Judges on Trial

Abstract

Extant research on the effects of judicial background characteristics suggests minimal influence from the race or gender of the sentencing judge in criminal cases. This raises at least two possibilities: (1) the combined influence of judicial recruitment, indoctrination, and socialization into the judgeship results in a homogenous body of criminal court judges; or (2) current approaches to identifying judge effects in criminal sentencing have methodological and conceptual flaws that limit their ability to detect important influences from judicial background characteristics. The current article examines this issue with data from the Pennsylvania Commission on Sentencing that is augmented to include information on sentencing judges and criminal court contexts. It argues that the mode of conviction shapes the locus of sentencing discretion in ways that systematically underestimate judge effects for pooled estimates of incarceration and sentence length. The empirical results support this interpretation, especially for incarceration in trial cases, where older, female, and minority judges are substantially less likely to sentence offenders to jail or prison terms. The article concludes with a discussion of future research directions and policy implications for judge effects and disparity in sentencing.

Keywords

sentencing judge effects disparity mode of conviction HLM

For decades, criminologists, legal scholars, and policy makers have been captivated by continuing attempts to unravel the causes, consequences, and loci of disparate treatment in criminal sentencing. A voluminous literature has developed examining the influence that individual offender characteristics, such as race and gender, exert on courtroom decision-making outcomes. A limited but related literature has also emerged focusing on the influence of courtroom actor background characteristics in sentencing. While the former finds continuing evidence of individual offender disparities, the latter reports relatively few important influences from judicial background characteristics (Zatz, 2000). This has led some scholars to conclude that the collective impact of formal sentencing innovations, judicial selection and socialization processes, and courtroom workgroup pressures for conformity largely nullify any preexisting differences tied to judicial background characteristics (Spohn, 1990a; Steffensmeier & Britt, 2001; Steffensmeier & Hebert, 1999).

While the advent of sentencing guidelines and the importance of judicial socialization should not be overlooked, recent theorizing on courtroom actor discretion may provide valuable insights into the importance of judge characteristics in sentencing. Contemporary sentencing research emphasizes the fact that courtroom decision-making processes are likely to vary depending on whether or not a case is plea-bargained or convicted at trial (Alschuler, 1979; Johnson, 2003; Padgett, 1985; Smith, 1986). If the underlying processes determining sentencing outcomes are tied to the mode of conviction, then it may be necessary to disaggregate sentencing outcomes to better capture the effects of different decision makers in the punishment process. In particular, if the exercise of judicial discretion is most prominent in the sentencing of trial cases, as recent research argues (Johnson, 2003), then the influence of judicial background factors should be most pronounced for these types of cases. Given that guilty pleas account for about 95% of all felony convictions (Durose & Langan, 2005), the general lack of significant findings for judge characteristics in prior research may simply reflect the high prevalence of pled cases. The present study investigates this possibility by examining variations in judge-level predictors of sentencing for cases convicted through different types of pleas and trials. Using data from the Pennsylvania Commission on Sentencing (PCS), this study evaluates hypotheses about the relative import of a variety of judge-level background characteristics across modes of conviction. Such an investigation holds the potential to clarify past discrepancies regarding the importance of judge factors in sentencing, while also providing useful directions for future theoretical and policy developments regarding the identification and control of court actor discretion and sentencing disparity.

Prior Research and Theorizing

The Importance of Judge Characteristics in Prior Research

Relative to the copious literature on individual offender attributes, research on the characteristics of organizational decision makers in criminal courts is rare. In the past three decades, only about a dozen studies have incorporated measures of judge characteristics in the study of criminal punishment. Table 1 provides a summary of this research. Although not necessarily exhaustive, it represents the most recent and influential investigations of judge effects in sentencing.¹ The limited scholarly attention devoted to this aspect of criminal punishment is partly the product of data constraints that preclude examination of court actor background characteristics, and partly the result of prior research reporting few or inconsistent findings for judicial background factors (e.g., Gruhl, Spohn, & Welch, 1981; Spohn 1990a; Steffensmeier & Hebert, 1999). Overall, though, the collective research on judicial background characteristics has been limited in important ways, including small and inadequate samples of judges, limited measures of judicial backgrounds, and relatively simplistic analytical approaches (Johnson, 2006).

Table 1.

Prior Studies of Judge Characteristics in Criminal Sentencing.

Author(s)	Year	Sample	Judge race	Judge gender	Judge age	Years on bench	Prior experience	Marital status	Religion
Kritzer and Uhlman	1977	Unspecified number of “Metro City” trial judges sentencing 8 felony offenses		X
Uhlman	1978	91 “Metro City” trial judges (16 Blacks) sentencing felony offenses	X
Gruhl et al.	1981	182 “metro city” trial judges (7 females) sentencing 14 felony offenses for peer review		X
Frazier and Bock	1982	7 Florida trial judges in 1 judicial district sentencing 309 criminal cases			X	X	X
Myers and Talarico	1987	Unspecified number of Georgia Superior Court Judges Sentencing Felony Offenses		X	X		X		X
Myers	1988	Unspecified number of Georgia superior court judges sentencing felony offenses			X		X		X
Welch et al.	1988	140 “Metro City” trial judges (10 Black, 130 White) sentencing felony offenses	X	X		X	X
Spohn	1990a	38 Detroit trial court judges (13 Black, 25 White) sentencing 11 violent felony offenses	X
Spohn	1990b	Detroit recorder’s court judges sentencing sexual assault cases	X	X
Holmes et al.	1993	15 District court justices in El Paso county (5 Hispanic) sentencing felony offenses	X
Steffensmeier and Hebert	1999	279 Pennsylvania trial judges (39 females, 231 males) sentencing all crimes in 18 counties		X	X	X	X	X
Steffensmeier and Britt	2001	90 Pennsylvania trial judges (10 Black, 80 White) sentencing all crimes in 4 counties	X		X	X
Muhlhausen	2004	290 Pennsylvania trial judges (40 female, 16 Black, 2 Hispanic) sentencing felony cases		X	X		X
Johnson	2006	303 Pennsylvania trial judges sentencing all crimes in all 67 Pennsylvania counties		X	X	X	X	X	X
Anderson and Spohn	2010	18 Federal judges in 3 districts (MN, NE, IA) sentencing all federal crime types					X	X
Tiede et al.	2010	504 Federal judges (78 female; 70 non-White), cases published in the federal supplement	X	X		X

X = study examined judicial characteristic.

Early investigations into the importance of judicial attitudes and sentencing philosophies argued forcefully that the sentencing judge was critically important for determining individual sanctions. In his theoretical exposition of judicial decision making, Gibson averred, “there can be little doubt that what judges prefer to do—i.e. their substantive attitudes and values—influences their decision-making behavior” (Gibson, 1983, p. 9). Similarly, Hogarth (1971) examined the influence of judicial attitudes, sentencing philosophies and organizational constraints on the sentencing behavior of a large sample of Canadian magistrates and found that “enormous variation exists among magistrates in the way in which they go about making sentencing decisions” (Hogarth, 1971, p. 91). He ultimately concluded that “one can explain more about sentencing by knowing a few things about a judge than by knowing a great deal about the facts of the case” (Hogarth, 1971, p. 350).

Early empirical work, however, failed to provide evidence for the importance of judge characteristics in sentencing. For instance, three early studies utilizing the same data from an unspecified large urban court reported few important differences along racial and gender lines (Gruhl et al., 1981; Kritzer & Uhlman, 1977; Uhlman, 1978). In their study of judge’s gender, Kritzer and Uhlman (1977, p. 86) concluded that “female judges behave no differently than their male colleagues.” This research was important for identifying the potential importance of judicial variations in punishment, but it was limited by small samples (e.g., seven female judges), a limited focus on race or gender, and simplistic analytical techniques (e.g., cross-tabulations). Later work by Welch, Combs, and Gruhl (1988) using the same data with additional control variables suggested that Black judges were slightly more lenient toward Black offenders than White judges.

The work of Myers and Talarico (1987) substantially advanced early work on the influence of judge characteristics in sentencing. Their investigation of contextual variations in Georgia incorporated a range of judicial background factors, such as their age, religious affiliation, prior prosecutorial experience, and tenure on the bench, along with various measures of county-level social context. Their findings for the direct effects of judicial backgrounds “conveyed the impression that these factors are relatively unimportant” (Myers & Talarico, 1987, p. 122), but evidence of indirect effects suggested that judge “variables operate as impressive conditioners of differential treatment” in punishment. However, they were unable to include any direct measures of some key characteristics such as the individual gender and race of the judge in their sample.

More recent research has focused on judge effects in additional jurisdictions. Spohn (1990a, 1990b) investigated the impact of judge factors in Detroit. Her first study examined the effect of judge’s race for violent felonies and uncovered only mild evidence that black judges were less punitive, with both Black and White judges punishing minority offenders more harshly. The second study examined the role of judge’s gender in sexual assault cases and reported that female judges sentenced offenders to longer terms of incarceration. The overall conclusion in both these studies, however, was that judicial backgrounds have minimal impact on punishment decisions in criminal courts. These studies were limited, however, by a lack of information on other judicial background characteristics.

Steffensmeier and Britt (2001) reported that Black judges were overall more severe in their treatment of criminal defendants. Given that the effect of judge’s race was relatively small, though, they concluded that “the job, not so much the individual, apparently makes the ‘judge’” (Steffensmeier & Britt, 2001, p. 749). In a related study, Steffensmeier and Hebert (1999) examined the influence of judge’s gender on variations in criminal punishments across 18 Pennsylvania counties. After controlling for age, tenure, prosecutorial experience, and marital status of the judge, they concluded female judges sentenced more harshly than male judges, particularly when faced with repeat Black offenders. Muhlhausen (2004) found contradictory effects for judge’s race, ethnicity, and gender in his study of Pennsylvania courts. Using Tobit regression with fixed effects for counties, he found that female and Hispanic judges were more lenient in sentencing, but that African American judges were more severe.

Johnson (2006) reexamined the influences of judicial race and gender in Pennsylvania using three-level hierarchical models incorporating individual-, judge-, and county-level sentencing measures. Contrary to early work using Pennsylvania data, he found minority judges were less likely to render incarceration sentences, especially when faced with minority defendants, and when they did so they meted out slightly shorter sentences. Furthermore, his analysis uncovered no significant relationships between judges’ gender and punishment decisions. Overall, judge-level predictors accounted for less than one quarter of the between-judge variation in both incarceration and sentence lengths. Wooldredge (2010) analyzed data on felony convictions in Ohio and reported relatively larger differences between judges. He found that even after controlling for compositional case characteristics, about 14% of the variation in the use of prison was attributable to judges. Moreover, the relative weight placed on different sentencing factors, including offender characteristics, differed significantly across judges.

Select work has also examined judge effects in federal sentencing. The United States Sentencing Commission (USSC, 2004, p. 101) conducted a comprehensive analysis of federal sentencing which concluded that only “relatively minor inter-judge and regional disparity” exists that is not attributable to case characteristics. However, using unique data collected from the federal District of Massachusetts, Scott (2009, p. 105) examined inter-judge sentencing disparity following the Booker decision and reported that “Sharp and growing differences are apparent in guideline sentencing patterns between judges in the same courthouse.” Similarly, Tiede (2009) demonstrated that variation in federal sentences was far more pronounced when judges depart from the sentencing guidelines, suggesting meaningful differences remain between judges for departure cases. Anderson and Spohn (2010) also found evidence of sentencing differences among federal judges. They examined inter-judge disparity in three federal district courts and found that significant variation existed between judges in mean sentence lengths after controlling for legally relevant case characteristics. Their results suggested that different federal judges evaluate individual offenders and cases differently, with some evidence emerging for the importance of judicial caseload and prior legal experience in the sentencing process.

Although a number of commentators have argued that the collective effects of judicial legal training, selection, socialization, organizational conformity pressures and presumptive sentencing guidelines largely negate preexisting differences in judicial attitudes, beliefs, and sentencing philosophies (Zatz, 2000), such a conclusion may be premature given recent evidence and the relative paucity of studies examining judge characteristics in sentencing. Because sentencing discretion can vary by case type, a true test of the influence of judicial characteristics in sentencing needs to account for variation in the exercise of judicial discretion across cases. One useful approach for gaining leverage on this issue is to disaggregate cases by the mode of conviction.

Judicial Discretion and JudgeCharacteristics Across Modes of Conviction

In his influential and eloquent diatribe chastising unfettered judicial discretion, Judge Marvin Frankel avouched, “the evidence is conclusive that judges of widely varying attitudes on sentencing . . . mete out widely divergent sentences . . . explainable only by the variations among the judges, not by material differences in the defendants or their crimes” (Frankel, 1973, p. 21). This assertion is consistent with early theoretical arguments that suggest judicial attitudes and sentencing philosophies act as filters through which sentencing judges selectively perceive their social worlds. According to Gibson (1983, p. 9), “judges’ decisions are a function of what they prefer to do, tempered by what they think they ought to do, but constrained by what they perceive is feasible to do.” What judges prefer to do is a product of their attitudes and sentencing philosophies, what they think they ought to do stems from their role orientation on the bench (i.e., their perception of what it means to be a judge), and what they perceive is feasible to do results from organizational constraints associated with their courtroom workgroup and community social environment.

However, the salience of any of these factors is restricted unless the sentencing judge is free to exercise significant individual discretion in the determination of individual sentencing outcomes. From a court community perspective, “judges do not exercise discretion freely even in sentencing. The power of individual judges is circumscribed by the role of ‘court work groups’” (Welch et al., 1988, p. 134). As Eisenstein and Jacob (1977, p. 8) argued long ago, “the singular focus on the judge cannot stand comfortably with the widespread use of plea bargaining.” For these cases “one would not expect the judge’s background or attitudes to be closely related to the disposition of charges against defendants” (Eisenstein & Jacob, 1977, p. 8). Their court community perspective suggested instead that criminal punishments result from the dynamic interaction of key court personnel, including the prosecutor, defense counsel, and presiding judge, and importantly, that different case processing strategies emerge for different modes of conviction that condition the relative influence of the sentencing judge.

Similar arguments have been made on less theoretical, more methodological grounds. Klepper, Nagin, and Tierney (1983, p. 102), for instance, explicitly criticize contemporary research for combining trial and plea dispositions into single regression equations, arguing that “This implicitly incorporates the assumption that the relationship between sentence . . . and various observational features of the case is the same for guilty pleas and trial convictions. If this is not correct, then the estimating equations . . . are misspecified.” They observe that because “guilty pleas generally dominate trial convictions,” analyses performed on aggregate data will provide information primarily on the factors influencing sentences for guilty pleas (Klepper et al., 1983, p. 102). In order to adequately separate and isolate the unique influence of different court actors and their background experiences on criminal punishment decisions, then, it becomes necessary to separate trial cases from guilty pleas.

In this respect, recent research that emphasizes the conditioning effects of different modes of conviction on the sentencing process provides useful insights into the influence of judicial factors in sentencing. Two recent studies are particularly apposite. The first, conducted by Lacasse and Payne (1999) using federal sentencing data, offers one of the few studies to compare sentencing variations across judges. These scholars examined the amount of variation attributable to judges for plea and trial cases before and after the implementation of the federal sentencing guidelines. Their results offered no evidence that inter-judge variation had been reduced by the guidelines (in fact it appeared to increase). Instead, their work offered convincing evidence that sentencing variations were intricately tied to the mode of conviction. Specifically, these researchers concluded that “the variation attributable to judge assignment represents . . . a higher proportion [for trials] than for pleas . . .” supporting “the notion that the prosecutor has an important influence on the sentence in plea bargains, while the judge has the determining influence on trial sentences, even in the post-reform period” (Lacasse & Payne, 1999, pp. 262-264).

In the second study, Johnson (2003) offered further support for the theoretical import of modes of conviction in criminal sentencing. He suggested that the exercise of judicial discretion is greatest for cases convicted at trial, while prosecutorial sentencing discretion is more likely to emerge in negotiated pleas. His empirical findings were largely consistent with this theoretical framework. Extralegal disparities in judicial departures from the sentencing guidelines tended to be most pronounced for modes of conviction in which judges and/or prosecutors exercised significant sentencing discretion. Nonnegotiated pleas, which he argued largely reflected “going rates” (Eisenstein & Jacob, 1977) applied to “normal crimes” (Sudnow, 1965), demonstrated less influence from offender characteristics. Collectively, these perspectives clearly suggest that sentencing discretion, and the associated sentencing disparities that potentially accompany it (at the individual or court actor level of analysis), may be tied to and reflected in different types of case disposition.

Theoretically, this raises important issues for research on judge effects in sentencing. If it is the case that judges exercise significantly more sentencing discretion for trials compared to pleas, then trial cases should provide greater opportunity for judicial background characteristics and career experiences to influence these sentencing outcomes. Conversely, if judicial decision making is constrained in cases disposed of through guilty pleas, then one would expect judge characteristics to exert weaker influences for these cases. Given that guilty pleas account for all but about 5% of criminal convictions, the general lack of significant findings for judge characteristics in prior research may be due to a failure to disaggregate sentencing outcomes by mode of conviction. All prior empirical work on judge effects jointly examines plea and trial cases. If, however, judge characteristics fail to demonstrate substantively important effects across modes of conviction, then the argument for the importance of judicial selection and socialization is strengthened.

The present work investigates this issue by reanalyzing the data examined in Johnson (2006) after disaggregating sentencing outcomes across modes of conviction. The guiding research question is the extent to which judge effects are more pronounced for trial cases compared to those convicted through nonnegotiated and negotiated pleas.² Specifically, the following two propositions are examined:

Hypothesis 1: Between-judge variation in criminal punishments is expected to be most pronounced for trial cases.

Hypothesis 2: Judicial background characteristics should have more pronounced effects in trial cases.

If trial cases allow for greater exercise of judicial discretion, they are likely to be associated with greater inter-judge variation in punishment. Although specific directional hypotheses are not specified due to the inconsistency of findings in prior work,³ the age, race, and gender of the judge are expected to have stronger effects for trial cases than for negotiated and non-negotiated pleas. Overall, then, this research presents an important opportunity to better evaluate the effects of judicial background characteristics in the meting out of criminal punishments while simultaneously enhancing our theoretical understanding of the potentially important intersection between courtroom actor discretion and various modes of conviction.

Data and Method

The current research examines the influence of judicial characteristics in sentencing across modes of conviction by utilizing 2 years of data (1999 to 2000) from the PCS. The PCS is required by law to collect information on sentenced cases in the state of Pennsylvania. The PCS data are among the most complete state-level resource for studying criminal sentencing. They contain a variety of theoretically important individual-level predictors of sentencing, such as offense severity and prior criminal history, as well as various offender characteristics such as age, race/ethnicity, and gender. Moreover, detailed information is available on whether an offender pled guilty outright, negotiated a plea, or was convicted at bench or jury trial. The Pennsylvania sentencing guidelines, enacted in 1982, are among the oldest presumptive guidelines in operation, and extensive prior work has been conducted in this research context, including previous studies of judge effects in sentencing (Johnson, 2006; Steffensmeier & Britt, 2001). The current results can therefore be evaluated relative to previous findings regarding judicial effect sizes.

The individual-level PCS sentencing data were extended to include judge and court characteristics using publicly available judicial biographies.⁴ In addition, select county- and courtroom-level control variables were also obtained through the U.S. Census Bureau and the Administrative Office of Pennsylvania Courts. Overall, this final data set provides a very rich and detailed body of information that is well suited for examining the influences of theoretically important judge-level factors in sentencing. In line with prior research, cases in this analysis are limited to the most serious offense per judicial transaction and to those cases sentenced under the 1997 guidelines, resulting in 148,590 total cases, 79,333 of which resulted in incarceration.⁵ Table 2 provides descriptions of the specific individual-, judge-, and county-level measures utilized in this study.

Table 2.

Statistical Coding and Descriptions of Dependent and Independent Variables.

Variables	Coding	Description
Dependent variables
Incarceration	1 = Incarcerated	Binary variable for sentences of incarceration compared to sentences of probation
Sentence length	Number of months	The minimum number of months of incarceration
Independent variables
Individual-level predictors
Year of sentence	1 = 1999	Dummy variable identifying cases sentenced in 1999 instead of 2000
Offense severity (OGS)	1-14	PCS Offense Gravity Score measured by 1- to 14-point scale
Criminal history (PRS)	0-8	PCS Prior Record Score measured by 0- to 8-point scale
Mandatory	1 = Mandatory applied	PCS dummy variable for cases receiving a mandatory minimum
Presumptive sentence	Number of months	The minimum months of incarceration prescribed by the sentencing guidelines
Offense type	Other = Referent	Four dummy variables for cases convicted for violent, property, drug, and other crimes
Offender age	Number of years	PCS continuous variable for age of the offender at time of sentencing
Offender gender	1 = Female	PCS dummy variable for gender of the offender
Offender race	White = Referent	PCS dummy variables for White, Black, Hispanic, and other offender race/ethnicity
Mode of conviction	Non-neg Plea = Referent	PCS dummy variables for non-negotiated plea, negotiated plea, bench, and jury trial
Selection bias correction	Inverse mills ratio	Control for selection bias calculated with the Heckman command in Stata 8.2
Judge-level predictors
Judge age	Number of years	Continuous variable measured as age of judge in 2000
Judge gender	1 = Female	Dummy variable identifying cases sentenced by a female judge
Judge race	1 = Minority	Dummy variable identifying cases sentenced by a minority judge
Judicial tenure	Number of years	Continuous variable measured as years serving on the Pennsylvania Court of Common Pleas
Judicial caseload pressure	Number of cases/2	Average number of criminal cases sentenced by a judge in a year (divided by 100)
County-level controls
Court size	Medium = referent	Three dummy variables for large, medium, and small courts, based on number of judges
Available jailspace	Number of jailbeds	Number of jail beds divided by the number of cases sentenced in the county
Crime rate	Crimes per 1,000 people	Number of serious crimes per 1,000 people obtained from Uniform Crime Reports
Racial composition	Percent minority	Percent of the county population identified as either African American or Hispanic

Analytic Procedures

The analyses separately examine the probability and length of incarceration using a combination of hierarchical generalized and linear modeling techniques. Sentence type is modeled as a binary outcome, whereas sentence length is a continuous measure based on the minimum number of months of detainment.⁶ Because the current data structure is characterized by the nesting of cases within judges and of judges within courts, three-level hierarchical models were utilized to simultaneously incorporate individual-, judge-, and court-level predictors. Useful discussions of the application and advantages of hierarchical modeling in the study of criminal sentencing are available in prior research (see Bontrager et al., 2005; Britt, 2000; Johnson, 2006, 2010; Kautt, 2002; Ulmer & Johnson, 2004; Weidner, Frase, & Schultz, 2005). For the sake of illustration, the simple three-level linear model for sentence length is presented below. Its application to the incarceration decision is straightforward.

Y_{i j k} = π_{0 j k} + π_{1 j k} (A_{i j k} - {\bar{A}}_{\dots}) + e_{i j k} where

π_{0 j k} = β_{00 k} + β_{01 k} (X_{j k} - {\bar{X}}_{\dots}) + r_{0 j k} and

β_{00 k} = γ_{000} + γ_{001} (W_{k} - {\bar{W}}_{\dots}) + u_{00 k}

In the multilevel model, Y_ijk is the minimum number of months of incarceration that the ith offender is sentenced to by the jth judge in the kth court. π_0jk is the individual-level intercept and π_1jk represents the effects of a vector of individual-level variables A_ijk centered around their grand means. e_ijkis the individual-level residual error term. Parameters in the individual-level Equation 1 are then modeled at higher levels of analysis using judge- and county-level predictors. Variation in the individual-level intercept π_0jk is modeled with a judge-level intercept β_00k and a vector of judge-level covariates X_jk also centered on their grand means. β_01k therefore represents the effect of judicial background characteristics on average sentence lengths after controlling for individual predictors, with r_0jk representing the judge-specific error term. In this way, judge characteristics are utilized to explain between-judge variation in sentence lengths. Variation in the intercept β_00k is further modeled using county court characteristics W_k centered on their grand means, where γ₀₀₁ symbolizes the effects of county-level predictors on sentence length with u_00k representing the county-specific error term. For all models, fixed effects were specified and robust standard errors were utilized.⁷

The application of hierarchical modeling procedures provides several advantages over many prior studies of judge effects in punishment. This approach provides standard error estimates that are corrected for the clustering of cases within judges and county courts, as well as statistical significance tests that are appropriately adjusted at each level of analysis. Hierarchical estimates also provide variance components that are useful for evaluating the variation in outcomes across higher units of analysis. Very little research on judges controls for aggregate court contexts, only one prior study utilizes multilevel models (Johnson, 2006), and no work has examined the moderating effects of modes of conviction. The current study therefore provides several advantages over prior work examining judicial characteristics and modes of conviction in criminal sentencing.

Results

Table 3 provides descriptive statistics for each level of analysis with individual predictors disaggregated by mode of conviction. The typical offender is a White male in his early 30s who is convicted though a negotiated guilty plea. Few differences emerge between offenders who negotiate a plea and those who plead guilty outright, but offenders who go to trial are distinct in several ways—they tend to be charged with more serious crimes, have longer prior records, and are disproportionately male, Black and Hispanic. These individual-level descriptive findings comport with prior research on who is most likely to exercise their right to trial (Albonetti, 1990).

Table 3.

Descriptive Statistics for Individual-, Judge-, and Court-Level Predictors.

	Total sample		Means by mode of conviction
Dependent variables	Mean	SD	Non-Neg Plea	Neg. Plea	Trials
In/out incarceration	0.55	0.50	0.59	0.52	0.73
Sentence length	8.75	17.46	7.74	7.31	28.04
Individual-level variables
Sentence year (1999)	0.48	0.50	0.51	0.49	0.51
Offense severity	3.46	2.17	3.41	3.37	5.46
Prior criminality	1.31	1.84	1.38	1.24	1.96
Presumptive sentence	3.68	10.18	0.31	0.29	0.61
Mandatory applied	0.23	0.42	0.31	0.22	0.17
Offender age	31.50	10.23	32.24	31.20	31.62
Female offender	0.17	0.37	0.18	0.17	0.09
Black offender	0.28	0.45	0.29	0.27	0.54
Hispanic offender	0.06	0.24	0.06	0.06	0.10
Violent offense	0.13	0.33	0.10	0.12	0.30
Property offense	0.21	0.41	0.19	0.22	0.13
Drug offense	0.20	0.40	0.21	0.20	0.23
Nonnegotiated plea	0.17	0.38	1.0	0.0	0.0
Negotiated plea	0.65	0.48	0.0	1.0	0.0
Trial	0.04	0.20	0.0	0.0	1.0
N	148,590		25,750	96,386	6,215
Judge-level variables (N = 303)
Judge age	55.97	6.58
Female judge	0.15	0.36
Minority judge	0.07	0.26
Judicial tenure	11.91	8.45
Judicial caseload pressure	2.45	2.17
Court-level variables (N = 60)
Large court	0.03	0.18
Medium court (reference)	0.23	0.43
Small court	0.73	0.45
Available jailspace	18.41	12.10
Crime rate	24.28	9.5
Percent minority	4.93	7.20

A typical judge in Pennsylvania is a White male in his mid-50s who has sat on the bench for nearly a dozen years. About 15% of judges were female and only 7% were minorities.⁸ Of the 303 judges in the data, 46 were female and 22 were minorities. The majority of criminal courts in the state are located in small rural counties, but the number of criminal cases sentenced is distributed about evenly across small, medium, and large courts. Typical county contexts were characterized by minority populations near 5% and average crime rates just over 24 crimes per 1,000 people.

Table 4 presents the results from the unconditional Hierarchical Linear Models (HLM) models that parcel the variation in sentencing across levels of analysis for each mode of conviction. These results provide some preliminary evidence that variation in inter-judge punishments is greatest for trial cases. For all cases, the between judge variance component for the incarceration decision is 0.17, but for trial cases it increases to 0.31. The estimate of the proportion of total variance in incarceration probabilities that is attributable to judges is about 8%, compared to an estimate of 5% when all modes of conviction are combined.⁹ Although these differences are not of large magnitude, they are consistent with the theoretical argument that judges have greater influence over individual punishment outcomes in trial cases. The probability of incarceration varies most starkly across judges for trial cases, with the differences being greatest between trials and negotiated pleas.

Table 4.

HLM Unconditional Models—Judge Characteristics Across Modes of Conviction.

Incarceration			Sentence length
All cases
Fixed effects	β	SE	Fixed effects	β	SE
Intercept	0.36	0.07	Intercept	7.53	0.70
Random effects	Variance	SD	Random effects	Variance	SD
Level 1	—		Level 1	253.35	15.92
Level 2	0.17	0.41	Level 2	55.26	7.43
Level 3	0.20	0.44	Level 3	14.74	3.84
Between judge proportion of variance			Between judge proportion of variance
0.05			0.17
Nonnegotiated pleas
Fixed effects	β	SE	Fixed effects	β	SE
Intercept	0.65	0.12	Intercept	7.30	0.90
Random effects	Variance	SD	Random effects	Variance	SD
Level 1	—		Level 1	215.38	14.68
Level 2	0.29	0.54	Level 2	50.90	7.13
Level 3	0.50	0.71	Level 3	20.25	4.50
Between judge proportion of variance			Between judge proportion of variance
0.07			0.18
Negotiated pleas
Fixed effects	β	SE	Fixed effects	β	SE
Intercept	0.27	0.07	Intercept	7.32	0.69
Random effects	Variance	SD	Random effects	Variance	SD
Level 1	—		Level 1	150.47	12.27
Level 2	0.14	0.38	Level 2	54.77	7.40
Level 3	0.22	0.47	Level 3	11.80	3.44
Between judge proportion of variance			Between judge proportion of variance
0.04			0.25
Trial cases
Fixed effects	β	SE	Fixed Effects	β	SE
Intercept	1.37	0.09	Intercept	26.30	1.85
Random effects	Variance	SD	Random effects	Variance	SD
Level 1	—		Level 1	1141.28	33.78
Level 2	0.31	0.56	Level 2	241.80	15.55
Level 3	0.13	0.36	Level 3	63.88	7.99
Between judge proportion of variance			Between judge proportion of variance
0.08			0.17

Examination of the absolute magnitude of inter-judge variance in sentence lengths reveals that it is nearly five times larger for trial cases compared to pled cases. The between-judge (level 2) variance components range between 50 and 55 for negotiated and nonnegotiated guilty pleas but increases dramatically to 241 for trial cases. Translated into concrete numbers, this suggests that a one standard deviation change in sentence length for pled cases results in a change of about 7 months, whereas for trial cases a one standard deviation change alters one’s sentence length between 15 and 16 months. Somewhat surprisingly, though, the relative proportion of total variation in sentence lengths attributable to judges is greatest for negotiated pleas, with an intraclass correlation of 0.25. This highlights the fact that sentence lengths following trial are substantially more varied at each level of analysis in the data. The proportionally larger judge-level variation for negotiated pleas may provide some evidence that plea bargaining occurs “in the shadow of the judge” (Lacasse & Payne, 1999) or it may reflect that fact that sentencing guidelines are less restrictive on sentence length decisions in negotiated pleas, but future work is needed to investigate these possibilities. Overall, the unconditional models for sentence length suggest that the relative variation attributable to judges compared to other levels of analysis is greatest in negotiated pleas, but that the absolute magnitude of inter-judge variance is nearly five times larger in trial cases compared to guilty pleas.

Table 5 presents results from the fully specified three-level model of incarceration across modes of conviction. Consistent with prior research, individual legal considerations exert the strongest effects on all incarceration decisions (e.g., Steffensmeier, Ulmer, & Kramer, 1998). Not surprisingly, offense severity and criminal history are strong predictors of incarceration, and mandatory minimums virtually guarantee time spent behind bars. Several individual offender characteristics also significantly influence the odds of incarceration though. Young, male, and minority offenders are disadvantaged across all modes of conviction. Females are approximately two thirds as likely to be incarcerated whereas Blacks and Hispanics are about 40% more likely to be sentenced to jail or prison. Also consistent with prior work, the odds of incarceration are especially pronounced in small rural courts and in counties with greater available jail capacity. Overall, individual- and county-level effects exert relatively consistent, anticipated effects across conviction categories.

Table 5.

HLM Judge Effects on Incarceration Across Individual Modes of Conviction.

	All cases			Non-neg plea			Neg plea			Trials
	Logit	SE	Odds	Logit	SE	Odds	Logit	SE	Odds	Logit	SE	Odds
Individual-level factors
Constant	0.76	0.09	—****	1.21	0.16	—****	0.57	0.09	—****	2.75	0.17	—****
Year of sentence	0.03	0.03	1.04	−0.02	0.06	0.98	0.04	0.03	1.04***	−0.04	0.05	0.96
Offense severity	0.50	0.02	1.65****	0.50	0.03	1.64****	0.50	0.03	1.65****	0.63	0.06	1.88****
Prior criminality	0.43	0.02	1.54****	0.43	0.04	1.54****	0.44	0.02	1.55****	0.61	0.10	1.85****
Presumptive sentence	−0.01	0.00	0.99**	−0.01	0.01	0.99	−0.01	0.01	0.99	−0.02	0.01	0.98
Mandatory applied	3.27	0.33	26.37****	3.72	0.55	41.36****	3.12	0.33	22.55****	4.04	0.62	56.64****
Offender age	−0.01	0.00	0.99***	−0.01	0.00	0.99***	−0.01	0.00	0.99***	−0.01	0.00	0.99**
Female offender	−0.41	0.03	0.67***	−0.30	0.06	0.74***	−0.41	0.04	0.67***	−0.66	0.11	0.52***
Black offender	0.33	0.07	1.39***	0.39	0.10	1.48***	0.33	0.08	1.39***	0.30	0.06	1.34**
Hispanic offender	0.36	0.09	1.43***	0.64	0.08	1.91***	0.29	0.12	1.34*	0.30	0.08	1.35**
Violent offense	0.24	0.06	1.27***	0.30	0.15	1.35*	0.26	0.08	1.30**	0.18	0.05	1.20***
Property offense	0.20	0.06	1.22**	0.11	0.08	1.11	0.19	0.08	1.21*	0.41	0.11	1.50***
Drug offense	−0.14	0.05	0.87**	−0.15	0.12	0.86	−0.20	0.07	0.82**	0.27	0.13	1.31*
Bench trial										−0.58	0.12	0.56***
Judge-level factors
Judge age	−0.01	−0.01	0.99*	−0.02	0.01	0.98	−0.01	0.00	0.99	−0.04	0.01	0.96****
Female judge	−0.06	0.11	0.94	−0.10	0.17	0.90	0.01	0.09	1.01	−0.46	0.12	0.63****
Minority judge	−0.36	0.10	0.69****	−0.35	0.08	0.71****	0−.28	0.07	0.76****	−0.25	0.13	0.78**
Judicial tenure	0.00	0.00	1.00	0.00	0.01	1.00	0.00	0.00	1.00	0.01	0.01	1.01*
Judicial caseload	−0.01	0.02	0.99	−0.01	0.03	0.99	0.00	0.02	1.00	0.01	0.02	1.01
County-level controls
Large court	−0.57	0.24	0.57**	−0.25	0.24	0.78	−0.60	0.32	0.55*	−0.85	0.44	0.43*
Small court	0.37	0.17	1.45**	0.75	0.27	2.13***	0.43	0.18	1.53**	0.67	0.30	1.96**
Available jailspace	0.03	0.01	1.03****	0.04	0.01	1.04***	0.03	0.01	1.03****	0.03	0.01	1.03***
Crime rate	−0.01	0.02	0.99	−0.04	0.02	0.97	0.00	0.02	1.00	−0.01	0.03	0.99
Percent minority	0.00	0.02	1.00	0.03	0.03	1.03	−0.01	0.02	0.99	0.01	0.03	1.01
Between judge R²	2.9%			2.2%			2.0%			18.1%
Between County R²	47.5%			43.7%			42.8%			33.7%
N	148,590			25,750			96,386			6,215

Note. The model for all cases includes controls for mode of conviction. The sample sizes for submodels do not equal the total sample size (N = 148,590) due to missing data on mode of conviction. The coefficients of determination are calculated as proportionate reduction in error (PRE) measures.

*p ≤ .10. **p ≤ .05. ***p ≤ .01. ****p ≤ .001.

Turning to the judge-level of analysis, minority judges are consistently less likely to incarcerate offenders, regardless of the mode of conviction. This may suggest that the race of the judge is influential in the plea negotiation process, or it may partially reflect the fact that minority judges are more likely to sentence cases in large urban courts. To tease out this relationship, future research is needed in additional contexts with more diverse samples of judges that span a broader range of county court environments.

The other judge characteristics exerted noticeably stronger effects in trial cases. The gender of the judge had no significant influences on negotiated and nonnegotiated pleas, but it emerged as a significantly stronger predictor of incarceration for trial cases (Z = 3.07).¹⁰ Female judges were 0.63 times as likely to incarcerate offenders convicted at trial. Similarly, the age of the judge had a significantly stronger effect at trial (Z = 2.99), with each additional year reducing the odds of incarceration by about 4%. Judicial tenure on the bench also emerged as a small and marginally significant predictor of incarceration for trial cases alone, but this effect was not statistically distinguishable from the plea effects. Taken together, these results offer some tentative support for the contention that judicial background characteristics have more influence on incarceration decisions for trial cases than for other modes of conviction. Theoretically, trial convictions should introduce greater opportunity for judges to exercise their individual sentencing discretion, resulting in stronger influence from demographic factors that may be tied to past experiences, role orientations and individual sentencing philosophies. Although the judge characteristics only explain 18% of the total between-judge variation in the use of incarceration for trial cases, this was substantially greater than the 2% of the variation explainable for cases convicted through guilty pleas.

Table 6 reports parallel analyses for sentence length. Overall, results from these models are less consistent with theoretical expectations. Mandatory minimums and offense severity again emerge as powerful individual-level predictors of punishment. While offender characteristics influence sentence lengths across all modes of conviction, though, it is interesting to note that their effects are strongest for trial cases. The effect of Hispanic ethnicity, for instance, is more than twice as great for cases convicted through trials. The only court context measure that emerged as a consistent predictor of sentence lengths was the size of the court. Larger courts meted out slightly shorter sentence lengths, though substantively, this effect was small.

Table 6.

HLM Judge Effects on Sentence Length Across Individual Modes of Conviction.

	All cases		Non–neg plea		Neg plea		Trials
Individual–level factors	β	SE	β	SE	β	SE	β	SE
Constant	9.09	0.14****	8.04	0.27****	7.41	0.14****	27.12	0.87****
Year of sentence	−0.14	0.18	0.03	0.14	−0.02	0.15	−0.77	0.37**
Offense severity	3.12	0.36****	3.51	0.64****	3.00	0.28****	3.73	0.47****
Prior criminality	1.36	0.18****	1.69	0.08****	1.41	0.18	0.89	0.52
Presumptive sentence	0.76	0.05****	0.73	0.01****	0.64	0.05****	0.83	0.08****
Mandatory applied	13.71	1.58****	17.49	2.70****	11.69	1.50****	11.50	1.95****
Offender age	−0.07	0.02****	−0.09	0.03***	−0.06	0.01****	−0.12	0.04****
Female offender	−2.55	0.36****	−3.13	0.66***	−2.08	0.25****	−4.91	1.11***
Black offender	0.48	0.18***	0.81	0.24****	0.18	0.15	2.36	0.48****
Hispanic offender	2.41	0.40****	2.29	0.57****	1.97	0.39****	5.66	0.84****
Violent offense	1.10	0.33****	−1.22	1.05	1.09	0.25****	1.91	1.12
Property offense	1.05	0.15****	1.00	0.94	0.94	0.17****	1.66	0.88
Drug offense	−0.55	0.33	−1.33	1.31	−0.08	0.18***	−1.65	0.85*
Bench trial	—	—	—	—	—	—	−9.96	1.31***
Inverse mills ratio	25.02	2.67	32.05	4.71****	20.99	2.47****	28.35	4.71****
Judge-level factors
Judge age	0.00	0.02	−0.05	0.02**	0.02	0.02	−0.07	0.06
Female judge	−0.63	0.21***	−0.42	0.37	−0.63	0.22***	−0.79	0.88
Minority judge	0.93	0.30***	1.74	0.58***	−1.43	0.47***	3.62	0.52****
Judicial tenure	0.02	0.03	0.02	0.02	0.00	0.03	0.12	0.09*
Judicial caseload	−0.21	0.11*	−0.17	0.10*	−0.29	0.11***	−0.04	0.37
County-level controls
Large court	−1.05	0.35***	−0.94	0.42*	−1.00	0.38**	−0.41	1.62
Small court	0.31	0.22	0.24	0.43	0.33	0.23	−3.14	1.89*
Available jailspace	0.00	0.01	−0.02	0.02	0.01	0.01	−0.03	−0.03
Crime rate	0.02	0.03	0.08	0.06	0.02	0.03***	0.13	0.10
Percent minority	0.03	0.03	−0.03	0.06	0.07	0.03**	−0.13	0.15
Between judge R²	8.1%		9.2%		9.6%		11.1%
Between county R²	—		—		—		36.0%
N	69,012		15,097		49,425		4,490

Note. The model for all cases includes controls for mode of conviction. R² estimates not computed for models with insignificant variation in sentence length.

*p ≤ .10. **p ≤ .05. ***p ≤ .01. ****p ≤ .001.

Judicial background characteristics had mixed and inconsistent effects on sentence lengths across modes of conviction. In line with theoretical expectations, judge characteristics explained the greatest proportion of judicial variation for trial cases. Judge factors accounted for 11% of this variance, but this was only marginally greater than for pled cases. The coefficient for female judges was negative for all modes of conviction but failed to reach statistical significance for trial cases (its effect was not statistically different across models). Substantively, this effect was negligible, amounting to less than a 1-month discount across all conviction types. Race of the judge produced stronger effects, but its influence was inconsistent. At trial, minority judges meted out sentences that were about 3.6 months longer than their majority counterparts. This effect was significantly stronger than the effect for both nonnegotiated pleas (Z = 2.42) and negotiated pleas (Z = 7.23), and for negotiated pleas the effect was negative. This highlights the importance of disaggregating judge effects and suggests that the race of the judge may have different meaning and different influence over negotiated pleas and trial cases. Overall magnitude of judge effects tended to be greatest in trial cases (excluding caseload), but the substantive impact of many of these predictors tended to be small and often statistically insignificant. Interestingly, the relative lack of strong judge effects in the sentence length models is consistent with research on individual sentencing disparities that reports stronger extralegal effects for incarceration decisions compared to sentence length outcomes (e.g., Steffensmeier et al., 1993). Future work is needed that investigates the reasons for this finding along with the potential influence of individual judge factors on additional outcomes of interest, such as guidelines departures, the use of intermediate sanctions, and the application of sentencing enhancements such as mandatory minimums.

Discussion

The past three decades have witnessed a virtual revolution in criminal sentencing practices in America (Tonry, 1996). Although contemporary changes in the justice system have been driven by myriad forces, one key concern associated with sentencing reform remains the presences of inter-judge disparities in punishment. Frankel (1973, p. 44) contended, “the splatter of varied sentences” among judges “nourishes the view that there is no justice in the law.” This viewpoint sparked early interest in the effects of judge’s race and gender, but a lack of substantial findings and data limitations soon led to stagnation. Modern attempts at revival have met with a similar fate, leading some scholars to conclude that the powerful forces of selective recruitment and judicial socialization result in a uniform “subculture of justice” among jurists (Frazier & Bock, 1982). Recent policy innovations like presumptive sentencing guidelines almost certainly contribute further to conformity in judicial punishment standards. Despite the considerable impetus for uniformity, though, extant research suffered from numerous shortcomings, including an implicit assumption that judges exercise equivalent sentencing discretion in all types of cases. The current research set out to test this assumption by disaggregating judge effects across modes of conviction.

If judges are free to exercise greater discretion in trial cases, as theoretical expectations suggest, then greater judicial variation should exist and judicial demographic factors should matter more for these cases. The current findings offer tentative support for this expectation. A greater proportion of variation in incarceration decisions emerged between judges for trial cases than for other dispositional outcomes. Similarly, the magnitude of between-judge differences in sentence lengths was twice as great for trials compared to pleas. Overall, variation among judges was greatest for trial convictions, but the magnitude of these differences was not overwhelming.

Regarding the influence of judicial background characteristics, strong support for the importance of modes of conviction emerged for the incarceration decision. Judge’s gender, for instance, had no discernible impact on overall incarceration decisions, nor did it significantly affect negotiated and nonnegotiated pleas. Yet it demonstrated sizeable effects in trial cases, reducing the odds of incarceration by a factor of 0.63 for female judges. Similarly, the age and tenure of the judge only appeared to matter for trial convictions, whereas the race of the judge exerted significant effects across all modes of conviction. One possibility is that the racial background of the judge conditions the negotiation process, though it should be noted that the majority of minority judges were clustered in only a few county courts, so it may be difficult to parcel out the unique effect of judge race from the broader influences of localized courtroom-sentencing norms. Future work is needed with more diverse samples in additional jurisdictions to better investigate these issues. Because trial cases are relatively rare, it was also necessary to combine jail and prison sentences into a single incarceration outcome in the current analysis. Future work on judge effects would benefit from examining these related but distinct dispositions separately (Holleran & Spohn, 2004), and additional work is also needed that begins to investigate the join impact, or intersectionality, of multiple judicial background characteristics in sentencing (Collins & Moyer, 2007). Overall, though, the current study found that judge characteristics tended to exert more pronounced effects on the incarceration decision in trial cases relative to guilty pleas.

The pattern of findings for models examining sentence lengths was less consistent and produced few notable judge effects with the exception of judge’s race. Interestingly, the pooled race effect was watered down by the fact that minority judges exert opposite effects on negotiated pleas and trials. This provides some suggestive evidence that the sentencing process, and the concomitant effect of judge characteristics, differs according to the mode of conviction. It may also suggest that judge characteristics exert indirect effects on the plea negotiation process, but future research is again needed to substantiate that possibility.

Taken together, these results are consistent with the theoretical argument that trial cases provide greater and perhaps different opportunities for judges and other court actors to exercise their sentencing discretion. Oftentimes, the effects of judge factors are on par with individual sentencing effects. For instance, being a minority offender at trial increases the probability of incarceration by about 35% whereas being sentenced by a minority judge at trial decreases it by 22%. Future research is needed that examines the interaction of judge and offender characteristics across modes of conviction. It may be that judge characteristics are particularly strong conditioners of individual disparities in punishment in trial cases where judges exercise the most discretion. Examination of these interactions would provide an additional test of the theoretical propositions outlined in the current work.

Future research is also needed that extends the gamut of punishment outcomes to additional decisions of consequence. Primary among them is the decision to depart from the sentencing guidelines. Because guidelines departures entail substantial exercise of judicial discretion, particularly for cases convicted through trials, judge effects are likely to have important influences over these outcomes. No research to date examines the influence of judge characteristics on guidelines departures. A similar outcome of interest is the use of intermediate sanctions (Johnson & DiPietro, 2012). In Pennsylvania, judges have carte blanche power to divert offenders from incarceration in specified cells of the sentencing guidelines matrix (i.e., the shaded cells in the Appendix). Because this decision is wholly discretionary, one should expect judge characteristics to emerge as potentially important predictors of these decisions. Future research should be guided by thoughtful consideration of how different court actors exercise sentencing discretion for different types of cases and different sentencing decisions. It is necessary to isolate decision-making outcomes that provide reasonable opportunities for individual sentencing philosophies and background experiences to be considered in order to provide a true test of the influence of judicial background characteristics.

Conclusion

As Lafree (1985, p. 291) suggested more than two decades ago, “Comparing sentencing severity for guilty pleas and trials is difficult because they represent different processes to which defendants are nonrandomly assigned.” Despite this poignant observation, the vast majority of research on criminal sentencing continues to utilize combined samples of guilty pleas and trials. If plea and trial convictions do in fact represent separate punishment processes, though, then it is ill-advised to aggregate them in pooled analyses. Moreover, disaggregating trials and pleas may provide important analytical leverage to better disentangle the separate sentencing processes that underlie different types of criminal case disposition (Bushway & Piehl, 2007; Piehl & Bushway, 2007). The present study investigated this issue through a reexamination of the influence that judge’s race and gender exert on criminal case outcomes.

The current findings provide some tentative support that judge characteristics matter more in trial cases because these cases entail greater exercise of judicial sentencing discretion. Overall, individual legal considerations, such as the severity of the offense, dominate individual punishment decisions, but beyond these influences, characteristics of sentencing judges and county court contexts also matter. Rather than impugning prior research, however, the current work serves as a call for additional work on variations in court outcomes across different social contexts of punishment. Ideally, background information on other court actors such as prosecutors and defense counsel would be included and the purview of study extended to other decision-making points in the justice system beyond the final sentence determination. More proximate measures of judicial sentencing philosophies, punitive attitudes, and role orientations are also needed. Judge’s race and gender are imperfect proxies for attitudinal variables that need to be analyzed. Future research should address these and other limitations of the present study.

From a public policy perspective, the current results provide a more nuanced test of inter-judge sentencing disparity under sentencing guidelines. Guidelines, such as other sentencing innovations that have been enacted in recent decades, were powered in large part by concern over unwarranted disparity in sentencing. Although a passel of studies examines the role of individual offender characteristics in punishment, relatively few studies incorporate relevant judge characteristics. The present study suggests that judicial discretion is an important conditioner of individual punishment decisions, but that it is manifest primarily in trial cases which constitute only a small minority of all criminal cases. Thus, greater attention should be devoted in future work to capturing the discretionary powers exercised by additional court actors such as the public prosecutor. The combined forces of judicial selection, socialization and formal sentencing rules may well serve to constrain judicial discretion and limit inter-judge sentencing variation, but very little is known about interprosecutor variation or potential disparities that may arise from early case processing decisions. Future research is therefore needed that identifies and measures these effects if we are to better inform future policy innovations aimed at increasing fairness and effectiveness in criminal punishment in America.

Footnotes

Appendix

1997 Pennsylvania Sentencing Guidelines

Prior record score
Level	OGS	0	1	2	3	4	5	RFEL	REVOC	AGG/MIT
Level 5	14	72-240	84-240	96-240	120-240	168-240	192-240	204-240	240	+/−12
State Incar	13	60-78	66-84	72-90	78-96	84-102	96-114	108-126	240	+/−12
	12	48-66	54-72	60-78	66-84	72-90	84-102	96-114	120	+/−12
	11	36-54	42-60	48-66	54-72	60-78	72-90	84-102	120	+/−12
		BC
	10	22-36	30-42	36-48	42-54	48-60	60-72	72-84	120	+/−12
		BC	BC
	9	12-24	18-30	24-36	30-42	36-48	48-60	60-72	120	+/−12
		BC	BC	BC	BC	BC
Level 4	8	9-16	12-18	15-21	18-24	21-27	27-33	40-52	NA	+/−9
State Incar/RIP trade		BC	BC	BC	BC	BC	BC
	7	6-14	9-16	12-18	15-21	18-24	24-30	35-45	NA	+/−6
		BC	BC	BC	BC	BC	BC	BC
Level 3 State/Cnty	6	3-12	6-14	9-16	12-18	15-21	21-27	27-40	NA	+/−6
Incar		BC	BC	BC	BC	BC	BC	BC
RIP trade
Level 2	5	RS-9	1-12	3-14	6-16	9-16	12-18	24-36	NA	+/−3
Cnty Incar			BC	BC	BC	BC	BC	BC
RIP	4	RS-3	RS-9	RS-<12	3-14	6-16	9-16	21-30	NA	+/−3
RS					BC	BC	BC	BC
	3	RS-1	RS-6	RS-9	RS-<12	3-14	6-16	12-18	NA	+/−3
						BC	BC	BC
LEVEL 1	2	RS	RS-2	RS-3	RS-4	RS-6	1-9	6-<12	NA	+/−3
RS	1	RS	RS-1	RS-2	RS-3	RS-4	RS-6	3-6	NA	+/−3

Note. Aggravated and mitigated ranges add or subtract, respectively, the months shown from the standard range.

Shaded areas of the matrix indicate restrictive intermediate punishments may be imposed as a substitute for incarceration.

When restrictive intermediate punishments are appropriate, the duration of the restrictive intermediate punishment program shall not exceed the guideline ranges.

When the range is RS through a number of months (e.g., RS-6), RIP may be appropriate

CNTY = county; RS = restorative sanctions; REVOC = repeat violent offender category; RFEL = repeat felony 1 and felony 2 offender category; RIP = restrictive intermediate punishments; <;> = less than; greater than; BC = boot camp.

Declaration of Conflicting Interests

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

Funding

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

Notes

Author Biography

Brian D. Johnson is an Associate Professor of Criminology and Criminal Justice at the University of Maryland. His areas of expertise involve social inequality in the justice system, with a particular focus on disparities in criminal case processing and sentencing. Dr. Johnson is the recipient of the 2008 ASC Ruth Shonle Cavan Young Scholar Award from the American Society of Criminology.

References

Anderson

Spohn

(2010). Lawlessness in the federal sentencing process: A test for uniformity and consistency in sentence outcomes. Justice Quarterly, 27, 362-393.

Albonetti

(1990). Race and the probability of pleading guilty. Journal of Quantitative Criminology, 6, 314-334.

Alschuler

A. W.

(1979). Plea bargaining and its history. Columbia Law Review, 79(1), 1-43.

Berk

R. A.

(1983). An introduction to sample selection bias in sociological data. American Sociological Review, 48, 386-398.

Bontrager

Bales

Chiricos

(2005). Race, ethnicity, threat, and the labeling of convicted felons. Criminology, 43, 589-622.

Britt

(2000). Social context and racial disparities in punishment decisions. Justice Quarterly, 17, 707-732.

Bushway

S. D.

Piehl

A. M.

(2007). Social science research and the threat to presumptive sentencing guidelines. Criminology and Public Policy, 6, 461-482.

Bushway

S. D.

Johnson

B. D.

Slocum

L. A.

(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.

Durose

Langan

(2005). State court sentencing of convicted felons, 2002. Retrieved from http://www.ojp.usdoj.gov/bjs/abstract/scsc02st.htm

Bureau of Justice Statistics

10.

Eisenstein

Jacob

(1977). Felony justice: An organizational analysis of criminal courts. Boston, MA: Little Brown.

11.

Frankel

M. E.

(1973). Criminal sentences: Law without order. New York, NY: Hill and Wang.

12.

Frazier

C. E.

Bock

E. W.

(1982). Effects of court officials on sentence severity: Do judges make a difference? Criminology, 20, 257-272.

13.

Gibson

J. L.

(1983). From simplicity to complexity: The development of theory in the study of judicial behavior. Political Behavior, 5, 7-50.

14.

Gruhl

Spohn

Welch

(1981). Women as policymakers: The case of trial judges. American Journal of Political Science, 25, 308-322.

15.

Hogarth

(1971). Sentencing as a human process. Toronto, Canada: University of Toronto Press in Association with the Centre of Criminology.

16.

Holleran

Spohn

(2004). On the use of the total incarceration variable in sentencing research. Criminology, 42, 211-240.

17.

Johnson

B. D.

(2003). Racial and ethnic disparities in departures across modes of conviction. Criminology, 41, 501-542.

18.

Johnson

B. D.

(2006). The multilevel context of criminal sentencing: Integrating judge- and county-level influences. Criminology, 44, 259-298.

19.

Johnson

B. D.

(2010). Multilevel Analysis in the Study of Crime and Justice. In Piquero

Weisburd

(Ed.), Handbook of Quantitative Criminology (Chapter 30). New York, NY: Springer.

20.

Johnson

B. D.

DiPietro

S. M.

(2012). The power of diversion: Intermediate sanctions and sentencing disparity under presumptive guidelines. Criminology, 50, 811-850.

21.

Kautt

P. M.

(2002). Location, location, location: Interdistrict and intercircuit variation in sentencing outcomes for federal drug-trafficking offenses. Justice Quarterly, 19, 633-671.

22.

Klepper

Nagin

Tierney

L.-J.

(1983). Discrimination in the criminal justice system: A critical appraisal of the literature. In Blumstein

Cohen

Martin

S. E.

Tonry

M. H.

, Research on Sentencing: The Search for Reform (Vol. 2). Washington, DC: National Academy Press.

23.

Kritzer

H. M.

Uhlman

T. M.

(1977). Sisterhood in the courtroom: Sex of judge and defendant in criminal case disposition. Social Science Journal, 14, 77-88.

24.

Kulik

C. T.

Perry

E. L.

Pepper

M. B.

(2003). Here comes the judge: The influence of judge personal characteristics on federal sexual harassment case outcomes. Law and Human Behavior, 27, 69-86.

25.

Lacasse

Payne

A. A.

(1999). Federal sentencing guidelines and mandatory minimum sentences: Do defendants bargain in the shadow of the judge? Journal of Law and Economics, 42, 245-269.

26.

Lafree

(1985). Trials and nonadversarial justice: A comparison of guilty pleas and trials. Criminology, 23, 289-312.

27.

Myers

Talarico

(1987). The social contexts of criminal sentencing. New York, NY: Springer-Verlag.

28.

Muhlhausen

D. B.

(2004). The determinants of sentencing in Pennsylvania: Do the characteristics of judges really matter? Center for Data Analysis (Report No. 04-02). Washington, DC: The Heritage Foundation.

29.

Nagel

(1961). Political party affiliation and judges’ decisions. American Political Science Review, 55, 843-50.

30.

Padgett

(1985). The emergent organization of plea bargaining. The American Journal of Sociology, 90, 753-800.

31.

Paternoster

Brame

Mazzerole

Piquero

(1998). Using the correct statistical test for the equality of regression coefficients. Criminology, 36, 859-866.

32.

Piehl

A. M.

Bushway

S. D.

(2007). Measuring charge bargaining. Journal of Quantitative Criminology, 23, 105-125.

33.

Raudenbush

S. W.

Byrk

A. S.

(2001). Hierarchical linear models: Applications and data analysis methods. Thousand Oaks, CA: Sage.

34.

Schanzenbach

M. M.

(2005). Racial and sex disparities in prison sentences: The effect of district-level judicial demographics. Journal of Legal Studies, 34, 57.

35.

Scott

R. W.

(2010). The effects of booker on inter-judge sentencing disparity. Federal Sentencing Reporter, 22, 14-108.

36.

Smith

(1986). The plea bargaining controversy. The Journal of Criminal Law and Criminology, 77, 949-968.

37.

Snijders

T. A. B.

Bosker

R. J.

(1999). Multilevel analysis: An introduction to basic and advanced multilevel modeling. Thousand Oaks, CA: SAGE.

38.

Songer

D. R.

Crews-Meyer

(2000). Does judge gender matter? Decision making in state supreme courts. Social Science Quarterly, 81, 750-762.

39.

Songer

D. R.

Tabrizi

S. J.

(1999). The religious right in court: The decision making of Christian evangelicals in state supreme courts. The Journal of Politics, 61, 507-526.

40.

Spohn

(1990a). The sentencing decision of Black and White judges: Some expected and unexpected similarities. Law and Society Review, 24, 1197-1216.

41.

Spohn

(1990b). Decision making in sexual assault cases: Do Black and Female judges make a difference? Women & Criminal Justice, 2, 83.

42.

Steffensmeier

Britt

C. L.

(2001). Judges’ race and judicial decision making: Do Black judges sentence differently? Social Science Quarterly, 82, 749-764.

43.

Steffensmeier

Herbert

(1999). Women and men policymakers: Does the judge’s gender affect the sentencing of criminal defendants? Social Forces, 77, 1163-1196.

44.

Steffensmeier

Kramer

J. H.

Streifel

(1993). Gender and Imprisonment Decisions. Criminology, 31, 411-416.

45.

Steffensmeier

Ulmer

Kramer

(1998). The interaction of race, gender, and age in criminal sentencing: The punishment cost of being young, Black, and Male. Criminology, 36, 763-797.

46.

Sudnow

(1965). “Normal crimes: Sociological features of the penal code.” Social Problems, 12, 255-264.

47.

Tate

C. N.

(1981). Personal attribute models of the voting behavior of U.S. Supreme Court justices: Liberalism in civil liberties and economics decisions, 1946-1978. The American Political Science Review, 75, 355-367.

48.

Tiede

L. B.

(2009). The impact of the federal sentencing guidelines and reform: A comparative analysis. Justice System Journal, 30, 34-48.

49.

Tiede

Carp

Manner

(2010). Judicial attributes and sentencing-deviation cases: Do sex, race, and politics matter? Justice System Journal, 31, 249-272.

50.

Tonry

(1996). Sentencing matters. Oxford, England: Oxford University Press.

51.

Uhlman

(1978). Black elite decision making: The case of trial judges. American Journal of Political Science, 22, 884-895.

52.

Ulmer

J. T.

Johnson

(2004). Sentencing in context: A multilevel analysis. Criminology, 42, 137-177.

53.

United States Sentencing Commission (USSC). (2004, November). Fifteen years of guidelines sentencing: An assessment of how well the federal criminal justice system is achieving the goals of sentencing reform. Washington, DC: United States Sentencing Commission.

54.

Walker

T. G.

Barrow

D. J.

(1985). The diversification of the federal bench: Policy and process ramifications. The Journal of Politics, 47, 596-617.

55.

Ward

Farrell

Rousseau

(2009). Does Racial Balance in Workforce Representation Yield Equal Justice? Race Relations of Sentencing in Federal Court Organizations. Law & Society Review, 43, 757-806.

56.

Weidner

R. R.

Frase

R. S.

Schultz

J. S.

(2005). The impact of contextual factors on the decision to imprison in large urban jurisdictions: A multilevel analysis. Crime & Delinquency, 51, 400-424.

57.

Welch

Combs

Gruhl

(1988). Do Black judges make a difference? American Journal of Political Science, 32, 126.

58.

Wooldredge

(2010). Judges’ unequal contributions to extra-legal disparities in imprisonment. Criminology, 48, 539-567.

59.

Zatz

(2000). The convergence of race, ethnicity, gender, and class on court decision making: Looking toward the 21st century. Criminal Justice: The National Institute of Justice Journal, 3, 503-552.