Gender and Sentencing in the Federal Courts: Are Women Treated More Leniently?

Abstract

Using data from the United States Sentencing Commission (2001-2003), we examine the role of gender in the sentencing of defendants in federal courts. We address two questions: First, can we explain the gender gap in sentencing by taking into account differences in legal and extralegal factors? And second, do legal and extralegal factors have the same impact for male and female defendants? Overall, we find that female defendants receive more lenient sentence outcomes than their male counterparts. Legal factors account for a large portion of the gender differences, but even after controlling for legal characteristics a substantial gap in sentencing outcomes remains. Also, despite their influence on sentencing outcomes in some instances, extralegal characteristics do not help to close the gender gap. Finally, when male and female defendants are examined separately, we find that not all legal and extralegal factors weigh equally for male and female defendants.

Full text

Criminal Justice Policy Review
XX(X) 1 –28
© 2012 SAGE Publications
Reprints and permission:
sagepub.com/journalsPermissions.nav
DOI: 10.1177/0887403412466877
http://cjp.sagepub.com
466877CJPXXX10.1177/0887403412466877Crimi
nal Justice Policy ReviewDoerner and Demuth
1University of Rhode Island, Kingston, RI, USA
2Bowling Green State University, Bowling Green, OH, USA
Corresponding Author:
Jill K. Doerner, Department of Sociology & Anthropology, University of Rhode Island, 505 Chafee Social
Science Center, 10 Chafee Road, Kingston, RI 02881, USA
Email: jdoerner@uri.edu
Gender and Sentencing
in the Federal Courts:
Are Women Treated
More Leniently?
Jill K. Doerner1 and Stephen Demuth2
Abstract
Using data from the United States Sentencing Commission (2001-2003), we examine
the role of gender in the sentencing of defendants in federal courts. We address two
questions: First, can we explain the gender gap in sentencing by taking into account
differences in legal and extralegal factors? And second, do legal and extralegal factors
have the same impact for male and female defendants? Overall, we find that female
defendants receive more lenient sentence outcomes than their male counterparts.
Legal factors account for a large portion of the gender differences, but even after
controlling for legal characteristics a substantial gap in sentencing outcomes remains.
Also, despite their influence on sentencing outcomes in some instances, extralegal
characteristics do not help to close the gender gap. Finally, when male and female
defendants are examined separately, we find that not all legal and extralegal factors
weigh equally for male and female defendants.
Keywords
sentencing outcomes, gender differences, sentencing guidelines
Federal sentencing guidelines are designed to encourage the uniform and proportional
treatment of defendants based on legally relevant factors. A main goal of the guide-
lines is to produce fair and honest outcomes that minimize unwarranted disparities
at UNIV OF RHODE ISLAND LIBRARY on July 19, 2013cjp.sagepub.comDownloaded from

2 Criminal Justice Policy Review XX(X)
based on defendants’ social characteristics. A large body of disparity research has
developed over time and, not surprisingly given America’s sordid racial history, the
overwhelming majority of studies focus on racial and ethnic differences in sentencing
outcomes (Demuth, 2002; Demuth & Steffensmeier, 2004; Spohn, 2000; Steffens-
meier & Demuth, 2006). What has not been a strong focus of past research is an argu-
ably more common, yet apparently less controversial, form of disparity based on
gender.
Like a defendant’s race, gender is considered to be an extralegal factor in decision
making at the sentencing stage. However, there are at least three factors that might
explain both the persistence of gender disparities in sentencing despite guidelines
designed to curtail them and a diminished concern for studying and remedying these
disparate outcomes. First, unlike claims of racism in the application of laws and sanc-
tions, there is no general presumption that women, the disadvantaged minority group,
have historically been subjected to a consistent pattern of discrimination resulting in
unwarranted harsher punishments (Nagel & Hagan, 1983). Second, in the context of
societal and court concerns about crime and public safety and given the known greater
propensity for crime among men, women are viewed as better recidivism risks and
more deserving of leniency than men (Spohn, 2002). Third, a major difference in the
social lives of men and women is the level of responsibility in caring for family, or
more specifically for their dependent children (Bickle & Peterson, 1991; Daly, 1987a;
Daly, 1987b; Daly, 1989). This practical consideration might make the court reluctant
to sentence women as harshly as men.
In sum, there is a tension in the guidelines between the goal of a gender-neutral
implementation of the law emphasizing uniform treatment based on offense severity
and criminal history and the realization that important differences exist between the
lives of men and women that might create a need or desire for differential treatment
(for a similar argument about race, see Tonry, 1996). In fact, the guidelines recognize
this dilemma and provide limited ways for judges and prosecutors to take gender into
account. For example, the guidelines allow for some discretion through the use of
departures, which enable factors such as family ties and responsibilities to be consid-
ered. But, overall, unexplained gender disparities persist despite policy changes
designed to minimize them. This suggests that reformers may have had unrealistic
expectations about the ability of guidelines to structure outcomes as intended
(Spohn, 2000).
For all these reasons, an underdeveloped body of scholarship exists that addresses
the topic of gender differences in sentencing. Much of this research is dated, having
been published in the 1970s and 1980s using smaller state data sets or single city
samples. Another shortcoming of many past studies is a lack of robust controls for
legal case characteristics such as offense seriousness and criminal history. Most
importantly, past research tends to examine only whether sex differences exist at the
sentencing stage and typically does not explore empirically how gender influences
outcomes (for a review see Chiricos & Crawford, 1995; Daly & Bordt, 1995).
Researchers who examine gender and court processing tend to treat gender as a fixed
at UNIV OF RHODE ISLAND LIBRARY on July 19, 2013cjp.sagepub.comDownloaded from

Doerner and Demuth 3
attribute of individuals; however work by Daly (1986, 1989, and 1994) and Kruttschnitt
(1984) explores how gender and patterned roles associated with gender can influence
court decisions. That male defendants tend to commit more serious offenses and have
more extensive criminal records than female defendants helps to explain why men
tend to receive harsher sentences than women (Bickle & Peterson, 1991; Daly, 1989;
Daly, 1994; Daly & Bordt, 1995; Doerner & Demuth, 2010; Spohn, 2000, 2002;
Steffensmeier, Kramer, & Streifel, 1993: Steffensmeier, Kramer, & Ulmer, 1995,
1998). But, legal factors alone do not appear to fully explain the gender gap and few
studies have attempted to account for the remaining, often sizeable, differences in
sentence outcomes between men and women.
In the present study, we use data from the Unites States Sentencing Commission
(USSC) to more fully explore the gender gap in federal sentencing and examine the
various ways in which gender continues to influence outcomes even within a system
of formal rules designed to minimize the impact of extralegal factors. We contribute
to the gender and sentencing literature in several important ways. First, we use data
that have rich and detailed measures of legal case characteristics. A concern in prior
research was that weak or incomplete measures of offense seriousness and criminal
history failed to adequately capture the real differences in offending between men and
women and made the gender gap in sentencing outcomes look larger than it actually
was. With more robust measures, we reduce the likelihood of finding a gender gap that
is simply an artifact of model misspecification.
Second, we examine a series of nested regression models to determine not just if a
gender gap in incarceration and sentence length outcomes exists, but why. We begin
by looking at the gender gap before accounting for differences in legal characteristics
between men and women. Next, we control for legal differences to see how much
gender differences in sentencing are explained by legal factors. Lastly, and most
importantly, we examine the gender gap after adding controls for extralegal character-
istics that are associated both with gender and sentencing outcomes: Education, mari-
tal status, and the number of dependents for which the defendant is responsible. Much
prior research tends to add all legal and extralegal variables to the model at the same
time making it difficult to compare the gender gap before and after controlling for
legal factors. And, central to our earlier criticism of existing gender-sentencing work
is that most prior studies examine gender as a fixed attribute and do not attempt to
address what aspects of gender influence sentencing. Building upon the research of
Daly (1986, 1989, and 1994) and Kruttschnitt (1984), we explore several gender-
related possibilities.
Third, in addition to examining the main effect of gender, we examine whether
legal and extralegal factors have different effects on sentencing outcomes for men and
women. Most prior studies focus on the main effect of gender and do not consider the
possibility that sentencing could be a gendered process. Prior research by Daly (1987a,
1987b, 1989) has shown that court personnel assumed gender divisions in the work
and family responsibilities of familied defendants, and this resulted in differential out-
comes during the sanctioning process. These court officials also viewed caretaking
at UNIV OF RHODE ISLAND LIBRARY on July 19, 2013cjp.sagepub.comDownloaded from

4 Criminal Justice Policy Review XX(X)
labor, most often provided by women, as more difficult to replace. In addition, Koons-
Witt (2002) found that the interaction between gender and number of dependents was
a significant predictor of incarceration decisions, with women with dependent children
significantly more likely to be sentenced to sanctions within the community. Thus, we
explore whether there are differences between men and women with respect to legal
and extralegal factors.
Gender and Sentencing Literature
The treatment of women in the courts has not been static in the United States (Farrell,
2004). Historically, female offenders were less likely to be arrested and often sen-
tenced more leniently than similarly situated male offenders. However, such judicial
discretion has often been a double-edged sword for women. Rafter (1990) and col-
leagues (with Stanko, 1982) documented a dual system of punishment for female
offenders during the middle of the 19th century. Women deemed “feminine” or “train-
able” by the court were most often sent to reformatories, while women viewed as
“bad” or “masculine” were subject to incarceration in penal institutions, often along-
side male prisoners (Butler, 1997). Gendered sentencing laws at the turn of the 20th
century still allowed judges to send women to prison for minor public order offenses
(e.g., alcohol-related offenses, DUI, and prostitution) for which men were rarely even
arrested (Rafter, 1990; Temin, 1980). Indeed, until the 1970s, state sentencing laws
allowed judges to sentence women differently than men because female offenders
were perceived to be more amenable to rehabilitation and would benefit from longer
indeterminate sentences (Pollock-Byrne, 1990).
Currently, a fairly persistent finding in the sentencing literature is that female
defendants are treated more leniently than male defendants (Bickle & Peterson, 1991;
Daly & Bordt, 1995; Doerner & Demuth, 2010; Griffin & Wooldredge, 2006; Spohn,
2002; Steffensmeier et al., 1993); however, one study reported no gender differences
(Kruttschnitt & Green, 1984). Doerner and Demuth (2010) showed that female defen-
dants were significantly less likely to receive incarceration sentences than male defen-
dants. The odds of incarceration for female defendants were approximately 42% lower
than the odds of incarceration for male defendants. Griffin and Wooldredge (2006)
found that female defendants in general were less likely than men to be sent to prison
both before and after the sentencing reform efforts in Ohio and that the magnitude of
this effect did not change significantly over time (.51 to .43 for men, and .38 to .34 for
women). Spohn (2002) reported that the odds of receiving a prison sentence were 2.5
times greater for male offenders than for female offenders after controlling for legally
relevant factors. Steffensmeier and Motivans (2000) found that female defendants
were sentenced less harshly than male defendants—on average they were about 14%
less likely to be incarcerated and received prison sentences about 7 months shorter.
Similarly, previous research by Steffensmeier et al. (1993) indicated that gender, net
of other factors, had a small effect on the likelihood of imprisonment, with female
defendants less likely to receive an incarcerative sentence than male defendants. But,
they found that gender had a negligible effect on sentence length outcomes.
at UNIV OF RHODE ISLAND LIBRARY on July 19, 2013cjp.sagepub.comDownloaded from

Doerner and Demuth 5
According to Gruhl, Welch, and Spohn (1984) female defendants were treated
more leniently than male defendants, based on a simple breakdown with no controls.
Even though they plead guilty and were convicted at about the same rates as males,
females were more likely to have their cases dismissed and were less likely to be incar-
cerated. When the authors controlled for legal and extralegal factors, significant gen-
der differences remained for dismissal and incarceration, even though the difference
between males and females was reduced somewhat.
In terms of gender, women are thought to be less dangerous, less blameworthy, less
likely to recidivate, and more likely to be deterred than men (Spohn, 2002). Therefore,
the more lenient sentences that are imposed on them might reflect the fact that judges
believe them to possess these qualities more than men. According to Belknap (2001),
studies consistently show that females generally commit fewer crimes than males but
also tend to commit offenses that are less serious and violent in nature. However, net
of case severity, charge severity, type of offense, prior record, and other defendant
characteristics, male and female defendants were still treated differently on the basis
of their ties to and responsibilities for others. Kruttschnitt (1984) found that control-
ling for gender-related statuses (i.e., being a wife or mother) mediated the length of
probation sentences. In addition, she concluded that women were more likely than
men to remain free, both prior to adjudication and after conviction, and that the deter-
minants of these two decisions varied significantly with the offender’s gender.
Therefore, her analysis provided some insight into why females receive preferential
treatment by criminal courtroom personnel.
Familial Responsibility Literature
It has long been observed that female defendants who are married or who have chil-
dren receive greater leniency from the courts than their male or unmarried and child-
less female counterparts (Bickle & Peterson, 1991; Daly, 1987a, 1987b, 1989; Eaton,
1987; Farrington & Morris, 1983; Kruttschnitt & Green, 1984; Kruttschnitt &
McCarthy, 1985; Simon, 1975). Early explanations of how and why gender-based
family roles were important in judicial decision making focused on the impracticality
of harsh sanctions for female offenders compared to their male counterparts
(Bernstein, Cardascia, & Ross, 1979; Simon, 1975). More specifically, Simon (1975)
reported that officials’ accounts of gender differentials in sentencing in both New
York (1963-1971) and California (1945-1972) emphasized that women have families,
both husbands and children, to care for and sending women to prison would seriously
disrupt the family unit.
Kruttschnitt (1982a, 1984), along with her colleagues (with Green, 1984; with
McCarthy, 1985) examined gender differentials in sanctioning, specifically pretrial
release and sentencing outcomes, using data from Minnesota. In addition to gender,
these analyses included either a composite measure of informal social control, or one
or more sex-based family role factors including family/household composition, num-
ber of children, employment status, and sources of support. Overall, the findings from
this research indicated that gender-based disparities were affected but not eliminated
at UNIV OF RHODE ISLAND LIBRARY on July 19, 2013cjp.sagepub.comDownloaded from

6 Criminal Justice Policy Review XX(X)
by including family role factors, and that when composite measures of informal con-
trol were considered, there was little support for the claim that familial social control
was a sex-specific determinant of criminal sanctioning.
In a more recent study of imprisonment decisions in Minnesota, Koons-Witt (2002)
found that gender alone did not have a significant influence on sentence outcomes
prior to the use of sentencing guidelines, but results indicated a significant interaction
between gender and the presence of dependent children. The presence of dependent
children for women significantly reduced their likelihood of going to prison. She also
found that the interaction between gender and number of dependent children was a
significant predictor of the incarceration decisions after sentencing guidelines were
enacted. In this instance, women with dependent children were significantly more
likely to be sentenced to a community sanction than were women without dependent
children.
In her 1989 study, Daly found that a defendant’s work–family relations affected the
sentencing of both men and women. Furthermore, she reported that what defendants
did for families, in terms of providing economic support or care for dependents, mat-
tered to judges. Familied men and women (those with dependent children) were less
likely to be detained pretrial, and they were less likely to receive the harsher types of
nonjail sentences than childless men and women. In addition, the mitigating effect of
being familied was stronger for women than men (Daly, 1987a). Furthermore, having
dependents, whether in a marital context or not, was generally the more determining
feature of whether defendants receive lenient treatment. For men, being married with-
out dependent children conferred no advantage at the pretrial release or the two sen-
tencing decisions; but having dependent children did. Married women, and especially
those with dependent children, were accorded greater leniency at the pretrial release
decision. In addition, at the sentencing stage, women with dependents received the
most lenient sentences.
What appears to matter most for court personnel is whether defendants have day-
to-day responsibilities for the welfare of others; such care or economic support can
occur with or without a marital tie, and the specific form of care and economic support
can vary by gender. In addition, the greater leniency accorded familied women than
familied men stems from contemporary gender divisions in work and family life, spe-
cifically that women are more likely to care for others. The mitigating effects of family
were found in both the pretrial release and nonjail sentencing decisions. Thus, familied
defendants may be accorded leniency even when decisions do not center on a defen-
dant’s loss of liberty (Daly, 1987a).
Daly (1987b) found that court officials consistently drew on the categories of work
and family in explaining why some defendants deserved leniency. One theme present
was that defendants who provide economic support or care for others deserve more
lenient treatment than those without such responsibilities. Leniency toward the fami-
lied defendants was therefore justified on the grounds that these defendants were more
stable and have more to lose by getting into trouble again. Court personnel assume
gender divisions in the work and family responsibilities of familied men and women.
at UNIV OF RHODE ISLAND LIBRARY on July 19, 2013cjp.sagepub.comDownloaded from

Doerner and Demuth 7
These differences, combined with the family profiles of defendants, foster discrepan-
cies in the treatment of familied men and familied women. In addition, officials often
justified treating familied defendants more leniently because of the social costs of
removing them and jeopardizing the family unit. Sex differentials in outcomes stem
from the perceived differential responsibilities of females versus males. Officials
viewed it as more costly or impractical to jail women with families than men with
families because breadwinning support, usually provided by males, was more readily
replaced than caretaking labor (Daly, 1987b).
Overall, research has shown that legal factors play a large role in the sentencing
outcomes of male and female defendants, but even after controlling for characteristics
like criminal history and offense severity, unexplained differences still persist. As a
result, our understanding of why women are sentenced more leniently than men
remains limited. In addition, research on familial responsibility indicates that having
dependents (more specifically, dependent children) creates leniency at sentencing,
especially for women. The present study sets out to explore how legal and extralegal
factors play a role in the sentencing of male and female defendants, using data from
the United States Sentencing Commission. We pay particular attention to whether
characteristics such as education, marital status and the presence of dependents help to
explain the remaining gap in sentencing outcomes, as previous research in this area
has discovered, after controlling for legally relevant variables outlined under the
Federal Sentencing Guidelines.
Theoretical Framework and Research Expectations
As previous research has shown, sentencing outcomes continue to be influenced by a
host of extralegal factors, even with sentencing guidelines in place (Doerner &
Demuth, 2010; Steffensmeier & Demuth, 2000; Steffensmeier et al., 1998; Ulmer,
1995). The focal concerns perspective developed by Steffensmeier (1980) serves as a
framework for understanding why extralegal factors such as gender, race/ethnicity,
and age might influence sentencing decisions, despite the implementation of formal
guideline systems. The theory outlines three focal concerns that are important to
judges and other criminal justice actors in reaching sentencing decisions: blamewor-
thiness, protection of the community, and practical constraints and consequences.
Grounded in research on organizational decision making, inequality and stratification,
and criminal stereotyping, Steffensmeier and colleagues (with Kramer & Streifel,
1993; with Kramer & Ulmer, 1998) argue that defendant status characteristics may
influence sentencing decisions insofar as stereotypes and behavioral expectations
linked to these characteristics relate to the focal concerns of legal agents.
Blameworthiness follows the principle that sentences should depend on the offend-
er’s culpability and the degree of injury caused. The primary factors influencing per-
ceptions of blameworthiness are legal factors such as the seriousness of the offense,
the defendant’s criminal history or prior victimization at the hands of others, and the
defendant’s role in the offense (Steffensmeier et al., 1998). Albonetti (1997) suggests
at UNIV OF RHODE ISLAND LIBRARY on July 19, 2013cjp.sagepub.comDownloaded from

8 Criminal Justice Policy Review XX(X)
that court officials attempt to achieve rational outcomes in the face of incomplete
knowledge by relying on stereotypes that differentially link defendant groups to recid-
ivism. Research by Daly (1994) indicates that judges, at least to some extent, share
common beliefs portrayed by the media and are influenced by them in their sentencing
decisions. In other words, when decisions have to be made quickly, judicial profes-
sionals may rely on limited resources to reach an outcome in the time available.
Protection of the community typically focuses on the need to incapacitate the
offender or to deter future crime. Albonetti (1991) argues that sentencing is an arena
of bounded rationality, in which court actors, particularly judges, confront the goal of
protecting the public and preventing recidivism in the context of high uncertainty
about offenders’ future behavior. Judges’ assessments of offenders’ future behavior is
often based on attributions predicated primarily on the nature of the offense and the
offender’s criminal history. However, these decisions may also be influenced by extra-
legal characteristics of the offender such as gender, race/ethnicity, age, and socioeco-
nomic status (SES). As mentioned previously, criminal justice professionals may give
in to stereotypical notions as a means of making decisions more quickly, espe-
cially in the face of pressure from the media, victim’s families, and members of the
community.
Practical constraints and consequences relate to how sentencing decisions impact
the functioning of the criminal justice system as well as the circumstances of individual
defendants, their families and communities. Organizational concerns include main-
taining working relationships among courtroom actors, ensuring the stable flow of
cases, and being sensitive to local and state correctional crowding and resources
(Dixon, 1995; Flemming, Nardulli, & Eisenstein, 1992; Steffensmeier et al., 1993,
1998; Ulmer, 1995; Ulmer & Kramer, 1996). Individual concerns include the offender’s
ability to do time, health conditions, special needs, the cost to the correctional system,
and disruption to children and family (Daly, 1987a; Hogarth, 1971; Steffensmeier,
1980; Steffensmeier et al., 1995).
Expectations
Guided by the focal concerns perspective and the findings of past research on the
effect of gender on sentencing outcomes, we develop several hypotheses for the pres-
ent study to answer two research questions. First, can the gender gap in sentencing be
explained by accounting for differences in legal and extralegal factors? Second, do
legal and extralegal factors have the same impact for male and female defendants?
Drawing on prior research, we expect to find that, on average, female defendants will
receive more lenient sentences than male defendants (H1), and that this finding will
hold true even after controlling for relevant legal and contextual factors (H2). In addi-
tion, we expect that defendants that have more education, more marital stability, and
dependents will be afforded greater leniency than defendants that have less education,
are single, or have no dependents (H3). Furthermore, we hypothesize that legal and
extralegal factors will exert similar effects on sentencing outcomes for both male and
female defendants (H4).
at UNIV OF RHODE ISLAND LIBRARY on July 19, 2013cjp.sagepub.comDownloaded from

Doerner and Demuth 9
Data and Method
In the present study, we use data from three years (2001-2003) of the Monitoring of
Federal Criminal Sentences program compiled by the USSC. The data include all
cases received by the USSC that had sentencing dates between October 1, 2000, and
September 30, 2003 and were assessed as constitutional (total = 194,521 cases). Data
from the three years were combined to create one large data set, thus providing larger
case sizes for both male and female defendant groups. These data are especially
appropriate as they contain some of the richest and most detailed information avail-
able on cases at the sentencing stage. Many of the single-city or state-level data sets
used in prior studies have lacked the large number of legal control variables found in
the federal guidelines data. Having these variables available enabled a more adequate
elimination of alternative explanations for extralegal effects on sentencing outcomes
(e.g., Demuth & Steffensmeier, 2004; Spohn & Holleran, 2000). Furthermore, the
federal sentencing guidelines provide a more rigid and conservative test of the impact
of extralegal factors on sentencing outcomes.
For this analysis, we eliminate several defendant groups from the sample. First,
noncitizens are deleted from the analysis. Federal sentencing of noncitizen defendants
often differs greatly from sentencing of citizen defendants in many ways and, as a
result, makes comparisons of sentencing outcomes between them difficult (Demuth,
2002). For instance, a large proportion of noncitizen cases involve immigration viola-
tions. Furthermore, because noncitizens can be deported, the sentencing process for
noncitizens is often qualitatively different (the goal being to send the defendant back
to his/her country of origin and not to punish) from that of U.S. citizens. Finally, case
information provided for noncitizens may be incomplete and this will most likely
result in an underestimation of prior criminal history.
Second, defendants under the age of 18 are excluded from the analysis because
their cases are substantively and legally different due to their juvenile status. Third,
defendants who receive upward departures are deleted from the analysis as they com-
prised only 0.8% of departure cases and made comparisons across departure type very
difficult. Fourth, using listwise deletion, all cases with missing information for all
variable used in the analysis are deleted. Analyses were run predeletion and postdele-
tion of missing information and the elimination of these cases did not significantly
change the overall results. The final analytic sample for the present study is 109,181.
Dependent Variables
The sentencing outcome is the result of a two-stage decision making process: The
decision to incarcerate and, once incarceration is selected, the sentence length deci-
sion (for discussion, see Spohn, 2002). In the present study, we use logistic regression
to model the incarceration decision. The in/out decision variable is coded dichoto-
mously, with 1 indicating a prison sentence and 0 indicating a nonincarceration sen-
tence (e.g., probation, community service). The sentence length decision is modeled
using ordinary least squares (OLS) regression and includes only those defendants who
at UNIV OF RHODE ISLAND LIBRARY on July 19, 2013cjp.sagepub.comDownloaded from

10 Criminal Justice Policy Review XX(X)
receive a prison sentence. Sentence length is a continuous variable representing the
logged length of the prison sentence in months. Logging sentence length helps to
normalize the distribution and taking the antilog of the coefficient in the logged sen-
tence length model provides a useful proportional interpretation. Sentence length is
capped at 470 months. Any sentence length beyond that duration is considered to be
life in prison.1
Extralegal Variables
Defendant gender is a dummy variable coded 1 if the defendant is female and 0 if the
defendant is male. Race/ethnicity is coded as four dummy variables: White non-
Hispanic, Black non-Hispanic, Hispanic of any race, and Other.2 Defendant age is a
continuous variable representing the age of the defendant at the time of sentencing and
ranges from 18 to 100. In this case, defendant age has been grouped in logical ranges
consistent with Steffensmeier et al. (1998) and is coded as a series of dummy vari-
ables (18 to 20, 21 to 29, 30 to 39, 40 to 49, 50 to 59, and 60 and over).
Education level is coded as three dummy variables: Less than high school, high
school, and more than high school, with those who graduated high school as the refer-
ence category. Marital status is coded as six dummy variables: Single, married, cohab-
iting, divorced, widowed, and separated. Those defendants who are single serve as the
reference category. Number of dependents3 is a continuous variable indicating respon-
sibility of support by the defendant of their dependents. For the purposes of this study,
number of dependents has been recoded into a dichotomous variable indicating that
defendants either have no dependents or have one or more dependents.4 Many studies
have shown that female defendants that are married or have dependents receive greater
leniency from the courts than their male or unmarried and childless female counter-
parts (Bickle & Peterson, 1991; Daly, 1987a, 1987b, 1989; Eaton, 1987; Farrington &
Morris, 1983; Kruttschnitt & Green, 1984; Kruttschnitt & McCarthy, 1985; Simon,
1975). Having dependents, whether in a marital context or not, is generally the more
determining feature of whether defendants receive lenient treatment. However, while
the majority of prior research uses the terms “child or children,” the present study uses
“dependent” as the data do not specify what type of dependent the defendant is respon-
sible for.
Legal Variables
Under the Federal Guidelines, federal judges retain discretion for sentencing indi-
viduals within the range determined by the offense level and criminal history of the
offender. Sentence ranges are determined using a grid that takes these two variables
into account, one on each axis. However, it has been argued (see Engen & Gainey,
2000) that a variable representing the presumptive guideline sentence, where criminal
history and offense severity are combined into a single measure, is a more appropriate
strategy and actually explains more of the variation in sentencing outcomes. This
at UNIV OF RHODE ISLAND LIBRARY on July 19, 2013cjp.sagepub.comDownloaded from

Doerner and Demuth 11
analytic strategy is also used by the USSC (2004). Therefore, we include a variable
representing the guideline minimum sentence, in months. We also include a measure
of criminal history, which ranges from 1 to 6 and indicates the final criminal history
category of the defendant, as assigned by the court. According to Ulmer (2000), mea-
sures of offense severity and prior record have important main, curvilinear, and inter-
active influences on in/out and sentence length that cannot be reduced to the effect of
presumptive sentence measures. This suggests that it is statistically and substantively
important to include offense severity and prior record even if one is including a pre-
sumptive sentence measure. However, Ulmer also points out that including all three
legally prescribed variables results in problematic multicollinearity in the OLS mod-
els of sentence length. As a result, an offense severity score variable is not included
in the analysis because it is highly collinear with the guideline minimum sentence
variable.
Case Disposition is a dichotomous variable, which indicates whether the offender’s
case is settled by plea agreement or trial. It is coded 1 for trial and 0 for guilty plea. We
also include a measure of multiple counts. A dummy variable is coded 0 for cases
involving a single count and 1 for cases that involve multiple counts. The defendant’s
offense type (see Appendix for a complete breakdown of categories) is coded as four
dummy variables: violent (i.e., murder, manslaughter, sexual abuse), drug (i.e., traf-
ficking, simple possession), white-collar (i.e., fraud, embezzlement, bribery), and
other (includes all other offenses in the federal data). Defendants committing other
types of offenses serve as the reference group. The variable departure indicates the
defendant’s departure status. Departure status is dummy-coded into 3 categories: No
departure (the reference), downward departure, and substantial assistance departure.
Upward departure cases were deleted from the sample as they only made up 0.8% of
the sample and deleting them does not significantly change the findings. The federal
sentencing statutes include provisions that permit judges to depart either above or
below the sentence prescribed by the guidelines. Judges may award these sentencing
departures based on a legitimate reason if they feel the defendant does not deserve the
sentence stated under the prescribed guidelines. Overall, however, the overwhelming
direction of departures is downward.
The narrow range of factors that judges may consider when sentencing either above
or below the prescribed guideline range makes the Federal Sentencing Guidelines
much more rigid than similar state structured sentencing systems (Farrell, 2004).
Consequently, federal courts are prohibited from departing from the Guidelines based
on the race, gender, religion, or class of an individual defendant. However, the
Sentencing Commission has deferred to the courts to interpret how extensively judges
may use offender characteristics to justify departures from the guideline range.
Several control variables are also included in the models. Since multiple years of
data were used in the present study, a dummy variable for each of the three years was
constructed. Prior studies have indicated that judicial circuit, as well as other court
contextual variables, may be important influences on sentencing outcomes (Peterson
& Hagan, 1984; Steffensmeier & Demuth, 2000). One cause of disparities is that not
at UNIV OF RHODE ISLAND LIBRARY on July 19, 2013cjp.sagepub.comDownloaded from

12 Criminal Justice Policy Review XX(X)
all states or judicial circuits have implemented guidelines systems. The variable judi-
cial circuit indicates the judicial circuit is which the defendant was sentenced. Judicial
circuits are broken down into 11 categories, which were then made into dummy
variables.
Results
In the present study, we analyze the data and present the results in several stages. In
the first section, we present descriptive statistics for all variables used in the models
(Table 1). Second, we use logistic and OLS regression (including only those defen-
dants who receive a prison sentence) to examine the independent effects of gender on
incarceration and sentence length decisions (Table 2) in three separate models. Third,
we partition the full model by gender, examining the differential influence of legal
and extralegal variables on sentencing outcomes of male and female defendants
(Table 3). It is important to note that the data set we use in the present study is not a
sample. It includes the entire population of defendants sentenced in the federal courts
during the period. As such, statistical tests of significance are not particularly mean-
ingful in that there is no sampling error and no need to make inferences (Berk, 2010;
Raftery, 1995). In our discussion of results, we focus mostly on the size and direction
of coefficients, but nonetheless include indicators of significance (p < .05) in the
tables.
Descriptive Statistics
Overall, men make up 83% of the sample. In terms of race, we found similar percent-
ages in each racial category for both men and women. The plurality of defendants in
the sample are White, approximately 44%, while34% are Black and 18% are Hispanic.
In terms of age, the largest portion of the sample fell in the 21 to 29 age range, fol-
lowed closely by the 30 to 39 year age range.
Looking at sentencing outcomes, a smaller percentage of women are incarcerated
than men, with 85% of men receiving a prison sentence while only 62 % of females in
the sample are incarcerated. The sentence length gap for incarcerated defendants is
also quite substantial between male and female defendants; male defendants receive
sentence lengths of roughly 70 months, while female defendants are sentenced to
approximately 34 months of incarceration. The average sentence length for the total
sample falls close to that for male defendants (approximately 65 months).
These large differences in sentencing outcomes may be explained by both legal and
extralegal factors. In terms of legal characteristics, male defendants have higher crimi-
nal histories, and they also receive higher recommended minimum guideline sentences
than do female defendants due to the greater severity of the offenses committed by
men. In addition, a higher percentage of male defendants are sentenced on multiple
counts. Furthermore, a smaller percentage of female defendants go to trial. However,
men and women receive sentencing departures at similar rates. A higher percentage of
at UNIV OF RHODE ISLAND LIBRARY on July 19, 2013cjp.sagepub.comDownloaded from

Doerner and Demuth 13
Table 1. Descriptive Statistics by Gender
Overall Males Females
Independent variables N Percentage NPercentage NPercentage
Gender
Male 90,297 82.70 90,297 100 — —
Female 18,884 17.30 — — 18,884 100
Race
White 48,003 43.97 39,568 43.82 8,435 44.67
Black 37,541 34.38 31,408 34.78 6,133 32.48
Hispanic 19,348 17.72 15,988 17.71 3,360 17.79
Other 4,289 3.93 3,333 3.69 956 5.06
Age
18-20 5,427 4.97 4,516 5.00 911 4.82
21-29 37,777 34.60 31,455 34.84 6,322 33.48
30-39 32,702 29.95 26,950 29.85 5,752 30.46
40-49 20,305 18.60 16,427 18.19 3,878 20.54
50-59 9,537 8.74 8,000 8.86 1,537 8.14
60 & over 3,433 3.14 2,949 3.27 484 2.56
Legal variables
Multiple counts 23,142 21.20 20,274 22.45 2,868 15.19
Trial 4,536 4.15 4,062 4.50 474 2.51
Prior criminal history (points) 2.40 — 2.57 — 1.60 —
Guideline minimum sentence
(months)
58.92 — 65.11 — 29.33 —
Offense type
Violent 6,092 5.58 5,609 6.21 483 2.56
Drug 48,688 44.59 41,626 46.10 7,062 37.40
White-collar 23,259 21.30 16,371 18.13 6,888 36.48
Other 31,142 28.52 26,691 29.56 4,451 23.57
Departures
No departure 72,938 66.80 60,816 67.35 12,122 64.19
Downward departure 12,866 11.78 10,289 11.39 2,577 13.65
Substantial assistance
departure
23,377 21.41 19,192 21.25 4,185 22.16
Education
Less than high school 38,587 35.34 32,794 36.32 5,793 30.68
High school 40,484 37.08 33,544 37.15 6,940 36.75
More than high school 30,110 27.58 23,959 26.53 6,151 32.57
Marital status
Single 48,909 44.80 41,349 45.79 7,560 40.03
Married 30,588 28.02 25,448 28.18 5,140 27.22
Cohabit 10,702 9.80 9,087 10.06 1,615 8.55
Divorced 12,529 11.48 9,817 10.87 2,712 14.36
(continued)
at UNIV OF RHODE ISLAND LIBRARY on July 19, 2013cjp.sagepub.comDownloaded from

14 Criminal Justice Policy Review XX(X)
Overall Males Females
Independent variables N Percentage NPercentage NPercentage
Widowed 626 0.57 315 0.35 311 1.65
Separated 5,827 5.34 4,281 4.74 1,546 8.19
Number of dependents
No dependents 44,677 40.92 37,411 41.43 7,266 38.48
One or more dependents 64,504 59.08 52,886 58.57 11,618 61.52
Dependent Variables
Incarcerated 88,647 81.19 76,979 85.25 11,668 61.79
Sentence length (months)a65.12 — 69.80 — 34.25 —
N109,181 90,297 18,884
aSentence length is for those who received an incarceration sentence.
Table 1. (continued)
males commit violent, drug, and other offenses, while a higher percentage of females
commit while-collar offenses compared to their male counterparts.
Looking at extralegal factors that might be related to gender, a slightly higher per-
centage of female defendants have one or more dependents. More specifically, about
62% of female defendants have at least one dependent, compared to 59% for male
defendants. Also, male defendants are more likely to be single than female defendants
(46% vs. 40%), but female defendants are more likely to be divorced (14% vs. 11%)
or separated (8% vs. 5%) than male defendants. Furthermore, a higher percentage of
female defendants, roughly 6% more, have more than a high school education com-
pared to their male defendant counterparts.
Independent Effects of Gender
Table 2 shows the main effects of gender in three nested models.5 Model 1 controls
only for basic defendant demographics including gender, race, and age. Overall,
female defendants have odds of incarceration roughly 74% lower than similarly situ-
ated male defendants. Hispanic defendants have the highest odds of incarceration,
while White defendants have the lowest, and Black defendants fall in the middle. The
odds of incarceration follow an upside-down U-shaped pattern with increasing age.
Defendants age 21 to 39 have odds of incarceration roughly 40% to 50% higher than
defendants age 18 to 20. After age 50, the likelihood of receiving an incarceration
sentence drops substantially, with defendants age 60 and over having odds of incar-
ceration roughly half that of the youngest defendants.
For the sentence length decision, female defendants receive sentences that are
about 50% (exp[b]) shorter than similarly situated male defendants. Black defendants
receive the longest sentence lengths, approximately 50% longer than White defen-
dants. Hispanic defendants fall in the middle when it comes to sentence length out-
comes. Overall, sentence lengths increase until age 30 to 39, then decrease thereafter,
at UNIV OF RHODE ISLAND LIBRARY on July 19, 2013cjp.sagepub.comDownloaded from

Doerner and Demuth 15
Table 2. Main Effects Model
Model 1 Model 2 Model 3
Variable In/out Ln(Length) In/out Ln(Length) In/out Ln(Length)
Gender
Malea— — — — — —
Female 0.26* −0.70* 0.61* −0.25* 0.61* −0.25*
Race
Whitea— — — — — —
Black 1.65* 0.40* 0.96 0.04* 0.95 0.03*
Hispanic 1.80* 0.13* 1.40* −0.03* 1.34* −0.03*
Age
18-20a— — — — — —
21-29 1.48* 0.26* 1.00 0.05* 1.08 0.06*
30-39 1.42* 0.32* 0.88* 0.04* 0.98 0.05*
40-49 1.05 0.20* 0.84* 0.04* 0.94 0.05*
50-59 0.73* 0.08* 0.74* 0.06* 0.82* 0.07*
60 & over 0.48* −0.06* 0.54* 0.02 0.59* 0.03
Legal variables
Multiple counts 1.64* 0.29* 1.65* 0.29*
Trial 1.68* 0.10* 1.71* 0.10*
Prior criminal history 1.66* 0.06* 1.62* 0.06*
Guideline minimum
sentence
1.12* 0.01* 1.12* 0.01*
Offense type
Violent 1.80* 0.39* 1.77* 0.39*
Drug 1.41* 0.26* 1.37* 0.26*
White-collar 1.18* −0.41* 1.23* −0.41*
Othera— — — —
Departures
No Departurea— — — —
Downward departure 0.27* −0.41* 0.27* −0.41*
Substantial assistance
departure
0.12* −0.44* 0.12* −0.44*
Education
Less than high school 1.35* 0.02*
High schoola— —
More than high school 0.99 −0.01*
Marital status
Singlea— —
Married 0.92* −0.01*
Cohabiting 1.07 0.00
Divorced 1.15* 0.01
Widowed 0.67* 0.02
Separated 1.07 0.00
(continued)
at UNIV OF RHODE ISLAND LIBRARY on July 19, 2013cjp.sagepub.comDownloaded from

16 Criminal Justice Policy Review XX(X)
Model 1 Model 2 Model 3
Variable In/out Ln(Length) In/out Ln(Length) In/out Ln(Length)
Number of dependents
No dependentsa— —
One or more dependents 0.92* 0.00
Max-resealed R20.13 — 0.59 0.59 —
Adjusted R2— 0.12 — 0.67 — 0.63
N109,181 88,647 109,181 88,647 109,181 88,647
Controls for circuit and year are included in all models.
aRepresents the reference category.
*p < .05.
Table 2. (continued)
with defendants age 60 and over receiving sentences similar to those received by
defendants age 18 to 20.
Model 2 builds on the baseline variables by adding legal factors indicating number
of counts, trial or guilty plea, prior criminal history, guideline minimum sentence
(which accounts for offense severity), offense type, and receipt of departure. As
expected, the legal factors are strongly related to whether a defendant receives a prison
sentence or probation. Defendants with longer criminal histories are more likely to be
sentenced to prison than defendants with shorter criminal records. In addition, defen-
dants that are sentenced for multiple offense counts have odds of incarceration that are
64% higher than defendants sentenced on only a single count. Furthermore, defen-
dants that go to trial are more likely to be sentenced to an incarceration term than
defendants that plead guilty (odds ratio = 1.68). Defendants who commit violent
offenses have the highest odds of incarceration, roughly 80% higher than defendants
in the other offense category. Defendants committing drug and white-collar offenses
are also more likely to be incarcerated (41% and 18 %, respectively) than the reference
group. Finally, defendants receiving a sentencing departure are less likely to receive
an incarceration sentence than defendants who do not receive a sentencing departure.
Looking at gender, net of legal factors, the odds of incarceration for females are 39%
lower than the odds of incarceration for males. This represents a substantial reduction
in the gender gap as compared to the findings presented in Model 1 where the odds of
incarceration for women are 74% lower for women than men.
Similar findings emerge for sentence length in Model 2. After controlling for legal
factors, female defendants receive sentences approximately 23% shorter than those
received by male defendants. As with the in/out decision, defendants with longer crim-
inal histories and those who go to trial receive slightly longer sentences. Those defen-
dants with multiple counts receive sentences approximately 34% longer than those
sentenced for only a single count. In addition, defendants who commit violent or drug
offenses receive significantly longer sentences (48% and 30% longer, respectively)
than those defendants in the reference group. However, defendants who commit
at UNIV OF RHODE ISLAND LIBRARY on July 19, 2013cjp.sagepub.comDownloaded from

Doerner and Demuth 17
Table 3. Main Effects Model—Males Versus Females
Males Females
Variable In/out Ln(Length) In/out Ln(Length)
Race
Whitea— — — —
Black 1.02b0.05*b0.81*b−0.04*b
Hispanic 1.44*b−0.03* 1.13*b0.00
Age
18-20a— — — —
21-29 1.19*b0.06* 0.85b0.04
30-39 1.03 0.05* 0.88 0.08*
40-49 1.00 0.03*b0.80* 0.15*b
50-59 0.90b0.04*b0.65*b0.23*b
60 & over 0.66*b0.01b0.46*b0.18*b
Legal variables
Multiple counts 1.60* 0.29* 1.84* 0.27*
Trial 1.74* 0.09*b1.69* 0.21*b
Prior criminal history 1.59*b0.06*b1.70*b0.09*b
Guideline minimum sentence 1.12* 0.01*b1.12* 0.01*b
Offense type
Violent 1.78* 0.37*b1.56* 0.65*b
Drug 1.27*b0.23*b1.76*b0.42*b
White-collar 1.07*b−0.40*b1.75*b−0.24*b
Othera— — — —
Departures
No Departurea— — — —
Downward departure 0.25*b−0.40*b0.31*b−0.47*b
Substantial assistance departure 0.13* −0.45* 0.12* −0.44*
Education
Less than high school 1.46*b0.01b1.13*b0.05*b
High schoola— — — —
More than high school 1.02 −0.02*b0.95 0.05*b
Marital status
Singlea— — — —
Married 0.90* −0.01 0.95 −0.03
Cohabiting 1.06 0.00 1.07 0.00
Divorced 1.10* 0.02 1.23* −0.03
Widowed 0.78 −0.06 0.57* 0.05
Separated 1.06 0.00 1.09 −0.03
Number of dependents
No dependentsa— — — —
One or more dependents 0.95 0.01 0.89* 0.02
Max-resealed R20.58 — 0.55 —
Adjusted R2— 0.67 — 0.58
N90,297 76,979 18,884 11,668
Controls for circuit and year are included in all models.
aRepresents the reference category.
bCoefficients are different between male and female defendants at p < .05 level (two-tailed z-test).
*p < .05.
at UNIV OF RHODE ISLAND LIBRARY on July 19, 2013cjp.sagepub.comDownloaded from

18 Criminal Justice Policy Review XX(X)
white-collar offenses, or receive a sentencing departure are given significantly shorter
sentence length outcomes than their respective reference categories. Notably, by
including legal variables in the model, the male–female gap in sentence length is
reduced from a 50% difference to a 23% difference.
Model 3 represents the full model and includes three groups of variables indicating
educational attainment, marital status, and number of dependents. These extralegal
variables were added separately because they can be considered gendered in nature.
The odds ratio for female defendants remains the same as in Model 2, indicating that
female defendants have odds of incarceration approximately 39% lower than male
defendants with similar characteristics. Defendants with less than a high school educa-
tion are more likely to be incarcerated than those with a high school education.
Furthermore, defendants that are divorced have higher odds of incarceration than
defendants that are single, while married and widowed defendants are less likely to be
incarcerated. In addition, defendants that have one or more dependents are signifi-
cantly less likely to be incarcerated than defendants who have no dependents. In terms
of the sentence length decision, female defendants receive the same sentence length
outcome as they did in Model 2, even after the addition of educational attainment,
marital status, and number of dependents. Overall, there remains a moderately large
gender gap that cannot be explained by legal and extralegal factors.
Main Effects Models by Gender
In Table 3, we present the results separately for the male and female defendants in the
sample. This is done to determine whether legal and extralegal factors differentially
influence the sentencing outcomes of male and female defendants.
In terms of race, incarceration outcomes appear to be influenced differently for men
and women. Hispanic male and female defendants have the highest odds of incarcera-
tion with defendants roughly 44% and 13% more likely to be incarcerated than their
respective White counterparts. On the other hand, Black female defendants have the
lowest odds of incarceration compared to White females. We use z-tests of difference
of means to compare coefficients between models. Z-tests of difference indicate that
the having prior criminal history plays a stronger role for female defendants than male
defendants. This also holds true for female defendants who commit drug and white-
collar offenses. Defendants, male and female, have lower odds of incarceration if they
receive a sentencing departure, but the magnitude of the effect appears to be similar
for both gender groups who receive substantial assistance departures. In addition,
being less educated hurts male defendants more than women. More specifically, male
defendants completing less than a high school education are 46% more likely to be
incarcerated than those male defendants with a high school education.
In terms of sentence length outcomes, the results for male and female defendants
are somewhat different. Black male defendants receive the longest sentence terms,
roughly 5% longer than similarly situated White defendants. On the other hand, Black
female defendants receive the shortest sentence length outcomes, approximately 4%
shorter than their White female counterparts. Defendants, both male and female, who
at UNIV OF RHODE ISLAND LIBRARY on July 19, 2013cjp.sagepub.comDownloaded from

Doerner and Demuth 19
go to trial and those with prior criminal history receive longer sentence lengths overall.
Defendants receiving sentencing departures are given significantly shorter sentences
than defendants who do not receive a sentencing departure however this appears to
play a slightly larger role for female defendants who receive downward sentencing
departures. Having anything but a high school education appears to play a stronger
role for females than males, with female defendants receiving sentences 5% longer
than female defendants that finish high school.
Overall, when it comes to the incarceration decision, several things were found to
weigh differently for male and female defendants. Racial differences were found among
defendant groups, with Hispanic males and females most likely to be incarcerated and
Black females least likely to be incarcerated. In terms of legal variables, having prior
criminal history plays a stronger role for women than men. For the extralegal measures,
having less education negatively effects the sentencing outcomes of men. Looking at
sentence length outcomes, racial differences were found. Black male defendants receive
the longest sentence lengths, while Black female defendants receive the shortest.
Educational differences were also found. Having anything but a high school education
leads to negative effects for female defendants (longer sentences).
Discussion and Conclusions
The current study had several major goals. First we wanted to perform a rigorous
analysis of the possible causes of gender disparities in sentencing outcomes. Gender
disparities are quite common and usually discouraged or prohibited by statute yet
receive relatively little attention in the literature. Furthermore, many past studies have
used older data, small localized samples, or have not had sufficiently robust legal
measures with which to provide adequate statistical control. In the current study, we
used some of the richest and most detailed data available to examine how differences
in the legal and extralegal case characteristics of men and women contribute to the
gender gap in sentencing.
Second, beyond explanations based on differences in legal case characteristics, we
wanted to gain a better understanding of how gender impacts sentencing outcomes
through other extralegal factors related to both gender and sentencing. Past studies
typically examine gender as a fixed attribute and do not consider how gendered
roles might impact court decisions. In the current study, we drew on research from the
areas of criminology, criminal justice, and family sociology to examine whether dif-
ferences in marriage, education, and the presence of dependents helped to account for
the gender gap. We also looked to see if there were gender differences in the impact of
extralegal and legal factors on sentencing outcomes.
Finally and more broadly, the current study set out to address the limitations of the
criminal justice system after the implementation of fixed sentencing reforms like for-
mal guidelines designed to reduce unwarranted extralegal disparities. Central to the
guidelines is the notion that defendant characteristics such as gender should not be
considered during the sentencing process. However, even with these guidelines in
place, gender disparities persist, calling into question the effectiveness of their
at UNIV OF RHODE ISLAND LIBRARY on July 19, 2013cjp.sagepub.comDownloaded from

20 Criminal Justice Policy Review XX(X)
implementation. In the current study, we examined possible mechanisms by which
gender may influence the sentencing process in spite of guidelines.
Consistent with prior sentencing research, we found that legal factors play an
important role in determining sentencing outcomes. Overall, regardless of gender,
defendants with more extensive criminal histories and those who committed more
serious offenses were more likely to receive harsher sentences than defendants with
less serious criminal pasts and current convictions. However, the findings of the cur-
rent research also showed that gender appears to have a significant effect on sentenc-
ing outcomes, after accounting for legal and extralegal factors. Female defendants
were less likely to receive an incarceration sentence than male defendants and also
received shorter sentence length terms.
Several important findings emerged from the analysis in relation to our research
questions and hypotheses. As expected in our first hypothesis, female defendants
received more lenient sentence outcomes than their similarly situated male counter-
parts. Second, legal factors accounted for a considerable portion of the gender gap in
sentencing. However, even after accounting for these legal factors, a sizeable gender
gap remained in that male defendants continued to be sentenced more harshly than
their female counterparts, as proposed in our second hypothesis. Third, although edu-
cation level, marital status, and number of dependents appeared to influence sentenc-
ing outcomes in some instances, they did not help to minimize the gender gap in
sentencing outcomes. Thus, our third hypothesis was supported in the expected direc-
tion in that defendants who have more marital stability and dependents received more
lenient sentence outcomes, but there were no significant advantages for defendants
with more than a high school education. One reason as to why this group of variables
may not be helping to narrow the sentencing gap between male and female defendants
is that judges on the federal level, compared to the state level, are more insulated from
community pressures and political forces and less able to exercise their discretion than
their state or local counterparts. Overall, the gender gap in sentencing outcomes can-
not be fully explained by accounting for legal and extralegal factors.
Finally, contrary to our expectations in hypothesis four, when each gender group
was examined separately we found that some legal and extralegal factors did influence
sentencing differently for male and female defendants. In terms of legal variables,
prior criminal history played a more important role in receiving an incarceration sen-
tence for female than male defendants. In terms of extralegal variables, having less
than a high school education negatively influenced the incarceration decision of male
defendants (raising their odds of incarceration). However, when it came to sentence
length outcomes, having less than, or more than, a high school education increased
sentence lengths for female defendants. Race also influenced male and female defen-
dants differently. For the incarceration decision, Hispanic male and female defendants
had the highest odds of being sent to prison, while Black females had the lowest odds
of incarceration. For the sentence length decision, Black males received the longest
sentence length terms and Black female defendants received the shortest terms.
Overall, legal and extralegal factors were found to have differential impacts on male
and female defendants.
at UNIV OF RHODE ISLAND LIBRARY on July 19, 2013cjp.sagepub.comDownloaded from

Doerner and Demuth 21
The results of the current study are consistent with the focal concerns perspective
(Steffensmeier, 1980; Steffensmeier et al., 1993, 1998) that argues that legal decision
making is organized around concerns of blameworthiness, protection of the commu-
nity, and practical constraints and consequences. Overall, the primary influences of
sentencing decisions are legal factors (e.g., prior criminal history, offense serious-
ness); however we also found that extralegal characteristics play an important role in
some defendant’s outcomes. The findings support the idea that judges attribute mean-
ing to past and present behavior of defendants, as well as stereotypes associated with
various gender or racial/ethnic groups. These extralegal sources of sentencing dispar-
ity indicate that these stereotypes may be very influential and that inequalities in the
application of the law and subsequent court proceedings may be taking place, despite
the existence of sentencing guidelines designed to avoid such unequal treatment.
One limitation of this study was that socioeconomic status (SES) information was not
available in the data set (Monitoring of Federal Criminal Sentences), and thus, could not be
included in the current analysis. It is not unusual for measures of SES to be missing from
sentencing research. In prior years of federal data a variable representing defendant income
was available, however over 50% of defendants listed their incomes as US$0, making it
difficult to analyze the true effects of this variable and how it might interact with gender (see
Steffensmeier & Demuth, 2000). Future research should explore the extent to which gen-
der disparities are truly a function of gender perceptions versus economic constraints that
limit the ability of defendants to resist legal sanctions and acquire appropriate counsel.
Another limitation of the current study is that the variable indicating number of
dependents does not differentiate between the types of dependents. In other words, it
is unclear as to whether the defendant is claiming responsibility for their dependent
children, their spouse or significant other, some other family member, or a combina-
tion of all of the above. Much of the prior research cited in the current study specifi-
cally explores the effect of children on sentencing outcomes, regardless of the
defendant’s marital context (Bickle & Peterson, 1991; Daly, 1987a, 1987b; 1989;
Eaton, 1987; Farrington & Morris, 1983; Kruttschnitt & Green, 1984; Kruttschnitt &
McCarthy, 1985; Simon, 1975). However, in this context, the definition leaves much
room for interpretation. This is especially true given the very different worlds of par-
enting across various racial/ethnic groups, including instances of multiple partner fer-
tility, mixed family households, extended family care, and responsibilities for aged
dependents. Therefore, future research would benefit from an analysis broken down
by marital status, specifically targeting single defendants, to determine if significant
differences are present when children are the only dependent examined. Furthermore,
future research should strengthen our understanding of different family forms, espe-
cially across racial/ethnic groups and same-sex partnerships.
In conclusion, the topic of differential treatment at sentencing will continue to be an
important topic, given the Supreme Court decisions (Blakely v. Washington; U.S. v.
Booker; U.S. v. Fanfan), which changed the sentencing guidelines from mandatory to vol-
untary. While the full implication of these changes are still to come, they will likely result
is significant changes in sentencing outcomes, and more specifically, the role that judges
and other members of the courtroom work group play in those sentencing decisions.
at UNIV OF RHODE ISLAND LIBRARY on July 19, 2013cjp.sagepub.comDownloaded from

Appendix
Breakdown of Offense Types by Category
Offense type Overall Males Females
Coding No. Name Number Percentage Number Percentage Number Percentage
Violent 6,092 5.58 5,609 6.21 483 2.56
1 Murder 145 0.13 128 0.14 17 0.09
2 Manslaughter 125 0.11 98 0.11 27 0.14
3 Kidnapping/hostage 65 0.06 58 0.06 7 0.04
4 Sexual abuse 585 0.54 566 0.63 19 0.10
5 Assault 1,032 0.95 920 1.02 112 0.59
6 Bank robbery/other robbery 4,140 3.79 3,839 4.25 301 1.59
Drug 48,688 44.59 41,626 46.10 7,062 37.40
10 Drugs: Trafficking 46,606 42.69 39,992 44.29 6,614 35.02
11 Drugs: Communication facilities 923 0.85 721 0.80 202 1.07
12 Drugs: Simple possession 1,159 1.06 913 1.01 246 1.30
White-collar 23,259 21.30 16,371 18.13 6,888 36.48
18 Fraud 14,837 13.59 10,535 11.67 4,302 22.78
19 Embezzlement 1,913 1.75 775 0.86 1,138 6.03
20 Forgery/counterfeiting 3,159 2.89 2,353 2.61 806 4.27
21 Bribery 375 0.34 329 0.36 46 0.24
22 Tax offenses 1,333 1.22 1,093 1.21 240 1.27
23 Money laundering 1,642 1.50 1,286 1.42 356 1.89
Other offenses 31,142 28.52 26,691 29.56 4,451 23.57
9 Arson 176 0.16 162 0.18 14 0.07
13 Firearms: Use/possession 13,339 12.22 12,832 14.21 507 2.68
15 Burglary/breaking & entering 118 0.11 111 0.12 7 0.04
16 Auto theft 370 0.34 352 0.39 18 0.10
(continued)
22
at UNIV OF RHODE ISLAND LIBRARY on July 19, 2013cjp.sagepub.comDownloaded from

Offense type Overall Males Females
Coding No. Name Number Percentage Number Percentage Number Percentage
17 Larceny 5,125 4.69 3,205 3.55 1,920 10.17
24 Racketeering/extortion 1,627 1.49 1,492 1.65 135 0.71
25 Gambling/lottery 271 0.25 246 0.27 25 0.13
26 Civil rights offenses 211 0.19 201 0.22 10 0.05
27 Immigration 3,299 3.02 2,597 2.88 702 3.72
28 Pornography/prostitution 1,714 1.57 1,689 1.87 25 0.13
29 Offenses in prison 741 0.68 638 0.71 103 0.55
30 Administration of justice-related 1,982 1.82 1,324 1.47 658 3.48
31 Environmental, game, fish, and wildlife
offenses
312 0.29 297 0.33 15 0.08
32 National defense offenses 16 0.01 12 0.01 4 0.02
33 Antitrust violations 39 0.04 38 0.04 1 0.01
34 Food and drug offenses 171 0.16 151 0.17 20 0.11
35 Traffic violations and other offenses 1,631 1.49 1,344 1.49 287 1.52
Overall totals 109,181 90,297 18,884
Appendix (continued)
23
at UNIV OF RHODE ISLAND LIBRARY on July 19, 2013cjp.sagepub.comDownloaded from

24 Criminal Justice Policy Review XX(X)
Acknowledgments
The authors wish to thank Steve Cernkovich, Al DeMaris, Joseph Jacoby, Laura Sanchez, and
Neal Jesse for their thoughtful comments on previous versions of this article.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship,
and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship,
and/or publication of this article: The core staff support and resources the Center for Family and
Demographic Research (CFDR) at Bowling Green State University provided were funded by a
NIH Population Research Center Grant. The only support that was directly provided for this
research was for data, coding, and other technical issues.
Notes
1. Many sentencing studies model the sentence length decision including a correction term
for selection bias stemming from the decision to incarcerate (Berk, 1983). This involves
controlling for the “hazard” of incarceration (estimated in the in/out model) in the sen-
tence length model. The hazard variable represents for each observation the instantaneous
probability of being excluded from the sample conditional upon being in the pool at risk.
However, Stolzenberg and Relles (1997) and Bushway, Johnson, and Slocum (2007) find
that this correction term can often introduce more bias into the sentence length model than
it eliminates due to high levels of collinearity between the correction term and other predic-
tors of sentence length. This is especially likely when the predictors of incarceration are
very similar to the predictors of sentencing length as in the present study. Also, Stolzenberg
and Relles (1997) argue that a correction term is often unnecessary when there is a low level
of selection. In the current data, because only 19% of defendants avoid incarceration, it is
unlikely that a selection bias will strongly influence the sentence length findings. For these
reasons, we do not include a correction term for selection bias in the sentence length model.
2. Defendants in the “Other” racial category have been included in the analysis models, but
were not included in the regression tables as they are not the focus of this study and only
constitute a small percentage of the sample (3.9%).
3. The “number of dependents” variable may not accurately represent a defendant’s potential
family responsibilities because the Sentencing Commission has not differentiated among
types of dependents (e.g., children, spouses, significant others, aged parents, or extended
family members, etc.).
4. Initial analyses were conducted using a full range of categories for this variable, but it was
found that no differences existed between higher levels of dependents.
5. All models in the analysis control for judicial circuit and year. Model fit for the full in/out
model as indicated by the area under the ROC curve (0.931) is very good. For the full
at UNIV OF RHODE ISLAND LIBRARY on July 19, 2013cjp.sagepub.comDownloaded from

Doerner and Demuth 25
sentence length model, an examination of variance inflation factor scores indicates that all
variables are well below 10, which is typically considered to be an acceptable cutoff.
References
Albonetti, C. (1991). An integration of theories to explain judicial discretion. Social Problems,
38, 247-266.
Albonetti, C. (1997). Sentencing under the federal sentencing guidelines: Effects of defendant char-
acteristics, guilty pleas, and departures on sentence outcomes for drug offenders, 1991-1992.
Law & Society Review, 31, 789-822.
Belknap, J. (2001). The invisible woman: Gender, crime, and criminal justice. Belmont, CA:
Wadsworth.
Berk, R. A. (1983). An introduction to sample selection bias in sociological data. American
Sociological Review, 48, 386-398.
Berk, R. A. (2010). What you can and can’t properly do with regression. Journal of Quantitative
Criminology, 26, 481-487.
Bernstein, I., Cardascia, J., & Ross, C. (1979). Defendant’s sex and criminal court decisions.
In R. Alvarez & K. G. Lutterman (Eds.), Discrimination in organizations (pp. 329-354).
San Francisco, CA: Jossey-Bass.
Bickle, G., & Peterson, R. (1991). The impact of gender-based family roles on criminal sentenc-
ing. Social Problems, 38, 372-394.
Bushway, S., Johnson, B., & Slocum, L. (2007). Is the magic still there? The use of the Heckman
two-step correction for selection bias in criminology. Journal of Quantitative Criminology,
23(2), 151-178.
Butler, A. (1997). Gendered justice in the American west: Women prisoners in men’s peniten-
tiaries. Champaign, IL: University of Illinois Press.
Chiricos, T., & Crawford, C. (1995). Race and imprisonment: A contextual assessment of the
evidence. In D.F. Hawkins (Ed.), Ethnicity, race, and crime: Perspectives across time and
place (pp. 281-309). Albany, NY: State University of New York Press.
Daly, K. (1986). Gender in the adjudication process: Are judges really paternalistic toward
women? Revised paper presented at the American Society of Criminology Annual Meeting,
November 1985, San Diego, CA.
Daly, K. (1987a). Discrimination in the criminal courts: Family, gender, and problems of equal
treatment. Social Forces, 66, 152-175.
Daly, K. (1987b). Structure and practice of familial-based justice in a criminal court. Law and
Society Review, 21, 267-290.
Daly, K. (1989). Rethinking judicial paternalism: Gender, work-family relations, and sentencing.
Gender and Society, 3(1), 9-36.
Daly, K. (1994). Gender, crime, and punishment. New Haven, CT: Yale University Press.
Daly, K., & Bordt, R. (1995). Sex effects and sentencing: An analysis of the statistical literature.
Justice Quarterly, 12, 141-175.
Demuth, S. (2002). The effects of citizenship status on sentencing outcomes in drug cases. Fed-
eral Sentencing Reporter, 14, 271-275.
at UNIV OF RHODE ISLAND LIBRARY on July 19, 2013cjp.sagepub.comDownloaded from

26 Criminal Justice Policy Review XX(X)
Demuth, S., & Steffensmeier, D. (2004). Ethnicity effects on sentence outcomes in large urban
courts: Comparisons among White, Black, and Hispanic defendants. Social Science Quarterly,
85, 994-1011.
Dixon, J. (1995). The organization context of criminal sentencing. American Journal of Sociol-
ogy, 100, 1157-1198.
Doerner, J., & Demuth, S. (2010). The independent and joint effects of race/ethnicity, gender,
and age on sentencing outcomes in U.S. federal courts. Justice Quarterly, 27(1), 1-27.
Eaton, M. (1987). The question of bail: Magistrate’s responses to applications for bail on behalf of
men and women. In P. Carlen & A. Worrall (Eds.), Gender, crime, and justice (pp. 95-107).
Philadelphia, PA: Open University Press.
Engen, R., & Gainey, R. (2000). Modeling the effects of legally relevant and extralegal factors
under sentencing guidelines: The rules have changed. Criminology, 38, 1207-1229.
Farrell, A. (2004). Measuring judicial and prosecutorial discretion: Sex and race disparities in
departures from the federal sentencing guidelines. Justice Research and Policy, 6(2), 45-78.
Farrington, D., & Morris, A. (1983). Sex, sentencing, and reconviction. Journal of British Crim-
inology, 21, 229-249.
Flemming, R., Nardulli, P., & Eisenstein, J. (1992). The craft of justice: Work and politics in
criminal court communities. Philadelphia, PA: University of Pennsylvania Press.
Griffin, T., & Wooldredge, J. (2006). Sex-based disparities in felony dispositions before versus
after sentencing reform in Ohio. Criminology, 44(4), 893-923.
Gruhl, J., Welch, S., & Spohn, C. (1984). Women as criminal defendants: A test for paternalism.
Western Political Quarterly, 37, 456-467.
Hogarth, J. (1971). Sentencing as a human process. Toronto, ON: University of Toronto Press.
Koons-Witt, B. (2002). Decision to incarcerate before and after the introduction of sentencing
guidelines. Criminology, 40, 297-327.
Kruttschnitt, C. (1982a). Women, crime, and dependency: An application of the theory of law.
Criminology, 19, 495-513.
Kruttschnitt, C. (1984). Sex and criminal court dispositions: The unresolved controversy. Jour-
nal of Research in Crime and Delinquency, 21, 23-32.
Kruttschnitt, C., & Green, D. (1984). The sex sanctioning issue: Is it history? American Socio-
logical Review, 49, 541-551.
Kruttschnitt, C., & McCarthy, D. (1985). Familial social control and pretrial sanctions: Does sex
really matter? Criminal Law and Criminology, 76, 151-175.
Nagel, I., & Hagan, J. (1983). Gender and crime: Offense patterns and criminal court sanctions.
In M. Tonry & N. Morris (Eds.), Crime and justice: An annual review of research (Vol. 4,
pp. 91-144). Chicago, IL: University of Chicago Press.
Peterson, R., & Hagan, J. (1984). Changing conceptions of race: Towards an account of anoma-
lous findings of sentencing research. American Sociological Review, 49, 56-70.
Pollock-Byrne, J. (1990). Women, prison, and crime. Pacific Grove, CA: Brooks/Cole.
Rafter, N. (1990). Partial justice: Women, prison, and social control. Piscataway, NJ: Transaction.
Rafter, N., & Stanko, E. (1982). Judge lawyer victim thief: Women, gender roles, and criminal
justice. Boston, MA: Northeastern University Press.
at UNIV OF RHODE ISLAND LIBRARY on July 19, 2013cjp.sagepub.comDownloaded from

Doerner and Demuth 27
Raftery, A. (1995). Bayesian model selection in social research. Sociological Methodology, 25, 111-163.
Simon, R. (1975). Women and crime. Lexington, MA: Lexington Books.
Spohn, C. (2000). Thirty years of sentencing reform: The quest for a racially neutral sentencing
process. Criminal justice 2000 (Vol. 3, pp. 427-501). Washington, DC: National Institute
of Justice.
Spohn, C. (2002). How do judges decide: The search for fairness and justice in punishment.
Thousand Oaks, CA: SAGE.
Spohn, C., & Holleran, D. (2000). The imprisonment penalty paid by young unemployed Black
and Hispanic male offenders. Criminology, 38, 281-306.
Steffensmeier, D. (1980). Assessing the impact of the women’s movement on sex-based differ-
ences in the handling of adult criminal defendants. Crime and Delinquency, 23, 344-356.
Steffensmeier, D., & Demuth, S. (2000). Ethnicity and sentencing outcomes in U.S. federal
courts: Who is punished more harshly? American Sociological Review, 65, 705-729.
Steffensmeier, D., & Demuth, S. (2006). Does gender modify the effects of race-ethnicity on
criminal sanctioning? Sentences for male and female White, Black, and Hispanic defen-
dants. Journal of Quantitative Criminology, 22, 241-261.
Steffensmeier, D., & Motivans, M. (2000). Older men and older women in the arms of criminal
law: Offending patterns and sentencing outcomes. Journal of Gerontology, 55, 141-151.
Steffensmeier, D., Kramer, J., & Streifel, C. (1993). Gender and imprisonment decisions. Crim-
inology, 31, 411-446.
Steffensmeier, D., Kramer, J., & Ulmer, J. (1995). Age differences in sentencing. Criminal
Justice Quarterly, 12, 701-719.
Steffensmeier, D., Kramer, J., & Ulmer, J. (1998). The interaction of race, gender, and age in
criminal sentencing: The punishment cost of being young, Black, and male. Criminology,
36, 763-797.
Stolzenberg, R., & Relles, D. (1997). Tools for intuition about sample selection bias and its cor-
rection. American Sociological Review, 62, 494-507.
Temin, P. (1980). Regulation and the choice of prescription drugs. American Economic Review,
70, 301-305.
Tonry, M. (1996). Sentencing matters. New York, NY: Oxford University Press.
Ulmer, J. (1995). The organization and consequences of social pasts in criminal courts. Socio-
logical Quarterly, 36, 587-605.
Ulmer, J. (2000). The rules have changed—so proceed with caution: A comment on Engen and Gainey’s
method for modeling sentencing outcomes under guidelines. Criminology, 38, 1231-1243.
Ulmer, J., & Kramer, J. (1996). Court communities under sentencing guidelines: Dilemmas of
formal rationality and sentencing disparity. Criminology, 34, 383-408.
U.S. Sentencing Commission. (2001, 2002, 2003). Monitoring of Federal Criminal Sentences
(computer files). ICPSR versions. Washington, DC: U.S. Sentencing Commission (pro-
ducer), 2001, 2002, 2003. Ann Arbor, MI: Inter-University Consortium for Political and
Social Research (distributor).
U.S. Sentencing Commission. (2004). Fifteen years of guidelines sentencing: An assessment
of how well the federal criminal justice system is achieving the goals of sentencing reform.
Washington, DC: U.S. Sentencing Commission.
at UNIV OF RHODE ISLAND LIBRARY on July 19, 2013cjp.sagepub.comDownloaded from

28 Criminal Justice Policy Review XX(X)
Bios
Jill K. Doerner is an Assistant Professor of Sociology in the Department of Sociology and
Anthropology at the University of Rhode Island. Her research interests include sentencing and
corrections, the effects of race/ethnicity, gender, and age on decision-making in the criminal
justice system, consequence of incarceration on aging prisoners, juvenile justice, and quantitative
research methods. Her publications have appeared in Justice Quarterly and Women and
Criminal Justice.
Stephen Demuth is an Associate Professor of Sociology and Director of Graduate Studies in
the Department of Sociology and Research Associate of the Center for Family and Demographic
Research at Bowling Green State University. His primary research interests include the effects
of race, ethnicity, citizenship, and gender on crime, delinquency, and criminal justice decision-
making. His publications have appeared in American Sociological Review, Criminology, Justice
Quarterly, and Social Problems.
at UNIV OF RHODE ISLAND LIBRARY on July 19, 2013cjp.sagepub.comDownloaded from