In the skill model (Mayer et al., 2001; Mayer et al., 2004; Mayer et al., 2016; Mayer & Salovey, 1997; Salovey & Mayer, 1990), “emotional intelligence” (EI) describes four domains that identify emotionally intelligent people. They can accurately perceive emotions, use them to facilitate thinking, understand emotional meanings, and regulate feelings in themselves and others (Mayer et al., 2016; Mayer & Salovey, 1997). EI facilitates emotional assessment and prognosis that assists in making judgments and anticipating emotional and behavioral responses (Barrett et al., 2011; Hess et al., 2016; Zeidner et al., 2009).

Meanwhile, “empathy” has been defined as an emotional and cognitive state associated with understanding and the ability to feel other people’s emotion (Dohrenwend, 2018). Empathy as a motivational process is determined by physiological and automatic mechanisms, makes use of information analysis, and integrates context determinants that affect the degree of empathic involvement (Longobardi et al., 2020; Zaki, 2014). It has been observed that in circumstances that would expose a person to suffering, material costs, or competitive losses, empathy decreases and tends to increase when there are positive affect, affiliation, and social convenience (Szuster & Jarymowicz, 2020; Zaki, 2014). Empathy is related to emotional regulation and other emotional skills because they include the skills of emotional perception, imitation, and self-reference that cause spontaneous resonance with another person’s emotions. It is also related to cognitive processes that facilitate the inference of another person’s emotional state and the ability to differentiate between their emotions and another person’s (Thompson et al., 2019).

Empathy is a psychological construct closely related to EI because understanding and regulating emotions are required to achieve empathy (Estévez et al., 2019; Sa et al., 2019). A relationship between empathy and EI has been observed in children and adolescents (Corbí et al., 2008; Gorostiaga et al., 2014). Empathy favors helpful behavior, altruism, and cooperation (Glen et al., 2020; Streit et al., 2020), and those behaviors necessarily involve decision-making (Marshall et al., 2020).

EI has been observed as negatively correlating with antisocial behaviors (Mavroveli & Sánchez-Ruiz, 2011), and fewer EI skills have been identified in adolescent offenders than in their normative peers (Contreras & Cano, 2016; Fariña et al., 2008; Hayes & Reilly, 2013; Kahn et al., 2016). In adults, it has been observed that when emotional regulation problems increase, more psychopathic traits are manifested in the affective, lifestyle, and antisocial domains (Garofalo et al., 2020). Socialization problems and trauma have also been identified as affecting EI in male offenders (Wang et al., 2019).

Empathy has been shown to be associated with fewer antisocial behaviors in children and adolescents (Estévez et al., 2019; Garaigordobil & García de Galdeano, 2006; Thompson & Gullone, 2008). It can also help adolescents inhibit antisocial behavior and maintain better interpersonal relationships (Olthof, 2012). Evidence indicates that teen offenders show less empathy than their peers in the community (Lardén et al., 2006; Robinson et al., 2007). Empathy has also been observed to be inversely related to crime, and low empathy may be a predictor of criminal behavior (O’Neill, 2020).

Evidence suggests that emotional responses influence decision-making more than cognitive evaluations (Bechara & Damasio, 2005; Damasio, 1994; Naqvi et al., 2006). Decision-making requires EI and empathy due to the process of self-awareness and the understanding of one’s own emotions that enables one to understand others (Brabec et al., 2012), and is affected by emotion, the pursuit of profit, problem-solving, relationship management, emotions, and life orientation (Ekel et al., 2020; Farnia et al., 2018).

Decision-making models propose that rational decision-making processes are more appropriate than intuitive ones and that non-systematic and irrational decision-making actions lead to personal losses and risky behaviors (Ceschi et al., 2019). In a previous study, Avsec (2012) analyzed whether EI was a positive predictor of decision-making styles, and the results showed that higher EI is associated with more frequent use of rational and intuitive decision styles, and lower EI is related to avoidant, dependent, and spontaneous styles. Rationality, intuition, and emotional life are also considered to play a determining and positive role in decision-making (Reyna, 2018). Analyzing the role of EI in decision-making can help to understand the nature of judgments and decisions (Farnia et al., 2018). EI has been observed to facilitate intuitive decision styles and shows a negative mediation effect on dependent and avoidant decision-making styles (El Othman et al., 2020).

Studies with somatic markers (Bechara & Damasio, 2005) have reported that the brain areas related to empathy are more active when making moral decisions (Reniers et al., 2012) and that low EI is associated with less physiological control and higher risk behavior (Yip et al., 2020). In adolescent offenders, a higher level of risk decisions and less vigilance and control in decisions have been observed than in their non-offending peers (Poon, 2020).

This study considers that having EI and empathy positively affects individual well-being, interpersonal relationships, and social adjustment (Nguyen et al., 2019; Salavera et al., 2019), and the cognitive and emotional variables of the dimensions that make up these constructs could influence more adaptive decision-making styles (Altman, 2017). In addition to helping the individual, they could also favor their context and society (Reyna, 2018). As background, EI in adolescents has been associated with better interpersonal relationships, higher levels of subjective well-being, better school performance, and mental health (MacCann et al., 2020; Nyarko et al., 2020). Meanwhile, empathy has been related to better interpersonal skills and greater prosocial behavior (Rodriguez et al., 2021; Taylor et al., 2020).

Outside of the legal field, the relationship and benefits of empathy and EI on decision making have been the subject of study in business and organizational contexts. From Goleman’s emotional intelligence model, it has been identified that EI and empathy contribute to ethical decision-making and leadership (Hess & Bacigalupo, 2013; Issah, 2018). It has also been identified that EI can be useful for decision-making in professions that require specific skills, for example, in negotiators and police officers (Grubb et al., 2018). It has been observed in the educational domain that EI favors decision-making when adolescents and young adults choose professions (Santos et al., 2018).

Decision-making is particularly important in adolescents due to its role in making judgments, predicting behavior, and planning goals (Fischhoff & Broomell, 2020). Risk decisions have been found to have foreseeable negative effects (Broniatowski & Reyna, 2018), and are more frequent in adolescent offenders (Poon, 2020). EI and empathy have also been observed to be related factors and for this research they have been considered as contributing to more adaptive decision-making styles. Studies that address EI and empathy as predictors of decision-making styles in adolescents have not been observed in the literature, so these constructs will be analyzed in offenders and non-offenders.

Hypotheses of the study: (1) adolescent offenders will present poorer performance in EI, empathy, and adaptive decision-making styles; (2) the variables of EI, empathy, and decision-making styles will be related in both populations; and (3) EI and empathy will be predictors of decision-making styles in adolescent offenders and non-offenders.

Concerning the first hypothesis, previous studies have indicated that adolescent offenders show less empathy than their normative peers (Lardén et al., 2006; Robinson et al., 2007) and fewer EI skills (Contreras & Cano, 2016; Kahn et al., 2016). Evidence indicates that adolescents with higher analytical decision-making skills show less antisocial behaviors (Crean, 2012). As for the second hypothesis, evidence indicates a positive relationship between EI and empathy (Corbí et al., 2008; Gorostiaga et al., 2014), and it has been indicated that EI could favor decision-making (Farnia et al., 2018). The third hypothesis is justified by previous studies that have suggested that EI and the emotional and cognitive components of empathy might influence decision making (Altman, 2017; Avsec, 2012).

This study could be useful for the prevention and intervention in socioemotional competencies in antisocial adolescents. Some explanatory and intervention models of criminal behavior have included socioemotional variables, such as the Risk-Need-Responsibility - RNR model (Andrews & Bonta, 2017). This model has identified that the development of self-control and problem-solving skills helps to control antisocial personality patterns. A positive effect is achieved through intervention in reducing dynamic criminological needs by “increasing self-control skills, fostering empathy, improving problem-solving skills, and controlling negative emotions” (p. 53). Addressing cognitive-emotional states associated with delinquency, such as anger and feelings of irritation or resentment, intervenes in criminal attitudes, a predictor of persistence. Intervention in the RNR model contemplates “developing problem-solving skills, self-control skills, anger management, and coping skills” (p. 288).

Eight hundred eight (808) adolescents from four cities in Colombia participated in a sample obtained by availability and convenience. Fifty percent (n = 404) were from Sistema de Responsabilidad Penal para Adolescentes (SRPA), aging between 14 and 18 years old (M = 16.6, SD = 1.04), 17.3% being girls (n = 70). They had been prosecuted for theft (33.2%, n = 134), aggravated robbery (16.6%, n = 67), drug manufacture and trafficking (20.5%, n = 83), physical injuries (5.6%, n = 23), attempted homicide (2.7%, n = 11) and homicide (3.2%, n = 13), sexual crimes (4.4%, n = 18), carrying weapons and ammunition (4.5%, n = 18), participation in organized crime (3.7%, n = 15), extortion (2.7%, n = 11), or other crimes (2.9%, n = 12). The other group of participants, 50% (n = 404), were regular school students, between 14 and 18 years old (M = 15.5, SD = 1.29), of whom 47.8% were girls (n = 193). The sample inclusion criteria were being between 14 and 18 years old and not having been diagnosed with serious mental health problems. Twenty surveys from the sample of adolescent offenders and 16 surveys from the sample of adolescents in school were not included in the final sample due to errors in completing the questionnaires. The sample for the study was estimated with a confidence level (1 - α) of 99% and accuracy (d) 3%. According to the analyses, the required sample for non-offenders was 350 participants. For offenders, 346 were required based on a population of 30.000 adolescents (the average number of individuals prosecuted per year). This indicates that the results obtained are representative of the population.

The Trait Meta-Mood Scale (TMMS-24) was used to assess emotional intelligence (Fernández-Berrocal et al., 2004). It consists of 24 items, with the following response options: strongly disagree (score 1), disagree (2), neither agree, nor desagree (3), agree (4), and strongly agree (5). Each subscale of the emotional intelligence measure is made up of 8 items that assess attention, clarity, and emotional repair. TMMS-24 reliability in the sample of offenders and non-offenders was .93 and .90, respectively. The reliability of each factor in each sample is presented in Table 1.

Empathy was evaluated with the Interpersonal Reactivity Index-IRI (Davis, 1980; Mestre-Escrivá et al., 2004). This scale evaluates cognitive and emotional factors of empathy and consists of 28 items distributed in 4 subscales: perspective taking, fantasy, empathic concern, and personal distress, with seven items each. The Likert-type format of the scale has five response options: does not describe me very well (1), does not describe me well (2) describes me more or less (3), describes me well (4), and describes me very well (5). The reliability of the IRI in the sample of offenders and non-offenders was .72 and .88, respectively.

Decision-making styles were assessed using the Melbourne Decision-making Questionnaire-MDMQ (Mann et al., 1997). This questionnaire is made up of 22 items, which have three response options: very true for me (score 2), somewhat true for me (score 1), and not at all true for me (score 0). It has four subscales that evaluate vigilance (6 items), hypervigilance (5 items), buck-passing (6 items), and procrastination (5 items). In decision-making styles, vigilance involves a careful, unbiased, thorough, and rational evaluation of alternatives; a hurried and anxious approach characterizes hypervigilance; procrastination refers to a delay in decision making; and buck-passing is a style that involves leaving decisions to others and avoiding responsibilities (Cotrena et al., 2018). The reliability of the MDMQ in the sample of offenders and non-offenders was .79 and .80, respectively.

The procedures used for the data collection were carried out following the guidelines of the Declaration of Helsinki (World Medical Association, 2013). This research was approved by the Ethics Committee of the University of Valencia and authorized by the Instituto Colombiano de Bienestar Familiar-ICBF (SRPA).

Written informed consent and permissions were requested from the directors of schools and centers, parents, and legal guardians (including judges). Adolescents were informed about the research and participated voluntarily, anonymously, and free of charge. The evaluation was conducted in classrooms in groups of 3 to 5 adolescents during school hours. Researchers and principals of each grade were present. The duration of the application was approximately 40 minutes, and surveys in paper format were used.

All statistical procedures were performed using the SPSS-25 program. First, reliability and descriptive analysis were performed for each of the factors evaluated in the study. Student’s t-tests were then used to determine the existence of statistically significant differences in the mean between groups. The effect size of differences was also estimated (Cohen, 1988).

Second, bivariate Pearson correlations were performed between the variables of interest. Fisher’s Z tests were used to examine whether the strengths of the correlations differed significantly between the groups. Finally, hierarchical linear regressions were performed to predict the decision-making styles in each of the samples, setting age, and gender as control variables; the dimensions of EI and empathy were the independent variables (Stockemer, 2019).

The descriptive results indicate that the group of adolescent non-offenders show greater EI, empathy, and rationality in decision-making. When comparing the two groups (Table 1), the variables vigilance, fantasy, empathic concern, and perspective taking present statistically significant differences and greater effect size. In EI, only emotional attention differences were observed between the two groups.

All decision-making styles were positively and significantly related to emotional attention, perspective taking, fantasy, empathic concern, and personal distress in offending participants. Only vigilance, which is the most adaptive style of decision-making, was positively associated with all EI variables. No relationship between clarity and emotional repair with less adaptive decision-making styles was observed. Interestingly, vigilance was the variable most closely related to the variables of EI and empathy. Personal distress was closely related to less adaptive decision-making styles.

In non-offending participants, vigilance was positively related to EI variables and negatively to personal distress. Hypervigilance was negatively related to clarity and emotional repair and positively to fantasy and personal distress. However, there is no relationship between hypervigilance and emotional attention, and perspective taking. Meanwhile, buck-passing was positively related to emotional attention and fantasy and negatively to clarity and emotional repair. Finally, procrastination was negatively related to clarity and emotional repair and positively to fantasy and personal distress (Table 2). Fisher’s Z-tests were performed to check for significant differences in the results of the correlations of adolescent offenders and non-offenders.

The results showed that some correlations between variables occurring in the sample of adolescent offenders are not present in adolescent non-offenders, such as the relationship between emotional clarity and personal distress and between reparation and empathic concern.

Statistically, significant differences are observed in the correlations of the EI and empathy variables when they occur simultaneously in offender and non-offender groups. No differences are observed in the relationship between emotional attention and buck-passing (Fisher’s Z = 0.015, p = .49). In fantasy with hypervigilance (Z = -0.86, p = .19), and buck-passing (Z = -1.19, p = .11). There was also no difference in the relationship between personal distress with hypervigilance (Z = -0.88, p =.19), buck-passing (Z = -1.25, p = .10), and procrastination (Z = -0.46, p = .32). Similarly, in hypervigilance with buck-passing (Z = -1.17, p = .12), and procrastination (Z = -0.64, p = .26). The most significant differences are between personal distress with the variables emotional clarity (Z = -9.37, p = .001), emotional repair (Z = -9.43, p = .01), perspective taking (Z = -9.37, p =.001), and empathic concern (Z = -11.3, p = .001).

All regression models performed to analyze the influence of emotional intelligence and empathy on decision-making styles were significant. Models were specified in three steps: sociodemographic variables gender (boys 0, girls 1) and age were introduced in the first step, IE variables in the second step, and finally, the empathy variables in the third step.

The regression to predict vigilance was significant the offenders’ sample, F(9, 393) = 15.286, p = .001, and in the non-offenders’ sample, F(9, 394) = 9.498, p = .001. The regression model explained 18% of the variance of the dependent variable in adolescent non-offenders, and 26% in the sample of offenders (Table 3). Emotional repair and perspective taking are variables that contribute to the prediction of vigilance in both samples. Gender (girls, β = -.17, t = - 3.47, p = .001) and less personal distress (β = -.16, t = -3.09, p = .002) were significant in adolescent non-offenders.

The regression to predict hypervigilance was significant in the sample of offenders, F(9, 394) = 10.484, p =.001, and in the non-offenders, F(9, 394) = 11.928, p = .001. The regression model explained 21% of the variance of the dependent variable in the sample of non-offenders, and 19% in the sample of offenders (Table 4). Emotional attention and personal distress are variables that contribute to the prediction of hypervigilance in both samples. Less repair (β = -.21, t = -3.88, p = < .001), less clarity (β = -.17, t = -3.15, p = .002) and fantasy (β = .14, t = -. 2.96, p = .003) were also significant in adolescent non-offenders.

The regression to predict buck-passing was significant in the sample of offenders, F(9, 394) = 9.228, p = .001, and in the non-offenders, F(9, 394) = 19.178, p = .001. The regression model explained 30% of the variance of the dependent variable in the sample of adolescent non-offenders and 17% in the sample of offenders (Table 5). Personal distress, attention, and less clarity are variables that contribute to the prediction of buck-passing in both groups.

The regression to predict procrastination was significant in the sample of offenders, F(9, 394) = 11.300, p = .001, and in the non-offenders, F(9, 398) = 9.971, p = .001. The regression model explained 19% of the variance of the dependent variable in the sample of adolescent non-offenders and 20% in the sample of offenders (Table 6). In both groups, attention, less clarity and emotional repair, and personal distress are variables that contribute to the prediction of procrastination in both groups. In offenders, fantasy (β = .19, t = 3.19, p = .001), and in non-offenders, less empathic concern (β = -.11, t = -2.88, p = .003) were significant.