Introduction

Intimate partner violence (IPV) is usually defined as the occurrence of different forms of violence (e.g., physical, psychological, sexual) between current or former intimate partners, resulting in harm for the victim (World Health Organization [WHO], 2017). The potentially severe short- and long-term effects on the physical and mental health of victims, or ultimately their death, have made IPV a growing concern as a public health problem (Miller & McCaw, 2019; Stöckl et al., 2013; WHO, 2013). Furthermore, it is a highly costly phenomenon, with billions of euros being annually spent in Europe on their direct and indirect consequences (European Institute for Gender Equality [EIGE], 2014).

Establishing the prevalence of IPV has been a major concern in western countries for a long time (e.g., Kury et al., 2004; Magdol et al., 1997; Straus & Gelles, 1986) and has turned into a global priority (e.g., Bott et al., 2019; Elghossain et al., 2019). The WHO (2017) estimated that the prevalence of physical and/or sexual IPV against women is lower in high-income countries (around 23.2%) when compared to other global regions: 37.7% in the South-East Asia region, 37% in the Eastern Mediterranean region, and 24.6% in the Western Pacific region (for a comprehensive review on IPV prevalence worldwide, see Esquivel-Santoveña et al., 2013). In Europe, the past-year prevalence of physical and/or sexual IPV towards women aged between 18 and 74 years old is roughly 8%, and lifetime prevalence (since the age of 15) is around 22% (European Union Agency for Fundamental Rights [FRA], 2014). Nevertheless, there has been a consensus that discrepancies in definitions, sampling, time frame, research design, and assessment methods may impair comparisons, even between European countries (e.g., Belén et al., 2018; Esquivel-Santoveña et al., 2013; Garcia-Moreno et al., 2006; Hagemann-White, 2001; Kury et al., 2004). Furthermore, many of the IPV prevalence estimates did not come from national surveys, and some were part of violence against women studies, excluding male victimization and the LGBTQ population, and/or ignoring the potential presence of bidirectionality of violence between partners (e.g., Belén et al., 2018; FRA, 2014; Gracia et al., 2019; Hagemann-White, 2001). Nonetheless, robust evidence has been showing symmetry in the perpetration of IPV by men and women either in the community (e.g., Archer, 2000; Chan, 2011; Hamel, 2020; Straus & Mickey, 2012) as in forensic or clinical heterosexual samples (e.g., Hamberger & Larsen, 2015; Straus, 2011). These findings point out the need to adopt a gender-inclusive questioning regarding victimization and perpetration experiences and to explore directionality patterns of IPV, regardless of sexual orientation.

The Case of Portugal

Portugal, one of the first signatory countries of the Istanbul Convention, has done several attempts to monitor the prevalence of IPV, which has an annually estimated cost of 2.552.698.493 euros (EIGE, 2014). Since 2000, IPV is a public crime in Portugal, typified as domestic violence (which includes various forms of family violence). It is of mandatory reporting, and it has been among the most common crime types reported in Portugal over the last years. Indeed, according to the latest available data of the Annual Homeland Security Report, in 2020 there were 27,637 events of domestic violence recorded by the police (Sistema de Segurança Interna, 2021). Being an atypical year due to the COVID-19 pandemic, in comparison to 2019, and specifically regarding IPV, a decrease of 5.5% was observed in the number of reports across the country. In comparison with the previous 4 years, a decrease of about 11% in the IPV reports was identified in urban areas alone (Capinha et al., 2021).

Besides criminal statistics, data on the prevalence of IPV in Portugal came mainly from studies focused on gender violence (which included other forms of violence and discrimination against women in different contexts) or on the criminal records of domestic violence (Lisboa et al., 2010). Two national studies were published (Lisboa et al., 2009; Lourenço et al., 1997); however, they did not include the autonomous regions (the Azores and Madeira Islands). Representing important steps to assess IPV prevalence, these studies collected data through interviews, carried out in private (e.g., the respondent’s home) or public (e.g., coffee) spaces, which may increase social desirability, shame, or even fear, interfering with the ability to report undergone behaviors. The mainland survey by Lisboa et al. (2009) was the first to include male victimization, but questions asked to men and women were not the same, and the methodology that was used to select and establish the first contact with participants was not clarified. In 2010, the Azores Islands Regional Government conducted a study with similar goals, aiming to compare the results with findings from the mainland territory (Rocha & Lalande, 2010). The authors analyzed data from police records and interviewed several professionals and women victims in shelters resorting to a non-structured set of questions. Lisboa et al. (2010) reviewed all the existing findings and identified a life prevalence of women victimization between 38 and 48% in the mainland and 53% in the Azores Islands. Regarding male victimization, the identified prevalence varied between 43% in the mainland and 46% in the Azores Islands.

Data about Portugal are also available from the study of FRA (2014), which estimated that 19% of Portuguese women have experienced physical and/or sexual violence by a current or previous partner since the age of 15. Other studies, representative of the north region of Portugal, assessed the past-year prevalence of IPV in men and women in heterosexual couples (Costa et al., 2015; Machado, et al., 2007) and found a prevalence ranging from 3.3 (injury victimization in women) to 54.3% (psychological aggression perpetrated by women). Costa et al. (2015) found no gender differences regarding victimization prevalence and frequency of the acts identified as occurring in the past year. Regarding perpetration, gender differences were found in frequency, with women more frequently perpetrating acts of “minor” psychological violence, while acts of minor sexual coercion were more frequently perpetrated by men. Most of the participants reported being involved both as victims and perpetrators (bidirectional violence). Machado et al. (2007) also found differences across gender, with a higher prevalence of perpetration of any violent act and physical aggression among males. More recently, a web-based study, targeting only Portuguese heterosexual men (Machado et al., 2019), identified perpetration rates of 70.6% and 77.6%, in the past 12 months and during the lifetime, respectively. In the same study, victimization rates were 69.7% in the past 12 months and 76.5% during lifetime. Bidirectionality was found in 73.7% of the cases. Similar results were obtained in student samples. Using the CTS-2, Paiva and Figueiredo (2006) found perpetration rates ranging from 5 (injury) to 54.2% (psychological aggression) and victimization rates ranging from 4.5 (injury) to 51.5% (psychological aggression). The self-report of sexual coercion being higher on the male sample was the only gender difference identified. Machado et al. (2010) used a different instrument, focused only on physical and emotional violence occurred in the past year. They found a 25.4% perpetration rate and a 30.6% victimization rate. Regarding gender differences, women reported more acts of IPV that men, but no gender differences were found regarding victimization.

As in other countries, less attention has been given to IPV among non-heterosexual individuals, with no consensus on its prevalence in the Portuguese population (De Barros et al., 2019). The IPV prevalence identified in non-heterosexual Portuguese couples in different studies ranged from 15.9 to 91.7% for perpetration and from 20.6 to 91.7% for victimization (see Santos & Caridade, 2017). This considerable variability calls for the need for further research on this topic.

The different range of findings regarding IPV in Portugal stressed the need to understand the real dimension of the phenomenon at a nationwide level. The lack of research specifically targeting IPV prevalence, its risk factors, within a gender-inclusive perspective and in a national sample, is a major constraint to an empirically guided development of actions to prevent IPV and to intervene with victims and/or perpetrators. Therefore, the main goals of this study were to estimate the prevalence and frequency of occurrence in the past year, of different forms of IPV in the Portuguese population, and to investigate gender and sexual orientation differences. This study also aimed to identify directionality patterns of violence and dyadic concordance types and to explore the association between IPV and relevant risk factors.

Method

Participants

The sample consisted of Portuguese adults (aged 18 years or older) who were in an intimate relationship by the time of the study. Participants were recruited across all municipalities of the mainland and the autonomous regions (Azores and Madeira) in Portugal. A total of 2133 participants accessed the study link. Of these, participants with missing responses in the full questionnaire/protocol (n = 737) or with over 20% of missing items of a questionnaire (n = 4) were excluded. The final sample comprised 1392 Portuguese adults from the community (77.4% female, 0.1% another gender), aged between 18 and 80 years old. All participants were in an intimate relationship, on average for 121.51 months (approximately 10 years). Most of the participants did not have children (58.4%), nor considered themselves as being financially dependent on their intimate partner (80.2%). Table 1 summarizes the characteristics of the participants by gender.

Table 1 Demographic, developmental, and relational characteristics of the complete sample and by gender
Full size table

Gender differences were found concerning age (t(464.907) = 11.96, p < 0.001, d = 0.79, 95% CI [8.30, 11.55]) and the duration of the relationship (t(431.364) = 5.80, p < 0.001, d = 0.40, 95% CI [34.41, 69.67]). Significant gender differences were also found regarding marital status (χ2(2) = 42.10, p < 0.001, Cramer’s V = 0.17), number of children (χ2(4) = 59.17, p < 0.001, Cramer’s V = 0.21), sexual orientation (χ2(2) = 11.50, p = 0.003, Cramer’s V = 0.09), and professional situation (χ2(3) = 101.93, p < 0.001, Cramer’s V = 0.27). Men were older than women and were more likely to be married, while most women were more likely to be dating. Men were more likely than women to have one or more children. Although the majority of men and women were heterosexual, more men identified themselves as homosexual, while more women identified themselves with other sexual orientation. Regarding the professional situation, women were more likely to be studying and less likely to be retired, in comparison to men.

Gender was also significantly associated with alcohol consumption (χ2(4) = 165.68, p < 0.001, Cramer’s V = 0.35) and drugs use (χ2(4) = 11.67, p = 0.020, Cramer’s V = 0.09), being victim of an aggressive/violent behavior by a previous intimate partner (χ2(1) = 21.02, p < 0.001, φ = 0.12), having perpetrated an aggressive/violent behavior towards a previous intimate partner (χ2(1) = 0.50, p = 0.480, φ =  − 0.02), and having received psychiatric or psychologic treatment for some reason (χ2(1) = 15.66, p < 0.001, φ = 0.11). Men were more likely to report a higher alcohol and drugs consumption and to have perpetrated an aggressive/violent behavior in a previous intimate relationship, in comparison to women. On the other hand, women were more likely to identify themselves as being victims of aggressive/violent behavior by a previous intimate partner and as having received psychological or psychiatric treatment.

No significant differences were found considering residence area (χ2(1) = 4.02, p = 0.045, φ =  − 0.05), financial dependency of the partner (χ2(1) = 0.85, p = 0.355, φ = 0.02), the existence of previous judicial processes as IPV victim (χ2(1) = 0.17, p = 0.677, φ = 0.01) or perpetrator (χ2(1) = 1.03, p = 0.310, φ =  − 0.03), and being a victim of maltreatment by an adult before the age of 15 (χ2(1) = 0.85, p = 0.358, φ = 0.03).

Procedures

This study has a cross-sectional design, and it was conducted through a web-based survey. Ethical approval was obtained from the Ethics Committee of the host institution. A national convenience community sample was collected. Using the available data from the last Census (2011), a minimum sample size of 1067 was determined considering a 95% confidence level and a 3% margin of error. Participants were recruited in 2020, between March and June, through an online data collection platform (LimeSurvey©), which was placed on the website of the host institution, and on which the set of questionnaires was available. A Facebook page was created to publicize the study, and 30 euros were spent on advertising and paid boosting campaigns. An invitation to participate was also sent through e-mail, containing a brief explanation of the study and the link to the survey. This e-mail was sent to the mailing lists of the research center/host institution, all the city councils of the country, several academic associations, senior universities, charities, and private organizations, as well as to the contacts of researcher’s acquaintances. All potential participants were also asked to share the study page or to forward the invitation e-mail to their contacts, to obtain a largest and diverse sample. Whenever a potential participant accessed the study link, the introductory page presented the purposes of the study, the research team, and the underlying ethical issues, including the voluntary and confidential nature of the study. Eligibility criteria (being over 18 years old, being Portuguese, and being currently in an intimate relationship) were clarified. Only those who gave their informed consent completed the questionnaires. No economic or other compensation was provided. At the end of the survey, a victim support freephone was provided for those wanting to ask for help or information about IPV.

Measures

Sociodemographic Questionnaire

Participants were asked to provide sociodemographic information and other relevant information related to the history of IPV, namely gender, age, current relationship duration, marital status, number of children, sexual orientation, professional situation, financial dependence from partner, district/region they were living in, area of residence (rural or urban), alcohol and drugs use, history of IPV (being identified in a court case as a victim or perpetrator or self-reporting having been a victim or perpetrator of aggressive/violent behavior by/towards a former intimate partner), victimization by an adult before the age of 15, or history of psychological/psychiatric treatment. These questions were developed by the researchers and designed to include variables pointed out in the literature as predictors of IPV in international studies (e.g., Capaldi et al., 2019; FRA, 2014; Garcia-Moreno et al., 2006).

Conflict Tactic Scales-Revised

The Conflict Tactic Scales-Revised (CTS-2; Straus et al., 1996; Portuguese version by Paiva & Figueiredo, 2006) is a 78-item self-report questionnaire widely used in IPV studies. It allows measuring the use of psychological aggression, physical assault, sexual coercion, injury, and negotiation, within the couple. Considering the past year, participants answered whether and how often they (perpetration) or their partner (victimization) had engaged in the described actions. Items are rated on an eight-point scale ranging from Once in the last year (1) to More than 20 times in the last year (6), including the option Not in the last year but have occurred previously (7) and Never occurred (8). Each scale can be divided into “minor” or “severe” acts of violence, indicating the injury risk (except negotiation, which does not include violent acts). In this study, only data of the global violence-related subscales were analyzed. The past-year prevalence was assessed as the rate of participants that reported having perpetrated or having been a victim of at least one act of violence or abuse during the 12 months before the present study (i.e., scores of 1–6 on the scale items). The lifetime prevalence of IPV was assessed as the rate of participants who reported having experienced at least one act of violence or abuse by an intimate partner (i.e., scores of 1–7 on scale items). Frequency was also assessed, considering the mean number of occurrences of the violent or abusive acts, among those participants who reported to have perpetrated or been the victim of at least one of those acts in the past year. Cronbach’s alphas of the original version ranged from 0.79 to 0.95 (Straus et al., 1996). In the present study, Cronbach’s alphas were computed based on frequency, for victimization and perpetration, and ranged from acceptable (α = 0.72) to excellent (α = 0.95) (Nunnally & Bernstein, 1994), except for sexual coercion (both victimization and perpetration presented α < 0.50) and perpetration of psychological aggression (α = 0.64). Nonetheless, all scales were used to describe the prevalence, as they represent the presence of specific behaviors. For comparisons and inferential statistics, only sexual coercion was excluded, as its reliability was not acceptable (Nunnally & Bernstein, 1994).

Data Analyses

IBM SPSS 22 was used to compute missing values analyses, for descriptive statistics, mean comparisons between genders, scales’ reliability, and prevalence rates. Only 1.44% of participants in the final sample presented missing values, and missing data accounted for 0.03% of the possible answers gathered with this sample. The Missing Completely at Random (MCAR) test for missing data was non-significant (χ2(1454) = 1435.14, p = 0.631), suggesting that, on average, data observed in cases with and without missing values are comparable with each other (Meyers et al., 2013). The method of linear interpolation was used to estimate the missing values.

The prevalence of IPV was assessed considering the past year and lifetime. To calculate both prevalence types, items were transformed into dichotomic variables, with a score of 1 assigned if one or more of the acts in the scale had been reported. Prevalence by gender only included males and females due to the small number of participants who identified themselves as another gender (n = 2). Prevalence by sexual orientation was assessed in heterosexual and non-heterosexual participants (those who identified themselves as homosexual or having other sexual orientation). Directionality patterns of violence were determined by combining the presence of reported past-year victimization and perpetration behaviors (perpetration-only, victimization-only, bidirectional violence). For heterosexual participants, dyadic concordance types (man-only perpetration, women-only perpetration; both-violent; Straus, 2011, 2015) were also analyzed. In order to estimate frequency, and following the guidelines suggested by Straus et al. (1996), the value of the original scale was transformed into the midpoint of each category and then added for each scale, with the total ranging from 0 to 825.

Finally, to examine which factors were associated with victimization or perpetration, separate univariate logistic regression analyses were run on the dependent variables (victimization and perpetration in the past-year). Nagelkerke’s R2 adjusted was used as an indicator of effect size, and the statistical significance of individual predictors was evaluated by calculating the odds ratios with 95% confidence intervals (Field, 2018). A p-value of 0.05 was set as the cut-off point for statistical significance.

Results

IPV Prevalence Rates

The past-year prevalence rate of any form of IPV was 64.4% for victimization and 64.6% for perpetration (Table 2). The lifetime IPV prevalence rate was 74.6% for victimization and 76.1% for perpetration in the total sample (Table 2). The most common type of violence in the past year, and consistently across groups (male, female, heterosexual, non-heterosexual), was psychological violence (e.g., insults, scream at). The same was observed for the lifetime data. Prevalence by gender, sexual orientation, and according to the type and severity of violence were also identified. No significant differences were found in the past-year or lifetime prevalence regarding gender or sexual orientation (heterosexuals vs non-heterosexuals), considering the total and each specific type of violence.

Table 2 Past-year and lifetime prevalence of total violence and by types of violence in the total sample, by gender and by sexual orientation
Full size table

Directionality of Violence

The majority of participants reported a bidirectional pattern of violence (defined by the report of simultaneous involvement in victimization and perpetration), with no significant differences between hetero- and non-heterosexual participants (χ2(2) = 1.99, p = 0.369, φ = 0.05). The analysis of dyadic concordance types showed a similar percentage of male and female perpetration (see Table 3).

Table 3 Dyadic concordance types and directionality patterns of violence in heterosexual and non-heterosexual participants
Full size table

Participants were also asked to identify who was the first to hit the other the last time any physical violence behavior occurred, if ever. Twenty-four participants did not answer this question. Considering gender, eight men and 16 women did not respond, and considering sexual orientation, 20 heterosexual and four non-heterosexual participants did not answer this question. For those who answered (n = 1368), the majority (93.2%) said that it had never happened, and the same percentage (3.4%) assumed that they had hit first or that the other had. Across groups, most participants (over 90%) reported that no physical violence has ever happened. In the female and the heterosexual groups, more participants reported that they did hit first, as opposed to those who reported their partners had done it (3.6% and 3.4% vs 3.1% and 3.2%, respectively). In the male and the non-heterosexual group, more participants reported that their partners had hit first than those who reported that they had done it in the first place (4.3% and 5.6% vs 3.0% and 3.7%, respectively).

IPV Frequency

The frequency of any form of IPV was on average 18.20 (SD = 30.69; range: 1–495) for victimization and 15.11 (SD = 23.80; range: 1–495) for perpetration, for the complete sample. Whenever present, and consistently across groups, sexual coercion, and psychological violence seem to be the most recurring types of violence in the past year. It should be noticed that in over half of the identified behaviors (57.6%), both for perpetration and victimization, only a single occurrence of the identified behavior was reported by most of the participants. This indicates that the other participants reported the occurrence of the behavior frequently enough to increase the means to the observed values. Table 4 presents the frequency for each type of violence by gender and by sexual orientation. No significant differences were found regarding gender and sexual orientation, considering the total and each type of violence.

Table 4 Frequency of violence victimization and perpetration in total and by types of violence in the total sample, by gender and by sexual orientation
Full size table

Risk Factors of Past-Year Prevalence of IPV

Regarding the factors associated with victimization, the univariate analysis revealed that being younger, cohabitating with the intimate partner, using drugs, having perpetrated an aggressive/violent behavior towards a previous intimate partner, having been a victim of maltreatment by an adult before the age of 15, and having had psychiatric or psychological treatment for some reason, was associated with an increased likelihood of having been a victim of any act of violence in the past year. In the multivariate model, all the abovementioned variables, except having had psychiatric or psychological treatment, were significantly associated with being a victim of any act of violence in the past year (see Table 5). The final logistic regression model was significantly reliable (χ2(6) = 69.61, p < 0.001; Nagelkerke R2 = 0.07), correctly predicting 65.7% of the cases. The results of the Hosmer–Lemeshow goodness-of-fit test indicated that the multivariable model fit the data well (χ2(8) = 4.81, p = 0.778).

Table 5 Univariate and multivariate logistic regression analysis of the association between sociodemographic, judicial, and violence-related variables and past-year victimization
Full size table

Regarding perpetration, the univariate analysis revealed that being younger, having perpetrated an aggressive/violent behavior towards a previous intimate partner, having been a victim of maltreatment by an adult before the age of 15, and having had psychiatric or psychological treatment for some reason, increased the likelihood of having perpetrated any act of violence in the past year. In the multivariate model, all variables but having had psychiatric or psychological treatment were significantly associated with increased odds of having perpetrated any act of violence in the past (see Table 6). The final model was significant (χ2(4) = 44.49, p < 0.001; Nagelkerke R2 = 0.04), correctly predicting 64.7% of the cases. The results of the Hosmer–Lemeshow goodness-of-fit test indicated that the multivariable model fit the data well (χ2(8) = 10.28, p = 0.246).

Table 6 Univariate and multivariate logistic regression analysis of the association between sociodemographic, judicial, and violence-related variables and past-year perpetration
Full size table

Discussion

This study presents findings on the extent of physical, sexual, and emotional intimate partner violence in Portugal. The use of a standardized measure widely used internationally is a strength, allowing for cross-country comparisons. To the best of our knowledge, this is the first study assessing the IPV prevalence (including victimization and perpetration) by self-report and in a large national and gender-inclusive adult community sample. The main findings of our study show that most of the participants have already been victims or perpetrators of at least one violent behavior in an intimate relationship throughout their lives. Indeed, a high prevalence of IPV (above 62.5%) was identified, regardless of gender or sexual orientation. Both past-year and lifetime prevalence of sexual and physical victimization are higher than those identified on average in other European countries and those previously estimated for Portugal (FRA, 2014). Nonetheless, comparisons between these studies’ findings are not possible due to differences in the assessment strategies and the definition of the included forms of violence, as well as participants’ characteristics. For example, the study conducted by the FRA (2014) only assessed violence against females. Moreover, the CTS2 is a measure that includes minor forms of psychological (e.g., “I screamed or yelled at my partner”) and physical violence (e.g., “I grabbed my partner by force”) and sexual coercion (e.g., “Insisted on sex when my partner did not want to (but did not use physical force)”), which may not have been considered in other studies. Accordingly, minor forms of violence may be more prevalent than more severe ones, which may have contributed to a higher prevalence rate in the current study. When comparing findings of this study concerning IPV past-year prevalence to those found by other studies using the same measures and definitions of types of violence, the findings for victimization and perpetration are more similar (Costa et al., 2015; Machado et al., 2019; Paiva & Figueiredo, 2006; Straus et al., 1996). As regards the comparisons with other studies conducted in Portugal (although not at a nationwide level), a higher prevalence is found, albeit slight, of psychological and physical violence. Also, there is a consistent identification of less than half of injury occurrences in comparison with findings from previous studies (Costa et al., 2015; Machado et al., 2019; Paiva & Figueiredo, 2006). Either way, findings of this study are particularly worrisome not only for the extensively known negative consequences of intimate partner violence (e.g., Miller & McCaw, 2019; WHO, 2013), but also because the existence of family violence increases the risk for other forms of violence (e.g., child abuse; Machado et al., 2007), and a substantial proportion of the current study’s sample (41.6%) had children. Furthermore, the absence of differences among individuals with distinct sexual orientation in the IPV prevalence and frequency also seems to suggest that all people are equally vulnerable to be in a violent intimate relationship, regardless of their sexual orientation.

The past-year prevalence of IPV also needs further reflection. First, it is important to take into consideration that the report of a single behavior is enough to identify a participant as a victim or a perpetrator. And despite the high IPV prevalence identified, more than half of the reported behaviors only occurred once in the past year. Therefore, the prevalence should be interpreted considering frequency (Costa et al., 2015). The most prevalent types of violence are also the most recurrent: psychological violence and sexual coercion. Nonetheless, the dispersion of data suggests that there are differences between participants on the frequency within each type of violence. The large majority of participants identifying one occurrence of the reported behavior in the past 12 months and the difference to the levels of frequency that are found allows to assume that different patterns and severities of violence are present. On the one hand, those who are engaged in repeated violence behaviors seem to repeat them enough to raise the mean of frequency. On the other hand, most participants identifying a single act of violence may not reflect a true context of intimate violence, in the sense of an existent pattern of violence within the relationship. Second, although many participants reported that physical violence never occurred when this was directly questioned, higher levels of physical violence were found in this study. People seem more prone to recognize the existence of violent behavior than to recognize that same behavior as being a violent one, reinforcing the idea that asking behaviorally specific questions enhances disclosure (Straus et al., 1996). The denial or rationalization of the behaviors as non-violent may contribute to the considerable gap between the rates of identified IPV and the rates of judicial intervention (only 2.2% of the participants were identified as victim of an IPV crime and 1.1% as a perpetrator).

Other relevant findings in the current study concern gender. Much controversy and resistance persist regarding the symmetry found in IPV perpetration by women and men. Factors such as differences between findings from community samples and those of criminal statistics (where males are predominantly identified as perpetrators), the concern that this recognition may underestimate the higher risk for women to suffer injury or to be murdered by an intimate partner or even its seeming incompatibility with gender-paradigm-driven theories and policies is probably among the most influential reasons accounting for this resistance (Hamel, 2020; Straus & Mickey, 2012). In the current study, no significant differences were found regarding gender neither in past-year prevalence nor in lifetime prevalence for the different types of IPV, including injury. This is opposed to the widely spread idea that women are more likely to get physically injured as a consequence of IPV (Archer, 2000; Hamel, 2020). However, current findings are in line with those from meta-analyses (Archer, 2000) and several reviews (Chan, 2011; Esquivel-Santoveña et al., 2013; Straus & Mickey, 2012) that pointed out to symmetric gender perpetration prevalence in community samples. Furthermore, the non-existence of differences regarding sexual orientation reinforces the idea that IPV in non-heterosexual relationships is similar (in prevalence, frequency, severity, impact, and dynamics) to IPV in heterosexual relationships, with the threat of “outing” being the mainly difference identified among these couples (Merrill, 1998; Nunan, 2004; Renzetti, 1998). These findings reinforce the recognition that the argument of sexism and unequal gender relations as main factors underlying IPV is not supported by research (Hamel, 2020; Straus & Mickey, 2012).

When the directional pattern of IPV is analyzed, symmetry in gender is also identified: exclusive male and female-only perpetration patterns occur in similar proportions, but most men and women, as well as most heterosexual and non-heterosexual participants, report being involved in both perpetration and victimization. These findings are also in accordance with those from meta-analyses (Archer, 2000, 2002) and review studies (Chan, 2011; Esquivel-Santoveña et al., 2013; Straus & Mickey, 2012) that tend to identify bidirectionality in community samples as the most prevalent pattern of IPV. These findings are also in line with past research in Portugal (Costa et al., 2015; Machado et al., 2019). Equally important is to consider that, via self-report methods, more women recognize having been the first to hit their partner when physical violence occurred, and more men recognize that their partner hit first when this occurred. This points out that, in these cases, physical violence should not be immediately interpreted as self-defense or retaliation, at least in a commensurate way (as physical violence may be used to retaliate after psychological aggression, for example). Thus, this finding reinforces the conclusions of previous research (Hamel, 2020; Straus & Mickey, 2012), showing no evidence that self-defense or retaliation is the primary motive for women to perpetrate physical aggression towards their partners. The bidirectionality of IPV poses a difficulty for legal systems, in which individuals are deemed to be identified either as victims or as perpetrators, and risk assessment and interventions are planned accordingly. It also challenges the victims’ protection systems (traditionally more dedicated to women) and the probation services (usually more focused on male perpetrators) to re-think their intervention and design gender-inclusive responses to accommodate reality. Additionally, legal systems are usually less prepared to identify emotional and psychological aggression. Therefore, the high rates of this form of IPV are also a challenge and call for an interdisciplinary work, aiming to assess and generate evidence of this type of violence and its consequences.

When examining which factors were associated with IPV victimization and perpetration, the strongest predictor of violence (in both cases) is having perpetrated IPV in previous relationships. The presence of violence in different intimate relationships raises questions about the vulnerability of some people to engage in non-healthy relationships and/or to cope in a dysfunctional way with marital conflict. Attention should also be given to the presence of victimization before the age of 15, which is significantly associated with both victimization and perpetration in current relationships. These findings agree with previous research (e.g., Widom et al., 2014), pointing out that childhood trauma may be an important factor in understanding both victimization and perpetration in intimate relationships. Future studies should consider various forms of traumatic childhood experiences, including neglect but also the absence of experiences of warmth and affection, when trying to relate early life experiences with current IPV. The mechanisms through which these and each of the other identified risk factors contribute to victimization and/or perpetration of IPV must be analyzed in more detail, since they may provide important targets for prevention and intervention efforts. Furthermore, the complex nature of IPV precludes the identification of single causes but rather points out to a multi-causality in which the variables associated with psychological characteristics seem to be better predictors of violence than the social-structural characteristics (Dutton, 2011; Ehrensaft et al., 2004). This may help to understand why neither perpetration nor victimization was strongly predicted by the analyzed models.

Despite the contributions introduced by the current study, some limitations should be considered. First, the sample was self-selected, recruited through non-probability sampling methods. People more prone to participate could be those to whom this topic generates more interest, for whatever reason, including personal experience, thus eventually contributing to higher prevalence rates. Furthermore, the web-based nature of the study excludes a proportion of the population, potentially older, less educated, without easy internet access or those who are less familiarized with internet and websites. Also, the data collection period included the first 4 months of the pandemic period. Specific questions about the impact of COVID-19 on intimate relationships could have been asked if it had been possible to anticipate the onset of the pandemic and consequent lockdown measures during data collection. This would be important to contextualize the discrepancy between the decrease of reports to the police during the first lockdown period in Portugal (Capinha et al., 2021) versus the increase of requests for help, during those months, to one of the main non-governmental organization (NGO) that support victims nationwide (Ribeiro et al., 2022). Such information would allow for a better understanding of how the COVID-19 pandemic may have influenced the current findings. Bearing in mind that criminal statistics point to an overall decrease in IPV (Capinha et al., 2021) and there was also a decrease of 4.9% in requests for help to the NGO, specifically regarding IPV, during 2020, one cannot exclude the possibility that less aggressive behavior occurred in this period. If that would be the case, it would be contrary to what was anticipated by some authors (e.g., Campbell, 2020), despite the initial increase that was observed on requests for help (Ribeiro et al., 2022). Future studies will have the opportunity to clarify whether these findings extend to other countries during the COVID-19 pandemic. Furthermore, future research should be designed considering the dyadic nature of IPV, using both partners’ reports. This approach has been identified as the best approach to minimize possible bias in the reports of IPV, although with recognized constraints regarding the possibility of being carried out (Straus & Mickey, 2012), particularly those related to victims’ security. Other forms of violence (e.g., online harassment or control) should also be investigated as the relative newness of these phenomena prevents the existence of proper knowledge about its prevalence in all genders. Addressing these issues will contribute to more effective policies and intervention/prevention efforts.

Despite the abovementioned limitations, this study adds to the growing body of evidence that supports the need to generate more gender-inclusive policies and interventions in the IPV field. Empirically driven policies would be more effective tackling IPV. Current findings also provide relevant insights into a more comprehensive knowledge of the IPV prevalence and patterns in the Portuguese population.