 317 out of 345 (91.90%), who were answering questions about a finalized legal process. Thus, we also wrote a retrospective version of the basic questionnaire, putting the questions into the past tense.

The Appendix shows that some of the questions from the questionnaire were open and required subsequent recoding. The country of origin was coded into categories: Spain, other Spanish-speaking countries and neighbors (Brazil), or non-Spanish speaking countries. Whether or not they received psychological support and the type of service which had given them the support was coded into categories: without support ( n = 169), public health services ( n = 29), public social services (SAVA, n = 10), or other social services that could be accessed freely regardless of the judicial process ( n = 116). Some data from women with private psychological support ( n = 2) or supported by an association ( n = 7) were excluded from the analysis. Finally, to analyze all public social services as a unique category, we checked beforehand that there were no significant differences between the percentage of disengagement between users of SAVA and other public social services, χ 2(1, N = 126) = 0.58, p = .448, ? = .07.

Some answers in which there was no agreement in the coding or where it was difficult to read the response were omitted, thus the level of inter-observer agreement among the final analyzed responses was 100%. We also merged categories of the educational level due to the low frequency of some values. The five categories were therefore reduced to just three: without studies ( n = 86), with compulsory studies or vocational training ( n = 211), and baccalaureate or university studies ( n = 47).

The data was analyzed using the PASW Statistics 18 package.

Before starting the research, the project was submitted for approval to the Consejería de Igualdad y Bienestar Social [Regional Department for Equality and Social Welfare] of the Andalusian Regional Government, which financed the project together with the Fundación de Investigación de la Universidad de Sevilla [Research Foundation of the University of Seville]. The authors confirmed to both institutions that the project complied with ethical regulations. Participants voluntarily responded to the questionnaire and were previously informed that all data would remain confidential. They were also told that they did not have to answer all the questions.

Three people with experience assisting women immersed in judicial processes helped with the application of the questionnaires; they worked on a regular basis for the Andalusian Victims Assistance Service (SAVA). In our study, the SAVA professionals were all asked to try and obtain information from women who had finished the judicial process, either because they had testified in the trial or because they had disengaged from the process prematurely. The reason for this was to have a large enough number of women who could be classified as having abandoned or not. In addition to collecting data from the service users during the study, they also contacted previous users who had finished the procedure and they were given the retrospective version of the questionnaire.

To make the interview process easier for these women, they were interviewed by SAVA staff looking after them as part of the victims’ care process. The questions were not always given in the same order; this depended on the stage they were at in the care process. When the victim care process was over, the women were asked more directly about the questions which had not been answered, particularly those items which should have scored between 0 and 10 (see Appendix).

The questionnaires applied in the other institutions and foundations taking part in the study were retrospective. In this case, the institutions’ staff administered the questionnaires following the same instructions given by the research team to the personnel of the Andalusian Victims’ Assistance Service.

First of all, we analyzed independently the relation between each of the variables included in the questionnaire and whether or not they had disengaged from the process. For the quantitative variables we used the ANOVA or Welch's F , depending on the assumption of homoscedasticity contrasted using Levene's F test. The effect size index calculated as a complement to these tests was R 2 partial . For the qualitative variables we used Pearson's chi-square test, together with the ? contingency coefficient as effect size index. All the significance tests used a level of significance of .05 and the effect size indexes were evaluated according to the conventional levels proposed by Cohen (1988) as small (R 2 = .01, ? = .10), medium ( R 2 = .06, ? = .30) or large ( R 2 = .14, ? = .50). Subsequently, all those variables whose relation with disengagement turned out to be statistically significant were introduced into a binary logistic regression model. Regarding collinearity, the lowest tolerance index was 0.61, and the highest VIF was 1.71. The step-forward method was used based on the likelihood ratio to reduce the number of predictor variables.

Table 1 shows the statistical data corresponding to the F or chi-square tests for the sociodemographic and psychosocial variables analyzed. As one can see, the relation between the country of origin and disengagement was statistically significant, albeit with a small effect size. The study of standardized residuals revealed that the greatest contribution to this significance was due to the high percentage, 60.7%, of women abandoning who were from non-Spanish speaking countries (Africa, Eastern Europe, etc.). The lowest percentage of disengagement occurred in the group of Spanish women, 34.8%, with this percentage being 46.7% in the group of women from other Spanish-speaking countries or neighbors (Brazil).

The relation between disengagement and the educational level was also significant, although again without reaching a medium effect size. The greatest standardized residual corresponded to women without studies, who abandoned in a significantly higher percentage than expected, 55.8%, as opposed to the 31.3% of the women with compulsory education or vocational training and 34% of those with baccalaureate or university studies.

In contrast, a statistically significant relation was not found between disengagement and the rest of the sociodemographic variables studied: number of children, woman's age, woman's monthly income, and living in a rural or urban setting.

Table 1 shows that the differences between the three levels of psychological support (without support, public health services, or social services) were statistically significant and had a medium effect size. The study of the standardized residuals which were higher than those expected for a contingency table with 6 boxes ( Z = 2.64, p = .05/6 = .008) revealed a lower percentage of disengagement from the judicial process amongst women who were receiving psychological support from a social service, 19%. In contrast, that percentage increased to 55.5% amongst those who were not receiving any support.

Significant differences were also found in the perceived level of support from relatives and friends between the group of women who disengaged from the process ( M = 6.04, SD = 3.96) and those who did not ( M = 7.21, SD = 3.21), although in this case the effect size was small. In contrast, the time that they had been receiving psychological support, counted in days, was not statistically related to disengaging from the legal process.

Table 2 shows the results of the statistical analysis of the emotional and motivational variables. As you can see, the mean differences in terms of the perception of fear, of his reaction to her and towards their children and the feeling of her life being in danger were statistically significant, although the effect size only reached the mean level in the last case. More specifically, the fear of his reaction towards her was lower amongst those women who gave up the legal process ( M = 7.58, SD = 3.40 as opposed to M = 8.70, SD = 2.40). Fear of his reaction towards their children was also lower amongst those women who abandoned ( M = 4.83, SD = 4.43 as opposed to M = 6.18, SD = 4.14). Similarly, the feeling that their lives were in danger was lower amongst women who gave up the legal process ( M = 5.87, SD = 3.86 as opposed to M = 7.56, SD = 2.94). In contrast, differences were not found in terms of concern about the lack of money or work.

As for concern about the possible incarceration of the aggressor, we found a statistically significant relationship between disengagement and concern about the possibility of him going to prison, with a medium effect size. More specifically, women who disengaged were more concerned about the possibility of him going to prison ( M = 7.00, SD = 4.00) than those who did not ( M = 4.32, SD = 4.14). Significant differences were also found with an effect size near the medium level in relation to the feeling of guilt for what might happen to him, with this feeling being relatively high amongst women who disengaged ( M = 6.71, SD = 3.83 as opposed to M = 3.62, SD = 3.80).

Finally, contact with the aggressor after the charges had been made (none, occasional, or frequent) and having thought about going back with him also had a statistically significant relation to disengagement, with an effect size nearing the large level. In relation to contact, the study of the higher than expected standardized residuals for a contingency table with 6 boxes ( Z = 2.64, p = .05/6 = .008) revealed a lower percentage of disengagement among women who did not have contact with their aggressor, 12.1%, while this percentage jumped to 74.4% amongst women who maintained frequent contact with him. We also found a significantly higher percentage of disengagement when the women had thought about going back with him, 73.7%, than when they had not 20.2%.

We studied the women's responses about their reasons for having pressed charges: manage to incarcerate him, get him to stop abusing her, give him a scare, get protection and separate from him, and we found a statistically significant relation between disengagement and each of the reasons analyzed (see Table 2 ). Thus, women were less likely to disengage from the judicial process when their motivation was to manage to incarcerate him, 16.7% compared to 39.1% for those without this motivation; get him to stop abusing them, 29.7% compared to 54.1% for those who did not include this reason; to separate from him, 28% as opposed to 45% for those who did not want to separate from him; or achieve protection, 26.2%, as opposed to 56.2% in women who did not include this reason. However, disengagement was more frequent among women whose reason for pressing charges was to give the aggressor a scare, 63.1% as opposed to 29.2% for those not giving this reason. The effect size index only came near the medium level of .30 in these two last cases: get protection and give him a scare.

Finally, all the variables whose relation to disengagement had been statistically significant and had given rise to at least a medium effect size were entered into a binary logistic regression model: the psychological support received, contact with the aggressor, thinking about going back with him, concern about the possibility of him going to prison, guilt, pressing charges to get protection and pressing charges to give him a scare.

Table 3 shows the results of this binary logistic regression analysis with the stepwise forward likelihood ratio method and orthogonal contrasts for social support and contact with the aggressor after pressing charges. As you can see, the resulting likelihood ratio model showed a statistically significant drop in deviation in relation to the observed data, and the Hosmer-Lemeshow statistic did not reveal significant differences between the observations and the predictions of the model. The variables selected for the prediction were the psychological support received, contact with the aggressor, thinking about going back with him and feelings of guilt. Other stepwise forward methods for the introduction of the variables (Wald and conditional) gave rise to the same solution as the likelihood ratio method.

As you can see in the odds ratios (OR ) column, the fact of not having received psychological support multiplied by 3.37 the probability of disengaging, and having frequent contact with the aggressor multiplied it by 4.34. The probability of abandoning was also 4 times greater if they thought about going back with him, and this increased as the feelings of guilt increased. Hence, an increase of just two points on the scale of guilt would multiply the probability of disengaging from the judicial process by more than 10 (e(2*1.17)=10.38).

Finally, classifying cases with probabilities over .30 as disengagement, the model only gave 20.5% of false positives (specificity) and 18.4% of false negatives (sensitivity). Overall, 80.2% of the cases were correctly classified.

The educational level and the country of origin were the only sociodemographic variables to produce significant results. However, as the effect size was small, they were not included in the regression analysis. Language fluency is probably an important factor behind the country of origin variable and this could merit specific study in future research. In turn, educational level is a variable which encompasses other factors such as economic status or the quality of social networks, and is strongly related to all psychosocial processes and the state of health. It is therefore interesting to see its small effect size, although it is the largest within the sociodemographic variables. The lack of significance of the monthly income variable is also important, because financial independence has been put forward as one of the factors which help victims break away from the aggressor ( Anderson & Saunders, 2003; Barnett, 2000 ). In this sense, we should highlight that the mean incomes in our sample were very homogeneous and low (72% of the cases were under 500€). While the literature provides significant but contradictory findings about the influence of children, our analysis did not come up with significant results. We believe that this variable should be understood in relation to another variable which was not covered in our study design: the type of aggression or the seriousness of the offense ( Cala et al., 2012 ). If the seriousness of the offense has endangered people's lives, the existence of children would mean that women were more likely to report the offense or continue with the judicial process to protect them. If the seriousness of the offense is low, some women might consider that it is not worth subjecting their children to a particularly stressful and uncertain legal scenario and decide not to report the offense in the first place or cut short the judicial process if it has already started. In short, from the analysis of the sociodemographic variables we cannot confirm the existence of a clear sociological profile to describe women who abandon the judicial process. While certain sociodemographic factors such as educational level may exercise some influence, the absence of a medium-high effect size seems to indicate that the phenomenon of IPV against women is highly transversal to all social classes.

It seems logical that the women who have greater contact with the aggressor and those who think about going back to him are more likely to cut short the legal process. These are the other two variables with high odds ratios, particularly the second, which are included in the final regression model. However, we are still faced with the question of why a woman would report her aggressor if she was thinking about carrying on her relation with him. As you can see in the results section, an important group of women report their partners to give him a scare. This type of expectation has already been found in the literature ( Ford, 1991; Hare, 2006; Hoyle & Sanders, 2000 ). Our results show that women's expectations are factors which help explain their behavior during the judicial process.

If we turn our attention to psychosocial factors, one variable which appears in the final regression model stands out: psychological support. Women who do not receive psychological support are 3.37 times more likely to cut short the judicial process. This figure corroborates the considerations of other studies which have pointed to the need for a coordinated psychosocial response to the issue of IPV ( Murphy, Musser & Maton, 1998; Wright & Johnson, 2009 ).

With regard to the importance of social and psychological support, it is worth noting that for offenders who perceived high levels of social support, recidivism is not reduced after an intervention program ( Lila, Oliver, Galiana, & Gracia, 2013 ). This may happen because offenders’ networks have attitudes of tolerance and acceptance towards violence. In contrast, victims’ social networks may offer low levels of support. This makes the psychological support from institutions all the more essential.

In other words, psychological requirements have to be met as well as legal ones. Our study has also shown that the support given by the social services is related to less disengagement. This coincides with the results of Dawson and Dinovitzer (2001) , who found that women who contact the social services are more likely to go through with the entire judicial process. This result would seem to indicate different approaches to working with women who have suffered IPV and emphasizes the importance of how we design our support services and which types of treatment are most effective.

All of the emotional variables turned out to be significant, although only guilt, contact with the aggressor, and thinking about going back with him had a medium-large effect size and appear in the final regression model. The emotion of guilt is included in the equation which best explains disengagement from legal procedures as it multiplies the likelihood of abandoning by 1.17. We would like to comment on two aspects related to emotions: the fact that the women who disengaged from the process usually felt less fear, and the power of the emotion of guilt to explain disengagement from the judicial process.

The literature corroborates the increase in anxiety amongst women who pursue the judicial process to the end. Thus, for example, 20% of the sample studied by Goodman, Bennett, and Dutton (1999) said they had been attacked or threatened by their ex-partners three months after the start of the process. However this is not always the case. Other studies ( Hare, 2006; Hoyle & Sanders, 2000; Weisz, 2002 ) have shown that many of the women who give up the judicial process understate the abuse received and the situation of danger in which they find themselves. They consider that the crime committed by their partners does not justify the potential sentence, especially if it means prison. Hoyle and Sanders (2000) found that the reason most widely reported for giving up the judicial process was that they did not want incarceration, but just to be left alone or for their aggressors to undergo rehabilitation. Furthermore, ambivalence towards the possibility of abusers going to prison may be particularly relevant in women from ethnic groups which look at the legal system and the police as sources of racism and discrimination ( Goodman et al., 1999a, 1999b ). Hence, women may feel guilty about handing their partners over to the system and feel they are betraying their community, with the danger of a knock-on effect leaving them even more isolated ( Gillis et al., 2006).

Similarly to what other studies have affirmed ( González-Méndez & Santana-Hernández, 2014 ), in our opinion fear or risk perception may act as a protective factor against violence by reducing contact with the aggressor. This way, although the decision of not continuing legal proceedings may be linked with a higher risk of suffering violence in the short term ( Goodman et al., 1999a, 1999b ), in the medium and long term the reduction of contact with the aggressor and his social context may lower the probability of suffering violence.

Guilt is defined in reference to the norms and rules we have internalized and which form part of our basic beliefs and give meaning to our existence. For that reason it is particularly relevant that guilt, being a second order emotion, explains the decision to cut short the judicial process more effectively than fear. In our opinion, guilt lies at the ideological core of the decision to disengage from the judicial process. Guilt can mediate the instinctive reaction of fear, escape, with such force that it overcomes it in effect size. In other words, the basic reaction of escape when a woman is attacked could be put on hold when it interacts with the beliefs and norms which have been internalized by some women.

It is the socialization of gender, the “perfect-love discourse” ( Towns & Adams, 2000 ), and the guilt associated to the non-compliance with their mandates which may make it harder for many IPV victims to terminate the relationship with their aggressor ( Barnett, 2001 ). But what gender mandates are women disobeying to explain the appearance of this guilt? We should not forget that in the process of gender socialization there is a tendency for some women to internalize certain beliefs, including ideas about love (give it everything you can, endure suffering, and make it last forever), and these lead some women to do everything within their grasp to keep the relationship going ( Barnett, 2001 ). Embarking on legal proceedings often coincides with the termination of the relationship and, in any event, means reporting someone with whom they have had an affective relationship and someone who may be the father of their children. This course of action is something which is not expected of women. In many cases, their partner relationship is expected to be the pillar of their life project, and they are supposed to surrender themselves to this in a selfless and unconditional manner ( Alberdi, 2005 ). This all means that “when women stop responding to the cultural mandates which hang over them they tend to see themselves as transgressors of the established norms and, as a result, they feel bad and to blame” ( Coria, 2005 , p. 75). This may be influenced by the prevalence in Spain of victim-guilt attitudes regarding partner violence against women ( Gracia & Tomás, 2014).

The importance of women's basic beliefs tells us that to reach a stable decision many IPV victims need to discuss and question these beliefs and reconstruct their identity in a different way. This requires significant cognitive and emotional effort and may clash with their social context. They will also need to confront the daunting task of planning the future on their own. This all helps to explain why a reflex action of escape from a serious attack, driven by the atavistic emotion of fear, is often not enough to sustain the legal process.

To understand our results, we need to consider the two sources of stress to which the women are exposed. On the one hand, the context of the abuser and the woman's life with him; on the other hand, the context of the judicial process, the definitive rupture with her previous life, and the uncertainty lying ahead. Some authors have referred to this type of stress as role restructuring stress ( Pearlin, 1989 ). Furthermore, they may perceive, and with good reason, that their lives are under greater threat from their ex-partner after the decision to continue with the judicial procedure.

Our results indicate that the feelings of vulnerability and fear are significantly higher in women who decided to continue with legal proceedings. Therefore, although our study is correlational, one important reading from our results is that the decision to continue with the judicial process may lead to an increment in the level of stress or fatigue endured by these women. This fear requires support and assistance from all the professionals involved if we do not want to see the process unravel. Obviously, this reading from our results is debatable. At the same time, it is possible that women suffering more severe situations of violence and feeling more negative emotions will be more likely to continue. However, two arguments support the plausibility of our interpretation. First, as we have described above, some studies have indicated that women become more vulnerable after the start of the legal process ( Goodman et al., 1999a,b ). Second, we have detected that victims who report their partner as the result of a traumatic event are more likely to disengage from legal proceedings. Our findings indicate that women who have enjoyed a period of reflection and psychological support in a relatively stable emotional state are more likely to go through with legal proceedings.