Violence Against Women: Gravity, Prevalence and Socioeconomic Factors in the Ecuadorian Andes

Abstract

Background: This study aims to identify the socioeconomic factors influencing the likelihood of women being victims of violence in the canton of Riobamba, located in the Ecuadorian highlands, as well as the gravity and prevalence of different types of violence. Methods: For each type of violence, a questionnaire was applied on social, economic and family structure characteristics, personal and family experiences of violence, and events of abuse in work, educational, community, public institution, intimate partner and family environments. From the gravity and prevalence estimates, multiple linear regression modelling (gravity) and discrete choice modelling (probability of being a victim and prevalence) were used. Results: Findings indicate that the likelihood of being abused increases among divorced women and with experiences of violence in the family. The prevalence and severity of all types of violence rise with higher income levels due to men’s loss of control. Conclusions: Strategies for access to education could influence the different types of violence, as it reduces the level of gravity in all cases.

1. Introduction

Violence against women in Ecuador constitutes a serious social and human rights problem that affects a large sector of the female population and manifests in various forms, such as physical, psychological, sexual, and economic violence. According to data provided by national and international organisations, gender-based violence in Ecuador is at worrying levels, despite the existence of regulatory frameworks that guarantee women’s rights and prevent violence, such as the Comprehensive Organic Law to Prevent and Eradicate Violence against Women.

Recent figures indicate that 60% of women have been victims of violence (INEC 2019), with psychological violence occupying the first place, with 56.9% of the female population, followed by physical violence with 35.4%, sexual violence with 32.7% and economic–patrimonial violence with 16.4%, which evidences the challenges in institutional and social terms.

Edeby and Sebastiån (2021) and Boira et al. (2021) indicate that the main factors that explain this behaviour are related to age, education, marital status and the patriarchal culture and social norms that perpetuate violence. In addition, regarding family and intimate partner violence, the causes identified in the work of (Cumbicus-Pineda et al. 2021) emphasise alcohol and drug use as well as childhood experiences of violence.

Several structural, cultural and social factors contribute to the persistence of gender-based violence in Ecuador. Firstly, gender inequalities are profound, and Ecuadorian society continues to be permeated by patriarchal norms that limit the full exercise of women’s rights. Secondly, the normalisation of violence in many social contexts makes it difficult to recognise violence and contributes to women not perceiving that they are being abused.

These social norms limit women’s empowerment, access to education, and economic autonomy, fostering dependence and violence. This study provides findings that help us understand the under-explored phenomenon of violence against women in the region, examining its different typologies and contexts through indicators such as gravity and prevalence. Another aim was to identify the factors that determine this behaviour, incorporating individual, economic, social and even family structure variables, such as the number of children. The latter has been scarcely studied in the literature on violence against women, although it is a relevant variable in studies on access to the labour market and empowerment, which, according to (Mosquera 2018), is central to curbing violence against women in countries such as Ecuador.

The Ecuadorian highlands are a geographically and culturally diverse region in the Andean zone of Ecuador, where the indigenous population plays a key role in the region’s social, economic, political, and cultural dynamics. It is characterised as one of the most traditional and culturally rooted areas of the country, where women face various forms of structural and gender-based violence, often exacerbated by the rural context, patriarchal customs and the scarce presence of support services.

In the case of the Andean province of Chimborazo, located to the south of Ecuador’s capital, the 2019 prevalence of violence was below the national average, at 5 out of 10 women, and no cases of economic, marital or sexual violence were reported, which could be because women are afraid to report violence or believe it is pointless because they cannot escape. Nevertheless, there were incidences of psychological (46.9%), physical (39.1%) and gynaecological-obstetric violence (58.2%), in this case higher than the national average. Specifically, intimate partner violence, which stood at 28.9% in 2019, was above the national average and the average for the Ecuadorian highlands (INEC 2019).

The figures for the province of Chimborazo and its capital, Riobamba, like the rest of the Ecuadorian highlands, are often explained by ethnic and multicultural factors, where traditional gender roles predominate in the development of productive and reproductive activities. Broadly speaking, women must dedicate themselves to caring for the home and to what Medina-Hernández et al. (2021) called the “feminization of agriculture”, while men dedicate themselves to productive work outside the home.

This research aimed to identify the predictors of violence against women, the gravity and prevalence of different types of violence (psychological, physical, sexual, and economic), and the contexts in which abuse occurs, whether in the community, public institutions, educational spaces, workplaces, couples or families.

Despite the social relevance of violence against women and the Ecuadorian reality, quantitative studies on the determinants are scarce but crucial to generating a deep understanding of the phenomenon, creating effective interventions and evidence-based public policies, and fostering a culture of justice, equality and respect for women’s rights.

Literature Review

Physical violence is conceived as an aggression that generates bodily harm, which can even lead to death. It is related to the so-called ‘cycle of violence’, in which there is an affective relationship with the aggressor that perpetuates the violence (Walker 1979; Watts and Zimmerman 2002; Krahé 2018), which is why it is usually related to family or intimate partner violence. Similarly, sexual violence is conceived as actions of sexual origin that are carried out without the consent of the victim and is also related to the desire for control or domination inherent to patriarchal social and cultural structures, in which strong patterns of inequality are evident (Segato 2016; Watts and Zimmerman 2002; Krahé 2018).

Psychological violence and gaslighting are linked to situations of humiliation or devaluation, including threats or stigmatisation (Biroli 2018), originating not only in family or intimate partner contexts but also in work, education, community or institutional contexts, repeated over time, with deteriorating mental health in the form of lowered self-esteem, anxiety, depression and even suicide (Hirigoyen 2006).

Economic or patrimonial violence directly affects economic empowerment, as decisions about the allocation of resources are made solely by the aggressor. Thus, the woman loses control over assets, and even over her own wages. Sometimes the money from work is physically stolen by the partner, which Lupica (2015) has called economic autonomy (part of economic empowerment and relates to the ability to have control over one’s own financial resources). This not only implies the ability to control income but also the possibility of carrying out work and household chores.

In any of its typologies, violence against women is a multifactorial phenomenon addressed by different theories or models that emphasise cultural, social, demographic, economic and individual factors as determinants. The ecological model (Heise 1998) considers that it is the result of antecedents of violence associated with the family and the couple and influenced by the community environment and the social and patriarchal structures that perpetuate inequality.

For its part, the theory of the cycle of violence (Walker 1979) emphasises that the relationship of aggression–repentance and reconciliation between victim and perpetrator reinforces dependence and limits the possibilities of leaving the circle of violence. This relationship is aggravated by the victim’s inaction or lack of control, resigning the woman to situations of abuse, as stated in the theory of learned helplessness (Walker 1979).

On the other hand, theories of hegemonic masculinity (Connell 1987) and the theory of male domination (Bourdieu 2000) argue that violence against women is the result of the social construction of male superiority and the need for control over women. Furthermore, gender roles exacerbate these inequalities by assigning certain reproductive activities to women, while any other productive responsibilities are considered inappropriate for men, which can lead to violent situations, as proposed by the gender socialisation model (Eagly and Wood 1999).

From an economic point of view, violence against women can be explained by models such as economic dependency (Anderson 1997), which proposes the idea that the lower the economic empowerment, the more difficult it is to overcome the environment of violence because of the absence of financial autonomy. The theory of human capital, formulated by (Becker 1993), proposes that when access to education is improved, it is possible to generate job opportunities and strengthen economic empowerment, reducing vulnerability and the probability of being a victim of violent situations. Additionally, the model of poverty and violence (Duflo 2012) establishes a direct relationship between poverty and violence against women.

In these theories or models, sociocultural, economic, individual and family factors stand out, linked to cultural and social structures and norms (Yoosefi Lebni et al. 2025; Alesina et al. 2016), which establish that gender roles and patriarchal systems mean that control must fall to men (Banarjee 2020; Mannell et al. 2022). Cultural factors include the idea that violence is not an isolated phenomenon but a product of the control exercised over women and a form of domination rooted in the culture in which the woman lives (Lagarde 2005), which contrasts with female empowerment that attempts to transform traditional gender roles (Hernández-Medina et al. 2023b).

In addition, family dynamics are considered predictors of these patterns of behaviour, including silence in the face of these situations or their normalisation. Other factors include the couple’s wellbeing (or lack thereof), drug and alcohol addictions, the number of children and violence experienced in childhood (Yoosefi Lebni et al. 2025; Mannell et al. 2022; Alkan et al. 2022; Abdi et al. 2021; García-Moreno et al. 2006; Mwalupani et al. 2024).

In this case, authors such as (Mojahed et al. 2020; Smith-Marek et al. 2015; Till-Tentschert 2017) have shown that experiencing violence in childhood, whether as a victim or witness, increases the likelihood of experiencing abusive behaviour and violence in adulthood. This is mainly associated with intimate partner violence, with important consequences for mental health, family structure and interpersonal relationships.

With respect to the number of children, studies address this as a risk factor for experiencing violence. In this regard, Awang and Hariharan (2011), for the case of Malaysia, identify that women with the highest prevalence of abusive situations are those who do not have children or who have many children. The number of children has a direct relationship with empowerment and employability. Studies such as those by (Hernández-Medina et al. 2023a) identify that women with children under 6 years of age report a lower probability of becoming employed and lower levels of empowerment. Moreover, many pregnancies in quick succession put a physical strain on women, which makes it harder for them to find the strength to leave an abusive situation. The lack of sleep and the mental load of caring for many children and keeping them fed and healthy make it difficult for a woman to plan an escape from an abusive situation.

Individual factors include age and marital status since divorced women are more likely to be victims of violence, as are women with limited interpersonal skills and low self-esteem. This is accentuated when there is a history of having been exposed to violence in the family nucleus during childhood (Chipokoso et al. 2024; García-Moreno et al. 2006; Yoosefi Lebni et al. 2025; Mannell et al. 2022).

From an economic point of view, women’s poverty and lack of economic independence, as well as their level of education, influence the probability of being a victim of violence (Briere and Jordan 2004; Montesanti and Thurston 2015). The greater economic autonomy achieved through education should tend to reduce the risk of violence, although the effects are often not so clear and are often conditioned by gaps or contexts that make them more complex and contradictory (Ince-Yenilmez 2020; Mannell et al. 2022). Indeed, authors such as Haobijam and Singh (2021), Dang and Le (2024) and Muluneh et al. (2021) identified a reduction in the risk of domestic violence as education and health levels rise, while the effect of income level does not always have an impact on reducing that risk.

Specifically, in studies such as (Zhou et al. 2021), an additional year of education reduced physical and sexual abuse in China, while (Oluwagbemiga et al. 2022) identified a decrease in violence in Nigeria amongst women who report formal education. These effects are evidence of the extent that education not only provides women with legal knowledge about abuse but also empowers them in decision-making, including partner decision-making. As (Erten and Keskin 2022) show in the case of Türkiye, training on rights and regulatory frameworks against violence did not result in a reduction in this behaviour, as it did not lead to changes in women’s control over their lives. In addition, Ackerson et al. (2008) warn of the need to address not only literacy from the women’s perspective but also to analyse the education of the couple, as higher levels of education in men also tend to reduce the risk of abuse and violent behaviour.

It is important to highlight the behaviour of educational gaps. Although (Rapp et al. 2012) hypothesised that a higher level of education amongst women compared to men makes them more likely to be subjected to intimate partner violence, their results indicate a different behaviour–that an educational gap in favor of women does not increase the risk of this behaviour.

Regarding the level of income, Vyas and Heise (2016) and Abramsky et al. (2019) warn that although generating one’s own financial resources increases women’s economic empowerment, this can generate family tensions. The idea of women working contrasts with traditional gender roles, in which women must assume reproductive and non-productive activities. It reduces men’s control and the perception of dependence, which is likely to result in higher levels of violence, especially in societies where gender norms and roles are permissive in terms of these patterns of behaviour (Muluneh et al. 2021). Hence, there is backlash from these men, which is largely determined by contexts in which divorce is not permitted, restricted or socially inappropriate (Bhalotra et al. 2019; Lenze and Klasen 2017). On the opposite effect of a larger income, Abramsky et al. (2019) reiterate that the risk of violence rises if the woman’s income level is higher than the man’s, the extreme case being when the man is unemployed. There is evidence that women’s employment can achieve a reduction in the risk of experiencing violence through women’s economic empowerment (Villarreal 2007; Iregui-Bohórquez et al. 2019; Dildar 2020); however, according to Alonso-Borrego and Carrasco (2016), this is related to both partners being employed. Akram (2021) reports another finding regarding women who are employed. Although this situation may increase the risk of suffering violence, if the women started working before marriage, they are less likely to experience this type of behaviour.

Regarding the type of occupation, Iregui-Bohórquez et al. (2019) propose that if the type of employment, services and/or commerce is less valued, under the idea that the woman’s contribution to the household is lower, the risk of violent behaviour in the household rises.

These factors can mostly explain the risks of becoming a victim of violence and the gravity and prevalence in family and couple contexts. However, there are other scenarios in which, although the severity could be lower, it does not imply a lower prevalence, such as educational, community, work, or public sector environments. In these cases, the social structures and norms condition actions, stimulated by a weak regulatory framework and questioned institutional quality and trust (Flood and Pease 2009).

Specifically in educational settings, the role of institutions is essential. Avoiding oversight and having policies that regulate the offender’s actions are fundamental, as established by authors such as Barbosa et al. (2024) and De Andrade et al. (2024). In work environments, women in lower positions, younger and less experienced in organisations with male leadership guided by patriarchal social norms are more likely to suffer some form of violence (Kaphle et al. 2014; Adams-Prassl et al. 2023).

2. Materials and Methods

In order to, first, analyse the gravity and prevalence of violence against women in any of its typologies and then to, second, identify the socioeconomic determinants that could explain this behaviour, this study considered a sample of 200 women from the urban sector of the Riobamba canton, located in the Ecuadorian highlands. It is a region characterised by a significant Indigenous population that determines different cultural and social norms to which Mestizo women are also subjected, thus shaping gender roles and the achievement of equality. The survey required the consent of each participant and ensured the anonymity of the respondents.

The information was collected through a survey on social and gender violence applied by the Geoprospectiva Information Centre within the framework of the Social Co-investment Programme in the areas of influence of the Observatories of Social and Gender Violence in Mexico (Geoprospectiva Information Centre 2008). This instrument was structured in two sections: the first was associated with questions on socioeconomic explanatory variables, related to education, marital status, occupation and income, marital status and number of children and the second on whether there are cases of violence in the family, whether they have personally suffered violence and, if so, what type(s).

The second section proposes a series of questions related to the frequency of situations associated with the different types of violence (psychological, sexual, economic and physical) that can be experienced in different environments, such as the community, public institutions, educational institutions, the workplace, with the partner and with the family. When we speak of public institutions as a setting for the severity and prevalence of violence against women, we are referring to public entities and organisations where women may be exposed to violence or discrimination, either by employees or even in their interaction with these institutions. These institutions include government agencies and public services, such as the police, the judiciary, hospitals, schools, social welfare agencies, and other public services. Responses are measured on a Likert-type scale that considers options such as never, only once, several times and many times and discusses some of the following items (Table 1).

Table 1. Main items in the questionnaire.

Violent situations performed by unknown men in the community

Said rude or derogatory words about me or women in general

Looked at me morbidly, making me feel uncomfortable

Said obscene or offensive compliments about my body or of a sexual nature

Forced me to have sex

Violent situations performed by public servants in public offices

They set requirements that are difficult for women to meet

They deny me the service because I am a woman

They behave arrogantly towards me because I am a woman

They make sexist or mocking comments about women

Violent situations carried out by managers or work partners

I am pressured or forced to do more work than men

I am humiliated or made fun of because I am a woman

They deny me permission to take care of my children or get angry when I do

I am paid less than men doing the same job

I am conditioned to have sexual contact at work

I have been retaliated against or punished for refusing sexual requests

Violent situations performed by teachers or teaching partners

They make fun of me because I am a woman

They make me feel less because I am a woman

Because I am a woman, they belittle my opinions or my ideas

They give me lower grades than men even though my work is equal

They touch me, touching, fondling or groping me without my consent

Violent situations performed by partner

Forbids me from seeing my acquaintances or forces me to tell him who I am always with

Unreasonably accuses me of cheating on him with other men

Insults me, calls me names or offends me

Has tried to kill me

Has used physical force to have sex

Forbids me to use contraception

Violent situations performed by relatives

They insult, humiliate or offend me

They take my things, valuables, houses or documents that prove they are mine

They take my money

They do not let me work or study

They force or pressure me to have sex with them

They force or pressure me to have sex with other people for money

Note. Taken from (Geoprospectiva Information Centre 2008).

The dependent variables considered from the data are having experienced some type of violence, which was answered directly by the women surveyed, and two additional variables estimated with the information collected in section two of the instrument. The former is related to the gravity of the violence, which is the result of the indicated frequency multiplied by the severity of the situation. This severity is proposed by (Zhou et al. 2021), who classify the situations on a scale of 1 to 4, where 1 represents the least serious and 4 is the most serious, such as having experienced death threats or attempted murder. The situations or events are conceptualised for each setting and each typology was grouped and associated with a level of severity.

The third dependent variable is the prevalence of violence, which is given by having experienced in some situation (event) and in some area, at least one case classified as severe, i.e., those classified as 3 and 4, in which case the prevalence will take the value of 1 or, otherwise, the value of 0.

From the estimation of the dependent variables and given their characteristics, different econometric models were proposed. In the first case, to estimate the probability of being a victim of violence, the dichotomous variable directly consulted (VICT) was considered, which requires the estimation of a discrete choice or probabilistic model, either probit or logit, with the following equation:

$${VICT}_i={\beta }_o+{\beta }_1{x}_{1i}+{\beta }_2{x}_{2i}+\cdots +{\beta }_n{x}_{ni}+{\epsilon }_i$$

(1)

where the set of variables $x_{1i}\dots x_{ni}$ are given by the socioeconomic variables of age, marital status, educational level, occupation, income, previous violence from a family member, number of children under and over 6 years old and being the head of the household.

For the selection of the optimal model that best classifies the data, the confusion matrix was used, which indicates the safety percentage of the model; in addition, the information criteria, the maximum likelihood and the pseudo R² were analysed. Once the model was selected, the marginal effects were analysed in terms of partial derivatives, as the explanatory variables (except for age) are qualitative, the specificity and sensitivity curve and the diagnostic plot.

The second variable, the gravity of violence, was estimated using multiple linear regression models, each focusing on a different type of violence. Since gravity (GRAV) was estimated for each typology, the modelling corresponded to four models, namely the gravity of physical violence, psychological violence, sexual violence and economic violence.

$${GRAV}_i={\beta }_o+{\beta }_1{x}_{1i}+{\beta }_2{x}_{2i}+\cdots +{\beta }_n{x}_{ni}+{\epsilon }_i$$

(2)

For the validation of each of the estimates, multicollinearity was assessed using the variance inflation factor, correct specification the Ramsey test and normality of residuals by the Jarque Bera test. The absence of heteroskedasticity was confirmed by the Breusch and Pagan test and by estimating robust errors.

With respect to the last prevalence variable (PREV), given that it is also dichotomous, a similar estimation to that of violence was used, with the difference that it was analysed for each typology. Hence, three prevalence models were developed under the estimation of a probabilistic model (logit or probit), as there is no evidence of prevalence in economic or patrimonial violence; thus, three estimations were made for the rest of the typologies.

3. Results

The analysis of the information obtained from the sample shows that 58.5% of the women surveyed were single and 23% were married, with a predominance of educational levels of at least primary school, with no illiterate women. Of the sample, 31.5% are heads of households and the main occupation was a worker in the public or private sector (47%) and housewife (18%); thus, 25% indicate that they do not receive an income and 40% obtain remuneration worth less than 2 minimum salaries, the minimum being 460 USD per month at the time of applying the instrument.

In terms of family and social structure, 49.50% have children, 77.7% of whom are older than 6 years and whose mothers’ average age is 43.72 years, as opposed to the average age of women without children, which is 24.35 years, indicating a statistically significant difference in terms of age. The results also show that Mestizo women prevail (90.50%), whereas Indigenous women represent 8%.

With regard to victims of violence amongst other family members, 53.50% report that a family member has been a victim of some type of violence. This proportion is slightly reduced when asked about personal experience, with 50% of women indicating that they have experienced violence in any of its forms. Among the types of violence, physical violence prevails (24%), followed by psychological violence (15.50%), sexual violence (7%) and economic violence (only 3%). It is worth mentioning that not all women are even aware that their relationship is abusive, particularly for these types of violence; many women subconsciously believe in male superiority.

Although there are no statistically significant differences in the average age of women who have or have not been victims of violence, when analysing marital status, the proportion of women who have or have had a partner, whether co-habiting, married or divorced, rises to 86.36% among divorced women. In the case of educational level, 100% of women with low levels of education (such as primary school only) have been victims of violence, which decreases to less than 50% with higher levels of education.

With respect to the economic variables of heads of households, income level and occupation, no statistically significant differences in the proportion of female victims of violence are evident.

A woman is not more likely to experience violence if another family member has been subjected to it; 70% of the women who indicate they have family members in abusive situations have also experienced some type of gender-based violence.

As concerns the gravity variables related to the frequency of situations in different settings and the severity of the situation, the scores were between 0 and 1, where the highest value indicated a high frequency and significant severity. The meaning ranges between 0.33 and 0.42 in the different typologies, as shown in Table 2. Gravity is reported to be highest for physical violence (0.424) and psychological violence (0.411), while sexual violence is 0.36 and economic violence is the lowest at 0.33.

Table 2. Gravity and prevalence by typology of violence.

	Gravity			Prevalence
	Average	Minimum	Maximum	Affirmative (%)	Negative (%)
Psychological	0.4119	0.2485	0.8502	89	11%
Economic	0.3322	0.1875	0.7976	-	-
Physical	0.4248	0.239	0.9428	55.5	44.5
Sexual	0.3673	0.246	0.7791	76.5	23.5

Regarding prevalence, the indicator is constructed by considering whether the woman has experienced at least once any of the situations in the contexts analysed that can be classified as having a high severity (greater than 3). Although the methodology for psychological violence does not propose situations classified as having a high severity, it does for psychological, physical, and sexual typologies.

Eighty-nine percent of women indicate that they have been victims of psychological violence in an area with a high prevalence of it. This figure is 55% of women for physical violence and 76.5% of women for sexual violence.

When analysing the environments in which approaches linked to some types of violence are proposed, the gravity differs between these contexts; in psychological violence, the highest levels of gravity are reported in public institutions (0.47) and the community (0.50), followed by educational spaces. In all cases the gravity is above 0.34, as can be seen in Table 3.

Table 3. Average gravity by typology of violence and proposed scenario.

	Psychological	Economic	Physical	Sexual
Public institutions	0.4726	-	-	-
	(0.1935)	-	-	-
Community	0.5055	-	0.525	0.3658
	(0.1488)	-	(0.2019)	(0.1016)
Educational spaces	0.3928	0.3412	0.253	0.3613
	(0.1906)	(0.1707)	(0.1259)	(0.1353)
Workplaces	0.3825	0.3244	-	0.4927
	(0.1560)	(0.1033)	-	(0.2517)
Partners	0.3730	0.3420	0.3012	0.285
	(0.174)	(0.1500)	(0.0914)	(0.0935)
Family	0.3451	0.3212	0.62	0.2748
	(0.1719)	(0.1669)	(0.3218)	(0.1151)

Note: Values in brackets are the standard deviation.

With respect to economic violence, there are no situations related to public institutions and the community, but in the rest of the scenarios, an average gravity of 0.32–0.34 is recorded. In the case of physical violence, there are no situations associated with public institutions or workplaces, but the highest levels of gravity are recorded for events in the family and the community, at 0.62 and 0.52, respectively.

When analysing the typology of sexual violence, it is striking that the highest gravity of events is not registered in the couple or with the family but in the work environment, which reaches a value of 0.49, followed by the community and educational spaces (0.36 in both cases).

From these results, it is possible to identify the socioeconomic factors that determine the probability of being a victim of violence along with the gravity of violence by typology and prevalence. Table 4 shows the results of the estimates of the logit and probit models for gender-based violence (GBV) both in terms of their coefficients and marginal effects in partial derivatives.

Table 4. Coefficients and marginal effects of discrete choice (probit–logit) model estimates for the probability of being a victim of GBV.

Explanatory Variable	Probit Model		Logit Model
	Coefficient	Marginal Effect	Coefficient	Marginal Effect
Marital status	0.3311 **	0.1048 **	0.5606 **	0.1051 **
	(0.1340)	(0.0406)	(0.2297)	(0.0409)
Educational level	0.0233	0.0074	0.0353	0.0066
	(0.1780)	(0.0563)	(0.3008)	(0.0564)
Being the head of the household	−0.0642	−0.0203	−0.0629	−0.0118
	(0.2635)	(0.0834)	(0.4480)	(0.0840)
Occupation	−0.8021 **	−0.0226	−1.4242 **	−0.0228
	(0.3446)	(0.0153)	(0.6286)	(0.0151)
Income	−0.0976	−0.0309	−0.1637	−0.0307
	(0.0981)	(0.0308)	(0.1649)	(0.0307)
Children over 6 years old	−0.6823 **	−0.2161 **	−1.2211 **	−0.2290 **
	(0.3191)	(0.0983)	(0.5598)	(0.1009)
Children under 6 years	−0.1220	−0.0386	−0.2869	−0.0538
	(0.2868)	(0.0908)	(0.5112)	(0.0957)
Ethnicity	−0.7407 **	−0.0491	−1.35 **	−0.0567
	(0.3376)	(0.0435)	(0.6403)	(0.0466)
Experience of family members who are victims of violence	1.2139 ***	0.3844 ***	2.0038 ***	0.3758 ***
	(0.2036)	(0.0447)	(0.3468)	(0.0407)
Interaction between occupation and ethnicity	0.1524 **		0.2715 **
	(0.0696)		(0.1266)
Constant	3.0252 *		5.6603 *
	(1.6850)		(3.1962)
Maximum likelihood	−111.588		−111.432
Pseudo R2	0.1951		0.1962
LR test	54.08 ***		54.39 ***
Akaike information criterion	245.1772		244.8643
Bayesian information criterion	281.4587		281.1458
Correct classification (accuracy)	72.50%		73%

Note: Marginal effects are considered as partial derivatives. Values in parentheses are standard errors. The significance levels are 1% (***), 5% (**) and 10% (*).

Although the results of both models are presented, the percentage of correct classification of the data is higher in the logit model (73%). In this model, a higher value of maximum likelihood, pseudo R² and lower values of the information criteria were also obtained, which is why this estimation was selected for the analysis. The marginal effects indicate that marital status, having children over 6 years old and having family members who have experienced violence are statistically significant in explaining the probability of being a victim of GBV.

If a woman is married or divorced, the probability of being a victim rises (by 0.10), as does having family members with similar experiences (0.37), while having children over 6 years of age reduces the probability of being subjected to this type of event.

Regarding the results for gravity by typology, given that it is a continuous quantitative variable that ranges between 0 and 1, multiple linear regression models were estimated for each typology. In the presence of heteroskedasticity problems, this required robust error correction, as shown in Table 5.

Table 5. Estimates for the severity of gender-based violence typologies (robust errors).

Explanatory Variable	Gravity of Psychological Violence	Gravity of Economic Violence	Gravity of Physical Violence	Gravity of Sexual Violence
Marital status	−0.0072	0.0132	−0.0008	−0.0153 *
	(0.0120)	(0.0088)	(0.0095)	(0.0092)
Educational level	−0.0435 ***	−0.0214 *	−0.0335 **	−0.0368 ***
	(0.0164)	(0.0127)	(0.0164)	(0.0138)
Being the head of the household	0.0308	0.0084	−0.0106	0.0243
	(0.0246)	(0.0148)	(0.0210)	(0.0169)
Work	−0.0100	0.00749	−0.0076	−0.0055
	(0.0337)	(0.0269)	(0.0301)	(0.0278)
Occupation	0.0053	0.0070	0.0009	0.0036
	(0.0065)	(0.0053)	(0.0052)	(0.0050)
Income	0.0269 ***	0.0176 **	0.0259 ***	0.0179 **
	(0.0087)	(0.0078)	(0.0091)	(0.0074)
Children over 6 years old	−0.0523 *	−0.0529 ***	−0.0186	−0.0166
	(0.0297)	(0.0202)	(0.0250)	(0.0218)
Children under 6 years	−0.0516 *	−0.0723 ***	−0.0344	−0.0178
	(0.0274)	(0.0166)	(0.0221)	(0.0211)
Ethnicity	0.0218 *	0.0049	0.0170 **	0.01675 **
	(0.0187)	(0.0091)	(0.0076)	(0.0071)
Experience of family members who are victims of violence	−0.0381 **	−0.0473 ***	−0.0358 **	−0.0210
	(0.0187)	(0.0151)	(0.0173)	(0.0159)
Constant	0.4457 ***	0.3659 ***	0.4442 ***	0.4155 ***
	(0.0914)	(0.0757)	(0.0735)	(0.0643)
F-test	2.70 ***	3.71 ***	2.14 **	2.52 ***
R2	0.1252	0.1396	0.0922	0.0951
Breusch and Pagan (Homoskedasticity)	10.77 ***	28.52 ***	13.25 ***	7.89 ***
Variance Inflation Factor	2.00	2.00	2.00	2.00
Ramsey (correct specification)	0.97	1.87	0.49	0.53
Jarque Bera (normality of residuals)	0.4926	−0.4431	0.2677	−0.2579

Note: Tests for validation of assumptions are performed for ordinary least squares estimation. Values in parentheses are standard errors. The significance levels are 1% (***), 5% (**) and 10% (*).

For all types of violence, gravity depends on the level of education, in that the more educated the woman is, the less severe the violence. Gravity is also influenced by income, but in a different way; the higher the income, the more severe the violence is in the different environments.

The presence of children older or younger than 6 years of age is only significant in explaining the gravity of psychological and economic violence in that older offspring lead to a reduction in violence. Although the inverse relationship holds for physical and psychological gravity, they are not statistically significant. Ethnicity indicates that a Mestizo woman is subject to greater gravity of violence, although ethnicity was significant in all types of violence except economic violence.

In terms of having family members who are victims of violence, the gravity of any type of violence is reduced, although it is not significant for sexual violence. While occupation is not statistically significant, being a stay-at-home wife increases the gravity of all types of violence. In the case of marital status, it was only statistically significant in the gravity of sexual violence, being higher amongst single women.

Considering the same explanatory variables but for prevalence by type of violence, Table 6 presents the estimates of the marginal effects of the logit model, which, according to the selection criteria, was the one that best classified the data.

Table 6. Marginal effects of discrete choice (probit–logit) model estimates for the prevalence of gender-based violence.

Explanatory Variable	Marginal Effects for the Selected Model
	Prevalence of Psychological Violence Logit	Prevalence of Sexual Violence Logit	Prevalence of Physical Violence Logit
Marital status	−0.0550 *	−0.01311	−0.0361
	(0.0303)	(0.0386)	(0.0441)
Educational level	−0.0925 **	−0.1108 **	−0.0161
	(0.0391)	(0.5172)	(0.0614)
Being the head of the household	−0.0116	0.1343 *	0.0398
	(0.0611)	(0.0760)	(0.0913)
Occupation	−0.0208	0.0083	0.0195
	(0.0164)	(0.0214)	(0.0250)
Income	0.0647 **	0.0636 **	0.0581 *
	(0.0289)	(0.0301)	(00329)
Children over 6 years old	0.0810	−0.1632 *	−0.1343
	(0.0751)	(0.0889)	(0.1104)
Children under 6 years	0.0063	−0.0964	−0.2417 **
	(0.0652)	(0.0855)	(0.0990)
Ethnicity	0.0138	0.0090	0.0793 *
	(0.0260)	(0.0356)	(0.0427)
Experience of family members who are victims of violence	0.0329	−0.0143	−0.0624
	(0.0443)	(0.0593)	(0.0691)
Maximum likelihood Logit	−60.56 +	−101.910 +	−128.111 +
Maximum likelihood probit	−61.243	−101.978	−128.139
Pseudo R2 logit	0.1260 +	0.0655 +	0.0677 +
Pseudo R2 probit	0.1163	0.0648	0.0675
LR Logit test	17.46 *	16.28 *	18.67 **
LR probit test	16.12 *	16.14 *	18.56 **
Akaike logit information criterion	143.138 +	225.821 +	278.222 +
Akaike probit information criterion	144.486	225.957	278.278
Bayesian logit information criterion	179.419 +	262.102 +	314.503 +
Bayesian probit information criterion	180.768	262.239	314.560
Accuracy (accuracy) logit	90%	78.50% +	63% +
Accuracy (accuracy) probit	90%	78%	62%

Note: The results show the marginal effects of each selected model (logit or probit) according to the ranking matrix or additional criteria. The ‘+’ sign indicates the model with the best result on the criterion. Values in brackets are standard errors. The significance levels are 5% (**) and 10% (*).

For all cases, the level of income is statistically significant, showing that an increase in remuneration increases the probability of prevalence in all the types of GBV analysed. The rest of the significant explanatory variables only manage to influence the behaviour of certain types of violence; for example, marital status has a negative impact in all cases, yet it is only significant in psychological violence, in that single women tend to have a higher probability of this type of violence. In addition, educational level, although also showing an inverse relationship, is significant in psychological and sexual violence because women with a higher educational level have a reduced likelihood of the prevalence of these typologies.

Being the head of the household increases the probability of the prevalence of sexual and physical violence, although the latter is not significant. Having children over 6 years of age reduces the probability of sexual violence and having children under 6 years of age reduces the probability of physical violence.

With respect to ethnicity, it only significantly affects physical violence, with the probability increasing if the woman is Mestizo.

4. Discussion

The results reflect the complexity and multidimensionality of violence against women in Riobamba, Ecuador. They show that the prevalence and severity of different types of violence vary depending on socioeconomic and cultural factors, as noted in previous research (Bensley et al. 2003).

One of the most relevant findings is the high prevalence of psychological violence, experienced by 89% of the women surveyed, consistent with previous research highlighting psychological violence as one of the most frequent forms of GBV, given its less visible but equally damaging character for the mental health and wellbeing of victims (Biroli 2018).

Psychological violence is most intense in community and institutional spaces, which suggests the persistence of cultural patterns that normalise this type of aggression in public and private spheres (Lagarde 2005).

Physical violence, although less prevalent than psychological violence, shows a significant gravity, especially in the family and intimate partner environment, which is in line with the theory of the cycle of violence (Walker 1979). Additionally, the relationship between marital status and risk of violence is evident in our analysis, as divorced women are more likely to be victims. This could be explained by the socioeconomic vulnerability they face after the dissolution of the relationship or by the continuity of previous abusive dynamics (Chipokoso et al. 2024; García-Moreno et al. 2006).

In terms of sexual violence, its prevalence is lower compared to other forms of violence, but its gravity is considerable, especially in work and community spaces. This is consistent with previous studies that identify the work environment as an environment prone to GBV, particularly for women in positions of lesser power or in sectors with a strong male presence and patriarchal hierarchies (Kaphle et al. 2014; Adams-Prassl et al. 2023). The presence of sexual violence in educational settings also highlights the need to strengthen regulations and prevention mechanisms within academic institutions (Barbosa et al. 2024; De Andrade et al. 2024).

In relation to socioeconomic factors, educational level is revealed as a key determinant in the gravity and prevalence of violence. As women’s level of education increases, the probability of being a victim of violence decreases, which supports (Becker 1993; Duflo 2012) theories on the positive impact of human capital on female autonomy and empowerment. However, increased income does not always translate into a reduction in violence. In certain contexts, it can lead to power conflicts within the household, a phenomenon identified in previous studies as a backlash or violent male reaction to a loss of economic control (Vyas and Heise 2016; Bhalotra et al. 2019). The number of children also emerges as a relevant factor. Women with children over the age of six are less likely to experience violence due to a greater degree of independence and economic stability. In contrast, those with younger children face greater barriers to leaving violent situations, which is consistent with previous research linking early motherhood with economic dependence and vulnerability to violence (Hernández-Medina et al. 2023a; Awang and Hariharan 2011).

On the other hand, Mestizo women experience greater gravity in physical and psychological violence, which could be related to living in urban environments with greater socioeconomic inequalities and less community cohesion (Medina-Hernández et al. 2021; Edeby and Sebastiån 2021).

Finally, the experience of growing up in an environment of family violence correlates strongly with the likelihood of being a victim of violence in adulthood, supporting theories of the intergenerational transmission of violence (Smith-Marek et al. 2015). This finding underscores the need for early interventions to break these cycles of violence and mitigate their impact on future generations.

5. Conclusions

The results of the study confirm that gender-based violence in the canton of Riobamba is a complex phenomenon, influenced by multiple socioeconomic and cultural factors. That is why public policies should focus on strengthening access to education, economic autonomy and community awareness to prevent and reduce the prevalence of violence against women in its various forms. It is also essential to strengthen institutions and complaint mechanisms, especially in the workplace and education, where elevated levels of violence persist with serious repercussions for the victims.

The intersection between economic, cultural and social factors in the perpetuation of violence underscores the urgency of implementing intervention strategies that include gender-sensitive education, access to economic resources for women and community awareness campaigns. Furthermore, it is crucial to improve protection (including women’s refuges, as there is often literally nowhere for a woman to flee to) and reporting systems, ensuring effective and safe responses for victims through the engagement of all sectors of society, promoting structural change that favours equity and respect for women’s rights.

Based on these findings, it is possible to expand the study geographically to compare the results with women in other areas of the Ecuadorian highlands, as well as to incorporate a larger sample of Indigenous women. It is important to further consider elements of gender-related social norms, as well as levels of social capital that could explain abusive behaviour in different settings.