Abstract
Male and female sexual victimization has been studied, although research has yet to strongly establish whether victimization occurs similarly across victim sex in the same types of locations, same areas of a city, or in places with similar structural factors. The current study fills these gaps by studying male and female sexual victimization in Los Angeles. Relying on official crime incident data, this study describes the types of locations unique types of sex crimes occur at for males and females, explores the spatial distribution of male and female victimization, and tests for structural factors as predictors of sex crimes in census tracts. Results of descriptive analyses suggested subtle differences in the types of locations (e.g., indoors vs. outdoors) for certain styles of sexual victimization for male and females, although a spatial point pattern test suggested male and female victimization followed similar spatial distributions throughout Los Angeles census tracts. Finally, spatial autoregressive models indicated theoretical variables, including a household activity ratio, land use type, and concentrated disadvantage were important predictors of both male and female sexual victimization. These findings suggest routine activities, crime pattern, and social disorganization theory policy applications could be effective for simultaneously reducing male and female victimization.
Introduction
In the United States, it is estimated that someone is sexually assaulted every 73 s (Brown et al., 2020). According to National Crime Victimization Survey (NCVS) data, the number of rape and sexual assault incidents rose by 63.9% between 2021 (324,500) and 2022 (531,810); thus countering the significant decline observed during the COVID-19 pandemic (Bureau of Justice Statistics, 2023). In California, a 2019 report by Johns et al. (2019) identified lifetime prevalence of sexual assault/harassment to be higher than the national average, with 86% of Californian women and 53% of Californian men experiencing victimization, compared to 81% of women and 43% of men nationally.
With rare exceptions, the majority of studies investigating sexual assault victims’ characteristics identified victims as most likely to be young, lower-income females (e.g., Felson et al., 2012; Felson & Cundiff, 2013). This is also reflected in the demographics of the population presenting to U.S. emergency departments after sexual assault (Vogt et al., 2022). Thus, and accordingly so, many researchers have dedicated time to understanding the nature of female sexual victimization, leaving the correlates of male victimization understudied.
Furthermore, the study of the spatial distribution of sexual assault and the structural correlates of sex crimes appear less documented, despite the relevance of such research for legal and practical actions to reduce crime. In the past, geospatial analyses have, for instance, been used to inform and decide on police strategies (e.g., Ferreira et al., 2012; S. Weisburd, 2021), support public planning solutions for crime prevention (e.g., Saraiva et al., 2022), and promote evidence-based policymaking (e.g., Wise & Craglia, 2007).
With growing interest and recent developments in empirical research on male sexual assault victimization (for a review see Thomas & Kopel, 2023), the possible (and probable) existence of differences between the spatial distribution of male and female sexual assault victimization is of interest. Indeed, reported differences in victims’ characteristics go beyond victims’ sex and span a whole array of relational, lifestyle, individual, societal, and emotional factors (e.g., Depraetere et al., 2018; Ioannou et al., 2017; Kimerling et al., 2002; Weiss, 2010) and there is thus no reason to believe that situational factors (i.e., location) would be the exception. This underscores the importance of studying both male and female sexual victimization, rather than simply focusing on one or the other, as well as the unique set of correlates of gender-based victimization. These are the primary objectives of this paper.
From a theoretical standpoint, gender-based differences in the spatial distribution of sexual assault could be explained through the lenses of crime pattern theory (CPT; Brantingham & Brantingham, 1982), routine activities theory (RAT; Cohen & Felson, 1979), and social disorganization theory (SDT; Shaw & McKay, 1942). CPT and RAT rely on a similar underlying assumption that crime is opportunistic by nature and shaped by the involved individuals’ movements in space and time, whereas SDT points to the role of structural factors that can break down cohesion and informal social controls in neighbourhoods. These theories have been useful for explaining predatory crimes like sex offenses (e.g., Franklin et al., 2012); however, whether the distribution of male and female sex crime victimization has different spatial patterns or varies due to neighbourhood characteristics is unknown.
The present study aims to answer the following questions about male and female sexual victimization using data from Los Angeles, California. First, do male and female victimizations for sex crimes take place in the same type of locations? Second, do male and female sex crime victimizations have similar spatial distributions in Los Angeles? Third, what neighbourhood characteristics are associated with male and female sex crime victimizations?
It is important to answer the current study’s research questions as, in the past, findings on hot spots overlap of different types of crime had implications for hot spots policing and its efficiency (Haberman, 2017). Understanding whether sex crime hot spots overlap for men and women provides valuable information for understanding the nature of these victimizations and how they might best be prevented.
Literature Review
Theoretical Frameworks
In studying the spatial distribution and structural correlates of female and male sexual victimization, this study relies on three theoretical frameworks: routine activities theory, crime pattern theory, and social disorganization theory.
First, routine activities theory (RAT) explains crime occurrence by the combination of three elements (i.e., conditions; motivated offender, suitable target, and absence of guardianship) in time and space (Cohen & Felson, 1979). According to the theory, certain locations host the conditions necessary for crime in the same place at the same time more often than other locations (Sherman et al., 1989). For example, bars have been identified as places that breed the RAT conditions and increase the risk of violence (Roncek & Maier, 1991; Savard et al., 2019; Sherman et al., 1989). Overall, previous empirical applications of RAT to sexual victimization recognized the explanatory power of the theory and supported its premise that the convergence of all three conditions necessary for crime to happen is more easily achieved in certain locations (Roncek & Maier, 1991; Savard et al., 2017).
Second, crime pattern theory (CPT) expands on RAT by proposing that when such locations enter an offender’s awareness space as either a node (i.e., key location) or place along their regularly travelled paths between nodes, there is a heightened risk of crime (Brantingham & Brantingham, 1995). According to CPT, land use type is an important variable to consider because crime generators (i.e., places many people go to without any criminal motivation) and crime attractors (i.e., places of known criminal opportunities that offenders frequent) are typically located in commercially zoned areas (Brantingham & Brantingham, 1995), whose higher concentrations of crime relative to residentially zoned areas may be explained by the intersection of offender and victim awareness spaces within those commercial areas (e,g., Kinney et al., 2008). Empirical applications of the CPT framework to the study of sexual victimization provided results supporting the theory as an explanation for sexual victimization patterns (Hewitt et al., 2018).
Third, social disorganization theory (SDT) relies on neighbourhood level factors, like residential instability (e.g., turnover in residents), ethnic heterogeneity (e.g., racial diversity), and low socioeconomic status (e.g., concentrated economic disadvantage), to explain the breakdown in collective efficacy and informal social controls that can suppress crime (Sampson et al., 1997; Sampson & Groves, 1989). The overall body of research on SDT suggests it holds substantial utility in explaining variation in crime across communities (Pratt & Cullen, 2005). Tewksbury et al. (2010) tested the validity of a SDT explanation of sex offenses with their study of Louisville, KY census tracts. Findings supported a positive statistical relationship between social disorganization, which was a composite score of eleven variables, and sexual assault. Similar support was provided by a study of dissemination areas, similar in size to census block groups in the US, in a large Canadian city, with some support for SDT concepts explaining sexual crime (Hewitt et al., 2018).
Spatial Distribution of Sexual Crimes
As previously stated, scholars have studied the spatial distribution of sexual crimes but there is more to discover about whether males and females are victimized in different areas or types of locations. With different defining characteristics (e.g., offender motivation, prerequisites for higher chances of “successful” crime commission, unique dynamics between the victim and offender), sexual crime hot spots could very well differ from non-sexual ones. Not unlike, perhaps, the spatial distribution differences between harm-weighted and raw crime reported by Fenimore (2019). Indeed, the latter study on “harm spots” (i.e., clusters of harm-weighted crimes) revealed the existence of unique, non-random distribution of crime harm, relative to raw unweighted crime. Furthermore, the types of places men and women are victimized at could also prove to be unique (Savard et al., 2017).
Macro Spatial Patterns of Sexual Crimes
Using time-geographic density estimations, Downs (2016) estimated activity spaces of 87 registered sex offenders based on their respective home and work anchor points. The obtained activity spaces were then combined into a single intensity surface highlighting areas of St. Louis most frequented by sex offenders, before rape incidents were spatially joined with the intensity surface to evaluate how well the latter predicted sexual crimes. Ultimately, as RAT and CPT predict, Downs’ (2016) results evidenced how sexual crimes were concentrated in offenders’ activity spaces. Given that low mobility has been consistently found to be a defining characteristic of (serial) sexual assaulters (Lundrigan & Czarnomski, 2006), motivated offenders may be more prone to act when potential targets are near their residencies.
Another approach to applying spatial (and temporal) analyses to sex crimes relies on the consideration of cultural characteristics of a given community to explain a pre-known heightened concentration of sexual crimes in given areas. Using kernel density estimation, time series, logistic regression, and random forest modelling, Clougherty et al. (2015) analysed sexual assault incidents at the University of Virginia and wider Charlottesville community between 1990 and 2015. Various spatial variables (bars, restaurants, fraternities, sororities, entertainment venues, hotels, residences of known sex offenders, downtown entertainment area [Main Street], and athletic fields) were incorporated in the tested models. Overall, the greatest and most significant predictor of sexual assault perpetration was the proximity to registered sex offenders’ homes, again lending credence to RAT and CPT. Proximity to Greek life, hotels, and restaurants also emerged as robust predictors of sexual crimes, echoing previous findings on the link between sexual assault and Greek life (e.g., Bannon et al., 2013; Franklin et al., 2012) and alcohol consumption (e.g., Franklin et al., 2012; Lisak, 2011).
Notably, both the studies by Clougherty et al. (2015) and Downs (2016) used cities as spatial units of analysis. Although macro-level and especially city-wide spatial analyses are common in criminological research (e.g., Ferreira et al., 2012; Saraiva et al., 2022), smaller geographical units also allow for the use of spatial analyses. The selection of a spatial unit is a function of a project’s aims and data availability but relying on sub-city spatial units of analysis is better suited for adequately understanding the spatial patterns of crime (Hewitt, 2021).
Micro Spatial Patterns of Sexual Crimes
Several empirical studies have investigated the spatial pattern of sexual crimes across smaller spatial units. The following studies do not make up an exhaustive list of empirical work linking space and sexual crimes but are valuable examples of the application of spatial analyses for studying sexual crime patterns.
Analyses at the neighbourhood and district levels by Muldoon et al. (2019) provided information regarding the characteristics of neighbourhoods where sexual and gender-based violence (SGBV) took place in the Ottawa-Gatineau metropolitan area. Of the eight identified neighbourhoods with a high SGBV concentration, three were in the downtown entertainment district, three were lower income neighbourhoods, and, surprisingly, one was a high-income area, and one was a suburb more than 20 km from downtown.
At the street and block level, a 2020 study by Miranda and van Nes produced interesting findings. Building upon previous work that has established the requirement of different types of spaces for different types of crimes (e.g., Hillier & Sahbaz, 2008), the authors detected correlations between sexual crimes, the number of people and women on the streets, local spatial integration, the land use of streets, and temporal aspects. At both the street and block levels, non-residential zones were the safest during the day due to high presence but were unsafe at night due to the lack of surveillance. Additionally, mixed land use areas emerged as safer than mono-functional areas (e.g., commercial only).
At the street segments (and intersections) level, Hewitt (2021) operated spatial concentration and kernel density analyses jointly with a spatial point pattern test to quantify the effects of victim’s characteristics (adult vs children) and sexual crime type (penetration, sexual contact, and sexual non-contact offences) on spatial distributions of sex crimes. Ultimately, analyses revealed that both child and adult sexual victimization was spatially concentrated in only a few street segments (and intersections) of Austin, TX. The degree to which sexual offences clustered was a function of sexual crime type, thus highlighting the importance of disaggregating sexual offences when possible.
In the sexual crime and space literature, the work of Ceccato and Paz (2017) stands out because of their choice of spatial unit: 62 metro stations in São Paulo, Brazil. In line with research at other geographic levels (e.g., Muldoon et al., 2019), the authors reported that high levels of activity in metro stations went hand in hand with high concentrations of sexual crimes. Sexual crimes in particular were committed during the morning and afternoon rush hours; potentially highlighting insufficient guardianship due to more-than-usual users’ presence. Finally, stations with high concentrations of sexual crimes were also those most likely to host other types of crimes and general public disorder. The latter finding suggests that the spatial distribution of sexual and non-sexual crimes may be similar because of the accumulation of environmental and societal risk factors for all types of crime.
Gender Differences in Sexual Victimization
Despite a growing interest in male sexual victimization in the last two decades (e.g., Carr & VanDeusen, 2004; Luetke et al., 2020; Turchik, 2012) that went along with societal changes of mores and the recognition that sexual assault is not only a male-on-female phenomenon, the vast majority of the literature on sexual assault focuses on female victims. This could be the result of voluntary methodological choices to focus on the population statistically most likely to be sexually victimized and a lack of available data on male victims. In any case, scholars have reported differences in victim, offender, and incident characteristics depending on whether the victims of sexual assault were male or female.
Sexual Victims’ Characteristics
In male-on-female sexual assault incidents, the victims’ age is most likely between 16 and 19 (Ioannou et al., 2017) with victims most likely belonging to lower economic classes (Vogt et al., 2022). Female victims are also most likely to have been victimized by someone they are/have been in a relationship or are acquainted with (Ioannou et al., 2017).
In male-on-male incidents, victims usually are between 20 and 30 (McLean, 2013) and White (Choudhary et al., 2011). Findings on the victims’ sexual orientation differ, with some scholars reporting a majority of male victims being heterosexual (Hodge & Canter, 1998; Ioannou et al., 2017; Isely & Gehrenbeck-Shim, 1997) and other researchers identifying queer individuals as more at-risk (Kimerling et al., 2002; Walker et al., 2005). Again, methodological limitations, such as samples made entirely of gay or bisexual men, are the cause for such disparities.
Much rarer and less documented are the cases of female-perpetrated sexual assault. Munroe and Shumway’s (2020) work with victims of such crimes informs us that childhood victimization is more common than adult victimization and that, like their male counterparts, female assaulters act alone and are close to their victims (especially a friend, dormmate, roommate, classmate, neighbour, or co-worker in 56.6% of the cases reviewed).
Crime Location and Victim-Offender Relationship
Importantly here, the findings that both the victims and perpetrators of male-on-male sexual assault tend to be heterosexual (Ioannou et al., 2017) could have implications for the spatial distribution of such incidents. This could, for instance, be because if the persons involved are less acquainted or not in a romantic relationship (relative to male-on-female incidents), the process of selecting the optimal location to commit the sexual crime could be impacted in one way or another. Again, two environmental criminology perspectives can be valuable in explaining the role location plays in sex crimes: routine activities and crime pattern theories.
In a study where more than 73% of the female victims of sexual assault were acquainted with their male offender, Hilden et al. (2005) reported that 60.4% and 35.7% of the incidents took place in a private home (i.e., offender or victim’s house) or a public space respectively. Similar findings have been reported with a sample of male serial sex offenders who mainly assaulted their victims (all strangers to the offenders prior to the assault and either female, children, or both) in private places (64.8% vs. 35.2% public places) that were either the offenders’ (29.1%), the victims’ (21.1%), or neither (49.9%; Hewitt & Beauregard, 2014).
An early study (Groth & Burgess, 1980) on male-on-male sexual assaults carried out by interviewing offenders themselves identified 75% (12 out of 16) of these offenders as strangers to their victims. It bears mentioning, though, that this finding clashes with most research on victim-offender relationships in male-on-male sexual assaults (Hodge & Canter, 1998; Isely & Gehrenbeck-Shim, 1997; Mezey & King, 1989). According to Ioannou et al.'s, 2017 systematic literature review of male-on-male incidents, offenders were acquainted to their victims in 67.4% of the cases, even though “being acquainted” sometimes referred to connections established less than 24 hr prior to the offence. Additionally, the most common locations where incidents took place were the victims’ homes (22.3%), the offenders’ homes (17.7%), and non-listed public areas (13.2%; Ioannou et al., 2017). In a rare study investigating female sexual offenders, Morgan and Long (2018) reported that of the 13 reviewed cases with solo female offenders, seven took place in the victim’s home (vs. one at the offender’s, four at school, and one outside).
Overall, despite few exceptions (e.g., McLean et al., 2005 who found that both male and female victims were more likely to be sexually assaulted in public spaces [around 30% of the cases in the sample for both sexes]), it would appear that there is little difference in terms of victim-offender relationship and assault location preferences based on the victims’ sex. Assaulters tend to know their victims and prefer inside locations, likely their or their victims’ private homes, to commit sexual crimes. Hot spots-wise, if assaulters of both male and female victims predominantly share the same characteristics (i.e., young, White, heterosexual) and spatial preferences for their sexual crime commission (i.e., indoor and familiar location), we could expect similar general sexual crime hot spots instead of gendered ones.
However, different dynamics in the victim-offender relationship in the case of male or female victims could still impact the actual crime location, rather than the initial preferences. For instance, if the assailant’s home is further away from the initial meeting point than the victim’s, and the victim offers more physical resistance (possibly the case with male victims), perhaps the latter option will be preferred by the offender. Likewise, the superposition of the victim and offender’s activity spaces will also impact the spatial distribution of sexual crimes in a way that can reflect gendered habits and/or partying and meeting up patterns/styles.
Correlates of Female and Male Sexual Victimization
Comparative work on the characteristics of victims (males vs. females) and the situational factors surrounding their victimization is scarce. One explored avenue is the connection between college lifestyles and sexual victimization.
Situational factors inherent to college campus life may facilitate sexual victimization for females. The female student body experiences sexual victimization at a higher rate than any other community, and it is estimated that one in five women will be the victim of sexual assault during their college years (Clougherty et al., 2015). Reported prevalence rates vary considerably based on methodology, and obtaining accurate and representative statistics is even harder for males given prevalent societal perceptions that sexual victimization is a gender-based problem (i.e., a robust barrier to reporting). Consequently, prevalence rates amongst college male students have ranged from as low as 0.2%, when only female-on-male completed or attempted rape was considered (Tjaden & Thoennes, 2000), to as high as 65.5% (Schuster et al., 2016), when the sexual violence definition was broadened and respondents were asked to recall past experiences for a longer timeframe (for a review, see Depraetere et al., 2018).
In a 2012 study, Franklin et al. (2012) used a sample of 2,230 female students from eight universities in Texas to apply RAT to sexual assault victimization. Their results identified: (a) self-control deficits, (b) increased number of days spent on campus, and (c) increased frequency of partying as linked with increases in sexual assault victimization. Increased partying frequency being linked with an increase in sexual victimization falls under the expectations of RAT as a simple matter of exposure to potential motivated offenders and opportunities for alcohol consumption and sexual miscommunication (see also Franklin, 2010; Schwartz & Pitts, 1995). Self-control deficits as a predictor of victimization also fall within expectations since previous studies linked them to sexual victimization incidents involving alcohol consumption (e.g., Franklin, 2011); itself being more likely in partying contexts. Finally, aligning with previous research on female date and acquaintance rape (e.g., Fisher et al., 2000), the finding that the number of days spent on campus was positively associated with an increased risk of sexual victimization not only supported RAT through the argument of exposure, but also emphasizes the view that situational factors inherent to campus life (e.g., guardianship, Greek affiliation, partying opportunities) may facilitate sexual assault perpetration.
Comparatively, Tewksbury and Mustaine (2001) revealed that, unlike female sexual assault victimization, victimization experienced by males was not correlated with alcohol consumption. Additionally, the authors also reported that Greek affiliation or frequently engaging in fraternity parties did not put men at greater risks of sexual assault victimization. These results, however, differ from those obtained by Luetke et al. (2020) who reported that over a quarter of their male fraternity respondents experienced any sexual assault victimization since entering college.
Eventually, Tewksbury and Mustaine (2001) concluded that “the patterns and sources of male and female sexual assault victimization risks are unique” (p. 174), hence warranting the need for further research on male and female sexual victimization. A closer inspection of the gendered spatial distribution of sexual victimization, as well as that of the spatial characteristics of sexual crime hot spots, is a first step in that direction.
Current Study
To date, research has not adequately described whether male and female sex crime victimizations happen in similar types of places, in the same parts of a city, or whether the same neighbourhood level factors explain the prevalence of both male and female victimization. The current study fills in these gaps in the literature by asking three research questions. First, did male and female victimizations for sex crimes take place in the same type of locations? To answer this question, Los Angeles sex offense data were disaggregated by victim sex, type of sex offense (penetration, contact, and non-contact), and location. Second, did male and female sex crime victimizations have similar spatial distributions in Los Angeles? A spatial point pattern test was used to assess whether there were different distributions of sexual victimization for male and female victims across Los Angeles. Third, what neighbourhood characteristics were associated with male and female sex crime victimizations? To answer this question, census and other data were collated to test for effects of social disorganization related and other variables in Los Angeles census tracts in a recent sub-sample of Los Angeles data.
Data and Methods
The city of Los Angeles provides a valuable context for our work. As of 2024, Los Angeles’ population of nearly 3.9 million made it the second largest city in the U.S. (Census Bureau, 2025). Of note is the quality of the crime data publicly available from the city of Los Angeles. Not all cities make sex crime data public, nor do they provide as many variables that were needed for the current study. The publicly available crime dataset from Los Angeles contained many important variables for sex crime incidents, including victim characteristics, location type, and geographic location. Given the size and data availability, Los Angeles is a valuable site for answering our research questions.
Crime data were sourced from the publicly available Los Angeles Police Department (LAPD) crime incident data. The dataset used was the combination of the historical LAPD crime dataset (2010–2019) with more recent data with complete records through 2023. 1 Data from 2010-2019 and 2022-2023 were used for a descriptive analysis of the location of penetration, sexual contact, and sexual non-contact offenses, as well as a spatial point pattern test (SPPT) of the distribution of male and female victimization. Data from 2020 and 2021 were excluded because the pandemic may have shifted spatial crime patterns during lockdown orders (Yim & Riddell, 2024). After restricting the sample window, a total of 47,765 sexual victimizations remained; 42,613 had a female victim and 5,048 had a male victim. 2 For these cases, the mean age of victims was 26.26 (male average = 26.32; female average = 26.25). The ethnicity breakdown for victims in these cases was 50.5% Hispanic (male = 49.4%; female = 50.7%), 22.9% white (male = 23.3%; female = 22.9%), 17.6% black (male = 18.6%; female = 17.5%). 2.9% Asian/Pacific Islander (male = 1.9%; female = 3.1%), and 6.1% other (male = 6.7%; female = 5.8%).
Los Angeles census tracts were the unit of analysis for the SPPT and for spatial simultaneous autoregressive (SAR) lag models for victimization rates. For estimating the SAR models, crime data from 2022 and 2023 were pooled together, with independent variables sourced from the American Community Survey (ACS), the California Department of Alcoholic Beverage Control, and Los Angeles City Planning. Crime data from 2022 and 2023 were combined to create a more stable measure of male sex crime victimization given the rarity and volatility of male victimization, and the selection of these years avoid possible COVID-19 related effects on sex offenses. Relying on just the most recent years of data allowed for joining the crime data with ACS estimates for independent variables that captured other concepts that may be related to explaining sex crime victimization patterns. After removing observations due to missingness, 993 census tracts were analysed using SAR models, which was approximately 89% of all 1118 whole or partial census tracts within the city boundary of Los Angeles.
Sex Offenses
One of the difficulties with studying male sex crime victimization was the low number of official crime reports. Aggregating multiple types of sex offenses together created a larger number of observations that made it easier to conduct comparison tests between male and female victimization patterns. To measure sex offenses, the categorization process used by A. N. Hewitt (2021) was employed. We have categorized sex offenses into penetration (forcible rape, sexual penetration with a foreign object, and sodomy), sexual contact (attempted rape, battery with sexual contact, oral copulation, lewd/lascivious acts with a child, lewd conduct), and sexual non-contact offenses (indecent exposure, peeping tom). 3 An aggregate sex crime index was also analysed using the hot spots maps and SAR models.
Routine Activities Theory Variables
Two independent variables were included to assess the utility of routine activities theory in explaining variation in male and female sexual victimization: the number of on-sale alcohol licenses per square mile and the household activity ratio. First, a list of active on-sale retail alcohol licenses was pulled from the California Department of Alcoholic Beverage Control. Licenses issued after 2023 were excluded, and the number of on-sale alcohol licenses per square mile was calculated to measure the density of bars in census tracts. It is important to account for bar prevalence given that studies have indicated predatory crimes like sex crimes can cluster near bars (Savard et al., 2017; Sherman et al., 1989). However, it has not been clearly determined if bars pose an equal risk of sexual victimization to males and females. We hypothesized that the prevalence of bars, measured as the number of bars per square mile, would be positively associated with both male and female victimization counts.
Second, a household activity ratio was calculated by dividing the sum of the number of married females in the labor force and the number of single-lead households by the total number of households. This ratio was then multiplied by 100 to create a percentage, and it was used to capture guardianship. These data came from the ACS 5-year estimates for 2023. Prior research has used household activity ratios to approximate the percentage of households that are at a greater risk of victimization due to less guardianship (i.e., only one capable adult; Pratt & Cullen, 2005). We hypothesized there would be a positive relationship between the household activity ratio and sexual victimization, given that higher values of the ratio indicate lower levels of guardianship.
Crime Pattern Theory Variables
Two independent variables were included to measure crime pattern theory concepts: percent of land in a census tract zoned commercial and percent of land in a census tract zoned residential. Zoning data from the Los Angeles open data portal were accessed to determine the land use type of areas within the city. 4 The zoning data was spatially joined to census tracts, and then the percent of land zoned commercial and the percent of land zoned residential were calculated. Tests of crime pattern theory have pointed to the importance of including the type of land use, especially the percent of commercially zoned land, in geospatial analyses due to higher crime concentrations in areas with a higher percentage of the area zoned as commercial (Kinney et al., 2008; Twinam, 2017). Given past research findings, we hypothesized a positive relationship between percent commercial and sexual victimization for males and females, and no statistical relationship between percent residential and sexual victimization for males and females.
Social Disorganization Theory Variables
A concentrated disadvantage index, a diversity index, and residential turnover were used to capture the socio-economic status, ethnic heterogeneity, and residential instability social disorganization theory concepts. All of these variables were created using ACS 5-year estimates for 2023. First, a standardized concentrated disadvantage index was formed following a confirmatory factor analysis process. This index was comprised of the poverty rate, percentage of residents that had dropped out of high school, the percentage of female headed households, the reciprocal of the median home value, and the reciprocal of the median household income. This concentrated disadvantage index was constructed using variables that have been included in various combinations in concentrated disadvantage indexes in prior research (Allen & Feldmeyer, 2022; Barnett & Mencken, 2002; Blumenstein & Jasinski, 2015; Holder et al., 2022; Sampson et al., 1997), and it had a Cronbach’s alpha of 0.8445, which indicates high internal consistency. We hypothesized higher levels of concentrated disadvantage would be associated with higher levels of male and female sexual victimization.
Second, a diversity index was calculated to capture racial heterogeneity. This index was calculated by summing the squared proportions of each race category and subtracting them from 1, as seen in Equation (1):
where pi is the proportion of the population represented by each racial group (Greenberg, 1956). The five racial groups for this study were non-Hispanic white, Black, Hispanic, Asian, and other. Values closer to 0 indicate a more homogenous population and values closer to 1 indicate a more heterogenous population. This is a common way past research has measured heterogeneity, and Pratt and Cullen (2005) found this to be a moderately strong and stable predictor across research. Based on social disorganization theory, we hypothesized a positive relationship between the diversity index and male and female sexual victimization.
Third, residential turnover was captured with the percentage of residents in a new residence versus the previous year. This measures the residential mobility within census tracts, and residential mobility has some empirical support as a macro-level predictor of crime (Pratt & Cullen, 2005). We expected more residential turnover to be associated with more male and female sexual victimization.
Control Variables
Additional control variables were collected from the ACS 5-year estimates for 2023. Controls included the ratio of male to female residents, the Gini index (i.e., indicator of income inequality where 0 = perfect equality and 1 = perfect inequality), the percentage of vacant dwellings, the percentage of the population under 18, the percentage foreign born (i.e., the percentage of residents born outside of the U.S.), and the percentage of residents that were divorced. Many of these covariates have been linked to crime levels by prior research, with varying degrees of support (Pratt & Cullen, 2005). Descriptive statistics for all dependent and independent variables used in the SAR models are in Table 1.
Descriptive Statistics for Los Angeles Census Tracts (n = 993).
Analytic Plan
The analyses for this paper proceeded as follows. First, there was a breakdown of the types of sex offenses (penetration, contact, and non-contact) by location type for both male and female victims. This analysis was descriptive, and there were no tests for statistically significant differences in male and female victimization locations. The purpose of this step was to simply understand the types of places where male and female victimizations occurred.
Second, crime density maps of male and female sexual victimization were compared and the results of a SPPT are presented. The estimated SPPT assessed whether there were statistically significant differences (at the .05 level of significance) in the proportional distribution of male and female sex crime victimizations in census tracts. Comparing differences in proportions, rather than raw counts, between the two sets of data was more desirable and reduced the likelihood of falsely identifying statistically significant differences in the spatial patterns of male and female victimization (Wheeler et al., 2018). This approach was used due to the disparity in male and female victimization, and the census tract (n=1,118) was selected as the areal unit in order to have enough statistical power to adequately detect differences in point distributions (see Drake et al., 2022 for a discussion on statistical power). 5
Last, SAR models assessed the effect of the same independent variables on male and female victimization counts in 993 census tracts in Los Angeles. Only crime data from 2022 and 2023 were used for these models, and some census tracts were dropped due to missing values for independent variables. The SAR models accounted for the lag of the dependent variables using a queen’s contiguity matrix, which is important due to the results of tests for spatial autocorrelation (Moran’s I = 0.131 for the male victimization rate; Moran’s I = 0.305 for the female victimization rate). Crime pattern, routine activities, and social disorganization theory variables were entered in separate models first before being combined, and then all theoretical and control variables were combined into a final model. Statistically significant effects at or below the .05 level of significance indicate enough evidence to suggest a statistical relationship between the independent variable and the outcome measure.
Results
Descriptive Analysis of Crime Locations
The breakdown of victimization type and location for males and females is presented in Table 2. Here, it is important to recall that in this section, frequencies not probabilities, are discussed. The information presented in Table 2 should thus be interpreted accordingly. Penetration offenses were the most common type of sexual victimization for males and females. Overall, there appears to be a similar breakdown of male and female victimization by indoor versus outdoor locations for penetration offenses only. A majority of penetration offenses for both males and females took place in indoor locations, particularly places that people sleep in (residences and hotels) rather than in businesses. A higher percentage of indoor penetration offenses with a male victim took place in business and other places, as well as at schools as compared to female victims.
Descriptive Analysis, Location Breakdown by Victimization Type.
Sexual contact offenses most commonly happened indoors for male victims relative to female victims. Of the indoor violations, there were similar splits between the types of indoor places across victim sex. Notably, a greater percentage of female sexual contact victimizations occurred in a bus or train station, parking lot or garage, or street, highway, road, or alleyway as compared to male victimizations.
For non-contact offenses, female victimizations most commonly took place indoors compared to male victimizations. When considering the type of indoor location, females were in places people might sleep at a higher rate than males, and a larger proportion of male victimizations were at business locations. A greater percent of males experienced non-contact victimization at various outdoor locations than females, including parks and playgrounds, parking lots and garages, and street, highway, road, or alleyways.
Male and female victims experienced penetration, contact, and non-contact offenses at different types of places. Regardless of victim sex, penetration offenses were primarily experienced indoors, followed by contact and non-contact offenses. For males, a greater proportion of non-contact offenses were experienced at business or other indoor locations, with a smaller proportion of contact and penetration offenses happening at such places. This was slightly different for women, as similar proportions of contact and non-contact offenses took place at businesses and other indoor places, with a greater percentage of penetration offenses happening at places where people might sleep as compared to the other sex crime categories.
SPPT Results
Figures 1 and 2 contain maps of the male (Figure 1) and female (Figure 2) sexual victimizations per square mile in Los Angeles census tracts (n = 1,118). When comparing these maps, male and female sexual victimization appear to occur in generally similar areas throughout Los Angeles. There are some subtle differences, but the highest concentrations of male victimization look to be overlapping with the highest concentrations of female victimization.
Male sexual victimization density in Los Angeles, all years of data.
Female sexual victimization density in Los Angeles, all years of data.
Results of the SPPT are displayed in Figure 3, with circles marking the census tracts that had a statistically significant difference in sexual victimization. The Global S value, which is an indicator of similarity, was 0.991, suggesting a high degree of similarity in the spatial patterns of male and female victimizations(Steenbeek et al., 2020) In fact, there were only 9 (<.1%) census tracts with a statistically significant difference identified. Based on these results, male and female sexual victimizations appear to have similar spatial distributions in Los Angeles.
SPPT results, sexual victimization in Los Angeles.
SAR Results
The results of SAR models for all male sex crime victimization are contained in Table 3. Theoretical variables were entered one at a time into models before combining them together in model 8 and then finally with controls in the full model (model 9). Models suggest multiple theoretical and control variables are associated with male victimizations.
SAR Results, Male Victimization Rates.
p < .10; *p < .05; **p < .01; ***p < .001.
First, the prevalence of bars and the household activity ratio remained positively signed, statistically significant predictors of male victimization across all models. These findings support our hypotheses and routine activities theory. Crime pattern theory was not as equally supported, given that percent commercial was not statistically significant after other theoretical variables and controls were included (models 8 and 9). Interestingly, percent residential remained a statistically significant predictor across all models, and higher percentages of land zoned as residential was associated with lower male sexual victimization rates.
There was also mixed support for social disorganization theory variables. The concentrated disadvantage index was also positively associated with male victimization rates in the full model, but racial heterogeneity and residential turnover were not statistically significant variables. Vacancy rates and percent divorced were also positively signed and statistically significant. The percentage of the population under the age of 18 in a census tract was negatively correlated with male victimization rates. Other variables failed to exert statistically significant effects at the .05 level of significance in the full model, including the male to female ratio, the Gini index, and percent foreign born.
Table 4 presents the SAR regression results for female victimizations. Routine activities theory was supported by the results of models 1, 2, 8, and 9, with the bar density and household activity ratio variables found to be positively associated with female sex crime victimizations. Crime pattern theory was better supported by the female victimization models, given that percent commercial had a statistically significant, positive association with female victimization rates as hypothesized. Percent residential exerted a similar, statistically significant negative effect on female victimization rates as with male victimization rates.
SAR Results, Female Victimization Rates.
p < .10; *p < .05; **p < .01; ***p < .001.
For social disorganization theory, there was once again a lack of full support. Only concentrated disadvantage was found to consistently exert a positive relationship with female victimization. Racial heterogeneity and residential turnover were less consistent predictors across models and were not statistically significant factors after including controls in model 9. The male to female ratio was a positive predictor of female victimization, as was the vacancy rate. Findings suggest the Gini index, the proportion of residents under 18, percent foreign born, and the rate of divorce were not significant predictors of female victimization.
When comparing the full models for male and female victimizations, there were some similarities in the effect of theoretical and control variables. The concentrated disadvantage index, prevalence of bars, percent residential, and vacancy rates were positive predictors of both male and female victimization. Racial heterogeneity, the Gini index, and percent foreign born were not associated with either outcome. For the remaining variables, there were different impacts on sex crime victimization by victim sex. Interestingly, the male to female ratio was positively associated with female victimization but not with male victimizations. The percent of the population under 18 was negatively associated with male victimization, but there was a null effect for female victimization. Percent divorced had no effect for female victimization but a positive relationship with male victimization.
Discussion
Scholars have dedicated time to studying the spatial distribution and contextual factors associated with sex crimes, but there is a great need to study potential differences in sex crime location and correlates across males and females. Our work begins exploring the spatial differences in sex crime victimization in Los Angeles by examining location types of sex offenses, spatial distributions of male and female victimization, and comparing SAR models predicting male and female victimization to assess any similarities in the correlates of sex crime victimization. Our findings suggest there may be some meaningful differences in the types of locations male and female victimizations occur. However, the male and female sexual victimizations clustered in similar areas in Los Angeles. While there were shared factors predicting sex crime victimization in census tracts, there were some differences as well. Each research question of this study provides valuable insight into the place-based characteristics of sexual victimization across sexes.
Do Male and Female Victimizations for Sex Crimes Take Place in the Same Type of Locations?
There were some differences in the types of locations male and females experienced sexual victimization in Los Angeles. As previously stated, the results from our descriptive analysis could be explained through the lens of routine activities theory. For example, one of the most notable differences in location types was for sexual contact offenses, with male victimizations happening indoors (63.38% of male sexual contact victimizations took place indoors) more frequently than for women (50.86% of female sexual contact victimization took place indoors). It could be that the conditions necessary to perpetrate sexual contact offenses against men intersect more frequently in time and space indoors than outdoors. Whilst virtually all (89%) victims of sexual incidents use some sort of resistance strategy to stop the attack (Weiss, 2010), it is understandable that physical resistance will bear greater chances of success in the case of male victims. This makes coercion as part of the attack all the more important for the completion of the act and, for instance, non-violent coercion tactics such as participating in drinking games; found to lead to higher male victimization rates (Johnson & Stahl, 2004), are likely to take place indoors.
Our result that a greater percentage of female sexual contact victimizations occurred in public places (bus or train station, parking lot or garage, or street, highway, road, or alleyway) as compared to male victimizations is altogether unsurprising given the prevalence of street harassment and unwanted sexual attention that women are subjected to on a daily basis (Fairchild, 2022). Indeed, it is frequent for unwarranted “compliments” or any similar request at attention, to be accompanied by physical gestures such as pinching, touching, and grabbing (e.g., Fairchild, 2022; diGennaro & Ritschel, 2019) that can be categorised under lewd conduct or, depending on the circumstances of the case, attempted rape (i.e., sexual contact offences in our methodology).
Interestingly, for non-contact offenses, female victimizations were more likely to have taken place indoors compared to male victimizations. When considering the type of indoor location, females were most likely victimised in places people might sleep (i.e., hotels, residences) at a higher rate than males, and a larger proportion of male victimizations were at business locations. The very nature of the two offences categorised as non-contact in our methodology (peeping tom and indecent exposure) may explain such differences.
On one hand, as a type of voyeurism, a peeping tom incident refers to the active viewing of a non-consenting person engaging in private activities for the purpose of sexual gratification (Lister & Gannon, 2023). Peeping toms particularly aim to observe others undressing, taking a shower, using the bathroom, or engaging in sexual activities (Långström & Seto, 2006); these such activities typically occur behind closed doors, making private places more prone to being targeted by voyeurs. On the other hand, as a type of exhibitionism, indecent exposure is a practice that involves exposing one’s genitals to a person who does not expect nor consent to it (American Psychiatric Association, 2013).
Importantly, a much greater proportion of the non-contact offenses experienced by women were peeping tom incidents (approx. 73%), whereas only about 15% of male non-contact cases were peeping tom events. This variation in victimization experience, with women experiencing peeping toms at higher rates and male victims having experienced more indecent exposure, could explain why female victims of non-contact offenses were more victimized indoors at places like residences and hotels at higher rates than males. The finding that a larger proportion of non-contact male victimizations were at business locations could simply be the result of greater opportunities for indecent exposure in public areas (e.g., public bathrooms, gym changing rooms). Overall, these tentative explanations to the descriptive findings we reported fall within the expectations of RAT and CPT and tap into the argument of overlapping activity spaces for victims and offenders.
Do Male and Female Sexual Victimizations Have Similar Spatial Distributions in Los Angeles?
From our constructed hot spot maps and SPPT, two realizations come to mind: (1) there was a great degree of overlap between male and female sexual victimization locations across Los Angeles, and (2) sexual victimization distributions for males and females are very similar. Differences in the distribution of victimization were identified in less than one percent of census tracts, suggesting males and females in Los Angeles were likely to experience victimization in the same general areas throughout the city. Focusing resources in areas to reduce sexual victimization may serve to reduce incidences for both males and females, although the actual types of locations within the census tracts at a higher risk for sex crimes may differ.
How Do Neighbourhood Characteristics Affect Male and Female Sex Crime Victimizations?
Some, but not all, neighbourhood characteristics predicted both male and female sexual victimization. There does appear to be clear support for routine activities, but mixed support for crime pattern and social disorganization theories when reviewing the results of our SAR models. As expected, lower guardianship levels (i.e., higher household activity ratios) were found to predict higher rates of victimization for males and females. The prevalence of bars was also positively associated with male and female victimization, suggesting that this type of crime generator can be equally predictive of sexual victimization across sexes. This again corroborates research conducted in this area that has found liquor establishments to generate a significant amount of sex crimes (Hewitt et al., 2018; Tewksbury et al., 2010).
Another finding that could be explained through routine activities theory was the positive association of vacancy rates with male and female victimization. Prior work suggests vacant buildings, especially those that have been abandoned, can play host to violent crimes (Fox et al., 2021; Locke et al., 2023; Spelman, 1993). Such locations often lack capable guardianship and would provide the conditions needed for sex crime to occur.
Interestingly, there was only partial support for crime pattern and social disorganization related variables in our models. Census tracts with a higher percentage of land zoned as commercial were at a higher risk for female, but not male, victimizations. The concentrated disadvantage index was positively associated with male and female sexual victimization, while residential turnover was only positively associated with female victimization in a partial model and the diversity index was not statistically significant in male or female SAR models. These findings suggest the conditions created by socio-economic disadvantage play host to sexual victimization, regardless of victim sex, yet other factors that are theorized to breakdown social controls have comparably less utility.
Given that male and female sexual victimizations seemed to cluster in the same areas of Los Angeles, there are clear policy implications. Hot spots policing should be an effective strategy for reducing sexual victimization (Braga et al., 2012; D. Weisburd & Telep, 2014). As Haberman (2017) noted, heterogeneity of hot spots by crime type limits the efficiency of hot spots policing in reducing crime in general and instead is more likely to affect just one or two types of crimes that do cluster in the policed hot spot. Our work suggests there do not need to be similar concerns about the heterogeneity of hot spots by victim sex, at least in the context of sexual victimization, and efforts to deter potential sex crimes in sexual victimization hot spots should be able to reduce the risk of victimization for both males and females.
Limitations and Future Research
Reliance on official data is one of the primary data limitations of this study. Previous research has identified sexual assault as one of the most underreported violent crime types. Using administrative data from the U.S. Bureau of Justice Statistics and the NCVS, Jaitman and Anauati (2019) estimated the dark figure of sexual crimes in the U.S. to be 62.7% (versus 35.9% for robberies for instance). This estimation means that, in the country, only around 37 out of 100 sexual crimes are known to the police. Although the U.S. fares better than the United Kingdom (79.9%) and Latin America and the Caribbean (90%) dark figure-wise (Jaitman & Anauati, 2019), alarmingly high rates of sexual assault (known to the police or not) is a worldwide problem that deserves substantial empirical attention. Furthermore, male victims of sexual assault may have additional reservations when it comes to reporting victimization. Indeed, a 2026 systematic review of sexual trauma disclosure in boys and men by Pilkington and colleagues identified cultural (perceived masculine norm violation) and societal (minimal public acknowledgement and validation of their trauma) barriers to sexual victimization reporting and discussion. Another limitation of the LAPD data is the coding of victim sex. The data were coded as male, female, or unknown, and there was not an additional variable to indicate whether a victim was transgender.
Bar prevalence was measured, but with the current analyses, we were not able to isolate the effect of specific bars on victimization by sex. In other words, while bars in general might increase the prevalence of sexual victimization, this study did not determine if male victimization clusters at different bars than female victimization. Similarly, with places experiencing concentrated disadvantage, it is unclear whether certain disadvantaged areas were more likely to host male victimization or female victimization over others.
This study has two common limitations of geospatial research, including the constraint of the data available being coded to the block level and the modifiable areal unit problem (Ratcliffe, 2010). Data being aggregated up to the block level could have masked spatial heterogeneity and shifted crime locations to be assigned to different census tracts than they occurred in. Furthermore, aggregating data to census tracts instead of a different areal unit, like equivalently sized grid cells, could have generated different representations of crime distributions (Ratcliffe, 2010).
An additional limitation is that this study used a convenience sample of one U.S. city predicated on the availability of crime data with victim characteristics. While the nature of the data and size of Los Angeles made this a quality option for answering our research questions, the unique demographic makeup, cultural differences, and city dynamics of Los Angeles limit the generalizability of our findings. Our descriptive analysis of location types is also limited in interpretability, and it does not indicate differences in probabilities or explain why male and female victimization happened across location types. Future research should replicate our approach to assess sex crime patterns in other cities. It would also be interesting for future research to investigate to what extent sexual and non-sexual hot spots overlap and whether situational and neighbourhood level factors carry the same weight in offender’s decision-making and location selection processes.
Conclusion
Overall, our findings indicated that there was some overlap in male and female victimizations, but there were some differences and unique predictors dependent upon victim sex that have important theoretical implications. We concur with Hewitt et al. (2018) in their conclusion that crime pattern theory and social disorganization theory can explain sexual crime locations as, taken together, our findings indicated both crime opportunity explanations and structural correlates of crime were needed to explain both male and female sexual victimizations. We also add that the opportunity-based model of routine activities theory provides a valuable explanation for sexual crime location. In particular, lower guardianship emerged as a reliable predictor of higher rates of sexual victimization for both males and females in our work.
Footnotes
Funding
The authors received no financial support for the research, authorship, and/or publication of this article.
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.