Risk and Protective Factors in Child-to-Parent Violence: A Study of the YLS/CMI in a Spanish Juvenile Court

Child to Parent Violence (CPV) is one of the crimes with the highest rates of increase in Spain in recent years, and has had a significant media and social impact. This phenomenon has been analyzed with different methodologies and samples, but very few studies have used a standardized instrument such as the Youth Level of Service/Case Management Inventory (YLS/CMI). This study examined the heterogeneous profile of youths in a judicial context. The aims of the study were: 1) to evaluate whether there were any differences in sociodemographic and criminogenic factors in a group of youths who committed CPV in comparison to a group who committed other type of crimes. 2) to analyze which risk factors in the YLS/CMI Inventory are the best predictors of CPV. The participants were 341 youths aged between 14 and 17 years old with a record in the Juvenile Court of a Spanish province. The sample covered crimes committed during a follow-up period from 2011 to 2017. The results showed that the CPV group had a higher risk profile than the comparison group. The Family circumstances, Substance abuse and Personality behavior subscales of the YLS/CMI were able to predict CPV among these youths. The CPV profile showed higher risk than the comparison group. A poor family environment combined with an aggressive personality, together with substance abuse predict the commission of a CPV offense. The education of the parents is of vital importance. The CPV profile showed higher risk than the comparison group. A poor family environment combined with an aggressive personality, together with substance abuse predict the commission of a CPV offense. The education of the parents is of vital importance.

The CPV profile showed higher risk than the comparison group.
• A poor family environment combined with an aggressive personality, together with substance abuse predict the commission of a CPV offense.
• The education of the parents is of vital importance.

Description and Prevalence of Child to Parent Violence
Child to Parent Violence (CPV) is one of the crimes with the highest rates of increase in Spain in recent years, and has had a significant media and social impact (Condry & Miles, 2012Pereira, 2006). The first definitions of the phenomenon were short and limited in operational terms. Harbin and Madden (1979) defined this type of violence as physical attacks, verbal and non-verbal threats or physical harm exerted by children towards parents. Aroca (2013) argues that the two key factors that must be considered when defining this phenomenon are intentionality and repetition, and maintains that in this type of violence, the child acts consciously and repeatedly with the intention of obtaining/maintaining power, control and dominance over their parents, causing harm to their victim, with the immediate aim of fulfilling their wishes, using psychological, economic and/or physical violence.
The Spanish Society for the Study of Child to Parent Violence , defines it as follows: a set of repeated behaviors of physical, psychological or economic violence (verbal or non-verbal) by sons and daughters against their parents, or the adults who take their place. This violence is associated with both the search for control and power, and with the achievement of specific objectives which are beneficial to the child. This violence, which is of an indeterminate duration, ends when absolute control over the parents is achieved. It increases gradually, and usually begins with insults, then continues with threats, and finally with increasingly serious physical assaults or even death (Pereira & Bertino, 2009).
However, rates of recidivism and prevalence are not always accurate, since parents have difficulty reporting that they are experiencing abuse by their son or daughter due to feelings of shame or blame (Downey, 1997;Edenborough et al., 2008). Rates of prevalence also depend on the definition used. The use of different methodologies, procedures and samples in studies, and even the different definitions of the violence taken into account in each one, hinder comparisons of results and rates, and hampers accurate measurement of this phenomenon. Studies have found prevalence of CPV ranging from 3 to 27% (Gallagher, 2008;Holt, 2012a, b), and in specific terms, the prevalence rate ranges from around 20% for physical assault, around 60% for verbal aggression, and increases to 90% for psychological violence (Calvete et al., 2014(Calvete et al., , 2017Ibabe, 2015;Jaureguizar et al., 2013;Loinaz & de Sousa, 2020). Based on the judicial data provided by the Annual Report of the Spanish State Attorney General's Office, the figures fluctuated between 4300 and 4800 reported cases between 2013 and 2019, and exceeded 5000 in 2020. This is not an excessive increase but it is revealing, as the figures for the entire decade are very high compared to other crimes. For example, there is a higher incidence of CPV than burglary. These data are even more alarming when the reluctance of parents to report cases is taken into account, and may only show the tip of the iceberg. This increased awareness of CPV in Spain has resulted in a greater investment in CPV research (Contreras et al., 2021). This increase in the frequency of reports has been accompanied by a certain treatment in the media, unknown in the rest of the countries, which has increased the social sensitivity of the problem (Altea, 2008). Some specific factors in this country wich could explain this situation are the following. Diverse studies show that Spanish families have swung from an authoritarian parenting style to an indulgent and permissive style which increasingly leads to symmetrical relationships (García & Gracia, 2009). This is associated with an important change in the power cycles within the family (Tew & Nixon, 2010). Furthermore, in Spain, in comparison with other European countries, youths tend to spend more time with their families and they maintain very high degrees of interrelation and family dependence (Alberdi & Escario, 2007). As a result, emancipation from the family occurs later than in other countries. Consequently the length of time youngsters remain living in the parental home may lead to a more probability of conducting CPV.

Risk Factors for Child to Parent Violence
Numerous studies in various countries show that child to parent violence is generally committed more frequently against mothers than fathers, albeit with some variations. In the first analysis of cases of adolescent to parent violence reported to the police in the United Kingdom, 77 per cent of the victims were female, and 66 per cent of cases involved a son-to-mother relationship (Condry & Miles, 2014;Pagani et al., 2004). In American studies, children were also more violent to mothers than to fathers at all ages, and in both genders (Ulman & Straus, 2003). Meanwhile, in a Spanish study, adolescents more frequently behaved violently in general towards their mother than towards their father, with significant differences (Ibabe & Jaureguizar, 2009).
Parent-child relations are dynamic interactions which are prone to conflict and turmoil. CPV and other crimes, for example parricide, are connected through a complex process of escalation, whereby CPV occurs at an earlier stage in the cycle of family violence (Ibabe & Jaureguizar, 2009;Walsh & Krienert, 2009). The literature has traditionally held that youths who commit parricide are frequently responding to ongoing and intolerable abuse, and parricide may be an extreme final stage in the escalation of violence (Walsh & Krienert, 2009). In comparison with CPV youths, patricidal youths are more likely to resort to extreme violence as a means of conflict resolution. In addition, they are more desensitized to violence, and more willing to resolve family conflicts through extreme violence.
In terms of the adolescent's who engage in CPV, gender studies show a higher prevalence for boys (with rates of between 60 and 80%) compared to girls (Aroca et al., 2014;Cottrell & Monk, 2004). Some studies found somewhat higher proportions for girls, but they did not exceed those reported for boys (Romero et al., 2005;Webster, 2008). When these data are compared with general samples of youths, the studies show a majority male profile, with rates of around 80% (Alcázar et al., 2015;Cuervo et al., 2017a, b, c;Iborra et al., 2011). As for the type of violence used, boys tend to engage in physical assault (Boxer et al., 2009;Walsh & Krienert, 2007), while girls use verbal and psychological abuse (Calvete et al., 2013;Estévez, 2016;Ibabe, 2015;Ulman & Straus, 2003). Mothers were the most common victims of aggression in this type of violence, particularly in one-parent families. They were often the main caregivers, and hence were more frequently exposed to the violence (Gallagher, 2004;Ibabe & Jaureguizar, 2011). The presence of gender violence in the history of the family may also have some influence in this type of abuse (Ibabe & Jaureguizar, 2011;Patró & Limiñana, 2005).
One crucial factor in this regard is the family and the factors related to it: the parents' gender, family structure, context, relationship, the gender of the offspring, etc. The family is the first and most important socializing context, and through it the child learns how to interact with other people and in other environments. It provides the basis for their communication and their interpretation of the behavior and intentions of others (Bjorklund & Hernández, 2011;Bornstein, 2012). Parenting styles are considered part of the relationship between parents and their children. Abuse was traditionally related to authoritarian parenting styles characterized by high levels of control over children, and the use of corporal punishment as a disciplinary method. Studies on the subject have found that the use of corporal punishment increases resentment and frustration in children, who take advantage of their abilities to engage in violence against their parents and obtain revenge (Aroca, 2013;Calvete et al., 2011;Cottrell & Monk, 2004;Gallagher, 2004;Pagani et al., 2004). This strict control may adolescents feel humiliated, infantilized and resentful (Strauss & Stewart, 1999). Cottrell and Monk (2004) present a common family pattern, in which youths use violent behavior against excessively controlling parents. These authoritarian methods were effective when children assumed that they needed to obey, but the tension started when the youth showed a need for autonomy. At this point, parents do not know how to negotiate or compromise, and attempt to maintain the same level of rigid control, and as a consequence, youths may use abusive behavior in an effort to obtain a sense of power.
However, this violence has been also associated with excessively permissive and forgiving educational practices or styles (Ibabe, 2015). A lack of clear boundaries and limited sanctions may lead to a poor control and inadequate supervision, which makes facilitates this behavior. These consequences have also been related to poor communication and poor quality of time spent as a family (Aroca & Robles, 2012;Aroca et al., 2012;Cottrell, 2005;Paulson et al., 1990;Robinson et al., 2004). Inconsistent rules lead parents to contradict each other, and to fail to provide security or impose limits, which means they are unable to stop their children's violent behavior (Aroca et al., 2014). It is difficult to impose discipline and rules and monitor compliance without any barriers or a defined hierarchical structure within the family (Calvete et al., 2011). Other researches have also highlighted the negative consequences of permissive parental styles. When the youth is used to getting what he or she wants and occasionally this does not happen, violent behavior appears (Cottrell, 2005;Paulson et al., 1990;Robinson et al., 2004). In these cases, the parent-child relationship is sometimes overly symmetrical, and when parents try to exert authority or to impose discipline, the son or daughter may not obey and respond with violence and disobedience (Agustina & Romero, 2013). As a result, both types of parenting are related to antisocial behavior in adolescence. In both cases, there are general difficulties in child-parent relationships, and in particular among parents who have unrealistic expectations, and a lack of adequate communication skills (Kennedy et al., 2010;Paulson et al., 1990;Peek et al., 1985).
Personality traits have been studied to a more limited extent than the factors mentioned above. CPV has been related to aggressive behavior in general, and to hostile perceptions in social relationships (Card & Little, 2006;Dodge & Pettit, 2003). Children who mistreat their parents present aggression, impulsiveness, traits of psychopathy, lack of empathy, tantrums and low self-esteem. Some of them present apathy, tend to have fewer friends and even suffer from social isolation (Aroca et al., 2014). These youths presented lower levels of tolerance of frustration, and more confrontational and aggressive behaviors even compared to children with problematic behaviors. Furthermore, these youths seem to be more demanding than children with problematic conducts (Cottle & Monk, 2004).
Another risk factor that seems to be related to this type of violence is low achievement at school. Academic difficulties are very commonly present in these youths, in 53% (Rechea-Alberola et al., 2008), and 93% of the cases (Ibabe et al., 2007). Likewise, academic failure varies between 32.7% reported by Rechea-Alberola et al. (2008) to the 67.2% by Romero et al. (2007). This misbehavior is also present in the employment sphere, as they are more likely to leave their jobs due to difficulties with complying with rules and schedules and behaving irresponsibly (Aroca et al., 2014). Lastly, in the psychological sphere, the following variables have been reported as possible risk factors: aggressiveness, impulsiveness, psychopathy, low levels of frustration, ADHD, lack of empathy, lack of anger control, low self-esteem and irritability. In addition, the presence of apathy, social isolation and the lack of a group of friends were also considered (Aroca et al., 2014).
The use of drugs has been frequently related to CPV youths. Drug abuse was present in 59% of the reported CPV cases (Romero et al., 2005). Substance abuse by adolescents increased the likelihood of verbally abusing the mother by 60%. Furthermore, drug use appeared to increase the severity of the violence (Aroca et al., 2014;Calvete et al., 2011;Cottrell & Monk, 2004;Pagani et al., 2004Pagani et al., 2009). This effect is particularly marked in males (Walsh and Krienert 2007). However, various studies concluded that the use of alcohol and drugs was not a predictor of CPV, but instead a factor that exacerbated the problem. (García de Galdeano & González, 2007;Garrido, 2010;Pantoja, 2005;Price, 1996). This abuse may therefore act as a catalyst for the aggression, which leads youths to higher rates of conflict and prevents them from using coping mechanisms that would stop the abuse (Calvete et al., 2012). This might explain the relationship with both types of aggression (i.e., physical and verbal) (Pagani et al., 2004). Meanwhile, peer influence has been considered a classic risk factor in delinquency (Farrington & West, 1993), and has also been present in child to parent violence. Aroca et al. (2014) found that an antisocial peer group correlated with this type of violence in half of the studies included in their metaanalysis. It is a fact that there is a strong social component to antisocial behavior in adolescence (Xu et al., 2020), and that these acts are more likely to occur in groups (Sijtsema & Lindenberg, 2018). Peer pressure has been hypothesized to be an important contributor to deviant and risky behaviors in adolescence, as minor delinquency, serious offending, reckless driving, and drug and alcohol use (De Somma et al., 2021;Yoon, 2020). Taking into account The Big Four Model (Andrews & Bonta, 2016) and the results obtained by Cuervo et al. (2017a, b, c), the presence of antisocial peer relationships is among the strongest predictors of recidivism, as these antisocial relationships can influence negatively on the decisions adolescents make when having the opportunity to engage in infringing acts.
In contrast, in the case of general judicial samples, the YLS/CMI areas that were found to be the best predictors of recidivism were Leisure/recreation, Education/employment and Substance abuse (Cuervo & Villanueva, 2015;Garrido, 2009;Graña et al., 2007). The most significant risk factors related to recidivism seem to be the factors related to the minors' social context, their education and management of their spare time (Cuervo & Villanueva, 2015).

Protective Factors for Child to Parent Violence
Protective factors can be considered as variables that predict a low probability of offending among youth exposed to risk factors, as the "mirror image" of a risk factor (White et al., 1989) or as variables that interact with a risk factor to nullify its effect (Farrington et al., 2016;Rutter, 1985). The main protective factors for CPV are those related to the family. A positive family environment, characterized by warmth, parental monitoring, support from family members to encourage positive development and family bonding in therapy were found to be protective against CPV (Beckmann et al., 2017;Ibabe et al., 2013;Loinaz et al., 2017). A secure attachment style and a democratic educational style, which are consistent among the parents, and family flexibility or adaptability have also been related to avoidance of CPV (Llamazares et al., 2013). As regards the social network, the presence of social support through friends, positive relationships at school, adherence to prosocial behaviors and the existence of prosocial academic or future employment plans stopped this type of abuse (Jaureguizar et al., 2013;Llamazares et al., 2013;Loinaz et al., 2017). Self-control, motivation for change and the therapeutic alliance are examples of individual features related to the avoidance of CPV (Beckmann et al., 2017;Loinaz et al., 2017).

Standardized Assessments
Various dynamic risk and protective factors have therefore been related to this type of violence. Some research about the family context, type of family and personality traits has been carried out, but no studies have used a standardized instrument for the assessment of risk factors in order to determine which specific factors are related to CPV. Several studies have recognized the predictive superiority of standardized methods compared to clinical assessments (Hoge, 2002;Steadman et al., 2000). These types of methods have provided professionals with an objective guide, unifying criteria in the aspects to be evaluated, and reducing possible interviewer biases. It is also difficult to interpret information from psychosocial teams and institutions, as there is no standardized protocol or instrument. The use of protocols or structured evaluation systems would therefore significantly increase the reliability and validity of the results (European Economic and Social Committee, 2006;San Juan & Ocáriz, 2009). Scientific findings should be used in each country's policies from the perspective of evidence-based policy .
The Youth Level of Service/Case Management Inventory (YLS/CMI) (Hoge & Andrews, 2006) is an objective scale which assesses various risk factors. It is considered a valid predictor of criminal recidivism Graña et al., 2006). Meanwhile, protective factors can be considered variables predicting a low probability of offending (Farrington et al., 2012;Hartman et al., 2009). Most studies have focused on risk factors rather than on protective factors, despite acknowledgment of their evident beneficial role. This study therefore includes both types of factors in order to determine how they are related to CPV (Haines & Case, 2008;Hoge et al., 1996). Identifying the risk needs linked to the persistence or otherwise of CPV among youths is essential for preventing or reducing these conducts in adulthood. In fact, some research has shown that this violence is related to other abuse in adulthood, such as gender violence. The use of standardized instruments gives courts an accurate method to detect the risk level of youths so that they can implement the appropriate interventions. The use of these instruments is therefore essential for preventing future crimes (Ibabe & Jaureguizar, 2010). The YLS/CMI is one of the most widely used recidivism risk assessment mechanisms for youths, and emphasizes the link between assessment and case management (Fourth-Generation risk assessment). It is widely used in forensic populations, and has been designed to aid probation officers, juvenile workers, psychologists and social workers in assessing adolescents.
In predicting criminal behavior, according to Andrews et al. (2006), there are four generations or stages, which have evolved over the past 30 years (Andrews & Bonta, 2016): In the First-Generation Risk Assesment predictions are based on clinical judgment which depends on the experience of the professional, using informal and unobservable criteria. The reasons for making the decisions may be subjective, sometimes intuitive and not empirically validated. The Second Generation consist almost entirely of static or historical items, that is, those belonging to the child's past or to circumstances that are difficult to modify. Therefore, items considered as predictors, such as crime history, have little credit to the opportunity of improvement or recovery. Whereas in the Third Generation of Assesment the offender dynamic needs are measured. Therefore, there is space for improvement or recovery, rather than focusing on static factors as in the previous generation and assessment are theoretically based. Finally, the Fourth Generation emphasizes the link between assessment and case management. This generation focuses the assessment of specific responsivity factors to maximize the benefits from treatment interventions, and the structured monitoring of the case from the begining to the end of the supervision. Furthermore, this generation acknowledges the role of personal strengths in building a prosocial orientation. Examples of other fourth generation tools include the Correctional Offender Management Profiling for Alternative Sanctions (COMPAS), Ohio Risk Assessment System (ORAS), and Wisconsin Risk and Needs tool (WRN) (Desmarais & Singh, 2013) and the Offender Assessment System (OASys) (Howard et al., 2006).
The detection of risk factors may play a critical role in predicting CPV in youth populations. The identification and detection of a particular need related to a specific type of crime could be useful for intervention purposes, thereby increasing the effectiveness of the programs. The Risk-Need-Responsivity [RNR] model (Andrews & Bonta, 2016), distinguishes between criminogenic and noncriminogenic needs. Criminogenic needs are the risk factors specifically related to recidivism and the model is used as an offender rehabilitation worldwide. This may be useful in regulating the intensity and urgency of the supervision. Very few studies in Spain related to child to parent violence among judicial populations have been carried out. To our knowledge, no study has analyzed CPV using a fourthgeneration instrument in judicial samples, which is a prerequisite (Ibabe & Jaureguizar, 2010). This study aims to provide an in-depth understanding of this type of violence using a standardized instrument, and to examine the heterogeneous profile of these youths in a judicial context. Bearing previous findings in mind, the aims of the study are: (1) To evaluate whether there are any differences in terms of sociodemographic and criminogenic factors (e.g. gender, age, recidivism, number of offenses, risk and protective factors) in a group of youths who committed CPV, in comparison to a group who committed other type of crimes.
(2) To analyze which risk factors in the YLS/CMI Inventory are the best predictors of CPV.
(1) We hypothesize that the CPV group will present a higher risk profile than the comparison group (Cuervo et al., 2017a, b, c;Kennedy et al., 2010;Moulds et al., 2019).

Participants
The data for this study were obtained from an analysis of the records of 341 youths with a disciplinary record from the Juvenile Court of a Spanish province from 2011 to 2017, covering 7 years of follow-up. The participants' ages ranged from 14 to 17 years (M = 15.86, SD = 1.02). The majority (248 participants) were male (72.7%), although 93 (27.3%) were female. All youths who had committed child to parent violence from 2011 to 2017 were included in this study as the CPV group, and youths who had committed other types of crimes were included as the comparison group. The youths included in this group include those committing all types of offenses, ranging from minor offenses to crimes committed against persons and property, such as shoplifting, fraud, reckless driving, robbery, as well as assault and sexual assaults. The child to parent violence group consisted of 153 youths (44.9%), and the other group was made up of 188 youths (55.1%). The comparison group was selected by deleting one youth in every 4, listed by their case number in the same period of time.

Instrument
The Youth Level Service of Case Management Inventory (YLS/CMI) Hoge & Andrews, 2006), is a recidivism risk hetero-assessment inventory, which consists of 42 items grouped into eight risk factors. Each item can be marked as present or absent. The eight factors are as follows: (1) Prior and current offenses/adjudications; (2) Family circumstances/parenting; (3) Education/ employment; (4) Peer relations; (5) Substance abuse; (6) Leisure/recreation; (7) Personality/behavior; (8) Attitudes/ orientation. The total score provides a total recidivism risk level for each youth, which can be classified as low (a score from 0 to 8 points); moderate (a score from 9 to 22 points); high (a score from 23 to 32 points); and very high (a score from 33 to 42 points). The recidivism risk can also be assessed on each subscale. The Inventory also allows factors of strength (protective factors) to be recorded. The assessor can indicate whether one specific factor might be considered as one of the young person's strengths. Protective factors are not considered as merely the absence of risk in a factor, but the explicit presence of a positive factor. This option exists for all subscales except Prior and current offenses, because the absence of offenses would be normative for all participants instead of protective. Part III of the YLS/CMI contains a section where the assessor can record information about a range of variables that may be relevant to disposition, intervention or case planning. These variables are considered as Responsivity factors, and are not included in the risk assessment. The assessor marks if the item applies to the client, and whether it is present or not (0-1). "Ethnic or cultural problems" refers to ethnic or religious problems or conflicts related to the lack of family adaptation. One of the most common cases in practice are Arab families who have just arrived in Spain without knowing the language or the culture, and who may have difficulty communicating, finding work, adapting to school or integrating with the rest of society. "Financial problems" refers to cases in which the family income is not high enough to cover basic needs. Some indicators for the evaluation would be receiving help from social services, residing in statefunded or supervised housing, living in precarious conditions, in the street or in overcrowded or squatted housing. The information was obtained from various sources by a member of the Psychosocial Team in the Juvenile Court.
These sources included interviews with the adolescent and his or her family, prior court records, and data from other social centers with which the youth was or had been associated.

Procedure
In Spain, minors must be from 14 to 17 years old to be charged under the youth legal system. According to the country's Organic Law 5/2000 (OL 5/2000), Spain has a specialized system for youths. A disciplinary record is created in the Juvenile Court when a youth is accused of committing a crime. In this study, records were compiled for all the charges that occurred in 135 municipalities, covering over 600,000 inhabitants. The disciplinary record is created with the criminal and personal information available for each youth. The Psychosocial Team of the Juvenile Court interview both the juvenile and his or her legal representatives about the individual, educational, family, and social aspects in the youth's environment. The YLS/CMI is scored based on the information from the interview. The authorization to apply the YLS/CMI to this sample was obtained from Multi-Health Systems (MHS). The interviews took place at the Juvenile Court around 3 to 6 months after the commission of the crime and before the trial, in order to assess the risk level of recidivism. This information was compiled in a report for the prosecutor, with the recommended interventions based on the risk level of the instrument.
Child to parent violence is included in the Spanish Criminal Code as an "intra-family violence" or "domestic violence" crime. When carrying out the study, the Psychosocial Team confirmed that all cases in the sample were related to CPV. These were all the cases of child-to-parent violence committed in this period of time. The criterion for classifying any youth (in both the CPV and the comparison group) as a reoffender was as follows: any youth who had committed any other crime(s) since his or her first recorded offense during the follow-up period.

Data analysis
The descriptive and predictive results for the CPV and comparison group were analyzed. For the descriptive results, X 2 analyses were used with dichotomous variables, and an ANOVA analysis was used for the continuous variables, in order to examine the statistically significant differences between the two groups. The different YLS/ CMI recidivism risk level scores and risk subscales were also examined. A logistic regression was used for the predictive results, as this strategy provides enough information about prediction (Flores et al., 2017).

Results
The following results are conducted to analyze the differences between the CPV group and the comparison group. Sociodemographic and criminological variables will be presented in Table 1. The general and subscales risk level are presented in Table 2 and Fig. 1. It is expected that the highest levels for the CPV group would be for Family circumstances/ parenting, Education/employment and Personality/behavior. Meanwhile, for the comparison group they would be Leisure/ recreation, Education/employment and Substance abuse. Table 1 shows the differences in sociodemographic and criminological variables. As shown in Table 1, the gender in both groups was mostly male. In the comparison group, boys accounted for a significantly higher percentage (80.3%) than in the CPV group (63.4%), and as such the difference is significant (Χ 2 (1) = 12.1, ***p < 0.001). There were no significant differences in age, ethnicity or financial problems between the two groups.

Descriptive Analyses
As regards judicial histories, there was a significant difference in recidivism between the two groups, as most child to parent violence youths were reoffenders (60.1 versus 30.3%). This differentiation was also clear in the number of cases. The  CPV group presented a mean of 2.67 cases, whereas the comparison group presented 1.64 cases (these cases refer to any type of crime). Furthermore, the maximum number of accumulated cases in the CPV histories was 15, whereas the maximum number of cases in the comparison group was 8. Table 2 shows the statistically significant differences in all the subscales of the YLS/CMI Inventory when comparing the two groups. The largest differences between risk factors were in the subscales of Family circumstances/parenting, Personality/behavior and Substance abuse. An analysis of the YLS/CMI total risk provides similar results. The CPV group showed a higher risk of recidivism than the comparison group (F(56) = 2.24, ***p < 0.001). Figure 1 presents the distribution of the recidivism risk levels identified by the YLS/CMI Inventory for each group. The highest percentage of youths belongs to the CPV group, with 68.2% of cases in the moderate recidivism risk level, and is followed by 55.5% of the youths in the low risk level in the comparison group. The high risk level of recidivism was 20.5% of the CPV group, in contrast to 6% in the comparison group.
The presence of protective factors in both groups was also analyzed. As Table 3 shows, the youths who had committed child to parent violence presented very few protective factors. When the two groups were compared, the comparison group presented higher and significant rates of protective factors than the CPV group in the following subscales: Peer relations, Substance abuse, Personality/behavior and Attitudes/orientation. When the total sum of the risk factors was analyzed, the CPV group presented a mean of 0.11 protective factors per youth, whereas for the comparison group it was 0.55, F(1, 272) = 13.76, ***p < 0.001.

Predictive Analyses
The following analyses were conducted to check the prediction of CPV with the YLS/CMI subscales. A binary logistic regression was carried out with the crime group of youths as dependent variables (CPV = 1; Non CPV = 0) and the subscales of the Inventory (Table 4). Three of the subscales of the YLS/CMI predicted the commission of child to parent violence CPV. These variables were Family circumstances/parenting, Substance abuse and Personality/ behavior, whereas the Antisocial peers subscale was negatively related to CPV. A larger score on this subscale indicates that the odds of belonging to the CPV group are reduced. This subscale therefore predicts the commission of general offenses.

Discussion
The first aim of this study was to examine a number of sociodemographic and criminogenic factors, such as the general risk level of recidivism with the YLS/CMI, the risk of each subscale, protective subscales, recidivism, etc. among youths who had committed child to parent violence, in comparison to a group of youths who had committed general offenses. The second objective was to analyze the predictive ability of YLS/CMI for CPV, identifying the specific risk subscales and checking whether the inventory is able to predict CPV.
A higher risk profile was anticipated in the CPV group. This hypothesis was confirmed. Youths in the CPV group present higher rates of recidivism, more crimes committed and higher total scores on the YLS/CMI Inventory and for its subscales. The gender was mostly male in both groups. However, boys accounted a big percentage from the comparison group (80.3 versus 63.4% in CPV). These results are consistent with the studies consulted, as some authors state that there is a higher proportion of girls in the child to parent violence group compared to youths who commit other crimes (Romero et al., 2005;Walsh & Krienert, 2007;Webster, 2008). Age did not differ significantly. It appears that a peak age in the commission of this crime cannot be established, since it is a continuous phenomenon over time and when compared with the various studies, it depends on the samples studied, the methodology used and the inclusion criteria. In relation to the gender of the aggressors and victims, the context of the incident when the arrest was made should not be overlooked. Armstrong et al. (2018a, b) found that police officers were 1.95 times more likely to arrest a male aggressor (i.e., the son) when the victim was female (i.e., their mother) and they were no more likely than females to be arrested when they victimized a male (i.e., their father). It seems that as children grow older, the number and cruelty of aggressive acts increases, possibly because of their increased physical capacity and the severity of the situation (Walsh & Krienert, 2007). In contrast, the opposite is true of daughters, i.e., as the years go by, the aggressions are less intense and less frequent (Pérez & Pereira, 2006). One possible explanation for this is the relationship of CPV with gender violence, and the perception that a boy is a more serious threat than a girl (Calvete et al., 2015;Ibabe & Jaureguizar, 2010). In the judicial sphere, it seems that sons are more likely to be reported that daughters (Gallagher, 2008). Another explanation could be related to earlier maturity in girls, which may lead to a reduction of punishment (González-Álvarez et al., 2011;Rechea-Alberola et al., 2008). The recidivism rate among the first group was 60.1%, whereas in the comparison group it was 30.3%. For general samples of youths, the studies show recidivism rates consistent with this percentage, of around 30% (Capdevila et al., 2005;Cuervo & Villanueva, 2013;García-España et al., 2011;Ortega-Campos et al., 2014). Armstrong et al. (2018a, b) found that around 70% of youths who had committed CPV had been in custody between 1 and more than 5 times, and around 95% had prior involvement or convictions in the juvenile justice system. Ibabe and Jaureguizar (2010) also reported higher rates of offending in the CPV group than in the group committing other crimes. However, not all the studies report differences in this regard. No differences were found for CPV juveniles in terms of the number of arrests or charges in the study by Gebo (2007) and Kennedy et al. (2010). In juvenile court records, Kennedy et al. (2010) also found that CPV youths did not significantly differ from non-CPV youths in terms of the number of prior arrests. Non-CPV involved youths are more likely to have a history of property offenses, whereas prior offenses tended to involve other forms of violence among CPV-involved youths.
The child to parent violence youths group present higher levels of risk in all the subscales of the inventory. All the differences are significant. The biggest differences were found for the following factors: Family circumstances/parenting, Personality/behavior, Substance abuse and Education or employment. Disruptive behavior at school was also found to be a predictor of these types of offenses, as various studies (Ibabe & Jaureguizar, 2010;Pagani et al., 2004) found that Spanish youth who committed CPV were more likely to have presented various difficulties at school, including disruptive behavior, than adolescents who were charged with other offenses. In fact, academic factors and student behavioral are related to parental involvement (Jeynes, 2011). Substance abuse and difficulties at school appeared to reflect the young person's psychological problems related to CPV, and affect them in aspects of their personality, such as aggressiveness and lower levels of tolerance of frustration (Ibabe et al., 2007). A similar phenomenon occurs if the YLS/CMI total risk is analyzed, as the CPV group presents a higher risk of recidivism than the comparison group. They also present almost double the risk score in the direct score for the test.
According to this score, higher percentages of CPV youths are classified in the moderate and high risk level than the comparison group. For example, 20.5% of the CPV group presented a high level of risk, compared to 6% among the comparison group. Other studies from Juveniles Courts who have used this Inventory also show low rates of high risk, between 0 and 9.10%. As stated before, the comparison group presented the highest percentage of classifications in the low and moderate risk, whereas the majority of CPV youths were classified at the moderate and high risk level. General youths in Court generally present a majoritarian low and moderate risk level and minoritarian high risk level (Cuervo & Villanueva, 2013;Garrido, 2009;Ortega et al., 2020;Takahashi et al., 2013). This percentages, together with the recidivism rates of CPV presented, show a higher-risk profile when compared to general youths.
The child to parent violence group presented significantly lower protective factors than the comparison group on the Peer relations, Substance abuse, Personality/  behavior and Attitudes/orientation subscales. The study of protective factors for youths who commit offenses in general is relatively recent, and the study of the protective factors for CPV in particular is even more novel, and they have been the subject of even less research. The key factors in reducing risk in these youths are those that stop the cycle of violence (Aroca et al., 2012;Pereira & Bertino, 2010). The areas of Personality/behavior and Antisocial attitudes would be therefore good candidates for this. Personal variables such as Self-control and the development of prosocial behaviors would be also useful (Beckmann et al., 2017). Some features of family relationships and bonding such as positive attachment figures, Family flexibility or adaptability and a Secure attachment style are also crucial (Llamazares et al., 2013). Indeed, parental warmth and supportive parent-child relations were important in stopping aggressive behavior towards parents (Beckmann et al., 2017;Ibabe & Bentler, 2016). These positive attachment figures may involve either people of the same age, people in these youths' peer group, or adult figures such as teachers or relatives who help the youth to focus on long-term goals and objectives (Contreras et al., 2011). In a study comparing clinical and judicial samples, Loinaz and De Sousa (2020) found that Family involvement was higher in clinical CPV cases. In the judicial sample, social support was found to be greater. This may be related to the deterioration of family relationships, leading to more frequent interaction or seeking help outside the family. The anticipated high risk subscales for the child to parent violence group in the YLS/CMI Inventory were Family circumstances/parenting, Education/employment, and Personality/behavior (Cuervo et al., 2017a, b, c;Garrido, 2010), while for the comparison group they were Leisure/ recreation, Education/employment and Substance abuse (Cuervo & Villanueva, 2015;Garrido, 2009;Graña et al., 2007). The predictive risk subscales related to CPV were Family circumstances/parenting, Antisocial peer relationships, Substance abuse and Personality/behavior. All the subscales predicted CPV with a very high Nagelkerke value. The Peer relationships subscale warrants particular attention, since it presented a negative relationship and therefore predicted the general commission of crimes. The Family circumstances subscale was the strongest predictor for CPV. In this case, this subscale is related to discipline and supervision by parents, and the relationships that the youths have with their parents (Cuervo, 2021). The level of cohesion and conflict in the family is related to a lack of inhibition as regards engaging in aggressive behavior towards the parents. In terms of communication and attachment models, the research highlights a weak parentchild bond, a lack of communication and a weak emotional connection between parents and adolescents (Agnew & Huguley, 1989;Duffy & Momirov, 1997;González et al., 2013, Paulson et al., 1990, which even become feelings of rejection towards the children in some cases (Aluja et al., 2005;Ibabe, 2015). Meanwhile, the study by Contreras and Cano (2016) highlights the factor of exposure to violence at home as the strongest predictor for CPV, and uses a methodology similar to the one used in this study, i.e., with one group that engages in CPV, and another that commits other types of offenses. These authors explain this relationship in terms of the hostile attributions that the youth adopts, and the lack of social problem-solving skills and biases in the attribution process.
Some authors argue that one of the risk factors that influences the onset and persistence of violent intrafamily reactions is substance abuse, such as cannabis use. This consumption has an influence, and increases verbal aggression in around 50% of cases. This seems logical, taking into account the general disinhibition that these substances create in situations involving confrontation. Shouting, insults, threats and violent behavior would increase in these cases, with an escalation of violence (Aroca et al., 2014;Pagani et al., 2004Pagani et al., , 2009. In this case, drug abuse is confirmed as a predictor, and not only a factor which increases or reinforces CPV, as reported by other authors (García de Galdeano & González, 2007;Pantoja, 2005;Price, 1996).
Another predictor of CPV is the Personality subscale. This factor in turn plays a crucial role in the difference with the comparison group in terms of protective factors. This subscale consists of items related to aggressiveness, poor frustration tolerance, the absence of guilty feelings and impulsiveness. These traits are obviously related to any type of aggression, and not only CPV. If we consider the classic distinction between proactive and reactive aggression, in which the former is an impulsive reaction with high levels of anger due to a perceived threat, and the latter is aggression motivated by achieving a goal, Pagani et al. (2009) argue that this type of violence is related to reactive aggression. This type of reaction occurs because these youths have only developed one way of dealing with boundaries, which they perceive as insulting. Meanwhile, in permissive patterns, proactive aggression is used in order to achieve some material benefit or obtain some reinforcement in behavior (Coogan, 2012). These behaviors are related to the entitlement that the youth feels when relating to their parents. They believe that they are entitled to exercise control over them and behave as they wish, which in some cases becomes tyranny (Howard et al., 2010). Cottrel and Monk (2004) affirm that drug use should be analyzed as a symptom of a deteriorated family dynamic. This negative use could also be due to the search for relief from family conflicts, as well as an attempt to obtain relaxing impulses. However, this is not always achieved, since irritability and loss of control would be discharged on the family in almost the same way (Ibabe, 2007). This abuse does not necessarily have to occur at the moment of the aggression, and therefore whether the relationship with CPV is direct or indirect remains to be determined (Cuervo, 2018).
Meanwhile, the increase in the Peer relationships risk subscale predicts the commission of other types of offenses. In fact, this factor is actually statistically non-significant when compared to the comparison group. The association with high risk youths is related to the commission of common crimes and recidivism in other general samples of youths (Bonta & Andrews, 2016;Cuevas et al., 2019;Cuervo & Villanueva, 2015). These youths therefore have more peers as positive models than those engaging in CPV, and at the same time would share leisure with antisocial peers, leading them to commit crimes. These youths therefore would present a greater heterogeneity and variability in their relationships. Meanwhile, the friends of the CPV group were not as positive, as is apparent in the analysis of the protective or negative factors shown in the predictive analyses. In more specific terms, youths who commit CPV tend to be more closely associated with youths who also present violent conduct at home (Agnew & Huguley, 1989).

Conclusion
As a general conclusion, a poor family environment combined with poor family relationships, inadequate and ineffective supervision and an aggressive personality, together with substance abuse, therefore predict the commission of a CPV offense. On the other hand, the Prior and current offenses criminal record, the Education/employment subscale and the Attitudes subscale do not predict this type of offense, since they do not differentiate between the two groups of youths. This study therefore not only examines the factors that specifically predict CPV, but also tests the factors that do not predict it. The results for risk factors could be used as objectives for intervention. In this case, Family circumstances/parenting, Substance abuse and Personality/behavior are the subscales on which intervention should be focused. There is no single risk factor that influences CPV. This type of violent behavior is instead caused by the interaction between various factors. Nevertheless, according to these results, the education of the parents is of vital importance. An early intervention would avoid the chronification of the conflict, and perhaps the involvement in the judicial system (Cuervo, 2018). There are several evidence-based parenting interventions which could be classified as skills training programs for parents. These interventions help to develop the necessary resources and train parents how to use them appropriately. Some of these programs are: Parent-Child Interaction Therapy (PCIT) (Fernandez & Eyberg, 2005); Incredible Years (IY) (Trillingsgaard et al., 2014); Parent Management Training (PMT) (Nock & Kazdin, 2005); Helping the Noncompliant Child (HNC) (Forehand et al., 2017) and TRIPLE P: Positive Parenting Program (Leung et al., 2003). All of them place emphasis on improving the quality of the parentchild relationship and changing parent-child interaction patterns. They are directed towards children and teenagers with different emotional and conduct difficulties. In particular, the PCIT is more specific for Oppositional Defiant Disorderand and Conduct Disorder (CD); IY for AHDH and CD; and the PMT for oppositional, aggressive, and antisocial behaviors. Moreover the long term goal of HNC and Triple P is to prevent emotional and conduct problems and teaching parents how to obtain compliance from the children, whereas the IY goes further and intends to prevent delinquency, drug abuse, violence and emphazises in improving academic success. All of them take place in an external clinic or organization with the help of trainers, while some activities are done by the parents at home without the professionals.
Second, an intervention focusing on reducing that risk might therefore reduce criminal trajectories, as well as future reoffending. In particular, according to Jeynes (2020), family factors, taking illegal drugs and mental illness are the leading cause of school shootings and violence. In overall terms, this is a particularly high risk profile, with very few protective factors and almost twice the rates of recidivism of general youths, which according to statistical data is increasing every year (Abadías, 2017; Spanish Attorney General's Office [Fiscalía General del Estado], 2019).
All the findings above may have practical implications for professionals who work with youths, as identifying the criminogenic differences and providing specific protective factors in order to reduce crime among youths who commit child to parent violence and youths who commit other types of crimes. These conclusions are essential for designing specific intervention programs. The explanation for this growing phenomenon involves major sociocultural changes that affect how family relationships are understood due to the postmodern society. Determining the family structure and history will be determinant for planning therapy. One of the therapeutic models with positive and confirmed scientific results is the systemic model which differentiates between functional and dysfunctional human relationships (Abadías & Pereira, 2020).
Finally, this study has some limitations. First, the data came from one specific Spanish province, and as such the results should not be generalized to other Spanish provinces, other countries or to the population of youths in general. Second, some YLS/CMI Inventory scores were lacking, as some of the youths studied were subject to precautionary measures and they were not usually assessed by the Psychosocial Team. Finally, an important issue within this specific sample is the parents' difficulties in reporting the assaults due to feelings of shame and guilt, as discussed in the introduction. This may mean the extent of the problem is underestimated.
Further research is needed for detailed analysis of this complex phenomenon, in which violence is sometimes mutual or defensive, and cases should therefore be treated differently based on those considerations (Ibabe & Jaureguizar, 2010). In future research, it would be interesting to analyze the factor of family circumstances, which is the strongest predictor, in order to determine whether this influence on prediction is due to attachment factors, domestic violence or the type of relationship that youths have with their parents, and to examine the presence of violence in the family environment, which is reported to be a predicting factor in the literature on the subject (Boxer et al., 2009: Contreras & Cano, 2016Langhinrichsen-Rohling & Neidig, 1995;Mahoney & Donnelly, 2000). Prevention and intervention policies must take this into account, as well as the fact that parents wait until the last possible moment to file a report. At present, there are apparently no other mechanisms or protocols to prevent this type of aggression ; Spanish Attorney General's Office [Fiscalía General del Estado], 2019).
The data that support the findings of this study, which were used under license, are available subject to restriction. Since the participants were minors, the data are available with permission from the Juvenile Court of Castellón, Spain.

Compliance with Ethical Standards
Conflict of interestḍ The authors declare no competing interests.
Ethical Approval This study did not include human or animal participants. Data were obtained from a database from the Juvenile Court.
Informed Consent Consent to participate: Verbal informed consent was obtained prior to the assessment interviews. Interviews were conducted by members of the Juvenile Court, not by the authors.
Publisher's note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons. org/licenses/by/4.0/.