The Joint Effects of Offender Race/Ethnicity and Sex on Sentence Length Decisions in Federal Courts

Abstract

The current study examined the main and interactive effects of offender race/ethnicity and sex on sentence length decisions for drug offenders convicted in three federal courts located in Iowa, Minnesota, and Nebraska. The additive model showed that females received shorter prison sentences than similarly situated male offenders, but there were no differences between white offenders and minority offenders. However, when the data were partitioned by sex, black males were found to receive lengthier prison terms than white males. There were no differences between white males and Hispanic males, and white females were treated no differently than either black or Hispanic females. Moreover, when the data were partitioned by race/ethnicity, white females were treated no differently than white males. However, black females received shorter sentences than black males and Hispanic females received shorter sentences than Hispanic males. Further analyses showed that black and Hispanic males also received longer sentences than white females and that black males received longer sentences than all other offenders (with the exception of Hispanic male offenders). These findings mesh with those gleaned from other sentencing studies, although they are at odds with theoretical notions that leniency at the sentencing stage is reserved only for white women.

Introduction

Minority offenders and males are disproportionately overrepresented in US prison populations. Of the estimated 1.5 million inmates who were in state or federal prisons at the end of 2007, approximately 59 percent were black or Hispanic and 93% were male (West and Sabol 2008, Table 5). Incarceration statistics further reveal how the degree of minority overrepresentation varies by an offender’s sex; a larger proportion of the male prison population consists of minority inmates. At yearend 2007, white inmates constituted 33% of the male prison population; by comparison, 48% of all females incarcerated in state and federal prisons were white (West and Sabol 2008, Table 5). Relative to their white counterparts, the rate of incarceration was 3.6 times higher for black women at mid-year 2007, and Hispanic women were about 1.5 times more likely to end up behind bars (Sabol and Couture 2008, Table 10). The rates were more disparate among male offenders. Based on data for the same time period, black men were about six times more likely to be in prison than white men, and the rate of incarceration was more than two times higher for Hispanic men than for white men (Sabol and Couture 2008, Table 10). In addition, researchers estimate that the lifetime chances of going to prison are highest for black males (32% or 1 in 3), followed by Hispanic males (17% or 1 in 6), and then white males (6% or 1 in 17) (Bonczar 2003, p. 8). Interestingly, however, the lifetime chances of going to prison for black females are nearly as high as they are for white males (5.6%) (p. 8). In contrast, about one in 50 (2.2%) Hispanic females and approximately one in 100 (0.9%) white females are expected to go to prison during their lifetimes (Bonczar 2003, p. 8). These statistical data suggest that criminal justice outcomes differ for minority offenders and for females and that race/ethnicity and sex interact to influence outcomes.

There are now a number of studies that examine the additive effects of race/ethnicity and/or sex on sentencing outcomes, and excellent literature reviews have been written about the effects of both variables. In the two most recent reviews of the effect of race on sentencing outcomes, Spohn (2000) and Mitchell (2005) reached the same conclusion—race matters. Both found that minority offenders received more severe punishments, even after offense seriousness and prior criminal record were taken into account (see also Chiricos and Crawford 1995). With regard to the effect of sex on sentencing outcomes, Nagel and Johnson (1994) reviewed the extant literature and noted that “female offenders consistently received more favorable sentences than similarly situated males” (p. 185). Moreover, Daly and Bordt (1995) reviewed research findings from studies published through the mid-1990s and “found no case in which the overall results showed that men were favored” (p. 144). For the most part, female offenders were more likely than males to receive lenient sentences. Similar conclusions are also reported in more recent reviews (see for example Brennan 2002; Rodriguez et al. 2006; Steffensmeier et al. 1998).

The above conclusions are based on a body of sentencing research that largely tests additive models, but additive models do not allow one to assess whether offender race/ethnicity and sex operate jointly to influence outcomes. In other words, past research is largely premised on the assumption that all women are likely to be treated more leniently than all men and that all minority offenders are likely to receive harsher punishment than all white offenders. But, findings of leniency for female offenders may be conditioned by race/ethnicity. And, findings of preferential treatment for white offenders (or more punitive treatment for minorities) may be conditioned by sex. Thus, as other researchers have warned, a failure to consider the intersection of sex and race/ethnicity may result in inaccurate conclusions about the effects of these variables on sentencing outcomes (Brennan 2002, 2009; Crew 1991; Daly and Tonry 1997; Steffensmeier and Demuth 2006; Steffensmeier et al. 1993, 1998; Spohn and Beichner 2000; Young 1986).

Are the effects of race/ethnicity on sentencing outcomes similar or different for male and female offenders; does race/ethnicity predict sentencing outcomes among women? Are the effects of offender sex invariant across offender race/ethnicity? Relative to other males and females of varying races/ethnicities, do black men receive the harshest sentences? Do white women, in particular, receive sentences that are less severe than those imposed on all other offenders? Empirical questions such as these have received limited attention in sentencing studies.

The current study adds to the extant research in the area of sentencing by examining the joint effects of offender race/ethnicity and sex on sentence length decisions for drug offenders convicted in three federal courts located in Iowa, Minnesota, and Nebraska. Our offense- and region-specific study replicates and modifies Steffensmeier and Demuth’s (2006) recent study of sentencing outcomes for offenders convicted of a wide array of offenses (i.e., both drug and non-drug) in 54 large urban counties. We focus on drug offenders because Steffensmeier and Demuth (2000) argue that sentencing outcomes are likely to be particularly severe for minority drug offenders (see also Crow and Johnson 2008; Portillos 1998; Steffensmeier and Demuth 2001; Demuth and Steffensmeier 2004; Mitchell 2005). Our focus on drug offenders also follows Mitchell’s (2005) recommendation that examinations of sentencing disparity should be conducted at lower levels of aggregation (e.g., offense specific).

Although there are now a number of studies that focus on the issue of racial and/or ethnic disparities in the sentencing of drug offenders, research findings are somewhat mixed (for a detailed review of this literature see Brennan and Spohn 2008). Some researchers find that black drug offenders are sentenced more severely than whites (see for example Albonetti 1997; Brennan and Spohn 2008; Demuth and Steffensmeier 2004; Rodriguez et al. 2006; Steen et al. 2005; Steffensmeier and Demuth 2000, 2001; Unnever 1982; Unnever and Hembroff 1988). However, other researchers find that sentencing outcomes do not differ between similarly situated black and white drug offenders (see for example, Chiricos and Bales 1991; Curry and Corral-Camacho 2008; Engen and Steen 2000; McDonald and Carlson 1993; Spohn 1999). With regard to white-versus-Hispanic differences, some find more lenient sentencing outcomes for whites (see for example, Albonetti 1997; Brennan and Spohn 2008; Demuth and Steffensmeier 2004; Rodriguez et al. 2006; Steffensmeier and Demuth 2000, 2001), but others find that Hispanic and white drug offenders are treated similarly (see for example Curry and Corral-Camacho 2008; Lagan 1996; McDonald and Carlson 1993; Spohn 1999). Thus, overall results are far from conclusive. Furthermore, most studies of drug offender sentencing do not allow for comparisons to be made between black, white, and Hispanic females. This is unfortunate because, as Steen et al. (2005) have suggested, stereotypes about drug offending are likely to be influenced by stereotypes about gender and race/ethnicity. The current study, therefore, adds to our understanding of the context by which discrimination may operate for male and female drug offenders of varying races/ethnicities.

Literature Review

Theoretical Framework

Stereotypes play a role in criminal justice decision-making. Labeling theorists, for example, contend that classifications based upon a variety of stereotypes (some of which may be derived from sex and/or race/ethnicity) influence both the attribution of a deviant label and reactions to that label (Becker 1963; Bernstein et al. 1977a, b; Schur 1971, 1983). Harris (1977) has similarly argued that perceptions of the types of behaviors or roles that are likely or unlikely for one to exhibit or perform are derived from how one is “type-scripted.” Type-scripts may be used to identify those who are likely to be deviant and, consequently, those who deserve harsher punishment.

In line with discussions of the influence of stereotypes on decision-making, Albonetti’s (1991, 1997, 2002) “uncertainty avoidance/causal attribution” perspective explains the harsher sentences imposed on minorities and males. Albonetti (1991) contends that when people are forced to make decisions but do not have all relevant information, they will attempt to reduce uncertainty by relying upon prior experience, stereotypes, and prejudice. Judges and prosecutors, she argues, rely on stereotypes of minorities and males as more dangerous and likely to recidivate to help them attempt to achieve “rational outcomes in the face of incomplete knowledge” (Albonetti 1997, p. 797).

“Focal concerns” theorists similarly argue that judges have limited case information and, therefore, develop a “perceptual shorthand” based on stereotypes linked to an array of offender characteristics, including race/ethnicity, gender, social class, and other social positions (Engen et al. 2003, p. 110; Kramer and Ulmer 2002, p. 904; Steffensmeier and Demuth 2000, p. 709; Steffensmeier et al. 1998, p. 768; Ulmer and Johnson 2004, p. 145). These offender characteristics (along with offense type and prior record) influence sentencing outcomes because images or attributions connect them to groups thought to be bad (or good) risks for rehabilitation, potentially dangerous (or not), and more (or less) culpable (Steffensmeier and Demuth 2000, p. 709). With regard to the effect of race/ethnicity, negative racial and ethnic stereotypes allow court actors to assume that minority offenders are more dangerous, more culpable, and more likely to commit future crime (see for example Steffensmeier and Demuth 2006). Such preconceived notions are believed to result in harsher sentences.

Much has been written about the prevalence of negative racial/ethnic stereotypes in our society (Barak 1994; Brennan 2002, 2006; Dates and Pease 1997; Entman 1990, 1992, 1994, 1997; Farr 2000; Humphries 1981; Hurwitz and Peffley 1997; Kurokawa 1971; Landrine 1985; Madriz 1997; Peffley et al. 1996; Sarat 1993; Surette 1992). Kurokawa (1971) points out that in American society, where whites comprise the majority, “white ethnocentrism prevails, attributing a positive image to the whites and a negative one to other racial groups” (p. 214). Although this statement was made decades ago, more recent studies lend support for this notion (for general studies see, for example, Entman 1997; Gladwell 2005; Smith 1990; for studies of media depictions of minorities and crime see Barak 1994; Barlow 1998; Entman 1990, 1992, 1994, 1997; Humphries 1981; Hurwitz and Peffley 1997; Madriz 1997; Sarat 1993; Smith 1990).

In a series of studies, Entman (1990, 1992, 1994, 1997) found that, across time, media depictions of criminal activity by African Americans were significantly more likely to (1) emphasize violent or drug crime (1992, 1994), (2) receive the greatest percentage of news coverage (1990, 1992), (3) be accompanied by a mug shot (1990, 1997; see also Peffley et al. 1996), (4) emphasize racial/ethnic differences between the offender and victim (1990), (5) have the case spoken about by a criminal justice official not of their own race (1992), and (6) show African Americans in police custody (1994, 1997). Unfavorable media portrayals such as these influence how minorities are perceived, which furthers their negative treatment (Hurwitz and Peffley 1997, p. 376).

While the arguments and conclusions drawn from the earlier mentioned literature are informative, the aforementioned discussions are limited in they reveal little about how gender and race/ethnicity combine to influence expectations. In those rare instances where sex and race are simultaneously considered, researchers who study media depictions of offenders suggest that the most common image of a criminal is that of a young, black (or other minority) male (Barak 1994; Barlow 1998; Chermak 1994; Humphries 1981; Madriz 1997). Sentencing scholars likewise argue that the brunt of negative stereotyping falls most heavily on young, unemployed, black males (see for example Spohn and Holleran 2000; Steffensmeier et al. 1998). But what does this say about perceptions of minority women? How are minority women viewed relative to others? And, how may these perceptions influence court processing outcomes?

A review of the extant literature reveals that minority women have been stereotyped more negatively than white women (Brennan 2002, 2006; Castro 1998; Farr 2000; Healey 1997; Landrine 1985; Madriz 1997; Portillos 1998; Young 1986). Landrine (1985), for example, found that white women were more likely to be stereotyped as “competent, dependent, emotional, intelligent, passive…and warm” (p. 72), whereas black women were more likely to be stereotyped as “dirty, hostile, and superstitious” (p. 71–72). Furthermore, other scholars have found a strong tendency for minority females, in general, to be stereotyped as “hyper sexed” (Farr 2000, p. 55; see also Madriz 1997; Young 1986) and as “welfare queens” (Hurwitz and Peffley 1997, p. 393). In short, racist conceptions of “femininity” more closely fit white women (Klein 1995). Consequently, some have speculated that women of color are less likely than white women to be accorded preferential treatment (Brennan 2002, 2006; Belknap 1996; Farnworth and Teske 1995; Griffin and Wooldredge 2006; Visher 1983; Young 1986). Indeed, studies of media portrayals of female offenders reveal that white women are more likely than minority women to have their criminal behavior excused in some way (Brennan and Vandenberg 2009; Bond-Maupin 1998; Farr 1997, 2000; Huckerby 2003). Thus, it is possible that negative racial and ethnic stereotypes for minority women may result in sentences that are similar to those given to male offenders. In other words, findings of leniency for female offenders may be conditioned by race/ethnicity.

Prior Empirical Research

Are all women (regardless of their race/ethnicity) more likely to receive preferential treatment relative to their male counterparts or is leniency reserved only for white women? Researchers have attempted to answer this question in one of two ways. First, some have constructed race/ethnicity-specific models of sentencing outcomes to determine whether offender sex emerges as a statistically significant predictor in separate models for blacks, whites, and Hispanics. Overall, findings from this body of research do not support the assertion that leniency at the sentencing stage bypasses minority women. Instead, findings from studies with race/ethnicity-specific models suggest that female offenders are likely to receive more lenient sentences than male offenders of the same race/ethnicity. Specifically, researchers have found that black females are less likely to be incarcerated than black males (Albonetti 1997; Gruhl et al. 1984; Spohn et al. 1985; Steen et al. 2005; Steffensmeier and Demuth 2006). Among whites, the likelihood of receiving a prison or a jail sentence is higher for males (Albonetti 1997; Gruhl et al. 1984; Steen et al. 2005; Steffensmeier and Demuth 2006). And, Hispanic females are less likely to be put behind bars than Hispanic males (Gruhl et al. 1984; Steffensmeier and Demuth 2006). Studies of the length of incarceration generally support these findings (Albonetti 1997; Steen et al. 2005; Steffensmeier and Demuth 2006).

Others have used a different approach to assess whether all women are equally likely to receive preferential treatment relative to males. Specifically, some researchers have included race/ethnicity-by-sex interaction terms or dummy variables in their sentencing models to make this assessment. Findings from these examinations generally support the notion that white women are treated more leniently than white men, black men, and/or Hispanic men (Curry and Corral-Camacho 2008; Spohn and Beichner 2000; Spohn and Spears 1997; Steffensmeier and Demuth 2006). They do not, however, support the notion that minority women are denied preferential treatment relative to males of the same (or other) race/ethnicity.¹ In fact, Spohn and Spears (1997) found that black females were less likely to be incarcerated and were more likely to receive shorter prison sentences than white males. Thus, relative to men, black and Hispanic women, like white women, appear to “benefit more from their female status than would be expected all else equal (i.e., given their racial/ethnic status)” (Steffensmeier and Demuth 2006, p. 257).

Black men, in contrast, are likely to be treated more punitively than females of varying races/ethnicities (Curry and Corral-Camacho 2008; Hartley et al. 2007; Spohn and Beichner 2000; Spohn and Spears 1997) and white males (Crew 1991; Curry and Corral-Camacho 2008; Hartley et al. 2007; Kruttschnitt 1984; Spohn and Beichner 2000; Spohn and Holleran 2000; Steffensmeier and Demuth 2006; Steffensmeier et al. 1993, 1998). This may especially be the case for young, black males (see Spohn and Holleran 2000; Steffensmeier et al. 1998, but see Curry and Corral-Camacho 2008). To summarize, findings from the extant literature indicate that relative to their male counterparts, females of all races/ethnicities seem to benefit as a result of their gender. And, relative to whites and females of varying races/ethnicities, black males seem to be penalized as a result of their race.

But, what does the extant literature reveal about how minority women are treated relative to white women? To answer that question some researchers have included race/ethnicity-by-sex interaction terms or dummy variables in their models of the sentencing outcome. There is little evidence from those studies to indicate that white women receive preferential treatment relative to other women. In fact, in two separate studies of sentencing outcomes that occurred before and after guidelines implementation, researchers found either no differences between black and white females or differences that favored black females (Griffin and Wooldredge 2006; Koons-Witt 2002). Others have also observed that black and white women are equally likely to be incarcerated (Spohn and Beichner 2000; Spohn and Spears 1997). And, research findings from studies with sex-specific models generally lend support to the conclusion that race/ethnicity does not make a difference among women (Bickle and Peterson 1991; Crew 1991; Farnworth and Teske 1995; Kruttschnitt 1984; Spohn and Beichner 2000; Steffensmeier and Demuth 2006). These research findings suggest that judges do not differentiate between women of varying race/ethnicities. Consistent with the focal concerns perspective, it would appear that judges perceive female offenders, irrespective of their race/ethnicity, as less dangerous, less blameworthy, and more amenable to rehabilitation than male offenders.²

However, in two separate examinations of sentencing outcomes in Pennsylvania, Steffensmeier et al. (1993, 1998) found that black females were more likely to be incarcerated and were incarcerated for longer periods than white women. Moreover, findings from examinations with exclusive focus on female offenders suggest that offender race may play a role among female drug offenders in particular. To date, however, only two researchers have published studies about sentencing outcomes for this specific population of offenders. Kruttschnitt (1980–81 and 1982) examined outcomes for 1,034 female defendants processed between 1972 and 1976 in a northern California county; she analyzed whether black females received more severe sentences (i.e., as measured by an 8-category sentence severity scale) than white females. Among drug offenders, black women were sentenced more severely than similarly situated white women (Kruttschnitt 1980–81 and 1982).³

In a more recent study, Crawford (2000) examined data for 1,103 incarcerated females in Florida who were eligible for sentencing as habitual offenders. In general, black women were almost twice as likely as white women to be “habitualized” (p. 268). Subsample analyses of drug offenders revealed larger race effects. Black females charged with drug offenses were more than nine times more likely than similarly charged white females to be sentenced as habitual offenders (p. 273). Moreover, “white females [were] rarely targeted for habitualization for drug offenses; nearly 96% (21 of 22) of the women sentenced as habitual offenders with a drug-related offense were black” (Crawford 2000, p. 274). Thus, in contrast to findings from other examinations that indicate that race does not affect the sentences imposed on female offenders, the results of other studies discussed herein suggest that negative racial stereotypes affect females as well as males.

The research conducted to date, then, provides inconsistent evidence regarding the joint effects of offender race and sex. Moreover, there are very few studies that compare sentences imposed on Hispanic women to those imposed on white women, and no studies that examine whether black and Hispanic female drug offenders are given lengthier prison sentences than white female drug offenders. The current study expands prior research by addressing both of these limitations.

Based on the extant theoretical and empirical literature, we test the following hypotheses:

Additive Analyses⁴

H1::

Black and Hispanic drug offenders will receive longer prison sentences than white drug offenders

H2::

Female drug offenders will receive shorter prison sentences than male drug offenders

Sex-Specific Analyses⁵

H3::

Black and Hispanic male drug offenders will receive longer prison sentences than white male drug offenders

H4::

Black and Hispanic female drug offenders will receive longer prison sentences than white female drug offenders

Race/Ethnicity-Specific Analyses⁶

H5::

White female drug offenders will receive shorter sentences than white male drug offenders

H6::

Black female drug offenders will receive shorter sentences than black male drug offenders

H7::

Hispanic female drug offenders will receive shorter sentences than Hispanic male drug offenders

The Joint Effects of Race/Ethnicity and Sex⁷

H8::

Black male drug offenders will receive longer sentences than all other drug offenders

H9::

White female drug offenders will receive shorter sentences than all other drug offenders

Data and Methods

The data for this study consist of a subset of data collected for a study of charging and sentencing decisions in three United States District Courts: the District of Minnesota, the District of Nebraska, and the Southern District of Iowa. The data file includes detailed information on all offenders sentenced in these courts during fiscal year 1998, fiscal year 1999, and fiscal year 2000.⁸ For this study, we selected all cases for which the offender was convicted of a drug-trafficking offense involving powder cocaine, crack cocaine, methamphetamine, or other drugs (i.e., heroin and marijuana). We eliminated cases (N = 25) with offenders who were Asian or Native American, as well as cases with missing data on the independent variables. This resulted in a data file with 1,547 cases: 530 from the Southern District of Iowa, 439 from the District of Minnesota, and 578 from the District of Nebraska.

We use the data from these three district courts, rather than data from all district courts in the United States, for two reasons. First, a number of the offender characteristics included in the analysis—whether the offender had a prior drug-trafficking conviction, the offender’s employment and marital status, and the number of charges initially filed against the offender—are not included in the publicly available data files from the United States Sentencing Commission. Because we had access to the presentence reports in these three district courts, we were able to collect this information. Second, and perhaps more importantly, there is mounting evidence that sentence outcomes vary significantly among federal district courts (Hofer et al. 1999; Kautt 2002; LaCasse and Payne 1999; Spohn 2005). This calls into question the conclusions of studies of federal sentencing decisions that use data aggregated across all district courts. As Weisselberg and Dunworth (1993, p. 27) have noted, “It is extremely difficult, and perhaps unhelpful, to draw general, system-wide conclusions about the effect of the guidelines upon the district courts.” Because of these concerns, we use data from three relatively homogeneous US District Courts and we control for the district in which the case was adjudicated.

Dependent and Independent Variables

Descriptive statistics for the dependent and independent variables are presented in Table 1. Whereas sentencing research often examines both the decision to incarcerate or not and the length of the sentence, we limit our analysis to the length of the prison sentence. This is because there were only 33 drug offenders in these three district courts who were not sentenced to prison. These offenders are not included in the analyses. The original dependent variable was sentence length, measured in months. Because the distribution of the values was positively skewed, we logged the sentence length variable.

Table 1 Descriptive statistics

The independent variables of interest are the offender’s race/ethnicity and sex. Race/ethnicity is measured with three dummy variables, Black, Hispanic, and White; white offenders are the reference category. The offender’s sex is measured with a dichotomous variable (Female) that is coded “1” if the offender is female and “0” if the offender is male. To measure the interaction between the offender’s race/ethnicity and sex, we created six dummy variables: Black Male, White Male, Hispanic Male, Black Female, White Female, and Hispanic Female. There are 324 black males, 516 white males, 445 Hispanic males, 60 black females, 151 white females, and 51 Hispanic females.

Our analyses control for other offender and case characteristics that previous research has shown to influence sentencing decisions. With regard to controls for offender characteristics, we include dummy variables for citizenship status (Non-citizen), employment status (Unemployed), marital status (Married), and whether the offender had dependent children (Kids). Three dummy variables are included to measure the offender’s educational achievement (No High School Degree, High School, Some College/College Degree), with those holding less than a high school degree as the reference category. The offender’s Age is a ratio-level variable that measures the offender’s age in years. In addition, dummy variables that indicate whether the offender had a prior drug-trafficking conviction (Prior Drug Trafficking) or was under the control of the criminal justice system as the time of arrest (CJ Control) are included.

With regard to controls for case characteristics, we include measures of the Presumptive Sentence, the number of counts filed by the US Attorney (Counts Filed), and the number of counts for which the offender was convicted (Conviction Counts). The series of analyses also include dummy variables that indicate whether the offender was in custody prior to the sentencing stage (In Custody), whether the offender received a regular downward departure (Downward) or a substantial assistance departure (Substantial Assistance), and whether the offender pled guilty to the charges (Pled Guilty). Moreover, we control for the type of drug involved in the offense, which is measured by four dummy variables (Powder Cocaine, Crack Cocaine, and Methamphetamine, with Other Drugs as the reference category), and for the jurisdiction in which the case was adjudicated (Minnesota and Nebraska, with Southern Iowa as the reference category).

Although prior research on federal sentence outcomes controlled for the offense seriousness score and the offender’s criminal history score, the present study controls for the presumptive sentence, which is the approach recommended by Engen and Gainey (2000) and the United States Sentencing Commission (2004).⁹ The presumptive sentence, which is based on the offense seriousness score and the criminal history score, is the minimum sentence that judge could impose without departing from the guidelines. In order to account for mandatory minimum sentences that were prevalent in drug cases, the presumptive sentence was measured as the guideline minimum unless a mandatory minimum sentence was triggered and indicated a longer sentence than the guideline minimum. In such cases, the presumptive sentence was measured as the mandatory minimum sentence. If there was a mandatory minimum sentence but the safety valve was applied, the presumptive sentence was the guideline minimum. Because the presumptive sentence was positively skewed (like prison sentence length), the log of the presumptive sentence was included in the analysis.

As shown in Table 1, the mean prison sentence imposed on offenders adjudicated in these three district courts was 88.37 months. Three quarters of the offenders were either white (43.1%) or Hispanic (32.1%); 24.8% of the offenders were black. Most offenders (83.1%) were male and their average age was about 32 years. The typical offender had at least one dependent child, and was a US citizen, employed at the time of the offense, and not married. Approximately, 35% of the offenders were under some type of criminal justice system control at the time of the offense and 25% had a prior conviction for a felony drug-trafficking offense.

The mean presumptive sentence was 115.31 months; this was 27 months longer than the mean sentence (88.37 months) that was imposed on these offenders. This disparity between the presumptive sentence and the actual sentence reflects the fact that 10% of the offenders received a regular downward departure and 39% received a downward departure for providing substantial assistance. The mean number of indictment counts (2.56) was more than twice the mean number of conviction counts (1.23). More than half (54.8%) of the offenders were convicted of an offense involving methamphetamine; most of the remaining offenders were convicted of an offense involving either crack (22.3%) or powder (14.5%) cocaine. Almost two-thirds (62.3%) of the offenders were in custody at the time of the sentence hearing and 92.0% pled guilty.

The correlation matrix, which is presented in Appendix 1, reveals that (with only five exceptions) all of the independent variables had a significant effect on the length of the prison sentence. The only variables that did not affect sentence length were whether the offender had a prior drug-trafficking conviction and the offender’s ethnicity, age, citizenship status, and marital status. With respect to the key independent variables, the bivariate correlations reveal that the offender’s race and sex had significant effects on sentence length; being black was associated with a longer sentence (r = .23; p = .00), being white is associated with a shorter sentence (r = −.22; p = .00), and being female is associated with a shorter sentence (r = −.22; p = .00). The correlation matrix also demonstrates that only two of the independent variables (Hispanic offender and non-citizen) were strongly related (i.e., with correlations over .70).

Analytical Procedures

We use ordinary least squares regression to analyze the dependent variable, the logged prison sentence length. To test our hypotheses regarding the additive effects of the offender’s race/ethnicity and sex, we analyze the sentence length decision for the full sample. We then partition the data, first by the offender’s sex and then by the offender’s race/ethnicity, to test our sex-specific and race/ethnicity-specific hypotheses. Next, we use six race/ethnicity-by-sex dummy variables to assess whether black males received the harshest sentences and whether white females received the most lenient sentences; we run the analysis twice—first with black males and then with white females as the reference category.¹⁰

Multivariate Findings

The results of the OLS regression analysis for the full sample of drug offenders are shown in Table 2. Because the sentence length variable is logged, the unstandardized regression coefficient (the B coefficient) is not easily interpreted.¹¹ The standardized coefficient (Beta) indicates the strength of the relationship between an independent variable and the dependent variable. With respect to the key variables of interest—the offender’s race/ethnicity and sex—the results reveal that, contrary to our first hypothesis, there were no differences in the sentences imposed on white offenders (the reference category) and either black or Hispanic offenders. On the other hand, and consistent with Hypothesis 2, female offenders received significantly shorter sentences than similarly situated male offenders. The only other offender characteristics that affected the length of the sentence were the offender’s employment status and criminal justice status; unemployed offenders received longer sentences than those who were employed, and offenders who were under the control of the criminal justice system at the time of the offense received longer sentences than those who were not under some type of criminal justice system control.

Table 2 Results of OLS regression analysis of logged sentence length: full sample

Not surprisingly, the strongest predictors of the length of the sentence imposed by the judge (as measured by the standardized regression coefficients) were the presumptive sentence (beta = .78), whether the offender received a substantial assistance departure (beta = −.40), and whether the offender received a regular downward departure (beta = −.11). Offenders facing longer presumptive sentences received significantly longer sentences, and offenders who received either type of downward departure received significantly shorter sentences. The sentences imposed on offenders convicted of offenses involving crack cocaine or methamphetamine were significantly longer than those imposed on offenders convicted of offenses involving marijuana or other drugs. In addition, offenders who were in custody at the time of sentencing received significantly longer sentences. Finally, offenders adjudicated in Minnesota and Nebraska received shorter sentences than did offenders adjudicated in Southern Iowa.

The results discussed thus far reveal that the offender’s sex, but not the offender’s race/ethnicity, affected the length of the prison sentence in these three US District Courts. However, as shown in Table 3, which presents the results of the analysis of the data partitioned by the sex of the offender, although the race/ethnicity of the offender did not affect the length of the sentence imposed on female offenders, offender race did affect the length of the sentence for male offenders. Among male offenders, black offenders received significantly longer sentences than white offenders. The results, in other words, lend only partial support for Hypothesis 3 (because black males were given longer sentences than white males, but Hispanic males were not) and fail to support Hypothesis 4.

Table 3 Results of OLS regression analysis of logged sentence length: data partitioned by offender’s sex

The other variables had, with some exceptions, similar effects on sentence length for male and female offenders. The presumptive sentence and both regular and substantial assistance departures had similar effects, as did the district in which the case was adjudicated. Both male and female offenders received longer sentences if they were facing longer presumptive sentences; they received shorter sentences if they received either type of downward departure or if they were sentenced in Minnesota or Nebraska rather than in Southern Iowa. The offender’s employment status, whether the offender was in custody at the time of sentencing, and the type of drug for which the offender was convicted, on the other hand, only affected the sentences imposed on male offenders.

The results of our analyses of the data partitioned by the offender’s race/ethnicity are shown in Table 4. Of particular interest is that finding that whereas the sex of the offender affected the sentences imposed on black and Hispanic offenders, it had no effect on the sentences imposed on white offenders. To elaborate, black and Hispanic female offenders received shorter sentences than similarly situated black and Hispanic male offenders, but there were no differences in the sentences imposed on white females versus white males.

Table 4 Results of OLS regression analysis of logged sentence length: data partitioned by offender’s race/ethnicity

Table 4 further shows that, consistent with the results of the analysis of the data partitioned by the sex of the offender, the presumptive sentence and both types of downward departures had significant effects on sentence length irrespective of the race/ethnicity of the offender, as did the jurisdiction in which the case was adjudicated. Variables that had inconsistent effects were the offender’s employment status (being unemployed affected sentences only for black offenders), whether the offender had dependent children (which affected sentences only for Hispanic offenders), the type of drug involved in the offense (white offenders convicted of powder cocaine or methamphetamine offenses got longer sentences than whites convicted of other drug offenses), and whether the offender was in custody prior to the sentencing hearing (which influenced the sentences of white and Hispanic offenders but not black offenders).

The analytical results discussed thus far reveal that the race (but not the ethnicity) of the offender affected sentence severity only for male offenders and that the sex of the offender affected the length of the sentence imposed on black and Hispanic, but not on white, offenders. This suggests that the effect of the offender’s race/ethnicity is conditioned by the offender’s sex and that the effect of the offender’s sex is conditioned by the offender’s race/ethnicity. Support for this conclusion is found in Table 5, which presents the results of the analysis with the race/ethnicity-by-sex dummy variables.

Table 5 Results of OLS regression analysis of logged sentence length: using race/ethnicity × sex dummy variables

To test our hypothesis that black male drug offenders would receive longer sentences than all other offenders, we ran the analysis with black males as the reference category. We found that the sentences imposed on white males, black females, white females, and Hispanic females were significantly shorter than those imposed on black males. Stated another way, all of the drug offenders adjudicated in these three district courts, with the exception of Hispanic males, received significantly shorter sentences than did black males. To test our hypothesis that white females would receive shorter sentences than other offenders, we ran the analysis with white females as the reference category. We found significant differences only for black males and Hispanic males. Contrary to expectations, there were no differences in the sentences imposed on white females and either black or Hispanic females; there were also no differences in prison sentence length for white females versus white males.

These similarities and differences are illustrated more clearly in Fig. 1. The data presented in Fig. 1 are based on the results of a sentence length model using the unlogged measure of sentence length¹²; we used these results to calculate the adjusted mean sentences for each type of offender.¹³ As shown in Fig. 1, the adjusted mean sentence for black male offenders (93.1 months) is more than 2 years longer than the adjusted mean sentence for black female offenders (66.5 months). In contrast, mean sentences for Hispanic male and Hispanic female offenders vary by only 10.6 months, and the sentences for white female offenders are only 5.3 months shorter than the sentences for white male offenders (a difference which is not statistically significant). There also are very small (and non-significant) differences between the mean sentences imposed on black and Hispanic male offenders and on white and Hispanic female offenders.

Fig. 1

Adjusted mean sentences

Discussion and Conclusion

Many researchers have examined the additive effects of offender race/ethnicity and sex on sentencing outcomes (for reviews of the literature on the effect of race/ethnicity see Mitchell 2005 and Spohn 2000; for reviews of the effect of sex see Daly and Bordt 1995 and Brennan 2002). Few to date, however, have examined how these two highly visible characteristics might operate jointly to influence the types of punishments that offenders receive. An examination of the joint effects of these variables is important because it allows one to determine whether all women are likely to be treated more leniently than all men and whether all minority offenders are likely to be given harsher punishments than all white offenders.

The purpose of this study, therefore, was to explore the additive and interactive effects of the offender’s race/ethnicity and sex on the length of the prison sentences imposed on drug offenders by judges in three US District Courts. We hypothesized that black and Hispanic drug offenders would receive harsher sentences than white drug offenders because, consistent with theoretical discussions of uncertainty avoidance/causal attribution and focal concerns, negative racial and ethnic stereotypes allow court actors to assume that minority offenders are dangerous, culpable, and likely to commit future crime. We also predicted that female drug offenders would receive more lenient sentences than male offenders because females do not fit prevailing stereotypes of offenders. We further hypothesized that the offender’s race/ethnicity would affect sentences for both male and female offenders, that the offender’s sex would influence sentences for white, black, and Hispanic offenders, and that sentences would be particularly harsh for black males and particularly lenient for white females. These latter predictions are based on findings from media and sentencing studies. Recall that researchers who study media depictions of offenders suggest that the most common image of a criminal is that of a young, black (or other minority) male (Barak 1994; Barlow 1998; Chermak 1994; Humphries 1981; Madriz 1997). Sentencing scholars similarly argue that the brunt of negative stereotyping falls most heavily on young, unemployed, black males (see for example Spohn and Holleran 2000; Steffensmeier et al. 1998). At the same time, however, a review of the extant literature also revealed that minority women have been stereotyped more negatively than white women. The media reinforce the stereotype that minority women defy gender-role expectations, which makes it more likely that minority women will be held accountable for their criminal actions, even when they are accused of similar offenses (Brennan and Vandenberg 2009; Bond-Maupin 1998; Farr 1997, 2000; Huckerby 2003).

As shown in Table 6, the results of our study provide more support for our hypotheses regarding the effect of the offender’s sex than for our hypotheses regarding the offender’s race/ethnicity. Consistent with our expectations, female offenders received shorter sentences than similarly situated male offenders (Hypothesis 2), black females received shorter sentences than black males (Hypothesis 6), and Hispanic females received shorter sentences than Hispanic males (Hypothesis 7); there were, however, no differences in the sentences imposed on white female and male offenders (Hypothesis 5).

Table 6 Summary of hypothesis tests

Our results suggest that findings of leniency at the sentencing stage do not bypass minority women. This conclusion meshes with findings from previous research. Specifically, other researchers have found that black females receive more lenient sentences than black males (Albonetti 1997; Gruhl et al. 1984; Spohn et al. 1985; Steen et al. 2005; Steffensmeier and Demuth 2006) and that Hispanic females receive less severe sentences than Hispanic males (Gruhl et al. 1984; Steffensmeier and Demuth 2006). However, our finding of no difference in the length of sentence imposed on white females versus white males is at odds with other research findings (Albonetti 1997; Steen et al. 2005; Steffensmeier and Demuth 2006; but see Zatz 1984) and suggests the need for further analyses of the effect of sex for offenders of varying races/ethnicities.

With regard to the effect of race/ethnicity, contrary to our predictions, there were no differences in the sentences imposed on black, Hispanic, and white offenders (Hypothesis 1), no differences in the sentences imposed on Hispanic male offenders and white male offenders (Hypothesis 3), and no differences in the sentences imposed on black, Hispanic, and white female offenders (Hypothesis 4). In fact, the only result consistent with our initial hypotheses about the effect of the offender’s race/ethnicity was the finding that black male offenders received significantly longer sentences than did white male offenders (Hypothesis 3).

These findings are somewhat consistent with other studies that focus on the issue of racial and/or ethnic disparities in the sentencing of drug offenders. While some find that black drug offenders are sentenced more severely than whites (Albonetti 1997; Brennan and Spohn 2008; Demuth and Steffensmeier 2004; Rodriguez et al. 2006; Steen et al. 2005), others find that sentencing outcomes do not differ between similarly situated black and white drug offenders (Chiricos and Bales 1991; Curry and Corral-Camacho 2008; Engen and Steen 2000; Spohn 1999). In addition, our finding of no differences in sentence length between similarly situated white and Hispanic offenders is consistent with some of the extant research (Curry and Corral-Camacho 2008; Lagan 1996; McDonald and Carlson 1993; Spohn 1999), although it is at odds with findings reported in other studies (see for example, Albonetti 1997; Brennan and Spohn 2008; Demuth and Steffensmeier 2004; Rodriguez et al. 2006; Steffensmeier and Demuth 2000, 2001). Overall, results from the extant literature are far from conclusive. While the results reported in the current study do not clarify the issue, they do provide a direction for future inquiries. Specifically, we recommend that future researchers examine the joint effects of offender sex and race/ethnicity before concluding that either variable influences the sentencing outcome. Most studies of drug offender sentencing have, instead, focused solely on the additive effects of these variables.

With regard to a specific sex-by-race/ethnicity interaction effect, we found partial support for our expectation that black males would receive longer sentences than all other offenders (Hypothesis 8). The sentences imposed on black males were significantly longer than those imposed on white males, black females, white females, and Hispanic females; there were no differences in the sentences imposed on black males and Hispanic males. Others have also found that black men are treated more punitively than females of varying races/ethnicities (Hartley et al. 2007; Spohn and Beichner 2000; Spohn and Spears 1997) and white males (Crew 1991; Hartley et al. 2007; Kruttschnitt 1984; Spohn and Beichner 2000; Spohn and Holleran 2000; Steffensmeier and Demuth 2006; Steffensmeier et al. 1993, 1998), but no differently from Hispanic males (Hartley et al. 2007).

Our hypothesis that white female offenders would receive shorter sentences than all other offenders, on the other hand, was not confirmed (Hypothesis 9). While white females received shorter sentences relative to both black and Hispanic males, there were no differences in the sentences that judges imposed on white females and white males, black females, or Hispanic females. Other researchers have also found that race/ethnicity does not make a difference among women (Bickle and Peterson 1991; Crew 1991; Farnworth and Teske 1995; Kruttschnitt 1984; Spohn and Beichner 2000; Steffensmeier and Demuth 2006). In short, our findings find support for the assertion that black and Hispanic males, but not black and Hispanic females, fit prevailing stereotypes of dangerous and culpable offenders.

The results of the current study illustrate the importance of testing for the additive and interaction effects of the race/ethnicity and the sex of the offender. When we ran the analysis using the full sample, we found that the length of the prison sentence was affected by the offender’s sex but not by the offender’s race/ethnicity. These results, while valid for the full sample, were misleading. The sex of the offender affected sentence severity only for black and Hispanic offenders, and the race of the offender affected sentence length for male, but not for female, offenders. In other words, the effect of the offender’s sex was conditioned by the offender’s race, and the effect of the offender’s race was conditioned by the offender’s sex. We urge future researchers to continue to explore the combined effects of offender race/ethnicity and sex on sentencing outcomes, for both drug and non-drug offenders.

We especially encourage future researchers to examine the influence of both variables for non-drug offenders, because the findings may be different. Although an examination of drug offenders was warranted in the current study, our sole focus on drug offenders limits the generalizability of our findings to other offenders. Other researchers have found that the effects of race/ethnicity are greater in drug than in non-drug cases (see for example, Engen et al. 2003; Mitchell 2005; Steffensmeier and Demuth 2000). Another limitation of this study is its focus on offenders convicted in three relatively homogeneous US District Courts located in the Midwest. Although there is a growing consensus among researchers that the district courts vary on a number of important dimensions and that it therefore is inappropriate to aggregate data across all district courts (Johnson et al. 2008; Kautt 2002; LaCasse and Payne 1999; Nagel and Schulhofer 1992; Spohn 2005), the patterns uncovered by this study may not apply to other district courts, especially those with different drug caseloads or different proportions of offenders who are racial minorities or women. For these reasons, we encourage replication of our study with data from other courts.

Appendix 1

See Table 7.

Table 7 Correlation matrix