Behavior problems and personality in Korean high school students

Introduction

Adolescence is a developmental period characterized by rapid growth of both body and mind, and establishment of psychological identity which may determine lifespan mental health through the process of interaction with society (Klimstra et al., 2010). And adolescents are at particular risk of developing behavior problems resulting from daily life stress as they adjust themselves as members of society. The behavior problems of Korean adolescents experiencing academic stress and rapid sociocultural changes have been a major issue (Oh & Lee, 1990), and various studies attempting to delineate this phenomena were conducted (Kim, Lee & Chae, 2017; Lee et al., 2017b; Lee et al., 2013).

Behavior problems demonstrate typical clinical characteristics depending on gender, age and developmental stage, and such complex clinical spectrums need to be measured from a wide-band perspective rather than focusing on specific problems (Oh & Lee, 1990). The Achenbach System of Empirically Based Assessment (ASEBA) (Achenbach, 2009; Achenbach & Rescorla, 2001), which was developed by Achenbach and Edelbrock (Achenbach & Edelbrock, 1978), provides clinical assessment of behavior problems in diverse age groups using a dimensional approach to understand the cause and maintenance of behavior problems (Achenbach, 2009).

The ASEBA categorizes behavior problems into three groups, namely Internalizing Problems, Externalizing Problems and Total Problems (Achenbach, 2009). The Internalizing Problems of Anxious/Depressed, Withdrawn/Depressed and Somatic Complaints are deemed to result from excessive internal control of emotional problems, and are reported to occur more frequently in females (Nolen-Hoeksema & Girgus, 1994; Rutter, Caspi & Moffitt, 2003). The Externalizing Problems refer to Rule-Breaking and Aggressive Behavior sought to represent excessive expression of psychological difficulties, and has been shown to be more prevalent in males (Archer, 2004; Else-Quest et al., 2006). The Total Problems is the sum of Internalizing and Externalizing Problems along with Social Problems, Thought Problems, Attention Problems and Other Problems.

Several previous studies showed that specific psychological traits may be useful in effective screening and preventive intervention of particular behavior problems (Caprara et al., 2017; Cooper et al., 2003; De Pauw, Mervielde & Van Leeuwen, 2009; Hoyle, 2000). As for vulnerability and resilience, the Five Factor Model (Costa & MacCrae, 1992) suggested that Internalizing Problems are associated with low extraversion and high neuroticism, and that Externalizing Problems are related to high extraversion, low agreeableness, and low conscientiousness (Slobodskaya & Akhmetova, 2010).

However, since the Five Factor Model is based on factor analysis of psychological characteristics of an individual (Costa & MacCrae, 1992), it holds limitations in that only mixture of expressed traits are measured without distinguishing relatively hereditary and stable temperament traits from acquired and changeable character traits across the lifespan. We need a biopsychosocial measure of the temperament facets along with character aspects which may be influenced by the environment and learning (Lee et al., 2014) because it may provide the theoretical grounds for intervention aimed at prevention and treatment of behavior problems (Kim, Lee & Chae, 2017; Lee et al., 2017b; Lee et al., 2013).

The Temperament and Character Inventory (TCI), a personality assessment tool based on Cloninger’s biopsychosocial personality model (Cloninger, 1987; Cloninger, Svrakic & Przybeck, 1993), consists of two domains measuring temperament and character, respectively, and explains the foundation and development of personality through their interaction.

Temperament, encompassing Novelty Seeking (NS), Harm Avoidance (HA), Reward Dependence (RD), and Persistence (PS), is an innate and biological foundation for character development based on the behavioral activation and inhibition system proposed by Gray (Gray, 1982) and the behavior maintenance system suggested by Sjobring (Sjobring, 1973). Character, comprised of Self-Directedness (SD), Cooperativeness (CO), and Self-Transcendence (ST), represents the value of the person regarding him/herself, society and universe, and is developed through interactions between the temperament and the environment (Cloninger, 1987; Cloninger, Svrakic & Przybeck, 1993).

Previous studies have reported correlations between personality traits and a wide spectrum of behavior problems using the ASEBA and Junior version of the TCI (JTCI). The HA temperament was shown to have a strong positive correlation with Internalizing Problems such as depression and anxiety (Copeland et al., 2004; Hemphälä, Gustavsson & Tengström, 2013; Kim, 2013; Kim et al., 2006; Lim, Choi & Choi, 2013; Rettew et al., 2006), while NS and RD temperaments (Cloninger, Svrakic & Przybeck, 1993; Kim et al., 2006; Lim, Choi & Choi, 2013) and SD character (Copeland et al., 2004; Kim, 2013; Kim et al., 2006) showed negative correlations with depression and anxiety. The NS temperament alone (Copeland et al., 2004; Ha & Kwon, 2016; Hemphälä, Gustavsson & Tengström, 2013; Kim & Lee, 2016; Kim et al., 2006; Rettew et al., 2006) or in association with HA (Kim et al., 2006) demonstrated a positive correlation with Externalizing Problems such as violence and rule-breaking, and the SD and CO characters (Copeland et al., 2004; Kim et al., 2006) showed a negative correlation.

However, the human personality is a complex adaptive system comprised of multiple dimensions interacting non-linearly, rather than a collection of individual components (Cloninger, 2008). Clinicians and researchers need a profile analysis providing person-centered perspectives appropriate for detecting such non-linear and complex dynamics of personality (Cloninger & Cloninger, 2011; Cloninger & Zohar, 2011), not only focusing on the average effects of separate traits in regard to vulnerability and resilience across groups of individuals (Cloninger et al., 2012b). Since healthy, resilient, and matured personality profile of adults was characterized by low HA and high SD and CO in previous studies (Cloninger & Cloninger, 2011; Cloninger & Zohar, 2011; Eley et al., 2013; Eley et al., 2016), this possibility must also be examined and confirmed in adolescents (Moreira et al., 2014).

For such reasons, we recruited a community sample of Korean adolescents, and analyzed the effects of temperament and character as both individual factors and complex profiles on behavior problems using the JTCI and Youth Self Report (YSR) (Oh & Kim, 2011; Oh & Lee, 1990). We used stepwise multiple regression to examine whether temperament and character traits explain behavior problem patterns, and also performed latent profile analysis (Eley et al., 2016; Lee, Choi & Chae, 2017a) to uncover personality profiles related to specific severity of behavior problems with consideration of gender (Moreira et al., 2014; Oh & Lee, 1990; Pursell et al., 2008) which has not been reported previously.

Adolescents are able to report their behavior problems, especially internalizing problems more reliably than primary caregivers and the development of their personality expands in this developmental period to play a pivotal role as a protective factor of psychological health in adulthood. Korean high school students who have relatively developed personalities and who can complete validated self-report measures (Sourander, Helstelä & Helenius, 1999; Verhulst & Ende, 1992) would be fit for this study as such findings may be generalizable to preventive interventions in the future (Roberts, Caspi & Moffitt, 2001).

This study may contribute to person-centered research and theory related to the correlation between psychopathology and complex personality characteristics in line with previous studies of vulnerable and resilient personality factors (Eley et al., 2016; Lee, Choi & Chae, 2017a; Moreira et al., 2014) which showed the importance of nurturing matured personality for coping with daily life stress.

Methods and Materials

Results

Latent profiles analysis with nine YSR subscale scores

Latent profile analysis identified three latent behavior problem groups as the best solution (Table 6) after controlling the effect of gender, and YSR subscale scores of three (high, middle, and low) latent groups were presented in Fig. 1. The three-latent group model (3-class solution) fit the current data significantly better than a 2-class solution, and was not worse than a 4-class solution considering the smaller BIC and adjusted BIC, the smaller p value of VLMR and LMR and the greater entropy. All nine YSR subscale scores of the high, middle, and low behavior problem groups in current study exhibited a decreasing order as shown in Fig. 1. The number of participants of the high, middle, and low groups in regard to behavior problems were 70, 235, and 365, respectively.

Table 6:

Results of Latent Profile Analysis with gender as covariate.

	BIC	adj.BIC	VLMRp	LMRp	BLRTp	Entropy
2 class	29616.64	29524.563	<0.0001	<0.0001	<0.0001	0.9300
3 class	28979.392	28855.564	0.0446	0.0460	<0.0001	0.8860
4 class	28741.791	28586.212	0.3298	0.3336	<0.0001	0.8920
5 class	28636.385	28449.056	0.3245	0.3273	<0.0001	0.8880

DOI: 10.7717/peerj.6106/table-6

Notes:

BIC

Bayesian Information Criteria

adj.BIC

Sample-Size Adjusted BIC

VLMRp

Vuong-Lo-Mendell-Rubin Likelihood Ratio test p-value

LMRp

Lo-Mendell-Rubin Adjusted LRT p-value

BLRTp

Parametric Bootstrapped Likelihood Ratio p-value

Differences of YSR and JTCI subscales among three latent behavior problem groups using profile analysis and ANCOVA

The significant differences in the YSR subscale profile among the three latent behavior problem groups were analyzed with profile analysis (Fig. 1), and there were significant differences between the three behavior problem groups. The YSR subscales of the three behavior problem groups were not flat (Greenhouse-Geisser correction, df = 6.078, F = 45.931, p < 0.001), and the interaction of three groups was significantly different in regard to parallelism (Greenhouse-Geisser correction, df = 12.156, F = 30.149, p < 0.001). There were significant differences in nine YSR subscales of Anxious/Depressed (F = 533.559, p < 0.001), Withdrawn/Depressed (F = 283.200, p < 0.001), Somatic Complaints (F = 182.751, p < 0.001), Social Problems (F = 577.616, p < 0.001), Thought Problems (F = 523.450, p < 0.001), Attention Problems (F = 259.036, p < 0.001), Rule-Breaking Behavior (F = 184.601, p < 0.001), Aggressive Behavior (F = 307.468, p < 0.001) and Other Problems (F = 186.481, p < 0.001) based on ANCOVA results. Nine YSR subscale scores of the three behavior problem groups were significantly different from each other as shown in Fig. 1.

The significant differences on the JTCI subscale profile among the three behavior problem groups were analyzed with profile analysis (Fig. 2), and there were significant differences between three latent groups. The JTCI personality profile of the three behavior problem groups was not flat (Greenhouse-Geisser correction, df = 3.949, F = 71.200, p < 0.001), and the interaction of three groups was significantly different in regard to parallelism (Greenhouse-Geisser correction, df = 7.897, F = 52.974, p = 0.001). There were significant differences in the six JTCI subscales of NS (F = 29.334, p < 0.001), HA (F = 73.850, p < 0.001), RD (F = 17.135, p < 0.001), PS (F = 43.357, p < 0.001), SD (F = 89.575, p < 0.001) and CO (F = 28.356, p < 0.001), but not ST (F = 1.048, p = 0.351). High, middle, and low latent groups of NS, HA were in decreasing order, while those of RD, PS, SD and CO were in increasing order with significant differences as shown in Fig. 2.

Discussion

This study recruited 670 Korean high school students, and measured behavior problems and personality characteristics with the YSR and JTCI to investigate which personality traits and profiles explain behavior problems in adolescents.

Demographic characteristics showed significant differences in the YSR and JTCI subscale scores between male and female high school students (Table 1). As for behavioral problems, female students showed higher scores on Internalizing Problems including Anxious/Depressed and Withdrawn/Depressed, and male students evidenced higher scores in Externalizing Problems such as Rule-Breaking Behavior (Table 1) which is consistent with results of previous study (Oh & Lee, 1990). Gender was found to account for the type of behavior problem (Table 5) such that the male gender explained Externalizing Problems and the female gender delineated Internalizing Problems. As for personality traits, females exhibited higher RD, CO and ST scores compared to male students, which are consistent with previous studies (Lee, Choi & Chae, 2017a; Lee et al., 2014; Zohar et al., 2018).

The effects of temperament and character traits on YSR Total, Internalizing and Externalizing Problems were analyzed with stepwise regression analysis (Table 5). As for the JTCI temperament domain with a biological basis, NS and HA were risk factors and RD was a protective factor for YSR Total and Internalizing Problems, and NS was a risk factor for YSR Externalizing Problems. These results are in accordance with previous studies which showed that high HA and low RD is associated with Total and Internalizing Problems (Copeland et al., 2004; Hemphälä, Gustavsson & Tengström, 2013; Kim, 2013; Kim et al., 2006; Lim, Choi & Choi, 2013; Rettew et al., 2006) and high NS is related to Externalizing Problems (Copeland et al., 2004; Ha & Kwon, 2016; Hemphälä, Gustavsson & Tengström, 2013; Kim & Lee, 2016; Kim et al., 2006; Rettew et al., 2006).

Although NS significantly explained Internalizing Problems unlike previous studies, considering that NS and Internalizing Problems showed a negative correlation after controlling for Externalizing Problems using partial correlation analysis, this result may have stemmed from comorbid Internalizing and Externalizing Problems. It was reported that high NS shows a relation with development of delinquent behavior and antisocial personality (Schmeck & Poustka, 2001), and NS was the best predictor of highly delinquent behavior (Tremblay et al., 1994). In other words, such results are similar to a study examining social anxiety disorder in an outpatient sample (Kashdan & Hofmann, 2008) indicating the need for a more careful interpretation in regard to comorbidity of adolescent behavior problems.

As for the JTCI character domain, high SD may be a protective factor for YSR Total, Internalizing and Externalizing Problems along and high CO for YSR Externalizing Problems in adolescents. This result is consistent with previous studies (Cloninger, Svrakic & Przybeck, 1993; Copeland et al., 2004; Hemphälä, Gustavsson & Tengström, 2013; Kim, 2013; Kim et al., 2006; Lim, Choi & Choi, 2013; Rettew et al., 2006) which reported that high SD and high CO explains resilience to behavior problems. The JTCI RD and CO might be crucial factors in determining the expression and progression of behavior problems considering its theoretical correlation with agreeableness and protective effects of agreeableness in Externalizing Problems (Laursen & Richmond, 2013; Slobodskaya & Akhmetova, 2010). In summarizing the relationship between behavior problems and personality traits, we can conclude that high HA, high NS and low RD of temperament domain and low SD and low CO of character domain may be a particular vulnerability factor for adolescent behavior problems.

Behavior problems in real life evolve through complex interactions between diverse factors, rather than through several problematic symptoms individually influenced by one or two personality traits. Therefore, the personality characteristics of an individual must be analyzed as a multi-dimensional and complex system using person-centered Latent Profile Analysis (LPA). The diagnosis, treatment, and prevention of behavior problems should be practiced with this whole-person perspective in mind as suggested previously (Eley et al., 2016; Lee, Choi & Chae, 2017a).

We performed LPA analysis on nine subscales of the YSR to extract distinctive latent profile groups (Table 6), and found three latent groups consisting of high, middle, and low behavior problem groups. These three behavior problem groups were distinctively and significantly different from each other in regard to YSR behavior problem profiles and also the nine YSR subscale scores (Fig. 1). There also were significant differences among the three behavior problem groups in regard to JTCI personality profiles and also the JTCI subscales scores with the exception of ST (Fig. 2).

These results provided information regarding the temperament and character profiles of community adolescents which may predict matured personality, positive affect, psychological well-being, and life satisfaction in adulthood (Garcia, 2011; Garcia et al., 2013b; Garcia & Moradi, 2012; Moreira et al., 2014). In current study, the high behavior problem group showing a vulnerable personality profile was characterized by high HA and NA and low RD, PS, SD and CO. However, the low behavior problem group showing a resilient personality profile was characterized by low HA and NA and high RD, PS, SD and CO which is in line with the healthy and resilient personality profile reported in previous studies (Eley et al., 2016; Lee, Choi & Chae, 2017a; Moreira et al., 2014) as Dependable temperament and Matured character profile (Zohar et al., 2018). The temperament traits of high NS and/or high HA in combination with character traits of low SD were repeatedly demonstrated as a major risk factor for many psychopathological conditions (Fassino et al., 2004; Lee, Choi & Chae, 2017a; Lilenfeld, 2010).

The temperament complex of high NS, high HA and low RD (NHr) represented the high behavior problem group in the current study, and the most vulnerable personality profile related to psychological immaturity (Basoglu et al., 2011; Cloninger, 1987; Lee et al., 2014) was characterized by an Explosive temperament profile which possesses dynamic psychological energy secondary to a simultaneous co-existing mixture of approach (high NS) and avoidance (high HA) behaviors (Cloninger, 1987; Lee et al., 2014). Individuals with an Explosive temperament profile tend to be bold and daring when others hesitate and/or behave according to the particular situation. There is also a high probability of conflicts in interpersonal relationships since they may express their impulsivity and emotions without filtering, and may lack empathic capability secondary to their temperament trait of low RD.

On the contrary, an individual with a low NS, low HA and high RD (nhR) representing the low behavior problem group in this study manifests a Reliable temperament profile which is potentially the least vulnerable to psychological immaturity (Cloninger, 1987; Lee et al., 2014). Individuals with Reliable and Dependable (low NS, low HA, high RD and high PS) temperament profile are calm and careful, and are generally optimistic, pleasant, and conscientious. They are compassionate and express soft and warm emotions, while not being easily angered, annoyed nor excited.

In these ways, when analyzing psychological characteristics from the perspective of interacting complexes, rather than as a collection of individual factors, we may find that the immature coping style of the Explosive temperament profile (Cloninger, 1987; Lee et al., 2014) could trigger maladaptive response styles to daily life stress, and potentially exacerbate behavior problems, and the Matured character profile of high SD and CO provides cognitive self-regulation of their behavior and adaptive responses to stress (Moreira et al., 2014; Zohar et al., 2018).

Adolescence is an important period for the development of autonomous identity, social relationships, and values and goals shaping their lifestyle (Erikson, 1968; Garcia et al., 2013b), and psychological maturity measured as high SD and CO is a predictor of appropriate adaptation to society and/or ability to bounce back from lifespan challenges (Cloninger, Salloum & Mezzich, 2012a; Eley et al., 2013; Moreira et al., 2014). It is a period of identity formation showing dramatic transition from external inhibition on emotional drive by parents, teachers and guardians to cognitive and intentional self-regulation based on their own goals and values (Erikson, 1968; Moreira et al., 2014; Zohar et al., 2018).

Adolescents and adults with high SD and CO character experience more frequent positive emotions and life satisfaction throughout their lifetime, and less psychosocial dysfunction and behavioral disorders which reportedly increase dramatically during adolescence (Garcia, Anckarsäter & Lundström, 2013a; Garcia et al., 2013b). The matured adults demonstrate higher executive functioning such as response inhibition, attention, long-range planning and organization, and flexible regulation of impulses and employment of cognitive emotional strategies emerge and grow by adolescence (Garcia et al., 2013b).

Since the most effective preventive intervention for behavior problems may be accomplished by enabling matured coping to stress through promotion of character development both at home and school (Eley et al., 2013; Eley et al., 2016; Moreira et al., 2014; Zohar et al., 2018), emphasis should be placed on fostering personal integrity, self-esteem, self-acceptance, responsibility, relatedness and effectiveness in society (Cloninger, 2006; Magen, 1998) which the JTCI SD and CO character traits accentuate.

Children with high RD, PS, SD and CO were shown to exhibit high intelligence and academic achievement (Copeland et al., 2004), self-discipline in adolescents was reported to be a better predictor of academic performance than intellectual functioning in school (Duckworth & Seligman, 2005; Moreira et al., 2012), and matured adolescents exhibited higher level of positive affect, life satisfaction, social support and subjective health (Moreira et al., 2014). Educating adolescents receiving and providing help as peer counselors may be an exemplary intervention which is useful for strengthening a sense of purpose and worth, feelings of usefulness, and satisfying interaction with others (Garcia et al., 2013b; Magen, 1998).

This study may hold a number of limitations in regard to generalization. First, participants in the current study were high school students from a metropolitan area, and the ratio of well-adjusted to problematically adjusted students in addition to rural or urban characteristics (Eley et al., 2016) may have affected the outcome. Although the sample size was sufficiently large enough to dampen potential biases, it will be needed to be validated in diverse aspects to generalize the findings.

Second, the current study examined the personality characteristics of high school students who exhibits relatively stable temperament and character and can complete reliable self-report measures. Considering that behavior problems and coping mechanism in response to stress situations develop in complex interaction with temperament, character, environment, and learning (Cloninger, Salloum & Mezzich, 2012a; Eley et al., 2013), the findings of the current study with high school students should be also be replicated in middle school and elementary school students as well. This would allow for early and proactive preventive interventions through reviewing lifelong developmental perspectives on personality and psychopathology.

Conclusion

The current study confirmed previous reports (Eley et al., 2013; Eley et al., 2016; Lee, Choi & Chae, 2017a) that high NS and HA and low RD temperaments and low SD and CO characters are risk factors for behavior problems in adolescents both as individual traits and as a complex profile (Moreira et al., 2014; Zohar et al., 2018). The results would contribute to studies examining effects of personality on behavior problems and benefits of improving resilience to stress and nurturing lifespan psychological well-being by developing matured personality (Eley et al., 2016).

Supplemental Information