Sentencing Recommendations by Probation Officers and Judges: An Examination of Adult Offenders Across Gender

Abstract

Connecting the courtroom workgroup model with attributions and stereotyping based on the focal concerns perspective and gender sentencing literature, the present study investigates the extent to which probation officer recommendations influence judicial sentencing, and whether the gender of the offender further conditions this relationship. Results from logistic and ordinary least squares regression indicate that there is concordance between probation officer recommendations and sentencing by judges. Offender gender has both direct and indirect effects on judicial sentencing through its relationship with probation officer recommendations, and Black males tend to receive lengthier sentences than other race/gender counterparts. These findings provide evidence that probation officer recommendations are an important part of the sentencing process and offer additional insight on how extralegal factors such as gender and race impact criminal justice decision making.

Keywords:

• courtroom workgroup model
• focal concerns
• gender
• judicial sentencing
• probation officer recommendations

Notes

Actuarial risk assessment scales were not part of the PSI report or used in general by the court at the time of the study (2003, 2004). Iowa’s criminal justice sentencing system operates under a minimum and maximum sentencing scheme (Iowa Code 901.5 2009) and does not have the death penalty as a potential sentencing outcome. The court considers the facts and circumstances in imposing judgment by taking into consideration various factors such as age, prior record, circumstances of the offense, and so on. The court must enumerate reasons for selecting the sentence imposed (i.e., State v. Freeman, 404 N.W.2d 188 (Iowa App. 1978); State v. Cooper, 403 N.W.2d 800 (Iowa App. 1987)). Data were not collected to capture sentencing departures. This is an omission that future research should incorporate.

If multiple charges were associated with crime type and crime severity, the most serious charge was included in the analyses.

Data were not available to discern if bail was denied, the defendant received release on one’s own recognizance (ROR), or whether bail was set and the defendant could not pay. In addition, the data do not differentiate between the type of plea bargain.

The total weighted sample size is 1,163. When predicting judicial sentence, there are 920 cases. The discrepancy is based on 243 cases that were dismissed prior to or after trial.

The log transformation was used to normalize the skewed distribution of length of incarceration (in months).

Heckman’s (1976) two-stage procedure was initially employed to create a hazard rate that was included to correct for possible sample selection bias (Berk, 1983). An examination of the diagnostic results, however, revealed the variation inflation factor (VIF) score and the hazard rate to be considerably more than the conventional level of 4 suggesting problems with multicollinearity. Multicollinearity is often a problem in attempting to correct for possible sample selection bias (Bushway, Johnson, & Slocum, 2007). We attempted to first center the mean, but this failed to change the VIF. Thus and following precedent, the hazard rate was omitted from the equations. Instead, we included theoretically relevant variables in the model that may influence the decision-making process. It is important to note that although correction for possible selection bias is widely used in the sentencing literature, there are those that question its appropriateness and indicate that it is often misused (e.g., Bushway et al., 2007; Stolenzberg & Relles, 1990).

We acknowledge that county and geographical differences could emerge in probation recommendations and sentencing outcomes. Unfortunately, there were not enough cases to make county and gender comparisons throughout the multivariate analyses, or when estimating two-way comparisons involving gender and the independent variables with decision making. Although unstable, supplemental models were estimated that controlled for county. Results indicated similar effects reported in the text.

To further validate findings due to the small number of females incarcerated (n = 48), additional analyses were conducted. Reduced models across gender were also estimated and the significant interaction with race and length of sentence remained.

Examination of the data revealed variation in the case distributions handled by probation officers and sentenced by judges. This finding lends support to the belief that the current findings are not based on the practices and decision making of a small number of probation officers and/or judges. To further test for independence, a variable was created that captured the number of probation officers in the data. The inclusion of this variable in supplemental analyses did not alter the statistical significance or direction of the effects involving probation recommendation, gender, and sentencing decisions. The variable did not also predict probation recommendation.

At the time of the study (2003–2004), Hispanics represented less than 2% of the population in this District. Hispanics accounted for a total of 23 cases or less than 1% of the cases studied. Although it is possible that some Hispanics were classified as White, we feel confident that the racial/ethnic classifications and distributions are accurate.