Based on what you read in the Haynes article, do you think the courtroom workgroup is good or bad for defendants? Why? If bad, what do you think can be done to solve it?
Courtroom Workgroups and Sentencing
The Effects of Similarity, Proximity, and Stability
Stacy Hoskins Haynes Barry Ruback Pennsylvania State University Gretchen Ruth Cusick Chapin Hall Center for Children and University of Chicago
Sentencing decisions are the product of a group of courtroom actors, primarily judges and district attorneys. Although the structure of the courtroom work- group and the interdependencies among members are assumed to be important determinants of sentencing decisions, the degree of this importance and the specific mechanisms through which workgroups affect these decisions have not been investigated. This study used data from the Pennsylvania Commission on Sentencing (PCS) for the years 1990 to 2000 to examine how three social psychological aspects of courtroom workgroups (similarity, proximity, and stability) affect sentencing decisions. Results indicated (a) that workgroups generally had very high levels of similarity in terms of race, gender, and polit- ical party but lower levels of similarity in terms of age, college education, and law school education and (b) that proximity and stability were generally high. Controlling for individual, case, and distal contextual factors, workgroup fac- tors affected the decision to incarcerate, the decision to impose fines, and the decision to impose restitution. In particular, proximity increased the use of economic sanctions relative to incarceration and stability was associated with a decrease in the imposition of economic sanctions. Similarity had inconsistent effects.
Keywords: courtroom workgroups; sentencing; contextual factors; incar- ceration; economic sanctions
Sentencing decisions represent the culmination of investigative, legal, practical, and policy determinations. Although sentencing is primarily a
function of two factors—the severity of the offense and the offender’s prior
Crime & Delinquency Volume 56 Number 1
January 2010 126-161 © 2010 SAGE Publications
10.1177/0011128707313787 http://cad.sagepub.com
hosted at http://online.sagepub.com
126
Haynes et al. / Courtroom Workgroups and Sentencing 127
record—individual, case, and contextual effects can contribute to variations in sentencing outcomes. In addition, the relationships between individuals involved in sentencing—the judge, prosecutor, and defense attorney—may be important because of shared beliefs about what constitutes appropriate sentencing. These individuals, who together form the courtroom work- group, share a common task environment and work together to achieve the common goal of disposing of cases.
This study uses hierarchical modeling techniques to examine whether and how sentences are affected by three aspects of courtroom workgroups: similarity (i.e., the degree to which workgroup members share the same characteristics), proximity (i.e., the location of workgroup members’ offices in relation to one another), and stability (i.e., the number of years workgroup members worked together in the same jurisdiction). These three quantitative measures of the proximal context (i.e., characteristics of the individuals responsible for the handling of cases), along with measures of individual, case, and distal contextual factors (i.e., characteristics of the jurisdiction in which cases are processed), are used to investigate the deci- sion to incarcerate, the decision to impose fines, and the decision to impose restitution. In addition, this study examines how workgroup factors affect statutory implementation, specifically restitution decisions before and after a 1995 statutory change making restitution mandatory.
Overview of Sentencing Research
Legally relevant factors, particularly crime seriousness and offense history, are the most important determinants of sentencing outcomes (Hofer, Blackwell, & Ruback, 1999; Myers & Talarico, 1987), but studies have also examined how offender characteristics, judge characteristics, and contextual characteristics affect sentencing.
Offender Characteristics
Although some studies have found that extralegal factors such as the offender’s race/ethnicity, gender, and age have little effect on sentencing,
Authors’ Note: This research was supported by grants from the National Institute of Justice (Grant No. 97-CE-VX-0001) and the Pennsylvania Commission on Crime and Delinquency and by funds from the Center for Research on Crime and Justice at Penn State University. The points of view expressed in this article do not necessarily reflect the opinions of the Department of Justice, the Pennsylvania Commission on Crime and Delinquency, or the Pennsylvania Commission on Sentencing. We thank Cynthia Kempinen for her thoughtful comments on an earlier draft.
128 Crime & Delinquency
independent of legally prescribed variables such as the severity of the offense and the offender’s prior record (Konecni & Ebbesen, 1982), other studies have found that offender characteristics contribute significantly to sentencing outcomes (Steffensmeier, Ulmer, & Kramer, 1998). Studies have focused primarily on three offender characteristics: race/ethnicity, gender, and age. Regarding race/ethnicity, studies indicate that Black and Hispanic offenders receive harsher sentences than White offenders (Kramer & Ulmer, 1996; Steffensmeier & Demuth, 2000). Although Black and Hispanic offenders are generally more likely to be incarcerated than White offenders, they do not necessarily receive lengthier sentences (Spohn, 2000). With regard to gender, evidence suggests that the criminal justice system treats female offenders more leniently than male offenders (Hofer et al., 1999; Steffensmeier et al., 1998). Regarding age, evidence suggests that both older and younger offenders receive more lenient sentences than offenders in the middle of the age distribution (Steffensmeier, Kramer, & Ulmer, 1995). In sum, research suggests that although offender characteris- tics contribute significantly to sentencing outcomes, the effects are small when compared to legally relevant factors.
Judge Characteristics
Most studies that have examined the effects of decision-maker charac- teristics on sentencing outcomes have focused primarily on judges. Judges’ socialization experiences, which are a product of their political affiliation, demographic characteristics, and educational experiences, influence their later values, attitudes, beliefs, and ideologies. For example, conservative judges embrace a more punitive stance toward crime and are more likely to attribute an individual’s criminal involvement to personal choice, whereas liberal judges focus more on rehabilitating the offender and are more likely to believe that factors external to the individual are responsible for their involvement in crime (Carroll, Perkowitz, Lurigio, & Weaver, 1987).
In addition to political beliefs, some research suggests that judges’ race and gender affect sentencing. Findings regarding the effects of race on sen- tencing are mixed regarding the imposition of incarcerative sentences. For example, some studies find that minority judges are more likely to incar- cerate offenders (Steffensmeier & Britt, 2001), whereas others find that minority judges are somewhat less punitive (Johnson, 2006). With regard to gender, studies consistently find greater similarities than differences in how male and female judges sentence offenders (Gruhl, Spohn, & Welch, 1981). However, some evidence suggests that female judges are somewhat harsher
Haynes et al. / Courtroom Workgroups and Sentencing 129
in their sentencing decisions, particularly toward repeat Black offenders, because they are more strongly influenced by offender characteristics and prior record than are male judges (Steffensmeier & Hebert, 1999).
In sum, there appear to be no consistent findings regarding the effects of individual judges’ characteristics on sentencing. Rather than continue to focus on the characteristics of individual judges, it might make sense to examine how those characteristics relate to those of people with whom judges interact.
Contextual Characteristics
The differential weighting of individual and case factors is probably the result of both the judge who handed down the decision and the jurisdiction in which the case was processed. Thus, in addition to examining how judi- cial discretion affects the decision-making process, studies have examined how the jurisdiction conditions that discretion. The court context refers to both the proximal context—the characteristics of the individuals responsi- ble for the handling of cases—and the distal context—the characteristics of the jurisdiction in which cases are processed.
Studies consistently find that otherwise similar offenders receive differ- ent sentences in rural and urban areas. Research at both the state and fed- eral levels finds that offenders sentenced in large urban courts receive less severe sentences than offenders sentenced in small rural courts (Hofer et al., 1999). Because urbanization correlates with other court and county vari- ables, however, its effect depends on the offense, the specific sentencing outcome under consideration, and the attributes and behavior of offenders (Myers & Talarico, 1986, 1987; Olson, Weisheit, & Ellsworth, 2001).
A county’s political ideology (i.e., the extent to which residents are con- servative or liberal) is important because judges’ decisions generally reflect the views of their constituents. Thus, judges in more conservative jurisdic- tions tend to impose longer sentences, particularly on repeat offenders and offenders convicted of more serious offenses (Eisenstein, Flemming, & Nardulli, 1988). Furthermore, a study of 337 jurisdictions in seven states revealed that male offenders and Black offenders received longer sentences in more conservative political environments (Helms & Jacobs, 2002).
In sum, although sentencing studies have examined legally relevant fac- tors, offender characteristics, judge characteristics, and contextual factors, many of these studies have examined only one type of factor (typically offender characteristics). There is a need for studies that test all of these fac- tors simultaneously. Moreover, there is a need to acknowledge the interplay of these factors. This study does that with regard to judges, by analyzing
how judges relate to the other primary member of the courtroom work- group, the district attorney.
A Social Psychological Framework for Studying Courtroom Workgroups
According to Eisenstein and his colleagues (Eisenstein et al., 1988; Eisenstein & Jacob, 1977; Nardulli, Eisenstein, & Flemming, 1988), court- room workgroups consist of individuals who share a common workplace, who interact in the performance of their jobs, and whose collective purpose is to dispose of cases. Those researchers conducted two studies that inves- tigated how courtroom workgroup traits, offender characteristics, and case factors affect the disposition process. In their first study, Eisenstein and Jacob (1977) examined more than 4,000 cases in felony courts in Baltimore, Chicago, and Detroit. Ten years later, Eisenstein et al. (1988) conducted a more extensive study that further explored how courtroom workgroups dispose of cases. In this second study, the researchers analyzed data on nearly 7,400 defendants and 300 judges, prosecutors, defense attorneys, and other participants in three medium-sized criminal courts in each of three states: Illinois, Michigan, and Pennsylvania.
On the basis of these studies, Eisenstein and his colleagues concluded that differential patterns of sentencing occur because courtroom work- groups perceive offenders and cases differently (Eisenstein & Jacob, 1977; Eisenstein et al., 1988). Furthermore, because sentencing is a shared deci- sion, the structure of the courtroom workgroup and the interdependencies among members explain much of the variance in sentencing outcomes across jurisdictions. The results suggested that factors such as the degree of familiarity among workgroup members, the context in which the court is located, and the county legal culture (i.e., workgroup members’ shared beliefs about both interpersonal relations and the manner in which cases should be disposed of) affect workgroup behavior.
The notion of workgroups is powerful, but there have been few quantita- tive studies that have examined how they affect sentencing, and there have been no real tests of their importance relative to individual and case charac- teristics. Moreover, Eisenstein and his colleagues’ research focused on courtroom workgroups in only 12 medium-sized counties, thereby neglect- ing how workgroups across the entire range of counties in a state—from small rural counties to very populous urban counties—might affect sentenc- ing (Eisenstein & Jacob, 1977; Eisenstein et al., 1988).
130 Crime & Delinquency
Haynes et al. / Courtroom Workgroups and Sentencing 131
A further limitation of the research on workgroups is that it ignores the social psychological literature on small groups, particularly work relating to the formation of groups and the manner in which groups carry out tasks. Three factors from the social psychological work are relevant to Eisenstein et al.’s research on courtroom workgroups: similarity (i.e., the degree to which individuals share the same characteristics), proximity (i.e., spatial closeness), and stability (i.e., interacting with the same individuals over time). We chose to focus on these variables because similarity and proxim- ity are two of the most important predictors of group formation. Thus, not surprisingly, because most group members come from the same social net- work there tends to be considerable overlap in members’ knowledge, expe- riences, and perspectives (Jehn, Northcraft, & Neale, 1999). Furthermore, stability affects group performance by reducing uncertainty about others’ intentions and probable behavior. We expected that quantitative opera- tionalizations of all three social psychological aspects of courtroom work- groups—similarity, proximity, and stability—would affect sentencing outcomes.
Similarity
Eisenstein et al. (1988) observed that members of courtroom work- groups are similar to each other. Given that similarity, they should like one another, as people are attracted to others who have similar attitudes, beliefs, and personal characteristics (Byrne, 1971; Newcomb, 1961), and they value the contributions of similar others more than the contributions of dissimilar others (Hinds, Carley, Krackhardt, & Wholey, 2000).
In a courtroom workgroup, there are several possible dimensions of sim- ilarity that might be relevant to how the workgroup operates, including demographic characteristics (race, gender, and age), background (college and law school education), and beliefs (political party). Although numerous studies in the organizational literature have examined how differences in age, gender, race, job tenure, and education affect performance (Jehn & Bezrukova, 2004; Tsui, Porter, & Egan, 2002), few studies have examined the role of these factors in sentencing. Our first step was to determine the similarity of workgroups, that is, the extent to which workgroup members (judges and district attorneys) share the same characteristics. Second, we wanted to determine whether that degree of similarity is related to sentenc- ing decisions. In general, we expected that similarity among workgroup members would be relevant primarily for decisions that involve discretion. In contrast, for decisions that are straightforward or automatic, workgroup
132 Crime & Delinquency
similarity should not be very important. Furthermore, although we expected all three types of similarity to be important, we expected education to matter more than demographic characteristics because research shows that over time age and gender become less relevant (Jehn et al., 1999).
Proximity
Individuals who are physically closer to one another generally interact more and are therefore more likely to like each other than are individuals who are more distant from one another (Festinger, Schachter, & Back, 1950). Because organizational settings provide such proximity, they create opportu- nities for individuals to form interpersonal network ties (McPherson & Smith-Lovin, 1987). Consistent with these findings, Eisenstein and his col- leagues (1988) observed that courtroom workgroups are highly dependent on their environments. The geography of the courthouse (e.g., the availability of informal meeting places) influences the structure of court communities and the opportunities available for workgroup members to interact. These oppor- tunities lead to greater familiarity, which, in turn, facilitates cooperation and improves the decision-making process.
It is likely that proximity relates to whether there are opportunities for workgroup members to discuss alternative sanctions. When workgroup members’ offices are in the same building, for example, there are more opportunities for workgroup members to come into contact with one another outside the courtroom. These informal meetings may affect sen- tencing outcomes because they provide workgroup members with opportu- nities to discuss the general appropriateness of different sanctions and, we believed, would make intermediate punishments like economic sanctions more likely to be imposed.
Stability
Workgroup members become more familiar with one another when they have more interactions over time, which are likely to increase when there are few changes in workgroup personnel. The stability of a group’s mem- bership determines the familiarity among members (Goodman & Leyden, 1991) and their patterns of communication (Katz, 1982). These factors exert a significant effect on the group’s performance because groups com- posed of more familiar individuals generally perform better than groups composed of less familiar individuals (Guzzo & Dickson, 1996).
Workgroup stability also reduces uncertainty about others’ intentions and probable behaviors. Because groups are generally more productive when members’ behaviors are predictable, individuals typically prefer to work with others whose personalities and work practices are familiar to the rest of the group (Hinds et al., 2000). This preference explains why Eisenstein and Jacob (1977) found that workgroup members who were more familiar with one another were more likely to dispose of cases by plea bargaining.
We expected that stability would affect the imposition of economic sanc- tions. Members of more stable workgroups have more joint experience with offenders and are therefore probably more realistic about offenders’ ability to pay economic sanctions. Furthermore, we expected that stability would affect statutory implementation. A statutory change is likely to be success- fully implemented when it is consistent with individuals’ beliefs and prac- tices and when the individuals and organizations responsible for its implementation are both capable and willing to transform the policy into practice (Goggin, 1986). In 1995, Pennsylvania implemented a statutory change mandating that courts order restitution for those victims who had suffered a physical injury or monetary loss as the direct result of a crime (18 Pa. C.S.A. §1106). The statute required judges to impose full restitution regardless of the offender’s ability to pay. Studies have found that the statute had an effect, as courts in Pennsylvania ordered restitution more often for offenses after the statutory change (Ruback, Ruth, & Shaffer, 2005; Ruback, Shaffer, & Logue, 2004). We expected that decisions by more stable workgroups would be more consistent across the pre– and post–statutory change periods than would decisions by less stable work- groups. That is, we expected more stable workgroups to be less likely to change their rate of imposition of restitution following the statutory change.
The Current Study
The current study used statistical techniques appropriate for multilevel data to examine individual, case, and contextual effects on sentencing outcomes across an entire state. Furthermore, we built on the work of Eisenstein and his colleagues (Eisenstein & Jacob, 1977; Eisenstein et al. 1988) by examining how three quantitative measures of the courtroom workgroup—similarity, proximity, and stability—affect the decision to incarcerate, the imposition of fines, and the imposition of restitution. We focused on these three sentencing outcomes because they range in type (i.e., one incarcerative sanction and two economic sanctions), severity, and amount of discretion. We also examined how workgroup characteristics
Haynes et al. / Courtroom Workgroups and Sentencing 133
134 Crime & Delinquency
affected courts’ response to a statutory change that made restitution manda- tory. Courtroom workgroups may consist of judges, prosecutors, defense attorneys, probation officers, police officers, and administrative personnel. Although all of these individuals play some role in the disposition process, our data are limited to judges and district attorneys, individuals who are employed by the state, who have a more permanent role in the courtroom workgroup, and who handle the most cases.1
For each of the analyses, we first analyzed all 67 Pennsylvania counties. Then we conducted those same analyses excluding Allegheny (Pittsburgh) and Philadelphia Counties, the two largest urban areas in the state, because urban areas are likely to be highly formal and bureaucratic and decision making is more likely to be based primarily on legal factors (Dixon, 1995). By excluding these counties, we could determine whether the effects of workgroups are stronger when only the less bureaucratized counties are examined.
Method
We used hierarchical modeling techniques to examine individual, case, and contextual (both proximal and distal) effects on sentencing outcomes. Because criminal cases are nested within different counties, similarities among cases at the county level are likely to occur (i.e., cases are likely to be more similar within counties than across counties). This similarity means that residual errors tend to be correlated within counties, which vio- lates the ordinary least squares assumption of independent error terms and risks the misestimation of standard errors. Hierarchical modeling tech- niques address this problem of dependence by partitioning the variance within and between units of analysis. Thus, multilevel models are appro- priate for this study because they allow simultaneous tests for individual and contextual effects on the dependent variable. The sentencing data were analyzed using the hierarchical linear modeling (HLM) program of Raudenbush and Bryk (2002).
The county-level (level-2) data for this study consisted primarily of contextual information (i.e., proximal and distal characteristics) from The Pennsylvania Manual (Commonwealth of Pennsylvania) for the years 1989 to 2000, the 1990 and 2000 U.S. Censuses, and the 1990 and 1995 Uniform Crime Reports (Federal Bureau of Investigation). We collected additional information from the Martindale-Hubbell Law Directory (Martindale- Hubbell, 2003), the Pennsylvania District Attorneys Association (2004), and personal communications with court personnel in all 67 Pennsylvania
Haynes et al. / Courtroom Workgroups and Sentencing 135
counties. The case-level (level-1) data consisted of sentencing information (i.e., individual and case characteristics) from the Pennsylvania Commission on Sentencing (PCS) for the years 1990 to 1994 and 1996 to 2000.2
Contextual Variables
The contextual variables included in this study were measures of both the proximal context and the distal context. The proximal context refers to characteristics of the courtroom workgroup (i.e., characteristics of the indi- viduals responsible for the handling of cases), whereas the distal context refers to characteristics of the county in which the court is located (i.e., characteristics of the jurisdiction in which cases are processed).
Proximal context. Three measures of the proximal context were included in these analyses: similarity among workgroup members, proximity of workgroup members’ offices in relation to one another, and stability of workgroup membership. The information for these variables came primar- ily from The Pennsylvania Manual (Commonwealth of Pennsylvania) for the years 1989 to 2000.
The first set of variables included six measures of workgroup member characteristics: race, gender, age, political party, and college and law school location. Race was a dichotomous variable coded 1 for Whites and 0 for nonwhites. Gender was coded 1 for males and 0 for females. Age was a continuous variable that represented a workgroup member’s age (in years) in 1990. Political party was a dichotomous variable coded 1 for Republicans and 0 for Democrats. College and law school location were coded 1 for Pennsylvania schools and 0 for non-Pennsylvania schools. We used these variables to create separate measures of similarity for each workgroup member characteristic. For all of the variables except age, similarity was whether the variable had the same value for each judge and the county dis- trict attorney (see Table 1). For example, if the district attorney was White (as they all were), our measure of race similarity reflected the percentage of judges who were White. Furthermore, if the district attorney attended col- lege in Pennsylvania, our measure of college similarity reflected the per- centage of judges who attended Pennsylvania colleges. Our measure of age similarity reflected the percentage of judges who were within 5 years of the district attorney’s age. For each type of similarity, we computed the per- centage similarity for each year and then averaged across the 11-year period for each of the 67 counties.
136 Crime & Delinquency
Table 1 Description of the Coding Scheme
Variable Coding Mean SD
County level Urbanization % of the population living in an urban area 45.30 27.03 Economic climate % of the population living below poverty line 11.28 3.58 Political climate % of voters who voted for Republican candidate 49.46 7.83 Crime rate Part I offenses per 100,000 population 2,411.72 1,047.56 Gender composition % of males in the population 48.80 1.17 Stability Number of years worked together 3.77 1.48 Proximity 1 = offices in the same building, 0.75 0.44
0 = offices not in the same building Race similarity % of judges the same 99.14 3.58
race as the district attorney Gender similarity % of judges the same gender 90.24 19.56
as the district attorney Age similarity % of judges within ± 5 years 32.84 25.95
of the district attorney’s age College similarity % of judges who went to 45.06 37.01
college in Pennsylvania Law school similarity % of judges who went to law 41.38 37.99
school in Pennsylvania Political party similarity % of judges the same political 68.45 30.58
party as the district attorney Case level
Prior record High prior record 0.18 0.38 (dummy variables) Low prior record 0.22 0.41
No prior record (reference) 0.60 0.49 Offense severity Range 0.07-1.00 0.31 0.18 Type of offense Person offense 0.21 0.41
(dummy variables) Property offense 0.35 0.48 Drug offense 0.20 0.40 Traffic offense 0.14 0.35 Other offense (reference) 0.09 0.29
Type of disposition Jury trial 0.02 0.15 (dummy variables) Bench trial 0.04 0.19
Plea (reference) 0.94 0.24 Type of sentence Prison 0.13 0.34
(dummy variables) Jail 0.42 0.49 Probation (reference) 0.45 0.50
Fines 1 = fines imposed, 0 = fines not imposed 0.55 0.50 Restitution 1 = restitution imposed, 0.36 0.48
0 = restitution not imposed Offender race 1 = White, 0 = non-White 0.62 0.49 Offender gender 1 = male, 0 = female 0.84 0.37 Offender age Age at time of offense 30.32 9.58
The second variable was a measure of proximity that controlled for whether the president judge (the administrative head of the county court elected by all of the judges) and district attorney had offices in the same building. This was a dichotomous variable coded 1 if their offices were in the same building and 0 if their offices were not in the same building.3 In all of the counties except Allegheny (Pittsburgh) and Philadelphia, all judges are in the same building as the president judge. In these two coun- ties, we coded that the president judge and the district attorney did not have offices in the same building.
The final variable was a measure of workgroup stability that represented the number of years the judge and county district attorney worked together within the same jurisdiction. For each county, we aggregated across all judges to compute the average workgroup stability for each year and then averaged across the 11-year period. For any given year, the average work- group stability ranged from 0 to 14 years. For any given judge, the number of years he or she worked with the county district attorney ranged from 0 to 22 years.
Distal context. We included five variables representing the social, politi- cal, and criminal justice context of each county. Three variables came from the 1990 and 2000 U.S. Censuses. The first variable, a measure of urbaniza- tion, was the percentage urban population within each county. The second variable, a measure of the county’s economic climate, was the percentage of the population living below the poverty level. The third variable, the percent- age of males within each county, controlled for the individuals most prone to engaging in crime. We calculated each of these variables separately using the 1990 (pre–statutory change) and 2000 (post–st