Deconstructing Incidents of Female Perpetrated Sex Crimes

Abstract

Very little is known about co-offending by female sexual offenders (FSOs), especially in terms of diverse forms of offender groupings. To address this gap in the literature, this study uses 21 years (1992-2012) of National Incident-Based Reporting System data to analyze incidents of sexual offending committed by four female groupings: solo FSOs (n = 29,238), coed pairs consisting of one male and one FSO (n = 11,112), all-female groups (n = 2,669), and multiple perpetrator groups that consist of a combination of three or more FSOs and male sexual offenders (MSOs; n = 4,268). Using a multinomial logistic regression model, the data show significant differences in offender, victim, and crime context incident characteristics. The data also indicate that incidents with solo FSOs and all-female groups have similar characteristics, coed pairs and multiple perpetrator incidents have similar characteristics, and these two categorizations are fairly distinct from one another. Implications of this research are discussed in addition to directions for future research on female sexual offending.

Keywords

female sexual offending co-offending National Incident-Based Reporting System sex crimes multiple perpetrator groups

In the United States, female offenders perpetrate thousands of sexual assaults each year (Bierie & Davis-Siegel, 2015; Black et al., 2011; Williams & Bierie, 2015). Little is known, however, about offender, victim, and offense similarities and differences between females offending alone versus with others in terms of sexual assault incidents. Because there is the need to develop a gender-specific approach toward female sexual offenders (FSO; Cortoni, 2010; Cortoni & Gannon, 2013; Turchik, Hebenstreit, & Judson, 2016), research that brings about a better understanding of similarities and differences between types of female sexual offending can aid in working toward this goal. A better understanding of females who sexually offend alone versus with others can also assist law enforcement, policy makers, and, in particular, treatment providers as they respond to this social problem.

Evidence suggests the majority of FSOs and male sexual offenders (MSOs) offend alone; however, FSOs are far more likely than MSOs to commit sex crimes with a co-offender, primarily males (Denov, 2003; Gillespie et al., 2015; Grayston & De Luca, 1999; Lewis & Stanley, 2000; Nathan & Ward, 2002; Vandiver, 2006; Williams & Bierie, 2015). Although several studies have documented this fact, few to date have compared females who offend alone with those offending with a male. For the few studies that have specifically compared these two groupings, solo to co-offending FSOs, the sample sizes have been limited, ranging from 20 to 104 identified co-offenders, which were then compared with 12 to 123 identified solo FSOs (Gillespie et al., 2015; Muskens, Bogaerts, van Casteren, & Labrijn, 2011; Vandiver, 2006; Wijkman, Bijleveld, & Hendriks, 2010).

In addition, no empirical research has gone beyond these two FSO groupings, solo versus co-offending, to compare other groupings such as females who sexually offend with other females or females who sexually offend in larger groups. As a result, it is unclear how these more nuanced groupings of FSO incidents vary with respect to offender features (e.g., demographics), victim characteristics (e.g., age, race, gender, and relationship to offender), and crime details (e.g., injury and sexual assault behaviors). The field has recently recognized this dearth of research as particularly important to address because females sexually offending in other groupings are more common than initially realized (Horvath & Woodhams, 2013; Morgan, Brittain, & Welch, 2012; Williams & Bierie, 2015). Williams and Bierie (2015) found that a sizable minority of female co-offending involved groups of three or more offenders, and that almost 6% of the FSO incidents reported to police were committed by all-female groups. This is similar to Vandiver (2006) who reported all-female groups in approximately 3% of sex crime incidents resulting in arrest. Because a number of FSO incidents each year derive from group offending, a better understanding is needed across these various groupings to continue to advance knowledge about female sexual offending dynamics and to continue the development of a gender-specific approach toward FSOs.

Group Composition and Offender Behavior

Four studies to date have compared solo and co-offending FSOs and when taken together showed significant similarities and differences in sexual offending patterns, but there were also areas where evidence was mixed. One area of mixed evidence was in regard to the number of victims involved in the sexual assault. Vandiver (2006) and Wijkman et al. (2010) found that co-offenders were significantly more likely to have multiple victims, whereas Muskens et al. (2011) found no significant difference in the number of victims.

Pertaining to victim gender, solo FSOs were more likely to have male victims (Muskens et al., 2011; Vandiver, 2006; Wijkman et al., 2010), whereas co-offenders were more likely to have female victims (Vandiver, 2006; Wijkman et al., 2010). Vandiver (2006) and Wijkman et al. (2010) also found that co-offenders were more likely than solo FSOs to victimize both males and females during a sexual assault, although Muskens et al. (2011) found no statistical difference relating to this victimization pattern. When looking at the victim–offender relationship, co-offenders were more likely to be related to their victim compared with solo FSOs (Gillespie et al., 2015; Muskens et al., 2011; Vandiver, 2006; Wijkman et al., 2010).

Vandiver (2006) investigated additional characteristics of FSO sexual assault incidents. Her study is the largest and broadest study to date that investigated and compared 123 solo FSOs incidents with 104 co-offending incidents in the United States. In her research, incidents that involved solo FSOs and co-offending pairs had similar propensities to commit rape, but forcible sodomy was significantly more like to be committed in incidents with co-offending groups (Vandiver, 2006). Pertaining to crime characteristics, there were no significant differences in terms of the use of a weapon during the sexual assault incident or in terms of the location of the sexual assault incident (Vandiver, 2006).

In addition to these studies, typology and theoretical work on female sexual offending have further elaborated on the co-offending dynamic. Two types of female co-offenders have been identified in prior research: male-coerced FSOs who participate due to some type of force exhibited by the MSO and male-accompanied FSOs who participate willingly with the MSO (Gannon, Rose, & Ward, 2008; Mathews, Matthews, & Speltz, 1989, 1991; Nathan & Ward, 2002). It is theorized that FSOs who offend with men do so for various reasons, such as fear of their partner leaving them, fear of abuse, revenge, anger, or jealously toward the victim, or sexual motivations such as establishing and maintaining intimacy with their partner (Gannon et al., 2008; Gannon et al., 2014; Mathews et al., 1991). If females are part of the sexual assault strategy of male partners, victims and other characteristics of these incidents may more strongly reflect patterns generally observed among MSOs compared with solo FSOs (Faller, 1995; Gannon et al., 2014; Mathews, Hunter, & Vuz, 1997; Nathan & Ward, 2002; Vandiver, 2006). Research so far suggests this is the case—females who offend with a male partner are more likely to abuse girls, a victim characteristic consistent with the preferences displayed by MSOs (Freeman & Sandler, 2008; Williams & Bierie, 2015).

Furthermore, female co-offenders may permit access to their children or other intrafamilial relatives; for example, Gannon et al. (2008) found that about half of the victims were related to the offender through marriage or blood. In recent work, Gannon et al. (2014) found that roughly 20% of their participants were biologically related to their victims. As a result, co-offending pairs may be more likely to have a dependent child or an intrafamilial victim. This has been suggested by prior work on solo FSOs and coed pairs (see, for example, Muskens et al., 2011; Vandiver, 2006; Wijkman et al., 2010), but needs further investigation.

Moving beyond the solo and co-offending comparisons, there is reason to believe other FSO groupings may demonstrate unique incident characteristics. For instance, although studies comparing solo and co-offending show similar levels of severity of abuse and suggest victims and offenders are likely to be intrafamilial (Muskens et al., 2011; Rudin, Zalewski, & Bodmer-Turner, 1995; Vandiver, 2006; Wijkman et al., 2010), other research suggests multiple perpetrator sexual assaults, such as those that involve male offender groups or male-accompanied groups, are characterized by more injuries, are more likely to involve stranger victimization, and have a greater diversity of offending (Lambine, 2013; Morgan et al., 2012). Scholars theorize this derives from the motivation of this type of assault—assault that is often about public displays of masculinity and aggression among gangs or cliques of offenders; it is about face work (Collins, 2005; Goffman, 1967). That is, “where women are implicated in the perpetrating group, they are invariably acting within the current constructs of masculinity as a bid for acceptance and power accorded to men and boys” (Kelly, 2013, p. xv). Taken as a whole, there is evidence that suggests incidents involving multiple perpetrator groups (MPGs) will victimize strangers at greater rates, have more victim injury, and commit their sexual assaults in conjunction with other crimes, but there have been few empirical tests of these assertions to date.

There is also a lack of empirical work on all-female groups that sexually offend together. Some literature discusses sexual assault by groups of females in jails due to the special nature of the segregated setting. Much of this discussion has centered on the use of group sexual assaults, which act as tools of harassment or retaliation for insulting members of a gang or clique (Alarid, 2000). If sexual assaults are used in jails as expressions of dominance, there may also be more instances of additional violence, such as physical beatings. Therefore, all-female sexual assaults compared with other sexual assault groupings may be more likely to occur in a jail setting and result in additional non-sex crimes. These also remain untested assertions.

As with other research, there are challenges in building the body of FSO literature. After a thorough search, we located only four studies that compared FSO groups. One study was based on data from the United States, whereas the others focused on the United Kingdom and the Netherlands (see Gillespie et al., 2015; Muskens et al., 2011; Vandiver, 2006; Wijkman et al., 2010). Not only is there a scarcity of research in this area but also a vast gap of knowledge on this phenomenon in the United States. The FSOs in these studies were drawn from arrest, conviction, or court-ordered psychiatric treatment groups. Samples drawn from these settings are limiting in that only a minority of FSOs are arrested (Williams & Bierie, 2015). Because of this filtering effect of the criminal justice system (Langevin et al., 2004; Patrick & Marsh, 2011), these samples offer a narrow view of this offending population. This limits the statistical tools available to analyze data or the power to detect differences. The field benefits by pursuing samples that are larger, more recent, from early points in the criminal justice system, and that operationalize more nuanced types of FSO groupings.

The Current Study

The current study compares sexual assault incidents perpetrated by four distinct FSO groupings: solo FSOs, coed pairs, all-female groups, and MPGs. By focusing on victim, offender, and crime characteristics, we address the following research questions:

Research Question 1: Are FSO sexual assault incidents involving a male co-offender more likely to have female victims in comparison with solo FSO sexual assault incidents?

Hypothesis 1: Compared with solo FSO sexual assault incidents, sexual assault incidents involving a female and male co-offending pair will increase the odds that the victim is a female.

Research Question 2: Are FSO sexual assault incidents involving a male co-offender more likely to involve relative victims, such as a dependent child or intrafamilial family member, in comparison with solo FSO sexual assault incidents?

Hypothesis 2: Compared with solo FSO sexual assault incidents, sexual assault incidents involving a female and male co-offending pair will increase the odds that the victim is a relative.

Research Question 3: Are FSO sexual assault incidents involving MPGs more likely to result in stranger victimization, result in victim injury, and involve other non-sex crimes in comparison with other FSO sexual assault incidents?

Hypothesis 3: Compared with other FSO sexual assault incidents, sexual assault incidents involving MPGs will increase the odds that the victim is a stranger and the victim is injured, and the sexual assault incident involves other non-sex crimes.

Research Question 4: Are FSO sexual assault incidents involving all-female groups more likely to occur in jails and involve other non-sex crimes in comparison with other FSO sexual assault incidents?

Hypothesis 4: Compared with other FSO sexual assault incidents, sexual assault incidents involving an all-female group will increase the odds that the sexual assault incident occurs in a jail setting and involves other non-sex crimes.

Method

This research analyzed data from the National Incident-Based Reporting System (NIBRS) over a 21-year period, from the beginning of 1992 through the end of 2012.¹ The NIBRS, managed by the FBI, collects incident-level crime data for 52 crime types reported to police in participating agencies. As of 2012, a total of 6,115 law enforcement agencies across 37 states submitted data to the NIBRS program, approximately 33% of all agencies that submit data to the Uniform Crime Reports (FBI, 2012). The NIBRS is the largest publicly available incident-driven data set in the United States that captures information on sex crimes, including the type of offenses committed in the incident, characteristics of the offenders and their victims in the incident, and incident context characteristics such as where the crime took place.

The NIBRS collects incident-level data on six sexual offenses: forcible rape, forcible sodomy, sexual assault with an object, forcible fondling, non-forcible incest, and non-forcible statutory rape. In addition, it includes two types of sex-associated crimes: pimping and prostitution. The analysis began with a total of 1,008,483 sex crime incidents reported to police between 1992 and 2012. To narrow the analysis, pimping that did not have a direct sex crime component was dropped, approximately 8% of incidents. Because FSOs are the focus of this research, we dropped any incidents where the sex of the offender was unknown, removing approximately 6% of the incidents. We also dropped any incidents that involved (a) a solo MSO or (b) all male groups of sexual offenders, approximately 81% of the incidents. The final sample size consisted of 47,287 sex crime incidents involving FSOs.

Dependent Variable: Incidents by FSO Grouping

To construct a detailed and comparative picture of incidents of female sexual offending, we created a categorical variable that separated incidents by type of female sexual offending dynamic: solo FSOs, coed pairs, all-female groups, and MPGs.

A solo FSO referred to incidents with a single female perpetrator. Solo FSOs accounted for approximately 62% of the sample (n = 29,238). Coed pairs consisted of one MSO and one FSO who both participated in the sex crime incident, approximately 24% of the sample (n = 11,112). All-female groups consisted of two or more FSOs who participated in the sex crime incident. All-female groups accounted for approximately 6% of the sample (n = 2,669), and the vast majority of these consisted of a female pair (93%). MPGs were a mixed gender group consisting of three or more perpetrators that included at least one female who participated in the sex crime incident. MPGs accounted for approximately 9% of the incidents in the sample (n = 4,268).

Independent Variables

All the independent variables included in the analysis are analyzed at the incident level. As described below, this means that although many items were recorded at the individual level (e.g., gender of each victim; age of each offender), the items used in the analyses are incident-level summarizations of these characteristics. If there was only one offender, for example, then these variables were merely the offender-level attribute. If there were multiple offenders, then these measures represented aggregations of individual-level attributes.

Offender characteristics

We included the offender demographic variables of age, race, and substance use. Age was a continuous variable measured in years, ranging from 10 years old to 99 years old.² It recorded the offender’s age at the time of the incident. Approximately 10% of the offender age data were missing. If more than one offender was present, we used the average age of offenders; although, it is important to note that offender’s ages were within 2 years of one another in 90% of cases with multiple offenders. Incident-level offender race was coded into three mutually exclusive dummy variables: White, Black, and Other (Other including Asian and Native Americans). Due to the design of the NIBRS, race categories White and Hispanic were both pre-populated into the Caucasian category; therefore, we could not disentangle Hispanic from Caucasian to create a separate ethnicity measure. We refer to this combined Hispanic-or-Caucasian grouping as White and use it as the comparison category. In addition, due to collinearity, we had to exclude the measure that captured multiple races of offenders if more than one offender was present during the incident. This occurred in less than 4% of the incidents. Approximately 5% of the offender race data were missing. Finally, we also included two dummy variables to signal whether any offender was under the influence of drugs or alcohol during the incident.

Victim characteristics

Victim characteristics included a count of the number of victims, victim age, victim race, victim gender, and the relationship of the victim to the offender.

We computed the number of victims in each incident. This item was a continuous variable that ranged from 1 to 10 victims. Seventy-eight percent of the incidents had one sexual assault victim. If there was more than one victim in the incident, at least one of those victims was sexually assaulted. Approximately 18% of the incidents had 2 victims, and less than 5% had 3 or more victims. Victim age was a continuous variable measured in years and referred to the victim(s) age at the time of the incident. If more than one victim was present, then the item referred to the average age of victims. Approximately 1% of victim age data were missing. Incident race was coded into five mutually exclusive dummy variables at the incident level: White, Black, Other, Hispanic, and multiple races present. White was the omitted comparison category. Approximately 4.5% of victim race and ethnicity data were missing. Female victim (1 = female, 0 = male) was included in the model, using male victims as the omitted comparison category. In addition, we included mutually exclusive binary items of male, female, or both genders of victim(s) were present during the sex crime incident.

The NIBRS includes more than 20 victim–offender relationship categories. To provide breadth and depth about victim–offender relationships within these incidents, we created six distinct victim–offender relationship categories that were coded as separate variables: dependent children, intrafamilial or within the family (excluding dependent children), extrafamilial or outside the family but known to the victim, significant other, homosexual relationship, and stranger or not known to the victim. Significant other was created using the categories spouse, common-law spouse, boyfriend/girlfriend, and ex-spouse. Homosexual relationship is its own distinct NIBRS category and was not collapsed with “significant other” because some literature suggests homosexual relationships are an understudied but important feature of some sex crimes (see, for example, Rothman, Exner, & Baughman, 2011; Vandiver & Kercher, 2004).

Crime characteristics

Features of the criminal incident included the type of sexual assault behavior, measures of force used against the victim, the severity of the injury sustained by the victim, the sex crime incident location, and summary information regarding non-sex crime offenses. The NIBRS contains information on each sexual act committed against each victim. We included all the NIBRS sexual assault categories in this research: forcible rape, forcible sodomy, forcible sexual assault with an object, forcible molestation, non-forcible incest, and non-forcible statutory rape. These were coded as dummy variables and were not mutually exclusive; effects were allowed to be additive in models.

The NIBRS contains 15 categories for types of force used against victims. We collapsed the 15 categories into six primary types: guns, knives, blunt objects, personal weapons, drugs, and other. Guns included firearms, handguns, shotguns, and any other firearms. Knives included other cutting instruments such as ice picks and screwdrivers. Blunt objects were items such as clubs and hammers. The offender was considered to use personal weapons as a source of force if the offender used his or her hands, feet, or teeth, including using body parts to asphyxiate the victim. Drugs included things such as narcotics, sleeping pills, and poison. The category called other contained types of force that were atypical weapons in sexual assaults such as motor vehicles, explosives, and fire.

The NIBRS has eight classifications for victim injury. Due to the overall low frequency of injury in the sample, we created three dummy variables to measure victim injury: none, minor, and major. Minor injuries were defined as minor by the NIBRS. We created a major injury dummy by combining broken bones, internal injury, loss of teeth, severe laceration, unconsciousness, and other major injury.

The NIBRS tracks where the offense occurred using 25 specific location categories. Due to the field’s interest on sexual abuse in the home, jails, and schools, we left these as distinct location categories in the model. We also left hotel as its own distinct location because hotels may signal interesting and unique pathways to offending (e.g., offending among college students on spring break, or pimping and prostitution), and the sample size was large enough to support these analyses. We then collapsed the remaining 21 locations into the following categories: service locations (e.g., hospital, church, government buildings such as libraries), business locations (e.g., bar, store, gas station), and outdoor locations (such as construction areas, woods, alleys, parking lots). We also included a measure to capture whether the sex crime moved from one location to another. Finally, we included measures to capture whether the sex crime incident occurred in conjunction with another crime: computer (e.g., pornography), pimping, drugs, robbery, and assault.

Analytic Strategy

Using the statistical software package Stata, the incidents comprising the four FSO grouping were first described across crime, offender, and victim characteristics. Bivariate comparisons on each item were provided to paint a rich descriptive picture of the data. Research questions were then addressed using a multinomial logistic regression model (MNLM) in which group membership in the incident was the nominal outcome category (see Long & Freese, 2006). With this type of regression model, the probability of membership in each of the other female sexual offending categories in the incident was compared with the probability of membership in the reference, or base, category in the incident. For this analysis, incidents with solo FSOs were the reference category.

The MNLM reported coefficients for the effect of each independent variable on incidents containing each female sexual offending category relative to solo FSO incidents (Long & Freese, 2006). Because multinomial log-odds coefficients lack a metric in which substantive results can be explained, significant findings were reported as odds ratios (Pampel, 2000). Odds ratios were interpreted as follows: For each additional unit change in x_k (p ≤ .05), the odds of m versus n are expected to change by a factor of exp (β_k, m | n) holding all other variables constant.

Due to the complexity of the model and the corresponding statistical output, results in the tables are shown only for the reference category, incidents involving solo FSOs, relative to incidents with other groupings, although we do discuss statistically significant comparisons between incidents involving all groupings; for example, comparing incidents involving coed pairs versus incidents involving MPGs. Estimates for comparing incidents by each grouping were obtained by running the MNLM and then running an additional user command written by Long and Freese (2006), which produces coefficient estimates and odds ratios for all combinations of outcome categories. These results are available on request.

Because this study involves a large number of analyses, we applied a post hoc adjustment to account for Type I error. We used the Benjamini and Hochberg (1995) method to adjust the alpha level to control the false discovery rate (FDR), or the proportion of false positives among all significant results. Because power is seriously affected by the Bonferroni method, this FDR method is advantageous in that the loss of power is relatively small, and it balances the need to control Type I error from multiple testing while maintaining sensitivity to detect effects (Benjamini & Hochberg, 1995; Verhoeven, Simonsen, & McIntyre, 2005). The alpha for this research is based on the FDR adjustment: p ≤ .03.

All multivariate models used a listwise deletion strategy for item non-response. After missing data were dropped, the final MNLM analyzed 34,468 (73%) incidents of sex crimes committed by FSOs (see Table 1 for descriptive statistics and bivariate comparisons between incidents by FSO grouping).

Table 1.

Descriptive Statistics (N = 47,287) and Bivariate Comparisons of Incidents by FSO Grouping (Reference Category: Solo Female Offender Incidents).

			Solo female incidents		Coed pair incidents		All-female group incidents		MPG (3+) incidents
	Minimum	Maximum	M	SD	M	SD	M	SD	M	SD
Offender
Age (years)	10	99	25.697	13.205	28.922***	11.766	24.385***	12.235	26.750***	10.093
Race
White-Hispanic	0	1	0.807	0.395	0.758***	0.428	0.743***	0.437	0.687***	0.464
Black	0	1	0.181	0.385	0.141***	0.348	0.211***	0.408	0.159***	0.366
Other	0	1	0.012	0.107	0.007***	0.081	0.014	0.119	0.005***	0.072
Substance use
Alcohol	0	1	0.026	0.160	0.066***	0.249	0.030	0.172	0.088***	0.283
Drugs	0	1	0.012	0.110	0.042***	0.200	0.015	0.122	0.067***	0.249
Victim
Count of victims	1	10	1.141	0.482	1.421***	0.666	1.776***	0.951	1.818***	1.261
Age (years)	0	94	12.289	10.697	14.835***	10.377	12.758*	10.444	15.729***	10.041
Race
White	0	1	0.757	0.429	0.787***	0.410	0.729***	0.445	0.786***	0.410
Black	0	1	0.187	0.390	0.165***	0.371	0.232***	0.422	0.169**	0.375
Other	0	1	0.010	0.101	0.016***	0.124	0.011	0.102	0.017***	0.128
Hispanic	0	1	0.053	0.224	0.072***	0.259	0.053	0.224	0.066***	0.248
Multiple	0	1	0.007	0.086	0.039***	0.194	0.023***	0.149	0.037***	0.189
Gender
Female	0	1	0.425	0.494	0.638***	0.481	0.469***	0.499	0.706***	0.456
Male	0	1	0.542	0.498	0.141***	0.348	0.481***	0.500	0.120***	0.325
Multiple	0	1	0.033	0.179	0.221***	0.415	0.050***	0.219	0.174***	0.379
Relation to offender
Dependent children	0	1	0.133	0.340	0.321***	0.0467	.0130	0.336	0.260***	0.438
Intrafamilial	0	1	0.187	0.390	0.157***	0.366	0.201	0.401	0.197	0.398
Extrafamilial	0	1	0.613	0.487	0.655***	0.476	0.698***	0.459	0.781***	0.414
Significant other	0	1	0.043	0.202	0.128***	0.334	0.034	0.182	0.091***	0.288
Homosexual relation	0	1	0.009	0.093	0.002***	0.039	0.015**	0.120	0.002***	0.048
Stranger	0	1	0.023	0.150	0.048***	0.214	0.033**	0.178	0.116***	0.321
Crime
Sexual assault behavior
Rape	0	1	0.130	0.336	0.334***	0.472	0.123	0.329	0.399***	0.490
Sodomy	0	1	0.095	0.293	0.090	0.286	0.114**	0.317	0.107*	0.309
Object penetration	0	1	0.071	0.256	0.059***	0.236	0.090***	0.286	0.059*	0.235
Forced molestation	0	1	0.591	0.492	0.406***	0.491	0.575	0.494	0.349***	0.477
Incest	0	1	0.024	0.155	0.033***	0.178	0.020	0.140	0.019	0.138
Statutory	0	1	0.096	0.294	0.107***	0.309	0.092	0.289	0.104	0.306
Use of force
Blunt	0	1	0.004	0.066	0.005	0.068	0.007*	0.086	0.007*	0.084
Drug	0	1	0.001	0.031	0.003***	0.058	0.001	0.039	0.006***	0.079
Gun	0	1	0.001	0.027	0.007***	0.083	0.003***	0.055	0.026***	0.158
Personal weapons	0	1	0.501	0.500	0.524***	0.499	0.518	0.500	0.530***	0.499
Knife	0	1	0.003	0.054	0.007***	0.082	0.006*	0.075	0.015***	0.120
Other	0	1	0.034	0.181	0.036	0.185	0.039	0.193	0.036	0.187
Injury
Minor	0	1	0.088	0.283	0.171***	0.376	0.101*	0.302	0.201***	0.401
Major	0	1	0.019	0.137	0.052***	0.222	0.021	0.142	0.069***	0.254
Location
Home	0	1	0.775	0.418	0.806***	0.395	0.749**	0.433	0.770	0.421
Jail	0	1	0.021	0.145	0.002***	0.040	0.029**	0.168	0.002***	0.048
School	0	1	0.039	0.193	0.022***	0.146	0.050**	0.218	0.019***	0.138
Hotel	0	1	0.010	0.099	0.028***	0.165	0.011	0.104	0.038***	0.191
Service	0	1	0.026	0.158	0.011***	0.105	0.025	0.158	0.015***	0.122
Business	0	1	0.016	0.127	0.018	0.134	0.018	0.132	0.023**	0.150
Outdoors	0	1	0.038	0.190	0.061***	0.239	0.044	0.206	0.092***	0.289
Moves	0	1	0.001	0.036	0.018***	0.134	0.008***	0.090	0.036***	0.187
Additional crimes
Computer involved	0	1	0.002	0.047	0.011***	0.106	0.006***	0.077	0.017***	0.130
Pimping/prostitution	0	1	0.001	0.036	0.006***	0.076	0.003	0.055	0.012***	0.111
Drugs	0	1	0.002	0.046	0.021***	0.142	0.003	0.058	0.040***	0.195
Robbery	0	1	0.001	0.030	0.005***	0.073	0.003*	0.051	0.025***	0.155
Assault	0	1	0.006	0.080	0.076***	0.265	0.022***	0.147	0.071***	0.257

Note. Statistical tests use solo FSO as the reference, or base, category. Intrafamilial victim excludes dependent children. Personal weapons are defined as hands, feet, teeth, and so on. FSO = female sexual offender; MPG = Multiple perpetrators groups.

p ≤ .03. **p ≤ .01. ***p ≤ .001.

Results

Research Question 1: FSO incidents involving co-offending pairs and victim gender

Bivariate analyses indicated that incidents that involved coed pairs (64%) more often had female victims compared with solo FSOs ( p ≤ .001). Based on the MNLM, compared with solo FSO sexual assault incidents, sexual assault incidents involving a female and male co-offending pair increased the odds that the victim was a female compared with male. If the victim was a female, the odds the sexual assault incident involved a coed pair relative to a solo FSO were 9.08 times greater, holding all variables constant ( p ≤ .001). Said differently, if the victim was a female, the probability the sexual assault incident involved a coed pair is .34 greater than if the victim was male, holding all other variables at their mean. In contrast, if the victim was a female, the probability the sexual assault incident involved a solo FSO decreases by .43, holding all other variables at their mean. Thus, this finding supports the first hypothesis that incidents involving coed pairs compared with solo FSOs are more likely to have female victims.

Research Question 2: FSO incidents involving co-offending pairs and victim–offender relationship

Bivariate analyses indicated that dependent children were more likely to be victimized in incidents that involved coed pairs (32%) compared with solo FSO incidents ( p ≤ .001). and that incidents involving coed pairs (16%) were more likely to have an intrafamilial family member victim compared with solo FSO incidents ( p ≤ .001). Based on the MNLM, compared with solo FSO sexual assault incidents, sexual assault incidents involving a female and male co-offending pair increased the odds that the victim is a relative. The odds an incident involved a coed pair relative to a solo FSO were 101.83 times greater if the victim was a dependent child, holding all other variables constant ( p ≤ .001). Said differently, if the victim was a dependent child, the probability the sexual assault incident involved a coed pair is .55 greater than if the victim was not a dependent child, holding all other variables at their mean. In contrast, if the victim was a dependent child, the probability the sexual assault incident involved a solo FSO decreases by .75, holding all other variables at their mean.

Pertaining to intrafamilial victims, the odds an incident involved a coed pair relative to a solo FSO were 34.98 times greater if the victim was an intrafamilial family member, holding all other variables constant ( p ≤ .001). Said differently, if the victim was an intrafamilial family member, the probability the sexual assault incident involved a coed pair is .39 greater than if the victim was not an intrafamilial family member, holding all other variables at their mean. In contrast, if the victim was an intrafamilial family member, the probability the sexual assault incident involved a solo FSO decreases by .67, holding all other variables at their mean. These findings support the second hypothesis that incidents involving coed pairs compared with solo FSOs are more likely to have relative victims measured as dependent children and intrafamilial family members.

Research Question 3: MPG incidents, victim–offender relationship, victim injury, and non-sexual assault criminal behaviors

Based on the bivariate analyses, incidents with MPGs (12%) were more likely to have stranger victims compared with solo FSO incidents ( p ≤ .001). Sexual assault incidents involving MPGs were more likely to involve injury, minor (20%) and major (7%), compared with solo FSO incidents ( p ≤ .001). MPG incidents were also more likely to be involved in computer crimes (2%), pimping/prostitution (1%), drugs (4%), robbery (3%), and assault (7%) compared with solo FSO incidents ( p ≤ .001).

Based on the MNLM, compared with other FSO sexual assault incidents, sexual assault incidents involving MPGs increased the odds that the victim is a stranger, the victim is injured, and the sexual assault incident involved other non-sex crimes. Stranger victimizations were most likely in incidents that involved MPGs relative to all other groupings—the odds increased 187.61 versus solo FSO incidents, 8.37 versus all-female group incidents, and 2.89 versus coed pair incidents, holding all other variables constant ( p ≤ .001). Said differently, if the victim was a stranger, the probability the sexual assault incident involved an MPG is .35 greater than if the victim was not a stranger, holding all other variables at their mean. In contrast, if the victim was a stranger, the probability the sexual assault incident involved a solo FSO decreases by .61, holding all other variables at their mean.

As hypothesized, injury was more likely to result from an MPG sexual assault incident. The odds the incident involved an MPG relative to a solo FSO, a coed pair, and an all-female group were 2.28, 1.34, and 2.27 times greater if the sexual assault resulted in a major injury, holding all other variables constant ( p ≤ .001). Said differently, if the victim sustained a major injury, the probability the sexual assault incident involved an MPG is .06 greater than if the victim did not sustain a major injury, holding all other variables at their mean. In contrast, if the victim sustained a major injury, the probability the sexual assault incident involved a solo FSO decreases by .13, holding all other variables at their mean. Minor injury was also more likely to result from an MPG sexual assault incident. The odds the incident involved an MPG relative to a solo FSO, a coed pair, and an all-female group were 1.83, 1.20, and 1.60 times greater if the sexual assault resulted in a minor injury, holding all other variables constant ( p ≤ .001).

As hypothesized, MPG sexual assault incidents were more likely to involve non-sex crimes. Compared with solo FSO and all-female group incidents, MPG incidents were more likely to involve the following crime types: computer-involved crimes (e.g., pornography), prostitution/pimping, drugs, and assault (e.g., aggravated assault, simple assault, and intimidation). For example, the odds an incident involved an MPG relative to a solo FSO or an all-female group were 1.83 and 1.55 times greater if the sex crime incident also involved an assault, holding all other variables constant ( p ≤ .001, p ≤ .03). In addition, MPG incidents were more likely to involve robbery compared with incidents involving all other groupings. If the sexual assault incident also involved a robbery, the odds the incident involved an MPG relative to a solo FSO, a coed pair, or an all-female group were 5.59, 2.41, and 3.09 times greater, holding all other variables constant ( p ≤ .001, p ≤ .001, p ≤ .03). These findings support the third hypothesis that incidents involving MPGs compared with other FSO groupings are more likely to have stranger victims, greater victim injury, and the co-concurrence of non-sex crimes in conjunction with the sexual assault incident.

Research Question 4: All-female group incidents, sexual assault location, and non-sexual assault criminal behaviors

Bivariate analyses indicated that incidents that involved all-female groups (3%) were more likely to occur in a jail compared with solo FSO incidents (p ≤ .01). When investigating non-sex crimes that occurred in conjunction with sexual assault incidents, incidents that involved all-female group sexual assault were more likely to involve computer crimes (e.g., pornography; 1%, p ≤ .001), robbery (0.3%, p ≤ .03), and assault (2%, p ≤ .001) compared with solo FSO incidents.

The final hypothesis was partially supported: Compared with other FSO sexual assault incidents, sexual assault incidents involving an all-female group did increase the odds that the sexual assault incident occurred in a jail setting. If the sexual assault occurred in a jail, the odds the incident involved an all-female group relative to a coed pair or MPG were 11.22 and 9.98 times greater, holding all other variables constant ( p ≤ .001). Said differently, if the incident occurred in a jail, the probability the sexual assault incident involved an all-female group is .05 greater than if the sexual assault incident did not happen in a jail, holding all other variables at their mean. However, there was no significant difference when comparing solo FSO and all-female group sexual assault incidents that occurred in jails. When compared with other sexual assault incidents involving other groupings, there were also no significant findings that incidents involving all-female groups were more likely to involve additional non-sex crimes compared with incidents with other groupings.

Discussion

This study drew on the nation’s largest publicly available incident-level data set on crimes reported to police, the NIBRS, to compare sexual assault incidents involving four groupings of FSOs in relation to offender, victim, and crime characteristics. Overall, a key finding was that incidents that involved solo FSOs or all-female groups mirrored each other in many offender, victim, and crime characteristics; a similar relationship was observed in incidents that consisted of females acting in consort with one or more MSOs. In addition to adding to the body of knowledge on female sexual offending, this study also contributes to the continuing effort to formulate a gender-specific approach to FSOs by better understanding female sexual offending behavior in terms of their offender grouping (Cortoni, 2010; Cortoni & Gannon, 2013).

With regard to victim characteristics, in accordance with prior studies (Vandiver, 2006; Wijkman et al., 2010), incidents involving coed pairs were more likely than incidents involving solo FSOs to victimize females. This lends support to the theoretical premise of the co-offending dynamic. If females participate with males, whether coerced or willingly, it makes sense that coed pair victims are more likely to be female than male. This victim selection exhibits MSO offending preferences (Freeman & Sandler, 2008; Williams & Bierie, 2015). What is unknown though is how the coed pair group dynamic unfolds in regard to victim selection and offender participation level in terms of sexual assaults (Porter, 2013). Further research is needed to understand these dynamics and better inform a gender-specific approach that can be used by community actors, such as treatment providers.

With regard to victim–offender relationship, there is evidence that co-offenders are more likely to sexually assault those who are known to them (Muskens et al., 2011; Vandiver, 2006; Wijkman et al., 2010). In line with this evidence, this research found that incidents involving coed pairs were more likely to involve dependent children compared with incidents that involve solo FSOs. This significance also held for incidents involving intrafamilial family members. Because of this dynamic, especially with dependent children, we were interested to see whether non-forcible incest played a significant role in co-offending incidents versus other type of offending incidents; it did not once controls were added (Table 2). In addition, non-forcible incest offending only comprised 3% or fewer of the incidents in any group.

Table 2.

Multinomial Logistic Regression Predicting the Likelihood of an Incident Involving a Coed Pair, All-Female Group, or MPG (Reference Category: Solo Female Offender Incidents; n = 34,468).

	Coed pair incidents		All-female group incidents		MPG (3+) incidents
	Coefficient (SD)	OR	Coefficient (SD)	OR	Coefficient (SD)	OR
Offender
Age (years)	0.014 (0.001)	1.01***	−0.014 (0.003)	0.99***	−0.004 (0.002)	0.99*
Race
Black	−0.419 (0.067)	0.66***	−0.169 (0.089)		−0.434 (0.092)	0.65***
Other	−0.960 (0.213)	0.38***	0.186 (0.251)		−1.183 (0.325)	0.31***
Substance use
Alcohol	0.510 (0.076)	1.67***	0.194 (0.138)		0.780 (0.091)	2.18***
Drugs	0.519 (0.113)	1.68***	−0.076 (0.221)		0.744 (0.131)	2.11***
Victim
Count of victims	0.156 (0.038)	1.17***	1.352 (0.037)	3.87***	1.089 (0.037)	2.97***
Age (years)	0.006 (0.002)	1.01***	0.008 (0.003)	1.01**	0.010 (0.003)	1.01***
Race
Black	0.152 (0.066)	1.16*	0.441 (0.087)	1.55***	0.074 (0.091)
Other	0.445 (0.173)	1.55**	0.019 (0.280)		0.516 (0.223)	1.68*
Hispanic	0.154 (0.071)	1.17*	0.002 (0.117)		0.016 (0.099)
Multiple	0.217 (0.133)		−0.940 (0.217)	0.39***	−0.061 (0.165)
Gender
Female	2.210 (0.045)	9.08***	0.261 (0.056)	1.29***	2.442 (0.066)	11.49***
Multiple	3.162 (0.079)	23.62***	−1.468 (0.136)	0.23***	2.044 (0.102)	7.72***
Relation to offender
Dependent children	4.623 (0.097)	101.83***	2.780 (0.124)	16.28***	4.868 (0.107)	130.09***
Intrafamilial	3.555 (0.097)	34.98***	2.972 (0.120)	19.55***	4.217 (0.106)	67.80***
Extrafamilial	4.095 (0.096)	60.05***	3.210 (0.119)	24.78***	4.952 (0.105)	141.41***
Significant other	4.117 (0.126)	61.37***	2.801 (0.182)	16.45***	4.08 (0.142)	59.32***
Homosexual relation	0.986 (0.352)	2.68**	3.194 (0.229)	24.39***	1.764 (0.434)	5.83***
Stranger	4.174 (0.132)	64.98***	3.109 (0.184)	22.41***	5.234 (0.142)	187.61***
Crime
Sexual assault behavior
Rape	1.989 (0.134)	7.31***	−0.059 (0.228)		1.743 (0.146)	6.64***
Sodomy	0.351 (0.132)	1.42**	−0.036 (0.223)		0.170 (0.146)
Object penetration	−0.084 (0.140)		0.086 (0.232)		−0.390 (0.159)
Forced molestation	−0.038 (0.136)		−0.239 (0.227)		−0.503 (0.148)	0.66**
Incest	0.485 (0.433)		0.434 (0.692)		−0.379 (0.604)
Statutory rape	0.794 (0.319)	2.21**
Use of force
Blunt	−0.221 (0.230)		0.487 (0.282)		0.203 (0.274)
Drug	0.073 (0.340)		0.234 (0.669)		0.802 (0.362)	2.23*
Gun	1.084 (0.319)	2.96**	1.017 (0.485)		1.944 (0.324)	6.99***
Personal weapons	0.101 (0.035)	1.11**	0.0002 (0.053)		0.146 (0.048)	1.16**
Knife	−0.062 (0.233)		0.461 (0.324)		0.351 (0.257)
Other	0.177 (0.087)	1.19*	0.032 (0.135)		0.287 (0.117)	1.33**
Injury
Minor	0.418 (0.050)	1.52***	0.136 (0.087)		0.604 (0.064)	1.83***
Major	0.525 (0.093)	1.69***	0.007 (0.180)		0.825 (0.111)	2.28***
Location
Home	0.098 (0.066)		−0.066 (0.098)		−0.026 (0.089)
Jail	−2.164 (0.263)	0.11***	0.253 (0.166)		−2.047 (0.387)	0.13***
School	−0.178 (0.117)		0.076 (0.152)		−0.320 (0.169)
Hotel	0.707 (0.137)	2.03***	0.091 (0.252)		0.783 (0.166)	2.19***
Service	−1.002 (0.152)	0.38***	−0.021 (0.183)		−0.903 (0.202)	0.41***
Business	−0.013 (0.134)		0.018 (0.206)		−0.250 (0.183)
Outdoors	0.453 (0.098)	1.57***	0.123 (0.151)		0.543 (0.123)	1.72***
Moves	0.945 (0.239)	2.57***	0.934 (0.325)	2.45**	1.151 (0.255)	3.16***
Additional crimes
Computer involved	1.220 (0.231)	3.39***	−0.854 (0.374)	0.43*	0.639 (0.259)	1.89**
Pimping/prostitution	1.104 (0.329)	3.02**	−0.446 (0.554)		0.815 (0.353)	2.26*
Drugs	1.460 (0.222)	4.30***	−0.894 (0.447)	0.41*	0.819 (0.239)	2.27**
Robbery	0.844 (0.294)	2.33**	0.593 (0.466)		1.722 (0.295)	5.59***
Assault	1.074 (0.114)	2.93***	0.167 (0.190)		0.602 (0.135)	1.83***

Note. For clarity, odds ratios are only shown if statistically significant. Intrafamilial victim excludes dependent children. Personal weapons are defined as hands, feet, teeth, and so on. MPG = Multiple perpetrators groups; OR = odds ratio.

p ≤ .03. **p ≤ .01. ***p ≤ .001.

Previous studies have reported that MPG sexual assaults take on distinguishing characteristics compared with other sexual assault offender groupings: stranger victimizations, more victim injury (Lambine, 2013; Morgan et al., 2012), and a greater diversity of offending that includes non-sexual crimes (Lambine, 2013). The data analysis here offered support to prior research findings. Morgan et al. (2012) found that multiple perpetrators were more likely to be strangers to their victims than single perpetrators. This research supports and extends that finding as MPG incidents were more likely to involve stranger victims compared with solo FSO incidents, but in general, we found that group size mattered. If the sexual assault incident involved a coed pair, an all-female group, or an MPG compared with a solo FSO, the odds of stranger victimization significantly increased. Future research should further investigate the dynamics between group size and victim selection. The MPG incidents also involved more injury than other groupings and more diverse offending beyond sexual crimes, especially robbery. However, these features were rare even if they were significantly greater among MPG incidents. For example, only approximately 3% of the sexual assault incidents involved a major victim injury regardless of the offender grouping. Also, it is important to note that some additional crimes, such as assault, were more likely among coed pair incidents relative to incidents that involved solo FSOs, all-female groups, or MPGs.

Last, our investigation of sexual assault incidents involving all-female groups supports that these types of incidents are more likely to occur in a jail setting compared with incidents involving one or more male co-offenders. This diverges from Vandiver (2006) who found no significant differences in terms of the sexual assault location. This may be, in part, due to Vandiver’s (2006) coding of location as residence and non-residence, whereas our research used the specific location of jail and did not collapse it with other location categories. Pertaining to sexual assaults in jails, it is difficult to know whether this is a function of inmates assaulting other inmates or a feature of female prison guards exploiting prisoners. Importantly, a sizable portion (42%) of solo female and all-female incidents occurring within jails or prisons involved females assaulting male victims. Presumably, these represent sexual abuse by female prison staff against male inmates, as few jails in the United States are coed. The remaining portion of cases could be comprised of inmate–inmate violence, inmates assaulting staff, or staff assaulting one another. Offending in jails represented a substantively small portion of the assaults in this data set (less than 3%). Still, understanding these offending patterns may represent an important and underexplored context to female sexual offending.

Limitations

Although this research adds to the body of knowledge on female sexual offending, this research is not without limitations. The NIBRS is limited in that it only collects incident-level data, and as a result, this analysis does not model the offenders themselves but the incidents and their characteristics. Therefore, future research should examine sexual assault behavior as individual-level analyses. In addition, there is no way to disentangle whether an FSO was involved with more than one incident; for example, one FSO could have been involved in a coed incident and an MPG incident. NIBRS does not collect information on certain variables that may be of interest, such as the criminal history of victims or offenders. In regard to crime types, the NIBRS collects data on six sexual offenses involving offender–victim contact, but it does not include non-contact crimes such as indecent exposure, voyeurism, and child pornography.

The NIBRS data over-represent small and medium sized police departments. Also, they do not include data from some areas of the nation with disproportionately high crime (e.g., New York City). This may limit the generalizability of data only to agencies that report to the NIBRS. Because this is a criminal justice sample, there is the issue of underreporting. Underreporting of female perpetrated sex crimes may be due to cultural factors such as lack of awareness about female sexual offending, gender role stereotypes, and/or traditional gender scripts, the perception sex offenses committed by women are less serious than those committed by a man, and/or a reluctance to report due to the gender of the perpetrator (Becker, Hall, & Stinson 2001; Bunting, 2007; Center for Sex Offender Management [CSOM], 2007; Denov, 2003, 2004; Vandiver & Kercher, 2004). Therefore, although this research uses the largest sample of FSO sexual assault incidents to date, it does not capture all sex crimes perpetrated by females and should not be construed as being inclusive of offending not reported to police.

The data system also required data entry by more than 6,000 different police agencies, a process that may generate important differences over time or place with respect to the meaning of a particular item. We might wonder, for example, whether police agencies differed with respect to the working definition of incest in these data, given that approximately 20% of cases involved the sexual assault of a dependent child, but only 3% were deemed incest. The difference may be that “incest” here excluded incidence with force, whereas the assault of a dependent child could include force. However, a portion of the difference could also be associated with differing definitions. The FBI reports that the NIBRS system is audited, and agencies are required to maintain an error rate below 3% to retain accreditation. Regardless, it is likely that interrater reliability remains an untested, yet plausible, challenge in these data. Perhaps the most striking limitation, however, is the lack of specific contextual or qualitative data. Unlike research drawing on case files or interviews, we were unable to seek out more nuance in patterns as they emerged. For example, we cannot obtain more detail on which offender (if any) took the lead during the crimes, how the situations transpired, and what occurred over time between offenders engaged in these crimes. Likewise, the data system provided no additional details on victims or offender histories that may have further illuminated how these crimes emerged.

Conclusion

As the body of research addressing FSOs continues to grow, there is still a lack of research discerning FSO offending behavior when comparing different types of FSO groupings. This research is one step toward filling that gap because we had the ability to analyze more than 30,000 sexual assault incidents committed by FSOs that span across a large portion of the nation, and which had accumulated more than 2 decades of time. Although this analysis used a more precise grouping of incidents of female sexual offending than in prior work, there is still much work to be done to better understand the sexual offending patterns of females. Future research should continue to investigate and compare FSO group behavior within subgroups of FSO offenders such as juvenile FSOs, adult FSOs, MPG FSO offenders, and all-female groups. By continuing to address FSO offending dynamics, we can better understand these variations in offending as they have implications for offenders, victims, and communities.

Footnotes

Acknowledgements

We would like to thank Dr. Michael Bourke, Paul J. Detar, and Dr. Sarah Mustillo for suggestions throughout the process of conducting this research. We would also like to thank Dr. Franca Cortoni, the action editor, and the anonymous reviewers for their invaluable feedback on the manuscript throughout the revision process.

Authors’ Note

The views and opinions of this research do not necessarily represent those of the U.S. Department of Justice or any component therein.

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) received no financial support for the research, authorship, and/or publication of this article.

Notes

References

Alarid

L. F.

(2000). Sexual assault and coercion among incarcerated women prisoners: Excerpts from prison letters. The Prison Journal, 80, 391-406. doi:10.1177/0032885500080004005

Becker

J. V.

Hall

S. R.

Stinson

J. D.

(2001). Female sexual offenders: Clinical, legal, and policy issues. Journal of Forensic Psychology Practice, 1(3), 29-50. doi:10.1300/J158v01n03_02

Benjamini

Hochberg

(1995). Controlling the false discovery rate: A practical and powerful approach to multiple testing. Journal of Royal Statistical Society. Series B Methodological, 57, 289-300.

Bierie

Davis-Siegel

(2015). Measurement matters: Comparing old and new definitions of rape in federal statistical reporting. Sexual Abuse: A Journal of Research and Treatment, 27(5), 443-459.

Black

M. C.

Basile

K. C.

Breiding

M. J.

Smith

S. G.

Walters

M. L.

Merrick

M. T.

. . . . Stevens

M. R.

(2011). The National Intimate Partner and Sexual Violence Survey (NISVS): 2010 summary report. Atlanta, GA: National Center for Injury Prevention and Control, Centers for Disease Control and Prevention.

Bunting

(2007). Dealing with a problem that doesn’t exist? Professional responses to female perpetrated child sexual abuse. Child Abuse Review, 16, 252-267. doi:10.1002/car.982

Center for Sex Offender Management. (2007). Female sex offenders. Washington, DC: A Project of the Office of Justice Programs, U.S. Department of Justice. Retrieved from http://www.csom.org/pubs/female_sex_offenders_brief.pdf

Collins

(2005). Interaction ritual chains. Princeton, NJ: Princeton University Press.

Cortoni

(2010). The assessment of female sexual offenders. In Gannon

T. A.

Cortoni

(Eds.), Female sexual offenders: Theory, assessment, and treatment (pp. 87-100). Chichester, UK: Wiley-Blackwell.

10.

Cortoni

Gannon

T. A.

(2013). What works with female sexual offenders. In Craig

Dixon

Gannon

T. A.

(Eds.), What works in offender rehabilitation: An evidence based approach to assessment and treatment (pp. 271-284). Chichester, UK: Wiley-Blackwell.

11.

Denov

M. S.

(2003). The myth of innocence: Sexual scripts and the recognition of child sexual abuse by female perpetrators. The Journal of Sex Research, 40, 303-314. doi:10.1080/00224490309552195

12.

Denov

M. S.

(2004, October). The long-term effects of child sexual abuse by female perpetrators: A qualitative study of male and female victims. Journal of Interpersonal Violence, 19, 1137-1156. doi:10.1177/0886260504269093

13.

Faller

K. C.

(1995). A clinical sample of women who have sexually abused children. Journal of Child Sexual Abuse, 4, 13-30. doi:10.1300/J070v04n03_02

14.

Federal Bureau of Investigation. (2012). National Incident-Based Reporting System: A word about NIBRS. Washington, DC: U.S. Department of Justice. Retrieved from www.fbi.gov/about-us/cjis/ucr/nibrs/2012

15.

Freeman

N. J.

Sandler

J. C.

(2008). Female and male sex offenders: A comparison of recidivism patterns and risk factors. Journal of Interpersonal Violence, 23, 1394-1413. doi:10.1177/0886260508314304

16.

Gannon

T. A.

Rose

M. R.

Ward

(2008). A descriptive model of the offense process for female sexual offenders. Sexual Abuse: A Journal of Research and Treatment, 20, 352-374. doi:10.1177/1079063208322495

17.

Gannon

T. A.

Waugh

Taylor

Blanchette

O’Connor

Blake

Ciardha

C. O.

(2014). Women who sexually offend display three main offense styles: A reexamination of the descriptive model of female sexual offending. Sexual Abuse: A Journal of Research and Treatment, 26, 207-224. doi:10.1177/1079063213486835

18.

Gillespie

S. M.

Williams

Elliott

I. A.

Eldridge

H. J.

Ashfield

Beech

A. R.

(2015). Characteristics of females who sexually offend: A comparison of solo and co-offenders. Sexual Abuse: A Journal of Research and Treatment, 27, 284-301. doi:10.1177/1079063214556358

19.

Goffman

(1967). Interaction ritual: Essays on face-to-face behavior. New York, NY: Anchor Books.

20.

Grayston

A. D.

De Luca

R. V.

(1999). Female perpetrators of child sexual abuse: A review of the clinical and empirical literature. Aggression and Violent Behavior, 4, 93-106. doi:10.1016/S1359-1789(98)00014-7

21.

Horvath

M. A. H.

Woodhams

(Eds.). (2013). Handbook of multiple perpetrator rape: A multidisciplinary response to an international problem. New York, NY: Routledge.

22.

Kelly

(2013). Naming and defining multiple perpetrator rape: The relationship between concepts and research. In Horvath

M. A. H.

Woodhams

(Eds.), Handbook on the study of multiple perpetrator rape: A multidisciplinary response to an international problem (pp. 67-81). New York, NY: Routledge.

23.

Lambine

(2013). Numbers matter: Characteristic difference between lone, duo, and 3+ group rapes. In Horvath

M. A. H.

Woodhams

(Eds.), Handbook on the study of multiple perpetrator rape: A multidisciplinary response to an international problem (pp. 67-81). New York, NY: Routledge.

24.

Langevin

Curnoe

Federoff

Bennett

Langevin

Peever

. . . Sandhu

(2004). Lifetime sex offender recidivism: A 25-year follow-up study. Canadian Journal of Criminology and Criminal Justice, 46, 531-552. doi:10.3138/cjccj.46.5.531

25.

Lewis

C. F.

Stanley

C. R.

(2000). Women accused of sexual offenses. Behavioral Sciences & the Law, 18, 73-81.

26.

Long

J. S.

Freese

(2006). Regression models for categorical dependent variables using Stata (2nd ed.). College Station, TX: Stata Press.

27.

Mathews

Hunter

J. A.

Vuz

(1997). Juvenile female sexual offenders: Clinical characteristics and treatment issues. Sexual Abuse: A Journal of Research & Treatment, 9, 187-199. doi:10.1177/107906329700900304

28.

Mathews

Matthews

J. K.

Speltz

(1989). Female sexual offenders: An exploratory study. Orwell, VT: Safer Society Press.

29.

Mathews

Matthews

J. K.

Speltz

(1991). Female sexual offenders: A typology. In Patton

(Ed.), Family sexual abuse: Frontline research and evaluation (pp. 147-161). New York, NY: Guilford Press.

30.

Morgan

Brittain

Welch

(2012). Multiple perpetrator sexual assault: How does it differ from assault by a single perpetrator? Journal of Interpersonal Violence, 27, 2415-2436. doi:10.1177/0886260511433514

31.

Muskens

Bogaerts

van Casteren

Labrijn

(2011). Adult female sexual offending: A comparison between co-offenders and solo offenders in a Dutch sample. Journal of Sexual Aggression, 17, 46-60. doi:10.1080/13552600.2010.544414

32.

Nathan

Ward

(2002). Female sex offenders: Clinical and demographic features. Journal of Sexual Aggression, 8(1), 5-21. doi:10.1080/13552600208413329

33.

Pampel

F. C.

(2000). Logistic regression: A primer. Thousand Oaks, CA: Sage.

34.

Patrick

Marsh

(2011). Sentencing outcomes of convicted child sex offenders. Journal of Child Sexual Abuse, 20, 94-108. doi:10.1080/10538712.2011.541356

35.

Porter

(2013). Leadership and role-taking in multiple perpetrator rape. In Horvath

M. A. H.

Woodhams

(Eds.), Handbook on the study of multiple perpetrator rape: A multidisciplinary response to an international problem (pp. 160-181). New York, NY: Routledge.

36.

Rothman

E. F.

Exner

Baughman

A. L.

(2011). The prevalence of sexual assault against people who identify as gay, lesbian, or bisexual in the United States: A systematic review. Trauma, Violence, & Abuse, 12, 55-66. doi:10.1177/1524838010390707

37.

Rudin

M. M.

Zalewski

Bodmer-Turner

(1995). Characteristics of child sexual abuse victims according to perpetrator gender. Child Abuse & Neglect, 19, 963-973. doi:10.1016/0145-2134(95)00058-G

38.

Turchik

J. A.

Hebenstreit

C. L.

Judson

S. S.

(2016). An examination of the gender inclusiveness of current theories of sexual violence in adulthood: Recognizing male victims, female perpetrators, and same-sex violence. Trauma, Violence, & Abuse, 17(2), 133-148.

39.

Vandiver

D. M.

(2006). Female sex offenders: A comparison of solo offenders and co-offenders. Violence and Victims, 21, 339-354. doi:10.1891/vivi.21.3.339

40.

Vandiver

D. M.

Kercher

(2004). Offender and victim characteristics of registered female sexual offenders in Texas: A proposed typology of female sexual offenders. Sexual Abuse: A Journal of Research and Treatment, 16, 121-137. doi:10.1177/107906320401600203

41.

Verhoeven

K. J. F.

Simonsen

K. L.

McIntyre

L. M.

(2005). Implementing false discovery rate control: Increasing your power. Oikos, 108, 643-647. doi:10.1111/j.0030-1299.2005.13727.x

42.

Wijkman

Bijleveld

Hendriks

(2010). Women don’t do such things! Characteristics of female sex offenders and offender types. Sexual Abuse: A Journal of Research and Treatment, 22, 135-156. doi:10.1177/1079063210363826

43.

Williams

Bierie

(2015). An incident-based comparison of female and male sexual offenders. Sexual Abuse: A Journal of Research and Treatment, 27, 235-257. doi:10.1177/1079063214544333