Increased Prescribing of Psychotropic Drugs or School-Based Services for Children with Disabilities? Associations of These Self-control-Boosting Strategies with Juvenile Violence at the State Level

Abstract

The increasing rates at which psychotropic drugs have been prescribed to children and adolescents in the USA in the last three decades (since the early 1990s) have prompted questions about whether this trend is associated with the “great American crime decline.” Medicalization can be considered one of the strategies to remedy children’s neuropsychological deficits and improve their self-control. Another possible remedy is school-based services for children with learning disabilities, mandated by the Individuals with Disabilities Education Improvement Act (IDEA) of 2004. Using state-level panel data analyses for years 1990–2014 (with the main focus on 2000–2014 outcomes), the current study estimates associations between these two developmental self-control remedies—medicalization and school-based services—and minor, moderate, and severe types of juvenile violence, while controlling for relevant covariates (both time-varying and time-invariant). The results of mixed-effects linear regression analyses accounting for powerful time trends show a strong association between increases in school-based services for children with learning disabilities and declines in all types of juvenile violence. Another strong and consistent finding that emerges in the analyses is the link between reductions in child poverty at the state level and decreases in juvenile violence, both contemporaneously and over time. Psychotropic drug prescribing to children (measured using Supplemental Security Income rolls of children with mental health conditions) exhibits inconsistent or insignificant effects. The findings of this study have substantial theoretical and policy implications and indicate the importance of strengthening school-based services for children with disabilities and reducing child poverty as essential violence prevention tools.

Introduction

The recent COVID-19 pandemic has been associated with an increase in violent crime, especially homicide and mass shootings, in the USA (Rosenfeld et al., 2021; Schildkraut & Turanovic, 2022). At the same time, evidence shows that these increases in violence largely did not involve the youth population (Mendel, 2022). In fact, violent crime among young people has been decreasing since the early 1990s, both based on official arrest data (Puzzanchera, 2019, 2020), National Crime Victimization Survey (Irwin et al., 2021; Kaylen et al., 2017), as well as Youth Risk Behavior Surveillance System (Perlus et al., 2014). In fact, during the 1990s, the most dramatic decrease in crime and delinquency happened among young people (Baumer et al., 2021; Cook & Laub, 2002; Mendel, 2022; Tcherni-Buzzeo, 2019). For example, the percentage of all arrests in the USA that involved youth under 18 more than halved—from 15% in 2000 to 7% in 2019 (Mendel, 2022), while arrests for homicide in this age group decreased by three quarters—from around 120 per 1 million in the early 1990s to under 30 per 1 million in the 2010s (Tcherni-Buzzeo, 2019, p. 314). Thus, the possible explanations about the “great American crime decline” must take into account this important fact (see a comprehensive list of such explanations, along with their categorization and critical analysis, in Tcherni-Buzzeo, 2019). Moreover, it is essential to note that changes in the demographic profile did not contribute much to the violence declines in the USA (Kaylen et al., 2017), despite playing a prominent role in the international comparisons of homicide declines (Rennó Santos et al., 2019).

Since the crime declines happened during approximately the same time—starting in the early-to-mid 1990s—not only in the USA,¹ but in other developed (and many developing) countries as well (Eisner et al., 2016; Tseloni et al., 2010), theories for the crime drop must be universal enough to go beyond localized explanations like policing strategies (Blumstein & Wallman, 2006) or abortion legalization (Donohue & Levitt, 2001). One such rather universal possibility is the increased prescribing of psychotropic medications to children and adolescents and the respective increased consumption of psychotropic drugs (Bouvy & Liem, 2012; Finkelhor & Johnson, 2017; Finkelhor & Jones, 2006; Marcotte & Markowitz, 2011; Pappadopulos et al., 2006). Importantly, the increased psychotropic medication prescribing applies to adults as well, even though the increase was much more drastic among juveniles (Olfson et al., 2002, 2006, 2012; Pappadopulos et al., 2006). Since psychotropic medications are often prescribed to children for conditions involving impulsivity, aggression, and mood changes, and are intended to increase children’s sense of control, their ability to follow directions and stick with the routine in day-to-day life (Comer et al., 2010; Connor et al., 2019; Turgay, 2004), it is reasonable to expect that increased medicalization reduces engagement in undesirable behaviors, including delinquency and violence (Finkelhor & Jones, 2006). This “calming” effect of psychotropic drugs on violence has been supported in some studies (Bouvy & Liem, 2012; Pappadopulos et al., 2006), though others found this effect to be less consistent and substantively small (Marcotte & Markowitz, 2011) or even opposite to what would be expected (Molero et al., 2015). It is important to note though that Molero and colleagues (2015) acknowledge they have not been able to account for time-varying factors in their study, which is an important gap remedied in the current study.

In the USA, the increased prescribing of psychotropic medications to children is closely tied to several legal changes:

• 1) In the 1990s, Supplemental Security Income (SSI) eligibility was expanded to include children with mental health impairments other than intellectual disability—for example, ADHD, bipolar disorder, etc.—in many states, qualifying the whole low-income family of the child for additional Medicaid coverage (Autor, 2015; Wiseman, 2011). Importantly, changes in SSI rolls closely track the prevalence of diagnosed mental disorders of children in the general population (Boat & Wu, 2015; Romig, 2017), likely for the following reasons: a) close to 40% of US households are eligible for SSI based on income eligibility (200% of the poverty threshold), and b) children with mental disorders other than intellectual disability are the largest share of SSI recipients—over 60% of SSI rolls (U.S. Social Security Administration, 2022)

• 2) In 2001, the No Child Left Behind (NCLB) Act was passed, which changed the rules of school accountability. In 2002–2003, its clarification and broader application in some states led to increasing pressure on schools competing for funding in those states to raise their students’ average scores by excluding low-performing students’ results from calculation (Bokhari & Schneider, 2011; Figlio & Ladd, 2015). The way to do this was suggested by the Individuals with Disabilities Education Improvement Act (IDEA) of 2004, which was interpreted in certain states to allow schools to exclude the scores of learning-disabled students from the calculation of school-wide performance measures (Yell et al., 2006). A lot of low-performing schools in those states were thus encouraged to recommend the families of academically challenged students to get these students diagnosed with mental health conditions like ADHD (Fulton et al., 2009, 2015; Hinshaw & Scheffler, 2014).

Even though both sets of legislative changes have been passed at the national level, they were implemented differently at the state level (Bokhari & Schneider, 2011; Hinshaw & Scheffler, 2014), which has created differential increases in demand for school-based services for children with learning disabilities (Hanushek & Raymond, 2004; Katsiyannis et al., 2007; Yell et al., 2006).

Thus, the current study takes advantage of the variations among US states in (a) the way the SSI rolls of children with disabilities due to mental health conditions other than mental retardation were expanded, and (b) different increases that happened in the share of children with learning disabilities who were provided with dedicated school-based services under the IDEA law. The associations between these changes over time and the declines in juvenile delinquency and violence at the state level are the main focus of the study. Thus, the corresponding research questions of the current study are these:

• 1) Are levels of juvenile delinquency and violence lower in states with more SSI funding for psychotropic medications for children, net of other important factors?

• 2) Are levels of juvenile delinquency and violence lower in states with more school-based services for children with learning disabilities, net of other important factors?

• 3) Which one of these two factors—psychotropic medicalization of children or availability of school-based services—has more prominent associations with juvenile violence declines?

It is important to acknowledge that the explanations that seem more suitable for testing at the individual level are tested in the current study at the aggregate level: the unit of analysis is a state. The main reason for this is the privacy concerns and protections prominent in the USA, which make it essentially impossible to conduct a large-scale analysis involving information about such sensitive individual matters as mental health conditions, medication prescribing, or individualized education plans. There is nothing in the USA that even approaches the design of national population registers of Scandinavian countries like Sweden and European countries like the Netherlands (Falk et al., 2014; Hjalmarsson et al., 2015; Mok et al., 2018; Molero et al., 2015; Nilsson et al., 2017), where information is assembled at the individual level about medical conditions, treatment, school performance, family poverty, and criminal justice outcomes. However, it is very important to be able to investigate these associations in the USA, considering the paucity of research in this country (Marcotte & Markowitz, 2011).

On the other hand, there are benefits to conducting analyses using aggregate data at the state level, besides the obvious ones of complete privacy protection and being able to use information from the country like the USA. First, aggregating data into larger units of analysis reduces the “noise” in the data (measurement error, nonlinearity, confounding, etc.) and makes data analyses more efficient (Guthrie & Sheppard, 2001; Schwartz, 1994). For example, in the current study, using fixed-effects panel data analyses, it is possible to rule out alternative factors contributing to the observed outcomes, by statistically controlling for time-stable differences among US states that could potentially confound the main variables of interest (Allison, 2009; Bell & Jones, 2015; Wooldridge, 2010; though also see Mummolo & Peterson, 2018 for important qualifiers of this approach).

Second, aggregate-data analyses allow the use of rare outcome events of extreme violence, like homicides committed by younger people, which would have been impossible with a smaller individual-level sample. In a way, the benefits of aggregation are most easily illustrated using true experiments (or randomized controlled trials) as an example. Even though each individual person in the experimental or control group may respond differently to the treatment or a placebo, by calculating the average aggregate response for each group and comparing them statistically, we are able to reduce the individual ‘noise’ and take advantage of the data analysis tools for comparing groups. Finally, there are more measures from various sources of information available at the aggregate level of the state, which allows for data triangulation and comparison.

The benefits of aggregation are the main reason why almost all research on the causes and determinants of crime declines is conducted at the aggregate level (Baumer et al., 2018; Sharkey et al., 2017; though see Nilsson et al., 2017 as a rare exception; also see explanations in Tcherni, 2011 on how individual-level regression coefficients translate into aggregate effects). Moreover, aggregate-level analyses can help determine which policy-level decisions are likely to lead to sustained declines in crime at the population level (Eisner et al., 2016), which has direct importance for policy implications. At the same time, there are also significant trade-offs with the use of aggregate data: one cannot achieve the level of detail that is possible when using individual-level datasets, and several important fallacies should be avoided when interpreting the results (Cooper & Patall, 2009). However, a comparison of meta-analyses using individual-level versus aggregate-level data demonstrates that the results and conclusions drawn from the two types of sources are very similar (Smith et al., 2016). There is also a well-supported argument that any discrepancies between aggregate-level and individual-level findings likely point to validity problems rather than to the ecological fallacy (Schwartz, 1994; Sideridis et al., 2018) and that aggregate data analyses can mitigate many of the sources of bias present in individual-level data (Guthrie & Sheppard, 2001).

Thus, to test the research questions posed, I estimated a series of mixed-effects linear regression analyses to model how changes in the prescribing of psychotropic medications (using SSI data)² and changes in school-based services to children with learning disabilities (using U.S. Department of Education data) are associated with juvenile delinquency and violence (using self-report-based data from the Youth Risk Behavior Surveillance System and official arrest data from the Federal Bureau of Investigations based on reporting from local police departments) mainly focusing on the time period of 2000–2014. The results of analyses show that, at the state level, psychotropic medication prescribing has no consistent associations with youth criminal outcomes, while increases in school-based services for children with learning disabilities are consistently associated with declines in youth violence over time. As expected, child poverty has the strongest link with youth violence, and decreases in child poverty rates at the state level have one of the most powerful associations with decreases in youth violence, both contemporaneously and over time. The concluding part of the article discusses limitations of the current study design and measures, outlines directions for future research, and draws implications of the study findings for crime prevention, mainly in terms of improving and expanding school-based services for children with learning disabilities and combating childhood poverty.

Literature Review

To properly situate the current study, several significant areas of research should be reviewed in more detail:

• • Crime declines happening in the last three decades and their possible causes and explanations

• • Upward trends in psychotropic drug prescribing and consumption, especially among children and adolescents in the USA, and the potential relationship of this factor with the declines in juvenile delinquency and violence

• • Connections between learning disabilities and aggression through the lens of school-based services for disruptive children, as well as the legal changes in the USA at the federal level and their translation into state policies and practices that have affected both psychotropic drug prescribing and educational practices for children with learning disabilities

• • The role of childhood poverty in youth violence

Then, these separate strands of research are weaved together into an overarching narrative, to situate the current study and demonstrate how it fills an important gap in the existing research literature.

Dramatic Declines in Crime and Youth Violence

Criminologists have long been puzzled by the sustained and widespread declines in crime and violence that started in the early 1990s in the USA (Baumer & Wolff, 2014; Tcherni-Buzzeo, 2019; Zimring, 2006). Especially puzzling is the fact that the crime declines are deepest among the traditionally most crime-prone age groups—juveniles and young adults (Baumer et al., 2021; Tcherni-Buzzeo, 2019).

A broad array of possible explanations—from legalized abortion (Donohue & Levitt, 2001) to changes in policing strategies (Blumstein & Wallman, 2006; Kubrin et al., 2010) to decreases in childhood lead exposure (Nevin, 2000, 2007; Reyes, 2007; Stretesky & Lynch, 2001)—has been offered and debated (Tcherni-Buzzeo, 2019). Most popular explanations for the crime decline of the 1990s (the waning crack epidemic, abortion legalization, changes in policing and punishment strategies, etc.—see Blumstein et al., 2000; Cook & Laub, 2002) do not extend well to explain the continuing declines in crime during the 2000s and 2010s. The wide and pervasive nature of the crime decline, as well as its unexpectedly long duration into the 2000s and 2010s (Baumer et al., 2021; Zimring, 2006), seems to defy most “localized” explanations, especially if we take into account the crime drops that the other developed countries have experienced following the “great American crime decline” (Eisner et al., 2016; Knepper, 2012; Rosenfeld & Messner, 2009; Tseloni et al., 2010; Van Dijk et al., 2007).

One possible explanation for the crime declines in the developed world in general, and in the USA specifically, was offered by David Filkelhor and his colleagues (Finkelhor & Johnson, 2017; Finkelhor & Jones, 2006): the increased prescribing of psychotropic drugs to children and adolescents. Next, the potential of this factor to explain the sustained declines in juvenile delinquency and violence is examined through a closer look into the main reasons for increased medicalization, the explosion of ADHD diagnosis and its relationship with aggression and criminal justice outcomes, and the results of evaluation studies comparing treated and untreated matched populations or periods of treatment and non-treatment within the same individuals in terms of adverse behavioral outcomes. Further in the paper, a second related factor is examined: the increase in school-based services to children with disabilities, how the increased diagnosing and medicalization of children is associated with law changes and school accountability processes at the state level, and how child poverty plays a role in many of these processes.

The Rise of Psychotropic Medications

The rate at which psychotropic drugs such as stimulants, antidepressants, and antipsychotic medications get prescribed to children and adolescents in the USA has increased dramatically during the 1990s (Mojtabai & Olfson, 2020; Moreno et al., 2007; Olfson et al., 2002, 2006, 2012; Pincus et al., 1998), and the increase has continued and, for certain drugs, accelerated after 1999 (Comer et al., 2010; Thomas et al., 2006). These trends clearly coincide with the declines in juvenile violence over the recent decades (Baumer et al., 2021), especially the continuing throughout the 2010s declines in homicide perpetrated by youth (Tcherni-Buzzeo, 2019), echoed by the continuing homicide declines in many European countries (Suonpää et al., 2022).

The impact of psychopharmacology on crime and deviance, according to Finkelhor & Jones, (2006), could be explained by a) the easing of chronic depression, despair, and discouragement that often accompanies mental diseases, and b) through the effect of drugs in curbing impulsive and risk-taking behaviors (this explanation hints at the improvement in self-control, which will be examined later in the paper). Without the relief of drugs, these behaviors are likely to lead to delinquency and violence (Connor et al., 2002, 2019; Lichtenstein et al., 2012; Zych et al., 2021). In addition, curbing these behaviors can lead to improving family relationships and more effective parenting involvement, which additionally helps reduce delinquent behavior among youths (Finkelhor & Johnson, 2017; Finkelhor & Jones, 2006). It is also possible that the increased consumption of psychotropic drugs curbs juvenile violence and delinquency indirectly, by reducing/substituting the use of illegal psychoactive drugs and alcohol (Grucza et al., 2018).

Who Gets Medicated?

The most common diagnoses associated with prescriptions of antipsychotic medications to youth are disruptive behavior disorders (Comer et al., 2010; Turgay, 2004), which include symptoms of several related conditions in children such as conduct disorder, attention deficit hyperactivity disorder (ADHD), and antisocial personality disorder. In fact, aggression is the most common reason for children and adolescents to be referred for psychiatric evaluation (Connor et al., 2019). The association of aggression, especially impulsive aggression, with psychiatric diagnoses and need for treatment, has been consistently reflected in research literature (Fahim et al., 2011; Moffitt & Silva, 1988; Retz & Rösler, 2009). Similarly documented is the persistence of child conduct problems into adolescent delinquency and adult criminal behavior (Dalsgaard et al., 2013; Kratzer & Hodgins, 1997; Mohr-Jensen & Steinhausen, 2016; Retz et al., 2021).

The childhood psychiatric disorders associated with disruptive behavior and aggression are rather persistent across the life course (Harty et al., 2009; Lahey et al., 2004; McKay & Halperin, 2001), more common among boys rather than girls (Frick et al., 1994; Loeber et al., 2000; Mowlem et al., 2019; Visser et al., 2014), among minority youths rather than white youths (Delbello et al., 2001; Nguyen et al., 2007), and among children in poverty (Choi et al., 2017; Pastor & Reuben, 2008). These are the patterns closely matched in juvenile delinquency involvement reflected in both self-reports and official statistics, especially when more serious and violent delinquency is concerned (Blum et al., 2003; Krivo & Peterson, 1996; Rekker et al., 2015).

ADHD, Its Treatment, and Criminal Justice Outcomes

ADHD is the most common psychiatric diagnosis among children and adolescents, with an estimated worldwide prevalence of 8–12% (Luo et al., 2019, though underdiagnosed in many countries—see Sayal et al., 2018). A comparable figure for the USA is over 9% of children 2–17 years of age who are officially diagnosed with ADHD (Danielson et al., 2018). ADHD is frequently comorbid with other psychiatric and learning disorders (Luo et al., 2019; Reale et al., 2017) and is often accompanied by aggression and criminality (Cherkasova et al., 2022; González et al., 2013; Gudjonsson et al., 2014; Lichtenstein et al., 2012; Zhang et al., 2021). ADHD is several times more common among criminal justice-involved juvenile populations (about 30% prevalence) and among adult prisoners (about 26% prevalence) compared to the general population prevalence of 3–7% among youth and 1–5% among adults (Young & Cocallis, 2021; Young et al., 2015).

Over 60% of US children with ADHD are treated for this condition with medication (Danielson et al., 2018; Visser et al., 2014). In fact, stimulant medications intended for treatment of ADHD and related conditions are the most common class of psychotropic drugs prescribed to children and adolescents (Comer et al., 2010; Visser et al., 2014), and they have been found effective in reducing ADHD symptoms and aggressive behaviors (Connor et al., 2002; Golubchik et al., 2018; Huang et al., 2021; Lichtenstein et al., 2012), as well as a wide range of other functional outcomes (Boland et al., 2020; though also see Lee et al. (2016) and Borgen et al. (2021) showing that alternative behavioral treatment programs in educational settings can partially substitute medication). In fact, improvements in the educational performance of children while receiving medication for ADHD have been confirmed using Danish register data (Keilow et al., 2018), as well as in a recent meta-analysis including a wide range of outcomes (Boland et al., 2020).

Moreover, using Swedish population registers, Campbell and colleagues (2019) show that the earlier the children get diagnosed and treated, the less likely they are to experience problems like aggression and criminal convictions later in life. At the same time, the effectiveness of treating adults with these medications and the impact on adult aggression is less clear: the results of randomized, double-blind placebo-controlled trials of stimulant medications (standard treatment for ADHD) in prison settings and the effects of these medications on ADHD symptoms and other outcomes such as violent attitudes and aggressive behaviors have been mixed (Asherson et al., 2023; Ginsberg & Lindefors, 2012; Ginsberg et al., 2015).

From the combination of research findings described above, and taking into account that a small percentage of offenders—about 5–6%—are responsible for a disproportionately high share of crimes—about 40–70% of all crimes (Farrington & West, 1993; Kratzer & Hodgins, 1997; Wolfgang et al., 1972), or as Falk and colleagues (2014) aptly put right into the title of their article, “1% of the population accountable for 63% of all violent crime convictions” (p. 559), it logically follows that adolescent delinquency and violence trends could be influenced significantly by the increased medicalization of a relatively small share of children and adolescents who would have been likely to engage in persistent delinquent and violent behaviors otherwise (i.e., without the curbing effect of medication).

Learning Disabilities, Aggression, and School-Based Services

ADHD is just one of the range of conditions associated with childhood disruptive behaviors that lead to learning disabilities and problems with educational performance (Brennan et al., 2015; Carbonneau et al., 2016; Pastor & Reuben, 2008). Children with neurological deficits who were exposed to adverse childhood experiences, or ACEs, are found to be especially likely to experience problems and exclusion in educational settings (Novak, 2022; Pierce et al., 2022). Students’ learning disabilities are associated with poor anger regulation and aggression, as has been confirmed and investigated further in research literature, though the mechanism of this connection is not fully understood and interventions for reducing such problems are still being evaluated (Bellemans et al., 2019; Jahoda et al., 2001; Willner, 2015). Moreover, high impulsivity, attention deficits, and low school achievement are some of the common factors predicting juvenile offending across multiple cohort studies in several developed countries (Zych et al., 2021). At the same time, students with learning disabilities in the USA do not face a higher likelihood of suspension compared to similarly situated non-disabled students (Morgan et al., 2019). Thus, these complicated relationships between learning disabilities, school-based services, and delinquent or violent outcomes need to be further investigated.

Each year since the passage of the Individuals with Disabilities Education Improvement Act (IDEA) in 2004, more children with disabilities are receiving additional services in schools (U.S. Department of Education, 2022). Moreover, a special emphasis is put on inclusion (Kirby, 2017; Rivera & McKeithan, 2021), with the vast majority of children with disabilities being included into general education classrooms and activities: for example, in 2019, over 95% of students aged 5 to 21 spent at least a portion of the school day in regular classrooms (U.S. Department of Education, 2022). At the same time, the share of students with disabilities served under IDEA varies widely among US states, as well as the pace of increase in such a share over time (U.S. Department of Education, 2022). The current study takes advantage of this variation by state and models these changes to estimate whether the increases in school-based services to students with disabilities are an important factor associated with the declines in juvenile delinquency and violence. For a proper background on the legal framework that significantly contributed to changes in both the diagnosing and the treatment of children with psychotropic medications, as well as the changes in school-based services to children with disabilities, the next section briefly reviews the relevant federal legislation and its implementation at the state level.

Variability of Policy Implementation by State

As mentioned earlier in the article, several important nation-wide policy changes have contributed to the trend of increasing prescribing of psychotropic medications to children and adolescents, especially the ones from low-income families, as well as increases in school-based services to children with disabilities: the 1990 expansion of Supplemental Security Income (SSI) eligibility for children with mental disorders, the No Child Left Behind (NCLB) of 2001, and the Individuals with Disabilities Education Improvement Act (IDEA) of 2004. These legislative changes, though applicable nationally, were implemented differently in different states. For example, in some states, if a child were eligible for Supplemental Security Income, the child’s family would automatically qualify for Medicaid health insurance (Wiseman, 2011). Moreover, some states undertook active efforts to shift poor families from the state financial assistance that comes with Temporary Assistance for Needy Families (TANF) to federal assistance that comes with SSI and Medicaid, even hiring companies that help potentially eligible families to qualify for such a transfer (Wamhoff & Wiseman, 2006). States win, families win (since being awarded SSI assistance means a few hundred dollars of extra cash payment per child per month and, in most states, Medicaid health insurance for the whole family), and the companies that help poor families to make this transition successfully get compensated for their work (Autor, 2015). In fact, the increases in mental health diagnoses of children (often combined with psychotropic medication treatment), and the corresponding expansion of SSI rolls, were so drastic that in 1996, the law was somewhat modified to clarify the range and severity of qualifying mental health conditions, though this modification did not substantially slow down the general trend in the increasing SSI rolls of children with mental health disorders (Autor, 2015; Wiseman, 2011; also see Fig. 1).

Fig. 1

Changes over time for US states in the percentage of mental-health-related supplemental security income (SSI) recipients among children under the age of 18 and percentage of students age 3–21 who are receiving school-based services under the Individuals with Disabilities Education Improvement Act (IDEA). Note: Data are not available by state on SSI recipients in the mental-health category for years 1991–1999 and 2001–2004 and on students with disabilities served in schools for years 1990–2001

Variability by state also applies to the No Child Left Behind (NCLB) Act and its effects (Fulton et al., 2015; Hinshaw & Scheffler, 2014). When financial rewards become tied to the test scores of students, schools are made to compete with one another for funding. The level of this financial pressure depends on the state’s decision regarding the rules of accountability and the reporting of academic scores for students with learning disabilities (Hanushek & Raymond, 2004; Katsiyannis et al., 2007; Yell et al., 2006). In some states, the intense pressure of competition for resources among schools was so severe that school administrators and staff were actively encouraging parents to get their children assessed by mental health professionals, which in turn led to increased diagnosing of children with ADHD and prescribing of psychotropic medication as treatment (Bokhari & Schneider, 2011; Figlio & Ladd, 2015; Hinshaw & Scheffler, 2014). In response, some states passed laws restricting or prohibiting the school personnel from recommending parents to get their children diagnosed, and these laws led to reductions in ADHD diagnoses among children (Fulton et al., 2015).

Importantly, this trend of increased use of psychotropic medication has not been confined only to students from lower-income families. It turns out that students from higher-income families are also more likely to use stimulant medication to improve their school performance, and this is especially true for higher-accountability schools (King et al., 2014). This serves as an additional reason for explaining why increased prescribing of psychotropic medications to children on SSI rolls reflects a wider psychotropic drug use in the age cohort (Boat & Wu, 2015; Romig, 2017).

Educational Factors and Criminal Justice Outcomes

The importance of school attendance and truancy prevention for reducing criminal justice involvement has been addressed in criminal justice research (Abeling-Judge, 2021; Cardwell et al., 2021; Jackson et al., 2022; Mazerolle et al., 2019; Rocque et al., 2017). There has also been a wealth of research by Gottfredson, (2001) and her colleagues on how school climate and structure contribute to delinquency and violence (Gottfredson et al., 2005; Gottfredson et al., 2014), how after-school programs play a role in “deviancy training” (Rorie et al., 2011), as well as evaluations of effectiveness of school-based crime prevention programs (Gottfredson et al., 2002). School climate has also been addressed by non-US-based researchers: for example, López and colleagues (2008) have demonstrated that classroom environment is an important factor in moderating Spanish adolescents’ aggression, and it is more pronounced for boys rather than girls. Most recently, Wang and colleagues (2020) estimated in their meta-analysis that classroom climate plays a role in decreasing aggressive behaviors of students.

At the same time, the role of school-based services for children with disabilities has not been studied as a specific factor in crime prevention. The current study addresses this gap and investigates whether school-based services to children with learning disabilities contribute to the declines in juvenile violence.

Self-control/Life-Course Perspective

One of the theoretical approaches providing an overarching explanation for the developmental process leading from childhood behavioral issues to adult criminality is Pratt’s (2016) self-control/life-course theory of criminal behavior. It draws upon extensive research evidence to integrate Gottfredson & Hirschi’s, (1990) general theory of crime, which posits low self-control as one of the most parsimonious and widely empirically supported explanations for delinquency and crime, with Moffitt’s, (1993) dual taxonomy, which explains the continuity of childhood behavioral and school troubles into adulthood among a small share of children with neuropsychological deficits compounding over the life course. Moffitt and her colleagues (2011) demonstrate convincingly the impact of childhood self-control on a variety of adult outcomes including adult crime, and Novak, (2022) shows the influence of neuropsychological deficits on the experience of exclusion at school.

The key concepts being investigated in the current study seem to be directly related both to Moffitt’s, (1993) life-course-persistent category whose troubles stem from their neuropsychological deficits and to the developmental perspective on self-control proposed by Pratt, (2016)³: since both interventions under study here—psychotropic medication use and school-based services to children with learning disabilities—reflect the goals of remediation of neuropsychological deficits in children and improvement of their executive functions.

Childhood Poverty and Juvenile Violence

The legislative processes associated with the key variables of the study—SSI eligibility expansion to children with mental health conditions other than intellectual disability and the NCLB/IDEA side effects of the increasing diagnosing of students with mental health conditions as learning-disabled—by design disproportionately affect youth in poverty. This has important implications for juvenile delinquency, particularly for violent offenses. Contrary to popular opinion and common sense, property offending is not related to income level, either at the individual or aggregate level (Elliott et al., 1985; Krivo & Peterson, 1996; Rekker et al., 2015; though see Jarjoura et al., 2002 demonstrating the pervasive effects of persistent childhood poverty on all kinds of youth offending). On the other hand, violent offending is closely linked with poverty, both at the individual level (Cunradi et al., 2000; Mok et al., 2018; Rekker et al., 2015) and various aggregate levels (Kaylen et al., 2017; Land et al., 1990; McCall et al., 2010; Messner et al., 2001; Parker & Pruitt, 2000; Tcherni, 2011).

Moreover, Marcotte & Markowitz, (2011) have found that the increased use of psychotropic drugs had an ameliorative effect on violent crime but not on property crime. It is possible that, when neuropsychological deficits are being remedied, impulsive aggression is affected the most rather than the general self-control that would have a more universal effect on all crime types. Thus, more research is needed to better understand the mechanisms of the link between remediation of neuropsychological deficits and decreased involvement in crimes of different types.

The interpretation of the poverty-violence link is also not straightforward. The mechanisms involved likely include genetic, neurodevelopmental, and educational/structural factors that are clustered among families in poverty (Fergusson et al., 2004; Galloway & Skardhamar, 2010; Richmond-Rakerd et al., 2020; Sariaslan et al., 2014; Wikström & Treiber, 2016). But the importance of this variable in predicting juvenile violence is hard to overestimate. Thus, controls for childhood poverty are included in the current study as one of the most salient factors for juvenile outcomes.

Poverty, Weapon Carrying, and Violence

One of the measures of juvenile violence employed in the current study is weapon carrying among public high school students, based on the Youth Risk Behavior Surveillance System (YRBSS)—data gathered on a regular basis by the U.S. Centers for Disease Control and Prevention (in the current study, data for years 2003, 2005, 2007, 2009, 2011, and 2013 are used). Weapon carrying has been clearly associated with aggressive and violent behaviors among youth (Docherty et al., 2019; Emmert et al., 2018) though its connection with family or neighborhood poverty is not clear (Docherty et al., 2019) and the trends in weapon carrying do not coincide with the declines in other types of youth violence (Perlus et al., 2014).

Thus, it is important to examine whether the independent variables being the focus of the current study—psychotropic medication prescribing and school-based services for children with disabilities—are associated at the state level over time with various types of juvenile violence: minor (fighting), moderate (weapon carrying), and severe (homicide offending), and which role child poverty rates play in this process.

Study Hypotheses

The research literature examined here demonstrates clear connections among increased psychotropic medicalization of children, school-based services to children with learning disabilities, and subsequent declines in juvenile delinquency/violence, with child poverty likely playing an important role in the process as one of the most consistent predictors of violent outcomes among youth. Thus, here are the research questions/hypotheses investigated in the current study at the aggregate level, through the analysis of changes in the US states over time:

• • Research Question/Hypothesis 1: Is there a relationship between increased diagnosing and psychotropic medication prescribing to children and adolescents on the one hand (measured through SSI rolls) and decreased juvenile aggression and violence on the other hand (measured through self-reports and official statistics), net of other important factors?

• • Research Question/Hypothesis 2: Is there a relationship between increased school-based services for children with learning disabilities and decreased juvenile aggression/violence, net of other important factors?

• • Research Question 3: Which one of these two factors—psychotropic medicalization of children or availability of school-based services—has more prominent associations with juvenile violence declines, net of child poverty?

Data and Methods

To test the three research questions of the study, a state-level panel data set has been assembled. It includes measures by year by state, for the years between 1990 and 2014 when data are available, for the following: juvenile delinquency and violence (dependent variables), mental health issues and psychotropic medication use among children and adolescents (independent variables), and relevant control variables associated with these two sets of measures at the state level (school accountability and sociodemographic variables). The general logic of the method is to exploit the differences in state trends in order to estimate the associations between the remediation of children’s mental health issues on the one hand (modeled through (a) psychotropic medication use, and (b) school-based services), and state-level changes in juvenile delinquency and violence on the other hand.

Dependent variables

• 1. Estimates of offending based on official data from Uniform Crime Reports (UCR)⁴:

  • a. crime rates by crime type for each state, produced by the FBI through weighting and data imputation, are used as a measure of crime in general, not specifically juvenile crime.

  • b. juvenile-to-adult arrest ratios by crime type, calculated based on the UCR-reported arrest data and weighted by the juvenile-to-adult population ratio.⁵ Even though the interpretation of these ratios is not straightforward, they provide a useful measure of changes over time and comparisons across crime types in terms of which types of crime can be more or less representative of juvenile offending patterns (essentially, the higher the ratio, the higher the share of juvenile arrests for the crime type, taking the share of juveniles in the population into account).

• 2. Estimated rate of juvenile homicide offending per 100,000 age-matched population was derived from Supplementary Homicide Report (SHR) data on solved homicides/known offenders: a proportion of offenders ages 12–24 among all offenders arrested for homicide was calculated and then weighted by the inverse of clearance rates (calculated by Hargrove, 2019) and by the population share of age-matched population.

• 3. A set of self-report-based measures of delinquency and violence from the Youth Risk Behavior Surveillance System (YRBSS) data. A percentage of public high school students (grades 9–12) who reported “having been in a physical fight at least once during the previous 12 months” is one of the most direct measures of juvenile delinquency/violence and the one that proved most useful for data analyses, even though it was only available for select years and states (between 2003 and 2013). Other YRBSS measures include alcohol use, marijuana use, and weapon carrying among public high school students. Additionally, a measure of firearm incidents at schools was used (available from the U.S. Department of Education).

Independent variables

• 1) Two measures were constructed based on data for supplemental security income (SSI) recipients under age 18 receiving federal benefits for mental impairments other than intellectual disability⁶ (henceforth, MIOTID): a) rate of such SSI recipients per 1000 aged-matched population, and b) percentage of children in poverty who are such SSI recipients. Data were provided by the Social Security Administration (SSA) in response to a Freedom of Information Request from the author and available for years 2002–2014.

• 2) Share of public-school students ages 3–21 with disabilities, i.e., served under the Individuals with Disabilities Education Act (IDEA), as percentage of enrollment are provided by the U.S. Department of Education and available by state for years 1990, 2000, and 2005–2014. Since changes in child disability rates are mostly driven by mental health issues, or learning disabilities (Pastor & Reuben, 2008; Wagner et al., 2005), the variations in percentage of students with disabilities can be used as a proxy for children with mental health issues (learning disabilities) receiving additional school-based services. The reason these variations reflect specifically learning disabilities due to mental health is because the proportion of students with physical disabilities does not fluctuate much by state or by year. Also, unlike the measures for SSI recipients with MIOTID, the school-based disability variable has an added advantage of applying to all children, not just to children in poverty.

See Fig. 1 visualizing the changes over time in the two main independent variables used in this study. Other independent variables potentially measuring the processes of interest were considered, but their inclusion was limited by the fact that these measures were available for very fewer years and states.⁷

Control variables

• 1) School accountability measures introduced by states before the No Child Left Behind (NCLB) Act of 2001 account for the pressure put on schools in some states to improve their average scores by pushing certain low-performing categories of students into the disability designation and excluding their scores from the school averages.

• 2) Relatedly, data on psychotropic medication laws are included—these laws were introduced by several states to make it more difficult for school personnel to suggest to parents that their children may need to be psychiatrically diagnosed and medicated and, alternatively, easier for the parents to refuse psychotropic medications for their kids.

• 3) Sociodemographic controls include several variables that previous research literature on juvenile delinquency and violence suggests are consistently important in impacting crime/delinquency outcomes: percentage of children in poverty, divorce rate, percentage of African Americans or Blacks, and percent urban population.⁸

The descriptive statistics for the variables listed above are presented in Table 1. Also, see Fig. 2 illustrating changes in child poverty and the estimated rate of juvenile homicide offending.

Table 1 Descriptive statistics for key variables included into the models, for selected years 1990–2014

Fig. 2

Changes over time for US states in the share of children under age 18 who live in families with incomes below the federal poverty level (Census data) and the estimated rate of homicide offending among ages 12–24 (based on SHR data). Note: *per 100,000 age-matched population, adjusted for clearance rate

Analytical Procedures

To ensure that the dependent variables are approximately normally distributed to avoid violating the assumptions for linear regression analyses and biasing the estimates, logarithmic transformation was applied to the dependent variables whose distributions had a positive skew (long right-tale and/or excessive zeros): the estimated rate of juvenile homicide offending has the most pronounced skew (see Table 1). In addition, since the independent variables refer to a wider range of children’s ages (ages 0–18 for SSI recipients and ages 3–21 for students with disabilities) while the dependent variables refer to adolescents and young adults (ages 14–18 for having been in a fight, ages 12–24 for estimated rate of homicide offending), lagged effects of independent variables were examined. For this purpose, for each dependent variable of interest, several lead/lag variables were created to examine the impact of current-year predictors on outcomes 3 years later, 5 years later, and—whenever applicable and allowed by the data—also 7 and 9 years later.

A series of data analyses for time-series cross-sectional data was performed to estimate the effects of time-varying and time-invariant factors on juvenile delinquency and violence. Initially, fixed effects (FE) panel data analyses were performed as a preferred method given the structure of the data to estimate the relevant models (Marvell & Moody, 2008; Wooldridge, 2010). FE analyses allow estimating how the changes over time in certain variables within the states impact the outcomes, but they do not allow estimating how the average differences among states on a certain variable affect the average outcome for that state. Moreover, the results of tests evaluating the model fit (Hausman test) showed that random effects (RE) models were more appropriate.⁹ The advantage of RE models is that they are more generalizable and flexible, as well as allow for estimation of time-invariant controls (Bell & Jones, 2015; Greene, 2008). The downside of RE models is a higher likelihood of omitted variable bias in the model if the model is misspecified (Bell & Jones, 2015). To guard against this possibility, theoretical considerations and an in-depth, research-based understanding of the important determinants of juvenile delinquency and violence, as well as knowledge of each subject area involved in the current project—from the key correlates of juvenile delinquency to policy issues on psychotropic drug prescribing for children to state policies on school accountability to mental health survey-based research—guided model specification.

Ultimately, a mixed-effects linear modeling approach was adopted, combining the fixed effects of key independent variables and time periods (since their influence is considered to be independent of the observed predictor variables), along with the random intercept by state. One of the most important features of the analyses performed in this study is the explicit modeling of the time effects, since time trends are very powerful in affecting delinquent and criminal outcomes. Moreover, any omission of the time effects from the models leads to the reversal of regression coefficients for most independent variables and thus would lead to serious model misspecification and erroneous conclusions about the impact of the key independent variables. In addition, various alternative models were estimated as supplementary checks and balances as well, to ensure the results are robust and hold to various reasonable model specifications.

Thus, the mixed-effects linear regression models used in the study include fixed effects of time and random intercepts by state. They are represented by the following formula:

$${Y}_{it}={\beta }_{0}+{\beta }_{1}{X\mathrm{l}}_{\mathrm{i}}+{\beta }_{2}{X2}_{\mathrm{it}}+\left[{\gamma }_{2}{T}_{2}+\dots +{\gamma }_{\mathrm{n}}{T}_{\mathrm{n}}\right]+{\alpha }_{\mathrm{i}}+{e}_{\mathrm{it}}$$

where Y_it is the dependent variable for each state i and time t, β₀ is a constant, β₁ is the coefficient for a time-constant independent variable X1 that varies by state (state-mean-centered), β₂ is the coefficient for a time-varying independent variable X2 that varies by time and state, γ_2…γ_n are the coefficients for the binary time regressors, T_2…T are the time binary variables for t − 1 time periods, α_i is a state-specific random intercept, and e_it is the error term.

All data analyses were performed using Stata 17.

Results

Since using the general crime rates and juvenile-to-adult arrest ratios as the dependent variables did not produce any consistent or significant results, we will focus on two main sets of outcomes that also have the advantage of most closely reflecting juvenile violence (see Table 2):

• a) measures related to public high school students having been in a fight or carried a weapon (reflects minor to moderate self-reported violence), and

• b) estimated homicide offending rate among 12–24-year-olds (reflects extreme violence as reported in official data).

Table 2 Results from mixed-effects models with panel data at the state level, for selected years 2000–2014

Surprisingly, the independent variables related to children-SSI recipients with mental impairments other than intellectual disability (MIOTID) mostly yielded results contrary to the initial expectations: the increases in the SSI rolls of children with mental health conditions as a share of age-matched population are associated with relatively small in magnitude but statistically significant increases in violent outcomes.¹⁰ When the SSI recipients with MIOTID as a percentage of children in poverty are included into the models instead, none of the regression coefficients for this variable are statistically significant. Thus, the answer to the first research question—whether the increased prescribing of psychotropic medications to children and adolescents is associated with juvenile violence declines at the aggregate level—seems to be “not clear” or “likely not”.

On the other hand, the percentage of students with disabilities served under the Individuals with Disabilities Education Act (IDEA) has clear and strong negative relationship with minor and moderate juvenile violence: as seen in Table 2, each percent increase in students with disabilities as a share of public-school enrollment (which also means increases in IDEA-mandated specialized services to these students) is associated with about a half-percent decrease in the minor violent outcome of fighting (b =  − 0.42, p < 0.001) and the moderate violent outcome of weapon carrying (b =  − 0.53, p < 0.01). The effect of the disability services variable (likely reflecting IDEA-mandated school-based services to students with learning disabilities) is consistent and independent of the effects of poverty and time trends in the models. It mainly applies to less serious violent outcomes (having been in a physical fight and, to a lesser degree, weapon carrying, though only to fighting if estimated over time), but not to extreme violence (estimated homicide rates among ages 12–24), where the regression coefficient for the contemporaneous effect is − 0.05 and barely significant.

The next important and expected predictor that emerges from the analyses is the percentage of children in poverty (as a share of age-matched population), which is almost as powerful a determinant of the violent outcomes as the services to students with disabilities: for each percent increase in child poverty, there is about a third of a percent increase in students’ physical fighting (b = 0.28, p < 0.001). One exception is the contemporaneous effect of child poverty on the log-transformed estimate of homicide offending—it is weak (b = 0.02) and not statistically significant. At the same time, strikingly, the effect of the child poverty variable does not diminish over time: with each lag (3, 5, and 7 years later), it impacts both the minor-to-moderate juvenile violence (fighting, weapon carrying) and extreme youth violence (homicide), as the regression coefficients in Table 2 show. In contrast, the disability services variable exhibits contemporaneous and lagged effects only on the minor form of violence (fighting), whereas child poverty has an increasingly strong impact on all forms of violence over time.¹¹

It is important to note that, among the only other juvenile-specific outcome measures—alcohol use and marijuana use, both in general and on school property, there are no clear and consistent effects of either child poverty or disability variables.¹² The same situation applies to the control variables included into the models: for the time-stable percent Black and urban population, as well as for the time-varying divorce rate variable at the state level—the coefficients are not significant or inconsistent. Neither school accountability laws nor psychotropic medication laws were significant in any of the models (most likely, because the effects of these laws are already essentially reflected in the school services to children with disabilities and in the mental-health SSI children recipients variables). The models were also re-estimated with and without Washington, DC, as well as using additional variables suggested by anonymous reviewers, such as high school dropout rates,¹³ but they did not substantially affect the findings. Interstate mobility was considered as well but ruled out as an unlikely contributor due to the increasingly low interstate mobility rates, hovering around 2% in the 1990–2014 time period (Foster, 2017). The results of additional analyses are available from the author upon request.

Discussion

The current study estimates the relationship between mental-health-related factors, such as psychotropic medication use and school-based services for children with learning disabilities, and the declines in juvenile delinquency and violence in the USA over the last few decades at the aggregate level. For this purpose, state-level panel dataset for 1990–2014 (mostly focusing on the 2000–2014 period due to data availability constraints) is analyzed using mixed linear regression modeling techniques. Three important main findings emerged as a result, with implications going beyond the criminal justice system:

• 1) There are no clear associations between medicating children or adolescents with psychotropic drugs (estimated through the changes in supplemental security income rolls of children with diagnosed mental health conditions) and juvenile violent outcomes of various levels of severity.

• 2) Increases in school-based IDEA-mandated services to children with disabilities at the state level are clearly associated with the declines in the most common forms of juvenile violence: minor violence (fighting) and moderate violence (weapon carrying).

• 3) Child poverty is the strongest correlate of juvenile violence in its various forms, from minor (fighting) to moderate (weapon carrying) to the most severe types (homicide offending), especially in terms of the delayed consequences of child poverty: decreases in child poverty at the state level are associated with the declines in juvenile violence not only contemporaneously, but also many years into the future.

Limitations

The current study has clear limitations stemming from several constraints inherent in the data. First, the nature of aggregate data merits caution in drawing any conclusions at the individual level (potential for ecological fallacy), which limits our ability to test cause-and-effect relationships. At the same time, as discussed above, aggregate data analyses have important advantages over the use of individual-level data (Guthrie & Sheppard, 2001; Schwartz, 1994; Sideridis et al., 2018)—they reveal some of the higher-level processes happening in the USA. Second, even though prior research confirms that changes in SSI recipient rolls reflect the levels of psychotropic medication prescribing and consumption among children and adolescents in general (Boat & Wu, 2015; Romig, 2017), this measure may still perform as subpar at the state level in more nuanced analyses due to its high correlation with child poverty levels.

Third, other important factors may be closely associated with the variables in the analyses, which can complicate the exact interpretation of the likely influences: for example, the increased percentage of public-school students with disabilities receiving IDEA-mandated services may reflect higher levels of inclusivity and general tolerance due to the integration of children with learning disabilities into regular classrooms (Kirby, 2017; Rivera & McKeithan, 2021), which may be the most consequential aspect of the school-based services in terms of preventing juvenile aggression and violence. In fact, this can be construed as part of the cultural ‘civilizing process’—a concept developed by Norbert Elias, (1978) and popularized (while also being criticized for its vagueness) by Eisner, (2003, 2008)—as possibly contributing to the long-term trends in violence declines. Alternatively, it could be that these school-based services increase the likelihood of school graduation for those children who would have otherwise dropped out, and that is the actual reason these services have an impact on subsequent decreases juvenile violence. Determining the significance of these related factors falls outside the scope of the current analysis. Thus, future research should clarify these and other related research questions that will help us better understand the underlying processes and potential causal links.

Since no clear answers have been received in the current study regarding the benefits of medicalization of children and adolescents with psychotropic drugs in terms of its association with decreased juvenile delinquency and violence or crime in general, further research is needed to understand these links. The current study findings imply that either aggregate data is not the best method to answer this specific research question or the operationalization of this variable in the current study is insufficient for answering it. In a way, these problems are similar to the ones emerging in research on lead and crime that has been studied both at the aggregate level (Boutwell et al., 2017; Nevin, 2000, 2007; Reyes, 2007; Stretesky & Lynch, 2001) and at the individual level (Aizer & Currie, 2019; Dietrich et al., 2001; Reuben et al.,; 2020). Important progress has been made in better understanding the lead-crime relationship from research at both of these levels, despite some contradictory evidence (Lauritsen et al., 2016; McCall & Land, 2004).

Implications and Directions of Future Research

It is important to consider the implications of these findings for the developmental self-control/life-course perspective (Pratt, 2016), as well as for its component theories of low self-control (Gottfredson & Hirschi, 1990) and developmental taxonomy (Moffitt, 1993). The remediation of children’s neuropsychological deficits, happening both through medicalization and school-based services, has differential associations with juvenile violent outcomes, which is possibly a result of limitations in the medicalization measure being highly correlated with child poverty. The role of child poverty and the mechanisms of its influence on juvenile violence also need to be further investigated and clarified when considering possible routes of remediation of neuropsychological deficits and the potential of such remediation in curbing aggressive behavior and subsequent violence.

The findings of this study have important implications for policy and practice as well, not only in the USA but possibly in other countries since the processes underlying crime trends and patterns are likely universal. The role of school-based services for children with learning disabilities as a potential tool for violence prevention emerges from the current study. This finding is in line with a large body of research showing the importance of educational processes and factors for delinquency (Abeling-Judge, 2021; Gottfredson et al., 2014; Jackson et al., 2022; Liu et al., 2021; Novak, 2018; Rocque et al., 2017; Wertz et al., 2018). At the same time, specifically school-based services to children with learning disabilities is a novel factor that has not been studied in the research literature so far, and thus its potential contributions to crime declines need to be investigated further. Though there are studies examining the impact of truancy on delinquency (Gerth, 2022; Rocque et al., 2017), evaluating truancy prevention in curbing juvenile delinquency (Cardwell et al., 2021; Mazerolle et al., 2019), as well as studies on the adverse processes in schools contributing to the school-to-prison pipeline (Cuellar & Markowitz, 2015; Gottfredson et al., 2014; Jacobsen, 2020; Novak, 2018; Pesta, 2018), we know very little about the potential beneficial criminal justice implications of school-based services to vulnerable students. It follows from the results of the current study that school-based services to children with learning disabilities, as well as various related processes and their changes over time in US schools, are very likely an important factor contributing to the declines in juvenile delinquency and violence.

Additionally, the importance of reducing child poverty as clearly tied to reductions in juvenile violent outcomes has been supported by the current study, echoing prior research findings (Jarjoura et al., 2002; Rekker et al., 2015). Moreover, the study makes an important contribution to this literature and its significant policy implications by specifically showing both contemporaneous and subsequent, long-lasting consequences of child poverty for juvenile violent outcomes. Thus, this “ripple effects” aspect of child poverty should be further studied and heeded in policy discussions. This line of research can be invaluable in improving our understanding of what drives trends in juvenile crime and violence. In addition, the practical implications of this research would be relevant for every area of applied criminal justice concerning juvenile offenders and at-risk youths—from developmental crime prevention and early intervention to corrections and reentry—as well as research in education, developmental psychology and psychiatry, and public policy.