Introduction
One of the typical events in adolescence and young adulthood is establishing and maintaining romantic relationships (Kansky et al., Reference Kansky, Allen and Diener2019; Lantagne & Furman, Reference Lantagne and Furman2020). Although the literature suggests that adolescents are relatively inexperienced with romantic relationship dynamics, which can be highly variable and emotionally intense, romantic involvement can produce significant coping challenges (Rogers et al., Reference Rogers, Ha, Updegraff and Iida2018). In some cases, an overly romantic view of love triggers violent dating relationships in which harmful behaviors are tolerated (Alcalá et al., Reference Alcalá, Cortes-Ayala and Vega-Cauich2021; García-Díaz et al., Reference García-Díaz, Lana-Pérez, Fernández-Fieto, Bringas-Molleda, Rodríguez-Franco and Rodríguez-Díaz2018). Violence in adolescent and young couples can continue into adulthood and is a predictor of marital violence (Vicario-Molina et al., Reference Vicario-Molina, Orgaz, Fuertes, González and Martínez-Álvarez2015). Furthermore, studies on violence in adolescent and young adult couples reveal some specific characteristics that do not apply to married couples (Bringas-Molleda et al., Reference Bringas-Molleda, Cortés-Ayala, Antuña-Bellerín, Flores-Galaz, López-Cepero and Rodríguez-Díaz2015; De la Villa et al., Reference De la Villa, García, Cuetos and Sirvent2017), highlighting the need to focus on this phenomenon to better understand it.
Psychological abuse is the most common form of violence in adolescents and young couples (Pazos et al., Reference Pazos, Oliva and Hernando2014; Sánchez-Hernández et al., Reference Sánchez-Hernández, Herrera-Enríquez and Expósito2020). According to O’Leary and Smith Slep (Reference O’Leary and Smith Slep2003), the indicators are verbal aggression, control and dominance behaviors, and jealousy behaviors. Furthermore, there is a new form of psychological violence involving the use of electronic media. This type of violence is defined as any act involving slandering, insulting, intimidating, pressuring, or controlling one’s romantic partner via electronic media (De los Reyes et al., Reference De Los Reyes, Jaureguizar and Redondo2022). Different studies have shown that offline and online psychological violence is bidirectional in adolescent and young couples, suggesting that men and/or perpetrators can assume the roles of victims and/or perpetrators (Cava et al., Reference Cava, Buelga and Carrascosa2015, Reference Cava, Martínez-Ferrer, Buelga and Carrascosa2020; Laforte et al., Reference Laforte, Paradis, Todorov and Cyr2023; Lorente-Anguís et al., Reference Lorente-Anguís, Vilariño and Lopez-Zafra2024; Rojas-Solís & Romero-Méndez, Reference Rojas-Solís and Romero-Méndez2023). The systematic review by López-Barranco et al. (Reference López-Barranco, Jiménez-Ruiz, Pérez-Martínez, Ruiz-Penin and Jiménez-Barbero2022) concluded that violence in couples is bidirectional, making it necessary to delve deeper into the factors that influence it.
To understand this phenomenon of bidirectionality, in this study, we further analyze the reasons expressed by adolescents and young people for perpetrating or receiving violence in their relationships. Previous studies indicate that the motives given are that they are concerned about their couple, masked by jealousy and mistrust (Borrajo et al., Reference Borrajo, Gámez-Guadix and Calvete2015a; Rodríguez-Castro et al., Reference Rodríguez-Castro, Alonso-Ruido, Lameiras-Fernández and Faílde-Garrido2018), feelings of frustration, games, and/or self-defense (Hettrich & O’Leary, Reference Hettrich and O’Leary2005; Rodríguez-Pérez, Reference Rodríguez-Pérez2015). However, despite the high prevalence of psychological violence online and offline, reaching rates as high as 97% according to previous studies (Rubio-Garay et al., Reference Rubio-Garay, López-González and Amor2017), there is limited knowledge of the risk factors that distinguish profiles of violence in adolescent and young couples. Moreover, previous studies focused only on the victim, perpetrator, or mutually violent profile (Alegría & Rodríguez, Reference Alegría and Rodríguez2017).
Accordingly, several psychosocial factors have been studied to explain the characteristics of the people involved in violent relationships. Relational or partner dependency is one of the variables related to online and offline psychological violence, and it was recently considered a predictor. Dependency is characterized by an excessive need for attention and protection from a partner and strong trust in the relationship to define one’s view and functioning of the self (Bell et al., Reference Bell, Howard and Cornelius2022; Beltrán-Morillas et al., Reference Beltrán-Morillas, Valor-Segura and Expósito2019). Some of the most relevant characteristics of partner dependency are (a) possession, (b) inability to end the relationship, (c) conditional love, (d) feeling of not being complete without the other person, (e) external locus of control, (f) a partial elaboration of the problem, (g) excessive need for love, (h) hyperdependency to compromise, and (i) negative feelings (guilt and fear of abandonment) (Moral & Sirvent, Reference Moral and Sirvent2009). Within this concept, we can distinguish different dimensions. Following the classification of Rathus and O’Leary (Reference Rathus and O’Leary1997), three components are considered: Anxious attachment, which is defined as separation anxiety, feelings of abandonment by the partner, concern about the other person’s feelings, and whereabouts (where is he/she and what is he/she doing). There is an over-activation of the attachment system, showing sensitivity to signs of loss of affection and fear of abandonment by the partner (Bowlby, Reference Bowlby1973; Gómez-Zapiain et al., Reference Gómez-Zapiain, Ortiz and Gómez-Lope2011; Goncy & Van Dulmen, Reference Goncy and Van Dulmen2016). Exclusive dependency refers to the person that only trust in her/his partner like a confident, and excludes other important relationships, social supports, activities, or interests (Bell et al., Reference Bell, Howard and Cornelius2022; Linville, Reference Linville1985; Rathus & O’Leary, Reference Rathus and O’Leary1997). Finally, emotional dependency describes a need for protection and support, as well as a strong reliance on the partner as an essential element of self-esteem, identity, and overall functioning (Bornstein, Reference Bornstein1993, Reference Bornstein2006). The results show that individuals involved in abusive relationships are more dependent than those involved in non-abusive relationships (Tan et al., Reference Tan, Arriaga and Agnew2018). The perception or idealization of romantic love in couple relationships is associated with dependency on one’s partner and staying in violent relationships (Aiquipa, Reference Aiquipa2015). Dependent individuals display manipulative behaviors such as self-harm or endangering their lives to prevent their partner from leaving them (Villegas & Sánchez, Reference Villegas and Sánchez2013). Furthermore, research has shown a positive relationship between partner dependency and couple violence (Perles et al., Reference Perles, San Martín and Canto2019; Pradas & Perles, Reference Pradas and Perles2012; Urbiola et al., Reference Urbiola, Estévez and Iruarrizaga2014). This association is also present in the case of online violence (De Los Reyes et al., Reference De Los Reyes, Jaureguizar and Redondo2022; Lorente-Anguís et al., Reference Lorente-Anguís, Vilariño and Lopez-Zafra2024). Moreover, this relationship has been corroborated by other studies, where people with emotional dependency have been found to use social networks and cell phones to have greater proximity and control of their partners (Morey et al., Reference Morey, Gentzler, Creasy, Oberhauser and Westerman2013).
As indicated by Pazos et al. (Reference Pazos, Oliva and Hernando2014), another variable that seems to be a determining factor in the use of psychological violence in adolescent and young couples is frustration tolerance. Frustration tolerance is the ability to confront and manage adverse situations, and a person who suffers from this tolerance does not feel completely overwhelmed or blocked from the situation despite difficulties (Rojas et al., Reference Rojas, Morales and Sandoval2022; Ventura-León et al., Reference Ventura-León, Caycho-Rodríguez, Vargas-Tenazoa and Flores-Pino2018). Frustration is one of the reasons why people tend to attack others when they are prevented from achieving their desires (Bar-on, Reference Bar-On1996). Thus, the lower the degree of frustration tolerance, the greater the risk of using violence among both boys and girls (Pazos et al., Reference Pazos, Oliva and Hernando2014). It has also been reported that dependency on a partner is characterized by a low frustration tolerance (De la Villa-Moral et al., Reference de la Villa-Moral, Sirvent, Ovejero and Cuetos2018); however, no previous studies have analyzed both factors in relation to violence in young couples.
The present study further contributes to knowledge by considering a person-centered approach, such as latent profile analysis (LPA), which is based on intraindividual analyses. Unlike traditional, non-latent clustering methods, LPA classifies individuals into clusters on the basis of membership probabilities estimated directly from the model. Moreover, all types of variables (continuous, categorical, etc.) and their combinations can be considered, and covariates can be used as antecedents and outcomes of the profiles obtained (Vermunt & Magidson, Reference Vermunt, Magidson, Hagenaars and McCutcheon2002). Thus, profiles of individuals sharing similar patterns of variables are identified and compared with other profiles. Like other statistical techniques used for segmentation, such as factor analysis or cluster analysis, latent class analysis is an exploratory method for populations or samples; however, over and above its exploratory nature, latent class analysis allows for all kinds of confirmatory research on the nature of the latent concept (Castro & Tenorio, Reference Castro and Tenorio2010).
On the basis of the above, the current study aims, first, to identify different psychosocial profiles according to dependency and frustration tolerance and, second, to describe the differences in violence according to the emerged profiles and the motives young people give for justifying violence. Specifically, the following hypotheses were tested:
H1: Profiles with greater dependency and low frustration tolerance will associate with higher levels of violence within the relationship.
H2: The identified profiles will show a relationship with bidirectional violence. This implies the perpetration and/or reception of violence in their relationships.
Moreover, in this hypothesis, we consider that H2a. Profiles will not be specifically associated with victims or perpetrators.
H3: The reasons expressed for the presence of violence in their relationship will differ between the profiles.
Method
Participants and Procedure
One thousand fifty-eight adolescents and young adults completed this study. Fourteen participants were eliminated due to incomplete questionnaires. The researchers asked for permission from the education delegation, and once authorized, the first author visited all the centers to explain and deliver the questionnaires. The study involved 49 high schools and 13 universities in different areas of Spain. The inclusion criteria were that all participants were students, aged 15–22 years, and had a partner of at least 1 month in the previous year. The final 1044 participants included 62.1% women and 37.9% men. The average age was 18.04 years for women (SD = 2.014) and 18.20 years for men (SD = 2.207). A total of 69.9% were involved in a relationship at the time of the evaluation. A total of 92.9% of the participants were heterosexual, and 7.1% were bisexual but were currently in a relationship with a person of the opposite sex. Among the participants who indicated their current level of education, 41.8% were studying at university, 23% at the baccalaureate level, 20.2% in compulsory secondary education, 5.7% in higher education, 3.9% in intermediate education, and 0.9% in basic vocational training. They agreed to participate voluntarily in the study. Once they volunteered, students under 18 years of age were given a consent form to be signed by their legal guardian. For students over 18 years of age, informed consent was included in the questionnaire. The protocol was approved by the Ethics Committee of the first author’s university (Ref. NOV.17/6.TES).
Variables and Instruments
Sociodemographic Variables
Information about sex and age, as well as information concerning the couple’s relationship status (in a present relationship or not), was included.
Jealousy and Control Offline
The Dominating and Jealous Tactics Scale (DJTS; Kasian & Painter, Reference Kasian and Painter1992), adapted to young Spanish individuals by Muñoz-Rivas et al. (Reference Muñoz-Rivas, Redondo, Zamarrón and González2019), was used to evaluate jealousy and control behaviors offline. The scale comprises two dimensions of 11 items each for perpetration and victimization. Each dimension is divided into two subscales: dominant tactics, which includes seven items (e.g., “I have threatened my partner that I will go with someone else”), and jealous tactics, made up of four items (e.g., “I have been jealous of other boys/girls”). The DJTS is rated on a 5-point Likert-type scale ranging from 1 (never) to 5 (very frequently). The reliability for the total scale was (α = .78) for perpetration and (α = .80) for victimization.
Verbal Aggression Offline
The Conflict in Adolescent Dating Relationships Inventory (Fernández-Fuertes et al., Reference Fernández-Fuertes, Fuertes and Pulido2006) was used to measure verbal aggression offline; thus, only the subscale of verbal aggression was used. It consists of 10 items each for perpetration and victimization, with responses ranging from 1 (never) to 5 (frequently) (e.g., “I threatened to leave the relationship”). The Cronbach’s alpha was .70 for perpetration and .79 for victimization.
Online Violence
The Cyber Dating Abuse Questionnaire (Borrajo et al., Reference Borrajo, Gámez-Guadiz, Pereda and Calvete2015b) was used to evaluate direct aggression and control behaviors online. The instrument is composed of 20 items each for perpetration and victimization (a total of 40 items). Additionally, each component is divided into two factors: direct aggression (11 items each, e.g., “I threatened my partner using new technologies to hurt him or her physically”) and control (9 items each, e.g., Using mobile applications, I controlled the hour of the last connection with my partner or my former partner”). The reliability for this sample was (α = .70) for perpetration, (α = .75) for the victimization of direct aggression, (α = .79) for perpetration, and (α = .86) for the victimization of control. The response options used a Likert scale ranging from 1 point (never) to 6 points (more than 20 times).
Partner Dependency
The Couple-Specific Dependency Scale (CSDS; Lorente-Anguís & Lopez-Zafra, Reference Lorente-Anguís and Lopez-Zafra2022) was used to measure partner dependency. This instrument is derived from the Spouse-Specific Dependency Scale (Valor-Segura et al., Reference Valor-Segura, Expósito and Moya2009) but allows its use in non-married young people. It is composed of 16 items for adolescents, which are divided into three subscales: anxious attachment (five items: e.g., “I look for signs that my partner still loves me”), exclusive dependency (five items: e.g., “If I lost my partner, I would not have anyone to turn to”), and emotional dependency (six items: e.g., “I find it difficult to separate from my partner”). The items are presented with six answer options ranging from 1 point (strongly disagree) to 6 points (strongly agree). The reliability for this sample was .76 for anxious attachment, .73 for exclusive dependency, and .80 for exclusive dependency. The overall reliability of the scale was .84.
Frustration
The Frustration Tolerance Scale (ETF; Oliva et al., Reference Oliva, Antolín, Pertegal, Ríos, Parra, Hernando and Reina2011) was used to evaluate this psychosocial factor. This scale is derived from the subscale of the Bar-On Emotional Quotient Inventory (Bar-On & Parker, Reference Bar-On and Parker2000). The instrument comprises eight items (e.g., “Some things make me very angry”). The items are presented in a 5-point format ranging from 1 (never) to 5 (always). The Cronbach’s alpha was .81.
Motives for Aggressive Behavior
To have a comprehensive measure of the reasons that young people give to justify responding aggressively to or receiving an aggressive response from their partners, we included the following two questions: “Why do you perform these behaviors?” “Why do you think your partner performs these behaviors?” They were given eight options derived from the literature: (1) jealousy, (2) control, (3) concern, (4) bidirectionality, (5) response, (6) play/joke, (7) personality, and (8) absence (Borrajo & Gámez-Guadix, Reference Borrajo, Gámez-Guadix and Calvete2015a; Muñoz-Rivas et al., Reference Muñoz-Rivas, Graña, O’Leary and González2007; Rojas-Solís & Romero-Méndez, Reference Rojas-Solís and Romero-Méndez2023). Additionally, they could respond to an open-ended space to express other possible motives or to explain their choice further. They had the option of scoring more than one reason.
Data Analysis
IBM SPSS v. 25 was used to compute all the statistics (means, standard deviations, reliability, asymmetry, and kurtosis), Spearman correlations, and Cronbach’s alpha values for the reliability of the scales. AMOS 23 was used to perform validation through confirmatory factor analysis via maximum likelihood estimation. Owing to the limitations of chi-square (2), we used the Norm Fit Index and Comparative Fit Index, in which values equal to or greater than .90 are considered acceptable (García-Ael et al., Reference García-Ael, Recio and Silván-Ferrero2017; Hu & Bentler, Reference Hu and Bentler1999; Muñoz-Rivas et al., Reference Muñoz-Rivas, Redondo, Zamarrón and González2019). Additionally, we computed the root-mean-square error of approximation with the 90% CI as a parsimony adjustment index. In this case, the model is considered to fit properly when values between .05 and .08 are obtained, with values below .05 considered indicative of an excellent fit (García-Ael et al., Reference García-Ael, Recio and Silván-Ferrero2017; López-Ramos et al., Reference López-Ramos, Navarro-Pardo, Fernández-Muñoz and Da Silva2017).
Using the LPA methodology, Jamovi v.2.4.2 software was used to identify the psychosocial profiles. The most commonly used measures to select the model with the best fit are the Bayesian information criterion (BIC) and the Akaike information criterion (AIC). Both are measures of the relative quality of a model, and the best solution is chosen by taking the smallest index (Fonseca-Pedrero et al., Reference Fonseca-Pedrero, Ortuño-Sierra and Pérez-Albéniz2020; Spurk et al., Reference Spurk, Hirschi, Wang, Valero and Kauffeld2020); the bootstrap likelihood ratio test (BLRT) index with p-values less than .05 is also considered. An additional entropy index was also calculated for accuracy, with values close to 1 signifying better homogeneity of the classes. The influences of sex, relationship status, and age on psychosocial profiles were also analyzed. To this end, a multinomial logistic regression was estimated. This command makes it possible to analyze whether a one-unit increase in the variable taken as antecedent makes it probable to belong to a given profile with respect to another. This probability is compared in groups of profiles two by two. In addition, odds ratios (ORs) are calculated to facilitate the interpretation of the coefficients obtained (Morin et al., Reference Morin, Meyer, Creusier and Biétry2016). ORs reflect the change in the probability of belonging to a target profile versus a comparison profile with each one-unit increase in the predictor. To facilitate the interpretation of the results obtained, the profiles being compared are expressed in the form “Profile X versus Profile Y,” with Profile Y being the reference or comparison profile, and Profile X the target profile.
Finally, to identify the motives, a content analysis (López, Reference López2002) was carried out for the reasons given in the open question. First, a categorical classification was performed, for which all the answers provided by the subjects were read, and different categories were created to distribute the different contributions of the subjects. Two coding judges (psychologists with research experience in content analyses and in the field of intimate partner violence, who were unaware of the objectives of the study) were subsequently selected to establish a correspondence between the initial responses and the categories elaborated independently of each other. The same process was then carried out by another two expert judges, who were involved in the study and knew its objectives. Once agreement between the responses and the categories was obtained, the analysis of agreement between judges and discrepancies was carried out, with agreement existing when it was reached by at least three of the judges, two of whom had to be the experts.
Results
Descriptive statistics and bivariate correlations by gender are shown in Table 1. All dimensions of violence have a positive, significant relationship. Despite the high magnitude of the value of the correlations between factors, especially between perpetration and victimization on the same scale, they are below the value of .85 (Caridade & Braga, Reference Caridade and Braga2019; Netemeyer et al., Reference Netemeyer, Bearden and Sharma2003), thus demonstrating discriminant validity (Marôco, Reference Marôco2014). There is no significant relationship between exclusive and emotional dependency with the violence scales, with the exception of online control perpetration and jealousy and control offline perpetration. However, the significant relationship between online control perpetration and jealousy and control offline perpetration with emotional dependency is significant only for women; that is to say, the boys’ emotional dependency did not significantly correlate with any dimension of violence. Frustration tolerance and anxious attachment are significantly related to all dimensions of violence (Table 1).
Correlations among variables of the study by sex
Note: M, mean; SD, standard deviation. The upper part of the table includes the correlations for men, whereas the lower part of the table shows the correlations among the variables for women. The mean and standard deviation for each variable are represented in the line for women, whereas for men, they are represented in the column.
* p ≤ .05.
** p ≤ .01.
To validate the structure of the scales for our sample, confirmatory factor analyses were carried out via maximum likelihood estimation (Table 2). In this sense, the scale of the dominant and jealous tactics was used as a single factor (jealousy and control offline). The confirmatory analyses of the different scales showed adequate indices. Furthermore, they had good reliability (α and ω above .70), demonstrating good internal consistency (Table 1). As in the adaptation of the CSDS instrument, the scalar invariance between being or not being in a relationship at the time of the assessment was not fulfilled; thus, we decided to replicate this analysis and obtained the same results. Despite this result, we decided to keep both groups to determine if the interpretation differences were reflected in the profiles.
Fit indices of confirmatory factor analysis
Abbreviations: CADRI, Conflict in Adolescent Dating Relationships Inventory; CDAQ, Cyber Dating Abuse Questionnaire; CFA, confirmatory factor analysis; CFI, Comparative Fit Index; CSDS, Couple-Specific Dependency Scale; DJTS, Dominating and Jealous Tactics Scale; NIF, Nor Fit Index; RMSEA, root-mean-square error of approximation; α, Cronbach’s alpha; Ω, omega.
Latent Profile Analysis
To obtain the profiles, the frustration tolerance scores and the partner dependency subscales were considered as antecedents for intimate partner violence. Sex and having a relationship at the time of the evaluation were used as covariables.
Table 3 shows the adjustment indices estimated for the different models. Models 1 (Class 2) and 2 (Class 3) adjust the data better. In Model 2, the AIC score decreases; although the BIC score is somewhat higher than that in the other model, the BLRT is more significant in this model, and its entropy index is closer to 1. Given the discrepancy between the two indices, Model 2 was chosen because the AIC consistently chooses more dynamic models than the BIC does (Medel & Salgado, Reference Medel and Salgado2013), and its BLRT and entropy are better. Moreover, Model 1 only differentiates between participants with higher scores in dependency on the partner in general and frustration tolerance and those with low scores in these factors; however, many of the studies that analyze dependency on the partner show variability in the scores of the different subscales that comprise it, and they have different impacts on other variables (Alcalá et al., Reference Alcalá, Cortes-Ayala and Vega-Cauich2021; Bell et al., Reference Bell, Howard and Cornelius2022; Lorente-Anguís & Lopez-Zafra, Reference Lorente-Anguís and Lopez-Zafra2022). These arguments indicate that Model 2 was of greater theoretical significance, supporting the selection of this model.
Latent class analysis model indices
Abbreviations: AIC, Akaike information criterion; BIC, Bayesian information criterion; BLRT, bootstrap likelihood ratio test.
* p ≤ .05.
*** p ≤ .001.
Figure 1 shows the three profiles according to partner dependency and low frustration tolerance. Profile 1 (39.1% of the sample) included people who were highly emotionally dependent, whereas anxious attachment, exclusive dependency, and tolerance frustration were moderate. This profile was labeled Anxious-dependent. Profile 2 (32.5%) is characterized by people who have no anxious attachment, no exclusive dependency, mild frustration tolerance, and modest emotional dependency; thus, it is labeled Low anxious and exclusive dependency. Finally, Profile 3 (28.4%) included individuals with low exclusive dependency, moderate anxious attachment, and frustration tolerance, and moderate but somewhat higher emotional dependency, labeled Low exclusive dependency.
Latent profiles solution.
A post hoc Kruskal–Wallis H test was performed to determine whether there were significant differences (p ≤ .05) between the scores obtained from the different profiles. Significant differences were found, except for the scores between Profiles 2 and 3 for exclusive dependency and anxious attachment between Profiles 3 and 1. When looking at the profiles and the proportion of individuals included in each profile, we observe a greater proportion of men characterized as anxious-dependent (51.3%), followed by low anxious and exclusive dependency (30.3%), and finally, a lower proportion of low exclusive dependency (18.4%). In comparison, among women, there was a greater proportion of low exclusive dependency (34.6%), followed by low anxious-exclusive dependency (33.8%), and finally anxious-dependent (31.6%). The same occurred with having or having had a partner for at least 1 month in the previous year (χ2 p ≤ .05). In particular, 50.3% of those who have a partner at the time of evaluation pertained to anxious-dependent, but only 17.3% to low exclusive dependency. The proportion of low anxious-exclusive dependency was equal (32.5%) to that of those who did not have a partner at the time of evaluation. For the latter, the percentages of anxious-dependent were 13.1% and 54.5% for low exclusive dependency. Finally, regarding age, significant differences were also found between those under 18 years of age and those over 18 (χ2p ≤ .05). Those under 18 years of age were distributed as follows: anxious-dependent (30.3%), low anxious-exclusive dependency (32%), and low exclusive dependency (37.5%), whereas 44.8% of individuals over 18 years of age were characterized by the anxious-dependent profile, followed by low anxious-exclusive dependency (32.8%) and low exclusive dependency (22.3%).
To test the role of sex, relationship status, and age, a multinomial logistic regression was performed (see Table 4). Compared with Profile 1 (anxious-dependent), girls were more likely to belong to the low exclusive dependency profile (OR = 4.27) than to the anxious-dependent profile, and were twice as likely to pertain to the low anxious and exclusive dependency profile than to the anxious-dependent profile (OR = 2.07). Moreover, participants who were not in a relationship at the time of the evaluation had more than 13 probabilities of belonging to the low exclusive dependency profile (OR = 13.44) compared with the anxious-dependent profile, and were more likely to belong to the low anxious and exclusive dependency compared with anxious-dependent (OR = 2.19). In fact, 54.5% of those who had no partner at the time of evaluation pertained to the low exclusive dependency profile, but only 13.1% pertained to the anxious-dependent profile. Finally, age had no statistically significant effect on the different profiles.
Multinomial logistic regression model of latent profiles
Abbreviations: OR, odds ratio; SE, standard error; β, standardized beta coefficient.
Next, the relationships among the profiles associated with intimate partner violence were analyzed. As shown in Figure 2, the participants did not score high in the different aggressive dimensions evaluated. Interestingly, all profiles perceived higher values of victimization than of perpetration, but no particular profile was associated only, or mainly, with a victim or a perpetrator role. Furthermore, all the dimensions had a similar pattern except for direct aggression in cyber dating abuse. In particular, the low anxious and exclusive dependency profile was associated with less psychological violence in the relationship, anxious dependency was related to intermediate violence, and the low exclusive dependency profile was associated with greater violence. Significant differences were found between the different profiles, with the exception of direct aggression perpetration and online control perpetration, between the anxious dependency and low exclusive dependency groups, and online direct aggression victimization between the anxious-dependent and low anxious-exclusive dependency groups.
Level of violence according to profile.
Note: C-P, control perpetration; C-V: control victimization; DA-P, direct aggression perpetration; DA-V, direct aggression victimization; DJ-P, dominant and jealousy tactics perpetration; DJ-V, dominant and jealousy tactics victimization; VA-P, verbal aggression perpetration; VA-V, verbal aggression victimization.
Content Analysis
To analyze which are the reasons given according to the profiles obtained, we considered the list of seven reasons given in the questionnaire and the open-ended question. From this question, a content analysis was carried out. The seven motives and five additional motives were extracted: distrust, self-doubt, anger, snooper/gossip, and consent. In the following, we list all the motives: (1) jealousy, indicating that the person feels fear/threat of abandonment by his or her partner by another person; (2) control, as they want to know what their partner does, with whom, where, and so on (e.g., “He or she only wanted me for him/herself and did not like me to leave the house without him or her”); (3) concern, when the person is worried about something bad that has happened or may happen to his or her partner; (4) bidirectionality, or behaving this way as their partner also uses these behaviors toward him or her; (5) response, indicating that the behaviors are performed in response to that same behavior previously made by their partner; (6) play/joke, meaning that they behave this way as a “normal” way of teasing the partner; (7) personality, the person attributes behaviors to the traits and qualities that make up his or her partner or her partner’s way of being; (8) distrust, when the person is afraid or thinks that his or her partner might do something to harm him or her; (9) self-doubt, meaning that the person lacks self-confidence; (10) anger, the person engaged in these behaviors because of feelings that arose at the time of an argument/fight; (11) snooper/gossip, meaning that they claim that they engage in these behaviors because they know this information, without any other reason; (12) consent, as a demonstration of trust (e.g., “trust”; “he or she leaves me his or her cell phone and I look at it”); and (13) absence, for individuals asserting these behaviors do not occur in their relationship.
For the ease of understanding the justifications given, we focus on each question and analyze the first motives expressed by the participants. For the perpetrating violence (“Why do you perform these behaviors?”), the most frequent reason expressed by the three profiles is concern about the couple (between 38.7% anxious-dependent, 36.7% low exclusive dependency, and 23.9% low anxious-exclusive dependency). For anxious-dependent, jealousy (26.5%) and game/joke (25.7%) are the subsequent reasons given. Similar results were found in a low exclusive dependency profile (30.6% for the last two). Among the low anxious-exclusive dependency, game/joke was the second justification (21.2%), followed by absence (20.6%). Analyzing in greater depth the subjects indicating concern as a justification, they do it together with jealousy in 31.6% of the anxious-dependent, 37.6% of the low exclusive dependency, and 4.9% of the low anxious-exclusive dependency.
Regarding the reasons for the question about receiving violent behavior (“Why do you think your partner performs these behaviors?”) within an intimate partner relationship (victimization), jealousy was the main motive in Profile 2 (22.7%) and Profile 3 (41.8%). In the case of Profile 2, it was followed very closely by personality (22.4%) and concern (21.3%). For low exclusive dependency, concern (31%) and game/joke (27.9%) were the justifications indicated afterward. In the case of anxious-dependent, the principal justifications were the same as in the perpetration: concern (35.3%), jealousy (29.4%), and game/joke (25.2%). Similarly, concern was associated with jealousy in a considerable percentage of anxious dependency and low exclusive dependency: Profile 1 (34%), Profile 2 (11.1%), and Profile 3 (34.8%). Justifications implying the same response to receiving these behaviors in a bidirectional way (i.e., “He or she behaves this way toward me” and “He or she did it first”) obtained very low percentages (between 4.4% and 0.4%).
Discussion
Interest in aggression among young people involved in a relationship is a great concern because of the high prevalence and figures indicating intimate partner violence at these ages (Bonilla et al., Reference Bonilla, Rivas and Vázquez2017; Rubio-Garay et al., Reference Rubio-Garay, López-González and Amor2017; Valdivia & González, Reference Valdivia and González2014). The way early relationships are constituted may help to acknowledge how they impact subsequent relationships. Research has shown that partner dependency is positively correlated with psychological violence in young couples. For example, Perles et al. (Reference Perles, San Martín and Canto2019) reported a positive relationship between dependency and psychological aggression in teen couples. Similar results have been reported in other studies (Alcalá et al., Reference Alcalá, Cortes-Ayala and Vega-Cauich2021; Perez et al., Reference Perez, Reategui, Vela, Aranda and Revelo2022). Furthermore, frustration tolerance has been correlated with violence in young couples (Pazos et al., Reference Pazos, Oliva and Hernando2014). Although both factors have been related to intimate partner violence, there are no previous studies analyzing profiles based on them.
Based on the above, the purpose of this study was to establish emerging profiles based on frustration tolerance and dependency toward the partner and its relationship with intimate partner violence. Three profiles according to partner dependency and low frustration tolerance emerged. Profile 1 Anxious-dependent (39.1% of the sample) included people who were highly emotionally dependent, whereas anxious attachment, exclusive dependency, and tolerance frustration were moderate. Profile 2 Low anxious and exclusive dependency (32.5%) is characterized by people who have no anxious attachment, no exclusive dependency, mild frustration tolerance, and modest emotional dependency. Profile 3 Low exclusive dependency (28.4%) included individuals with low exclusive dependency, moderate anxious attachment, and frustration tolerance and moderate but somewhat higher emotional dependency.
Once we had the profiles, we tested the hypothesis. Hypothesis 1 is only partially fulfilled, since the low anxious-exclusive dependency profile was associated with less violence in the relationship as expected. However, the low exclusive dependency profile correlates with greater violence than the anxious-dependent profile yielded higher levels of dependency in all its subscales. This might seem contradictory; however, a closer look at the data revealed that a very high proportion of participants who were not in a relationship at the time of the evaluation were in the low exclusive dependency profile (54.5%). Indeed, this profile score shows a very low dependency on exclusivity; therefore, people who were not in a relationship at the time of the evaluation may identify less exclusive dependency than the other group. People who are not in a current relationship may have an altered perception of their exclusive dependency on their partner. In fact, Lorente-Anguís and Lopez-Zafra (Reference Lorente-Anguís and Lopez-Zafra2022) reported that subjects who were and were not in a relationship at the time of the evaluation interpreted the CSDS differently, indicating that the latter could have different perceptions of the dependency in their last relationship. This is an important contribution of this study because previous studies on this topic do not report data on whether the participants were in a relationship at the time of the assessment, which precludes possible comparisons (Rubio-Garay et al., Reference Rubio-Garay, López-González and Amor2017). Another possible explanation of these results could be that most of the studies analyzing the relationship between partner dependency and the violence exercised or received in the relationship use the total score of the scale (Garrido-Macías et al., Reference Garrido-Macías, Valor-Segura and Expósito2020; Perles et al., Reference Perles, San Martín and Canto2019; Valor-Segura et al., Reference Valor-Segura, Expósito and Moya2014). However, when partner dependency is analyzed by observing its different aspects/dimensions, the fear of being alone is one of the aspects that most correlates with partner violence. The results of Alcalá et al. (Reference Alcalá, Cortes-Ayala and Vega-Cauich2021) indicated that avoiding being alone predicts the degree of violence experienced by both men and women. Similar results were found for perpetrating for men, but no prediction was found for the other three factors of partner dependency. De los Reyes et al. (Reference De Los Reyes, Jaureguizar and Redondo2022) reported similar results. Cybervictimization was positively correlated with partner dependency, particularly avoiding being alone and being involved in an asymmetrical relationship. Moreover, cyberperpetration was positively correlated with the need for exclusivity and avoiding being alone. Bell et al. (Reference Bell, Howard and Cornelius2022) reported that anxious attachment was positively related to psychological aggression in women, but the emotional dependency subscale did not correlate with it. Individuals with a high level of anxious attachment show a strong need for intimacy and fear of being rejected by their partners (Bonache et al., Reference Bonache, Gonzalez-Mendez and Krahé2016). Furthermore, Bartholomew and Allison (Reference Bartholomew, Allison, Mikulincer and Goodman2006) indicated that people in coercive relationships tend to be anxious. People with anxious attachment exhibit more violent behaviors toward their partners due to their vulnerability and fear of abandonment (Mikulincer & Shaver, Reference Mikulincer, Shaver, Forgas, Kruglanski and Williams2011). Additionally, another interpretation of these results could be that people who are not involved in a couple relationship at the time of the evaluation could present less social desirability when responding to violence in their past relationship. In conclusion, as our study focused on profiles, the results pertaining to Profile 3 were the most related to violence, which could indicate that individuals with less tolerance of frustration together with overt anxious attachment behaviors are associated with greater violence when exclusive dependency is low and emotional dependency is moderate. High scores on these two expressions of partner dependency could result in individuals who do not display or perceive their partner’s behaviors as violent.
Thus, Hypothesis 2 was fulfilled because none of the profiles correlated solely with only receiving aggressive behaviors or perpetrating them. Instead, they were associated with bidirectional violence, and the difference between them was the frequency of reported violence. As other studies have also indicated, violence in young couples is often bidirectional (De la Villa et al., Reference De la Villa, García, Cuetos and Sirvent2017; Laforte et al., Reference Laforte, Paradis, Todorov and Cyr2023; Rojas-Solís et al., Reference Rojas-Solís and Romero-Méndez2022). The justifications reported by the participants of the different profiles are logical, as justifications labeled as unidirectional, such as “He or she exerts these behaviors on me” and “He or she did it first,” obtained low percentages.
The different profiles expressed similar justifications in terms of perpetration; therefore, Hypothesis 3 was not fulfilled. The concern was the most frequent justification; however, it was mostly associated with jealousy. The main justifications for victimization behaviors were jealousy and concern. Similar results have been reported in other studies (Bonilla et al., Reference Bonilla, Rivas and Vázquez2017; Borrajo & Gámez-Guadix, Reference Borrajo and Gámez-Guadix2015; Rojas-Solís & Romero-Méndez, Reference Rojas-Solís and Romero-Méndez2023). This could be explained by the myth of romantic love, which includes some unreal and distorted beliefs about love (Martínez et al., Reference Martínez, Lopez-del Burgo, Albetos and Ibabe2024), normalization of jealous behaviors, and support of beliefs such as the myths of romantic love (Ramírez-Carrasco et al., Reference Ramírez-Carrasco, Ferrer-Urbina and Ponce-Correa2023). This perception could lead young people to consider jealousy as a sign of love and affection (Rojas-Solís & Romero-Méndez, Reference Rojas-Solís and Romero-Méndez2023).
In summary, this study increases knowledge about the psychosocial factors associated with offline and online psychological violence in young couples. This study further contributes in several aspects. First, it is important to take into account the status of the relationship when measuring dependency on the partner, as well as the separate evaluation of each of its dimensions, since, as we have seen in this study, each one influences in different ways. This result opens a new line of research on approaching dependency analyses in violence in couples. On the other hand, it indicates that the conclusions of previous studies that have not included these issues should be viewed with caution. It also highlights the importance of considering frustration tolerance in the study of violence in young couples, which has received little attention so far. Finally, it allows us to delve deeper into the motives that young people express to justify violence in their relationships.
This study has several limitations. First, the evaluation of the participants was self-reported, which may have generated social desirability biases in the participants, which were minimized through anonymity. Second, the levels of violence reported by the participants were not very high, since most of them reported that they rarely engaged in these behaviors. This is interesting, as many of the studies of young people emphasize the prevalence of violent behaviors. Our sample was representative of young people; thus, identifying profiles may help detect who might be more prone to engage in violent behaviors. In our study, all participants were in a heterosexual relationship; thus, it would be advisable in future studies to extend this study to a diverse sample while considering other sexual orientations. All the participants were students, although there were several educational centers in different regions of the country, and the participants had different levels of education, thus including a heterogeneous sample. Finally, in our study, we only had information from one of the partners; thus, for future research, the sample could be composed of engaged couples to allow for comparing both profiles and the motives expressed. Nevertheless, these results can be valuable when carrying out prevention programs to address the dependency on the partner and frustration tolerance in relation to psychological violence in young couples.
Data availability statement
Data are available by reasonable request.
Author contribution
Conceptualization: A.L.-A., E.L.-Z.; Data curation: A.L.-A.; Formal analysis: A.L.-A.; Funding acquisition: E.L.-Z.; Investigation: A.L.-A., E.L.-Z.; Methodology: all authors; Supervision: E.C.-G., E.L.-Z.; Validation: E.L.-Z.; Visualization: all authors; Writing—original draft: A.L.-A.; Writing—review and editing: E.L.-Z.
Funding statement
This study was founded by the “Pacto de Estado contra la Violencia de Género” (State Pact against Gender Violence) through the Consejería de Economía, Innovación, Ciencia y Empleo, Junta de Andalucía (EI_SEJ7_2017).
Competing interests
The authors declare no competing interests.
Open access
