Juvenile’s Delinquent Behavior, Risk Factors, and Quantitative Assessment Approach: A Systematic Review : Indian Journal of Community Medicine

Secondary Logo

Journal Logo

Review Article

Juvenile’s Delinquent Behavior, Risk Factors, and Quantitative Assessment Approach

A Systematic Review

Gupta, Madhu Kumari; Mohapatra, Subrajeet; Mahanta, Prakash Kumar1

Author Information
Indian Journal of Community Medicine 47(4):p 483-490, Oct–Dec 2022. | DOI: 10.4103/ijcm.ijcm_1061_21
  • Open

Abstract

INTRODUCTION

Juvenile delinquency is a habit of committing criminal offenses by an adolescent or young person who has not attained 18 years of age and can be held liable for his/her criminal acts. Clinically, it is described as persistent manners of antisocial behavior or conduct by a child/adolescent repeatedly denies following social rules and commits violent aggressive acts against the law and socially unacceptable. The word delinquency is derived from the Latin word “delinquere” which described as “de” means “away” and “linquere” as “to leaveor to abandon.” Minors who are involved in any kind of offense such as violence, gambling, sexual offenses, rape, bullying, stealing, burglary, murder, and other kinds of anti-social behaviors are known as juvenile delinquents. Santrock (2002) defined “an adolescent who breaks the law or engages in any criminal behavior which is considered as illegal is called juvenile delinquent.”[1] In India, Juvenile Justice (J. J.-Care and protection of Children) Act of 2000 stated that “an individual whether a boy/girl, who is under 18 years of age and has committed an offense, referred or convicted by the juvenile court have considered a juvenile delinquent.”

PREVALENCE RATE: JUVENILE DELINQUENCY IN INDIA

According to the National Crime Records Bureau (India, 2019), statistical data of crimes in India show that overall, 38,685 juveniles were placed under arrest in 32,235 cases, among 35,214 juveniles were taken into custody under cases of IPC and 3471 juveniles were arrested under cases of special and local laws (SLL) during 2019. About 75.2% of the total convicted juveniles (29,084 out of 38,685) were apprehended under both IPC and SLL belonging to the age group 16–18 years. In 2019, 32,235 juvenile cases involving and recorded, indicating a slight increment of 2.0% over 2018 (31,591 cases). The rate of crime also indicates a slight increase from 7.1 (2018) to 7.2 (2019).[2] The total registered cases against juvenile delinquents are calculated as crime incidence rate per one Lakh population as shown in Figure 1.

F1-5
Figure 1:
The graphical view of registered cases against Juveniles in conflict with law under Indian penal code and special and local laws crimes during 2014–2019 of all the State (s) and union territories of India Sources: Crime in India National (2014-2019), National Crime Records Bureau (NCRB), Ministry of Home Affairs, 2019

RISK-FACTORS AFFECTING DELINQUENT BEHAVIOR

Studies identify that multiple risk factors are responsible for delinquent behavior categorized as individual, parental, family, community, society, schools/educational, financial, mental as well as psychological factors of the individual and the family shown in Table 1. Adolescents involve themselves in various anti-social activities to fulfill their basic needs. Basically, “delinquency” is just a recreational activity for earning money. These risk factors differ from person to person during the early childhood period and very crucial because children, who are involved in any kind of deviant activity at an early stage, have a higher chance to adopt delinquent tendencies chronically.[33]

T1-5
Table 1:
Developmental phases, risk-factors and developing delinquent behaviours of the child

Juvenile delinquency is caused by a wide range of factors, such as conflicts in the family, lack of proper family control, residential environmental effects, and movie influence, along with other factors are responsible for delinquent behavior.[3] Family and environmental factors, namely restrictive behaviors, improper supervision, negligence, criminal activities of parents, improper motivation by peers, fear of peer rejection, poverty, illiteracy, poor educational performance at school, lack of moral education may turn the individual personality into delinquents. Moreover, in the environment, deteriorated neighborhood, direct exposure to violence/fighting (or exposure to violence through media), violence-based movies are considered major risk factors.[4] In India, a higher level of permissive parenting in low-income families had so many family members and due to economic conditions, the adolescents had pressure to search various income sources to sustain the family, and it has affected parental behavior toward adolescents.[5] The children who belong to the lower middle-socio-economical class and are rejected by society showed more aggressive behavior.[6]

Juvenile gang members exhibit significantly higher rates of mental health issues such as conduct disorders, attention-deficit-hyperactivity-disorders, antisocial personality disorder, posttraumatic-stress-disorders, and anxiety disorders.[7] As well as the intellectual level of young offenders is significantly different from nonoffenders. Emotional problems on adolescents are related to delinquent behavior and impulsivity directly associated with antisocial behavior among adolescents.[8] Poor self-control of adolescents involved them in substance use, affected harmfully, and increased involvements in anti-social activities.[9] Nonviolent people, who not involved in any gang, are less likely to utilize mental-health services, having lower levels of psychiatric morbidity, namely antisocial personality disorders, psychosis, and anxiety disorders, when compared with the group of violent offenders.[10]

MACHINE LEARNING: A NEW QUANTITATIVE EVALUATION APPROACH

Machine learning (ML) is belonging to the multidisciplinary field that includes programming, math, and statistics, and as a new and dynamic field that necessitates more study. It is a branch of computer science that emerged through pattern recognition and computational learning theory of artificial intelligence. ML is exploring researches and development of algorithms that can learn and genera tea prediction besides a given set of data through the computer. It is a scope for the study that gives computers the capability to learn without being principally programmed.[11] Tom M. Mitchell explained ML as “a computer-based program to learn from action of “E” concerning any task of ‘T’s, and some performance evaluates “P,” if its performance on “T,” as assessed by “P,” improves with action of E.”[12] The goal of ML is to mimic human learning in computers.[13] Humans learn from their experiences and ML methods learn from data. The user provides a portion of a dataset designated to train by the algorithm. The algorithm creates a model based on the relationships among variables in the dataset, and the remaining dataset is used to validate the ML model. In simple words, ML approach is for risk indicator is meant to magnify the potential of current knowledge.[15] ML sits at the common frontier of many academic fields, including statistics, mathematics, computer science, and engineering.[14,17] ML models principally categorized into three categories, namely supervised, unsupervised, and reinforcement based on their task which they are attempting to accomplish. Supervised learning is relying on a training set where some characteristics of data are known, typically labels or classes, and target to find out the universal rule that maps inputs to outputs. Unsupervised learning has no design to give to the learning algorithm, balance itself to find out the patterns through inputs. In reinforcement, interaction with a dynamic environment happens during which a particular target such as driving a vehicle is performed without a driver principally involved in any activities, namely comparison. In numerous studies, pattern classification approaches based on ML algorithms are used to forecast human beings into various categories by maximizing the distance among data groups. ML generally refers to all actions that train a computer algorithm to determine a complicated pattern of data that is conceivable used for forecast category of membership into a new theme (e.g., individual vs. controls).[32]

RATIONAL OF THE STUDY

In the last decade, various researchers have been attracted to the use of quantitative computer-based techniques for analyzing various psychological and clinical aspects, which have greatly contributed to the area of modern psychology. In this analysis, most of the works are devoted to the use of various quantitative analysis techniques, namely ML and statistical methods which has utilized by the researchers for evaluating various risk and protective factors of juveniles. Henceforth, studies on the application of the ML model for risk-assessment of delinquent behavior on juveniles are limited as compared to other techniques, namely logistic regression. Hence, this review paper may explore the utilization of ML to get an easy and quick assessment on juveniles and helpful for future studies. It may help to determine the most significant risk factors and establishment of a successful treatment program that prevents juveniles from delinquent activities and stops them from recidivism.

In this review, all these studies carried out which has used various quantitative techniques to detected juvenile delinquency with specially emphasis on ML and statistical approaches. The review is organized into four sections follows as: Section-I gives an overview of juvenile delinquency, prevalence rates in India, and various behavioral risk factors during the developmental period. It also provides general information about ML as a new approach and their application. Section-II included information about the methodology of the present review. Section-III explores the results and discusses which explore the ML and statistical methods for detecting juvenile behaviors and Section-IV concludes the extant research of the present review and the implications for future work.

METHODOLOGY

This review paper aim is to find the various quantitative techniques (computer-assisted techniques) ML and statistical approaches which have been used for assessing/predicting delinquent behaviors, traits, and risk factors among juveniles.

Sources of information

For this review article, a total of 15 research articles were identified and selected through Google-scholar, Web of Science, Academia, PubMed, and Research-Gate, using the keywords, namely juvenile-delinquency, ML, Risk-factors, and delinquent-behavior. All relevant studies were selected for review of the quantitative approaches for identifying delinquent behavior and risk factors of adolescents and the preferred reporting items for systematic reviews and meta-analyses (PRISMA) flow diagram for articles search process as shown in Figure 2.[34]

F2-5
Figure 2:
Preferred reporting items for systematic reviews and meta-analyses flow diagram for search outcomes of quantitative assessment of juvenile delinquent behaviors

Inclusion criteria

Research studies published since 2011–2019, case studies, empirical, quantitative, qualitative, and cross-sectional studies published in English were included, which used ML and statistical models to analyze behaviors, risk and associated factors among juveniles.

Exclusion criteria

Protocol, dissertations, prototype studies, and studies which published in other languages were excluded.

RESULTS

Studies on machine learning and statistical methods among juvenile delinquency

In this review, we performed a rigorous search of the literature to provide a narrative description of the various quantitative computer-based approaches which are applicable to assess and identify the delinquent behaviors and risk factors on juveniles. Initially, the search identified 150 articles through various databases, search outcomes show in the PRISMA flow diagram [Figure 2]. One hundred and thirty-five articles were removed by screening through the title, text, removal of duplicate articles and based on inclusion and exclusion criteria, we identified 15 research articles in full text and these selected articles comprising through expert opinions. The findings of these articles tabulated the diverse approaches on the current state of knowledge about assessment of early diagnosis of delinquent behaviors and risk factors and tried to provide a summary which based on computer-based quantitative analysis [Table 2].

T2-5
Table 2:
Summary table of relevant studies which used quantitative approach to detect delinquent behaviors and risk factors among juvenile behaviors

DISCUSSION

In this systematic review, we performed a rigorous search of the literature to provide a narrative picture of various methods used to identify juveniles’ behaviors. We identified 15 articles, with the objective to analyze the application of ML and other quantitative approaches to assess various delinquent behaviors and risk factors of juveniles. The studies revealed ML is a new quantitative method to identify the risk factors and delinquent behavior henceforth; there very few studies are conducted. In this study, we tried to provide a summary of selected articles on the current state of knowledge about quantitative analysis for assessment of delinquent behaviors of juveniles and there only few articles have used ML as quantitative analysis. The City Social Welfare Development Office of Butuan, Philippines, used a dataset to create predictive models for analyzing the minors at risk and children in conflict with poor financial status. And found children with age range 12–17 years are victims of maltreatment, and adolescents between the ages of 15–17 years commit severe crimes.[16] Kim etal.[18] used traditional regression, ML method and certified the predictive validity of the models in numerous ways, along with traditional hold-out validation k-fold cross-validation, and bootstrapping to examine the present practice and policy for assessment, treatment, and management of delinquents who have a history of sexual conviction in multiple jurisdictions from New York, Florida, Oregon, Virginia, and Pennsylvania. Results revealed that important risk factors among juveniles had some criminal history, sexual offending experiences, and delinquent peers. Some dynamic factors viz. performance in school, peer connection, sorrowful feelings, impulsiveness, mental health, and substance abuse are important anticipating factors among sexual offenders for recidivism.

Rokven etal.[19] used multinomial logistic regression technique to compare four types of delinquent groups: online delinquents, offline delinquents, nondelinquents, and delinquents who belong to both online and offline categories and found juveniles who having both online and offline criminal records are more likely to commit crimes. Delinquency is indirectly linked with sleep deprivation, with poor self-control acting as a catalyst proved by regression models with latent factors.[20] Violent video games directly associated with anti-social behavior, even though several correlates, such as psychopathologies has present in youth analyzed by negative binomial regression (extended version of Poisson regression).[22]

Fernández etal. analyzed through multivariate logistic regression and found, school dropouts’ teenagers had a higher level of irresponsibility, substance, and illicit drug abuse compare then nondropouts.[23] In addition, lack of parental supervision plays a significant role in the prediction of deviant behaviors on school dropouts. School dropout teenagers have multi-dimensional problem that requires proper parental supervision and proactive school policies to reducing drug and alcohol abuse.[23] Fifty-two percent of juvenile offenders had issues with academic performance, 34% had family history of psychiatric disorders, 60% of juveniles involved in property crime and 54% of offenders involved in drugs and alcohol use-related offenses had some deficiency in academic achievement evaluated by multiple regression techniques.[24] Wu (2015) created a multidimensional scaling model and found students used a complex cognitive-mechanism measured and compared their position to friends and others.[25]

Sexually assaulted history has strongly associated and one of the most powerful variables associated with the intensity of psychoactive substances using by juveniles.[26] Parks[28] has used binary logistic regression and multivariate models revealed that no major variations in violent juveniles belong to cohabiting families and other families. However, teenagers of cohabiting families have marginally higher risk to involving in nonviolent forms of crime.[28] Economic conditions of the family has strongly linked to the influences of parents, siblings, and peers at risk and delinquency. Economic stress, having an active sibling aggression, harmful, and more destructive events affected seriously on adolescent delinquent behaviors who belongs to economically poor families.[29] Coercive parents are directly associated with violent delinquency of adolescents on both ways as explicitly and indirectly and transformed shame on adolescents. As opposed to articulated guilt, shame conversion is the major cause for more violence.[30]

It is very difficult to evaluate all possible outcomes and explain a single quantitative approach as ML to early identification of delinquent behaviors and risk-factors of juveniles for intervene in the affected factors. Our study has several limitations. First, other studies rather than the English language were we not included in the study. Second, counties like India have very less evidence-based studies in the field of early detection of juveniles and computer-based assessment approaches as ML for quantitative analysis. Third, only 15 articles were considered which fulfilled the inclusion criteria.

IMPLICATION

The modern world is fully based on computers and technology for making works easy and faster. ML model is an emerging future technology in the field of health and mental health. It has the potential to predictive ability to detect health/mental health-related problems as well as for early diagnosis of problems behaviors. This review is acknowledging the use of quantitative analysis focused on ML algorithm as a new research area for early identification of delinquent behaviors of children, to prevent the deviant behaviors and related risk-factors and may be beneficial for future studies and contribute to make a peaceful society and worthful young generation for the nation.

CONCLUSION

This review showed that available literature based on ML and other quantitative methods to identify the risk factors and delinquent behaviors of juveniles. Young peoples are at a higher risk to learn maladaptive/deviant behaviors as violent, aggressive, hyperactive, and easily involved in criminal activities. According to studies, individual factors, family environment, family structure, size/type of the family, parental status (single/separate/divorces) are highly affected adolescent’s behaviors. In addition, social, environmental, and economic conditions are lead to adapt conductive and delinquent behaviors. There highly need to identify delinquent behaviors in the initial stage to prevent with affected risk factors. It is very crucial for early screening and intervention.

Financial support and sponsorship

Nil.

Conflicts of interest

There are no conflicts of interest.

Acknowledgment

Authors acknowledge to Department of Science and Technology- Cognitive Science Research Initiative (DST-CSRI) for sponsored the project in the Department of Computer Science and Engineering, Birla Institute of Technology, Mesra, India, which explores the technology-based approach in multidisciplinary works. The authors also would like to thank Mr. Abhinash Jenasamanta and Mr. Devesh Upadhyay, Research Scholars, Department of Computer Science and Engineering, BIT, Mesra, Ranchi, for technical and motivational support.

REFERENCES

1. Santrock JW, Sumiharti Y, Sinaga H, Damanik J, Chusairi A. “Life-Span Development (Perkembangan Masa Hidup) jilid”. Published by Jakarta, Erlangga; 2002: 1.
2. National Crime Report Bureau (NCRB), Ministry of Home Affairs, Government of India. Chapter- A & B, Statistical. 1:403–48. Available from: https://ncrb.gov.in-2019. Last accessed on 2022 Jan 27.
3. Haveripet P. Causes and consequences of juvenile delinquency in India. Recent Research in Science and Technology 2013;5:29–31. Available Online: http://recent-science.com/. Last accessed on 2022 Mar 18.
4. Ahmed U, Murtaza A. Factors affecting juvenile delinquency in Punjab, Pakistan:A case study conducted at juvenile prisons in Punjab province. Mediterr J Soc Sci 2016;7:372.
5. Moitra T, Mukherjee I, Chatterjee G. Parenting behavior and juvenile delinquency among low-income families. Vict Offender 2018;13:336–48.
6. Bhatti AG. Difference in parental acceptance-rejection and personality organization in children of Hyderabad. Bahria J Prof Psychol 2013;12:64–85.
7. Madden V. Understanding the Mental Health Needs of Young People Involved in Gangs: A Tri-Borough Public Health Report Produced on Behalf of the Westminster Joint Health and Wellbeing Board. The Westminster Joint Health and Wellbeing Board 2013. Available from: https://www.jsna.info. Last accessed on 2020 Feb 17.
8. Overbeek G, Vollebergh W, Engels R, Meeus W. Juvenile delinquency as acting out:Emotional disturbance mediating the effects of parental attachment and life events. Eur J Dev Psychol 2005;2:39–46.
9. Holt TJ, Bossler AM, May DC. Low self-control, deviant peer associations, and juvenile cyberdeviance. Am J Crim Just 2012;37:378–95.
10. Coid JW, Ullrich S, Keers R, Bebbington P, Destavola BL, Kallis C, et al. Gang membership, violence, and psychiatric morbidity. Am J Psychiatry 2013;170:985–93.
11. Samuel AL. Some studies in machine learning using the game of checkers. IBM J Res Dev 1959;3:210–29.
12. Mitchell TM. Does machine learning really work?. AI Mag 1997;18:11.
13. Dutton DM, Conroy GV. A review of machine learning. Knowl Eng Rev 1997;12:341–67.
14. Friedman JH. Data mining and statistics:What's the connection?. Comput Sci Stat 1998;29:3–9.
15. Kim K, Duwe G. Improving the performance of risk assessments: A case study on the prediction of sexual offending among juvenile offenders. Handbook on Risk and Need Assessment. New York. https://doi.org/10.4324/9781315682327 .Routledge; 2016: 130–55.
16. Castro ET, Hernandez AA. Developing a predictive model on assessing children in conflict with the law and children at risk:A case in the Philippines. 2019 IEEE 15th International Colloquium on Signal Processing & Its Applications (CSPA). Penang, Malaysia: IEEE; 2019: 243–8.
17. Marsland S. Machine Learning: An Algorithmic Perspective. CRC Press, Taylor & Francis Group, Boca Raton, London, New-York; 2015.
18. Kim K, Duwe G, Tiry E, Oglesby-Neal A, Hu C, Shields R, et al. Development and Validation of an Actuarial Risk Assessment Tool for Juveniles with a History of Sexual Offending. CrimRxiv [Internet] 2019. Available from: https://www.crimrxiv.com/pub/0ai9byv6. Last accessed 2021 Aug 31.
19. Rokven JJ, Weijters G, Beerthuizen MG, van der Laan AM. Juvenile delinquency in the virtual world:Similarities and differences between cyber-enabled, cyber-dependent and offline delinquents in the Netherlands. Int J Cyber Criminol 2018;12:27–46. DOI:10.5281/zenodo.1467690.
20. Meldrum RC, Barnes JC, Hay C. Sleep deprivation, low self-control, and delinquency:A test of the strength model of self-control. J Youth Adolesc 2015;44:465–77.
21. Castellana GB, Barros DM, Serafim Ade P, Busatto Filho G. Psychopathic traits in young offenders vs. non-offenders in similar socioeconomic condition. Braz J Psychiatry 2014;36:241–4.
22. DeLisi M, Vaughn MG, Gentile DA, Anderson CA, Shook JJ. Violent video games, delinquency, and youth violence:New evidence. Youth Violence Juv Justice 2013;11:132–42.
23. Fernández-Suárez A, Herrero J, Pérez B, Juarros-Basterretxea J, Rodríguez-Díaz FJ. Risk factors for school dropout in a sample of juvenile offenders. Front Psychol 2016;7:1993.
24. Margari F, Craig F, Margari L, Matera E, Lamanna AL, Lecce PA, et al. Psychopathology, symptoms of attention-deficit/hyperactivity disorder, and risk factors in juvenile offenders. Neuropsychiatr Dis Treat 2015;11:343–52.
25. Wu CI. The impact of co-evolution of dynamic networks upon adolescent deviant behaviors. Proceedings of the 2015 IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining. Association for Computing Machinery, New York, NY, United States. https://doi.org/10.1145/2808797.2809319;2015.p.1411-27.
26. Brunelle N, Tremblay J, Blanchette-Martin N, Gendron A, Tessier M. Relationships between drugs and delinquency in adolescence:Influence of gender and victimization experiences. J Child Adolesc Subst Abuse 2014;23:19–28.
27. Gordon RA, Rowe HL, Pardini D, Loeber R, White HR, Farrington DP. Serious delinquency and gang participation:Combining and specializing in drug selling, theft and violence. J Res Adolesc 2014;24:235–51.
28. Parks AB. The Effects of Family Structure on Juvenile Delinquency; Paper 2279. https://dc.etsu.edu/etd/2279. Last accessed on 2020 Dec 14.
29. Low S, Sinclair R, Shortt JW. The role of economic strain on adolescent delinquency:A micro-social process model. J Fam Psychol 2012;26:576.
30. Gold J, Sullivan MW, Lewis M. The relation between abuse and violent delinquency:The conversion of shame to blame in juvenile offenders. Child Abuse Negl 2011;35:459–67.
31. Sumalatha V, Santhi R. An improved Bayes classification approach to reduce affliction of Juvenile. IEEE International Conference on Computational Intelligence and Computing Research (ICCIC). IEEE, Madurai, India; 2018: 1–4. DOI:10.1109/ICCIC.2018.8782351.
    32. Bishop CM. Pattern recognition. Mach Learn Springer, New York; 2006;128:1–58.
    33. Sage Publication; International Educational and Professional Publisher, Thousand Oaks, London, New Delhi 1998.
    34. Page MJ, McKenzie JE, Bossuyt PM, Boutron I, Hoffmann TC, Mulrow CD, et al. The PRISMA 2020 statement:An updated guideline for reporting systematic reviews. BMJ 2021;372:n71.
    Keywords:

    Delinquent behavior; juvenile-delinquency; machine learning; risk-factors

    © 2022 Indian Journal of Community Medicine | Published by Wolters Kluwer – Medknow