Drawing the Line

Abstract

Every state maintains some mechanism by which youths can be tried as adults in criminal courts. While scholars have long debated the inherent benefits or detriments of prosecuting youths as adults, empirical studies of actual outcomes have provided mixed findings and have been limited by problems of selection bias and jurisdictional differences in processing. The current research aims to further inform this literature by capitalizing on a policy change in Connecticut that raised the age of criminal responsibility from 16 to 17 on January 1, 2010, creating a natural experiment to assess the recidivism differences for youths based upon the system of processing: juvenile versus adult court. Findings from a 2-year follow-up reveal that 16-year-olds processed in juvenile courts had substantially reduced rates of recidivism with odds of rearrest that were between .462 and .630 less than for 16-year-olds processed in adult courts dependent on model specification.

Keywords

juvenile justice waiver/transfer age of criminal responsibility recidivism

The last quarter of the 20th century has come to be known as the “get-tough” era of juvenile justice in the United States. A real increase in youth violence in American cities (Sickmund & Snyder, 1999) coupled with media portrayal of these youths as a new breed of “super predator” put enormous pressure on policy makers and law enforcement officers to get tough on adolescent offenders (Butts & Mitchell, 2000; Cook & Laub, 2002). The response was a national movement to change both the purpose and the processing of the juvenile justice system. One of the most notable changes was a proliferation in mechanisms to transfer youths to adult courts for what was intended to be a greater punishment (Feld, 1984; Zimring, 1991). From 1992 to 1995 alone, 40 states and the District of Columbia modified legislation for trying juveniles as adults (Griffin, Torbet, & Szymanski, 1998). These laws lowered age thresholds for processing youths as adults—in some places to as young as 12 years of age—designated large categories of offenses for transfer, and shifted power in transfer decisions from judges to legislatures (statutory exclusion laws) and prosecutors (direct file provisions; Feld, 1999; Torbet et al., 1996).

The consequences of these laws on the number of juveniles being processed in adult courts were substantial. Research indicates these laws resulted in significant increases in the number of youths penetrating the adult criminal justice system (Feld, 2000; Torbet et al., 1996). Yet, these policies were based on no sound empirical evidence that processing youths as adults would indeed increase punishment or ultimately reduce youth violence. Rather, existing literature at the time would have suggested the opposite—that youths processed in adult courts go on to become more criminal than youths process in the more rehabilitative juvenile court (Bishop, Frazier, Lanza-Kaduce, & Winner, 1996; Fagan, 1991, 1996; Goodstein & Sontheimer, 1987; McNulty, 1996; Podkopacz, 1996; Winner, Lanza-Kaduce, Bishop, & Frazier, 1997).

Interestingly, within the past two decades, many states have begun to reconsider their transfer policies and are reexamining laws governing the criminality of youths. As predicted by Bernard and Kurlychek (2010) in the “Cycle of Juvenile Justice,” the pendulum of juvenile justice seems to be shifting away from the get-tough philosophy and again back to a more lenient treatment of youths. Unfortunately, this shift back to leniency may be as empirically uninformed as was the get-tough movement of the late 20th century.

Part of the reason for such uninformed policy choices is the tendency of legislators and government officials to respond to perceived public sentiment over science. Another reason is the difficulty in determining the true impact of processing youths in the juvenile justice system versus the adult justice system. The difficulty emerges from the fact that most states maintain policies that select only the most severe and chronic youths for transfer, which introduces problems of selection bias into attempts to empirically evaluate such policies. Social scientists have employed various methods to try to overcome these biases including comparing youths across state borders to capitalize on different state laws/policies (Fagan, 1991, 1996) and using propensity score matching techniques to try to match youths processed in the adult system to youths retained in the juvenile system who are similar in important legal characteristics such as prior criminal history and seriousness of the offense (Bishop, 2006). The current research further advances knowledge in this realm by capitalizing on a natural experiment in which one state changed the age of maximum jurisdiction for the juvenile justice system from 15 to 16 on January 1, 2010, thus impacting an entire population of youths within one jurisdiction. That is, prior to 2010, all 16-year-olds in Connecticut had been processed in adult court regardless of their criminal record. The “Raise the Age” legislation in Connecticut thus provides a natural experimental setting in which we can assess within one jurisdiction (Connecticut) the direct impact on recidivism of processing an entire population of 16-year-olds in the more rehabilitative juvenile justice system (post January 1, 2010) as compared to processing 16-year-olds in the more deterrence- and retributive-based adult criminal justice system (prior to January 1, 2010).¹ To set the stage for the current study, we next discuss why one would expect different outcomes based upon the system of processing.

Theoretical and Scientific Roots of Juvenile Processing

The Developing Juvenile

The original impetus for bifurcating the treatment of adults and juveniles under law was rooted in the idea that youths are still developing and thus are highly susceptible to influences—both positive and negative—and therefore need different treatment than adult offenders. To best help adolescents displaying early signs of criminality develop into law-abiding and productive adults, it was reasoned these vulnerable youths should be kept out of adult jails and prisons where they would learn to be more criminal (Bernard & Kurlychek, 2010; Platt, 1977). Instead, the youths were to be provided more age-appropriate treatment and services to promote positive development. In other words, where adult methods of crime control embodied rationales based in deterrence, retribution, and incapacitation, the response to adolescent crime focused on rehabilitation (Bernard & Kurlychek, 2010; Fagan, 1991; Feld, 1987, 1999).

In addition to emphasizing rehabilitation over punishment, the juvenile justice system was designed to keep youths away from any potential negative effects resulting from mixing youths with older, more culpable adult offenders. This rationale, which is in line with differential association theory (Sutherland, 1942), was eventually formalized into national legislation in the Juvenile Justice and Delinquency Prevention Act of 1974 that mandates juveniles be kept out of sight and sound contact with adult offenders and not be placed in adult jails or lockups. However, the act does not require a standard national definition of who qualifies as a “juvenile.” Thus, a 16-year-old in one state can be defined as a juvenile and the state required to abide by such regulations, yet in another state, a 16-year-old can be defined as an adult and therefore legally not afforded the same protections.

In addition to simply shielding youths from the negative influence of older more hardened criminals, there is widespread belief that adult system processing is simply not appropriate for youths based on their mental maturity and cognitive abilities. Psychologists have long postured that adolescents have not achieved the same levels of cognition and reasoning as adults and, by nature, are not as culpable for their acts (Hall, 1904; Keating, 1990; Klaczynski, 1993; Moshman, 1998; Piaget, 1964; Steinberg & Cauffman, 1996). Thus, in addition to the argument for separation from hardened adult criminals to avoid further criminalization, the juvenile justice system has historically striven to provide both a process and a consequence more developmentally appropriate to the cognitive and emotional maturity of youths.

More recently, this science has crossed the bounds from psychology to biology, garnering greater empirical support. On the biological front, research on adolescent brain development finds that the left frontal cortex of the brain does not fully develop until age 21 for females and age 24 for males (Lenroot & Geidd, 2006). This portion of the brain is responsible for rational decision-making and impulse control. In lieu of a fully developed left frontal cortex, research suggests adolescents’ decision-making and judgment are controlled by the amygdala or the part of the brain central to emotion and impulse (Baird & Fugelsang, 2004). These studies were the first to suggest that there are clear biological reasons youths may not be as responsible or culpable for their actions as adults.

Further research that directly addresses adolescent decision-making capabilities supports this notion, finding that although the basic understanding of principals may be intact by age 15 or 16, youths do not have the psychosocial maturity to make reasoned decisions in actual social situations (Steinberg & Cauffman, 1999). Specifically, their decision-making is impacted by a short temporal focus, making them less likely to understand long-term consequences such as lengthy prison sentences. Additionally, in social situations, they are more likely to act on impulse than reason and to be highly influenced by their peers and social circumstances (Steinberg & Cauffman, 1999).

Thus, while the rhetoric of the get-tough movement was largely embraced at face value in previous decades, theory and growing scientific evidence suggests that adult system processing may indeed be developmentally inappropriate for youths and, in the end, do more harm than good.

Stigma and Adoption of Criminal Values

While the philosophies discussed above suggest that a separate juvenile justice system might provide more developmentally appropriate responses to youths, it is also important to understand the theories that have influenced and directed the practice of juvenile justice such as labeling (Lemert, 1951), social bond (Hirschi, 1969; Sampson & Laub, 1993), and differential association (Sutherland, 1937) theories. For example, as first proposed by Lemert in 1951, contact with, and penetration of, the criminal justice system may lead to adaptation of a delinquent identity and a subsequent increase rather than decrease in delinquency. Referred to as secondary deviance, this deviance that comes after a shift in identity is proposed to be a defense mechanisms through which the youth responds to the “overt and covert problems created by the consequent societal reaction” (Lemert, 1951, p. 76).

One overt problem is that the stigma created by having a criminal record can lead others to react differently and negatively to the individual (Becker, 1963; Goffman, 1963). Thus, juvenile justice practices often involve diversion either from arrest or from further system penetration as well as keeping juvenile court records confidential. While research suggests that any system contact and arrest, even juvenile arrests, can lead to labeling and the development of secondary deviance (Huizinga, Esbensen, & Weiher, 1996; Huizinga & Henry, 2008; Liberman, Kirk, & Kim, 2014), it is also posited that the further one penetrates the system and the more public the criminal record, the greater the resulting stigma.

For example, the stigma of a juvenile or criminal record can in turn impact social bonds. As explained by Link (1982), social exclusion can come through the actual exclusion of the individual by others or the individual’s own withdrawal in anticipation of exclusion. Sampson and Laub (1993) further show how this process can work throughout the life course, weakening bonds to family and friends and limiting other life opportunities such as prospects for employment and marriage. Excluded by prosocial others, the delinquent may then be forced to associate only with deviant others, which further confirms the deviant identity and supports the adoption of definitions favorable to the violation of the law (Paternoster & Iovanni, 1989; Smith & Brame, 1994). Recent studies support the notion that justice system involvement may increase delinquency through the stigmatization process that decreases positive social bonds and increases association with deviant others (Kirk & Sampson, 2013; Lopes et al., 2012).

Thus, because the juvenile justice system is generally designed to provide a response that is more developmentally appropriate and has a greater emphasis on diversion and confidentiality of records than the adult system, we hypothesize that youths processed in the juvenile justice system should have more positive outcomes than youths processed in adult courts in relation to later recidivism. However, as further presented in the following sections, the findings to date are still mixed. Moreover, the juvenile justice system’s historical tendency toward diversion and less formal processing makes assessments of “recidivism” tricky as findings can be an intricate mix of actual youth behavior and system behavior.

Existing Studies of Jurisdictional Boundaries

The distinction in purpose and practice between juvenile and adult court justice systems has led many scholars to study this phenomenon by comparing the later recidivism outcomes of youths processed in juvenile and adult courts. Most early studies seemed to suggest that juveniles punished in adult courts did indeed fare worse in terms of likelihood of future criminality (Bishop et al., 1996; Fagan, 1991, 1996; Goodstein & Sontheimer, 1987; McNulty, 1996; Podkopacz & Feld, 1995; Winner et al., 1997).

However, on the surface, it is easy to argue that this may simply be because it is a group of worse youths (e.g., more serious and chronic offenders) that are targeted for adult court processing. Studies have employed various techniques to attempt to overcome this barrier. Fagan (1991), for example, used a cross-jurisdictional approach comparing violent youth offenders aged 15–16 adjudicated originally in juvenile court in New Jersey to those adjudicated originally in adult court in New York. Fagan (1991) found lower recidivism rates for youths processed in juvenile court. Specifically, a lower percentage of youths processed in the juvenile court were rearrested. Additionally, for those youths processed in the juvenile court who did experience a new arrest, the average time to rearrest was longer than for youths processed in the adult court. Regarding the finding that youths treated in adult court had higher crime rates, Fagan (1991) suggests treating juvenile delinquency as crime might itself precipitate future criminality. In support of this notion, Fagan (1996) also found that those youths who actually experienced incarceration in the adult system had even greater likelihood of later recidivism and violence.

Researchers exploring Florida’s transfer/waiver policy have taken a different approach using matched samples of youths across the systems to try and determine the impact of transfer. The earliest study by Bishop, Frazier, Lanza-Kaduce, and Winner (1996) found that juveniles retained in juvenile court recidivated less juveniles transferred to adult court in terms of prevalence and incidence. Additionally, the adult sample was arrested more quickly on average than the juvenile sample. A related study (Winner et al., 1997) also found that those juveniles waived to adult court recidivated more often than those retained in juvenile court within a 2-year follow-up period.

A follow-up study by Lanza-Kaduce, Frazier, Lane, and Bishop (2002) capitalized on variation in the discretionary decision by the judge to waive a youth to adult court and/or the prosecutor to file a case in juvenile versus adult court to create a matched sample of youths across the two systems. The researchers used propensity score matching techniques to conduct a “nearest neighbor match” which matches each individual juvenile transferred to adult court with a similar juvenile retained in the juvenile system. This research found about 35% of the youths processed in juvenile court to recidivate after age 18 as compared to 50% of the youths processed in adult court. This research also discovered that when youths in the matched samples did recidivate during the follow-up period, the youths processed in adult court were rearrested more quickly and were more likely to be rearrested for a serious felony.

Several more recent studies using econometric methods of analyses, however, have found null effects, or even lower rates of recidivism, for youths processed in adult courts bringing back into question whether these prior studies have adequately controlled for issues of selection bias. For example, Loughran et al. (2010) found overall null effects when comparing reoffending of a matched group of serious juvenile offenders in Pennsylvania and Arizona adjudicated in either juvenile court or adult court via transfer. Lee and McCrary (2009), using a regression discontinuity design, found adolescents processed in the adult system in Florida immediately after aging into the legal age of majority (18 years) had slightly lower rearrest rates than those processed in the juvenile system immediately before turning 18. Hansen and Waddell (2014) reported a similar finding in their study utilizing regression discontinuity to examine reoffending effects of juvenile transfer in Oregon. Loeffler and Grunwald (2015), also employing a regression discontinuity design, found processing juveniles as adults resulted in 3–5% reduction in the likelihood of recidivism when they compared felony drug offenders processed in juvenile and adult courts in Chicago, IL, over the course of 4 years.

What is one then to make of such mixed findings? Zane, Welsh, and Mears (2016) attempt to answer this question by conducting a meta-analysis of the most methodologically sound studies of juvenile transfer. Using meta-analytic techniques to combine the effect sizes across studies, these researchers conclude that transfer to adult court has a small but statistically insignificant impact on recidivism. Perhaps, more importantly, however, the authors conclude that this null effect appears to be tied to heterogeneous effects across studies with outcomes highly dependent upon the exact transfer policy influencing the study sample. That is, transfer exerted “different effects, whether because of the type of transfer, offender, offenses, or sanctions that youth receive” (p. 916).

Limitations of Existing Studies

While these mixed findings may seem confusing, this last point regarding heterogeneous treatment effects is important as each study dealt with different populations of youths based on state-specific transfer policies. Put simply, laws and policies guiding juvenile court jurisdiction operate on a statewide basis. Therefore, when comparing youths within one state processed as either juveniles or adults, the samples are necessarily segregated on the basis of the law and thus are not equivalent from the start. For example, if the law requires youths over the age of 15 charged with a violent felony to be processed in adult court, then there should be no directly comparable youths retained in the juvenile court. Even when the law is less straightforward and places a discretionary waiver decision in the hands of the juvenile court judge, the judge will consider such elements as offense severity, past delinquent behaviors, and amenability to treatment, thus resulting in a sample of more “serious” youths being transferred to adult court than being retained in juvenile court. Some studies have attempted to overcome this selection bias by looking across state boundaries (Fagan, 1991). However, herein enters the problem of jurisdictional differences in arrest practices, definitions of crime, and criminal justice procedures that make all but the most general of statements suspicious. For example, Fagan was comparing youth in New Jersey to youth in New York. However, New York has much more liberal sealing policies, particularly for youths, making only the most serious offenses/offenders visible through administrative data obtained from New York. Other studies, especially those cited above as having used more sophisticated methods of analyses (Hansen & Waddell, 2014; Lee & McCrary, 2009; Loeffler & Grunwald, 2015), are likely able to overcome problems associated with selection bias and exogeneity but also have relied upon artificial and synthetic control groups that present another potential source of unobserved error.

Current Study

Research Context and Questions

Our primary research question is quite simple: “Does processing youths in juvenile court versus adult court reduce later recidivism outcomes?” The research to date suggests that it should based on potential criminogenic and stigmatizing effects of adult court processing and the developmental differences that make juvenile court practices more appropriate. However, the existing empirical evidence is mixed. The current study is able to overcome many of the difficulties in selection bias faced by previous empirical research in this area by capitalizing on a natural experiment resulting from the change in law in the state of Connecticut that raised the maximum age of juvenile court jurisdiction from 15 to 16 on January 1, 2010.² Because the change impacts a population of youths rather than a select chronic and serious sample selected for transfer because of their serious criminality, and because it occurs within one jurisdiction, changes in the recidivism rates of these youths can be more directly related to the change in the court of jurisdiction than characteristics of the youths processed. Although in a natural experimental setting, one would expect the populations pre- and posttreatment to be similar, we thoroughly examined our samples, found slight differences, and incorporate these differences into our statistical models and sensitivity analyses as described below.

Data and Sample

To obtain data for this study, we forged a relationship with the Judicial Branch of Connecticut state government, which allowed us to enter into a data use agreement to explore the immediate and 2-year outcomes of all 16-years-olds processed as adults in 2009 (the year prior to the change) and in 2010 (the year immediately following the change). The result was the collection of four data sets as follows: File 1, client pool that included all 16-year-olds arrested in the state of Connecticut in 2009 and 2010; File 2, instant offense file that included all charges associated with the arrest, arrest dates, and police municipality of the arrest; File 3, criminal history that provided all prior arrests and convictions of the youth; and File 4, recidivism file that provided a 2-year recidivism follow-up for each youth. The data use agreement did not allow us to collect certain identifying information such as name and exact date of birth; however, youth were identified through a crosswalk procedure in which we were provided a scrambled unique identifier for each youth that allowed us to match across the four files.

The original sample from Connecticut included all 16-year-olds arrested in the year 2009 (4,050), who were defined as adults at that time, and all 16-year-olds arrested in the year 2010 (3,237), who were defined as juveniles.³ The first file noted above was a demographic “client pool” file that provided basic characteristics such as gender, race, and cohort year. The second file constructed was an instant charge file that included all arrest charges for the first time an individual was arrested in either 2009 or 2010, all details on the arrest charges, and the conviction and disposition/sentence outcome. The third file was a complete prior arrest and conviction history for each individual. The fourth file then followed each individual for 2 years (730 days) from the time of the first arrest date and conviction date recording all future arrests and convictions.

Upon merging the files, we realized the 134 youths in the 2010 file had previously been arrested and processed as adults in 2009. Thus, we decided to eliminate the youths from the 2010 file but maintain them in the 2009 sample then coding the 2010 arrest as a recidivism event. Also, because the theories noted above suggest that further system penetration beyond arrest might lead to further criminality regardless of court type, we create a second sample of only those youths who were convicted of the original offense. We refer to this smaller sample as our “conviction sample,” which has 1,360 individuals in the 2009 cohort and 901 individuals in the 2010 cohort. We conduct all analysis first on the all arrest sample and second on the smaller all conviction sample.

Dependent Variable

The key dependent variable of interest for this study is recidivism within 2 years of the initial arrest. Each youth in the all arrest sample was tracked for 2 years from the original event that led to his or her entrance into the sample and for the all conviction sample, for 2 years from the time of the conviction. The 2-year follow-up period was selected for several reasons including similar time periods being used in many previous studies of juvenile transfer and the timing of Phase II of the Raise the Age legislation that subsequently altered the status of 17-year-olds to juveniles, which could have confounded our results. Based on the dichotomous nature of the dependent variable (1 = rearrest, 0 = no new arrest), we utilize multivariate logistic regression controlling for any variables that differed significantly between the two populations of youths (see Control Variable section). As the data were provided to us by police jurisdiction, we utilized fixed effect models to account for any differences in processing across jurisdictions as well.

Independent Variable

Our independent variable of interest is simply whether the youth was processed in juvenile or adult court, which in this instance coincides with the timing of their arrest due to the nature of the policy change. All original arrests of 16-year-olds in 2009 were coded as “adult” and all original arrest of 16-year-olds in 2010 were coded as juvenile.

Control Variables

Although this is a natural experiment, it is important to assess whether there are any unexpected differences between the samples on variables known to influence court outcomes and recidivism. Thus, we created a battery of variables and tested for equality across the samples and controlled for those variables that were different, even if slightly, between the two samples. First, regarding the instant offense, we created two separate files: an “all arrest” and an “all conviction” sample, the latter being a subset of the first.⁴ Then, the most serious charge per arrest or conviction cycle was selected to code the seriousness and type of the current offense.⁵ Characteristics of the instant offense were coded by seriousness (Felony A through Misdemeanor B) and by type of crime (person, property, drug, sex crime, and other).⁶ Prior offenses were included only if the arrest led to conviction and the most serious conviction per court cycle was coded to determine type and seriousness of prior offenses. Inclusion of prior offense history in statistical models was explored several ways, with each having similar results. For the current analyses, we present total number of prior convictions as well as count variables for total number of prior felonies and total number of prior misdemeanors to capture seriousness.

Gender was simply coded as 1 for male and 0 for female. Race was coded as Black, White, Hispanic, Asian/American Indian, and missing.

Tables 1 and 2 provides bivariate comparisons on all variables for the 2009 and 2010 cohorts. As Table 1 shows, we did find several minor differences in this initial arrest sample such as the 2010 “juvenile cohort” being slightly more likely to be female, to have been arrested for a property crime, to be missing information on race due to differential reporting practices in the juvenile and adult systems, and to be less likely to have the arrest lead to a formal conviction. Because any such difference, not matter how small, could impact our findings, these variables are controlled for in all analyses. Table 2 then provides a similar comparison for the smaller conviction sample.

Table 1.

Sample Descriptive Statistics Arrest Sample.

Categorical Variables	Adult (n = 4,050)		Juvenile (n = 3,103)
Categorical Variables	#	%	#	%	p
Gender
Male	2,640	65.19	2,129	68.61	.002
Female	1,401	34.59	974	31.39	.004
Missing	9	0.22	0	0.00	.009
Race
Black	1,195	29.51	859	27.68	.091
White	2,153	53.16	1,288	41.51	.000
Hispanic	633	15.63	555	17.89	.011
Otder	16	0.40	25	0.81	.023
Unknown	53	1.31	376	12.12	.000
Offense type
Person	695	17.16	545	17.56	.652
Property	884	21.83	610	19.66	.026
Drug	493	12.17	390	12.57	.611
Sex	26	0.64	21	0.68	.856
Traffic/violation/otder	1,952	48.20	1,537	49.53	.259
Offense category
Felony	844	20.84	622	20.05	.412
Misdemeanor	3,088	76.25	2,315	74.61	.114
Convicted	1,360	33.62	901	29.78	.001
Continuous Variables	Mean	SE	Mean	SE	p
Priors
Prior felonies	0.18	0.01	0.15	0.01	.066
Prior misdemeanors	0.48	0.03	0.47	0.02	.712

Note. N = 7,153.

Table 2.

Sample Descriptive Statistics Convicted Sample.

Categorical Variables	2009 (n = 1,360)		2010 (n = 901)
Categorical Variables	#	%	#	%
Gender
Male	993	72.69	691	76.69
Female	369	27.01	210	23.31
Missing	4	0.29	0	0.00
Race
Black	479	35.07	275	30.52
White	818	59.88	356	39.51
Hispanic	37	2.71	195	21.64
Asian/American Indian	10	0.73	5	0.55
Unknown	22	1.61	70	7.77
Offense type
Person	193	14.13	114	12.65
Property	386	28.26	164	18.20
Drug	162	11.86	159	17.65
Offense category
Felony	339	24.82	193	21.42
Misdemeanor	892	65.30	648	71.92
Continuous Variables	Mean	SE	Mean	SE
Priors
Prior felonies	0.268	0.023	0.225	0.023
Prior misdemeanors	0.657	0.075	0.796	0.042

Note. N = 2,261.

Results

We first examined 2-year recidivism outcomes of 16-year-olds processed as either adults (2009) or juveniles (2010) when recidivism is defined as a new arrest. Overall, unconditioned by any controls, about 42% of the adult sample was rearrested whereas only 26% of the juvenile sample was rearrested during the follow-up period. The logistic regression presented in Table 3 controls for those variables noted above to differ slightly across the cohorts such as gender, race, and offender type.

Table 3.

Fixed Effect Logistic Regression: Two-Year Recidivism as Any New Arrest by Sample.

Arrest Sample (n = 6,965)	Coefficient	SE	Odds Ratio	p	Conviction Sample (n = 2,181)	Coefficient	SE	Odds Ratio	p
Juvenile	−.758	.060	0.469	.000	Juvenile	−.567	.111	0.567	.000
Convicted	.363	.061	1.438	.000
Male	.544	.061	1.723	.000	Male	.818	.117	2.266	.000
Black	.465	.069	1.592	.000	Black	.522	.123	1.685	.000
Hispanic	.264	.078	1.302	.001	Hispanic	.159	.138	1.172	.248
Asian/American Indian	−.260	.398	0.771	.512	Asian/American Indian	−.132	.624	0.876	.833
Unknown race	−.717	.157	0.488	.000	Unknown race	−.121	.266	0.886	.648
Property	−.302	.073	0.739	.000	Property	−.282	.121	0.754	.020
Felony	−.047	.072	0.954	.509	Misdemeanor	−.133	.105	0.875	.205
Prior felonies	.349	.049	1.418	.000	Drug	−.239	.144	0.787	.097

Here we find that controlling for all differences in the cohorts, youths processed as juveniles had an odds of rearrest that is about half the odds of youths processed as adults (coefficient = −0.758, odds ratio = 0.469). We interpret the odds ratio as 1–0.469 or 0.531, as the coefficient is negative indicating that this is a reduction in the odds. Similar interpretations are used in the remaining analyses. For the conviction sample, the difference is slightly less (coefficient = −0.567, odds ratio = 0.567) with the odds of rearrest of a 16-year-olds processed as juveniles being about 0.43 of that of the odds of a rearrest for 16-year-olds processed as adults. That is, regardless of whether the juvenile further penetrated the system after arrest, youths processed as a juvenile did indeed have significantly lower rates of recidivism.

Sensitivity Analysis 1: Accounting for Reduction in Arrests Across Years

However, as previously noted, there is a particularly sticky point here in that the arrest rate of 16-year-olds dropped significantly immediately after the change in legislation, which may imply police were less likely to arrest 16-year-olds when defined as juveniles. While on the one hand this may be seen as a desirable result as the juvenile justice system focuses more heavily on diversion than the adult system, and the research noted above suggests any arrest may be stigmatizing (Huizinga et al., 1996; Huizinga & Henry, 2008; Liberman et al., 2014), it also introduces a particularly sticky aspect into our study. That is, our proposed reduction in recidivism may be a result of changing police practices rather than changing youth behaviors. Moreover, many youths in the sample would alternate between juvenile and adult status depending on the date they entered the sample and their birthday (e.g., the date they turned 17 years of age). Therefore, it is possible that youths who spent a greater portion of time in the sample as juveniles might recidivate less not because they were less likely to commit acts that could lead to arrest but rather because police, at least in certain jurisdictions, were less likely to formally arrest them for these actions. While these recidivism results are overall promising regardless of the cause (police behavior or juvenile behavior), it is important to ascertain if this behavioral effect on 16-year-olds remains after accounting for any possible policing effect.

Perhaps serendipitous for the analyses, the drop in arrests of 16-year-olds was primarily centered in four jurisdictions: New Haven, Hartford, West Hartford, and East Hartford. That is, while overall there were less arrests in Connecticut in 2010 than in 2009, there were only four jurisdictions in which this drop was greater for juvenile than for adults. Thus, we are able to repeat the analyses removing these four jurisdictions.

As shown in Table 4, removing these jurisdictions from the sample changed neither the overall patterns in arrest observed in the full sample nor the variables that needed to be controlled. Controlling for all differences in the cohorts and excluding the outlier jurisdictions from the model, our analysis revealed that youths processed as a juvenile had an odds of rearrest that is again about half the odds of youths processed as adults (coefficient = −0.773, odds ratio = 0.462). For the conviction sample, the difference is again slightly less and very similar to the original model (coefficient = −0.555, odds ratio = 0.574) with the odds of rearrest of 16-year-olds processed as juveniles being about 0.43 the odds of a rearrest for 16-year-olds processed as adults.

Table 4.

Fixed Effect Logistic Regression: Two-Year Recidivism by Sample Minus Outlier Jurisdictions.

Arrest Sample (n = 5,834)	Coefficient	SE	Odds Ratio	p	Conviction Sample (n = 1,859)	Coefficient	SE	Odds Ratio	p
Juvenile	−.773	.066	0.462	.000	Juvenile	−.555	.122	0.574	.000
Convicted	.376	.067	1.456	.000
Male	.566	.069	1.761	.000	Male	.951	.131	2.588	.000
Black	.496	.078	1.642	.000	Black	.585	.135	1.795	.000
Hispanic	.312	.086	1.366	.000	Hispanic	.273	.149	1.314	.067
Asian/American Indian	−.069	.404	0.933	.864	Asian/American Indian	.095	.653	1.100	.885
Unknown race	−.840	.174	0.432	.000	Unknown Race	−.298	.293	0.742	.310
Property	−.312	.079	0.732	.000	Property	−.209	.129	0.811	.107
Felony	−.008	.081	0.992	.920	Misdemeanor	−.086	.116	0.918	.458
Prior felonies	.341	.057	1.406	.000	Drug	−.199	.163	0.820	.220

While this might seem surprising, it is important to remember that both cohorts are turning 17 and even 18 during the follow-up period and thus were “adults” for much of the follow-up period. As a result, the likelihood of a rearrest may be less influenced by jurisdictional status during the follow-up period than originally assumed. This further bolsters our confidence that processing these youths as juveniles does significantly reduce the likelihood of recidivism through changes in actual offending behaviors and not merely shifts in police behavior.

Sensitivity Analysis II: Recidivism as Any Felony Arrest

While examining the overall arrest patterns, we also noted that the drop in juvenile arrests in these four urban jurisdictions was primarily centered on misdemeanor arrests with no statistically significant differences in felony arrests. That is, while police in these jurisdictions might have been diverting youths for more minor offenses, they were still making arrests for the more serious criminal activity. Again, this is a practice in line with the philosophy of the juvenile court and labeling theory—to divert minor offenders to reduce stigmatization—but it could again confound our results. Thus, as a second sensitivity analysis, we reassessed the regression model this time returning to the full sample (i.e., including the four outlier jurisdiction) but redefine recidivism only as any new arrest for a felony. For the all arrest sample, this final specification did not significantly alter the results with the odds of a felony rearrest for 16-year-olds defined as juveniles being 0.55 the odds of a rearrest for 16-year-olds defined as adults or, again, about half. However, for the all conviction sample, while remaining significant (p = .030), our coefficient did decrease from −0.567 to −0.462 for an odds ratio of 0.630. Interpreting this reduction in the odds as 1–0.630, one would say that the odds of a felony rearrest 16-year-olds processed as juveniles are about 37% less as compared to 43% less in our original model. Thus, the policing behavior of formally arresting 16-year-olds less for misdemeanor offenses when defined as juveniles did partially account for our overall reduction in recidivism (Table 5).

Table 5.

Fixed Effect Logistic Regression: Two-Year Recidivism Defined as New Felony Arrest by Sample.

Arrest Sample (n = 6,258)	Coefficient	SE	Odds Ratio	p	Conviction Sample (n = 1,849)	Coefficient	SE	Odds Ratio	p
Juvenile	−0.814	.137	0.443	.000	Juvenile	−0.462	0.213	0.630	.030
Convicted	0.264	.126	1.302	.036
Male	1.253	.174	3.501	.000	Male	1.429	0.309	4.175	.000
Black	0.150	.143	1.162	.295	Black	0.101	0.223	1.106	.652
Hispanic	0.222	.159	1.249	.164	Hispanic	0.141	0.250	1.151	.572
Asian/American Indian	0.550	.761	1.733	.470	Asian/American Indian	0.399	1.184	1.490	.736
Unknown race	−0.922	.473	0.398	.051	Unknown race	−0.513	0.629	0.599	.415
Property	−0.058	.152	0.944	.700	Property	−0.001	0.227	0.999	.999
Felony	0.288	.135	1.334	.032	Misdemeanor	−0.197	0.186	0.821	.289
Prior felonies	0.288	.056	1.334	.000	Drug	0.075	0.257	1.078	.769

Sensitivity Analysis III: Controlling for Time in Placement

Finally, as a last sensitivity analysis, we addressed the issue of street exposure time during the follow-up period as 90 youths in the sample (1%) did spend some time in confinement subsequent to their instant offense. This number may appear small compared to other studies of adult court processing; however, remember that here we have the entire population of arrested youths and not merely those transferred to adult court for serious and chronic behavior. Although this number is small, we wanted to account for the reduced risk of recidivism posed by an individual who may have been incapacitated for a portion or all of the follow-up period, thus being unable to reoffend. Removing these individuals from the sample and reestimating the models result in a coefficient of −0.772, odds ratio of 0.462 for the all arrest sample and a coefficient of −0.672, odds ratio of 0.511 for the all conviction sample. Again, our findings reveal a large and significant reduction in the odds of recidivism for 16-year-olds processed as juveniles as compared to the odds of recidivism for 16-year-olds process as adults.

Discussion and Conclusions

This study began with the given fact that every state maintains policies allowing youth to be processed in adult courts, yet there is little empirical evidence to support the benefit of such policies. Capitalizing on a natural experiment resulting from a policy change in the state of Connecticut, our research was able to overcome many of the problems of selection bias and cross-state jurisdictional processing differences that have limited earlier work. Specifically, analyzing data from Phase I of Connecticut’s Raise the Age legislation, we sought to determine whether processing the state’s population of 16-year-olds as juveniles rather than adults resulted in lower levels of recidivism.

The recidivism analyses revealed that when processed as juveniles, the population of 16-year-olds arrested in 2010 had reduced odds of rearrest as compared to the population of 16-year-olds arrested in 2009 that were processed as adults. Specifically, when defined as juveniles, the odds of rearrest was about half in the all arrest sample and about 0.43 in the all conviction sample. Additionally, we explored several sensitivity analysis to ensure our findings were not an effect of policing, rather than youth offending, behaviors. In all instances, our sensitivity analysis confirmed our original findings, with the only difference being that results were of a slightly smaller magnitude, although still statistically significant, when defining recidivism only as a felony rearrest in the conviction sample.

In addition to the sensitivity analysis, we feel compelled to note that arrest rates in general have been declining consistently in the 21st century. Thus, from 2009 to 2010, all arrests decreased and similarly from 2010 to 2011 and onward. While we have no immediate reason to believe this overall trend would have impacted our study as the samples were followed over similar time periods with similar declines in arrest rates each year, to the extent that these falling arrest rates may impact overall estimates of recidivism, it is important for future research to account for such historical trends.

Although we are confident our findings provide a significant advantage over prior studies, our study is of course not without limitations of its own. In particular, we were unable to track the outcomes of 77 of the youths who were transferred to the adult system for processing. Thus, there was some inherent selection bias here in that our 2010 sample of all youths processed as juveniles is still missing a handful of the most chronic or serious youths who were transferred to adult courts. Interestingly, though, it is this type of serious offender who is typically transferred to adult court that has been the topic of much of the previous literature in the field with those results also showing similar benefits for juvenile processing.

Additionally, we were unable to obtain further information on the types of placements and treatments youths received in the two systems; therefore, while we can note a clear differences in recidivism, we cannot causally link it to any one type of placement or treatment provided by the juvenile system. This might be of interest since one of the proposed benefits of being processed in the juvenile system is the emphasis on rehabilitation over punishment. Indeed, in preparation for the Raise the Age legislation, Connecticut assessed and expanded the treatment capabilities of its juvenile system to prepare for a great influx of youths and to provide a higher level of services. However, less than 1% of the sample spent any time in placement suggesting that the impact of the system processing might be felt more through diversion and probation services than any type of placement outcome.⁷ Moreover, the ability to generalize these findings to other states may also be inherently tied to that state’s adherence to the original purposes of juvenile justice such as diversion and rehabilitation. For states in which there is less distinction between the practices of juvenile and adult courts, the subsequent impact of system process may be of a lesser consequence.

In addition, we define recidivism as a new arrest or new felony arrest and we did not track technical violations of probation as these data were not available. If youths in either sample were more likely to be placed in confinement for a period of time due to technical probation violations, this would impact their time at risk of a new offense and could potentially downwardly bias our measure of recidivism.

Finally, it is impossible to account for any and all unobserved or exogenous elements that may have, to some degree, influenced our recidivism findings. While representatives in Connecticut assured us that there were no additional official policy changes that would have influenced recidivism outcomes, and we used fixed effects models to account for differences across police jurisdictions, important actors such as probation officers, police departments, or youth services organizations may have made informal and internal adjustments in anticipation of subsequent changes. Moreover, we acknowledge, like most other studies of juveniles in adult courts, that our findings are limited to one state, and thus the policies in this one state (e.g., how rehabilitative its juvenile system truly is compared to other states) may limit the generalizability of our findings to other jurisdictions.

Despite these data limitations, we believe our study is a valuable contribution to the ongoing empirical debate regarding the benefits of processing youths in either the juvenile or adult criminal justice systems. Our findings may be particularly relevant, given that the most recent meta-analysis of prior work shows findings of prior work to be highly dependent on a state’s transfer policy. Thus, the ability to study a population of 16-year-old subject to a policy change, rather than a small subsample of only select youths targeted for transfer, is a significant contribution to the literature. Directions for future research include expanded data collection efforts that would not only link later recidivism outcomes with the system of processing but to the actual services and treatment provided by each system thus examining not only if the system of processing matters, but why it matters and what processes and services may be correlated with reductions in recidivism.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Notes

References

Baird

A. A.

Fugelsang

J. A.

(2004). The emergence of consequential thought: Evidence from neuroscience. Philosophical Transactions of the Royal Society B, 359, 1797–1804.

Becker

(1963). Outsiders: Studies in the sociology of deviance. New York, NY: Free Press.

Bernard

T. J.

Kurlychek

M. C.

(2010). The cycle of juvenile justice (2nd ed.). Oxford, England: Oxford University Press.

Bishop

D. M.

(2006). Public opinion and juvenile justice policy: Myths and misconceptions. Criminology & Public Policy, 5, 653–664.

Bishop

D. M.

Frazier

C. E.

Lanza-Kaduce

Winner

(1996). The transfer of juveniles to criminal court: Does it make a difference? Crime and Delinquency, 42, 171.

Butts

J. A.

Mitchell

(2000). Brick by brick: Dismantling the border between juvenile and adult justice. Criminal Justice, 2, 167–213.

Cook

P. J.

Laub

J. H.

(2002). After the epidemic: Recent trends in youth violence in the United States. Crime and Justice, 29, 1–37.

Fagan

(1991). The comparitive impacts of juvenile and criminal court sanctions on adolescent offenders. Washington, DC: US Department of Justice, Office of Justice Programs, Office of Juvenile Justice and Delinquency Prevention.

Fagan

(1996). The comparative advantage of juvenile versus criminal court sanctions on recidivism among adolescent felony offenders. Law and Policy, 18, 77–114.

10.

Feld

B. C.

(1984). The decision to seek criminal charges: Just deserts and the waiver decision. Criminal Justice Ethics, 3, 27–41.

11.

Feld

B. C.

(1987). The juvenile court meets the principle of offense: Legislative changes in juvenile waiver statutes. Journal of Criminal Law and Criminology, 78, 471–533.

12.

Feld

B. C.

(1999). Bad kids: Race and transformation of the juvenile court. Oxford, England: Oxford University Press.

13.

Feld

B. C.

(2000). Legislative exclusion of offenses from juvenile court jurisdiction: A history and critique. In Fagan

Zimring

F. E.

(Eds.), The changing borders of juvenile justice: Transfer of adolescents to the criminal court (pp. 83–144). Chicago, IL: University of Chicago Press.

14.

Goffman

(1963). Behavior in public place. Glen Cove, NY: The Free Press.

15.

Goodstein

Sontheimer

(1987). A study of the impact of 10 Pennsylvania residential placements on juvenile recidivism. Shippensburg, PA: Center for Juvenile Justice Training and Research.

16.

Griffin

Torbet

Szymanski

(1998). Trying juveniles as adults in criminal court: An analysis of state transfer provisions. Washington, DC: U.S. Department of Justice, Office of Justice Programs, Office of Juvenile Justice and Delinquency Prevention.

17.

Hall

G. S.

(1904). Adolescence. New York, NY: Appleton.

18.

Hansen

Waddell

(2014). Walk like a man: Do juvenile offenders respond to being tried as adults. Unpublished Manuscript, University of Oregon, Eugene.

19.

Hirschi

(1969). Causes of delinquency. Berkeley: University of California Press.

20.

Huizinga

Esbensen

Weiher

(1996). The impact of arrest on subsequent delinquent behavior. In Loeber

Huizinga

Thornberry

T. P.

(Eds.), Program of research on the causes and correlates of delinquency (pp. 3–20). Washington, DC: Office of Juvenile Justice and Delinquency Prevention.

21.

Huizinga

Henry

K. L.

(2008). The effect of arrest and justice system sanctions on subsequent behavior: Findings from 962 WILEY ET AL. longitudinal and other studies. In Liberman

A. M.

(Ed.), The long view of crime: A synthesis of longitudinal research (pp. 220–254). New York, NY: Springer.

22.

Keating

D. P.

(1990). Adolescent thinking. In Feldman

S. S.

Elliott

G. R.

, At the threshold: The developing adolescent (pp. 54–89, x, 642). Cambridge, MA: Harvard University Press.

23.

Kirk

D. S.

Sampson

R. J.

(2013). Juvenile arrest and collateral educational damage in the transition to adulthood. Sociology of Education, 86, 36–62.

24.

Klaczynski

P. A.

(1993). Reasoning schema effects on adolescent rule acquisition and transfer. Journal of Educational Psychology, 85, 679.

25.

Lanza-Kaduce

Frazier

C. E.

Lane

Bishop

D. M.

(2002). Juvenile transfer to criminal court study: Final report. Tallahassee: Florida Department of Juvenile Justice.

26.

Lee

D. S.

McCrary

(2009). The deterrence effect of prison: Dynamic theory and evidence. Cambridge, MA: National Bureau of Economic Research.

27.

Lemert

E. M.

(1951). Social pathology: A systematic approach to the theory of sociopathic behavior. New York: McGraw Hill.

28.

Lenroot

R. K.

Giedd

J. N.

(2006). Brain development in children and adolescents insights from anatomical magnetic resonance imaging. Neuroscience and Biobehavioral Reviews, 30, 718–729.

29.

Liberman

A. M.

Kirk

D. S.

Kim

(2014). Labeling effects of first juvenile arrests: Secondary deviance and secondary sanctioning. Criminology, 52, 345–370.

30.

Link

B. G.

(1982). Mental patient status, work, and income: An examination of the effects of a psychiatric label. American Sociological Review, 47, 202–215.

31.

Loeffler

C. E.

Grunwald

(2015). Processed as an adult: A regression discontinuity estimate of the crime effects of charging non-transfer juveniles as adults. Journal of Research in Crime and Delinquency, 52, 890–922.

32.

Lopes

Krohn

M. D.

Lizotte

A. J.

Schmidt

N. M.

Vá squez

B. E.

Bernburg

J. G.

(2012). Labeling and cumulative disadvantage: The impact of formal police intervention on life chances and crime during emerging adulthood. Crime & Delinquency, 58, 456–488.

33.

Loughran

T. A.

Mulvey

E. P.

Schubert

C. A.

Chassin

L. A.

Steinberg

Piquero

A. R.

… Losoya

(2010). Differential effects of adult court transfer on juvenile offender recidivism. Law and Human Behavior, 34, 476.

34.

McNulty

E. W.

(1996). The transfer of juvenile offenders to adult court: Panacea or problem? Law and Policy, 18, 61–76.

35.

Moshman

(1998). Cognitive development beyond childhood. Educational Psychology Papers and Publications, 48.

36.

Paternoster

Iovanni

(1989). The labeling perspective and delinquency: An elaboration of the theory and an assessment of the evidence. Justice Quarterly, 6, 360–394.

37.

Piaget

(1964). Part I: Cognitive development in children: Piaget development and learning. Journal of Research in Science Teaching, 2, 176–186.

38.

Platt

A. M.

(1977). The child savers: The invention of delinquency. Chicago, IL. University of Chicago Press.

39.

Podkopacz

M. R.

(1996). The juvenile court’s final decision: An empirical examination of transferring juveniles to adult court (Doctoral dissertation, University of Minnesota).

40.

Podkopacz

M. R.

Feld

B. C.

(1995). End of the line: An empirical study of judicial waiver. Journal of Criminal Law and Criminology, 86, 449.

41.

Sampson

Laub

(1993). Crime in the making. Cambridge, MA: Harvard University Press.

42.

Sickmund

Snyder

H. N.

(1999). Juvenile offenders and victims: 1999 national report. Washington, DC: Office of Juvenile Justice and Delinquency Prevention.

43.

Smith

D. A.

Brame

(1994). On the initiation and continuation of delinquency. Criminology, 32, 607–629.

44.

Steinberg

Cauffman

(1996). Maturity of judgment in adolescence: Psychosocial factors in adolescent decision making. Law and Human Behavior, 20, 249.

45.

Steinberg

Cauffman

(1999). Forward: The elephant in the courtroom: A developmental perspective on the adjudication of youthful offenders. Virginia Journal of Social Policy & the Law, 6, 389–653.

46.

Sutherland

E. H.

(1937). The professional thief. Chicago, IL: University of Chicago Press.

47.

Sutherland

E. H.

(1942). Development of the theory. Edwin Sutherland on Analyzing Crime, 30–41.

48.

Torbet

Gable

Hurst

IV Montgomery

Syzmanski

Thomas

(1996). State responses to serious and violent juvenile crime. Pittsburgh, PA: National Center for Juvenile Justice.

49.

Winner

Lanza-Kaduce

Bishop

D. M.

Frazier

C. E.

(1997). The transfer of juveniles to criminal court: Reexamining recidivism over the long term. Crime and Delinquency, 43, 548–563.

50.

Zane

S. N.

Welsh

B. C.

Mears

D. P.

(2016). Juvenile transfer and the specific deterrence hypothesis. Criminology & Public Policy, 15, 901–925.

51.

Zimring

(1991). The treatment of hard cases in American juvenile justice: In defense of discretionary waiver. Notre Dame Journal of Law, Ethics and Public Policy, 5, 267–280.