Skip to main content
SearchLoginLogin or Signup

The Right to a Good Defense: Investigating the Influence of Attorney Type across Urban Counties for Juveniles in Criminal Court

Juvenile defendants in criminal court represent an especially vulnerable group for whom quality legal representation is critical. While some juvenile defendants are able to obtain private counsel, indigent defendants are provided an attorney by the government. One ...

Published onFeb 10, 2020
The Right to a Good Defense: Investigating the Influence of Attorney Type across Urban Counties for Juveniles in Criminal Court
·

Abstract

Juvenile defendants in criminal court represent an especially vulnerable group for whom quality legal representation is critical. While some juvenile defendants are able to obtain private counsel, indigent defendants are provided an attorney by the government. One longstanding concern is that these court-appointed attorneys are less effective. Using data on juveniles in criminal court across 37 large, urban counties, the present study examines conviction and sentencing outcomes by comparing private counsel, public defenders, and assigned counsel. Results indicate that defendants with public defenders were less likely to be convicted, less likely to be incarcerated in prison, and served shorter prison sentences compared to defendants with assigned counsel. Contrary to hypotheses, however, the effect of attorney type was not conditioned by court urbanism. The findings suggest that public defenders provide effective legal representation for juveniles in criminal court. Research is needed to determine whether this holds across different contexts (e.g., rural).

Introduction

The 6th amendment to the U.S. Constitution (1791) provides, “In all criminal prosecutions, the accused shall enjoy the right to a speedy and public trial, by an impartial jury of the State and district wherein the crime shall have been committed . . . and to have the Assistance of Counsel for his defence.” For more than a century this last clause was interpreted to mean that the federal government (and later states) could not prevent a criminal defendant from obtaining legal representation.[1] It was not until 1932 that the 6th amendment was interpreted to mean that indigent defendants must be provided legal representation in capital cases (Powell v. Alabama, 1932). Shortly thereafter, this right to counsel for indigent defendants extended to all federal felony cases (Johnson v. Zerbst, 1938), and then to state felony cases (Gideon v. Wainwright, 1963) and misdemeanor cases (Argersinger v. Hamlin, 1972). Today, the right to counsel for those who cannot afford legal representation is a cornerstone of the U.S. criminal justice system.

In a series of landmark decisions, the Supreme Court extended this right to counsel to juveniles as well. For much of the juvenile court’s history, the right to counsel was absent. Due to its special role as parens patriae, the juvenile court was guided by a paternalistic philosophy where the court determined the child’s best interests in a non-adversarial setting (Tanenhaus, 2004). This approach lasted almost 70 years, but came under attack by the 1960s. Three years after Gideon, in its Kent v. United States (1966) decision, the Court held that juvenile defendants had the right to counsel at waiver hearings because the decision to transfer youth to criminal court was too significant to leave up to the juvenile court judge’s discretion in a non-adversarial setting. Kent narrowly focused on the waiver hearing, but only one year later the Court held that certain due process rights, including the 6th amendment right to counsel, extended to all juvenile court proceedings (In re Gault, 1967). In this historic “constitutional domestication” of the juvenile court (Gault, p. 22), the Court in Gault observed that, “The child requires the guiding hand of counsel at every step of the legal proceeding against him” (p. 36).

This “guiding hand of counsel” is no less important for juveniles who are transferred to criminal court. Since its inception, the juvenile court has determined that some juvenile offenders should be handled in adult court because of their dangerousness to other juveniles or to the public (Tanenhaus, 2004). Toward the end of the 20th century, however, the number of youth eligible for criminal court processing expanded dramatically—what has been called the “criminalization” of modern juvenile justice (Feld, 2017). This expansion—likely due to an increasing concern about violent juvenile crime and the moral panic regarding “super predators” (Bernard & Kurlychek, 2010, p. 157)—peaked with 13,000 judicially waived cases in 1994 (Hockenberry & Puzzanchera, 2015). Guided by the idea of juvenile offenders as “hardened criminals” as opposed to children (Bernard & Kurlychek, 2010, p. 5), prosecutorial and legislative waiver also became more prominent, transforming the practice of transfer from a decision guided by consideration of the best interests of youth by the juvenile court judge to one guided by criminal justice officials such as the prosecutor (Zimring, 2010). Estimates for the number of juveniles transferred by these non-judicial mechanisms are not available (see Griffin et al., 2011), but some estimate that as many as 200,000 juveniles per year were transferred to the adult justice system during the 1990s (Woolard et al., 2005). Given these changes, the right to counsel is of paramount importance for juveniles in adult court. An important follow-up inquiry is thus: Given their special status as a vulnerable population and a shift in perceptions of juveniles as hardened criminals deserving of criminal court processing, what kind of representation is most effective for juveniles in adult court?

The present study examines whether type of attorney influences outcomes for juveniles transferred to criminal court. We define an effective defense in criminal court as one that results in non-conviction, non-incarceration (among those convicted), or shorter sentence length (among those incarcerated). While most prior research on this topic compares public defenders and private attorneys, we seek to build upon the literature by examining three types of legal representation: public defenders, assigned counsel, and private (retained) counsel. In addition to examining the direct influence of attorney type on these outcomes, the present study assesses whether the influence of attorney type is conditioned by the urbanism of the court and surrounding community. We use data from multiple state jurisdictions so that our findings are not limited to particular (possibly idiosyncratic) jurisdictions. Our theoretical expectations are that public defenders will be more effective than assigned counsel and private (i.e., retained) counsel, and that this will be especially pronounced in larger, more urban jurisdictions where public defender systems are thought to be more efficient due to larger working groups with more routinized procedures (see Nardulli, Eisenstein, & Flemming, 1988). We also expect that assigned counsel will be less effective than public defenders as well as private counsel, representing the worst of both worlds: subject to public resource constraints as well as isolated from the efficiencies of tightly coupled public defender systems.

Background

The most basic distinction in attorney representation for criminal cases is between attorneys hired and paid by their clients and attorneys provided by the government to clients who cannot otherwise afford representation. There are different ways for state criminal justice systems to satisfy this latter 6th amendment guarantee, and there are no federal constitutional requirements regarding implementation (Davies & Worden, 2017). First, states can set up public defender organizations that specialize in representing indigent defendants. These public employees represent the counterpoint to the prosecutor’s office, originally established to process criminal cases in a more routinized and less adversarial fashion, even if these organizational goals sometimes took precedence over the client’s best interests (Felice, 2001). Following Gideon, however, there was an “explosion in the number of public defender offices and a fundamental change in the offices’ philosophy of representation” toward more zealous advocacy of clients’ best interests (Felice, 2001, p. 981).

Public defenders are perhaps the most visible form of indigent defense, yet they are not necessarily the most common (Worden, 1991, p. 392). In a majority of jurisdictions, private attorneys are appointed by the court to provide indigent defense on a case-by-case basis, usually for a below-market fee (Burnett, Moore, & Butcher, 2001). Referred to as “assigned counsel,” others have described this arrangement as “a kind of cross between public defenders and privately-retained defense counsel” (Hoffman et al., 2005, p. 233). That is, assigned counsel are attorneys with private practices (as opposed to being government employees), but are paid below-market rates for representing indigent clients, similar to public defenders. Assigned counsel can take many forms, from judges selecting attorneys based on an available list (i.e., “ad hoc” selection) to private law firms bidding for contracts (i.e., “contract attorneys”) for the representation of indigent clients in a certain court (see, e.g., Felice, 2001).

The Role of Public Defenders

Despite the promise of Gideon, many commentators—including some public defenders—believe that most indigent defendants still lack access to effective counsel (see, e.g., Bibas, 2013; Dripps, 2013; Gross, 2013; Lucas, 2005; Mosteller, 2013). It is often observed that “[a]ppointed defense counsel are underpaid, undersupported, and overworked” (Bibas, 2013, p. 1291). Due to high caseloads and limited resources, as the argument goes, public defender organizations cannot possibly provide quality representation for all indigent cases (Williams, 2013). In addition to lack of resources, there is also a perception that private attorneys are more qualified than public defenders because public defenders represent attorneys who were “unable to succeed in the private sector” (Rattner, Turjeman, & Fishman, 2008, p. 43), contributing to a “low status” in the legal community (Cole, 1970, p. 338).

On the other hand, it may be that public defenders offer some distinct benefits in providing legal representation. Public defenders are salaried employees who operate in a complex bureaucratic setting with different incentives compared to private attorneys operating under market forces (Davies & Worden, 2017). The criminal court and its component parts, such as prosecutors and public defenders, are bound by organizational goals that may take precedence over other concerns (see Dixon, 1995). As a result, patterns of routinization emerge within courtroom workgroups (Myers & Reid, 1995). Despite ideological differences (such as commitment to due process versus crime control models; see Packer, 1968), bureaucratic constraints often necessitate the prioritization of organizational goals—creating a less adversarial process than is often imagined (see Blumberg, 1967; Feeley, 1972).

As an integral part of the criminal courtroom workgroup, public defenders share organizational goals and develop expertise and working relationships with other members of the court (such as judges and prosecutors; see Metcalfe, 2016) that may prove beneficial to their clients—what has been recently referred to as “relational expertise” (Sandefur, 2015, p. 911). Compared to assigned counsel, who operate in relative isolation and are subject to the same (or greater) financial constraints (Anderson & Heaton, 2012), public defenders can draw on the resources of their organization, their experience, and their court relationships to provide a more effective defense for their indigent clients (Myers & Reid, 1995). In fact, some have even suggested that public defenders “potentially provide specialized service of high quality, equaling or surpassing that offered by privately retained lawyers” (Worden, 1993, p. 614). A key observation here involves the role of plea bargaining, which accounts for the vast majority of convictions and is the result of negotiations between the prosecutor and defense attorney (see Metcalfe, 2016). Due to their familiarity in the courtroom workgroup, public defenders may be in a better position to negotiate pleas than retained private attorneys as well as assigned counsel.

In the case of juveniles transferred to criminal court, our present focus, the advantages of public defender representation are even more salient. Juveniles are clearly less capable of navigating the criminal justice system (Monahan, Steinberg, & Piquero, 2015). As is the case with adults, prosecutors have the power to decide how they wish to charge, or whether to bring a case at all (see Spohn, 2018). But unlike with adults, they also have the power to reverse waive a juvenile’s case back to juvenile court (Singer, 1997). Others have observed that prosecutors appear to be especially punitive in their orientation toward juveniles compared to other court actors, such as probation officers and judges (Myers et al., 2011). Juvenile defendants in criminal court thus need zealous advocacy from their legal representation. While it may seem reasonable to assume that privately retained attorneys would be most effective under the basic logic that you “get what you pay for” (Rattner et al., 2008, p. 44), there are reasons for hesitation.

In addition to the organizational benefits of public defenders outlined above, private attorneys might involve a greater range of motivations for taking a juvenile’s case.[2] Most likely, the juvenile’s family will be paying for legal representation, creating possible conflicts of interest (Moore, 1996). Although all attorneys have an ethical obligation to represent their clients’ interests (see Model Rules, r. 1.7–1.8), there may nevertheless be subtle differences in the way a case is presented by a private attorney hired and paid by parents. For example, attorneys may not be in a position to draw testimony that can degrade the character of parents, such as where a possible mitigating defense would be parental abuse. With public defenders, it is clear that their only responsibility is to provide zealous advocacy for their clients.

Urbanism and Organizational Structure 

Courts vary in how organizational goals are shared among organizational actors, what others have referred to as tight versus loose coupling (Singer, 1997). One goal of the bureaucracy is to produce a more tightly coupled system with greater routinization in the behavior of organizational actors—leading to more efficient case processing (see Feeley, 1972). But in less tightly coupled courts—such as small, rural courts—this may not be possible. As Weber (1954) observed, while increasing bureaucratization will tend to create a more rational, formal legal system where extralegal considerations matter less and less, this may only occur in large bureaucracies. Smaller and more rural courts face more limited resources, often necessitating less formal or routinized processes (see Hester, 2017; Worden, Shteynberg, Morgan, & Davies, 2019). Empirical work has borne this out in the context of juvenile court, generally finding that urban courts tend to be more due process oriented (Feld, 1991, 2017) and that formal outcomes are more likely (Rodriguez, 2010).

Consistent with these more general observations about urban versus rural courts, others have remarked that public defender systems are most cost-effective in large, urban jurisdictions (Worden, 1993). As a result, public defender systems are more common in large, urban counties, while assigned counsel for indigent defendants remains more likely in rural counties (Worden, 1991). As Worden (1991) points out, this means that while there are fewer public defender systems than assigned counsel systems, more indigent defendants overall are represented by public defenders. It stands to reason that the benefits of public defender systems will be especially pronounced in more densely populated courts with greater resources and more routinization of courtroom workgroups (see Metcalfe, 2016).

Literature Review

Most research on the relationship between attorney type and court outcomes focuses on adult defendants in criminal court.[3] Early research was quite mixed. Some studies found that private attorneys secured better outcomes for their clients (see, e.g., Gitelman, 1971; Nagel, 1973; Spohn, Gruhl, & Welch, 1981), others found that public defenders secured better outcomes (see, e.g., Willison, 1984), and still others found no difference (see, e.g., Nardulli, 1986; Taylor et al., 1972; Wheeler & Wheeler, 1980).[4] More recent research utilizing sophisticated statistical approaches reports similarly mixed findings.

Some research suggests that private attorneys are more effective than public defenders. In an analysis of 476 capital murder cases in Georgia, Beck and Shumsky (1997) found that defendants with appointed counsel were more likely to receive the death penalty than defendants with retained (private) counsel. Martinez and Pollock (2008) examined the influence of attorney type for all criminal defendants processed in the largest county in Texas during 2004, finding that incarceration was significantly more likely for appointed attorneys (including public defenders) than for retained (private) counsel. Williams (2013) examined the influence of attorney type on court outcomes in the four largest Florida counties in 2006. The author found that defendants with public defenders fared worse than defendants with private attorneys across several outcomes: less likely to have their charges dismissed, more likely to be detained prior to trial, and more likely to be convicted. More recently, Williams (2017) examined bail decisions for a sample of Florida felony defendants from 1990–2004, finding that clients with public defenders were more likely to have bail denied and were less likely to be released.

Still other studies have found no relationship between attorney type and court outcomes. Williams (2002) examined sentencing outcomes for 420 felony convictions in a Florida county and found that type of attorney was not significantly related to probation, incarceration, or sentence length. Similarly, in an analysis of 2,850 felony convictions in Cook County, Illinois, Hartley, Miller, and Spohn (2010) found no significant differences between public and private attorneys in terms of charge reduction or ROR release.

At least one study has found that public defenders are more effective than private attorneys. Anderson and Heaton (2012) took advantage of a natural experiment in Philadelphia where every fifth indigent murder case was randomly assigned to the public defender’s office, with all other murder cases referred to assigned counsel. Examining 3,173 defendants charged with murder between 1994 and 2005, the authors found no differences in guilty verdicts regardless of attorney type, but did find significant differences in average incarceration sentence length and likelihood of receiving a life sentence. Specifically, sentences were on average 14 percent shorter, and life sentences 25 percent less likely, for defendants represented by the public defender’s office.[5] This is an especially interesting finding in light of the possibility that the association between public defender and more punitive outcomes may be a result of selection bias. Specifically, given that poor defendants are more likely to be represented by public defenders, the association between public defenders and punitive outcomes could be a class effect rather than an attorney effect (Nagel, 1973; see also Hermann, Single, & Boston, 1977). In Anderson and Heaton (2012), however, the entire sample is indigent and defendants are randomly assigned to public defenders—virtually eliminating any possible selection effect.

Noting that most of the research on attorney effectiveness has focused on the binary distinction between private and public attorneys, a recent study of attorney type in the criminal justice system examined case outcomes where the defendant was represented by a private attorney, public defender, or assigned counsel (Cohen, 2014). Using the 2004 and 2006 State Court Processing Statistics (SCPS), a multi-state sample of 40 of the largest 75 U.S. counties, Cohen (2014) found that conviction, incarceration, and sentence length were not significantly different for defendants represented by public defenders versus private attorneys—but that defendants with assigned counsel fared worse in terms of higher likelihood of conviction, higher likelihood of prison, and longer incarceration sentences. One earlier study performed by Hoffman and colleagues (2005) also compared the influence of private attorneys, public defenders, and court-appointed counsel for every felony case filed in Denver, Colorado, in 2002. The authors found that even after controlling for offense seriousness, public defenders were less effective than private attorneys and assigned counsel in terms of incarceration outcomes.

Unlike these last two studies, research does not often distinguish between types of court-appointed counsel—public defenders versus assigned counsel—and instead groups them together as “appointed” versus “retained” (i.e., privately hired). As such, much of the research that indicates that appointed counsel is less effective than retained counsel does not make clear whether public defenders are less effective than private attorneys. Moreover, contrary research indicating no differences between private and public attorneys may also be due to the conflation of public defenders and other forms of indigent defense (Frederique et al., 2015).

Study Hypotheses

The present study seeks to address this gap in prior research by comparing the effectiveness of three kinds of legal representation—public defenders, assigned counsel, and private (retained) attorneys—in terms of conviction and sentencing outcomes for an especially vulnerable group, juveniles in criminal court. It also examines whether attorney effectiveness is conditioned by the urbanism of the jurisdiction, something that prior research has not investigated. The following hypotheses will be tested:

H1a: Juvenile defendants represented by public defenders will be less likely to be convicted than defendants represented by (retained) private counsel or assigned counsel. Further, defendants represented by private counsel will be less likely to be convicted than defendants represented by assigned counsel. That is, we expect that public defenders will provide the most effective representation (in terms of reduced odds of conviction), while assigned counsel will provide the least effective representation.

H1b: The influence of attorney type on likelihood of conviction (H1a) will be greater in more urbanized courts. Specifically, the likelihood of conviction for defendants with public defenders (compared to others) will be lower in more urbanized courts (i.e., public defenders more effective in more urbanized courts).

H2a: Convicted juvenile defendants represented by public defenders will be less likely to be incarcerated in jail or prison (compared to community sentence) than defendants represented by (retained) private counsel or assigned counsel. That is, we expect that public defenders will provide the most effective representation (in terms of reduced odds of incarceration), while assigned counsel will provide the least effective representation.

H2b: The influence of attorney type on likelihood of sentence type (H2a) will be greater in more urbanized courts. Specifically, the likelihood of incarceration for defendants with public defenders (compared to others) will be lower in more urbanized courts (i.e., public defenders more effective in more urbanized courts).

H3a: Incarcerated juvenile defendants represented by public defenders will serve shorter sentences than defendants represented by (retained) private counsel or assigned counsel. Further, incarcerated defendants represented by private counsel will serve shorter sentences than defendants represented by assigned counsel. That is, we expect that public defenders will provide the most effective representation (in terms of jail and prison sentence length), while assigned counsel will provide the least effective representation.

H3b: The influence of attorney type on length of sentence (H3a) will be greater in more urbanized courts. Specifically, sentence lengths received by defendants with public defenders (compared to others) will be shorter in more urbanized courts (i.e., public defenders more effective in more urbanized courts).

Methods

Data

The present study uses a combination of individual-level case data and county-level contextual data. The court processing data comes from 1998 Juvenile Defendants in Criminal Courts (JDCC), the only publicly available dataset including official criminal court statistics for transferred juvenile defendants across multiple jurisdictions. The JDCC is a subset of the 1998 State Court Processing Statistics; it represents 40 large, urban counties selected from that larger dataset (for more details, see Steiner 2009). Each case (N = 7,135) was followed through adjudication and sentencing for at least one year. Some cases were still pending at the time of data collection and were treated as data missing completely at random and dropped from the analysis (n = 581). In addition, three counties (n = 32) were dropped due to missing data not collected by those counties. Within the remaining 37 counties, cases were also dropped if missing the main independent variable of interest, attorney type (n = 1,315), or missing data on one or more control variables (n = 186).

The JDCC data were merged with a level-2 dataset that includes county demographic factors based on official Census data and Uniform Crime Reports (UCR) data. The three full samples correspond to the three outcomes of interest: conviction (N = 5,019), sentence type (N = 3,691), jail length (N = 717), and prison length (N = 1,350) The conviction and sentencing samples include juvenile defendants in criminal court in 37 counties (across 19 states), while the prison length sample includes 36 counties and the jail length sample includes 26 counties. The conviction sample represents the full sample of juvenile defendants transferred to adult court, the sentencing sample represents a subsample of convicted defendants, and the prison and jail length samples represent subsamples of incarcerated defendants (in prison and jail). While it has become common to use the Heckman two-step procedure to account for the selection bias that may be associated with examining only those who are convicted from the original sample in sentencing models, it is problematic to do so without including exclusion restrictions (see Bushway, Johnson, & Slocum, 2007; Stolzenberg & Relles, 1997).[6] Here, no exclusion restrictions could be identified because no variables were theoretically predictive of conviction but not also theoretically predictive of incarceration and sentence length (a good candidate would be strength of evidence, which is not available). Running sensitivity analyses using the heckprobit command in Stata confirmed the issue, as the error terms for the selection and substantive equations were highly correlated (ρ =.99). As such, the Heckman correction is not appropriate for the present study. Descriptive statistics for all four samples are presented in table 1, indicating that the proportion of cases with different forms of representation did not differ appreciably among the samples.

Measures

Dependent variables

Table 1 includes descriptive statistics across the samples corresponding to the four criminal court outcomes examined in the present analysis. First, conviction was coded as a dichotomous variable indicating whether the defendant was convicted of the charged offense; convictions included guilty pleas and guilty trial verdicts while non-convictions included dismissal, nolle prosequi (decision not to prosecute), transferred to juvenile court, and not guilty verdict. As shown in table 1, the majority of defendants were convicted (n = 3,775; 75.21 percent). Second, following suggestions to move beyond the simple “in/out” dichotomy of incarceration (see Holleran & Spohn, 2004), sentence type for convicted defendants (N = 3,691) is coded as community sentence (n = 1,116; 31.59 percent), jail sentence (n = 768; 20.81 percent), or prison sentence (n = 1,757; 47.60 percent). Prison sentences included sentences to juvenile facilities (n = 108). Third, sentence length for jail and prison were coded as days sentenced, based on minimum and maximum sentences.[7] Jail sentences (N = 717) ranged from 1 day to 730 days, with a mean (SD) sentence of 208.0 days (130.86)—approximately seven months. For prison sentences (N = 1,350), life sentences (n = 59) were re-coded as 18,250 days (50 years). Of these, 34 received a life sentence only (i.e., minimum or maximum), while 25 received a life maximum as part of an indeterminate sentence range (i.e., < 18,250 days). Two prison sentences did not have any sentence length information and were dropped from the analysis. Prison sentences ranged from 60 to 18,250 days, with a mean (SD) sentence of 3037.60 days (3791.02)—approximately 8.3 years.

[Table 1 about here]

Attorney type

The main explanatory variable is attorney type, which was originally coded as a series of five values: public defender, private attorney, assigned counsel, self-representation, and other. Approximately 18 percent of cases (n = 1,305) in the full sample were missing information for attorney type and were dropped from the analysis. Further, cases coded as self-representation (n = 10) were also dropped. (No cases were coded as “other.”) The remaining values were recoded as a factor variable with three categories: public defender (n = 3,199; 63.74 percent), private retained counsel (n = 910; 18.13 percent), and assigned counsel (n = 910; 18.13 percent). Public defender representation serves as the reference category in subsequent analyses since the main comparison of interest is between public defenders and other forms of representation. Only four counties had no cases represented by public defender, while two counties had no private (retained) attorneys and 12 counties had no assigned counsel.

Urbanism

The contextual variable of interest is urbanism. Urbanism serves as a good proxy for the organization of the court, specifically the routinization of the courtroom workgroup (see Metcalfe, 2016). This follows from prior work showing that more urban courts are more tightly coupled, formal, and due process oriented, often leading to more punitive outcomes (see Bray, Sample, and Kempf-Leonard, 2005; Feld, 1991; Rodriguez, 2010). This is not necessarily true of county size, since counties can be low density and thus not necessarily tightly coupled (i.e., large, rural counties). Based on the closest decennial Census (2000), urbanism is measured as a continuous measure of population density, and was recoded as 1,000 persons per square mile for interpretation purposes. While all counties in our sample are large, mostly urban jurisdictions, they still differ considerably in level of urbanism. Across 37 counties, population density ranges from 90 to 66,940, with a mean average of approximately 4,230 persons per square mile.

Case-level control variables

Additional individual-level variables were included as controls (see table 1). Defendant race/ethnicity was coded as a series of mutually exclusive dummy variables for Black, Hispanic (non-Black), Other,[8] and White (non-Hispanic); white serves as the reference category. In the full sample (model 1), Black defendants made up the majority of defendants (n = 2,974; 59.25 percent), followed by Hispanic defendants (n = 987; 19.67 percent) and White defendants (n = 956; 19.05 percent). Age is a continuous variable coded when charges were filed in criminal court (rather than at arrest), ranging from 13 to 26 years in the full sample (mean = 16.32). Defendant sex was coded as “1” for female and “0” for male, with female defendants making up only 2.95 percent of the full sample (n = 148).[9]

Due to the possibility that different types of attorneys are assigned to different types of cases (e.g., public defenders may receive less serious cases), we controlled for offense type coded as a series of dummy variables for 11 criminal offenses: murder, rape/sexual assault, assault, robbery, burglary, larceny, motor vehicle theft, drug/alcohol offenses, weapons offense, other felonies, and misdemeanors; felony assault served as the reference category. With the exception of the misdemeanor category (n = 202; 4.02 percent), all offenses were felonies. A binary indicator for whether multiple charges were filed against the defendant was also included, indicating that most cases faced multiple charges (n = 3,343; 66.61 percent). Rearrest was coded as a binary variable indicating whether the defendant was rearrested after his or her initial charge (n = 313; 6.24 percent). A measure for pretrial detention (also coded as binary) indicated whether the defendant was detained without bail (or was unable to pay bail) prior to trial; this was true for about half of cases (n = 2,600; 51.80 percent). Some research also suggests that type of waiver may be associated with court outcomes (see, e.g., Zane, 2017). Type of waiver indicates how a juvenile defendant ended up in criminal court: by traditional judicial waiver (n = 1,398; 27.85 percent), prosecutorial direct-file (n = 2,099; 41.82 percent), or statutory exclusion (n = 1,522; 30.32 percent). Finally, a dummy variable for “guilty plea” is included given the evidence that pleas result in lower odds of incarceration and shorter sentences, likely through negotiated plea bargains (see, e.g., Johnson 2019). Most cases involved pleas (n = 3,488; 69.50 percent), with 92 percent of convictions resulting from pleas rather than trial.[10]

Contextual control variables

Several county-level variables are also included as controls. First, concentrated disadvantage is included as a measure of court resources, on the assumption that courts in disadvantaged counties will tend to have fewer resources (due to the local tax base). This could result in differential processing outcomes, as well as differential attorney effectiveness (e.g., a tightly coupled public defender system may be less effective in a more disadvantaged county). Concentrated disadvantage is measured as a weighted index (using principles components analysis) based on unemployment rate, families using public assistance, families below poverty, and percentage of female-headed households (Cronbach’s alpha = .92). For interpretation purposes, standardized scores are included in all analyses (re-standardized for each model). Prior to standardization, concentrated disadvantage scores ranged from -3.62 (indicating relative low concentrated disadvantage) to 7.77 (indicating relative high concentrated disadvantage), with a mean (SD) of 0 (2.19). Second, county-level index crime rates were included as a measure of the surrounding crime problem. More crime could plausibly be related to attorney effectiveness, with greater demands on limited resources. Data from 1998 was obtained from the UCR and an index crime rate was created as an average for the years 1997 through 1999. Crime rates ranged from 23.64 to 103.82 (per 1,000), with a mean rate of 71.04 crimes per 1,000 persons.

Finally, jurisdictional controls were added. First, since the composition of juveniles in criminal court is a function of juvenile court jurisdiction, upper age of juvenile court jurisdiction as of 1998 was included: age 15 (n = 267; 5.32 percent), age 16 (n = 619; 12.33 percent), and age 17 (n = 4,133; 82.35 percent). Also, whether counties were in states with sentencing guidelines was included (n = 784; 15.62 percent), as this likely influences sentencing outcomes.

Analytical Strategy

Due to the nested nature of the data, multilevel modeling is used to estimate the effects of attorney type on conviction, sentencing decision, and sentence length for juvenile defendants in criminal court. For the binary conviction outcome, random coefficients logistic regression was employed, while random coefficient multinomial logistic regression was employed for the trichotomous sentencing outcome. Random coefficients linear regression (using maximum likelihood estimator) was employed for jail and prison length, where sentence lengths were log transformed due to violation of linearity by positive skew (i.e., log-linear model).[11] Measures of concentrated disadvantage were standardized for interpretation purposes. Continuous control variables (age and crime rate) were grand mean centered. Correlations between variables were generally weak to moderate (< .5) and tests for multicollinearity (variance inflation factor [VIF]) indicated no serious concerns across all models (VIF < 2.5).[12]

For all three outcomes, a parallel analytical strategy was adopted. Unconditional models were first estimated, followed by random intercept and random coefficient models (see Johnson, 2010). In the random coefficient models, attorney type was designated as a random effect, and cross-level interactions between attorney type (whose slope was allowed to vary across counties) and urbanism were estimated.[13] In the context of a nonlinear dependent variable, interaction effects become more complicated and the product term in regression output does not represent a sufficient test of the interaction (see Mustillo, Lizardo, & McVeigh, 2018; for more details, see Mize, 2019). Following Mize (2019), we calculated predicated probabilities and marginal effects in post-estimation for the conviction and sentence type models. We then tested second differences in marginal effects of attorney type across different levels of urbanism from 0 to 70,000 persons per square mile (bounded by the range in the data). Specifically, for each sentencing outcome, we estimated the difference in the marginal effect of attorney type across levels of urbanism (in cutoffs of 5,000)—a total of 168 comparisons.[14] For continuous dependent variables (i.e., sentence length), the product term accurately captures the interaction effect. All analyses used 2-sided hypothesis testing with an a of .05 and were performed using Stata 15.

Results

Conviction

To assess whether conviction outcomes vary significantly by county, an unconditional multilevel logistic regression model was first estimated. The analysis (not shown) indicates that there is significant variation in conviction odds across counties, with a level-2 variance component (ψ) of 1.13 and intraclass correlation (ρ) of .26.[15] This indicates that before any predictors are added to the model, approximately 26 percent of the variation in conviction outcomes is attributable to between-county differences. As expected, a multilevel modeling strategy is justified.

To examine the effects of attorney type on conviction outcomes, a series of multilevel logit models were next estimated (see Table 2). Model 1 includes the full random coefficients model with attorney type designated as a random effect. Model 2 includes the cross-level interaction between attorney type and urbanism. Results of model 1 indicated that attorney type was partially associated with conviction outcomes for juveniles in criminal court (see table 2, model 1). Specifically, defendants with assigned counsel were 34 percent more likely to be convicted than defendants with public defenders (OR = 1.34), while there were no significant differences in odds of conviction between public defenders and private counsel.

Urbanism was not significantly associated with conviction likelihood, and cross-level interactions between attorney type and urbanism did not reach statistical significance. Following the random coefficients model with the interaction term (i.e., model 2), predicted probabilities and marginal effects of attorney type on conviction were estimated (full results available upon request). Tests of second differences revealed no significant differences in the marginal effects of attorney type at different levels of urbanism, indicating that the influence of attorney type on conviction outcomes was not conditioned by level of urbanism. 

[Table 2 about here] 

Sentence Type  

To assess whether sentencing outcomes vary significantly by county, an unconditional multilevel multinomial logistic regression model was first estimated. The analysis (not shown) indicates that there is significant variation in likelihood of sentence type across counties, with a level-2 variance component (ψ) of 3.74 (ρ = .53) for jail sentence (versus community) and a level-2 variance component (ψ) of 2.46 (ρ = .43) for prison sentence (versus community). This indicates that before any predictors are added to the model, approximately 53 percent of the variation in jail sentences and 43 percent of the variation in prison sentences (compared to community sentences) are attributable to between-county differences. As expected, a multilevel modeling strategy is justified.

To examine the effects of attorney type on sentencing outcomes, a series of multilevel logit models were next estimated (see table 3). Model 1 includes the full random coefficients model with attorney type designated as a random effect, while model 2 includes the cross-level interaction between attorney type and urbanism. Results of model 1 indicated that attorney type was partially associated with sentence type for juveniles in criminal court (see table 2, model 1). While there were no significant differences in jail versus community sentences, defendants with assigned counsel were 96 percent more likely to be sentenced to prison (compared to probation) than defendants with public defenders (OR = 1.96). (There were no significant differences in type of sentence between public defenders and private counsel.)

Urbanism was not significantly associated with conviction likelihood, and cross-level interactions between attorney type and urbanism largely did not reach statistical significance. Following the random coefficients model with the interaction term (i.e., model 2), predicted probabilities and marginal effects of attorney type on sentence type were estimated (full results available upon request). While the product terms in table 3 show no significant interaction, tests of second differences revealed several significant differences in the marginal effects of attorney type at extreme levels of urbanism. Specifically, there were significant differences between the probabilities of jail sentences for assigned counsel and public defenders at lower versus high population density. At the lowest possible urbanism (i.e., 0 persons per square mile,), assigned counsel are associated with a .19 probability of jail sentence while public defenders are associated with a .13 probability of jail sentence, a significant marginal difference (i.e., -.06). At high levels of urbanism (> 50,000 persons per square mile), however, the probability of jail sentence is .01 for assigned counsel and .00 for public defenders, a (nonsignificant) difference of .01. The second difference between these marginal effects (i.e., -.06 and .01) was significant, indicating that assigned counsel were more likely to receive jail sentences than public defenders at the lowest levels of urbanism compared to highest levels of urbanism. All other second differences were nonsignificant, however.

[Table 3 about here]

Jail Length

To assess whether jail sentence length varies significantly by county, an unconditional multilevel linear regression model was first estimated. The analysis (not shown) indicates that there is significant variation in jail length across counties, with a level-1 residual variance of .76 and a level-2 variance component of .10 (ρ = .12). This indicates that before any predictors are added to the model, approximately 12 percent of the variation in jail length is attributable to between-county differences. As expected, a multilevel modeling strategy is justified.

To examine the effects of attorney type on jail length, a series of multilevel linear regression models were estimated (see table 3). Model 1 includes the full random coefficients model with attorney type designated as a random effect. Model 2 includes the cross-level interaction between attorney type and urbanism. Results of model 1 indicated that attorney type was not associated with jail length for juveniles in criminal court (see table 4, model 1).

Results of model 2 indicate that while urbanism was not significantly associated with length of jail sentence, the influence of attorney type was partially conditioned by urbanism. Specifically, cross-level interactions between assigned counsel and urbanism reached statistical significance, indicating that in more urban counties, defendants with assigned counsel received slightly shorter jail sentences than defendants with public defenders (Exp(b) = .94). (Notably, this is not the direction of the effect anticipated by H3b). The influence of private counsel on jail sentence length was not conditioned by urbanism, however. 

[Table 4 about here]

Prison Length

To assess whether prison sentence length varies significantly by county, an unconditional multilevel linear regression model was first estimated. The analysis (not shown) indicates that there is significant variation in prison length across counties, with a level-1 residual variance of .73 and a level-2 variance component of .23 (ρ = .23). This indicates that before any predictors are added to the model, approximately 23 percent of the variation in prison length is attributable to between-county differences. As expected, a multilevel modeling strategy is justified.

To examine the effects of attorney type on prison length, a series of multilevel linear regression models were estimated (see table 3). Model 1 includes the full random coefficients model with attorney type designated as a random effect (a likelihood ratio test confirmed that a random coefficient model improved model fit compared to a random intercept model). Model 2 includes the cross-level interaction between attorney type and urbanism. Results of model 1 indicated that attorney type was associated with prison length for juveniles in criminal court (see table 4, model 1). Specifically, length of prison sentence was 18 percent longer for defendants with assigned counsel (Exp(b) = 1.18), and 20 percent longer for defendants with private counsel (Exp(b) = 1.20), compared to public defenders. Results of model 2, however, indicate that urbanism was not significantly associated with length of prison sentence, and that the influence of attorney type was not conditioned by urbanism. Specifically, cross-level interactions between attorney type and urbanism did not reach statistical significance.  

[Table 5 about here]

Discussion and Conclusions

The research hypotheses were partially supported. For each criminal court outcome, there was some support for the influence of attorney type in the expected direction. First, H1a was partially supported, as juvenile defendants with assigned counsel were more likely to be convicted than defendants with public defenders (but with no differences between private counsel and public defenders). Second, incarceration was partially associated with attorney type (consistent with H2a). While type of legal representation was not associated with odds of jail (compared to community sentence), prison sentences were more likely for defendants represented by assigned counsel than public defenders (but with no differences between private counsel and public defenders). Third, jail sentence length was not associated with type of attorney until the interaction term was added to the model (see below). Fourth, prison sentence length was associated with attorney type, with longer prison sentences for defendants with assigned counsel and private counsel compared to public defenders (i.e., support for H4a), but no differences between private and assigned counsel. In sum, defendants with public defenders tended to receive better outcomes in terms of conviction, prison sentences, and prison sentence length, compared to defendants with assigned counsel, while differences between public defenders and private counsel only showed up for length of prison sentence.

However, and contrary to the second set of hypotheses, these attorney effects were not conditioned by urbanism. The only exception was length of jail sentence, where attorney type was not related to jail length until an interaction between jail and urbanism was added to the model, at which point assigned counsel was positively associated with longer jail sentences (contrary to H3a) and negatively moderated by urbanism (contrary to H3b). The overall takeaway is that urbanism did not moderate the influence of attorney type in the expected direction. On the one hand, this might indicate that public defender systems are not in fact more effective in more urban, routinized courts (as we expected). On the other hand, our sample included only urban counties (as defined by the Census), ranging from low density urban (90 persons per square mile) to high density urban (66,940 persons per square mile). It may just be that among urban courts where such routinization is already present, higher population density does not translate into greater routinization and hence higher effectiveness. It could still be possible that public defenders in non-urban counties are less effective. Future research using urban and non-urban samples could delineate between these possible interpretations (cf. Worden et al., 2019).

Before discussing these results, several study limitations should be addressed. The first involves the measurement of the main independent variable—attorney type. Since assigned counsel is only constitutionally required for indigent defendants, others have worried that attorney type may function as a proxy for class (Hermann et al., 1977; Nagel, 1973; Shinall, 2010). This presents a possible selection bias; that is, reported effects of attorney type may actually be class effects. In the current study, however, we were able to distinguish between two types of attorney provided to indigent defendants, and compare them to each other as well as to private retained counsel. The results show significant differences in conviction and sentencing outcomes between public defenders and assigned counsel (as predicted), a result that is difficult to explain in terms of class effects. Additionally, while the present study was able to distinguish between private (retained) attorneys and two types of appointed counsel, there may nonetheless be other attorney qualities that are relevant to assessing effective assistance of counsel, such as experience (Abrams & Yoon, 2007; but see Miller, Keith, & Holmes, 2015). It may also be that courtroom workgroup routinization varies across jurisdictions in subtle ways that are difficult to capture in terms of structural variables such as urbanism (Myers & Reid, 1995). As Ulmer (2012, p. 34) observes, administrative data cannot “directly tell us what was going on in courtroom workgroup members’ heads, or the content of their interactions with each other.” Future research should address these issues, possibly through more qualitative approaches (see, e.g., Metcalfe, 2016). 

A second limitation, and related to the last point, is our measure of effectiveness. A large proportion of cases are negotiated; in the current sample, 92 percent of convictions are the result of guilty pleas rather than trial. In addition to conviction and sentencing outcomes, downward charging may be seen as effective in avoiding a harsher penalty. Our measures of effective counsel in terms of conviction, incarceration, and sentence length outcomes do not include this downward charging because it was not available in the data, and more research is needed to define effectiveness and the best defense for juveniles in the adult criminal justice system. It is worth noting here that among convicted defendants, bivariate associations revealed that those with public defenders were significantly more likely to enter a guilty plea (93 percent) than those with assigned counsel (89 percent) or private counsel (91 percent). It may be that public defenders’ increased effectiveness is in part a function of having more cooperative plea negotiations with the prosecutor and judge, a point discussed further below.

A final limitation is the age of the dataset. The present study was motivated by the importance of legal representation for juveniles in criminal court, the generalizability of multijurisdictional sampling, and the need to distinguish among public defenders, assigned counsel, and private counsel. The JDCC represents the most recent publicly available dataset that fits these research needs. The multi-jurisdictional nature of the dataset is particularly important, as arriving at general conclusions about effectiveness of different types of counsel requires moving beyond single state (or county) analyses (Worden, 1993). In the current analysis, 37 separate criminal justice systems—across 19 states—are included, permitting the testing of general hypotheses about the effectiveness of different counsel systems. Still, an important research priority going forward is to enhance current data collection efforts so that researchers can evaluate the impact of attorney type on juveniles in criminal court today (Soler, 2010). This is especially important in light of the increased attention to juveniles in criminal court, with the U.S. Supreme Court recently holding certain punishments to be unconstitutional because juveniles were, as a class, viewed as different and less culpable than adult defendants (e.g., Graham v. Florida, 2010; Miller v. Alabama, 2012; Roper v. Simmons, 2005).

Implications for Research and Policy

The present study’s findings should stimulate others to look more carefully at how juveniles can receive the best possible defense in criminal court. The conviction of juveniles in criminal court has considerable implications for their development as law-abiding adults. A criminal court conviction (and possibly incarceration) means that a juvenile will have a criminal record and will no longer be eligible for a wide range of occupational and educational pursuits. It could be a turning point in an adolescent’s life, leading to chronic adult criminality (Monahan et al., 2015; Scott, Duell, & Steinberg, 2018). The juvenile court was established to save adolescents from a criminal label (Tanenhaus, 2004), yet with the expanded shape of waiver today there are thousands of juveniles who are routinely brought into criminal court—subject to possible labeling effects (Myers, 2016). Access to quality legal representation for youth in adult court (and as early as initial contact with police; see Feld, 2013) is thus critically important. 

The present study found that attorney type was associated with odds of conviction, incarceration in prison, and length of prison sentence. In each case, defendants with public defenders received better outcomes than defendants with assigned counsel, as well as shorter prison sentences than defendants with private counsel. This suggests that public defenders—often depicted as less effective due to lower qualifications (e.g., Rattner et al., 2008) or severe resource constraints (e.g., Bibas, 2013; Pfaff, 2017)—are indeed as effective, and possibly more effective, than other forms of legal representation for juveniles in criminal court. Also consistent with our theoretical expectations, we found that assigned counsel, an alternative form of representation for indigent defendants, presents the least effective form of legal representation. Interestingly, recent research has also indicated that assigned counsel may be less cost-effective than an organized public defender office (see Davies & Worden, 2017). Our expectations that these effects would be greater in more urbanized courts were largely unsupported, however, possibly because the sample included only large, urban counties (i.e., no comparison to rural counties). Future research should directly compare urban and rural counties to assess the influence of urbanization.

In terms of policy implications, greater investment in public defender offices might result in even more effective legal representation for juveniles in adult court. According to Pfaff (2017), state and local governments spent almost 30 percent less on indigent defense (including both public defender systems and assigned counsel) than on prosecutors’ offices in 2007–2008. As part of a larger argument about reducing mass incarceration by reigning in the power of prosecutors, Pfaff (2017, p. 207) notes that, “Perhaps the most effective way to regulate prosecutors would be to adequately fund public defenders and other indigent counsel.” In addition to devoting more resources to indigent defense more generally, the current findings suggest that more of these resources should be devoted to public defender systems specifically—rather than assigned counsel.[16] Especially in light of the significant (and growing) power of prosecutors to steer case outcomes via charging decisions and plea bargain negotiations (Davis 2007), providing more resources to public defenders would seem to represent an important step forward in providing juveniles with the best possible defense in criminal court.

Nevertheless, more research is needed to understand the effectiveness of counsel. What constitutes a good defense? In juvenile court this will often involve rehabilitative services or diversionary programs that can divert a troubled youth away from a life of crime. In a criminal court context, however, we have defined the best possible defense in terms of less contact with the criminal justice system—whether through non-conviction, a non-incarceration sentence, or shorter sentence for those who are incarcerated. Future research could move beyond this and explore types of public defender systems in different jurisdictions, and how those agencies may be more effective in enabling charged juveniles to avoid a criminal court conviction (or divert them away from formal processing). The best possible defense for juveniles in criminal court will also account for the developmental needs of adolescents (Monahan et al., 2015), something that larger, more routinized public defender offices may be in the best position to do (and assigned counsel are likely not). Our findings suggest that policymakers interested in improving legal representation for juvenile offenders in criminal court (and more generally for indigent defendants) should focus on greater investment in public defender systems and transition from assigned counsel systems to public defender systems.

References 

Abrams, D. S., & Yoon, A. H. (2007). The luck of the draw: Using random case assignment to investigate attorney ability. University of Chicago Law Review, 74, 1145–1177.

Anderson, J. M., & Heaton, P. (2012). How much difference does the lawyer make: The effect of defense counsel on murder case outcomes. The Yale Law Journal, 122, 154­–217.

Armstrong, G. S., & Kim, B. (2011). Juvenile penalties for ‘lawyering up’: The role of counsel and extralegal case characteristics. Crime & Delinquency, 57, 827–848.

Beck, J. C., & Shumsky, R. (1997). A comparison of retained and appointed counsel in cases of capital murder. Law and Human Behavior, 21, 525–538.

Bernard, T. J., & Kurlychek, M. C. (2010). The cycle of juvenile justice (2nd ed.). New York, NY: Oxford University Press.

Bibas, S. (2013). Shrinking Gideon and expanding alternatives to lawyers. Washington & Lee Law Review, 70, 1287–1308.

Blumberg, A. S. (1967). The practice of law as a confidence game: Organizational cooptation of a profession. Law & Society Review, 1, 15–40.

Bray, T. M., Sample, L. L., & Kempf-Leonard, K. (2005). Justice by geography: Racial disparity and juvenile courts. In D. F. Hawkins & K. Kempf-Leonard (eds.), Our children, their children: Confronting racial and ethnic differences in American juvenile justice (270–299). Chicago, IL: University of Chicago Press.

Burrus, G. W., & Kempf-Leonard, K. (2002). The questionable advantage of defense counsel in juvenile court. Justice Quarterly, 19, 37–68.

Bushway, S. D, Johnson, B. D., & Slocum, L. S. (2007). Is the magic still there? The use of the Heckman two-step correction for selection bias in criminology. Journal of Quantitative Criminology, 23, 151–178.

Burnett, C. G., Moore, M. L., & Butcher, A. K. (2001). In pursuit of independent, qualified, and effective counsel: The past and future of indigent criminal defense in Texas. South Texas Law Review, 42, 595–693.

Cho, H. C., & Abe, S. (2013). Is two-tailed testing for directional research hypotheses test legitimate? Journal of Business Research, 66, 1261–1266.

Cohen, T. H. (2014). Who is better at defending criminals? Does type of defense attorney matter in terms of producing favorable case outcomes. Criminal Justice Policy Review, 25, 29–58.

Cole, G. F. (1969). The decision to prosecute. Law & Society Review, 4, 331–344.

Davis, A. J. (2007). Arbitrary justice: The power of the American prosecutor. New York, NY: Oxford University Press.

Davies, A. L. B, & Worden, A. P. (2017). Local governance and redistributive policy: Explaining local funding for public defense. Law & Society Review, 51, 331–345.

Davies, A. L. B., & Worden, A. P. (2009). State politics and the right to counsel: A comparative analysis. Law & Society Review, 43, 187–220.

Dixon, J. (1995). The organizational context of criminal sentencing. American Journal of Sociology, 100, 1157–1198.

Dripps, D. A. (2013). Why Gideon failed: Politics and feedback loops in the reform of criminal justice. Washington & Lee Law Review, 70, 883–925.

Eisenstein, J., & Jacob, H. (1977). Felony justice: An organization analysis of criminal courts. Boston, MA: Little, Brown & Co.

Feld, B. C. (2017). The evolution of the juvenile court: Race, politics, and the criminalizing of juvenile justice. New York, NY: New York University Press.

Feld, B. C. (2013). Real interrogation: What actually happens when cops question kids. Law & Society Review, 47, 1–35.

Feld, B. C. (1991). Justice by geography: urban, suburban, and rural variations in juvenile justice administration. Journal of Criminal Law and Criminology, 82, 156–210.

Feeley, M. M. (1972). Two models of the criminal justice system: An organizational perspective. Law & Society Review, 7, 407–426.

Feeney, F., & Jackson, P. G. (1991). Public defenders, assigned counsel, retained counsel: Does the type of criminal defense counsel matter? Rutgers Law Review, 22, 361–456.

Felice, D. A. (2001). Justice rationed: A look at Alabama’s present indigent defense system with a vision toward change. Alabama Law Review, 52, 975–1002.

Frederique, N., Joseph, P., & Hild, R. C. C. (2015). What is the state of empirical research on indigent defense nationwide? A brief overview and suggestions for future research. Albany Law Review, 78, 1317–1340.

Gitelman, M. (1970). The relative performance of appointed and retained counsel in Arkansas felony cases: An empirical study. Arkansas Law Review, 24, 442−452.

Griffin, P., Addie, S., Adams, B., & Firestine, K. (2011). Trying juveniles as adults: An analysis of state transfer laws and reporting. Washington, DC: U.S. Department of Justice, Office of Juvenile Justice and Delinquency Prevention.

Gross, J. P. (2013). Too poor to hire a lawyer but not indigent: How states use the federal poverty guidelines to deprive defendants of their Sixth Amendment right to counsel. Washington & Lee Law Review, 70, 1173–1219.

Guevara, L., Spohn, C., & Herz, D. (2004). Race, legal representation, and juvenile justice: Issues and concerns. Crime & Delinquency, 50, 344–371.

Hartley, R. D., Miller, H. V., & Spohn, C. (2010). Do you get what you pay for? Type of counsel and its effect on criminal court outcomes. Journal of Criminal Justice, 38, 1063–1070.

Hermann, R., Single, E., & Boston, J. (1977). Counsel for the poor: Criminal defense in urban America. Lexington, MA: Lexington Books.

Hester. R. (2017). Judicial rotation as centripetal force: Sentencing in the court communities of South Carolina. Criminology, 55, 205–235.

Hockenberry, S., & Puzzanchera, C. (2015). Juvenile court statistics, 2013. Pittsburgh, PA: National Center for Juvenile Justice.

Hoffman, M. B., Rubin, P. H., & Shepherd, J. M. (2005). An empirical study of public defender effectiveness: Self-selection by the marginally indigent. Ohio State Journal of Criminal Law, 3, 223–255.

Holleran, D., & Spohn, C. (2004). On the use of the total incarceration variable in sentencing research. Criminology, 42, 211–240.

Johnson, B. D. (2019). Trials and tribulations: The trial tax and the process of punishment. Crime & Justice, 48, 313–363.

Johnson, B. D. (2010). Multilevel analysis in the study of crime and justice. In A. R. Piquero & D. Weisburd (Eds.), Handbook of quantitative criminology (pp. 615–648). New York, NY: Springer.

Lucas, L. S. (2005). Effectively ineffective: The failure of courts to address underfunded indigent defense systems. Harvard Law Review, 118, 1731–52.

MacDonald, J. M., & Lattimore, P. K. (2010). Count models in criminology. In A. R. Piquero & D. Weisburd (Eds.), Handbook of quantitative criminology (pp. 683–698). New York, NY: Springer.

Martinez, P., & Pollock, J. M. (2008). The impact of type of attorney on criminal sentencing. Criminal Law Bulletin, 44, 652–678.

Metcalfe, C. (2016). The role of courtroom workgroups in felony case dispositions: An analysis of workgroup familiarity and similarity. Law & Society Review, 50, 637–673.

Miller, B., Keith, L. C., & Holmes, J. S. (2015). Leveling the odds: The effect of quality legal representation in cases of asymmetrical capability. Law & Society Review, 49, 1–239.

Mize, T. D. (2019). Best practices for estimating, interpreting, and presenting nonlinear interaction effects. Sociological Science, 6, 81–117.

Mustillo, S. A., Lizardo, O. A., & McVeigh, R. M. (2018). Editors’ comments: A few guidelines for quantitative submissions. American Sociological Review, 83, 1281–1283.

Myers, D. L., et al. (2011) “Practitioner Perceptions of Juvenile Transfer in Pennsylvania,” 9 Youth Violence and Juvenile Justice 222–240.

Model Rules of Professional Conduct, Rule 1.7—Conflict of Interest: Current Clients (2018).

Model Rules of Professional Conduct, Rule 1.8—Conflict of Interest: Special Rules (2018).

Moore, N. J. (1996). Conflicts of interests in the representation of children. Fordham Law Review, 64, 1819–1856.

Monahan, K., Steinberg, L., & Piquero, A. R. (2015). Juvenile justice policy and practice: A developmental perspective. Crime and Justice, 44, 557–619.

Mosteller, R. P. (2013). Potential innocence: Making the most of a bleak environment for public support of indigent defense. Washington & Lee Law Review, 70, 1345–1362.

Myers, D. L. (2016). Juvenile transfer to adult court: Ongoing search for scientific support. Criminology & Public Policy, 15, 927–938.

Myers, L. B., & Reid, S. T. (1995). The importance of county context in the measurement of sentence disparity: The search for routinization. Journal of Criminal Justice, 23, 223–241.

Nagel, S. S. (1973). Effects of alternative types of counsel on criminal procedural treatment. Indiana Law Journal, 48, 404−442.

Nardulli, P. F. (1986). Insider justice: Defense attorneys and the handling of felony cases. Journal of Criminal Law and Criminology, 77, 379−417.

Nardulli, P. F., Eisenstein, J., & Flemming, R. B. (1988). The tenor of justice: Criminal courts and the guilty plea process. Chicago, IL: University of Illinois Press.

Packer, H. (1968). The limits of the criminal sanction. Stanford, CA: Stanford University Press.

Paternoster, R., Brame, R., Mazerolle, P., & Piquero, A. R. (1998). Using the correct statistical test for the equality of regression coefficients. Criminology, 36, 859–866.

Pfaff, J. (2017). Locked in: The true causes of mass incarceration and how to achieve real reform. New York, NY: Basic Books.

Rattner, A., Turjeman, J., & Fishman, G. (2008). Public versus private defense: Can money buy justice? Journal of Criminal Justice, 36, 43–49.

Rodriguez, N. (2010). The cumulative effect of race and ethnicity in juvenile court outcomes and why preadjudication detention matters. Journal of Research in Crime and Delinquency, 47, 391–413.

Rydberg, J., & Carkin, D. M. (2017). Utilizing alternate models for analyzing court outcomes. Crime & Delinquency, 63, 61–76.

Sandefur, R. L. (2015). Elements of professional expertise: Understanding relational and substantive expertise through lawyers’ impact. American Sociological Review, 909–933.

Scott, E., Duell, N., & Steinberg, L. (2018). Brain development, social context, and justice policy. Washington University Journal of Law & Policy, 57, 13–74.

Shinall, J. B. (2010). Slipping away from justice: The effect of attorney skill on trial outcomes. Vanderbilt Law Review, 63, 267–306.

Singer, S. (1997). Recriminalizing delinquency: Violent juvenile crime and juvenile justice reform. Cambridge, UK: Cambridge University Press.

Steiner, B. (2009). The effect of juvenile transfer to criminal court on incarceration decisions. Justice Quarterly, 26, 77–106.

Soler, M. (2010). Missed opportunity: Waiver, race, data, and policy reform. Louisiana Law Review, 17–33.

Spohn, C. (2018). Reflections on the exercise of prosecutorial discretion 50 years after publication of The Challenge of Crime in a Free Society. Criminology & Public Policy, 17, 321–340.

Spohn, C., Gruhl, J., & Welch, S. (1981). The effect of race on sentencing: A re-examination of an unsettled question. Law & Society Review, 16, 71–88.

Sterne, J. A., White, I. R., Carlin, J. B., Spratt, M., Royston, P., Kenward, M. G., Wood, A. M., & Carpenter, J. R. (2009). Multiple imputation for missing data in epidemiological and clinical research: Potential and pitfalls. BMJ, 338, p.b2393.

Stolzenberg, R. M., & Relles, D. A. (1997). Tools for intuition about sample selection bias and its correction. American Sociological Review, 62, 494–507.

Tanenhaus, D. S. (2004). Juvenile justice in the making. New York, NY: Oxford University Press.

Taylor, J. G., Stanley, T P., DeFlorio, B. J., & Seekamp, L. N. (1972). An analysis of defense counsel in the processing of felony defendants in San Diego, California. Denver Law Journal, 49, 233–275.

Ulmer, J. T. (2012). Recent developments and new directions in sentencing research. Justice Quarterly, 29, 1–40.

Weber, M. (1954). Max Weber on law in economy and society (trans. Max Rheinstein). Cambridge, MA: Harvard University Press. 

Wheeler, G. R., & Wheeler, C. L. (1980). Reflection on legal representation of the economically disadvantaged: Beyond assembly line justice. Crime & Delinquency, 26, 319−332.

Williams, M. R. (2017). The effect of attorney type on bail decisions. Criminal Justice Policy Review, 28, 3–17.

Williams, M. R. (2013). The effectiveness of public defenders in four Florida counties. Journal of Criminal Justice, 41, 205–212.

Williams, M. R. (2002). A comparison of sentencing outcomes for defendants with public defenders versus retained counsel in a Florida circuit court. Justice System Journal, 23, 249–257.

Willison, D. (1984). The effects of counsel on the severity of criminal sentences: A statistical assessment. Justice System Journal, 9, 87–101.

Woolard, J. L., Odgers, C., Lanza-Kaduce, L., & Daglis, H. (2005). Juveniles within correctional settings: Legal pathways and developmental considerations. International Journal of Forensic Mental Health, 4, 1–18.

Worden, A. P. (1993). Counsel for the poor: An evaluation of contracting for indigent criminal defense. Justice Quarterly, 10, 613–637.

Worden, A. P. (1991). Privatizing due process: Issues in the comparison of assigned counsel, public defender, and contracted indigent systems. Justice System Journal, 15, 390–418.

Worden, A. P., Shteynberg, R. V., Morgan, K. A., & Davies, A. L. B. (2019). The impact of counsel at first appearance on pretrial release in felony arraignments: The case of rural jurisdictions. Criminal Justice Policy Review. Doi: 10.1177/0887403419873018.

Zane, S. N. (2017). Do criminal court outcomes vary by juvenile transfer mechanism? A multi-jurisdictional, multilevel analysis. Justice Quarterly, 34, 542–569.

Zimring, F. E. (2010). The power politics of juvenile court transfer: A mildly revisionist history of the 1990s. Louisiana Law Review, 71, 1–15.

Cases

Argersinger v. Hamlin, 407 U.S. 25 (1972)

Gideon v. Wainwright, 372 U.S. 335 (1963)

Graham v. Florida, 560 U.S. 48 (2010)

In re Gault, 387 U.S. 1 (1967)

Johnson v. Zerbst, 304 U.S. 458 (1938)

Kent v. United States, 383 U.S. 541 (1966)

Miller v. Alabama, 587 U.S. 460 (2012).

Powell v. Alabama, 287 U.S. 45 (1932)

Roper v. Simmons, 541 U.S. 1040 (2005)

Tables

Table 1. Descriptive statistics

 

Conviction

(N1 = 5,019; N2 = 37)

Sentence Type

(N1 = 3,691; N2 = 37)

Jail sentence

(N1 = 717; N2 = 26)

Prison sentence

(N1 = 1,350; N2 = 36)

Dependent variables

n (%)

n (%)

n (%)

n (%)

Conviction

3,775 (75.21)

-

-

-

Sentence type

 

 

 

 

     Community

-

1,116 (31.59)

-

-

     Jail

-

768 (20.81)

-

-

     Prison

-

1,757 (47.60)

-

-

Jail length, mean days (SD)

-

-

208.50 (130.86)

-

Prison length, mean days (SD)

-

-

-

3037.60 (3791.02)

Case-Level variables

 

 

 

 

Attorney type

 

 

 

 

     Public defender

3,199 (63.74)

2,345 (63.53)

530 (73.92)

788 (54.01)

     Private retained counsel

910 (18.13)

670 (18.15)

100 (13.92)

381 (26.11)

     Assigned counsel

910 (18.13)

676 (18.31)

87 (12.13)

290 (19.68)

Race/ethnicity

 

 

 

 

     White

956 (19.05)

755 (20.45)

189 (26.36)

201 (13.78)

     Black

2,974 (59.25)

2,097 (56.81)

328 (45.75)

844 (57.85)

     Hispanic

987 (19.67)

754 (20.43)

181 (25.24)

369 (25.29)

     Other race/ethnicity

102 (2.03)

85 (2.30)

19 (2.65)

45 (3.08)

Female

148 (2.95)

98 (2.66)

16 (2.23)

37 (2.54)

Age, mean years (SD)

16.32 (.99)

16.35 (.99)

16.55 (.86)

16.32 (1.02)

Offense Type

 

 

 

 

     Murder

179 (3.57)

121 (3.28)

-

108 (7.40)

     Rape

118 (2.35)

65 (1.76)

5 (.70)

44 (3.02)

     Assault

1,112 (22.16)

718 (19.45)

142 (19.80)

360 (24.67)

     Robbery

1,325 (26.40)

957 (25.93)

82 (11.44)

558 (38.25)

     Burglary

466 (9.28)

392 (10.62)

95 (13.25)

130 (8.91)

     Larceny

316 (6.30)

267 (7.23)

85 (11.85)

59 (4.04)

     Motor vehicle theft

167 (3.33)

118 (3.20)

41 (5.72)

59 (4.04)

     Drug/Alcohol

843 (16.80)

644 (17.45)

139 (19.39)

111 (7.61)

     Weapon offense

153 (3.05)

109 (2.95)

23 (3.21)

20 (1.37)

     Other felony

138 (2.75)

114 (3.09)

41 (5.72)

21 (1.44)

     Misdemeanor

202 (4.02)

186 (5.04)

64 (8.93)

27 (1.85)

Multiple charges

3,343 (66.61)

2,454 (66.49)

447 (62.34)

1,033 (70.80)

Rearrest

313 (6.24)

257 (6.96)

46 (6.42)

80 (5.48)

Pretrial Detention

2,600 (51.80)

2,015 (54.59)

337 (47.00)

1,150 (78.82)

Judicial waiver

1,398 (27.85)

1,134 (30.72)

183 (25.52)

642 (44.00)

Direct file

2,099 (41.82)

1,611 (43.65)

468 (65.27)

404 (27.69)

Statutory exclusion

1,522 (30.32)

946 (25.63)

66 (9.21)

413 (27.69)

Guilty plea

3,488 (69.50)

3,407 (92.31)

704 (98.19)

1,267 (86.84)

Contextual variables

 

 

 

 

Urbanism (1000 persons/sq mi), mean (SD)

4.23 (13.09)

3.67 (13.08)

1.22 (8.67)

4.06 (13.08)

Concentrated disadvantage, mean (SD)

0 (2.19)

0 (2.28)

0 (3.04)

0 (2.20)

Crime rates (per 1000 persons), mean (SD)

71.04 (20.58)

70.07 (20.58)

75.21 (21.72)

64.51 (20.58)

Sentencing guidelines

784 (15.62)

424 (11.49)

45 (6.28)

227 (15.56)

Upper age of jurisdiction: 15

267 (5.32)

197 (5.34)

5 (.70)

77 (5.28)

Upper age of jurisdiction: 16

619 (12.33)

473 (12.81)

27 (3.77)

233 (15.97)

Upper age of jurisdiction: 16

4,133 (82.35)

3,021 (81.85)

685 (95.54)

1,149 (78.75)

Table 2. Random coefficients logistic regression, conviction outcome (N1 = 5,019; N2 = 37)

Variable

Model 1

Model 2

 

OR (95% CI)

OR (95% CI)

Intercept

 

1.17 (.66, 2.06)

Level-1 variables

 

 

Assigned counsela

1.34* (1.05, 1.72)

1.34* (1.02, 1.76)

Private counsela

1.21 (.95, 1.52)

1.16 (.88, 1.53)

Blackb

1.04 (.83, 1.30)

1.04 (.83, 1.30)

Hispanicb

.98 (.75, 1.28)

.98 (.75, 1.28)

Other race/ethnicityb

1.27 (.65, 2.46)

1.27 (.65, 2.47)

Female

.95 (.62, 1.46)

.95 (.62, 1.46)

Agec

1.11* (1.02, 1.22)

1.11* (1.02, 1.22)

Murderd

1.26 (.84, 1.90)

1.26 (.84, 1.90)

Raped

.58* (.37, .91)

.58* (.37, .92)

Robberyd 

1.64*** (1.32, 2.03)

1.64*** (1.32, 2.03)

Burglaryd

2.47*** (1.79, 3.44)

2.48*** (1.79, 3.44)

Larcenyd

2.14*** (1.45, 3.14)

2.14* (1.45, 3.14)

Motor vehicle theftd

1.64* (1.08, 2.49)

1.64* (1.08, 2.49)

Drug/Alcohold

2.19*** (1.67, 2.86)

2.19*** (1.67, 2.87)

Weapon offensed

2.03** (1.31, 3.12)

2.03** (1.32, 3.12)

Other felonyd

1.62 (.96, 2.73)

1.62 (.97, 2.73)

Misdemeanord

6.86*** (3.76, 12.50)

6.88*** (3.77, 12.54)

Multiple charges

1.38*** (1.16, 1.63)

1.38*** (1.16, 1.63)

Rearrest

2.31*** (1.58, 3.38)

2.31*** (1.58, 3.37)

Pretrial Detention

1.77*** (1.49, 2.10)

1.77*** (1.49, 2.10)

Direct filee

1.36 (.86, 2.14)

1.35 (.86, 2.13)

Statutory exclusione

.97 (.69, 1.38)

.97 (.68, 1.38)

Level-2 variables

 

 

Urbanism

.98 (.93, 1.03)

.98 (.93, 1.04)

Concentrated disadvantagef

.63* (.42, .92)

.62* (.42, .92)

Crime ratesc

.99 (.97, 1.01)

.99 (.97, 1.01)

Age 15g

21.15*** (1.75, 255.32)

19.50* (1.59, 239.77)

Age 16g

1.56 (.64, 3.78)

1.56 (.65, 3.79)

Cross-level interactions

 

 

Assigned counsel x urbanism

-

1.00 (.98, 1.02)

Private attorney x urbanism

-

1.01 (.97, 1.06)

Random effects (variance components)

 

 

Random intercept

.71

.72

Intraclass correlation (ρ)

.18

.18

Random slope, attorney type

.01

.01

Intraclass correlation (ρ)

.00

.00

* <.05 ** <.01 *** <.001

a Reference category: Public defender

b Reference category: White

c Grand mean centered

d Reference category: Assault (felony)

e Reference category: Judicial waiver

f Standardized variable

g Reference category: Age 17 jurisdiction

  

Table 3. Random coefficients multinomial logistic regression, sentence type (N1 = 3,691; N2= 37) 

Variable

Model 1

Model 2

 

Jail

Prison

Jail

Prison

 

OR

(95% CI)

OR

(95% CI)

OR

(95% CI)

OR

(95% CI)

Intercept

.25*

(.08, .78)

1.15

(.49, 2.70)

.26*

(.08, .81)

1.14

(.49, 2.68)

Level-1 variables

 

 

 

 

Assigned counsela

.96

(.70, 1.32)

1.96***

(1.48, 2.59)

.92

(.65, 1.31)

2.02***

(1.48, 2.76)

Private counsela

.85

(.61, 1.18)

1.09

(.82, 1.46)

.76

(.52, 1.11)

.98

(.71, 1.35)

Blackb

1.20

(.89, 1.60)

1.66***

(1.26, 2.19)

1.19

(.89, 1.60)

1.65***

(1.25, 2.18)

Hispanicb

1.18

(.84, 1.66)

1.63**

(1.17, 2.27)

1.18

(.84, 1.65)

1.61**

(1.16, 2.25)

Other race/ethnicityb

.82

(.36, 1.89)

.82

(.38, 1.77)

.81

(.36, 1.86)

.81

(.37, 1.74)

Female

.80

(.41, 1.56)

.65

(.37, 1.16)

.81

(.41, 1.57)

.67

(.38, 1.20)

Agec

1.02

(.90, 1.16)

.95

(.84, 1.07)

1.02

(.90, 1.16)

.95

(.84, 1.07)

Murderd

-

8.94**

(1.88, 42.60)

-

8.94

(1.87, 42.69)

Raped

.35

(.08, 1.43)

.98

(.45, 2.17)

.34

(.08, 1.43)

.99

(.45, 2.19)

Robberyd 

.52***

(.35, .77)

.85

(.62, 1.16)

.52***

(.35, .78)

.86

(.32, 1.17)

Burglaryd

1.14

(.77, 1.70)

.81

(.56, 1.17)

1.15

(.77, 1.72)

.82

(.57, 1.18)

Larcenyd

.89

(.58, 1.35)

.38***

(.25, .59)

.89

(.58, 1.35)

.38***

(.25, .59)

Motor vehicle theftd

.90

(.53, 1.54)

.44**

(.24, .82)

.90

(.53, 1.54)

.45**

(.24, .82)

Drug/Alcohold

1.05

(.72, 1.52)

.27***

(.19, .39)

1.05

(.72, 1.53)

.27***

(.19, .39)

Weapon offensed

.71

(.39, 1.29)

.16***

(.08, .30)

.71

(.39, 1.30)

.16***

(.08, .30)

Other felonyd

.85

(.49, 1.48)

.23***

(.12, .44)

.85

(.49, 1.48)

.23***

(.12, .44)

Misdemeanord

1.32

(.82, 2.12)

.08***

(.05, .15)

1.33

(.82, 2.14)

.08***

(.05, .15)

Multiple charges

.82

(.66, 1.04)

1.32***

(1.07, 1.64)

.83

(.66, 1.04)

1.33***

(1.07, 1.65)

Rearrest

1.39

(.91, 2.14)

2.13***

(1.44, 3.15)

1.39

(.90, 2.13)

2.11***

(1.43, 3.13)

Pretrial Detention

1.69***

(1.34, 2.13)

5.59***

(4.50, 6.93)

1.70***

(1.34, 2.14)

5.60***

(4.51, 6.94)

Direct filee

1.27

(.86, 1.88)

.95

(.61, 1.47)

1.27

(.86, 1.88)

.95

(.61, 1.47)

Statutory exclusione

1.40

(.83, 2.36)

1.65*

(1.05, 2.61)

1.40

(.83, 2.35)

1.64*

(1.04, 2.59)

Guilty plea

3.19***

(1.69, 6.04)

.77

(.52, 1.15)

3.21***

(1.69, 6.07)

.78

(.52, 1.16)

Level-2 variables (N = 37)

 

 

 

 

Urbanism

.91

(.77, 1.08)

.99

(.93, 1.06)

.89

(.74, 1.07)

.99

(.93, 1.06)

Concentrated disadvantagef

.80

(.38, 1.69)

1.10

(.67, 1.81)

.83

(.39, 1.75)

1.09

(.66, 1.81)

Crime ratesc

.97

(.93, 1.00)

.96**

(.93, .99)

.97

(.93, 1.00)

.96**

(.94, .99)

Sentencing guidelines

.82

(.18, 3.73)

1.85

(.57, 5.98)

.82

(.18, 3.72)

1.85

(.57, 6.00)

Cross-level interactions

 

 

 

 

Assigned counsel x urbanism

-

-

1.04

(.96, 1.13)

1.00

(.98, 1.02)

Private attorney x urbanism

-

-

1.05

(.95, 1.16)

1.03

(.99, 1.06)

Random effects

 

 

 

 

Random intercept

1.06

.82

1.05

.82

Intraclass correlation (ρ)

.24

.20

.24

.20

Random slope, attorney type

1.14

.82

1.12

.83

Intraclass correlation (ρ)

.26

.20

.25

.20

* <.05 ** <.01 *** <.001

a Reference category: Public defender

b Reference category: White

c Grand mean centered

d Reference category: Assault (felony)

e Reference category: Judicial waiver

f Standardized variable

g Reference category: Age 17 jurisdiction

 

 Table 4. Random coefficients linear regression model, logged jail sentence length (days)

(N1 = 717; N2 = 26)

Variable

Model 1

Model 2

 

Exp(b) (SE)

Exp(b) (SE)

Intercept

246.68*** (67.50)

249.64*** (68.29)

Level-1 variables

 

 

Assigned counsela

1.17 (.12)

1.28* (.14)

Private counsela

1.27 (.16)

1.25 (.17)

Blackb

1.03 (.10)

1.02 (.09)

Hispanicb

.89 (.08)

.89 (.08)

Other race/ethnicityb

.97 (.20)

.98 (.20)

Female

.66 (.14)

.65 (.14)

Agec

.97 (.04)

.97 (.04)

Murderd

-

-

Raped

.73 (.28)

.73 (.28)

Robberyd 

.75* (.10)

.75* (.10)

Burglaryd

.79* (.09)

.79* (.09)

Larcenyd

.68*** (.08)

.69*** (.08)

Motor vehicle theftd

.82 (.12)

.83 (.12)

Drug/Alcohold

.67*** (.08)

.67*** (.08)

Weapon offensed

.68* (.13)

.68* (.13)

Other felonyd

.55*** (.08)

.55*** (.08)

Misdemeanord

.46*** (.06)

.46*** (.06)

Multiple charges

.93 (.06)

.93 (.06)

Rearrest

1.52** (.20)

1.51** (.20)

Pretrial Detention

1.48*** (.10)

1.48*** (.10)

Direct filee

1.19 (.12)

1.19 (.12)

Statutory exclusione

1.10 (.16)

1.07 (.16)

Guilty plea

.67 (.16)

.64 (.15)

Level-2 variables (N = 37)

 

 

Urbanism

.98 (.08)

1.02 (.08)

Concentrated disadvantagef

1.28*** (.08)

1.24** (.09)

Crime ratesc

.99 (.00)

1.00 (.00)

Sentencing guidelines

1.32 (.25)

1.33 (.25)

Age 15g

.26 (.64)

.23 (.54)

Age 16g

.72 (.17)

.71 (.17)

Cross-level interactions

 

 

Assigned counsel x urbanism

-

.94* (.03)

Private attorney x urbanism

-

.99 (.04)

Random effects (variance components)

 

 

Random intercept

.00

.00

Intraclass correlation (ρ)

.00

.00

Random slope, attorney type

.01

.01

Intraclass correlation (ρ)

.02

.02

Level-1 residual variance

.65

.65

* <.05 ** <.01 *** <.001

a Reference category: Public defender

b Reference category: White

c Grand mean centered

d Reference category: Assault (felony)

e Reference category: Judicial waiver

f Standardized variable

g Reference category: Age 17 jurisdiction

Table 5. Random coefficients linear regression model, logged prison sentence length (days) (N1= 1,350; N2 = 36)

Variable

Model 1

Model 2

 

Exp(b) (SE)

Exp(b) (SE)

Intercept

2486.15*** (283.42)

2453.92*** (280.31)

Level-1 variables

 

 

Assigned counsela

1.18** (.07)

1.21** (.07)

Private counsela

1.20* (.08)

1.17 (.09)

Blackb

1.00 (.06)

1.00 (.06)

Hispanicb

.92 (.06)

.92 (.06)

Other race/ethnicityb

.95 (.12)

.95 (.12)

Female

.82 (.10)

.83 (.10)

Agec

.95* (.02)

.95* (.02)

Murderd

2.88*** (.23)

2.88*** (.23)

Raped

1.56*** (.18)

1.56*** (.18)

Robberyd 

1.02 (.05)

1.02 (.05)

Burglaryd

.73*** (.06)

.73*** (.06)

Larcenyd

.53*** (.05)

.53*** (.05)

Motor vehicle theftd

.59** (.11)

.59** (.11)

Drug/Alcohold

.59*** (.05)

.59** (.05)

Weapon offensed

.38*** (.06)

.38*** (.06)

Other felonyd

.56*** (.09)

.56*** (.09)

Misdemeanord

.45*** (.07)

.45*** (.07)

Multiple charges

1.20*** (.05)

1.21*** (.05)

Rearrest

1.05*** (.10)

1.05*** (.05)

Pretrial Detention

1.27*** (.07)

1.27*** (.07)

Direct filee

1.01 (.09)

1.01 (.09)

Statutory exclusione

1.07 (.08)

1.07 (.08)

Guilty plea

.54*** (.03)

.54*** (.03)

Level-2 variables (N = 37)

 

 

Urbanism

1.00 (.01)

1.00 (.01)

Concentrated disadvantagef

1.00 (.04)

1.00 (.04)

Crime ratesc

.99* (.00)

.99* (.00)

Sentencing guidelines

.84 (.08)

.84 (.08)

Age 15g

.38** (.12)

.38** (.12)

Age 16g

1.36** (.15)

1.36** (.15)

Cross-level interactions

 

 

Assigned counsel x urbanism

-

1.00 (.01)

Private attorney x urbanism

-

1.01 (.01)

Random effects (variance components)

 

 

Level-2 random intercept

.00

.00

Intraclass correlation (ρ)

.00

.00

Level-2 random slope, attorney type

.02

.02

Intraclass correlation (ρ)

.04

.04

Level-1 residual variance

.45

.45

* <.05 ** <.01 *** <.001

a Reference category: Public defender

b Reference category: White

c Grand mean centered

d Reference category: Assault (felony)

e Reference category: Judicial waiver

f Standardized variable

g Reference category: Age 17 jurisdiction

 


[1] In Powell v. Alabama (1932), the Supreme Court stated, “What, then, does a hearing include? Historically and in practice, in our own country at least, it has always included the right to the aid of counsel when desired and provided by the party asserting the right” (p. 68, emphasis added).

[2] Eisenstein and Jacob (1977, p. 26) observe that “[r]etained attorneys face a more complicated set of incentives” than public defenders, including maintaining “a reputation for vigorous defense in order to attract new clients.”

[3] There are also some studies that examine the influence of attorney type on juvenile court outcomes, with mixed results (see, e.g., Armstrong & Kim, 2011; Burruss Jr. & Kempf-Leonard, 2002; Guevara et al., 2004). No research, however, has examined juveniles in criminal court.

[4] For a full review of earlier, non-multivariate studies, see Feeney & Jackson (1991), and Hartley, Miller, & Spohn (2010).

[5] Defendants with public defenders were also significantly more likely to plead guilty (38.4 percent vs. 28.1 percent), likely related to the sentencing differences due to the “trial tax” (see Johnson, 2019).

[6] An exclusion restriction involves a variable that is included in the selection model but not in the substantive model of interest (see, e.g., Miller, Keith, & Holmes 2015, p. 234). As Bushway and colleagues (2007, p. 162) observe, “when the same variables are used to model the selection and substantive equations (i.e. when exclusion restrictions are not utilized), the model is only identified by the nonlinearity inherent in the inverse Mills ratio.”

[7] Different types of sentences were combined as follows. Minimum/maximum sentences (i.e., indeterminate) were coded as the average of the minimum and maximum sentence length, while minimum only and maximum only were coded as the sentence length. While maximum only sentences may be considered less punitive than minimum only (e.g., reduced incarceration sentence length through good behavior), without parole information this could not be assumed. Sensitivity analyses revealed no significant differences when maximum only sentences were independently examined, for jail as well as prison sentences.

[8] This includes Asian, American Indian or Alaskan Native, and Native Hawaiian or Pacific Islander.

[9] Due to the small number of female juveniles transferred to adult court, we ran sensitivity analyses with all-male samples. Parameter estimates did not differ in significant or direction, and females were retained in the final samples. 

[10] There was considerable missing data for prior record (approximately 23 percent) and court relationship (approximately 13 percent); as dropping more than a quarter of cases was not desirable, these variables were not included in the final analysis. Sensitivity analyses revealed that including these controls did not change direction or significance of the parameter estimates of interest. Missing cases were not imputed due to the potential for biasing estimates without any clear benefit gained by modeling uncertainty of two control variables (see Sterne et al., 2009).

[11] Sentence length can also be modeled as a count variable skewed to the right and truncated at zero, using truncated negative binomial regression (see Rydberg & Carkin, 2017). Nevertheless, “when the average counts are large, all else equal, it is better to estimate crime outcomes using OLS because the squared standard deviations from the mean are always the minimum estimator” (McDonald & Lattimore, 2010, p. 684). We also employed linear regression for parsimony and ease of interpretation. Sensitivity analyses using truncated negative binomial regression revealed nearly identical parameter estimates to those shown in tables 4–5.

[12] There was one exception to this: population density was strongly associated with upper age of jurisdiction since all “age 15” jurisdictions were in highly dense, New York City counties. This resulted in a strong correlation (r = .87) and VIF values of 4.85 and 5.08 for “urban” and “age 15”, respectively. Sensitivity analyses revealed that “urban” was significantly associated with sentence type as well as sentence length outcomes (but not conviction) before the addition of the jurisdictional controls (which largely confounded these effects, as indicated by high VIF values). However, the interactions between attorney type and urban did not differ in direction or significance between the models including and excluding jurisdictional controls. As such, these controls were included in the final models.

[13] Likelihood ratio tests indicated that random coefficient models did not improve model fit compared to the random intercept models for conviction, sentence type, and jail length, suggesting that effect of attorney type did not vary significantly across counties for these outcomes. Nevertheless, attorney type is designated a random effect for purposes on computing the cross-level interaction in model 2. (For prison length, a likelihood ratio test indicated that a random coefficient model improved model fit compared to a random intercept model.)

[14] We compared all levels of urbanism against one another (  multiplied by the number of outcomes and number of attorney type comparisons: 28 x 3 x 2 = 168 comparisons.

[15] Since multilevel logit models do not include a level-1 variance term, intraclass correlation (ρ) can be estimated using p2/3 as the level-1 variance, assuming that the level-1 variance follows the logistic distribution (see Johnson 2010).

[16] One caveat is that the present study focuses on large, urban jurisdictions. Public defender systems are not identical across jurisdictions, however, and may be less effective in less urban jurisdictions. As Davies and Worden (2017, p. 340) observe, “While public defender programs are often touted as superior, this judgement may be simplistic. Our results suggest that where public defender programs operate with at least average levels of funding, they have full-time staff, well-established administrators. But where funding is meager, the ‘public defender’ label may disguise nothing more than an assortment of county-contracted attorneys operating largely as independent agents.” One response to this important observation is that indigent defense should be provided at the state- rather than county-level to ensure that all public defender systems are adequately supported (Davies & Worden, 2009, 2017).

Comments
0
comment
No comments here
Why not start the discussion?