Content uploaded by Chris Melde
All content in this area was uploaded by Chris Melde on Mar 24, 2015
Content may be subject to copyright.
Gender and Robbery: A National Test
Callie Marie Rennison1
School of Public Affairs, University of Colorado Denver
School of Criminal Justice, Michigan State University
Robbery is a quintessential male crime. Female involvement in this crime is important, however,
as they are responsible for a non-trivial amount of robbery each year. While some qualitative
studies have examined female involvement in robbery, patterns discerned from these studies
have not been examined using alternative methods. The current study adds to our understanding
of the gendered nature of robbery by examining patterns associated with male and female
robbery perpetration from the victim’s perspective using the National Crime Victimization
Survey. Findings suggest the role of gender in robbery is not as influential as extent research
Keywords: Robbery, Gender, Stratification, Victim’s Perspective, Urban, NCVS
Running Head: GENDER AND ROBBERY
**We would like to thank Dr. Heith Copes and Dr. Carole Gibbs for their help on earlier drafts
of the paper. We would especially like to thank Jody Miller for her work with us on the project.
Her guidance was especially helpful. We are solely responsible for any mistakes or omissions
1 Address correspondence to Callie Marie Rennison, University of Colorado Denver, SPA, Campus Box
142, P.O. Box 173364, Denver, CO 80217-3364. Email: firstname.lastname@example.org
Feminist discourse on interpersonal behavior suggests gender is more than an individual
attribute, but instead operates as a social institution that is reproduced through social interaction
(Lorber 1994; West and Fenstermaker 1995; West and Zimmerman 1987). This view has been
used by criminologists to explain sex differences in levels of violent offending (e.g., Braithwaite
and Daly 1994; Messerschmidt 1993) and to a lesser extent how gender shapes the interpersonal
dynamics involved in criminal events (see Decker et al. 1993; Jacobs and Miller 1998; Maher
1997; Miller 1998). Research on the interpersonal dynamics of criminal events (e.g., victim and
offender behavior, offense setting) however, has relied exclusively on qualitative interview data
to understand how gender influences these situations. Using data from the National Crime
Victimization Survey (1993-2011), the current study expands this tradition by examining gender
patterns associated with the enactment of urban street robbery with a large-scale quantitative data
source. We examine victimization level dynamics associated with robbery perpetration by male,
female, and group-based offenders to determine whether and how gender influences the nature of
The case of female perpetrated robbery presents a compelling circumstance for scholars
studying crime and violence. Robbery has been described as the “ideal opportunity to construct
an ‘essential’ toughness and ‘maleness’” (Messerschmidt 1993: 107), which means that women
involved in this crime must transgress both legal and social boundaries to commit these acts.
Indeed, roughly 87 percent of all robbery arrests recorded in the 2011 Uniform Crime Reports
were of males; although the proportion of female perpetrated robbery is up over two percent
from 2001 levels (Federal Bureau of Investigation 2012), consistent with evidence that female
violence is increasingly coming to the attention of the criminal justice system (Steffensmeier,
Schwartz, Zhong and Ackerman 2005). If gender influences, and is reproduced through, social
interaction, examination of the incident level characteristics of urban street robbery, a
particularly gender-differentiated crime, may prove fruitful in documenting these social
processes in action.
Extant research on gender and robbery (Baskin, Sommers and Fagan 1993; Brookman,
Mullins, Bennett and Wright 2007; Miller 1998; Sommers and Baskin 1993) has provided
detailed accounts of the situational context of armed robbery from the perspective of the
offender. Results reveal, to a greater or lesser degree, a gender differentiated pattern of behavior
based on the gender of the perpetrator(s). The current study builds on this literature by examining
the gendered nature of male and female robbery perpetration from the victim’s perspective using
the National Crime Victimization Survey (NCVS). The NCVS is particularly useful in this
regard given the detailed accounts of robbery situations, and thus allows us to examine the
interactional nature of robbery, including whether gender influences such factors as successful
completion of the crime, the use of weapons, victim resistance, and the extent and severity of
GENDER AND THE ACCOMPLISHMENT OF ROBBERY
From a social interactionist perspective, all violent crime is a consequence of a decision-making
process (Felson 1993). Given the act of robbery involves the interaction of offender(s) and
victim(s), both of which are active decision-makers during the incident, the criminal event “is
surrounded by a history and an environment” (Block 1981:743) that impinges upon the behavior
of the individual parties involved. Wright and Decker (1997) suggested that “armed robberies
invariably include a strong interactional component; offenders and victims must develop “a
common definition of the situation” and co-orient their actions to meet the demands of the
offense (Luckenbill 1981: 25)” (p. 95-96). It is, thus, important to understand the social
processes that influence this interaction. Because robbery, from a social interactionist
perspective, is not driven solely by processes internal to the offender (e.g., anger, frustration),
gendered social structures likely influence how robbery offenders and victims interact. Offender
decision-making, from target selection to the anticipated and actual use of force, according to
Miller (1998), is likely to indicate and reproduce “gendered social hierarchies” (p. 38).
Though women who rob transcend both legal and normative gender codes, this does not
mean that gender fails to influence the commission of such crimes. Miller (2000) suggests that
“gender inequality is a salient feature of urban street scenes,” and is a barrier that women
engrossed in such environments must deal with regularly. How “gender inequality and
stratification within criminal networks shape women’s patterns of offending” (Miller 2000: 26),
however, has received little empirical attention (for exceptions see Decker et al. 1993; Jacobs
and Miller 1998; Maher 1997; Miller 1998).
Evidence from interviews with offenders suggests gender norms influence robbery
incidents, although researchers disagree on the nature of such effects. Research on female and
male robbers suggests that the genders of both the robber and his or her victim strongly influence
how street robbers commit their crimes (Brookman et al. 2007; Miller 1998, 2000; Sommers and
Baskin 1993). For instance, Miller (1998) and Brookman and colleagues (2007) found that both
male and female robbers believed that women were easier to rob, because women were
perceived both as less likely to be armed and as weaker and thus more willing to comply. In spite
of such beliefs, however, only 30 percent of the men in Miller's (1998) sample admitted to ever
having robbed a woman.
Two pervasive explanations are given by male robbers for the non-utilitarian decision to
avoid female victims even though they are considered easy targets (Miller 2000). First, according
to Miller (2000), male robbers generally believed that females were unlikely to have much
money on them, meaning the risks involved in robbing females, however small they may be, are
not worth the trouble. Second, robbery was viewed as a mechanism for achieving status on the
streets. Similar to other acts of violence, victimization of females is unlikely to be admired or
increase one’s social standing in the masculine social environment of the street, and thus doing
so may actually hurt one's reputation (Brookman et al. 2007; Miller 2000). Together these
explanations provide some rationale for why the males in Miller’s (1998) sample reported
robbing few women.
The prototypical robbery committed by males, according to respondents in Miller’s
(1998) study, included other decidedly masculine elements besides the strong focus on male
victims. For instance, Miller characterized her male respondents’ accounts of robbery as being
strikingly similar across subjects, and included “the routine use of guns, physical contact with the
victim, and (in some cases) physical violence” (2000:35). Several of the hypotheses proposed in
Miller’s (2000) discussion of the gendered nature of robbery are listed in Figure 1.
--FIGURE ONE ABOUT HERE--
While male robbers have offered a rather circumscribed set of situational characteristics
related to the enactment of robbery (Miller 1998, 2000), studies of females who engage in this
crime suggests a much more nuanced explanation of their methods for accomplishing the same
crime. In particular, females tend to describe unique characteristics related to the robbery of
female versus male subjects. Female respondents reported that they typically targeted other
females, and used physical force, or the threat thereof, to intimidate victims into compliance.
Like some male respondents, females suggested that women were easier targets and could be
robbed without much threat of resistance. Females reported that they rarely used guns when
robbing other females because they thought that women were particularly unlikely to be armed.
Instead, they were much more likely to use no weapon at all, or use a knife when a weapon was
deemed necessary. Finally, while women reported the use of female accomplices when robbing
other females, they stated that they never used men in such circumstances.
Female robbers tend to report being especially careful when robbing males (Brookman et
al. 2007; Miller 1998, 2000; Sommers and Baskin 1993). For instance, Miller (1998) suggested
that the threat of resistance by male victims led to the near universal use of guns and the
maintenance of a safe distance from male victims because of a perceived high likelihood of
resistance, as female robbers did not want to risk being overpowered. Respondents in Miller’s
study (1998, 2000) also reported that they typically lured male victims to non-street settings
where they believed they could gain better control of the situation, suggesting that differences in
offender tactics are partly the product of perceived practicalities in garnering victim compliance.2
Although studies by Sommers and Baskin (1993) and Brookman and colleagues (2007)
also suggest women take special care when robbing males, their findings indicate fewer
differences between male and female robbers than found by Miller (1998, 2000). While Miller
concluded that women predominately victimize other women, Sommers and Baskin (1993:147)
concluded that “robbery victims…were as likely to be males as females.” The reason for parity
In contrast, female offenders reported robbing other women in the street, as the dynamics of this
setting were not viewed as problematic in such instances.
2 Even though it was common for women in Miller’s (1998, 2000) study to report having appeared sexually
interested in their male victims in order to manipulate the situational characteristics in their favor, our study is not
well suited to examine this particular feature of female perpetrated robbery.
in their study was not because gender failed to play a part in robbery perpetration, rather
respondents in Sommers’ and Baskin’s (1993) research described targeting men who were in the
company of women, because men in such circumstances were particularly willing to give up
their valuables to protect the women from violence; they used gender norms of chivalry to their
advantage. They also described targeting men who had been drinking because it limited their
capacity to resist. Thus, female-on-male robbery may be far more prevalent than suggested by
Miller (1998, 2000).
Overall, research suggests that male and female robbers enact their crimes in gendered
ways, and in accord with common gender stereotypes. Females are universally viewed as passive
victims who are unlikely to carry weapons or put up much of a physical fight, making them
relatively easy to victimize. Men, on the other hand, are expected to have a weapon and
forcefully resist, meaning they must be controlled through other means. Unlike male offenders
who overwhelmingly report the use of weapons and proactive physical force to control male
victims, females report manipulating the situation to avoid physical contact with male victims,
including using a firearm and keeping a safe distance from the victim. Exploration of these
behavior patterns with alternative, and more generalizable, data sources is imperative, however.
The Need for Alternative Data Sources
Face-to-face interviews are invaluable, especially in the genesis of highly detailed data and the
formulation of hypotheses and theories on crime and criminality. Still, they are limited as are all
methodologies. Face-to-face interviews are limited in two specific ways pertinent to the current
study. First, studies utilizing face-to-face interviews with male and/or female robbery offenders
have a propensity to have relatively few respondents, and are drawn from homogenous
populations and/or locations (e.g., single city, prison). For instance, Miller’s (2000) hypotheses
were drawn from 14 women and 23 men located in one inner-city urban area (i.e., St. Louis,
MO) in the United States. While a substantial portion of the St. Louis populace is not black,
nearly all of the respondents in the data used by Miller were of this race; a consequence of the
snowball sampling technique of the original study (Wright and Decker 1997). Hypotheses based
on such data, therefore, should be examined using different samples to determine if the patterns
found among a particular group of offenders in specific locations can be generalized to the
A second limitation of relying on face-to-face interviews in studying the gendered nature
of robbery relates to the influence of gender norms on the interview process itself (see e.g.,
Miller 2011). As Lynch (1991:98) stated, “If there is no time out from society, then
measurements of things are measurements in, of, and as social practices.” If robbery is an ideal
setting for demonstrating masculine identities on the streets (Messerschmidt 1993), then the
accounts of male robbery offenders in an interview setting are likely also part of this same
mechanism of demonstrating masculinity. In this sense, just as violence against women on the
streets is not part and partial of a masculine identity, there may be a tendency for males to
“excuse, rationalize, justify, and minimize” their violence against women (Anderson and
Umberson, 2001:361) in order to present themselves as “morally decent” in the interview setting
(Presser 2004: 83; see also Hochstetler, Copes and Williams 2010; Presser 2009; Sandberg
With respect to the current study, we might expect that male-on-female robbery
perpetration may be particularly impacted by the desire of male respondents to portray their
criminal exploits within “proper” gender parameters. As such, males may have minimized,
denied, or avoided discussing the details of their robberies of female victims. In fact, Brookman
and colleagues (2007) provided a specific example of just such a practice among their male
respondents in the United Kingdom, where Thomas stated “I never done anything like that
before, that’s not really me … I feel terrible that I robbed that woman so I don’t want to talk
about it really … I am so ashamed of myself” (Brookman et al. 2007:874). For a male to discuss
their robberies of women, even in the context of an interview about crime, is something that may
diminish the subject’s ability to effectively portray a masculine identity, and may lead to a
systematic difference between what robbers do in practice, and what they say they do in the
context of an interview about their offending. Using data from victims disallows the possibility
of an offender trying to “save face” or minimize the perpetration of robbery against a female.
While all research strategies have limitations, including our own—which we discuss below, we
attempt to add to this growing body of literature on the influence of gender on robbery by using
nationally representative victimization data on robbery in the United States from 1993 to 2011.
DATA, SAMPLE, MEASURES AND ANALYTIC STRATEGY
This study utilizes 1993 to 2011 NCVS data to investigate the relationship between gender and
3 These years used in the analyses include all NCVS data for which there is a full year of data available
following the 1992 split-sample redesign.
NCVS data are sponsored by the Bureau of Justice Statistics (BJS), collected by the
Census Bureau and are publicly available through the National Archive of Criminal Justice Data
(NACJD). The data are collected using a stratified, multistage cluster design (Hubble 1995;
Rennison and Rand 2007) and fielded at a sample of housing units and groups quarters in the
United States and the District of Columbia. All persons age 12 or older are interviewed every six
months for a three year period in each selected sample housing unit. The data are representative
of the non-institutionalized population of persons age 12 or older in the United States (Garofalo
1990; Rennison and Rand 2007).
The time period studied reflects all data available after a major series-breaking redesign
implemented with a split sample design in 1992. The changes were numerous, including
modifying the survey instruments (e.g., screener questions, new crimes added, improved cues),
changes in protocols regarding measurement of certain types of victimization (e.g., series
victimizations), and cost-saving changes. The net effect of the redesign was to improve the
survey's ability to measure victimization. Since post-redesign data were collected using a
different screening strategy, post-redesign data are not comparable to data gathered earlier
(Rennison and Rand 2007). The NCVS sample is characterized by relatively high response rates.
On average each year during the 1993 to 2011 period approximately 84,000 households and
about 153,000 persons were interviewed twice each year for the survey. Response rates during
this period averaged to 93 percent for households, and about 88 percent for individuals.4
The data used in these analyses are based on “collection year data,” a format of the
NCVS that have not been readily available to the public in the past. Collection year data also
known as reference year data or calendar year data includes information on criminal
victimizations based on the year in which the interview took place. That is, if one used the 2007
collection year data file, it would contain information on victimizations that were collected
during 2007 though the actual victimizations may have occurred in 2006 given the six month
reference period. The second format used for NCVS data historically is “data year data.” Data
year data is the format that has been made available to the public in past years. Data year data
include information on criminal victimizations based on the year in which the victimization took
4 Readers interested in additional information about the NCVS are directed to Rennison and Rand (2007).
place regardless of when the interview occurred. Thus, if one had the 2007 data year data file, all
victimizations included in that file occurred in 2007. Because of the six month reference period
used in the NCVS, this meant the data took longer to release to the public. Given that our work
aggregates all full year post redesign NCVS data, and that the year in which the victimization
took place is not used in the analysis, the use of collection year data should have little to no
substantive influence on the findings presented.
A second way these analyses differ from much extant research using the NCVS is the
way that series victimizations are treated. At times, respondents sustain a series of victimizations
that are so similar in nature or occur so often that respondents are unable to provide details of
each individual event. Because in the NCVS details of the victimizations are required to classify
an event as a crime or non-crime, and to classify crimes into types of crime, this presents special
problems. The literature offers a few ways to deal with this difficult problem (see e.g., Rand and
Rennison 2005). In the past, BJS did not include series victimizations in annual estimates of
victimization, and counted each series as one victimization in other reports. Extant research
published outside of BJS using the NCVS generally counted a series of victimizations as a single
victimization. This approach is widely understood to have undercounted some types of
victimizations. In 2012 BJS changed their NCVS series victimization protocol (see Lauritsen et
al. 2012). It now enumerates series victimizations based on the victim’s estimate of the number
of times the victimizations occurred during the six month reference period, with a maximum of
10 victimizations per interview. The present analyses follows the lead of BJS and counts series
victimizations based on the victim’s estimate of the number of times the violence occurred up to
a maximum number of ten.
Advantages and Disadvantages of the NCVS
NCVS data offer several advantages for the present robbery study. First, NCVS data include an
extensive range of incident-level characteristics of robbery such as weapon presence, use of
force, and victim’s resistance. Second, NCVS data include all robbery victimizations – not just
those reported to the police. Third, studies exploring robbery and gender have been limited to
small non-probability samples that primarily focused on urban, poor, incarcerated, and/or
African American respondents (e.g., Anderson 1999; Brookman et al. 2007; Mullins 2006;
Oliver 1994; Wright and Decker 1997). Further, much of this work focuses on the offender’s
perspective. The NCVS offers the opportunity to examine the relationship between gender and
robbery from the victim’s perspective using a nationally representative sample of the non-
institutionalized population of persons age 12 or older in the United States. Finally, while gender
affects all social interactions, including the interaction between interviewer and respondent, the
NCVS offers a unique advantage in the study of gender on the accomplishment of robbery in that
it relies upon the victim for information on the criminal incident. In particular, female victims are
not likely to feel the same social pressures in their discussion of male perpetrated robbery as
might be the case for male offenders (e.g., Brookman et al. 2007). For instance, Cook’s (1987)
findings regarding the use of force by male offenders on female victims in the National Crime
Survey was decidedly different than was discussed by respondents in Miller’s (1998, 2000)
study. Data in the NCVS, therefore, provide another vantage point from which to examine male-
NCVS data are limited in certain ways. First, information on victims younger than age 12
is excluded from the data. Second, the NCVS is designed to provide national-level information
on victimization and is unable to provide insight into smaller geographic regions such as states,
counties or cities. Thus, our study cannot provide for a direct test of previous work on the role of
gender on the accomplishment of robbery in particular locations. Third, the NCVS only provides
data on violence against individuals and do not include robberies committed against businesses.
Fourth, because the data are collected directly from victims, they cannot include robberies that
resulted in the victim’s death. Lastly, while the NCVS might provide certain advantages in
studying male-on-female robbery incidents, these same gender norms might impinge on the
reliability of information collected on incidents involving female offenders and male victims.
That is, norms of masculinity might prevent the full disclosure of incident-level characteristics
involving some male victims of female perpetrated robbery. For instance, Stanko and Hobdell’s
(1993: 413) research on male responses to victimization suggests that norms of masculinity lead
men to view victimization as a sign of weakness, and thus men who are victimized are “weak
and helpless.” Despite these limitations, the NCVS provides an excellent opportunity to examine
the hypotheses suggested by extant literature regarding the gendered nature of robbery.
To examine the role of gender in robbery, this study uses attempted and completed robberies.
Robbery victimizations are identified using the standard NCVS definition: “a completed or
attempted theft, directly from a person, of property or cash by force or threat of force, with or
without a weapon, and with or without injury” (Bureau of Justice Statistics 2010). Restricting the
data to attempted and completed robberies results in an unweighted sample size of 4,660 robbery
victimizations used in the analysis.
The analyses focus on features of incidents directly related to the interaction between offender
and victim in the accomplishment of robbery, including features that are the result of offender
decision making. One of the major characteristics of a robbery an offender controls is the
victim’s gender. Miller’s (1998) research makes clear the important role that victim’s gender
plays in a robber’s offense. This variable is investigated using two categories of victim’s gender:
male (coded as 1) and female (coded as 0).
To account for offender’s gender, several categories are used that distinguish between
male(s), female(s), a group of males and females, don’t know offender’s gender, and missing
data. Male(s) serves as the reference category. Aside from the offender’s gender, we utilize
additional measures which reflect offender decision making and the interaction of offender and
victim. These additional situational characteristics include: the number of offenders, whether the
robbery was attempted or completed, weapon presence and type, physical force used by the
offender, victim resistance, injury to the victim, and the place of the robbery. For detailed
information on variables used in the analyses, please refer to the Appendix.
Multiple strategies are used to investigate this topic, including estimating basic descriptive
statistics such as percentages, a comparison of estimates to ascertain differences, and finally a
logistic regression to identify those predictors of robbery that are associated with either a male or
female victim. First, the NCVS data used are weighted to account for non-response and other
aspects of the data collection. Weights are found on the data file. Second, comparisons of
estimates calculated from the NCVS account for the complex sample design. Because the NCVS
data are not gathered using a simple random sample, estimation techniques that assume simple
random sampling underestimate the standard errors and result in incorrect inferences about
statistical significance. Therefore all comparisons of estimates using NCVS data are tested using
formulae that take into account the complex sample of the NCVS. The comparisons of estimates
presented in the contingency table use generalized variance function constant parameters to
calculate variance estimates, standard errors, and confidence intervals. And finally, the NCVS
binary logistic regression analyses presented below use Taylor series linearization to account for
the complex sample design used in the NCVS and allow for accurate significance testing of the
coefficients (Levy and Lemeshow 1999). One disadvantage with adjusting for the complex
sampling is that traditional goodness-of-fit diagnostics cannot be estimated for these models
(Hosmer and Lemeshow 2000).
Before presenting results of our analysis, Table 1 offers a description of the sample of robberies
used in the analysis. Most robberies are committed against males (62%), by males (80%), and
have a lone offender (53%). Most victims live in urban areas (51%), most take place on the street
(42%), and tend to involve an armed offender (55%), an offender using non-proactive physical
force (54%), and a resisting victim (63%).
TABLE 1 ABOUT HERE
Gender and Robbery
A simple contingency table (see Table 2) shows that robbery tends to be an intra-sexual form of
violent crime. From 1993 to 2011, 65% of robbery incidents involved a male robber and a male
victim. Still 35% of victims robbed by males were female. Sixty-two percent of female
perpetrated robbery victims were female, while 38% of these victimizations involved male
victims. When a group of males and females worked together to commit a robbery, male and
female victims were equally selected (50% and 50%, respectively). A clear pattern emerges with
respect to gender and missing data, where a large majority of cases where the victim either does
not know the gender of the offender (69%) or does not report this attribute (82%) result from
interviews with male victims, although the overall prevalence of such missing data is limited
(Don't know = 3%; Missing data = 2%, see Table 1).
--TABLE 2 ABOUT HERE--
Factors Influencing the Selection of a Male or Female Robbery Victim
A question derived from the literature focuses on whether or not situational characteristics
influence the probability male or female victims are selected for a robbery. To begin to address
this question, we present a binary logistic regression that includes victim’s gender as the
dependent variable and offender’s gender as the independent variable.5
TABLE 3 ABOUT HERE
regression indicates a significant relationship between offender and victim gender. Specifically,
female robbery offenders are 67 percent less likely than male robbers to select a male as a victim
(AOR=0.33). This model also shows that a mixed group of robbers is less likely than male
robbers to target a male victim (AOR=0.55).
Our next step is to include the relevant correlates presented in the literature to determine
if this basic relationship maintains, or if controls for situational elements lead to relative parity in
the willingness of male and female robbers to select victims of the opposite sex. Findings in
Table 4 utilizes the full sample of robbery victimizations (i.e., all victims and all offenders) to
5 While initially the use of victim’s gender as the “dependent” variable appears unusual, it is an appropriate
strategy for our purposes. This method follows approaches used in fields such as taxonomy and biology, and
increasingly used in the criminology literature (see e.g., Swat and He 2006). Like that work, our research does not
treat victim’s gender as a dependent variable in the classic—causal—sense, but rather as a measure for the selection
of a victim based on their gender which is in line with the extant qualitative research on urban robbery (see Harrell
2001 for more information).
ascertain the factors associated with the victimization of male versus female victims, and
whether or not controlling for the situational factors associated with the accomplishment of
robbery can help explain the overall gendered pattern of robbery perpetration discussed above.
After controlling for the situational correlates of robbery, offender sex continues to be
associated with the gender of the targeted victim. In the base model female offenders were 67
percent less likely to target male victims than are male offenders. Once other relevant factors are
included in the model, the odds ratio was reduced to 46 percent. What factors mitigate the
influence of offender gender on the selection of a male versus a female victim? Findings show
that offenses utilizing more than one offender (AOR= 2.29), a firearm (AOR = 1.58), knife
(AOR = 1.58) or some “other” type of weapon (AOR = 2.32) (compared to no weapon) are more
likely to target a male victim. In addition, results show that non-proactive force by the offender
(versus no force) (AOR = 1.58), robberies that take place on the street (AOR = 2.57), in or near a
friend’s home (AOR = 1.81) or some other place (AOR = 2.63), (compared to in/near the
victim’s home) are significantly more likely to include male victims. In contrast, a completed
(versus attempted) robbery (AOR = 0.58), one in which proactive physical force is used (versus
no force) (AOR = 0.74), and one that occurs to a suburbanite (versus an urban dweller) (AOR =
0.94) is more likely to involve a female victim.
An interesting finding in Table 4 is that male victims continue to be significantly more
likely not to respond to the question about the gender of the offender (AOR = 3.16), even after
relevant situational controls are included. More specifically, it is more than three times as likely
for a male victim to not disclose the sex of the person(s) who robbed them than a female victim.
The reason the victim selects not to disclose this information is not clear in the data.
TABLE 4 ABOUT HERE
The previous findings suggested that after accounting for situational characteristics of robbery
incidents the offender’s gender continues to be significantly associated with the sex of the
victim. While an important finding, it cannot inform regarding whether the situational correlates
differed based on the gender of the offender. The next series of models, presented in Table 5,
address if and how situational factors associated with victim gender differ among male, female,
and a mixed-group of robbery offenders.
Factors Influencing the Selection of a Male or Female Robbery Victim by Male Offenders
Turning first to male offenders, the regression results displayed in Panel A of Table 5 shows
several significant associations. Specifically, when two or more males (compared to a lone male
offender) commit a robbery, the victim is more than three times likely to be a male (AOR=3.28)
than a female. Males are significantly more likely to complete a robbery when the victim is a
female (AOR = 0.53). When a male offender uses a firearm (AOR=1.42), a knife (AOR = 1.87),
or some “other” type of weapon (AOR=2.53), the odds are greater that the victim is a male.
Findings indicate that males are significantly more likely to use proactive physical force (versus
no force) when robbing a male victim (AOR = 1.64). Finally, robberies committed by males
away from the victim’s home were more likely to be committed against a male (AOR ranges
from 3.30 to 4.93).
TABLE 5 ABOUT HERE
Factors Influencing the Selection of a Male or Female Robbery Victim by Female
Panel B of Table 5 offers findings for female perpetrated robberies. Findings indicate the
importance the number of offenders, and the location of the robbery only. Results show that
robbery committed by two or more female offenders (versus a single female offender) is
significantly more likely to target a female versus a male victim (AOR = 0.26). A second finding
is that when a female commits a robbery away from the victim’s home, the odds are greater that
the victim will be a female versus a male (AORs range from 0.41 to 0.61).
Factors Influencing the Selection of a Male or Female Robbery Victim, by a Group of
Male and Female Offenders
Results from this final regression presented in Panel C indicate three significant associations.
First, when a firearm is used in this form of robbery, the victim is far more likely to be a male
(AOR= 4.82). That is, when a group of males and females commit robbery, they are almost five
times more likely to brandish a gun when robbing a male versus a female. Second, when a group
of males and females commit a robbery in which the victim resists, odds are increased that the
victim will be a female (AOR = 0.34). And finally, when a group of robbers commit this crime in
some “other” place, odds are increased that the victim will be a male (AOR = 4.68).
SUMMARY AND DISCUSSION
Males represent an overwhelming majority of robbery offenders. Females, however, are arrested
for a non-trivial amount of robberies on a yearly basis (Federal Bureau of Investigation 2012).
This study examined robbery contexts using a nationally representative sample of non-
institutionalized persons age 12 or older, and was informed by findings from interview data with
active (Miller 1998, 2000) and institutionalized (Brookman et al. 2007; Sommers and Baskin
1993) male and female offenders. Based on the idea that gender stratification shapes the
decision-making process of both male and female offenders and victims, we expected to find
several differences in the manner in which these groups accomplished robbery. This research
offers support for several of our hypotheses drawn from the research literature; there were a
number of interesting departures, however (see Figure 2).
FIGURE 2 ABOUT HERE
Support for a number of our hypotheses, which drew heavily from Miller's (2000)
discussion of the research literature about the gendered nature of robbery, emerged. First,
robbery is predominately an intra-sexual criminal incident, as males rob other males 65 percent
of the time they commit robbery, while females target female victims in 62 percent of incidents.
This is despite the fact that male robbery offenders suggest that female victims are easy targets
because they are especially unlikely to resist during the incident. Such a view of passivity in
criminal incidents might suggest females should be a more frequent target for robbery
perpetration, but research suggests both practical and social reasons why this is not the case. For
practical purposes, male offenders have suggested that females are likely to carry less cash, and
thus robbing them is less profitable. Perhaps more importantly, there is a social stigma associated
with males using force against females, and thus males receive no status enhancement for
engaging in such behaviors. In fact, doing so may lose social standing for males. Current
findings, shown in Table 5 Panel A, support the view that social norms are a powerful deterrent
against victimizing females among male robbery offenders. Specifically, while male offenders
prefer to rob male victims overall, when male offenders rob as a group and thus are subject to
immediate judgment and reinforcement of masculine norms from their peer group members,
male offenders are more than three times as likely to rob a male victim than a lone male
offender. The fact that male offenders are willing to rob a female victim, but substantially less
likely to do so in the company of their peers, may be viewed as a compelling example of how
gender norms shape criminal behavior.
Consistent with previous research, males are statistically more likely to use guns when
committing robbery (25%) than are females (2%) (15% of robbery committed by males and
females acting together involve a firearm). Interestingly, the sex differences in the rate by which
males and females utilize firearms in robbery incidents far exceeds the differences in population
level estimates of overall patterns of gun ownership in the United States, where 45 percent of
males own a weapon compared to 15 percent of women (Jones 2013). Although it was expected
that female robbers would utilize firearms more frequently when robbing males than when
robbing females, female offenders are equally as unlikely to use a gun when robbing males (2%)
or females (2%). Contrary to expectations, supplementary analyses suggest the sex of the victim
does influence the use of a weapon by female offenders as they are less likely to use any weapon
when robbing a male (62%) compared to a female (68%). This study suggests that Miller’s
(2000: 38) conclusion that when robbing males “the woman always used a weapon” is not
generalizable to other areas throughout the United States. While this disparity is likely due to
site-specific idiosyncrasies, it is clear that, more generally, women rarely use a firearm in the
commission of robbery, and gender of the victim does not appear to substantially impact the
decision to use weapons, overall.
Consistent with expectations, males used proactive physical force during some robbery
victimizations (10%). Proactive physical force was used by males against male victims
significantly more often (12%) than during male-on-female robbery (7%). While males in
Miller’s (2000) study reported that males never used proactive violence when robbing females,
the present findings suggest that males used preemptive physical force in a non-negligible
percent of robbery victimizations. This finding is consistent with the assertions of Anderson and
Umberson (2001: 361) who find that men routinely “minimize” their descriptions of violence
used against women, and is consistent with research that suggests that violent male offenders
work to convey a “decent” (Sandberg 2009) or “moral” (Presser 2004) identity in the interview
The tendency for males to minimize their accounts of robbing females is consistent with
the notion of situational accountability (Fenstermaker, West and Zimmerman 1991;
Messerschmidt 1993), in that even respondents in research interviews might “realize that others
hold them accountable to behavioral dictates consistent with immediate circumstances” (Copes
and Hochstetler 2003: 283).6
Inconsistent with expectations, females were equally likely to use physical force when
robbing males as they were when robbing females, which suggest that gender alone is not the
best predictor of whether or not offenders use physical force during the commission of robbery.
Future research on male offenders who rob females as part of their
criminal repertoire should examine the inconsistency in adherence to cultural norms regarding
the use of violence against women. That is, while these individuals appear able to disregard or
neutralize cultural norms against the use of violence against women in the context of robbery,
they are keenly aware of, and abide by, these same cultural norms while discussing their exploits.
Unfortunately, as found in Brookman and colleagues’ (2007) study of gender and the
accomplishment of robbery, men may not be particularly willing to discuss these incidents, and if
they do they may present themselves in a manner that does not accurately reflect their actual
behavior. Therefore, particular care and attention may be needed in discussing these issues in a
way that allows for meaningful discourse on the influence of gender across contexts.
6 Copes and Hochstetler (2003:299) continued “[t]he disreputable young men who occupy the bottom of the
socioeconomic ladder and are responsible for most street crime probably do not have the same conventions and
attitudes toward women as middle-class college professors.” We agree with this general statement, and also believe
that such respondents are keenly aware of this. Further it is our contention that interviews with male offenders are
likely systematically hampered by the pressure of these individuals to adhere to more conventional norms of
masculinity, which frowns upon the use of violence against women; a classic case of social desirability bias (Groves
et al. 2004).
This may indicate that females target males who appear particularly weak or vulnerable, whereas
the same caution in target selection is not taken when robbing females, consistent with the
assertions of Brookman and colleagues (2007) and Baskin and Sommers (1998). This remains
speculative however, as data on target vulnerability is unavailable in the NCVS.
While it is commonly believed that female victims pose little threat of resistance during
robbery attempts, NCVS data suggest that female victims regularly resist robbery attempts by
male or female perpetrators. Further, findings show that females resist in equal percentages
against male (63%) and female (62%) offenders. Again, this result may be indicative of more
careful target selection processes for male victims as suggested by Brookman et al. (2007) and
Baskin and Sommers (1998). Perhaps offenders are more careless when robbing women, leaving
more opportunity for active resistance by the victim in such circumstances. The fact that women
are no more or less likely to be injured by male or female assailant(s) suggests victim-offender
interchanges during robbery incidents, regardless of the gender composition of the respective
parties, may be more similar than suggested by offenders. The views shared by offenders in
previous research may stem from common stereotypes related to femininity, whereby women are
viewed as more passive than males. More research on the situational dynamics that lead to victim
resistance during robberies is needed in order to understand these exchanges more completely.
Better understanding of how gender shapes the enactment of robbery may be gleaned
from the rape literature. Burgess and Holmstrom (1974) identified three primary approaches to
rape based on the victim/offender relationship: “confidence rape,” “capturing the victim” and
“blitz rape.” An offender with an existing intimate relationship of the victim is most likely to
engage in a “confidence rape” in which they use the existing relationship to gain access to the
victim. In “capturing the victim” the offender engages in interactions such as simple
conversation or manipulating the situation in their favor, and uses verbal means rather than
physical force to gain access to the victim. And third, a “blitz rape” occurs without any prior
interaction between the victim and offender, and it involves weapons, force or the threat of force
to subdue the victim. Findings based on robbery and gender suggest that female robbers may be
more inclined to engage in robbery by means of capturing the victim using verbal means. In
contrast, male robbers appear more likely than female robbers to engage in a blitz robbery.
In general, these results suggest the importance of considering gender when examining
robbery without overemphasizing its effects on interpersonal behavior. We find no evidence that
“women commit crimes like men today” (Muraskin 2005:41). For example, female robbers
rarely use firearms (or weapons in general) while males do in large proportions. Yet, even
without firearms or weapons, female robbers gain victim cooperation in contrast to previous
suggestions (Brookman, et al. 2007). On the other hand, we also find no evidence for the notion
that differences in robbery offending are due to females being weaker or less courageous than
their male counterparts. Results indicate that, in contrast to expectations of a weak or cowardly
offender, female robbers are not afraid to use force against their victims.
These findings highlight the need to focus on the situational context of robbery, as gender
alone is not a sufficient predictor of robbery related behaviors. Influences aside from gender, and
likely the interactions among several variables with gender, appear to be operating. It may be
that the role of gender is not constant across situational context. From a law enforcement
standpoint, this research suggests that investigations into robbery incidents should not rely on
stereotypes related to gender norms when compiling leads on possible offenders. While male
offenders commit the majority of offenses, female offenders commit robbery, and do so against
male victims in roughly one-third of their offenses. Future research should closely examine the
myriad combinations of situational contexts associated with robbery in an effort to untangle the
complex influence of gender on such incidents.
This research has a number of limitations that should be kept in mind when interpreting
the results. First, the NCVS does not gather information on the motivation(s) of the offender(s).
It could be that using a nationally representative sample of robbery would reveal that motivations
may differ by offender’s gender. If that were the case, additional insight into the gendered nature
of robbery may be revealed. Second, there may be limits to the generalizability of the findings
with respect to race as it is not a part of the analysis. The purpose of this research was to examine
the situational characteristics of male and female robbery, and in particular to determine the
generalizability of prior research on gender and the accomplishment of robbery. Prior research
on female involvement in robbery in the United States has largely not considered the role of race,
and thus the importance of this characteristic remains unclear. Future research should explore the
potential moderating effects of race (if any) on the association between gender and the
accomplishment of robbery.
Finally, while one of the goals of our study was to overcome problems associated with
studying the impact of gender on the accomplishment of robbery solely through interviews with
robbery offenders, because the interview itself is impacted by these very same norms, we found
evidence to suggest that victimization studies with male victims may also be influenced by
gender norms in a way that impacts the validity of the data. Specifically, male victims of robbery
were substantially more likely to not report the offender's sex than female victims. While missing
data on this item was not a substantial problem in the NCVS, in 68.5 percent of cases where the
offender's gender was unknown, the victim was male. When the data on offender sex was simply
missing, the respondent was male 82.3 percent of the time. Interestingly, there were no
differences between male and female victims in their capability or willingness to report offender
race or age in the NCVS, meaning significant differences in missing data related to offender
demographics was limited to offender gender. Thus, while qualitative interviews may produce
biased results with respect to interpersonal behavior between men and women, similar processes
may hamper the collection of quantitative survey data. Future research may seek to compare the
influence of gender norms on the collection of qualitative and quantitative data in situations
where reports of behavior may violate common gender codes.
Anderson, Elijah. 1999. Code of the Street: Decency, Violence and the Moral Life of the Inner
City. New York: W.W. Norton.
Anderson, Kristen L. and Debra Umberson. 2001. “Gendering Violence: Masculinity and Power
in Men’s Accounts of Domestic Violence.” Gender & Society 15:358-380.
Baskin, Deborah and Ira Sommers. 1998. Casualties of Community Disorder: Women’s Careers
in Violent Crime. Boulder, CO: Westview.
Baskin, Deborah, Ira Sommers and Jeffrey Fagan. 1993. “The Political Economy of Female
Violent Street Crime.” Fordham Urban Law Journal 20:401-417.
Block, Richard. 1981. “Victim Offender Dynamics in Violent Crime.” Journal of Criminal Law
and Criminology 72(2):743-761.
Braithwaite, John and Kathleen Daly. 1994. “Masculinities, Violence and Communitarian
Control.” Pp. 189-213 in Just Boys Doing Business, edited by T. Newburn and B. Stanko.
Brookman, Fiona, Christopher Mullins, Trevor Bennett and Richard Wright. 2007. “Gender,
Motivation and the Accomplishment Of Street Robbery In The United Kingdom.” British
Journal of Criminology 47:861-884.
Bureau of Justice Statistics. 2006. Criminal Victimization in the United States – Statistical
Tables. http://bjs.ojp.usdoj.gov/content/pub/pdf/cvus/cvus05mt.pdf. retrieved October 15,
Burgess, Ann and Lynda Holmstrom. 1974. Rape: Victims of Crisis. University of Michigan:
R.J. Brady Co.
Cook, Philip. 1987. “Robbery Violence.” Journal of Criminal Law & Criminology 70(2):357-
Copes, Heith and Andy Hochstetler. 2003. “Situational Construction of Masculinity Among
Male Street Thieves.” Journal of Contemporary Ethnography 32(3):279-304.
Decker, Scott, Richard Wright, Allison Redfern, and Dietrich Smith. 1993. “A Woman’s Place is
in the Home: Females and Residential Burglary.” Justice Quarterly 10:143-162.
Federal Bureau of Investigation. (2012). Crime in the United States 2011.
Retrieved June 27, 2013.
Felson, Richard B. 1993. “Predatory and Dispute-Related Violence: A Social Interactionist
Approach.” Advances in Criminological Theory 5:189-235.
Felson, Richard B., Eric P. Baumer and Steve F. Messner. 2000. “Acquaintance Robbery.”
Journal of Research in Crime and Delinquency 37(3):284-305.
Fenstermaker, Sarah, Candace West, and Don H. Zimmerman.1991. “Gender inequality: New
conceptual terrain.” Pp. 298-307 in Gender, Family and Economy, edited by R.L.
Blumberg. Newbury Park, CA: Sage.
Garofalo, James. 1990. “The National Crime Survey, 1973-1986: Strengths and Limitations of a
Very Large Data Set.” Pp 75-97 in Measuring Crime: Large-Scale, Long-Range Efforts,
edited by D.L. MacKenzie, P.J. Baunach and R.R. Roberg. Albany: State University of
New York Press.
Groves, Robert M., Floyd J. Fowler, Jr., Mick P. Couper, James M. Lepkowski, Eleanor Singer,
and Roger Tourangeau. 2004. Survey Methodology. Hoboken, NJ: John Wiley & Sons.
Hochstetler, Andy, Heith Copes and J. Patrick Williams. 2010. “"That's not who I am:" How
offenders commit violent acts and reject authentically violent selves.” Justice Quarterly
Hubble, David L. 1995. The National Crime Victimization Survey Redesign: New Questionnaire
and Procedures Development and Phase-In Methodology. Paper presented at the Annual
Meetings of the American Statistical Association, Orlando, FL.
Jacobs, Bruce A. and Jody Miller. 1998. “Crack Dealing, Gender and Arrest Avoidance.” Social
Jones, J.M. 2013. Men, Married, Southerners Most Likely to Be Gun Owners. Gallup 2013 [cited
June 27 2013]. Available from http://www.gallup.com/poll/160223/men-married-
Kandiyoti, Deniz. 1988. “Bargaining with Patriarchy.” Gender & Society 2:274-290.
Lorber, Judith. 1994. Paradoxes of Gender. New Haven: Yale University Press.
Luckenbill, David. 1981. "Generating Compliance: The Case of Robbery." Urban Life 10:25-46.
Lynch, Michael. 1991. “Method: Measurement--Ordinary and Scientific Measurement as
Ethnomethodological Phenomena.” Pp. 77-108 in Ethnomethodology and the Human
Sciences, edited by G. Button. Cambridge, U.K.: Cambridge University Press.
Maher, Lisa. 1997. Sexed Work: Gender, Race, and Resistance in a Brooklyn Drug Economy.
Oxford, UK: Clarendon.
Muraskin, Roslyn. 2005. “Feminist Theories: Are they Needed?” Pp 31-43 in It’s A Crime:
Women and Justice, 4th Edition, edited by R. Muraskin. Upper Saddle River, NY:
Messerschmidt, James W. 1993. Masculinities and Crime. Lanham, MD: Rowman & Littlefield.
Miller, Jody. 2011. "Grounding the Analysis Of Gender And Crime: Accomplishing and
Interpreting Qualitative Interview Research." Pp. 49-62 in The Sage Handbook of
Criminological Research Methods, edited by D. Gadd, S. Karstedt, and S. F. Messner.
London: Sage Publications.
--------. 2000. “Feminist Theories of Women‘s Crime: Robbery as a Case Study.” Pp. 25-46 in Of
Crime & Criminality: The Use of Theory in Everyday Life edited by Sally S. Simpson.
Thousand Oaks, CA: Pine Forge Press.
--------. 1998. “Up It Up: Gender and the Accomplishment of Street Robbery.” Criminology
Mullins, Christopher. 2006. Holding Your Square: Masculinities, Streetlife and Violence.
Oliver, William. 1994. The Violent Social World of the Black Male. New York: Lexington
Presser, Lois. 2004. “Violent Offenders, Moral Selves: Constructing Identities and Accounts in
the Research Interview.” Social Problems 51:82-101.
--------. 2009. “The Narratives of Offenders.” Theoretical Criminology 13:177-200.
Rennison, Callie M. and Michael R. Rand. 2007. “Introduction to the National Crime
Victimization Survey.” Pp. 17-54 in Understanding Crime Statistics: Revisiting the
Divergence of the NCVS and the UCR, edited by J.P. Lynch, J.P. and L.A. Addington.
Cambridge, UK: Cambridge University Press.
Sandberg, Sveinung. 2009. “A Narrative Search for Respect.” Deviant Behavior 30:487-510.
Sommers, Ira and Debra R. Baskin. 1993. “The Situational Context of Violent Female
Offending.” Journal of Research in Crime and Delinquency 20(2):136-162.
Stanko, Elizabeth A. and Kathy Hobdell. 1993. “Assault on Men: Masculinity and
Victimization.” British Journal of Criminology, 33:400-415.
Steffensmeier, Darrell, Jennifer Schwartz, Hua Zhong and Jeff Ackerman. 2006. “An assessment
of recent trends in girls’ violence using diverse longitudinal sources: Is the gender gap
closing?” Criminology 43(2): 355-406.
Swatt, Marc L. and Ni “Phil” He. 2006. "Exploring the difference between male and female
intimate partner homicides: Revisiting the concept of situated transactions." Homicide
West, Candace and Sarah Fenstermaker. 1995. “Doing Difference.” Gender & Society 9(1):8-37.
West, Candace and Don H. Zimmerman. 1987. “Doing Gender.” Gender & Society 1(2):125-
Wright, Richard T. and Scott H. Decker. 1997. Armed Robbers in Action. Boston, MA:
Northeastern University Press.
Figure 1. Hypotheses Regarding the Influence of Gender on the Accomplishment of
Male Street Robbery
H1: Male offenders are significantly more likely to target male versus female
H2: Males are significantly more likely to use guns during the commission of robbery than
H3: Males are more likely to use proactive physical force against male compared to female
H4: Females are less likely to resist than male victims in male perpetrated robbery.
Female Street Robbery
H5: Female robbers are significantly more likely to target female versus male
H6: Female robbers are significantly more likely to use physical force to commit robbery
female victims than against male
H7: Females are significantly more likely to use a gun when robbing male versus female
H8: Females are significantly less likely to rob males in a street setting than
Mixed Gender Street
H9: When there is a group of male and female offenders, males are significantly more likely to
targeted than females.
H10: Mixed gender groups of robbery offenders are significantly more likely to use a firearm
against male victims.
Figure 2. Level of Support for Hypotheses Regarding Gender and the Accomplishment of
Male Street Robbery
H1: Male offenders are significantly more likely to target
male versus female victims. (Tables 2, 3, and 4)
H2: Males are significantly more likely to use guns during
the commission of robbery than females. (Supplementary) X
H3: Males are more likely to use proactive physical force
against male compared to female victims. (Table 5; Panel
H4: Females are less likely to resist than male victims in
male perpetrated robbery. (Table 5: Panel A) X
Female Street Robbery
H5: Female robbers are significantly more likely to target
female versus male victims. (Tables 2, 3, and 4)
H6: Female robbers are significantly more likely to use
physical force to commit robbery against female victims
than against male victims. (Table 5; Panel B)
H7: Females are significantly more likely to use a gun
when robbing male versus female victims. (Table 5; Panel
H8: Females are significantly less likely to rob males in a
street setting than females. (Table 5; Panel B)
Mixed Gender Street Robbery
H9: When there is a group of male and female offenders,
males are significantly more likely to be targeted than
females. (Table 2)
H10: Mixed gender groups of robbery offenders are
significantly more likely to use a firearm against male
victims. (Table 5; Panel C)
note: "Supplementary" denotes that additional analyses, not in table format, confirmed this
Table 1. Variable Descriptions, 1993-2011 NCVS, unweighted n= 4,660
Variable/Measure Percent Variable/Measure Percent
Total Robbery 100.0 Weapon presence
No Weapon 44.6
Gender of victim Weapon 55.4
Female 38.5 Firearm 22.5
Male 61.5 Knife/cutting implement 12.2
Other/Unknown Type Weapon 10.5
Gender of offender(s) Unknown if Offender Had Weapon 10.2
Female 10.3 Physical force used by offender
Both 5.0 No physical force 36.9
Don't know 2.9 Non-proactive physical force 53.5
Missing data 1.5 Proactive physical force 9.6
Number of offenders Victim resistance
One offender 57.3 No resistance 36.3
Two or more offenders 40.0 Resistance 63.3
Don't know 1.6 Don't know 0.4
Missing data 1.5
Completed/Attempted Robbery No injury 66.6
Attempted 32.2 Minor injury 27.0
Completed 67.8 Serious injury 6.2
Missing data 0.2
Urban 50.5 Place
Suburban 39.6 In/near victim's home 29.9
Rural 9.9 On street 41.5
In/near friend's home 6.0
Commercial place, school, field 22.6
Table 2. Percent, Victim's gender by offender's gender, 1993-2011 NCVS
Don't know 31.5 68.5 100.0
Total 38.5 61.5 100.0
Unweighted n = 4,660
Table 3. Binary logistic regression model predicting robbery victim's gender, by
offender gender, 1993-2011 NCVS data
Male (reference category) --- ---
Female -1.09 0.19 * 0.00 0.33
Mixed group -0.61 0.22 * 0.01 0.55
Don't know gender 0.18 0.19
Missing data 0.93 0.34 * 0.01 2.54
Constant 0.60 0.05 * 0.00
Dependent Variable: Male = 1; Female = 0
Table 4. Binary logistic regression model predicting robbery victim's
gender,1993-2011 NCVS data
Variables b SE
Male (reference category) --- ---
Female -0.62 0.25 * 0.01 0.54
Mixed group -0.90 0.23 * 0.00 0.41
Don't know gender 0.08 0.21
Missing data 1.15 0.35 * 0.00 3.16
Number of offenders
One offender --- ---
2 or more offenders 0.83 0.10 * 0.00 2.29
Completed robbery -0.55 0.11 * 0.00 0.58
No weapon --- ---
Firearm 0.46 0.14 * 0.00 1.58
Knife 0.46 0.18 * 0.01 1.58
Other type weapon 0.84 0.23 * 0.00 2.32
Unknown if a weapon was present 0.00 0.16
Offender's use of force
No physical force --- ---
Force, but not proactive -0.13 0.15
Proactive physical force 0.46 0.21 * 0.03 1.58
Victim resisted -0.31 0.12 * 0.01 0.74
No injury --- ---
Minor injury 0.01 0.13
Serious injury 0.14 0.20
In/near victim's home
On street 0.94 0.12 * 0.00 2.57
In/near friend's home 0.59 0.17 * 0.00 1.81
Commercial place, school, field 0.97 0.17 * 0.00 2.63
Urban --- ---
Suburban -0.24 0.11 * 0.03 0.78
Rural -0.07 0.18
Constant 0.09 0.20 0.64
* p<0.05; unweighted n=4,660
Dependent Variable: Male = 1; Female = 0
Panel A Panel B Panel C
Male Offenders Female Offenders Male and Female Offenders
Variables bSE p-value AOR bSE p-value AOR bSE p-value AOR
Number of offenders (reference = One offender)
2 or more offenders 1.19 0.11 *0.00 3.28 -1.36 0.46 *0.00 0.26 --- --- --- ---
Completed robbery -0.64 0.12 *0.00 0.53 -0.05 0.37 0.90 0.95 -0.57 0.51 0.26 0.56
Weapon presence (reference = No weapon)
Firearm 0.35 0.16 *0.03 1.42 0.71 0.94 0.45 2.04 1.57 0.54 *0.00 4.82
Knife 0.63 0.18 *0.00 1.87 0.70 0.61 0.26 2.01 0.30 0.56 0.60 1.34
Other type weapon 0.93 0.21 *0.00 2.53 0.70 0.69 0.31 2.01 0.05 0.58 0.94 1.05
Unknown if a weapon was pre
-0.01 0.18 0.98 0.99 -0.40 0.72 0.58 0.67 1.05 0.77 0.18 2.87
Offender's use of force (reference = No physical force)
Force, but not proactive -0.29 0.16 +0.07 0.75 0.31 0.48 0.51 1.37 0.07 0.48 0.88 1.08
Proactive physical force 0.49 0.24 *0.04 1.64 -0.56 0.73 0.44 0.57 1.13 0.75 0.13 3.08
Victim resisted -0.19 0.11 0.10 0.83 -0.54 0.39 0.16 0.58 -1.08 0.46 *0.02 0.34
Victim's injury (reference = No injury)
Minor injury -0.03 0.12 0.81 0.97 -0.28 0.41 0.50 0.76 1.05 0.52 +0.05 2.87
Serious injury 0.26 0.23 0.26 1.29 -0.69 0.76 0.36 0.50 1.12 0.77 0.15 3.06
Place (reference = In/near victim's home)
On street 1.43 0.13 *0.00 4.16 -1.15 0.41 *0.01 0.32 0.24 0.43 0.57 1.28
In/near friend's home 1.19 0.20 *0.00 3.30 -1.35 0.60 *0.03 0.26 -0.96 1.16 0.41 0.38
Commercial place, school, f
1.60 0.17 *0.00 4.93 -1.53 0.61 *0.01 0.22 1.54 0.54 *0.01 4.68
MSA (reference = Urban)
Suburban -0.17 0.11 0.15 0.85 -0.12 0.41 0.77 0.89 -0.69 0.43 0.11 0.50
Rural 0.01 0.20 0.96 1.01 0.46 0.59 0.44 1.58 -0.63 0.55 0.25 0.53
Constant -0.36 0.19 +0.06 0.35 0.52 0.50 0.22 0.69 0.75
* p < 0.05; + p < .10
Dependent Variable: Male = 1; Female = 0
Table 5. Binary logistic regression model predicting robbery victim's gender, by offender's gender, 1993-2011 NCVS data
n = 3,859
F(16, 308) = 24.78
Prob > F = 0.0000
n = 387
F(16, 196) = 2.18
Prob > F = 0.0068
n = 207
F(15, 131) = 1.86
Prob > F = 0.0329
Variable Name Variable Description __
Robbery A completed or attempted theft of property or cash directly from a
person, by force or threat of force, with or without a weapon, and with
or without injury.
Victim’s gender Gender is the central concept of interest in this research. Victim’s
gender is measured using a single variable with two attributes: “male”
and “female”. In the regression analysis, female serves as the reference
group. There are no missing data on this variable in the NCVS.
Offender’s gender Five categories capture the concept of offender’s gender: male, female,
group of male and female offenders, offender’s gender unknown, and
missing data. Male indicates the robbery was committed by a single
male or groups of males. Female indicates the robbery was perpetrated
by a female acting alone, or a group of females. A group of males and
females indicates that there were both male and female offenders
engaged in committing the robbery. A total of 2.9% of the victims
could not identify the gender of the offender. And 1.5% of the sample
included victims who failed to respond to this question.
Number of offenders Research points to differences in robberies committed by multiple and
single offenders (see e.g., Miller 1998). To account for this, three
categories are used: “single offender,” “two or more offenders” and
“unknown number of offenders”. Victims were unable to identify the
number of offenders in 1.6% of robbery victimizations. And additional
1.5% did not respond to this question.
Completed/Attempted Robbery To account for the role of whether a robbery was completed
or attempted, a binary variable is utilized in the analysis: 0 =
attempted; 1 = completed.
Weapon presence The use and type of a weapon is a key component in extant literature
(see e.g., Miller 1998). In our analyses, weapons are accounted for
using a five category variable: “no weapon,” “firearm,” “knife/cutting
implement,” “other/unknown type of weapon,” and “don’t know if
offender had weapon”.
Physical Force used by Offender Several conclusions in extant research indicate the
importance of the use of force by the offender to gain compliance. To
account for this, a three-category variable was utilized: “no physical
force,” “non-proactive physical force” and “proactive physical force.”
When there was actual or attempted physical force by an offender
used, and when that force was used first by the offender, the variable
is categorized as “proactive physical force by the offender.” When the
offender hit, attacked or tried to attack the victim but the force was not
first used by the offender, the incident is characterized as one in which
the offender used “physical force.”
Victim resistance Research indicates the importance of considering the resistance of the
victim. To account for this in our analysis, three categories are
generated for use: no resistance, resistance and don’t know if there was
resistance. Resistance includes a wide range of behaviors from
screaming for help, trying to run away and attacked the offender.
Victim injury Research describes differences in victim injury given the gender of the
victim (see Miller, 1998). To assess the influence of this correlate, a
three-category variable is included in this work: “no injury,” “minor
injury,” “serious injury.” Respondents failed to provide information on
this measure for 0.2% of the victimizations recorded.
Location of robbery The literature indicates that male and female robbers prefer different
settings to commit their crime (see e.g., Miller, 1998). Specifically,
findings suggest that male robbers are more inclined to act in an
outdoor location while females are more inclined to commit a robbery
indoors. To test this, we utilized a four category variable: “in or near
the victim’s home” “on the street” “in or near a friend’s home” and
finally a “commercial place, school, field, public transportation”. The
final category is also described in this research as an “all other places”
or “other” places.
MSA A large proportion of research focused on robbery is limited to urban
settings. Our research includes robberies regardless of MSA. To
account for the potential role that MSA may offer, three categories are
utilized: urban, suburban and rural. The designation of these areas is
determined by the U.S. Office of Management and Budget (OMB).
OMB classifies geographic areas into three groupings based on their
relationship to a Metropolitan Area: central city, outside central city,
and nonmetropolitan area.7
7 See http://www.census.gov/geo/www/GARM/Ch13GARM.pdf, http://www.census.gov/population/
www/metroareas/metrodef.html, and http://www.census.gov/population/ and www/estimates/metroarea.html
for additional information.