Federal Sentencing of Hispanic Defendants in Changing Immigrant Destinations

Abstract

Research shows that federal punishment disparity between Hispanic and white defendants varies according to federal district court contexts. Recent sociological scholarship has also drawn distinctions between traditional Hispanic immigrant destinations and new or emerging immigrant destinations, and depicted the reception of Hispanic immigrants in new immigrant destinations as less welcoming and supportive than in traditional destinations. We examine whether federal courts in different Hispanic immigrant destinations exhibit differing levels of Hispanic citizen and non-citizen sentencing disadvantage circa 2000, and then circa 2010. We find that traditional destinations exhibit little or no Hispanic versus non-Hispanic disparity, while Hispanic citizens and non-citizens both received longer sentences in new destinations and in non-immigrant destinations circa 2000. By the early 2010s, emerging destinations did not sentence Hispanic citizens and non-citizens significantly differently from traditional destinations. However, new destinations, along with non-immigrant destinations, sentenced Hispanic non-citizens more harshly, especially the undocumented.

Keywords:

• Sentencing
• federal courts
• disparity
• immigration

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes on contributors

Jeffery T. Ulmer is Professor of Sociology and Criminology at Penn State University. His work spans topics such as courts and sentencing, criminological theory and symbolic interactionism, religion and crime, and violent crime rates. He received the 2001 Distinguished New Scholar Award and the 2012 Distinguished Scholar Award from the American Society of Criminology's Division on Corrections and Sentencing.

Brandy R. Parker is a PhD candidate in the Department of Sociology and Criminology at Penn State University. Her research broadly focuses on stratification, crime, and criminal justice contact. Her current projects center on education, employment, and crime; extralegal factors in criminal case processing; and consequences of criminal justice contact.

Notes

1 Shihadeh and Barrano (2010a; 2010b; 2013; Shihadeh & Winters, 2010) argue that Hispanic communities experience relatively low violence rates only in places that are traditional Hispanic immigrant destinations. By contrast, Hispanic communities may experience violence at higher levels in new immigration destinations, though findings by Light (2017) contest this argument.

2 Research has shown that that the percentage of African Americans in the population is positively related to several dimensions of formal social control, such as the size of local police forces (Kent & Jacobs, 2004), the number of black death penalty sentences (Jacobs, Carmichael, & Kent, 2005), and the lynching of blacks (essner, Baller, & Zevenbergen, 2006; Tolnay & Beck, 1995).

3 Our analyses utilize data from time periods when the U.S. Sentencing Guidelines differed in their legal status, known as the Koon era (from 1998 to 2003) and the post Booker/Gall era (roughly after 2005; Stith, 2008). These two time periods coincide with two U.S. Supreme Court decisions (U.S. vs Koon and U.S. vs Booker/Fanfan along with U.S. vs Gall) that greatly affected the amount of federal judicial sentencing discretion. However, regarding judicial discretion and the constraint of the guidelines, these two time periods are more comparable than they appear. The 1996 Koon decision substantially loosened judicial sentencing discretion under the guidelines, and then with the 2003 PROTECT Act, Congress again tightly restricted judicial discretion (Stith, 2008). Our analysis of sentencing circa 2000 involves the less restrictive Koon period. Then, the 2005 Booker decision and the 2007 Gall decision essentially voided the restrictions of the PROTECT Act, rendered the guidelines advisory, and again made the guidelines less constraining of judicial discretion (Stith, 2008). Our 2010 analysis takes place in the Booker/Gall era. Thus, in both time periods of our analysis, federal courts enjoyed relatively greater sentencing discretion under the guidelines, compared to the PROTECT Act era (see Ulmer et al., 2011).

4 For example, immigration cases are commonly handled in “fast track” immigration-focused court proceedings and are often eligible for and given “fast track” departures (Hartley & Tillyer, 2012).

5 Logging the sentence length and re-transforming the regression coefficient by taking its antilog (Hannon, Knapp, & DeFina, 2005) allows us to examine the proportional differences in sentence lengths associated with our variables of interest. The untransformed coefficients will be presented in the tables; however, throughout the text we discuss the effects of our independent measures using the antilog transformation for ease of interpretation, which reports the proportional difference in a one-unit increase in our independent measure at the mean of the dependent variable (Agresti & Finlay, 2009).

6 Many prior federal sentencing studies control for various guideline departures in examining imprisonment and length, while some do not, and others treat them as mediating variables (Ulmer et al., 2016). Here, we control for substantial assistance departures (5K1) in our models. Substantial assistance departures are an identifiably specific reason for a more lenient sentence. In our main analyses, we do not control for judge-initiated departures under Rule 5K2 or government sponsored downward departures, for two reasons. First, the latter two departures represent more open-ended discretionary decisions, by judges and prosecutors respectively, that are more undefined in their specific reasons compared to substantial assistance departures. Since they represent more undefined discretion, it is possible that immigrant threat might operate through such departures, differentially affecting Hispanic defendants’ sentence. Second, we are examining federal sentencing in two time periods when the legal meaning of the sentencing guidelines, and the freedom to depart from them changed substantially (due to U.S. Supreme Court rulings in 2005 that rendered the guidelines advisory rather than mandatory). Also, the USSC data classified non-5K1 departures differently in the late 1990s to early 2000s compared to the latter 2000s, when they differentiated departures into more specific categories. To make our two time period models more comparable, we include 5K1 departures but not the other departures because the meaning and legal basis of the latter changed across time, and also because the USSC coding of departures in the period circa 2000 is not as specific as it is in the later 2000s. In supplemental models discussed below, we leave out substantial assistance departures, and also include all types of departures.

7 The inclusion of criminal history did not result in problematic collinearity with the presumptive sentence measure, and did not alter our findings with regard to Hispanic ethnicity and citizenship or immigration destination type (the correlation between logged Guideline minimum and criminal history is 0.35 and 0.33 for 2000–2002 and 2010–2012, respectively).

8 We chose the four percent cutoff because these districts intuitively map onto federal districts where Hispanics have lived for decades. In addition, to this cutoff aligning with immigrant destination classifications at the state level in past research (e.g. Shihadeh & Barranco, 2010). However, because we are focusing on Hispanic immigration in federal court districts, under our specification districts within states generally considered traditional destinations may have non-traditional classifications. For example, Florida and Texas are generally considered traditional destinations. Here, Florida South is classified as a traditional destination, while Florida Middle and North are classified as new and emerging destinations, respectively. Additionally, Texas North, South, and West are classified as traditional destinations, while Texas East is classified as a new destination (see Appendix). We also considered a five percent Hispanic foreign born cutoff for traditional destinations, but it was too restrictive given the distribution of percent Hispanic foreign born at baseline and growth from 1990 to 2000 and 2000 to 2010.

9 We used a two percent Hispanic foreign-born baseline to avoid categorizing districts with low base rates and small absolute numbers, but high percentage increases in Hispanic foreign born populations as new or emerging destinations, when these would be more accurately categorized as non-destinations. While our main analyses are based on the 75% growth criteria, we run supplemental analyses with 50% and 100% growth cutoffs to assess the sensitivity of our findings to alternative immigrant destination specifications, discussed later. Under the 50% growth cutoff, new destination classifications are unchanged because all such districts experienced at least 75% growth. However, four additional districts are classified as emerging destinations (n = 15; Iowa South, Oklahoma North, Minnesota, and Wisconsin West). Under the 100% cutoff, five fewer districts are classified as new destinations (n = 15; Maryland, Massachusetts, Rhode Island, Virginia East, and Washington East). Massachusetts is re-classified as an emerging destination, while the other five districts are classified as non-destinations. Additionally, four fewer destinations are classified as emerging destinations under the 100% growth cutoff (n = 8; Georgia South, Georgia Middle, Tennessee West, Washington West, though Massachusetts was added as an emerging destination).

10 The data here are not a random sample, but rather the population of federal districts and sentencing cases in these time periods. Thus, we refer to statistical significance as an interpretive convention.

11 For 2010–2012, the new destinations effect is no longer statistically significant, while the non-citizen effect becomes statistically significant (b= −0.02, p < 0.01). Moreover, the cross-level interactions hold with two exceptions: (1) the Hispanic citizen effect for new destinations (b = 0.03, p < 0.05) and (2) the Hispanic non-citizen effect for emerging destinations (b = 0.03, p < 0.05) become statistically significant.

12 Note that the prevalence of Fast Track programs at the district level strongly correlate with immigration caseloads, which we control for. In fact, Fast Track departures heavily cluster in 18 districts that overlap substantially with our traditional destination and new destination of 2000 variables, with smaller numbers and little variation across the other districts. Thus, including a district level predictor of the presence of Fast-Track programs is problematic. Also, Fast Track departures concentrate heavily among immigration offenses, which we omit in our main analyses. Specifically, 82% of Fast Track departures are immigration offenses, and 18% (4498 cases) are non-immigration offenses.