Full Terms & Conditions of access and use can be found at
https://www.tandfonline.com/action/journalInformation?journalCode=wwcj20
Women & Criminal Justice
ISSN: (Print) (Online) Journal homepage: https://www.tandfonline.com/loi/wwcj20
Intimate Partner Violence against Women in the
EU: A Multilevel Analysis of the Contextual and
Individual Impact on Public Perceptions
Matilda Karlsson , Maria Wemrell , Juan Merlo & Anna-Karin Ivert
To cite this article: Matilda Karlsson , Maria Wemrell , Juan Merlo & Anna-Karin Ivert (2020): Intimate Partner Violence against Women in the EU: A Multilevel Analysis of the Contextual and Individual Impact on Public Perceptions, Women & Criminal Justice, DOI: 10.1080/08974454.2020.1835792To link to this article: https://doi.org/10.1080/08974454.2020.1835792
© 2020 The Author(s). Published with license by Taylor and Francis Group, LLC.
Published online: 09 Nov 2020.
Submit your article to this journal
Article views: 91
View related articles
Intimate Partner Violence against Women in the EU: A
Multilevel Analysis of the Contextual and Individual Impact on
Public Perceptions
Matilda Karlssona, Maria Wemrella,b, Juan Merloa, and Anna-Karin Iverta,c
a
Unit for Social Epidemiology, Faculty of Medicine, Lund University, Malm€o, Sweden;bDepartment of Gender Studies, Faculty of Social Sciences, Lund University, Lund, Sweden;cDepartment of Criminology, Faculty of Health and Society, Malm€o University, Malm€o, Sweden
ABSTRACT
Intimate partner violence against women (IPVAW) poses severe threats to women’s health and rights. This study investigates the role of country con-text and gender equality in shaping individual perceptions of the severity of IPVAW. Multilevel logistic regression analyses of a Eurobarometer survey on attitudes toward IPVAW from 27 EU states showed that male gender, young age, low education, low self-assessed social position and particularly perceiving IPVAW as uncommon were associated with perceiving IPVAW as less severe. The likelihood of perceiving IPVAW as less severe was higher in countries with low gender equality. Between-country variance accounted for 14% of the variability, while country-level gender equality accounted for 22% of the between-country variance. We conclude that efforts toward strengthening perceptions of IPVAW as a severe issue should focus on awareness-raising and on increasing country-level gender equality.
KEYWORDS
European Union; gender-based violence; gender equality; intimate partner violence against women; multilevel analysis; public perceptions
INTRODUCTION
Violence against women is a global phenomenon and a concern for public health, human rights and social policy (Devries et al., 2013; Krantz, 2002). Intimate partner violence against women (IPVAW) is the most common type of violence against women and puts women at risk of severe direct injury as well as long term illness (Campbell, 2002; World Health Orgainization, 2010). It constitutes a serious threat to women’s security, human rights and equal participation in society. Despite measures taken by the European Union (EU) to combat IPVAW (European Commission, 2010), one in five women in the EU still experience IPVAW during their lifetime (FRA,2014).
As noted by Gracia and Lila (2015, p. 13), IPVAW “is a complex phenomenon that needs to
be understood within the wider social context and within the social and cultural norms that per-meate it. Public attitudes and responses regarding violence against women reflect these norms and play an important role in shaping the social climate in which the violence occurs.”
Public beliefs and attitudes form an important part of the social context of IPVAW (Copp et al.,2019; Martın-Fernandez et al.,2018; Waltermaurer,2012) and can influence IPVAW preva-lence, help-seeking behavior among victims and responses from judicial systems and support organizations as well as ambient communities (Gracia et al.,2009, 2014; Gracia & Herrero, 2006; CONTACTAnna-Karin Ivert anna-karin.ivert@mau.se Department of Criminology, Faculty of Health and Society, Malmo University, Malmo 205 06, Sweden.
ß 2020 The Author(s). Published with license by Taylor and Francis Group, LLC.
This is an Open Access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivatives License (http:// creativecommons.org/licenses/by-nc-nd/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited, and is not altered, transformed, or built upon in any way.
Rizo & Macy,2011; West & Wandrei,2002). Notably, victim-blaming attitudes can reflect public tolerance toward IPVAW and have often been used to explain or justify IPVAW (Boethius,2015; Gottzen & Korkmaz,2013; Gracia,2014; WHO,2002). Previous research has also shown that per-ceived severity of IPVAW is associated with the willingness to intervene when gaining awareness of a case of IPVAW (Gracia et al., 2018). How incidents of IPVAW are perceived in society is thus important for future prevention and policy, as well as for available support services or sanc-tions (Gracia & Herrero,2006; Waltermaurer,2012).
Evidence from previous research related to IPVAW suggests that prevalence of and attitudes toward IPVAW seem to be influenced by contextual-level factors (Gracia et al., 2015; Heise & Kotsadam, 2015; Herrero et al., 2017; Kovacs, 2018; Sanz-Barbero et al., 2018). However, the prevalence of IPVAW does not cut equally across alla sociodemographic groups within countries
(Kovacs, 2018). For example, research on differences in IPVAW prevalence between
socioeco-nomic and demographic groups within the country of Sweden has not always shown clear and consistent results (Lundgren et al.,2002; Nybergh et al., 2013). This indicate that individual-level characteristics and contextual factors work together toward the persistence of IPVAW, and pos-sibly also the attitudes justifying it. Such an interplay between factors at different levels in the causation of IPVAW has previously been described in what has been called the integrated eco-logical approach (Heise,1998).
In the international research, Individual-level socioeconomic status has proven to be correlated with both the prevalence of IPVAW (Herrero et al., 2017; Lauritsen & Schaum, 2004; Sanz-Barbero et al., 2018) and with the attitudes that justify it (Gracia & Herrero, 2006; Ivert et al., 2018; Stickley et al., 2008; Tran et al., 2016; Waltermaurer, 2012; Waltermaurer et al., 2013). Previous research has also shown that education is an important individual level predictor for
IPVAW (Herrero et al., 2017; Sanz-Barbero et al., 2018) as well as attitudes toward IPVAW
(Gracia & Herrero, 2006; Ivert et al.,2018; Stickley et al.,2008; Tran et al., 2016; Waltermaurer, 2012; Waltermaurer et al., 2013). Furthermore, research has also shown that the type of residen-tial area (e.g. urban vs rural) correlates to attitudes that justify IPVAW (Gracia & Tomas, 2014; Ivert et al., 2018; Waltermaurer, 2012; Waltermaurer et al., 2013). However, according to a sys-tematic review by Gracia et al. (2020) gender was the most frequently reported factor correlating with attitudes. They found that, broadly speaking, the results were consistent in showing that males in general tend to accept, justify and perceive IPVAW as less severe than women. Regarding noted inconsistencies in associations between age and attitudes, Gracia et al. (2020) suggest that this could be the result of the varying methods for sampling, measuring and analyz-ing these parameters. Moreover, country of residence does seem to provide a relevant contextual lens for understanding these types of attitudes (Ivert et al.,2018; Uthman et al.,2010). Literature investigating contextual-level factors is, however, scarce. This is particularly the case when it comes to the EU (Ivert et al., 2018). It remains important to look not only at the fact that con-textual factors seem to matter, but also to what extent they are useful in predicting individual attitudes (Merlo et al.,2017).
Feminist theories consider IPVAW to be a result of unequal gender norms and patriarchal structures in society (Lawson, 2012), meaning that gender inequality should be considered as a central and fundamental driving force behind gender-based violence such as IPVAW. Promoting
increased gender equality is often put forward as a way of reducing IPVAW (Garcıa-Moreno
et al.,2015; Jewkes, 2002; Uthman et al., 2010). Several studies resonate with feminist theories in that gender equality is negatively related to IPVAW and attitudes toward IPVAW (Archer, 2006; Heise & Kotsadam,2015; Sanz-Barbero et al., 2018; Uthman et al.,2010). Still, some counterintui-tive results have shown that the EU states with the highest levels of gender equality, such as the Nordic countries, have disproportionally high prevalence of IPVAW. This phenomenon has been referred to as the Nordic Paradox (Gracia & Merlo, 2016). Moreover, victim-blaming attitudes have been found to be higher in Nordic countries compared to other EU countries with lower
levels of gender equality (Gracia & Tomas, 2014). Thus, findings indicating that higher levels of gender equality are associated with higher rates of IPVAW contradict theories postulating that IPVAW is a result of gender inequality. Several hypotheses and theories attempt to explain this relationship. One explanation discussed in the literature is that of issues related to reporting (Gracia & Merlo, 2016). Another frequent explanation, often discussed in rape research, is that increased gender equality or demands for gender equality can result in a backlash against women, and thus in increased rates of violence against women (Whaley,2001; Wemrell et al.,2020). This so-called backlash hypothesis has proven to be useful when analyzing changes in IPVAW rates over time in certain contexts (Xie et al.,2012). Nonetheless, regardless of the reasons for the rela-tionship between the prevalence of IPVAW and gender equality, the issue remains that the preva-lence of attitudes that justify IPVAW is also high in countries with high levels of gender equality. To better understand these relationships, further research on the public’s attitudes toward IPVAW is needed.
The perceived severity of IPVAW among the public is an under-researched subject. Therefore, this study aims to investigate the effect of country context and gender equality on the perceived severity of IPVAW. Moreover, it aims to evaluate the ability of contextual and individual level factors to accurately discriminate between someone who perceives IPVAW as severe and someone who does not. The following three research questions were posed:
1. To what extent does the country of residence, in addition to individual sociodemographic
characteristics, affect individuals’ perceptions of the severity of IPVAW?
2. How accurately can individual sociodemographic characteristics and country context discrim-inate between those who find IPVAW to be severe and those who do not?
3. To what extent can between-country variance in perceptions of the severity of IPVAW be
explained by the country level of gender equality? MATERIAL AND METHODS
Material
Data was drawn from the 2010 Special Eurobarometer 344, which is a survey of domestic violence against women commissioned by the European Commission’s Directorate-General for Justice,
Freedom and Security (European Commission, 2010). The survey included data from 26,800
respondents above 15 years of age living in the 27 EU member states, and it measured public opinion on domestic violence against women together with a range of individual sociodemo-graphic factors. The sampling and interview methods have been described elsewhere (European Commission, 2010). A total of 23,322 respondents were eligible for analysis after excluding the respondents with missing values on any of the variables used in this study.
Outcome
Perceived severity of IPVAW was measured by using the survey question that asked respondents: “Please tell me whether you consider each of the following forms of domestic violence against women to be“very serious,” “fairly serious,” “not very serious” or “not at all serious.” This ques-tion was asked for five different types of violence: physical violence, psychological violence, sexual violence, threats of violence and restricted freedom. Responses were coded as 1¼ very serious, 2¼ fairly serious, 3 ¼ not very serious and 4 ¼ not at all serious.
In order to include all types of violence in the same analysis, an index was created by adding the values for all five questions, resulting in a score for each individual ranging from 5 to 20. The Cronbach’s alpha for this index was 0.845 for all countries together, ranging from 0.727 (the
Netherlands) to 0.884 (Czech Republic). The result formed a continuous variable with a skewed distribution, with a median of 5 and a mean of 6.59. The majority of respondents had a low value, and just above 50 percent of the respondents had the lowest value possible, 5, indicating that they reported finding all types of IPVAW to be very severe. This skewedness of the outcome variable makes it inappropriate for linear regression analysis, and thus the outcome was recoded into a binary variable in order to make it suitable for logistic regression. In order to define the group of respondents who found fewer types of violence to be severe, two groups were created from the index. These were the ones that had a score of one standard deviation or more above the mean (¼1), and those who had a score below that (¼0).
Individual-Level Variables
Individual sociodemographic characteristics previously shown to be correlated with attitudes toward IPVAW were used as predictors: gender, comparing male (¼1) to female (¼0); age, divided into six different groups: 15–24, 25–34, 35–44, 45–54, 55–64 and 65 and above; self-assessed social position in society, categorized as low, medium or high on a scale from 1–10 where low ¼ 1–4, medium ¼ 5–7 and high¼ 8–10; education, which defines the age until which respondents received education, catego-rized into still studying, no full-time education, below 15 years of age, 16–18 or 19 and older; area of residence, categorized into rural area or village, small/middle town or large town; whether the respond-ent knew at least one victim of IPVAW among friends, family, neighbors or at work, coded as 1¼ yes and 0¼ no and finally; how common the respondent perceived IPVAW to be in their country of
resi-dence, categorized as “very common,” “fairly common,” “not very common” and “not common
at all.”
Contextual Variables
The specific contextual effect in this study is country level gender equality, which was meas-ured by the gender equality index (GEI) created by the European Institute for Gender
Equality (EIGE, 2017a). This index measures gender gaps in either direction in the six
domains of work, money, time, health, power and knowledge (EIGE, 2017b). It presents a
possible range of 1–100, with high values indicating strong gender equality. The 27 EU mem-ber states analyzed in this study had values that ranged between 48.6 (Greece) and 81.1 (Sweden). The countries were categorized according to their GEI values into low (<59), mid-dle (60–69) or high (>70) level of gender equality. This variable was used as a measure of the actual gender gaps, but also as a proxy for the societal gender norms and values. In order to match the point in time of the Eurobarometer survey, the GEI data for this study was drawn from the year of 2010.
Statistical Analysis
Considering the nature of the research questions, the skewedness of the outcome variable, as well as the clustered structure of the data, we used a stepwise multilevel logistic regression analysis (Merlo et al.,2016).
Model 1: The Specific Individual Effects Model
In the first step of the analysis, a single-level logistic regression model that included only the indi-vidual effects, represented by the sociodemographic characteristics at the indiindi-vidual level, was applied. In this case the country level was ignored, giving us the standard single-level logistic
regression formula:
logitð Þ ¼ bpi 0þ b1X1iþ b2X2i::: þ bnXni (1)
Here, logitðpiÞ is the logit probability that the individual i perceives IPVAW as less severe,
given their individual characteristics, whileb0 is the intercept, i.e. the probability of the reference
person to perceive IPVAW as less severe. b1X1i, b2X2i, etc. are the log-odds for each covariate
holding each other covariate constant. From this model, odds ratios (OR) and their 95% confi-dence intervals (CI) were calculated. In addition, the discriminatory accuracy (DA) of the model was estimated to assess the ability of the individual characteristics included in the model in order to accurately classify individuals according to outcome (Merlo et al., 2016), i.e. to discriminate between those who find IPVAW to be more or less severe. The DA was assessed by measuring the area under the receiving operating characteristic curve (AU-ROC). The AU-ROC takes a value between 0.5 (no DA) and 1 (perfect DA), where a higher value indicates better DA (Merlo et al.,2016).
Model 2: The General Contextual Effects Model
This step provides an extension of the first model, by including the addition of a second level: the country level. In this model the random effect of the country of residence was added:
logitð Þ ¼ bpi 0þ b1X1ijþ b2X2ij::: þ bnXnijþ uj (2)
Here uj is the random effect of country j. It should be noted that the transformation into a
multilevel model results in the addition of j to each of the individual characteristics. This value thereby represents the coefficient for respondent i in country j. The DA (AU-ROC value) of this model was compared to the DA of Model 1, showing whether, and to what extent, the two-level model increases the ability to identify the outcome of an individual compared to the single-level model. If the change in AU-ROC between the models is large, so is the contextual effect (Merlo et al., 2017). To further examine to what extent the general contextual effect of which country one lives in helps us understand individual perceptions of IPVAW, the intraclass correlation coef-ficient (ICC) for Model 2 was calculated. The approach used for calculating the ICC in this study assumes an unobserved latent response variable that is continuous and represents a propensity for having one outcome rather than the other (Merlo et al.,2006). In this study, that translates to an assumption that each individual has an underlying continuous propensity for not perceiving IPVAW as severe, but that a person will start having this perception only at a certain threshold. The ICC is a measure of clustering and can be explained as“the share of the total outcome vari-ance which lies at the contextual level, having adjusted for any covariates” (Merlo et al., 2016, p. 3). A high ICC value means that the country of residence is highly relevant for understanding the variation in individual perceptions of IPVAW (Merlo,2018).
Model 3: The Specific Contextual Effects Model
This third step focuses on the importance of the specific contextual effect for predicting the out-come. In this model, the specific contextual effect of country-level gender equality was added to the second model:
logitð Þ ¼ bpi 0þ b1X1ijþ b2X2ij::: þ bnXnijþ bnGEIjþ uj (3)
HerebnGEIj is the contextual gender equality variable discussed above. When adding a specific
contextual variable, it is possible to investigate the proportion of the previously observed between-country variance, that can be explained by this specific contextual variable. This was done by calculating the proportional change in variance (PCV) after adding the specific
proportion of the contextual-level variance that is explained by the specific contextual effect (Merlo et al.,2006;2016). In the case of this study, that corresponds to the extent to which coun-try-level variance is mediated by the GEI. The AU-ROC value cannot be improved by adding fur-ther specific contextual variables. Since the random effect included in the second model already accounts for all observed and unobserved specific effects, this model will not have a higher
AU-ROC value than Model 2 (Merlo et al., 2016). For Model 3 we also obtained the Proportion of
Opposed Odds Ratios (POOR). The POOR can assume a value of 0%–50% and indicates the
per-centage of ORs that are in the opposite direction compared to the overall OR (Merlo et al., 2016). A lower value indicates more homogeneity, i.e. that a larger share of the ORs are in the same direction, while a high POOR indicates heterogeneity (Merlo et al.,2016). In this study the POOR was measured for the specific contextual variable of gender equality (bnGEIj). The value is
used as an indication of individual heterogeneity, complementing the average measure of the OR. A Note on the Model Estimation
The statistical software used for analysis were MLwiN and SPSS 25.0. The models were estimated using Markov Chain Monte Carlo (MCMC) methods.
RESULTS
Description of Study Population
Table 1 shows the percentage of respondents who perceived IPVAW to be less severe, compared to those who perceived it as more severe in each country. The countries are ranked by their GEI
score. Table 2 shows the proportion of respondents in each GEI group (low, middle and high)
that represent the different individual-level characteristics. For example, the rate of males/females in the study is fairly equal in all three categories, with slightly more females than males. All three Table 1. Description of the percentage of respondents who perceive IPVAW to be less vs. more severe in each country, organ-ized by gender equality index value (from low to high).
GEI group Country GEI Less severe % More severe %
Low Greece 48.6 9.3 90.7 Cyprus 49 9.1 90.9 Romania 50.8 24.9 75.1 Hungary 52.4 25 75 Slovakia 53 18.6 81.4 Italy 53.3 9.8 90.2 Estonia 53.4 26.6 73.4 Portugal 53.7 22.3 77.7 Malta 54.4 8 92 Lithuania 54.9 30.6 69.4 Bulgaria 55 22 78 Latvia 55.2 37.9 62.1 Poland 55.5 26.1 73.9 Czech Republic 55.6 15.9 84.1 Austria 58.7 22.3 77.7 Middle Luxembourg 61.2 10.3 89.7 Germany 62.6 15.7 84.3 Slovenia 62.7 20.8 79.2 Ireland 65.4 6.6 93.4 Spain 66.4 15.2 84.8 France 67.5 6.1 93.9 United Kingdom 68.7 5.9 94.1 Belgium 69.3 12.1 87.9 High Finland 73.1 16.5 83.5 The Netherlands 74 8.7 91.3 Denmark 75.2 7.5 92.5 Sweden 81.2 4.8 95.2
categories have a relatively large proportion of respondents above 65 years of age, while a larger proportion of respondents in the high than in the low or middle GEI countries reported having a higher education, occupying a high social position in society and living in a large city. More than half of the respondents in the high GEI countries claimed to know a victim of IPVAW, either at their workplace, in their neighborhood or among their family and close friends. However, a larger proportion of respondents in both the low (24.6%) and medium (29.1%) GEI countries believed IPVAW was very common, as compared to those in countries with high GEI (18.9%).
Multilevel Analyses
The results of the multilevel analysis are presented inTable 3, which shows the ORs and 95% CIs for perceiving IPVAW as less severe compared to the reference in each respondent group. In addition to the ORs and CIs, the POOR, ICC, PCV and AU-ROC for all three models are shown when relevant.
Model 1
The single-level model shows the ORs for perceiving IPVAW as being less severe compared to the reference group for all the individual factors respectively (seeTable 3). Males were more likely Table 2. Percentage of respondents by individual characteristics in the countries with low, medium and high GEI.
% Within groups of country level Gender Equality
Low Medium High % Missing
Number of countries (individuals) 15 (14,267) 8 (8,510) 4 (4,023)
Perceived severity 3.1 Less severe 21.2 11.7 9.4 More severe 78.8 88.3 90.6 Gender 0 Female 55 53.9 51.8 Male 45 46.1 48.2 Age 0 15–24 13.3 11.2 8.8 25–34 16.4 15.3 10.8 35–44 18 16.5 14.7 45–54 17 16.8 16.9 55–64 16.1 16.2 21.2 65 and above 19.2 24 27.7 Education (age) 2 Still studying 9.1 6.7 7.8 No full-time education 0.3 0.5 0.7 15 years or less 19.7 23.8 10 16–18 years 37.5 35.9 18.8 19 years or more 33.3 33.1 62.7
Position in society–self perceived 3.1
Low 28.9 16.7 8.6
Medium 62.1 72.6 67.1
High 9.1 10.8 24.3
Community 0.2
Rural area or village 35.7 38.4 29.9
Small/middle town 33.3 35.3 45 Large town 31 26.4 25.1 Know victim 3.2 Yes 42.4 39.7 52.1 No 57.6 60.3 47.9 Perceived commonness 4.2 Very common 24.6 29.1 18.9 Fairly common 52.4 52.8 57.9
Not very common 21.3 17.2 22.4
than women to perceive IPVAW as less severe, as were persons in the youngest age group, per-sons in a low self-assessed social position in society and with no full-time education, and perper-sons living in a rural area. Those perceiving IPVAW to be not at all common displayed a much higher likelihood of perceiving IPVAW as less severe, while those who did not know a victim showed a slightly higher likelihood. The AU-ROC value for Model 1 is 0.673. The ROC curve for Model 1 is represented by the blue curve inFigure 1.
Table 3. Multilevel logistic regression analysis of the perceptions of severity of IPVAW among citizens in the EU. Simple logistic regression Multilevel logistic regression
Model 1 Model 2 Model 3
Individual average effects OR (95% CI) OR (95% CI) OR (95% CI) Gender Male vs. female 1.45 (1.35–1.56) 1.52 (1.41–1.64) 1.52 (1.41–1.63) Age 15–24 Reference 25–34 0.76 (0.63–0.93) 0.77 (0.65–0.92) 0.76 (0.64–0.90) 35–44 0.77 (0.62–0.95) 0.81 (0.69–0.97) 0.80 (0.67–0.96) 45–54 0.79 (0.64–0.96) 0.83 (0.70–0.99) 0.82 (0.69–0.99) 55–64 0.87 (0.71–1.07) 0.96 (0.81–1.13) 0.94 (0.79–1.13) 65 and above 0.86 (0.70–1.05) 0.97 (0.81–1.16) 0.95 (0.80–1.14) Education (age)
19 years or more Reference
16–18 years 1.26 (1.15–1.38) 1.18 (1.07–1.29) 1.18 (1.07–1.29) 15 years or less 1.32 (1.18–1.48) 1.48 (1.31–1.67) 1.48 (1.31–1.67) No full-time education 1.54 (0.90–2.63) 1.85 (1.05–3.25) 1.88 (1.09–3.24) Still studying 0.75 (0.60–0.94) 0.68 (0.56–0.83) 0.67 (0.55–0.83) Position in society–self perceived
High Reference
Medium 1.38 (1.20–1.58) 1.17 (1.02–1.34) 1.16 (1.02–1.32) Low 2.13 (1.84–2.46) 1.42 (1.21–1.66) 1.40 (1.22–1.62) Community
Large town Reference
Small/middle town 1.15 (1.05–1.27) 1.24 (1.13–1.37) 1.24 (1.12–1.37) Rural area or village 1.25 (1.13–1.37) 1.28 (1.15–1.41) 1.27 (1.15–1.40) Perceived commonness
Very common Reference
Fairly common 2.80 (2.49–3.15) 2.73 (2.42–3.09) 2.80 (2.49–3.15) Not very common 4.50 (3.97–5.11) 4.49 (3.91–5.14) 4.59 (4.01–5.25) Not at all common 7.28 (5.57–9.51) 7.76 (5.82–10.35) 7.93 (6.00–10.49) Know victim
No vs. yes 1.10 (1.02–1.19) 1.21 (1.12–1.31) 1.21 (1.12–1.31) Specific contextual average effect
Gender equality index
High Reference
Medium 1.59 (0.89–2.84)
Low 3.31 (2.26–4.85)
POOR % Medium GEI 31
POOR % Low GEI 9
General contextual effects
Country level variance 0.532 0.417
PCV % 22 ICC % 14 11.3 AU-ROC 0.673 (0.664–0.682) 0.742 (0.733–0.750) 0.741 (0.733–0.750) AU-ROC change 0.069 –0.001 Goodness of fit DIC 19168.03 18022.57 18022.15 DIC change –1145.46 –0.42
A higher OR indicates a higher likelihood of perceiving IPVAW as less severe compared to the reference. Expressed as the change compared to the previous model.
Model 2
When adding the general contextual effect of the country of residence to the model, there was a slight change in some of the ORs of the individual predictors, but overall the directions of the relationships remain unchanged. The change in ORs from Model 1 to Model 2 can in itself be an indication that the contextual effect of the country of residence does explain some of the variance. This is also shown in the increased AU-ROC value compared to Model 1. The ROC-curve for Model 2 is represented by the red line inFigure 1(which overlaps with the green line represent-ing Model 3). The higher AU-ROC value and the larger area under the curve shows that Model 2 has better DA than Model 1. The ICC value demonstrates that 14% of the individual variance lies between countries, rather than within them.
Model 3
In model 3 the GEI has been added as a specific contextual effect. While the ORs for the individ-ual characteristics are similar in Models 2 and 3 (Table 3), individuals residing in countries with a lower GEI have conclusively higher odds of perceiving IPVAW as less severe compared to indi-viduals in countries with medium or high GEI. As seen by examining the CI for the OR in the medium vs high GEI group, the difference in odds between these two groups is inconclusive. The lower ICC value in Model 3 (11.3%) compared to that of Model 2 indicates that some of the between-country variance observed in Model 2 can be explained by the contextual effect of the GEI. This is further demonstrated by the PCV value, which shows that 22% of the between-coun-try variance can be attributed to the counbetween-coun-try-level gender equality. The POOR for the group of countries with low GEI compared to those with high GEI was low (9%), indicating homogeneity. For the medium GEI group it was considerably higher (31%), which is likely a result of the inconclusive difference between the medium and high GEI groups.
DISCUSSION
IPVAW poses a serious threat to women’s health, security and human rights globally. To contrib-ute to an increased understanding of the attitudes that can potentially affect the prevalence of and the community responses toward acts of IPVAW, the aim of this study was to investigate the effect of country context and gender equality in addition to individual sociodemographic charac-teristics on perceived severity of IPVAW. Moreover, the study aimed to evaluate the ability of these contextual and individual factors to accurately discriminate between those individuals who perceive IPVAW as severe and those who do not.
Conclusive relationships were found between the outcome and all individual sociodemographic predictors, meaning that male gender, young age, low education and low self-assessed social pos-ition in society as well as habitation in a rural area was associated with a higher likelihood of per-ceiving IPVAW as a less severe issue. Perceptions of IPVAW as not at all common was the stronger individual predictor, the influence of which was much greater than that of not knowing an IPVAW victim.
Our results show that 14% of the total variance in individual perceptions lie at the contextual level, and thus represent the between-country variability. While no definite interpretation exists for the magnitude of the ICC, according to a recently proposed framework for preforming geo-graphical comparisons (Merlo et al.,2019), an ICC between 10 and 20% falls into the category of large geographical differences. Thus, our results support the importance of the societal context, alongside that of individual factors, for the shaping of perceived severity of IPVAW (Heise, 1998). Furthermore, individuals residing in countries with a lower GEI were more likely to per-ceive IPVAW as less severe, compared to individuals in high GEI countries. The DA of the mod-els was fair, and the study suggests that disparities exist between, as well as within, nation state borders.
A particularly interesting finding of this study is the relationship between the specific context-ual effect of GEI and the individcontext-ual perceptions of the severity of IPVAW. The respondents in countries with low GEI were more likely to perceive IPVAW as less severe compared to those in countries with high GEI. In support of the understanding of IPVAW as strongly related to patri-archal structures in society (Lawson,2012), increased country-level gender equality is here associ-ated with citizens’ perceptions. However, the inconclusive difference in perceptions between the countries in the middle and high GEI groups indicates that the association with GEI is only rele-vant to a certain level. The fact that the country-level GEI explains 22% of the total contextual-level variance means that it is an important factor for understanding the variance in attitudes, even though some of the contextual-level variance is still unaccounted for. The low POOR for the low GEI group indicates a homogeneous relationship. Only 9% of the time, an individual from a low GEI country was less rather than more likely to perceive IPVAW as less severe than someone from a high GEI country. Meanwhile, the high POOR for the medium GEI group is coherent with the inconclusiveness of the difference in odds between the medium and high GEI groups. Here the heterogeneity is so large that in almost half of the cases the relationship is opposite to the average. It may here also be mentioned that, as noted above, a larger share of the respondents in low and medium than in high GEI believed IPVAW to be very common.
Nevertheless, the importance of the GEI is coherent with Heise (1998) integrated ecological approach. In this study the GEI represents the societal structure and gender norms at the context-ual level (representing the macrosystem in Heise’s model) in which individuals are embedded. Our results show that the contextual level works conjointly with relational, situational and per-sonal characteristics in relation to individual perceptions. Considering the Nordic paradox (Gracia & Merlo, 2016) and the previous difficulties in establishing a conclusive relationship between GEI and attitudes toward IPVAW in the EU (Ivert et al.,2018), the results of this study are more surprising than they may appear.
IPVAW persists despite being illegal and, according to the results of this study, socially undesirable, particularly in countries with high GEI (FRA, 2014; Gracia & Merlo, 2016). The inconsistency between people’s perceptions and their actions raise questions which merit further investigation. Why is the prevalence of IPVAW so high in the high GEI countries, even though the inhabitants claim to find it very severe? We could assume that in a context where people find IPVAW to be less severe, fewer cases of IPVAW would be reported, which could explain the lower prevalence. Conversely, in countries with high GEI, where it is perceived as very severe by a larger share of the population, more cases will be taken seriously and thus be reported (Gracia & Merlo,2016). This could be one explanation as to why the high GEI countries also have high prevalence of IPVAW (FRA, 2014; Gracia & Merlo, 2016). However, that does not explain why there is such a high prevalence of IPVAW cases to begin with in more gender equal contexts where more people perceive it to be more severe.
The fact that people in countries with high a GEI tend to find IPVAW to be more severe is also surprising if we apply the backlash hypothesis. One possible explanation of the inconsistency between perception and action, or between theory and practice (SOU, 2004, p. 121), in the high GEI countries could be that respondents here feel a stronger obligation to answer in a certain way, to avoid stigmatization associated with downplaying the severity of IPVAW. Wemrell et al. (2020) investigate the discrepancy between gender equality and the relatively high prevalence of IPVAW in the specific context of Sweden. One of their suggestions for this discrepancy is that the image of Sweden as a gender-equal country may in itself have hindered an appropriate response to IPVAW, therefore allowing its continuation. Following this reasoning it is plausible that respondents in countries with high GEI would respond that they find IPWAW to be very severe because it is a societal norm, and because it corresponds to a gender-equal self-image, even if this does not correspond with their actions. This could indicate a type of self-monitoring behavior, to ensure that one’s responses agree with the social context (Snyder, 1987): a phenom-enon previously referred to as social desirability bias (Brace,2008).
Limitations
Since a cross-sectional study design was used, this study identifies correlations but not causal relationships. Moreover, the data used is nearly 10 years old, and we can assume that societies have undergone changes during these years. For example, IPVAW has received more attention in public debate, especially following the viral spread of the MeToo movement in 2017. Outdated numbers do not present a serious limitation, however, as our aim is not to demonstrate IPVAW prevalence levels but to investigate how factors at different levels impact individual perceptions, in ways which are likely to remain consistent over time.
The Eurobarometer survey does not specify how well different ethnic minorities are repre-sented in the sample, or the amount of time respondents had lived in the country of residence into which they were recorded. We thus have no way of knowing if this may have resulted in a sample bias. Moreover, the survey, and in extension this study, excludes the population with gen-der identities other than male or female. Hence, no conclusions can be drawn concerning this part of the population. The response rates varied between different countries and the Eurobarometer survey does not include any information concerning the characteristics of the non-responders, meaning that a non-response bias cannot be ruled out.
Moreover, the data includes only two levels: the country level and the individual level, although in a previous study on prevalence of IPVAW in Valencia, Spain the community level has proven to be important (Gracia et al.,2015). The community level would represent the exole-vel in Heise’s ecological model, and adding this third level to this study could have added inter-esting and useful insights. However, we were unable to do so since the survey data did not include community data on the respondents’ areas of residence.
CONCLUSIONS AND IMPLICATIONS
The association between higher GEI scores and a stronger likelihood for individuals to perceive IPVAW as a severe issue, found in this study, has some implications for further attempts to com-bat IPVAW. As public perceptions may impact the societal response to acts of IPVAW, this indi-cates that efforts to address IPVAW will benefit from increasing gender equality, despite previous difficulties in linking gender equality to decreased prevalence in the EU. The study also shows that although there is a substantial amount of variance between countries, large differences are
also seen within countries. This means that efforts to change people’s perceptions of IPVAW
need to target disparities both across and within nations in the EU. As perceived commonness of IPVAW is the strongest individual predictor of perceived severity found in this study, it is likely that increased awareness of the magnitude of this issue can lead to strengthened perceptions of its severity. Hence, awareness-increasing campaigns are an appropriate way of combatting IPVAW and making people realize its severity.
This study found that the GEI explains 22% of the country-level variance in perceptions. Further research is needed to investigate which other specific contextual-level predictors account for the other, still unexplained, portion of the country-level variance. Since the individual level was also proven to be important in determining individual perceptions of severity, additional research focusing on specifying further individual predictors is also necessary. Moreover, this study raises important questions about the perceived severity and the prevalence of IPVAW. Why does IPVAW prevalence remain so high, even when the public perceives it to be very severe? Why do perceptions of high severity not correspond to fewer IPVAW cases? To answer these questions, further research is needed.
Funding
This research was supported from the Swedish Research Council (VR) by Grants #2017-03093 (AKI, MW) (https:// www.vr.se/english.html).
ORCID
Anna-Karin Ivert http://orcid.org/0000-0001-7749-9549
REFERENCES
Archer, J. (2006). Cross-cultural differences in physical aggression between partners: A social-role analysis. Personality and Social Psychology Review: Review, 10(2), 133–153.https://doi.org/10.1207/s15327957pspr1002_3
Boethius, S. (2015). M€an, våld och moralarbete: rapporter från m€an som s€okt behandling f€or våld i n€ara relationer. Lund University.
Brace, I. (2008). Questionniare design. 2nd ed. Kogan Page Ltd.
Campbell, J. C. (2002). Health consequences of intimate partner violence. The Lancet, 359(9314), 1331–1336.
https://doi.org/10.1016/S0140-6736(02)08336-8
Copp, J. E., Giordano, P. C., Longmore, M. A., & Manning, W. D. (2019). The development of attitudes toward intimate partner violence: An examination of key correlates among a sample of young adults. Journal of Interpersonal Violence, 34(7), 1357–1387.https://doi.org/10.1177/0886260516651311
Devries, K. M., Mak, J. Y. T., Garcıa-Moreno, C., Petzold, M., Child, J. C., Falder, G., Lim, S., Bacchus, L. J., Engell, R. E., Rosenfeld, L., Pallitto, C., Vos, T., Abrahams, N., & Watts, C. H. (2013). Global health. The global prevalence of intimate partner violence against women. Science, 340(6140), 1527–1528. https://doi.org/10.1126/ science.1240937
EIGE. (2017a). Gender Equality Index 2017 – Measuring gender equality in the European Union 2005–2015. European Institute for Gender Equality.
EIGE. (2017b). Gender Equality Index 2017: Measurement framework of violence against women. European Institute for Gender Equality.
European Commission. (2010). Domestic violence against women: Special Eurobarometer 344. Retrieved April 17, 2020, from:https://ec.europa.eu/commfrontoffice/publicopinion/archives/ebs/ebs_344_en.pdf
FRA. (2014). Violence against Women: An EU-wide survey. European Union Agency for Fundamental Rights. Garcıa-Moreno, C., Zimmerman, C., Morris-Gehring, A., Heise, L., Amin, A., Abrahams, N., Montoya, O.,
Bhate-Deosthali, P., Kilonzo, N., & Watts, C. (2015). Addressing violence against women: A call to action. The Lancet, 385(9978), 1685–1695.https://doi.org/10.1016/S0140-6736(14)61830-4
Gottzen, L., & Korkmaz, S. (2013). Killars våld mot tjejer i n€ara relationer. I Unga och våld: en analys av maskuli-nitet och f€orebyggande verksamheter. Ungdomsstyrelsen.
Gracia, E. (2014). Intimate partner violence against women and victim-blaming attitudes among Europeans. Bulletin of the World Health Organization, 92(5), 380–381.https://doi.org/10.2471/BLT.13.131391
Gracia, E., Garcia, F., & Lila, M. (2014). Male police officers’ law enforcement preferences in cases of intimate part-ner violence versus nonintimate interpersonal violence: Do sexist attitudes and empathy matter? Criminal Justice and Behavior, 41(10), 1195–1213.https://doi.org/10.1177/0093854814541655
Gracia, E., & Herrero, J. (2006). Acceptability of domestic violence against women in the European Union: A multilevel analysis. Journal of Epidemiology and Community Health, 60(2), 123–129.https://doi.org/10.1136/jech. 2005.036533
Gracia, E., Herrero, J., Lila, M., & Fuente, A. (2009). Perceived neighborhood social disorder and attitudes toward domestic violence against women among Latin-American immigrants. European Journal of Psychology Applied to Legal Context, 1, 25–43.
Gracia, E., & Lila, M. (2015). Attitudes towards violence against women in the EU. Publication Office of the European Union.
Gracia, E., Lila, M., & Santirso, F. A. (2020). Attitudes toward intimate partner violence against women in the European Union: A systematic review. European Psychologist, 25(2), 104–121.https://doi.org/10.1027/1016-9040/ a000392
Gracia, E., Lopez-Quılez, A., Marco, M., Lladosa, S., & Lila, M. (2015). The spatial epidemiology of intimate part-ner violence: Do neighborhoods matter? American Journal of Epidemiology, 182(1), 58–66. https://doi.org/10. 1093/aje/kwv016
Gracia, E., Martın-Fernandez, M., Marco, M., Santirso, F. A., Vargas, V., & Lila, M. (2018). The Willingness to Intervene in Cases of Intimate Partner Violence Against Women (WI-IPVAW) Scale: Development and valid-ation of the long and short versions. Frontiers in Psychology, 9, 104–121. https://doi.org/10.3389/fpsyg.2018. 01146
Gracia, E., & Merlo, J. (2016). Intimate partner violence against women and the Nordic paradox. Social Science & Medicine, 157, 27–30.https://doi.org/10.1016/j.socscimed.2016.03.040
Gracia, E., & Tomas, J. M. (2014). Correlates of victim-blaming attitudes regarding partner violence against women among the Spanish general population. Violence against Women, 20(1), 26–41. https://doi.org/10.1177/ 1077801213520577
Heise, L. L. (1998). Violence against women: An integrated, ecological framework. Violence against Women, 4(3), 262–290.https://doi.org/10.1177/1077801298004003002
Heise, L. L., & Kotsadam, A. (2015). Cross-national and multilevel correlates of partner violence: An analysis of data from population-based surveys. The Lancet Global Health, 3(6), e332–e340. https://doi.org/10.1016/S2214-109X(15)00013-3
Herrero, J., Torres, A., Rodrıguez, F. J., & Juarros-Basterretxea, J. (2017). Intimate partner violence against women in the European Union: The influence of male partners’ traditional gender roles and general violence. Psychology of Violence, 7(3), 385–394.https://doi.org/10.1037/vio0000099
Ivert, A. K., Merlo, J., & Gracia, E. (2018). Country of residence, gender equality and victim blaming attitudes about partner violence: A multilevel analysis in EU. European Journal of Public Health, 28(3), 559–564.https:// doi.org/10.1093/eurpub/ckx138
Jewkes, R. (2002). Intimate partner violence: Causes and prevention. The Lancet, 359(9315), 1423–1429.https://doi. org/10.1016/S0140-6736(02)08357-5
Kovacs, R. J. (2018). The macro-level drivers of intimate partner violence: New evidence from a multilevel dataset. Global Public Health, 13(7), 944–956.https://doi.org/10.1080/17441692.2017.1317010
Krantz, G. (2002). Violence against women: A global public health issue. Journal of Epidemiology and Community Health, 56(4), 242–243.https://doi.org/10.1136/jech.56.4.242
Lauritsen, J. L., & Schaum, R. J. (2004). The social ecology of violence against women. Criminology, 42(2), 323–357.https://doi.org/10.1111/j.1745-9125.2004.tb00522.x
Lawson, J. (2012). Sociological theories of intimate partner violence. Journal of Human Behavior in the Social Environment, 22(5), 572–590.https://doi.org/10.1080/10911359.2011.598748
Lundgren, E., Heimer, G., Westerstrand, J., & Kalliokoski, A.-M. (2002). Captured Queen: Men’s violence against Women in“Equal” Sweden— A prevalence study. Uppsala University.
Martın-Fernandez, M., Gracia, E., & Lila, M. (2018). Assessing victim-blaming attitudes in cases of intimate partner violence against women: Development and validation of the VB-IPVAW scale. Psychosocial Intervention, 27(3), 133–143.https://doi.org/10.5093/pi2018a18
Merlo, J. (2018). Multilevel analysis of individual heterogeneity and discriminatory accuracy (MAIHDA) within an intersectional framework. Social Science & Medicine, 203, 74–80.https://doi.org/10.1016/j.socscimed.2017.12.026
Merlo, J., Chaix, B., Ohlsson, H., Beckman, A., Johnell, K., Hjerpe, P., Råstam, L., & Larsen, K. (2006). A brief con-ceptual tutorial of multilevel analysis in social epidemiology: Using measures of clustering in multilevel logistic regression to investigate contextual phenomena. Journal of Epidemiology and Community Health, 60(4), 290–297.https://doi.org/10.1136/jech.2004.029454
Merlo, J., Mulinari, S., Wemrell, M., Subramanian, S. V., & Hedblad, B. (2017). The tyranny of the averages and the indiscriminate use of risk factors in public health: The case of coronary heart disease. SSM– Population Health, 3(C), 684–698.https://doi.org/10.1016/j.ssmph.2017.08.005
Merlo, J., Wagner, P., Ghith, N., & Leckie, G. (2016). An original stepwise multilevel logistic regression analysis of discriminatory accuracy: The case of neighbourhoods and health. PLoS One, 11(4), e0153778.https://doi.org/10. 1371/journal.pone.0153778
Merlo, J., Wagner, P., & Leckie, G. (2019). A simple multilevel approach for analysing geographical inequalities in public health reports: The case of municipality differences in obesity. Health & Place, 58, 102145.https://doi. org/10.1016/j.healthplace.2019.102145
Nybergh, L., Taft, C., Enander, V., & Krantz, G. (2013). Self-reported exposure to intimate partner violence among women and men in Sweden: Results from a population-based survey. Bmc Public Health, 13(1), 845.https://doi. org/10.1186/1471-2458-13-845
Rizo, C. F., & Macy, R. J. (2011). Help seeking and barriers of Hispanic partner systematic review of the literature. Aggression and Violent Behavior, 6, 250–264.
Sanz-Barbero, B., Corradi, C., Otero-Garcıa, L., Ayala, A., & Vives-Cases, C. (2018). The effect of macrosocial poli-cies on violence against women: A multilevel study in 28 European countries. International Journal of Public Health, 63(8), 901–911.https://doi.org/10.1007/s00038-018-1143-1
Snyder, M. (1987). Public appearances, private realities: The psychology of self-monitoring. W.H. Freeman. SOU. (2004). Slag i luften. En utredning om myndigheter, mansva˚ld och makt. Fritzes, 121.
Stickley, A., Kislitsyna, O., Timofeeva, I., & Våger€o, D. (2008). Attitudes toward intimate partner violence against women in Moscow, Russia. Journal of Family Violence, 23(6), 447–456. https://doi.org/10.1007/s10896-008-9170-y
Tran, T. D., Nguyen, H., & Fisher, J. (2016). Attitudes towards intimate partner violence against Women among Women and Men in 39 low- and middle-income countries. PLoS One, 11(11), e0167438. https://doi.org/10. 1371/journal.pone.0167438
Uthman, O., Lawoko, S., & Moradi, T. (2010). The role of individual, community and societal gender inequality in forming women’s attitudes toward intimate-partner violence against women: A multilevel analysis. World Health & Population, 12(2), 5–17.https://doi.org/10.12927/whp.2010.22007
Waltermaurer, E. (2012). Public justification of intimate partner violence: A review of the literature. Trauma, Violence & Abuse, 13(3), 167–175.https://doi.org/10.1177/1524838012447699
Waltermaurer, E., Butsashvili, M., Avaliani, N., Samuels, S., & McNutt, L.-A. (2013). An examination of domestic partner violence and its justification in the Republic of Georgia. BMC Women’s Health, 13(1), 44. https://doi. org/10.1186/1472-6874-13-44
Wemrell, M., Lila, M., Gracia, E., & Ivert, A.-K. (2020). The Nordic Paradox and intimate partner violence against women (IPVAW) in Sweden: A background overview. Sociology Compass, 14(1), e12759.https://doi.org/10.1111/ soc4.12759
West, A., & Wandrei, M. L. (2002). Intimate partner violence– A model for predicting interventions by informal helpers. Journal of Interpersonal Violence, 17(9), 972–986.https://doi.org/10.1177/0886260502017009004
Whaley, R. B. (2001). The paradoxical relationship between gender inequality and rape: Toward a refined theory. Gender & Society, 15(4), 531–555.
WHO. (2002). World report on violence and health. Retrieved April 17, 2020, from:https://www.who.int/violence_ injury_prevention/violence/world_report/en/
World Health Orgainization. (2010). Preventing intimate partner and sexual violence against women: Taking action and generating evidence. Retrieved April 17, 2020, from: https://www.who.int/violence_injury_prevention/publi-cations/violence/9789241564007_eng.pdf
Xie, M., Heimer, K., & Lauritsen, J. L. (2012). Violence against women in U.S. metropolitan areas: Changes in women’s status and risk, 1980–2004. Criminology, 50(1), 105–143. https://doi.org/10.1111/j.1745-9125.2011. 00245.x
