The organizational culture, social context and local judiciary influences on criminal sentencing decisions in Pennsylvania from 1997-1999 are the subject matter for this multi-level, quantitative analysis conducted by Ulmer and Johnson.

[Ulmer, J. T., & Johnson, B. (2004). Sentencing In Context: A Multilevel Analysis. Criminology, 42(1), 137-178.]

The authors elaborate on two pioneering studies by Britt (2000) and Kautt (2002) that offered key methodological and theoretical contributions. This study focuses on state sentencing and includes a wide array of offenses. It also extends Britt's analysis on the effect of race in sentencing data in 3 key ways:

1. broader focus,
2. more extensive local contextual measures (e.g. court characteristics) and
3. more recent data.

The authors proposed a total of 12 hypotheses.

1. Severity of sentencing will vary between counties.
2. The effects of key predictors will vary across counties.
3. County size will be negatively related to sentencing severity.
4. Counties with more conservative politics will issue more severe sentencing.
5. A practical constraint of jail capacity is positively related to incarceration odds.
6. More severe and violent offenses will have a greater effect on incarceration odds in counties with less jail capacity.
7. A defendant's prior record will have a greater effect on incarceration odds in counties with less jail capacity.
8. Heavier county caseloads will be negatively related to sentence severity.
9. Trial conviction (a positive effect on sentence severity) will be greater in counties that have heavier caseload pressure.
10. In counties with lower trial rates, there will be positive effect of trial conviction on sentence severity.
11. Positively related to sentencing severity at the county level will be the volume of Blacks and Hispanics.
12. Blacks and Hispanics will be sentenced more severely in jurisdictions with greater volumes of those minorities, respectively.

Research Design Method:

Individual level sentencing data and county level contextual data from county criminal trial courts in Pennsylvania from 1997-1999 (three years) were collected via the Pennsylvania Commission on Sentencing (PCS). These data contain information on all misdemeanor and felony sentences in the state plus type of offense, race, gender and age of the defendant, as well as case-specific attributes such as the application of mandatory minimum sentences. Contextual data from the U.S. Census, Uniform Crime Reports and the 1999 County and City Extra were also used. Cases in the analysis were limited to the most serious offense per judicial transaction and to those utilizing the 1997 guidelines.

Dependent Variables: Two dependent variables were utilized:

• the in/out incarceration decision;
• the number of months those incarcerated were sentenced.

Incarceration was coded 1 if the offender was sentenced to any length of confinement in a county jail or state prison. Non-incarceration options such as probation, restitution, etc were coded as 0. The sentence length variable was coded to equal the number of months the offender was sentenced to serve.

Independent Variables: Several individual case and contextual level factors served as independent variables including:

• severity of the current offense (measured using the Offense Gravity Score ranging from a 1 as least serious to a 14 as the most severe),
• offense type (measured with 3 dummy variables: 1=violent offense, 2=property offense, 3=drug offense),
• prior criminality of the offender (measured using the Prior Record Score, an 8 category scale ranging from 0 - 8 with the last two categories reserved for repeat felons and repeat violent offenders),
• sentence recommendations (what the guidelines indicate as an appropriate sentence -- for in/out models coded as 1 if incarceration was recommended and 0 otherwise -- for sentence length models coded to equal the minimum number of months recommended per guidelines);
• presence or absence of mandatory minimums (coded as a dummy variable to control for the mandatory minimums). Offender demographics, as noted earlier, were also examined. Case-processing factor, the mode of conviction, was also included. This was measured with 2 dummy variables: 1 if convicted via negotiated plea or a trial, 0 if through non-negotiated plea. Data not collected included: defendant's socioeconomic status, type of attorney and bail.

In addition to the individual factors, a variety of aggregate level contextual measures for the 67 counties in Pennsylvania were included:

• court size,
• judicial caseload,
• trial rate,
• available incarceration capacity of each county.

Court size was trichotomized into large, medium and small based on both the number of trial judges and the proportion of cases adjudicated in each county (modeled after Ulmer, 1997). Judicial caseload was determined by dividing the number of total criminal cases in each county by the number of sentencing judges. Trial rate was measured as the percentage of cases determined via jury trial.

Other variables included: political context of the county (based off voting recordings in the 1996 presidential election), available jail space (total number of beds divided by the number of cases in that county, higher ratio equally higher jail capacity), county poverty rates, amount and type of crime in each county.

Sample Population

Pennsylvania offers valuable jurisdiction information because it has operated under sentencing guidelines since 1982; presenting a potentially strict test of contextual variation in sentencing relative to non-guideline jurisdictions. Sentencing guidelines offer uniformity such as:

• quantifying/standardizing sentence decision criteria;
• mandate courts to consider these criteria;
• recommend uniformity in sentencing ranges.

In contrast, Pennsylvania contains two of the largest U.S. largest cities (Philadelphia and Pittsburgh) as well as numerous medium-sized cities and small rural communities, so there is wide variance in terms of resources and crime rates. In addition, there are wide variances in prosperity, racial, ethnic, religious, political affiliation and cultural diversity across the state. This is key as judges and district attorneys are elected and not appointed.

Measurements

Hierarchical linear modeling (HLM) was utilized as it provided several advantages over traditional analytical strategies such as ordinary least squares (OLS). The authors utilized this analytical strategy to investigate various complexities surrounding the influence of individual and contextual factors on sentencing outcomes. Over several pages, they discuss the benefits and strengths of this type of analysis and provide sources for greater detail on hierarchical logistic modeling.

A one-way random effects ANOVA was utilized to examine the unconditional models and to partial out the amount of variation in each sentencing outcome occurring at each level of analysis. This allowed the authors to determine the amount of variance between versus within counties, and provided a baseline from which later models could be evaluated. Next, level 1 explanatory variables were introduced via random coefficients ANCOVA with individual level predictors to estimate the effects of individual characteristics on sentencing outcomes. This allowed the authors to evaluate reductions in variance at each level of analysis due to individual level characteristics. It also allowed the fixed and random effects of level 1 explanatory variables to be examined.

Level 2 predictors were included next (random coefficients ANCOVA models with level 1 and level 2 covariates) to provide key information about mean differences in sentencing patterns across counties and attempt to explain via the aggregate variables. Finally, the authors estimated interactive models with level 1 and level 2 variables and cross level interactions fully specified (random coefficients ANCOVA models with cross level interactions).

Findings

In both the incarceration and sentence length models, significant variation existed between counties. The sentence length model demonstrated that this variation was small compared to the amount of within-county variation. Legal factors such as offense severity and prior record strongly related to both the likelihood of incarceration and length of sentence. Violent offenses were associated with increased sentencing severity. Offender characteristics also significantly influenced sentences; Black, Hispanic, male and younger offenders received more severe sentences. Also, mode of conviction was an important level 1 determinant of sentencing severity; offenders that go to trial have increased likelihood to be incarcerated and longer sentences. While most of the variation in sentencing did occur at the individual case level, there were significant between-county variations in sentencing not explained via individual case factors.

Major findings include:

• considerable sentencing variation noted between counties (hypothesis 1 support),
• significant variation existed between counties in the effects of all individual level predictors (support for hypothesis 2),
• large courts incarcerated less and gave shorter sentences (support for hypothesis 3),
• local jail capacity was positively related to incarceration odds (support for hypothesis 5),
• counties with heavy caseloads were less likely to incarcerate (support for hypothesis 8),
• trial penalties were greater in counties with heavier caseloads (support for hypothesis 9 -- caseload did not account for all the between-county variation),
• Blacks were given longer sentences in counties with greater black population percentages and Hispanics were given longer sentences in counties with greater Hispanic population percentages (support for hypothesis 12).
• Of the 12 hypotheses tested,  seven are supported or partially supported in the direction predicted (furnished in a well-constructed table in the Discussion section); hypotheses #4, #6, #10 were not supported; #11 was supported for Hispanics; and #7 was significant but in the direction opposite of the hypothesis.

Review Summary

This organizational culture research article has four merits that are demonstrated throughout:

1. appropriately delineating a hierarchical linear model (HLM) of research (a template) for examining outcome variables suspected of being unduly and inappropriately influenced by social/human services cultures (i.e., accounting for interaction effects between organizational culture and independent (intervention) variables;
2. the need to translate/communicate the research to the widest possible audience (professionals and lay readers alike);
3. presenting a must-read for CJS professionals in general;
4. reinforces the often noted influence local context (of which organizational culture is a component) has on outcome measures across all disciplines and markets.

This study was framed using a term borrowed from Salvesberg (1992): focal concerns theory; i.e., that culture-specific concerns embedded within any local court community lead to substantive rationalities in how they carry out sentencing; instead of there being a world view, a localized cultural view governs.

Implications

Dilemmas exist between due process rights and the organizational structure of the court system as it plays out in the community.  It was found that defendants who invoke their right to a trial and are found guilty are sentenced more severely than those who plead guilty through the plea bargain process, thus undermining the constitutional right to trial.  Additionally, where one is sentenced has a measurable effect on the length of one's sentence, which runs contrary to the notion of equal justice under the law.

Tools/Knowlecge Objects/Resources

• Britt, C. (2000). Social context and racial disparities in punishment decisions. Justice Quarterly, 17(4), 707-732.
• Kautt, P. M. (2002). Location, Location, Location: Interdistrict and Intercircuit Variation in Sentencing Outcomes for Federal Drug-Trafficking Offenses. Justice Quarterly, 19(4), 633.
• Salvesberg, J. (1992). Law that does not fit society: sentencing guidelines as a neoclassical reaction to the dilemmas of substantivized law. American Journal of Sociology, 97, 1346-1381.
• Ulmer, J. T., & Kramer, J. H. (1996). Court Communities under Sentencing Guidelines: Dilemmas of Formal Rationality and Sentencing Disparity. Criminology, 34(3), 383-408.