Sentencing Native Americans in US Federal Courts: An Examination of Disparity

Abstract

Existing sentencing literature largely focuses on the study of white, African-American, and to a lesser extent, Hispanic offenders. Unfortunately, very little is known about the sentencing of Native American offenders, especially in the federal courts. To address this shortcoming, the current study employs United States Sentencing Commission data for the fiscal years 2006-2008 to examine the comparative punishment of Native Americans. Consistent with the focal concerns perspective and its reliance on perceptions of race-based threat, findings demonstrate that Native Americans are often sentenced more harshly than white, African-American, and Hispanic offenders. Moreover, race-gender-age interactions indicate that during the incarceration decision, young Native American males receive the most punitive sentences, surpassing the punishment costs associated with being a young African-American or Hispanic male. These findings highlight the importance of directing increased attention toward the sentencing of this understudied offender population.

Keywords:

• sentencing
• Native American
• race
• disparity

Notes

1. Though it would be desirable to include Asian offenders in the current analysis, the geographic distribution of these offenders created a scenario where too few federal districts sentenced both Native American and Asian offenders to allow for meaningful comparisons between the two groups.

2. The following 28 districts are included in the analysis: New York North, New York West, North Carolina East, North Carolina West, Mississippi South, Texas West, Michigan West, Wisconsin East, Iowa North, Minnesota, Nebraska, North Dakota, South Dakota, Alaska, Arizona, Idaho, Montana, Nevada, Oregon, Washington East, Washington West, Colorado, New Mexico, Oklahoma East, Oklahoma North, Oklahoma West, Utah, and Wyoming.

3. The guidelines apply to most felony cases heard in the federal courts, but are also applicable to Class A misdemeanors.

4. Studies that examine the in/out and sentence length decision typically calculate and include a hazard rate in the sentence length model to correct for sample selection bias. A recent study by Bushway, Johnson, and Slocum (2007) has demonstrated that the hazard rate is rarely used correctly by researchers, with a frequent and substantial problem being the omitted use of exclusion restrictions in the selection model (i.e. predictors of the in/out decision that are not used in the prediction of sentence length). In the absence of useful exclusion restrictions, the hazard rate frequently introduces substantial multi-colinearity into the models, creating estimates that are less accurate than those uncorrected for selection bias. Since the current study was unable to employ theoretically sound exclusion restrictions, the advice of Solzenberg and Relles (1990) is followed, and uncorrected estimates are presented in the models.

5. This measure is logged for the analyses that examine the logged sentence length variable.

6. As argued by Ulmer (2000) and expanded upon by Doerner & Demuth (2010), the criminal history and offense severity scales employed by the guidelines have both main and interactive effects that are not able to be fully accounted for by the presumed sentence measure. Thus, it would be ideal to incorporate all three of these measures into analyses of federal sentencing guidelines, but doing so creates a problem with multicolinearity. To solve this problem, researchers (e.g. Doerner & Demuth, 2010; Johnson & Betsinger, 2009) typically include the presumed sentence measure along with the six-point criminal history scale, but exclude the offense severity scale. The current study follows this strategy.

7. Upward departures were captured through the use of two measures as indicated in Table 1 (i.e. upward departure and upward departure with Booker). Since the occurrence of these departures in the sample is relatively rare, modeling their unique effects in the analyses produced unstable parameter estimates. As a result, these two measures were collapsed to form a single measure of upward departures for use in the analyses.

8. Early disposition departures (also referred to as “fast track” departures) were developed to help reduce caseloads, particularly in jurisdictions dealing with large numbers of immigration related crimes. The justification is grounded in the financial savings produced by securing prompt cooperation with the defendants.

9. In a recent study of trial penalties in the federal courts (Ulmer, Eisenstein, & Johnson, 2010), it was demonstrated that plea versus trial sentence differences are in part explained by three guidelines approved factors. These included whether or not the defendant has accepted responsibility for their offense, whether or not they cooperated with or assisted the prosecution, and alternatively, whether or not they obstructed justice through the trial process (e.g. committing perjury). Since each of these factors is included in the presumed sentence measure used in this analysis, the remaining effect of trial is largely due to other unapproved factors, such as penalizing the offender for choosing a more difficult and uncertain trial processes.

10. To account for both inter-district variation in sentencing as well as intraclass correlation among cases processed within the same district courts, the models include a series of dummy variables that identify the relevant district. This approach has been discussed by previous researchers and is frequently used in lieu of more complex hierarchial modeling (see Helms & Jacobs, 2002; Johnson & Betsinger, 2009 for examples).

11. Following the lead of recent sentencing research (Johnson & Betsinger, 2009), for the purpose of creating appropriate race/ethnicity-gender-age categories, age is operationalized as young (less than 30 years of age) versus old (30 years of age or older). This strategy helps to ensure a large enough number of cases within each race/ethnicity–gender–age category for analysis, but also attempts to delineate between those most likely viewed as being in their crime-prone years from those who are not.

12. It should also be noted that, similar to most studies of sentencing outcomes, the current analysis is unable to consider the effect of the victims’ race/ethnicity on the sentencing outcomes. To the extent that the inter-racial nature of offending varies from one racial ethnic group of offenders to the next, sentencing outcomes could be effected. Future research is needed to address this concern in the sentencing of federal offenders.

*p < .05.

13. Upward departures were captured through the use of two measures as indicated in Table 1 (i.e. upward departure and upward departure with Booker). Since the occurrence of these departures in the sample is relatively rare, modeling their unique effects in the analyses produced unstable parameter estimates. As a result, these two measures were collapsed to form a single measure of upward departures.

*p < .05.

14. The percent change in the logged sentence length variable is calculated from the exponentiated coefficients reported for the sentence length models using the following formula: Y = 100(e ^b  – 1). As described by other sentencing researchers (see Ulmer, Eisenstein, & Johnson, 2010), when the sentence length measure is logged, taking the antilog of the regression coefficients allows for a proportional interpretation of their effects on sentence length. Using this approach provides a more exact estimation as compared to the common alternative approach of simply multiplying the non-exponentiated coefficients by a factor of 100.

15. Economic offenses comprise a variety of different offending types including fraud, embezzlement, forgery, bribery, tax offenses, money laundering, gambling/lottery-related offenses, burglary, larceny, and auto theft. While it may not be ideal to aggregate all of these offenses together for analysis, the number of Native Americans in the sample limits the degree to which offense types can be individually examined.

*p < .05.

*p < .05.

16. It should be noted that the absence of racial disparity when examining economic offenses could be, in part, due to the relatively broad array of behaviors captured by this offense type. Examining some of these behaviors separately would certainly be useful for future research, as the current study had too few Native Americans in the sample to address this concern.