Prevalence of Psychological Violence in Young People in the South of Spain: Implications for Prevention

The present study shows the prevalence of psychological violence received across ten dimensions in young people aged 18 to 25 years (55.8% female) and the perception of severity, in order to design more effective dating violence prevention programs. We used a sample taken in southern Spain of 771 people. The instrument used was an adaptation of Perceived Gender Violence Scale (VGP), a perception of dating violence scales towards women through 47 behaviours. It was adapted to assess psychological violence received in men and women, maintaining psychometric properties from original (α > .94) and a factor analysis without rotation resulting in a KMO-Barlett .94. The results show that 84.4% of women and 80.2% of men have received some type of psychological violence behaviour from their last partners. The most frequent are Affective indifference (67.7%), Discrediting (51.8%) and Control (50.6%), followed by Emotional manipulation and Sexual pressure. Differences were found by age, with more frequent at the age of 20 years old. Differences were found by sex in the type of psychological violence received, but the most important predictor variables were having a current partner (β = 17.7; p < .001; R 2 = .07): where women with a partner perceive a lower level of violence received. The data suggest that it is necessary to incorporate these behaviours to be developed as contents in prevention programs; the research provides the most frequent behaviors.

Dating violence (DV) is considered a public health problem internationally both due to its prevalence and the consequences it has on the development of young people including depression, substance use, suicide ideation, and injury (McNaughton Reyes et al., 2016). DV is a concept that describes intimate partner violence (IPV) perpetrated by at least one member of an unmarried couple on the other member in the context of dating or courtship, as applied to young people. Some research claims that it tends to be symmetrical and bidirectional between young men and young women (Straus, 1979); other research claims that this does not correspond to the ratio of adult female victims of gender-based violence (GBV) who require social, health and legal intervention (Delgado, 2014). Using the concept of DV (IPV in the case of adults) implies describing that IPV is perpetrated by men and women, not focusing on the gender perspective. Talking about GBV in young couples implies that violence is perpetrated by male-boys because of the patriarchal domination system that places the masculine as a comparative value. Beyond a possible conceptual controversy, in this research we understand that any prevention program should address both boys/men and girls/women because the context of interaction includes both and what is important is the meaning given to perpetrated or received violent acts, as well as the exercise of power in the relationship and the sexist beliefs that are part of them. We believe that this idea contributed by the feminist perspective is essential to understand the development of IPV, regardless of who perpetrates the violence. We understand that many contributions in research on GBV analyze the phenomenon in depth and it is useful and necessary to consider these contributions in the prevention of this problem. However, in this research we use DV as a broader concept used internationally.
The concern to design comprehensive preventive programs in the field of DV means that different content and techniques can be found, based on different empirical evi dence. But recent research has revealed certain limitations in existing programs when evaluating real behavioural changes (Fernández-González, Calvete, & Sánchez-Álvarez, 2020) that suggest that new intervention approaches are needed. A meta-analysis indi cates than although these programs produce changes in beliefs, these do not always translate into behaviour among youth couples (De La Rue et al., 2017). Following the Ecological Model (Heise, 2011), beliefs would be factors or individual characteristics (microsystem); but the behaviour in couples would be in social network factors (mesosys tem). It appears that the preventative programs in DV affect the microsystem, but not the mesosystem. We understand that prevention programs should affect both systems.
Evidence from Spain shows that, currently, 98% of adolescents and young people re ject sexism and the justification of GBV (De Miguel-Luken, 2015). Spain ranks 25th in the Gender Development Index worldwide, in group called Very High Human Development (UN, 2020). In addition, a recent study at international level carried out in 36 countries with eighth-grade students, places Spain as the 4th country in the level of attitudes towards gender equality (Dotti Sani & Quaranta, 2017).
However, other research shows that young Spanish people show anxiety if they have to contradict sexist stereotypes, at high levels among girls; and lower levels among boys; and this violence continues to be justified as a general form of conflict resolution, a proposal that almost one in four boys seem to agree with (Díaz-Aguado et al., 2020). The problem is that sexist beliefs are a clear risk factor in the emission of IPV (Ferrer et al., 2006), and we know the prevalence of sexist beliefs, and, in countries such as Spain, they are relatively low, because this means that even in the most advanced countries in terms of equality, social pressure towards behaviour continues to be sexist, that generates certain emotions and may represent risk behaviours for violence is not overcome. Thus, complementary proposals are needed to investigate other elements that can contribute to this behavioural change finally taking place, so it is important to continue research into programs that can be truly effective.

Defining Psychological Violence
ONU defines for types of IPV: physical, psychological, sexual and economic/neglect (Krug et al., 2002, p. 6). Among Spanish young people, psychological violence (PV) is the most common type of violence (Díaz-Aguado, 2014). Straus (1979) defined PV as verbal and non-verbal acts, which symbolically hurt the other, or the use of threats to hurt the other. Others also defined PV is a set of tactics designed to regulate the behaviour of another person, restricting the personal freedom to achieve a certain outcome for the benefit of the exerciser (agent), but which does not necessarily result in conflict and is not perceived or defined as violent by the recipient; it can extend to severe forms of deprivation and exploitation (Crossman & Hardesty, 2018;Stark, 2007). In the context of a couple's relationship, these tactics are used with the aim of restricting the other person's actions based on behavioural gender stereotypes with damaging implications for the recipient. There is no strong theoretical agreement on the definition of PV. In the context of young people, we used a possible set called PV that includes control, isolation, harassment, jealousy, discrediting, affective indifference, emotional manipulation, sexual pressure and neglect, threats, and domination. This set draws from authors who make a theoretical and qualitative review of young people's relationships (Cantera et al., 2009). Some researchers show similar figures for men and women in the percentage of offences carried out on young people. These data are often obtained with instruments such as the Tactic Conflict Scale (Muñoz-Rivas et al., 2007;Sánchez Jiménez et al., 2008;Straus, 2004) which study PV together with other types of violence such as physical, sexual, economic, etc., which are also very important, but have two disadvantages in this case. In our opinion may bias the answers: firstly, they may not be representative of what happens in relationships between 18 and 25 years old but specific to older people (economic, use of children, ...) or to the group of people classified as victims from IPV. Secondly, if physical and sexual violence are perceived as more serious than PV, the relative comparison can lead to minimizing or not giving importance to behaviours that cause harm, but do not produce a physical outcome, being more difficult to perceive the long-term consequences. For this reason, we believe that analyses using specific scales of PV are required. Moreover, in the opinion of some researchers, the use and interpretation of these tools can lead to the denial or minimization of GBV (Ferrer et al., 2006;Ferrer et al., 2011).
Another problem we point out is since violence in relationships is context-dependent (Delgado et al., 2015) and, as we said before, currently, in our context relationships between young men and young women are subject to change, it is necessary to delimit the type of behaviours that are happening in order to represent what is happening in that context. Therefore, knowing the current prevalence of violent behaviour among young people, will be very useful in order to prioritize and provide specific content to the programs designed, under the hypothesis that adapting the content will increase ad herence to the messages of the preventive programs and will improve the effectiveness of preventive actions. Thus, the main contributions of this work is to identify the contents of PV that are occurring in young people in order to include them specifically in the prevention program through a specific assessment tool (Delgado & Mergenthaler, 2011), which allows us to identify typical behaviours according to the prevalence in our sample. On the other hand, it also provides specific rates in these age ranges in both men and women, which we can used to compare similar samples in similar regions.
However, this type of violence is difficult to detect because it is not seen and does not have a short-term impact. As Delgado explains, GBV initially takes on psychological manifestations (Delgado & Mergenthaler, 2011), and moves up the continuum to include increasingly intense physical aggression (Crossman & Hardesty, 2018) because it is a process, not a one-time event. It is therefore a continuum, and it is difficult to detect the transition from one type to another. We consider this process to be the same for girls/women and for boys/men who receive violence and believe it may be critical to highlight behaviors involving PV through prevalence to help make preventive programs more effective.
The main objective of this work is, therefore, to explore VP behaviors that will contribute to the selection of content and make prevention programs more effective because they will be better targeted to what really affects young people. To this end, we have collected the current prevalence of IPV behaviors received in young people aged 18 to 25 years, and we have analyzed them by comparing 10 dimensions, considering sex, age, type of studies, sexual orientation, current partner and number of last partners.
More specifically, we set out to: 1. Determine the percentage frequency of VP re ceived in total, using the dimensions that are part of VP and disaggregating by sex; 2. Obtain the overall average rate of VP received and disaggregate it by sex, giving a measure of violence in women and men, calculating possible significant differences; 3. Disaggregate the overall rate according to the different dimensions of PV relevant to the development of prevention programs and establish whether there are significant dif ferences, considering age; 4. Explore differences in the scores of the different dimensions taking into account independent samples at origin. 5. To specify the most important items that could constitute the bulk of the activities of a specific program in case of significant dimensions.

Measures: Instrument and Variables
We applied the Perceived Gender Violence Scale (VGP) (Delgado, 2014;Delgado & Mergenthaler, 2011;Delgado et al., 2015) which is a specific scale of PV with previous qualitative study with young people undertaken by Cantera et. al. (2009), conducted a comprehensive review on different classifications of psychological abuse behaviours and completed it with a qualitative analysis in adolescents and young adults aged 15-25 years. In IPV research, qualitative research has been identified as fundamental because it shows elements that may go overlooked in quantitative research (Crossman & Hardesty, 2018). This scale used measures ten dimensions identified in GBV, when it takes the form of PV, excluding physical aggression. It uses a battery of 47 items in a Likert format, with responses ranging from 0 (no perceived violence) to 6 (clearly perceived violence). The original VGP Scale reformulated the language of expression in order to be able to compare data between groups of men and women. Dimensions: 1. Control: knowing everything they do, with whom, controlling the way they dress and groom themselves; 2. Harassment: watching them, following them, waiting for them at the exit of the places they go to without them knowing it, harassing them on the phone...; 3. Isolation: preventing them from having a social life, forcing them to relate only to them, making relationships with his/her network of friends and/or family difficult; 4. Jealousy: accusing them of flirting with others, of being interested in others, of being provocative; mistrusting their fidelity and displays of affection...; 5. Discrediting/disqual ification: objecting what they do or say, making derogatory comments about her/him and about women/men in general, ridiculing her/him, undermining her/his self-esteem...; 6. Affective indifference: being insensitive, not talking to you, inattentive, ignoring you; 7. Sexual pressure and neglect: imposing unwanted sexual relations through anger or accusations, ignoring what he/she wants, not taking responsibility for the risks, consid ering that protective measures are only the other person's responsibility; 8. Emotional manipulation: emotional blackmail to get what he/she wants, making him/her feel bad if he/she does not do what he/she wants...; 9. Threats: threat of physical violence, of abandonment, of harming him/her through objects or loved ones...; 10. Domination: imposing the rules of the relationship, considering him/her as your possession, believing that he/she belongs to you. The Cronbach's α in original scale was .96, which is very reliable.
In this work, Delgado's VGP Scale (Delgado, 2014;Delgado & Mergenthaler, 2011) was adapted to apply it to DV received in young men and young women, to measure the occurrence and frequency of each of the 47 items that make up the survey. In this adaptation, the questionnaire contained the following introduction: "Below is a list of situations or behaviours that can occur in a couple's relationship. For each one, mark the degree to which you have experienced this behaviour in your current relationship or in your last relationship (if you are not currently in a relationship) and how often. Put an X in the appropriate box and indicate how severity this is for you from 0 to 6". The criteria to guide the answers were: "0 -If you think that this behaviour has not occurred; 1 -If you think that this behaviour has occurred very little (it happened once); 2 -If you think that this behaviour has occurred little (it has happened approximately between 3 and 5 times in several months); 3 -If you think that this behaviour has occurred sometimes (it happens at least once a month); 4 -If you think that the behaviour has happened frequently (at least once a week); 5 -If you think that the behaviour has happened a lot (several times a week); 6 -If you think that the behaviour has happened always (every day it happens at least once)". The 47 items of the original scale were included, with a Cronbach's α of .94, for the total scale. The properties in the ten dimensions were maintained, as can be seen in Table 3.
In addition, we added after each item a severity rating on the evaluated behavior: "How serious do you consider this behaviour to be?", following the same scale from 0 to 6. This measure had a Cronbach's α = .99.

Procedure
It was applied in online format by five final year students of the Psychology Degree dur ing the 2018-2019. It was disseminated through social network contacts used by young people of this age group, with the only requirement being that they had to be between 18 and 25 years old, ensuring the randomness of the sample. All questionnaires were administered under the same instructions. At start, questionnaire voluntary participation was required. No personal data were collected, only social variables.

Data Analysis
Data were analyzed using IBM SPSS.25 with a frequency analysis and inferential analysis which included the calculation of mean difference with Student's t and ANOVA because the sample is high and Levene's test indicated homoscedasticity. The Kolgomorov-Smir nov tests showed that there is no normality in the samples, so we performed Kruskal-Wallis analysis of variances for non-parametric samples, showing the same distribution with similar medians for all dimensions. Since non-normality is very common in the Social Sciences and since the sample has statistical power, due to the large sample size, we used parametric statistics that provide more information. On the other hand, a factor analysis of the total scale was applied, to check that the adapted scale had not changed psychometric properties with respect to the original. We first performed a factor analysis without rotation resulting in a KMO-Barlett index of .94, which maintained the unifactorial structure with communalities between .52 and .78 for the 47 items, and then an oblimin rotation resulting in eight of the components explaining more than 66.7% of the variance. Similar data to the original scale, which showed 66.6% (Delgado, 2014). We also used regression analysis to test the effect of the independent variables age, sex, type of studies, sexual orientation, current partner and number of past partners. G*Power 3.1.9.7 was applied to calculate the effect size.

Results
1. We found that 82.6% (Table 1) of young people have received at least one type of PV behaviour from their last partner; 17.4% of young people aged 18-25 have not received any PV behaviour from a last partner. The data column representing "Absence of PV" are participants who have evaluated the behaviours as "None" (absolute zero) present in their last intimate partner relationship. Considering the rest of the frequencies, the most frequent dimensions of PV (Table  2) are affective indifference (65.5%), discrediting (51.8%), control (50.6%) and emotional manipulation (44.5%). The least received behaviours are isolation (86.5%), harassment (78.1%) and threats (71.7%). If we look at the frequencies of people who have received PV according to sex, and considering the three most received dimensions: women receive PV in the form of affective indifference (67.7%), discrediting (55.3%) and control (45.6%); men receive PV in the form of affective indifference (62.5%), control (57.3%) and emotional manipulation (48.8%). Jealousy, domination, threats, harassment and isolation remain in the same order between men and women, and are the least received. 2. Regarding a general mean index of PV received, data shows a mean (SD) of aggressions received of 21.3 (33.7). Taking into account that the minimum and maximum range would be [0-282], the mean in our study may be small, although the Standard Deviation offers a high variability. According to the results shown in Table 3, although women receive more PV on average [21.8 (35.2) > 20.8 (31.7)], ANOVA F(1,769) = .18; p = .66, expresses that the differences found are not statistically significant. However, this does not indicate anything about the consequences being similar.
3. Respect to the mean scores assigned to the dimensions, Table 3 shows that the mean of the dimensions are 2.14 (M = 21.4), there are significant differences between the dimensions, as reflected in ANOVA F(9, 770) = 99.94; p < .001. Although again, if the dimensions are considered according to the sex of the participant, the differences are not significant in any of the dimensions (p > .05): i.e. both men and women tend to value the situations of PV received in the same way. No significant differences were found for any category of sexual orientation. Significate differences were found on some dimensions for age in Discrediting F(7, 689) = 3.36; p = .002; Affective Indifference F(7, 689) = 2.87; p = .006; Emotional Manipulation F(7, 689) = 2.26; p = .02; and Jealousy F(7, 689) = 2.80; p = .007.
Differences were analyzed with post hoc test of multiple comparisons with HSD Tu key and Bonferroni (samples of 24 and 25 years-old) and we found: In Jealousy there are differences between 20 and 21-22 years old groups. In Discrediting, there are differences between 20 years old and 18 and 22 years old groups; In Affective Indifference 20 to 22 years old. In Emotional Manipulation there are differences between 18 and 20. Figure 1 shows dimensions for which significant differences were found by age.

Figure 1
Significant Differences of Means in Some Dimensions by Age 4. In order to observe whether there may be differences between the total data analyzed (N = 711) and what happens in some of the samples taken at the origin independently, but which are related samples when compared to the total sample. One of them, called Subsample 3, with an n = 194 (100 women; 94 men) was targeted. Table 4, above, indicates different mean scores, even though the same sampling type: subsample 3 (M 3 ) shows differences in scores between women and the total sample (M 1 ) in all dimensions of PV: t(193) = 3.66; p < .05. We calculated the effect size d, using Wilcoxon's W, in the women's group for PV Overall, and found mean differences comparing mean scores for women with an effect size d = .6, which is interpreted as a too large average (Cárdenas Castro & Arancibia, 2014). Scores between men and the total sample, no significances t(193) = 0.8; p > .05.  We used linear regression analysis to explore the origin of these differences in Subsample 3 by analyzing the predictive ability of the independent variables. Participant's sex was used to predict PV Overall score, but it was not significance F(1, 192) = 3.57; p = .6; Age was added, and it was not significance: F(2, 191) = 2.82; p = .13. Then, the type of studies was also added but it was also not significant, either: F(3, 190) = 1.44; p = .23. We then analyzed the predictive ability of participant's sex on the dimensions sepa rately and found that it is predictive of Disqualification (R 2 = .02), Affective indifference (R 2 = .02) and Emotional manipulation (R 2 = .02). Although the predictive capacity of the participant's sex was small with 2%, we decided to add age and type of studies, but they were not significant. We analyzed the variable current partner and then it was found that it is a variable that contributes to the different perception of IPV for women in all dimensions, as reflected in Table 5. The mean score for the perception of PV in women who do not have a partner increases. However, in all scores (Table 6) there is a large variability, which encourages us to incorporate other variables related to personal interaction (mesosystem) as predictors of the total score: sexual orientation and number of partners. Sexual orientation does not appear as significant for either men or women. Table 6 Means ( However, the number of partners for men results as a significant predictor in the percep tion of the total score β = 11.7; eS: 2.0; F Men (1, 87) = 33.9; p < .001; R 2 = .3; For women the number of partners is not significant F Women (1, 90) = 1.0; p = .3. All these results are for Subsample 3. However, we have found that all these results are reproduced in the total sample. Then, one difference we find is that as the variable having a partner influences the response of women and the ratio of women who do not have a partner is higher than in the total sample (Table 6). However, this is an issue to be considered in the following investigations since the sample is not normal and the sample size decreases as new variables are incorporated.
5. Table 7 shows items of PV most frequently received, indicating the severity attrib uted to each item, complete sample and disaggregated by sex, in order to determine the items that should be worked on as content in Dating violence (DV) preventive programs. The format expresses the mean of the PV behaviours suffered by the last partner (M) and the standard deviation (SD) of the group. It should be noted that the severity attributed to the events is quite similar for all items.

Discussion
This study reports the prevalence of psychological violence (PV) in young people aged 18 to 25 years old through ten dimensions: control, threats, isolation, affective indifference, emotional manipulation, domination, jealousy, sexual pressure and neglect, discrediting and harassment; taking into account the frequencies of the behaviours in a differential way between men and women. The main contribution of this study is to specify the contents of VP to be worked in prevention programs with young people and to offer rates that can be established in a comparative manner in different regions.
Regarding the frequency of PV received, we highlight that 84.4% of women have received some kind of PV from their last current partner, and 80.2% of men have received some kind of PV. The age of 20 years is the most VP behaviours are received. These data suggest the need for continued intervention with effective programs in the field of PV after secondary or obligatory education, in regions where the level of sexism has decreased in recent years.
Taking into account the overall mean rate of PV received, and given the minimum and maximum rate in the responses, we can interpret the mean of PV received as small. However, there are several complementary elements to be considered in the interpretation: 1. Standard deviation offered a high variability, which is explained by the differences that the data reflect if they are analyzed as independent samples at the origin of the data collection. 2. The effect of social desirability in the case of the study of intimate partner violence (IPV) tends to minimize the phenomenon in the case of wom en. Feminist research has shown that when we talk about gender violence, women do not like to identify themselves as victims because it is considered a socially stigmatized group and they do not want to be identified with it because it is a socially undesirable concept (Ferrer et al., 2006). Women also minimize because they don't want to upset or devalue the image of men. This could be the explanation why the Subsample 3 presents a higher score in young women, since there is a higher proportion of women without a partner than in the total sample; by contrast, men minimize their own aggressions and tend to maximize the aggressions they receive (Rubio, 2009). 3. It should also be taken into account that we have asked about what happens in a current or previous partner, leaving previous histories of violence hidden. For these reasons, the data obtained should be interpreted as a trend rather than as a real incidence. Furthermore, the samples do not present normality and although taken as a whole they can be evaluated, when analyzed with several variables the statistical power does not allow conclusions to be generalized. To improve this aspect, in the future we should control the samples according to the number of previous partners and current partners according to sex.
We can argue that men and women exercise behaviours of PV in a different way: women exercise more controlling and jealous behaviours; and men use more emotional indifference and disqualification. In fact, traditional sexist stereotypes already proposed women as controlling and men as emotionally indifferent in their relationships, so our results are showing that the sexes behave according to what is expected from them, and this idea of abandoning sexist beliefs and stereotypes does not seem to be fulfilled in practice. This supports the results obtained in the research by Díaz-Aguado et al. (2020) which expresses the effect of sexist role pressure, much greater in young women, who appear with high anxiety to comply with sexual stereotypes, which continue to objectify them and pressure them to comply with a feminine standard. And young men, who also feel pressure to conform to their male stereotype, but to a much lesser degree. As couple relationships and the way in which violence is expressed changes as new commitments and new stages are acquired (Crossman & Hardesty, 2018), it is essential to take as a starting point these types of aggressions, which may seem subtle due to their average incidence, but which express that the problem has not disappeared and that they are a breeding ground for more evident violence, given that young men continue to justify the use of violence in the resolution of conflicts (Díaz-Aguado et al., 2020).
All in all, taking the dataset, we can conclude that the PV dimensions most received by young people are Affective Indifference, Discrediting, Control and Emotional Ma nipulation, followed closely by Sexual Pressure. Indicating that these should be the priority aspects to work on with young people. Working with these contents is very important because it is the way in which gender power imbalances and attitudes that sustain violence are being manifested. Previous studies had revealed that control would be a starting step in the escalation of GBV and DV (Cevallos Neira & Jerves, 2018;Ruiz-Repullo, 2016), but our data suggest that even before control, some behaviours such as affective inference, discrediting and emotional manipulation could appear that we had not considered sufficiently important and on which we must intervene from an individual and interpersonal perspective (Heise, 2011). On the other hand, the dimen sion of affective indifference has not been developed as an essential component in IPV prevention programs, and this study highlights its importance as an initiating factor, which, combined with control, can increase the negative consequences for the victim. The consequences for men and women of receiving this type of PV have not been taken into account in this study and should be incorporated in future research; although it is understood that prevention programs should begin working on content related to affective indifference, considered as a problem of social skills (individual or microsystem) or essentially as the behavioural exponent of sexist roles that continue to be shown as part of the relationship between the members of young couples (interpersonal or mesosystem).

Funding:
The authors have no funding to report.