Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
3 Gender and Violence in the United States: Trends in Offending and Victimization Karen Heimer and Janet L. Lauritsen There has been increasing attention in social science to the recent U.S. decline in violent crime, which followed a period of large increases in vio- lence (e.g., Blumstein and Wallman, 2000; Zimring, 2006). Interestingly, almost all of the analyses of crime trends over the past few decades have been silent on the issue of gender (for an exception, see Rosenfeld, 2000). While it is true that female offending accounts for a relatively small per- centage of very serious violent offending, such as homicide and robbery, women accounted for roughly 25 percent of arrests for simple assaults and 21 percent of arrests for aggravated assaults in 2004, according to the FBIâs Uniform Crime Reports (UCR). Moreover, by 2004 women accounted for 44 percent of simple assault, 34 percent of aggravated assault, and 33 percent of robbery victimizations, according to the National Crime Victim- ization Survey (NCVS) (Bureau of Justice Statistics, 2006). The experiences of women and girls therefore are important for understanding crime in the United States. Some scholars suggest that an examination of changes in crime over time does not require attention to gender because the gender composi- tion of the population does not change rapidly enough to affect aggregate crime rates substantially (Blumstein and Wallman, 2000, p. 10). However, this argument presumes that the âgender gap,â or relative rates of female and male crime, remain constant over time. Perhaps it is reasonable to ignore gender in examinations of short-term trends, but research on long- term trends reveals important gender differences in both victimization ( Â Lauritsen and Heimer, 2008) and offending (OâBrien, 1999). Moreover, most researchers would argue that examining long-term trends is essential 45
46 UNDERSTANDING CRIME TRENDS for contextualizing shorter term spikes and drops in crime rates. Under- standing crime trends in the United States would therefore seem to require consideration of female as well as male experiences with crime over a sub- stantial period of time. In addition, a full understanding of crime trends necessitates attention to victimization as well as offending. Focusing on female and male experi- ences with violence highlights this point. Women consistently are less likely than men to be both violent offenders and victims. Yet the gender difference in rates of victimization is smaller than in rates of offending, for the most part. For example, women are much less likely than men to kill or rob. They are also less likely than men to be killed or robbed, but the difference between female and male rates is smaller in the case of victimization. This emphasizes the need for research addressing female as well as male trends and offending as well as victimization. Shifts in female victimization and offending may be of little unique significance if they simply mirror male shifts. Thus, a National Academies report on violence against women concluded that careful research comparing long-term trends in female and male violence is a priority (National Research Council, 2004). This chapter seeks to broaden knowledge of offending and victimiza- tion trends in the United States by reporting and examining changes in (1) female and male violent offending, (2) female and male violent victim- ization, and (3) gendered patterns of victim-offender relationships in violent incidents. We produce estimates of annual rates of female and male violent offending and victimization for 1980 through 2004 by pooling the National Crime Survey (NCS) and NCVS data. We also examine gendered patterns of violence across victim-offender relationships. There is no published study to date that examines all three of these aspects of gendered crime trends because research has relied heavily on arrest data from the UCR, which do not include information on victims. Using the pooled NCS-NCVS data, we estimate and report trends that have not been published previously and are free from potential criminal justice system bias. In addition, the NCS-NCVS data allow for important disÂaggregations that are not possible with UCR arrest data on nonlethal violence, such as by victim-offender relationships, and thus can be used to reveal factors that may be associated with crime trends. Our assess- ment of the data uncovers similarities and differences between gendered trends in victimization and offending. The detailed examination of these trends is a necessary first step toward better understanding violence in the United States.
GENDER AND VIOLENCE IN THE UNITED STATES 47 Gender, Violence, and Victims: Previous Research on Trends Two undisputed findings in criminology are that men are more likely than women to commit violent crime and, with the exception of rape, men are more likely to be the victims of violent crime. Although the gender gaps in violent offending and victimization are established, there is uncertainty about whether these gaps have changed in a meaningful way over time. Public perception seems to be that women are becoming more similar to men in terms of criminal violence. Over the past three decades, the popular press has warned periodically of a changing female criminal, who is more violent than her predecessors (e.g., Leach, 2004; Scelfo, 2005). The media and activist groups have highlighted the seriousness of violence against women (e.g., the National Organization for Women), and the increased attention to the problem has helped to bring this issue into public aware- ness. Indeed, the federal government responded by passing the Violence Against Women Act of 1994, reauthorized in 2005. But media treatment of women, violence, and victimizationâas well as some scholarly and textbook treatmentsâtend to blur the critical distinc- tion between two very different questions. The first is âHas violence by and against women increased over time?â The second question is âHas the gender gap in violent offending and victimization narrowed over time?â Of course, the answers to these questions can differ. For example, female rates of violent offending and victimization could have increased (or decreased) at a time when male rates changed similarly. When this occurs, the gender gaps in violent offending and victimization would be constant, and the changes in female trends would not be unique. By contrast, female rates of violent offending and victimization could have increased more or decreased less than the corresponding male rates, which would result in a narrowing of the gender gaps, with women accounting for an increasing portion of violent offending and victimization over time. In other words, womenâs and menâs patterns of victimization and offending would differ over time, which would highlight the importance of seeking gender-specific explanations of offending and victimization trends. The distinction between the two ques- tions is critical and illustrates the importance of examining both shifts in female rates of violent offending and victimization as well as replacing and comparisons of female and male rates. Yet some may ask whether decreasing gender gaps in violent offending and victimization are important in the current context of declining crime trends. In other words, would it be practically significant if female rates of offending and victimization remained stable while male rates declined, or if female rates decreased more slowly than male rates? The answer clearly seems to be âyes.â In the first scenario, the finding that womenâs offending
48 UNDERSTANDING CRIME TRENDS and victimization holds steady at prior levels when male offending or victim- ization declines undoubtedly would be of both scientific and policy impor- tance. This pattern would indicate that social forces affecting menâs exposure to violence seem to have little impact on womenâs experiences with violence. In short, womenâs lives do not improve as menâs do in this regard. The second scenario similarly highlights a situation that should be of both scientific and policy relevance. In it, womenâs exposure to violence (in the form of either offending or victimization) is reduced, but to a lesser extent than menâs exposure. Interestingly, this situation is analogous to cur- rent trends in death from heart disease in Western nations. Women have lower rates of mortality from heart disease than men, and the rates for both sexes have been declining over time. Yet there has been a narrowing in the gender gap over time in the United States and other nations because female rates have not been dropping as quickly as male rates (Lawlor, Ebrahim, and Smith, 2001). This has been identified as an issue of concern; menâs health is improving at a faster rate than womenâs health. The same logic applies to the case of female and male exposure to violence. If the gender gap in violent offending and victimization is narrowingâeven during a period of declining crimeâthis would suggest that social environmental changes have benefited men more than women. Research on long-term trends in the gender ratio of violent offending has produced mixed findings. Moreover, there has been a paucity of research on long-term trends in gender ratios of violent victimization. In the remain- der of this section, we review existing research on trends in female-to-male offending and victimization, with an eye to limitations of previous research and unanswered empirical questions. Later in the chapter, we present data on these trends. Gender and Trends in Violent Offending Some studies of changes in gender ratios of offending report that women have accounted for an increasing proportion of all arrests over time (e.g., Heimer, 2000; OâBrien, 1999; Simon and Landis, 1991), but other studies report little change in gender rate ratios (e.g., Steffensmeier and Allen, 1996; Steffensmeier and Cobb, 1981). One reason for these seemingly disparate findings in the case of violent offending may be that trends in gender rate ratios of arrest vary depending on the years under investigation. Studies of the 1960s through the early 1980s tend to report little meaningful change, while studies including more recent years are more likely to find significant increases in gender rate ratios of arrests (Heimer, 2000; OâBrien, 1999). More specifically, recent research that includes the crime decline since the mid-1990s reports that the gender gap in arrests for violence (namely aggravated and simple assault) continued to narrow
GENDER AND VIOLENCE IN THE UNITED STATES 49 because female rates either remained stable or dropped more slowly than male rates (Steffensmeier et al., 2006). Most researchers, however, have considerable concerns about relying exclusively on arrest data in studies of gender ratios of offending. It is possi- ble that the relative violence of women and men changed little over time, and the increasing gender rate ratios (i.e., narrowing of the gender gap) instead reflect changes in policing. For example, the increasing equality of the gen- ders may have shaped the way that police view female offending over time. In the past, police may have viewed womenâs violence as less serious or as less in need of criminal justice intervention. As time passed, however, police may have become more likely to view womenâs violence as problematic, and thus more likely to arrest female offenders. Or the criteria used in decisions about arrests for aggravated assault may have shifted over time, with police becoming more likely to âcharge upâ offenses that previously would have resulted in arrests for simple assault; this would disproportionately inflate the figures for aggravated assault over time (see Blumstein, 2000; Rosenfeld, 2006). Furthermore, cases that in the past would not have entered the offi- cial systemâparticularly domestic violence casesâincreasingly have resulted in arrests for aggravated assault Â(Blumstein, 2000, p. 17). Similar arguments can be made with regard to simple assaults. If shifts in police discretion in arrests for violence operate similarly for both female and male offenders, then the changes in gender gap or gender rate ratios of arrests for violence would not be biased. However, if police use their discretion in substantially different ways in arresting women and men, then the observed narrowing of the gender gap in arrests may be an artifact of changing police practices (Steffensmeier et al., 2005, 2006). An assessment of whether recent reports of increases in the gender rate ratios of violent offending represent real change in womenâs and menâs violent behavior can be answered by examining victimsâ reports of the gen- der of offenders in the NCVS. The NCVS is unaffected by criminal justice system policies and potential bias in arrest decisions, yet it has been used in only two studies of trends in gender ratios of offending (Steffensmeier et al., 2005, 2006). Part of the difficulty in assessing the comparability of arrest and victimization data on female and male offending over time has been that the NCS was redesigned in 1992, when it became the NCVS. The data can be used to create a single time series, but doing so requires specific computational procedures, which we describe in our data section (see Lynch, 2002). Gender and Trends in Violent Victimization At the time of this writing, there was almost no published research on long-term trends in nonlethal violent victimization against women. One
50 UNDERSTANDING CRIME TRENDS exception is an early study by Smith (1987), which used the NCS over a 10-year period (1973-1982) and reported some increases in the proportion of all robberies that had female victims, but no appreciable change in the proportion of assaults with female victims. However, there are good studies of long-term trends in the homicide victimization of women (e.g., Batton, 2004; Browne and Williams, 1993; Dugan, Nagin, and Rosenfeld, 1999, 2003; LaFree and Hunnicutt, 2006; Rosenfeld, 2000; Smith and Brewer, 1995). Most studies use the UCRâs Supplemental Homicide Reports and show that while homicide offenders and victims are disproportionately male, the magnitude of the gender gap is smaller for victimization than offending. Moreover, homicide victimization rates declined during the 1990s for both genders, with very little change in the gender gap. Indeed, a recent cross-national study of homicide victimization trends by LaFree and Hunnicutt (2006) shows little evidence that the gender gap changed significantly in the United States over the period 1950-2001, despite the broader changes in womenâs lives. However, recent evidence suggests that there have been changes in the gender gap in victims of homicides involving intimate partners (e.g., Browne and Williams, 1993; Dugan, Nagin, and Rosenfeld, 1999, 2003; Rosenfeld, 1997). Although intimate partner homicide rates declined for both women and men, the declines were greater among men. Given that female rates of intimate partner homicide were consistently higher than male rates over the past 30 years, the greater decline among men resulted in a widening of the gender gap in intimate partner homicide (Lauritsen and Heimer, 2008). Of course, it is difficult to rule out competing explanations of these changes with national-level data; researchers have thus turned to city-level analyses to try to determine how various factors might explain gender-specific changes in intimate partner homicide (Dugan, Nagin, and Rosenfeld, 1999, 2003). These studies suggest that the declines in female and male rates were significantly related to falling marriage rates. In addi- tion, the greater decrease in male rates relative to female rates may reflect the improved economic status of women, as well as the expansion of domestic violence intervention programs. Yet it is unclear whether patterns in gender rate ratio of homicide can be generalized to other forms of violent victimization (see Lauritsen and Heimer, 2008). In short, little is known about long-term changes in violence against women other than homicide. This gap in knowledge is attributed to poor integration between studies of violence against women and research on crime and violence more generally, as well as the difficulty of finding mea- sures of violent victimization that are reasonably valid and reliable over â owever, H this research examines female and male homicide as outcomes and does not analyze patterns in the gender rate ratio.
GENDER AND VIOLENCE IN THE UNITED STATES 51 time (National Research Council, 2004). Researchers concur that police data are problematic for this purpose because much violenceâespecially violence against womenâis poorly measured by police data (National Research Council, 2004). Rape and sexual assaults, nonstranger incidents, and intimate partner incidents against women are often the least likely crimes to be reported to the police (Catalano, 2006). Yet even if reporting rates were higher, police-based UCR data would be of limited use because, for victimizations other than homicide, the UCR data lack information about the sex of the victim. The NCVS, by contrast, is designed to produce data that allow for the assessment of long-term trends in violent victimization, for crimes other than homicide. As mentioned above, the NCVS data can be pooled with the earlier NCS data to estimate a continuous series of violence rates by using specific weighting procedures. Indeed, these are the only available source of continuous information about violent victimization and details about violent crime incidents. Generating Gender-Specific Estimates Using the NCS and the NCVS The NCS and the NCVS are rich sources of information on gender- specific rates of both violent offending and victimization (U.S. Department of Justice, ICPSR Study Numbers 8608, 8864, 4276). We use these data to create national-level estimates of gender-specific rates of aggravated assault, simple assault, and robbery offending using victimsâ reports of the gender of offenders. Similarly, we derive estimates of gender-specific aggravated assault, simple assault, and robbery victimization rates. We do not com- pare rates of rape in our analysis because preliminary analyses showed that almost all perpetrators of rape are male and almost all victims are female, and that there were no detectable changes in the gender gap in the rape offenders or victims over time. The NCS/NCVS has been used to gather self-report survey data about peopleâs experiences with violence and other forms of victimization con- tinuously since 1973. Using a nationally representative sampling frame, interviews are conducted with persons age 12 and older in each sampled household to determine whether respondents have been the victim of an attempted or completed violent crime. Persons who report an incident of â The NCVS by design does not include information on homicide. â or F example, about 96-97 percent of all rapes and sexual assaults since 1992 involve male offender(s) only. â he annual sample size has varied over the years, ranging from approximately 248,000 T interviews in 1980 to 148,000 interviews in 2004. Persons and households are selected for participation on the basis of Census Bureau information (rather than random-digit-dialing pro- cedures, which may produce biased samples). Person-level response rates are very high, ranging
52 UNDERSTANDING CRIME TRENDS violence over the six-month recall period are then asked a series of ques- tions about the incident, including the sex of the offender(s). In 1992, the NCS survey began using a redesigned questionnaire and henceforth became known as the NCVS. The redesigned survey instrument was phased into the data collection process in a way that makes it possible to assess the effects of the new format on victimization or offending esti- mates (Kindermann, Lynch, and Cantor, 1997; Lynch and Cantor, 1996; Rand, Lynch, and Cantor, 1997). Prior analyses of data from the phase-in period showed that the new questionnaire significantly increased the report- ing of victimization and the magnitude of the change varied according to crime type. Rape reporting increased most, followed by aggravated assault and simple assault. Robbery victimization rates were not significantly higher in the NCVS compared with the NCS. Generating Gender-Specific Rates of Violent Offending In order to use the NCS and NCVS data together, it is necessary to take into account this break in the series and weight the earlier NCS data in ways that are informed by research on the effects of methodological and content changes to the survey. Lynch (2002) details the appropriate procedures for estimating long-term offending trends using the NCS and the NCVS. We follow these procedures to generate gender-specific estimates of assault and robbery offending from 1980 through 2004. Some recent from 97 percent in 1980 to 86 percent in 2004. Census-created sampling weights are used to take into account possible differences in response rates according to the age, race, sex, and residential location of the respondent. Interviews are conducted in English and in Spanish. â ollowing a series of cues and questions about the possible occurrence of a victimization F event, detailed questions are asked about what happened during the incident. The answers to these questions are used to place the incident into crime type categories. Subsequent questions about the incident arise in the following order: the number of times it occurred, when and where the incident took place, the nature of the incident (threatened, attacked, completed), whether the offender had a weapon, the extent of injuries and subsequent medical care, victim protective actions during the incident, whether bystanders were present, whether the victim knows anything about the offender, the number of offenders, the sex and age of the offender, whether the offender was a member of a street gang, under the influence of alcohol or drugs, the victimâs relationship to the offender, and the race of the offender. â ey reasons for the changes in the survey were the difficulties of obtaining estimates of K events that were not commonly thought of as âcrimesâ and discoveries about the extent of family, intimate partner, and sexual violence from other surveys about violence against women (Kindermann, Lynch, and Cantor, 1997). For the purposes of estimating violent victimization, the 1992 Âredesign was the only major methodological break in the NCS-NCVS series. â e begin our analyses with data from 1980 because the victim-offender relationship mea- W sures in the NCS changed in the late 1970s. By starting with 1980, the same time series can be compared across consistently defined victim-offender categories. These years also correspond to the years addressed in two recent publications on gender and violent offending Â(Steffensmeier
GENDER AND VIOLENCE IN THE UNITED STATES 53 research presents gender-specific estimates of adolescent and overall offend- ing that depart from ours (Steffensmeier et al., 2005, 2006). We therefore describe our estimation procedure in detail. When an incident of attempted or completed violence is reported to an interviewer, respondents are asked a series of follow-up questions about the incident, including the number of offenders and the sex of those offenders. Estimates of the number of incidents involving female offenders depend on how one treats incidents involving single and multiple offenders. Because of this, we created two sets of measures to study female offense involvement. Our estimate of female involvement in violence includes single-offender incidents in which the offender was reported by the victim to be female and multiple-offender incidents in which any of the offenders were reported to be female. We also replicate our analyses using a more conservative measure of female violent offending that includes only single-offender incidents to assess whether changes in female involvement might be only as secondary offenders.10 For the 1992-2004 NCVS period, our annual gender-specific violent offending rates are defined as follows: et al., 2005, 2006). Although the definitions of stranger and nonstranger Â offenders did not change during this period, the additional categories of boyfriend/ex-boyfriend and girlfriend/ ex-girlfriend made it possible to distinguish such incidents from those involving other friends and acquaintances to better define incidents involving intimate partners. â The estimation procedure used to produce gender-specific rates of offending by ÂSteffensmeier et al. (2005) is described in a footnote 5, in which the authors state: âWe use three years of data surrounding the transition to calibrate upwards pre-redesign surveys to account for the expanded range of behaviors measured by the revised survey. See Figure 2 for the formulaâ (p. 369). The formula in Figure 2 reads: âMultiplier = (n92 + n93 + n94)/(n90 + n91 + n92)â (p. 380). The same multiplier is noted in the later study (Steffensmeier et al., 2006, p. 87). â e found a slight increase over time in the percentage of incidents involving a single W o Â ffender. Over the 1980-2004 period, approximately 54 percent of robberies, 72 percent of aggravated assaults, and 80 percent of simple assaults involved a single offender. Incidents in which the victim did not report the sex of the offender(s) were rare and are excluded from our estimates. About 1 percent of single-offender incidents are missing such information, as are about 2 percent of multiple-offender incidents. 10â esearchers must also decide how to treat series victimizations in their rate estimations. R Victimizations of a similar nature that occur more than six times during a recall period and for which the victim cannot recall sufficient detail are referred to as series victimizations. (During the NCS period, series victimizations were defined by three rather than six incidents.) To reduce respondent burden, series victims are asked to report the details (including sex of the offender) for the most recent event of the series. Victimsâ estimates of the number of times the event occurred tend to be rounded approximations that can have substantial influences on overall rates (see Planty, 2006; Rand and Rennison, 2005) as well as gender-specific offending rates. Because of this, we decided to count series victimizations as one incident. While male and female offending rates would certainly be higher if series victimizations were counted as three (NCS) or six (NCVS) or more incidents, we found that counting these crimes as one incident will not bias our conclusions about the gender gap in offending. Preliminary Âanalyses showed that the proportion of robbery, aggravated assault, and simple assault incidents
54 UNDERSTANDING CRIME TRENDS Male offending rate = Number of violent incidents with male offender(s) only *1,000 Number of men ages 12 and above in the population Female offending rate = Number of violent incidents with any female offender(s) *1,000 Number of women ages 12 and above in the population To estimate comparable offending rates for the 1980-1991 NCS period, we examined the data from the redesign overlap period and weighted the NCS data accordingly. Because the NCS and NCVS instruments were administered concurrently, estimates from the two surveys can be compared according to a variety of crime or victim characteristics. If the NCS/NCVS- ratio of the rate estimates from the overlap period are found to be statisti- cally significant, that ratio is then applied to the NCS estimates to make them comparable to NCVS estimates.11 We found that the gender-specific offending estimates did not differ significantly within crime type; therefore, we use the same crime-specific ratios developed in earlier analyses of the design change and used by the Bureau of Justice Statistics (Kindermann, Lynch, and Cantor, 1997). Thus, for the 1980-1991 period, the crime- s Â pecific offending rates were multiplied by wc, where wc = 1.00 for robbery, wc = 1.23 for aggravated assault, and wc = 1.75 for simple assault. Of course the NCS and NCVS estimates of offending are not without limitations and two caveats should be noted. First, the sample excludes per- sons who are unattached to households, and thus the data exclude incidents that are experienced by homeless and institutionalized persons. We do not know whether men and women offend against these persons in propor- tions that are different from their offending against others. Second, the use of weights to adjust NCS data to make them comparable to NCVS data assumes that the effect of the methodological change is constant across the NCS years. Although it cannot be determined whether this assumption is true, Rand, Lynch, and Cantor (1997) and others (Lynch, 2002) argue that r Â eported to be series was low and declined slightly from 1980 to 2004, and that the proportion of series incidents with female offenders remained fairly stable over time. 11â lthough prior research suggests that additional adjustments beyond crime type may not A be necessary (e.g., Lynch and Cantor, 1996), we assessed whether this was true for gender- and crime-specific rates of offending. We compared the gender-specific offending estimates of robbery, aggravated assault, and simple assault for the NCS/NCVS overlap period and found small but statistically insignificant differences in the ratio according to the gender of the o Â ffender. Thus the weights for our gender-specific offending estimates for the NCS period are the same for female and male rates, consisting of the crime-specific ratios developed in earlier analyses of the design change (e.g., Kindermann, Lynch, and Cantor, 1997). Lynch (2002) similarly found that the NCS adjustment rates for crimes involving juvenile offenders did not vary by gender but did vary according to crime type and the presence of adult co-offenders.
GENDER AND VIOLENCE IN THE UNITED STATES 55 it is probably the case that any potential weighting error is correlated with time and that estimates for distant years may be more problematic than those for years closer to the redesign. While the first caveat warns that the rates will be underestimates, the second urges caution if conclusions about the gender gap are driven by data from the earliest years of the series. Generating Gender-Specific Rates of Violent Victimization We use similar procedures to create gender-specific estimates of aggra- vated assault, simple assault, and robbery victimization for the period 1980 to 2004.12 We also assessed how the NCS data should be weighted for the purpose of comparing crime- and gender-specific victimization rates. As with the offending data, small gender differences were associated with the new design for some types of victimization; however, these differences were not statistically significant.13 Therefore, the final weights for the victimiza- tion estimates in the NCS period consist of the crime-specific ratios used in our earlier analyses of the gender gap in offending. An important second strength is that NCVS estimates of violence against women have been shown to be externally valid when compared with estimates from the 1995 National Violence Against Women Survey (NVAWS). Rand and Rennison found that the rape rate was higher in the NVAWS than in the NCVS, but that the difference was not statistically significant due to the large standard error for the NVAWS estimate. The difference in womenâs assault rates was also higher in the NCVS, and this difference was statistically significant (Rand and Rennison, 2005, pp. 278- 280). Thus, despite important differences in sampling method and the use of alternative questions, cues, and prompts, estimates from the NVAWS suggest that the NCVS data provide valid and reliable information about violence against women. Also, it is important to remember the key limitations of the survey data regarding the sampling frame, the potential correlation between weighting error and time, and the fact that series victimizations are treated as a single incident. For reasons discussed earlier, we think that these limitations will lead to underestimated rates but are unlikely to bias our estimate of the trend in the gender rate ratio of violent victimization. Of course, all crime rates include measurement error, and sampling error is a component of the NCVS estimates. Changes in sampling error will not bias our trend esti- 12â he issue of how to treat single-offender versus multiple-offender incidents is not appli- T cable in these analyses because the victim is a single individual. 13â When gender- and crime-specific weights are used, the adjustments make the gender gap appear slightly larger during the NCS time period. The use of such weights would suggest greater decreases in the gender gap over time.
56 UNDERSTANDING CRIME TRENDS mates, but readers should consider such errors before drawing conclusions about the difference between specific rates.14 Data on Victim-Offender Relationships by Gender The last set of results presented in this chapter goes beyond the exami- nation of offenders and victims by gender to further disaggregate by the relationship between the victim and the offender. This allows further elabo- ration of how the nature of violent incidents has changed over time. To this end, we use NCVS data to estimate rates of stranger, nonstranger, and intimate partner violence committed by and against women and men. For these analyses, rape, robbery, aggravated assault and simple assault incidents are combined to create a measure of violence that allows us to produce reliable estimates of the above rates.15 Rape and sexual assault are an important part of violence against women, so it is important to include rape in a composite measure of violent victimization. Trends in the Gender RATE Ratios of Violent Offending and Victimization Offending Figures 3-1 through 3-3 show the female and male trends in aggravated assault, simple assault, and robbery, as well as the trends in the gender rate ratios for each offense. Following the literature on trends in the gender gap in offending, we compute the gender rate ratio as the female population- adjusted rate over the male population-adjusted rate of offending for each violent crime type (e.g., Heimer, 2000; OâBrien, 1999). This measure intui- 14â e do not present tests of statistical significance between specific rates because these W would be quite numerous and thus would overwhelm the presentation of results. Rather, our focus is on the description of general patterns in the rates. 15â As with our analyses of offending and victimization, the data from the NCS years were weighted according to crime type. NCS estimates of rape were weighted by a factor of 2.57 (Kindermann, Lynch, and Cantor, 1997). Incidents in which the victim was unable to provide information on the victim-offender relationship were necessarily excluded. Victimization estimates are based on all (multiple and single offender) incidents, and in multiple-offender incidents the relationship was coded stranger if all of the offenders are reported to be strangers. Multiple-offender incidents involving intimate partners were rare and were coded as intimate partner incidents. Offending rate estimates presented in this section are limited to incidents involving single Â offenders. This is because in multiple-offender incidents it is impossible to match mixed-gender groups of offenders with each personâs relationship to the victim.
GENDER AND VIOLENCE IN THE UNITED STATES 57 TABLE 3-1â Percentage Change in Female and Male NCVS Offending Rates 1984-1994 1994-2004 Female Male Female Male Aggravated assault +1 +11 â42 â67 Simple assault +51 +8 â50 â57 Robbery +66 +2 â78 â65 tively captures the relationship between female and male offending rates.16 The general patterns reveal some similarities as well as differences across offense type. Because male rates of both offending and victimization are higher than female rates, the variability in female trend lines cannot be fully appreci- ated from examining these figures alone. Indeed, there is substantial change in the female rates that is masked in figures that depict female and male offending and victimization trends together. To illuminate these patterns more fully, we present the percentage changes in the female and male rates during the decades 1984-1994 and 1994-2004 in Table 3-1. This allows us to compare periods of equal length over time and across gender. We chose these specific periods because, at the time of this research, 2004 was the most recent year of available data and 1994 is near the peak of the crime rates in our study. We recognize that any choice of years to estimate per- centage change is somewhat arbitrary; we use this strategy only to reveal changes in the trends that are difficult to see through visual inspection of the figures (e.g., changes in female offending rates). To ensure that this procedure did not produce misleading findings, we examined alternative 10-year periods, varying the end points of the decades. This did not change our general conclusions about the patterning of female rates, male rates, or gender rate ratios. Figure 3-1 and Table 3-1 show that, from 1984 to 1994, the rate of male aggravated assault offending reported in the NCVS increased by about 11 percent and then plummeted by about 67 percent between 1994 and 2004. The NCVS offending data show little consistent trend and an overall trivial change in aggravated assaults by women between 1984 and 1994; 16â We used the gender rate ratio of offending rather than the female percentage of all offend- ing. The âfemale percentageâ must also be population adjusted and thus must be described not as the âfemale percentage of total offensesâ but rather as âthe population-adjusted percentage of offending incidents accounted for by womenâ because female and male populations (the rate denominators) are not equal for all years.
58 25 0.30 Male Gender rate ratio 0.25 20 0.20 15 0.15 Rate per 1,000 10 0.10 5 Female 0.05 0 0.00 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 FIGURE 3-1â NCVS aggravated assault offending by gender, 1980-2004. Figure 3-1, landscape
GENDER AND VIOLENCE IN THE UNITED STATES 59 after 1994, the NCVS female offending rate dropped by 42 percent. (Note that because the female rate is so much lower than the male rate, fluctua- tions in the female rate are difficult to see in Figure 3-1.) Although the rate of aggravated assault offending dropped for both genders, the decline was more pronounced for men. The result is that the female-to-male rate ratio of aggravated assault offending increases from about .11 in 1980, to .15 in 1992, to .25 by 2004.17 Female rates had reached 25 percent of male rates by 2004. It therefore appears that the NCVS data on offendersâwhich should be unaffected by potential bias in justice system ÂprocessingâÂproduces an upward trend in the gender rate ratio and therefore some decrease in the gender gap. Interestingly, this pattern parallels changes in the gender rate ratios of aggravated assault reported by analyses of UCR arrest data (e.g., Heimer, 2000; Steffensmeier et al., 2006). Figure 3-2 and Table 3-1 show that, unlike in the case of aggravated assault, female rates of simple assault increased by a much larger percentage than did male rates between 1984 and 1994, by 51 percent among women and 8 percent among men. After 1994, the simple assault offending rates decreased for both genders. The male rates dropped by a greater percent- age than the corresponding female rates, although the difference is not substantial (57 and 50 percent, respectively). These combinations of trends in female and male simple assault offending produce gender rate ratios that increases over time, from .19 in 1984, to .27 in 1994, and to .32 in 2004, with the largest increases occurring before the mid-1990s. The overall increase is sizableâwhereas female rates were about 19 percent of male rates in 1984, they had grown to 32 percent of male rates by 2004. The third offense that we examine, robbery, is well known to be the most male of crimes, with female robbery involvement being extremely low (Miller, 1998). Research on changes in the gender rate ratios of rob- bery arrests using UCR data has shown significant increases between 1960 and the middle 1990s (Heimer, 2000; OâBrien, 1999). Figure 3-3 and Table 3-1 indicate that there has been a similar upward shift in gender rate ratios of robbery offenders based on the NCS-NCVS data as well. Male robbery offending rates changed little between 1984 and 1994 (2 percent). As with simple assault, female robbery offending rates increased by a much greater percentageâabout 66 percentâduring this same time period. How- ever, the female robbery offending rate dropped more (78 percent) than the male rate (65 percent) between 1994 and 2004, during the time of the great 17â ur figures and tables present the victimsâ reports of the sex of offender in all victimization O incidents, including those perpetrated by a single offender and multiple offenders. Replicating our analyses with the data from incidents with only a single offender, we found that the gen- der rate ratios did not change much and showed quite similar trends. We present the former measure because a substantial portion of violent incidents involves multiple offenders.
60 60 0.35 Male Gender rate ratio 0.30 50 0.25 40 0.20 30 0.15 Rate per 1,000 20 0.10 Female 10 0.05 0 0.00 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 FIGURE 3-2â NCVS simple assault offending by gender, 1980-2004. Figure 3-2, landscape
16 0.25 14 0.20 12 Male Gender rate ratio 10 0.15 8 Rate per 1,000 0.10 6 4 0.05 2 Female 0 0.00 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 FIGURE 3-3â NCVS robbery offending by gender, 1980-2004. Figure 3-3, landscape 61
62 UNDERSTANDING CRIME TRENDS crime decline. Yet Figure 3-3 shows a slightly increasing trend in gender rate ratios for robbery, growing from an average of .08 for the years 1980 to 1982 to an average of .15 for the years 2002 to 2004. Overall, our data indicate that the gender gap in offending has narrowed over time, although violent offending has clearly declined among both gen- ders since the mid-1990s. This parallels some previous findings, based on UCR arrest data. However, the NCVS data used here are Â unaffected by procedural shifts and gender bias in the criminal justice system. Compar- ing across the three offenses addressed here, one can see that in the case of simple assault and robbery, the narrowing of the gender gap stems in part from the greater increases in womenâs involvement in these offenses during the 1980s and early 1990s. In the case of aggravated assault, by contrast, the narrowing of the gap seems to stem from the more pronounced decline in male than female offending. Victimization This section focuses on change in the gender rate ratios in violent vic- timization from 1980 through 2004. As with offending, gender rate ratios are computed as female rates divided by male rates of victimization for each violent crime type. Figures 3-4 through 3-6 show gender-specific victimiza- tion rates for aggravated assault, simple assault, and robbery, as well as the gender rate ratios of these victimization rates. Table 3-2 presents the figures on the percentage change over time in female and male rates. Figure 3-4 shows that there has been some narrowing of the gender gap in aggravated assault victimizations over time. Male rates declined some in the early 1980s and then increased slightly in the early 1990s. As Table 3-2 reports, male rates of aggravated assault victimization showed a net decline of about 4 percent between 1984 and 1994, but this was fol- lowed by a dramatic decline of about 62 percent between 1994 and 2004. By comparison, female rates did not decline in the 1980s, but rather were fairly stable until the early years of the 1990s when they increased. Thus, as TABLE 3-2â Percentage Change in Female and Male NCVS Victimization Rates 1984-1994 1994-2004 Female Male Female Male Aggravated assault +22 â4 â65 â62 Simple assault +23 +7 â54 â55 Robbery +3 +4 â68 â64
. 20 0.7 Gender rate ratio 18 0.6 Male 16 14 0.5 12 0.4 10 0.3 Rate per 1,000 8 6 0.2 Female 4 0.1 2 0 0.0 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 FIGURE 3-4â NCVS aggravated assault victimization by gender, 1980-2004. 63 Figure 3-4, landscape
64 UNDERSTANDING CRIME TRENDS Table 3-2 shows, female rates of aggravated assault victimization increased by about 22 percent between 1984 and 1994, whereas male rates showed only a small decrease. The decline in aggravated assault victimization fol- lowing the middle 1990s, however, was fairly comparable among women and men. Overall, gender rate ratios increased showing that female rates of victimization escalated from about 34 percent of male rates at the beginning of the series to about 55 to 65 percent of the male rates by the early 2000s. As the patterns discussed above indicate, much of this change is traceable to the period before the great crime decline of the 1990s and early 2000s. Figure 3-5 shows that compared with aggravated assault, rates of simple assault victimization are much higher for both women and men. The figure also shows that gender rate ratios of simple assault are much higher than the gender rate ratio of aggravated assault victimization. In terms of trends over time, there are some similarities to the patterns of aggravated assault victimizations. Specifically, like aggravated assault, male rates of simple assault declined somewhat during the 1980s, whereas female rates did not. Also like aggravated assault victimization, both men and women experienced increased simple assault victimization in the first few years of the 1990s, followed by subsequent declines. Between 1984 and 1994, male rates of simple assault victimization showed a net increase of about 7 percent (because of increases in the early 1990s following a period of decline in the 1980s), followed by a decline of about 55 percent between 1994 and 2004. Female rates of simple assault showed a net increase of approximately 23 percent between 1984 and 1994 (due to stability in the 1980s followed by the early 1990s increase) and, like male rates, declined by about 54 percent between 1994 and 2004 (see Table 3-2). Together, these patterns produce increasing gender rate ratios of simple assault vic- timization, from about .59 in the early years of the series to about .75 in 2004 (female rate was 75 percent of the male rate), with even higher ratios in years 2000 and 2001. As in the case of aggravated assault, the increase in gender rate ratios for simple assault victimization appears to be the result of the stabil- ity (rather than decline) of simple assaults against women in the 1980s, coupled with larger percentage increases in female victimization rates in the early 1990s. The declines after the mid-1990s are similar in percentage across gender. This is an important finding, because it indicates that for both aggravated and simple assault victimizations, womenâs experiences have become increasingly like menâs experiences mainly because of changes that occurred before the recent crime decline. Later we show that intimate partner victimization is an exception to this pattern. The patterns of robbery victimization, perhaps not surprisingly, look quite different than assault victimizations. Figure 3-6 shows that the risk for robbery decreased substantially for both women and men between 1980
45 1.0 Gender rate ratio 40 0.9 Male 0.8 35 0.7 30 0.6 25 0.5 20 Female Rate per 1,000 0.4 15 0.3 10 0.2 5 0.1 0 0.0 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 FIGURE 3-5â NCVS simple assault victimization by gender, 1980-2004. 65 Figure 3-5, landscape
. 66 12 0.8 0.7 10 Gender rate ratio 0.6 Male 8 0.5 6 0.4 Rate per 1,000 0.3 Female 4 0.2 2 0.1 0 0.0 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 FIGURE 3-6â NCVS robbery victimization by gender, 1980-2004. Figure 3-6, landscape
GENDER AND VIOLENCE IN THE UNITED STATES 67 and 2004. Slight differences in the two trends can be foundâfor example, male rates increased somewhat in the early 1990s, whereas female rates did not. Beyond this, the trends are quite similar. Between 1984 and 1994, male rates of robbery victimization increased by about 4 percent, followed by a decline of about 64 percent between 1994 and 2004. Similarly, female rates of robbery increased approximately 3 percent between 1984 and 1994 and declined by about 68 percent between 1994 and 2004. Thus, the proportional changes over time are fairly similar for women and men. In the beginning of the series, the gender rate ratio was about .47; in 2004, it was about .45. Lauritsen and Heimer (2008) recently compared these patterns to those of homicide victimization and found little change in the gender rate ratios of either homicide or robbery. Gender differences in rob- bery victimization trends are similar to those of lethal violence but different from those of aggravated and simple assaults. This may reflect a gender difference linked to the use of guns, which is most common in robberies and homicides. The female and male violent victimization findings lead us to conclude that the gender gap in assault victimization (but not robbery) has narrowed over the past several decades, with female victimizations constituting an increasing proportion of assaults.18 The risks of victimization for simple assault became more similar for men and women in the late 1990s and into the early 2000s (see Figure 3-5). The risk for aggravated assault vic- timization also was more similar for women and men during this time (see Figure 3-4). This suggests that researchers must be careful not to assume that the factors that influence the aggregate trends necessarily apply equally to female and male violent victimization. Indeed, the factors influencing assault victimizations may have differed across gender or, if similar factors were at work, their impact may have varied across gender. Comparing gender-specific trends in victimization with the trends in offending highlights several points. First, it appears that women have become more likely to encounter violent interactions over time, both as perpetrators and as victims. This is perhaps expected due to the fact that victimization and offending are correlated at the individual level and also share many of the same predictors when studied at the individual, situ- ational, or community level (Sampson and Lauritsen, 1994). A second parallel across the victimization and offending trends is that the crime drop 18â n earlier analyses, we found that these results are not due to the fact that intimate part- I ner incidents are included in these trends. Those analyses showed that the trends in intimate partner violence against women closely match those for stranger violence against women and the crime-specific patterns shown here. We observed no declines in intimate partner violence against women until approximately 1993-1994. This issue is discussed later in the chapter.
68 UNDERSTANDING CRIME TRENDS of the late 1990s and 2000s appeared to affect women and men in fairly comparable ways. While the most recent crime decline seemed to have a more or less similar impact across gender, a comparison of the percentage changes in Table 3-1 (offending) and Table 3-2 (victimization) suggests that, when narrowing of the gender gap occurs, it tends to be due to gender differences in the trends in the 1984 to 1994 period. More specifically, the data show either a greater proportionate increase in female than male rates (e.g., sim- ple assault offending and victimization, robbery offending), or an increase in female rates while male rates show a net decrease decline (aggravated assault offending). These observations raise questions about how and why female and male violent offending and victimization may have been differ- ently influenced in the 1980s and early 1990s. In the next section, the data are disaggregated by victim-offender relationships, as well as gender, to help shed light on what may have occurred. Victim-Offender Relationships and Gendered Patterns of Offending and Victimization It is possible that there may have been disproportionate changes over time in stranger, nonstranger, and intimate partner offending and victimization that would account for the shifts in the gender rate ratios in Â victimization and offending that we note above. Homicide research reports that men and women experience these types of violence in differ- ent proportions. For example, homicide offending and victimization rates vary by Â gender and by victim-offender relationship, and the gender gap is contingent on the victim-offender relationship in the homicide incident (Rosenfeld, 2000). We therefore disaggregate trends in female and male violent victimization as well as female and male violent offending by victim- offender relationship using the NCVS data, as described previously. Here we combine rape, robbery, aggravated assault, and simple assault incidents to create a measure of violence that allows us to produce reliable estimates of the above rates. Nonstranger violence includes incidents committed by intimate partners. Figures 3-7a and 3-7b show the rates of female and male victimization disaggregated by victim-offender relationship for the period 1980 through 2004, and Figures 3-8a and 3-8b display the comparable trends in female and male offending. Table 3-3 summarizes these patterns across the same decades examined above, 1984 to 1994 and 1994 to 2004. These trends again show differences across gender in victimization risks in the earlier 1984-1994 period (see Table 3-3). During this earlier period, female risk of stranger violence increased by about one-quarter (24 percent) and male risk of stranger violence increased by 10 percent. Female risk of violent
25 20 15 Nonstranger Rate per 1,000 10 Stranger 5 Intimate partner 0 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 FIGURE 3-7aâ Female violent victimization by victim-offender relationship, 1980-2004. Figure 3-7a, landscape 69
70 50 45 40 35 30 25 Rate per 1,000 20 Stranger 15 Nonstranger 10 5 Intimate partner 0 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 FIGURE 3-7bâ Male violent victimization by victim-offender relationship, 1980-2004. Figure 3-7b, landscape
10 9 8 7 6 Nonstranger 5 Rate per 1,000 4 3 Stranger 2 1 Intimate partner 0 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 FIGURE 3-8aâ Female violent offending by victim-offender relationship, 1980-2004. 71 Figure 3-8a, landscape
72 40 35 30 25 20 Nonstranger Rate per 1,000 15 Stranger 10 5 Intimate partner 0 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 FIGURE 3-8bâ Male violent offending by victim-offender relationship, 1980-2004. Figure 3-8b, landscape
GENDER AND VIOLENCE IN THE UNITED STATES 73 TABLE 3-3â Percentage Change in Female and Male Victimization and Offending Rates by Strangers, Nonstrangers, and Intimate Partners Across Decades: NCVS Victimization Offendinga 1984-1994 1994-2004 1984-1994 1994-2004 Female Male Female Male Female Male Female Male Stranger +24 +10 â66 â68 +243 +28 â52 â63 Nonstranger +17 â8 â52 â39 +20 â4 â46 â57 Intimate partner +5 + â54 â + +13 â â59 NOTE: Includes rape, robbery, aggravated assault, and simple assault. aBased on single-offender incidents only. + = Rates too low to assess percentage change; however, data suggest increase. â = Rates too low to assess percentage change; however, data suggest decrease. victimization by nonstrangers and intimate partners also increased between 1984 and 1994 (17 and 5 percent, respectively), but at lower rates than the risk of stranger violence. Nonstranger victimization against men decreased only slightly between 1984 and 1994 (see Table 3-3), when the comparable female rates were increasing. This suggests that prior to 1994, female risks of violence increased more than male risks in all relationship categories, and the largest percentage increases among women occurred in violence by strangers. After 1994, violent victimization declined for both genders across all relationship categories. The drop was quite sizable and comparable across gender in violence by strangers (66 percent for women, 68 percent for men). There was a less marked decline in nonstranger violence across gender, but women did experience a greater proportionate reduction in risk than men (52 and 39 percent, respectively). Women also experienced a sizable reduction in the risk of violence by intimate partners during this period (54 percent). We conclude from this that the crime drop of the late 1990s and early 2000s affected female victimization similarly across relationship categories, but it had relatively greater consequences for male experiences with victimization by strangers. Table 3-3 summarizes similar disaggregated trends based on offending. Female offending against strangers increased between 1984 and 1994; then it declined at a rate similar to male offending against strangers between 1994 and 2004. While female violence against strangers has always been very low, Table 3-3 (and Figure 3-8a) show that it increased by 243 percent between 1984 (1.1 per 1,000) and 1994 (4.2 per 1,000). Female offending against nonstrangers increased by a smaller percentage (20 percent) during this period, from 6.9 per 1,000 in 1984 to 8.2 per 1,000 by 1994. The per-
74 UNDERSTANDING CRIME TRENDS centage increase in male violence against strangers was 28 percent between 1984 and 1994, and male violence against nonstrangers declined by 4 per- cent during this same decade (Figure 3-8b). Although the rates of female violence against strangers are very low and thus a small change in absolute numbers produces a very large percentage increase, the steeper upward slope in female stranger (Figure 3-8a) than male stranger (Figure 3-8b) offending would seem to flag a noteworthy change. We explore this further in the concluding section. Turning to intimate partner offending and victimization, Table 3-3 shows that the decline in male intimate partner offending (and, equivalently, female intimate partner victimization) occurred between 1994 and 2004. In fact, the data suggest that male intimate partner offending increased slightly during the 1984 to 1994 period. Since these peak years, male rates of offending against strangers, nonstrangers, and intimate partners have all declined at similar magnitude, which could suggest that there may be some common causes of the declines in these types of offending during this period. We are unaware of any research that has assessed the factors associated with the decline in nonlethal intimate partner violence during the 1990s, but we suggest that future investigations of such trends should con- sider additional factors beyond changes in domestic violence policies and practices, which, along with womenâs economic and marriage rates, have been the focus of intimate partner homicide trends (e.g., Dugan, Nagin, and Rosenfeld, 1999, 2003). The trends by gender of offender, type of violence, and gender of victim suggest a few basic conclusions. First, the modal category of violent crime in 2004 is not the same as it was in 1994 or 1984. Figures 3-7a and 3-7b show clearly that the risk for stranger and nonstranger violent victimization has declined substantially for women and men since 1993, but it also shows that these declines have been proportionately greater for stranger violence than for nonstranger violence. Indeed, while male victimization by Âstrangers was by far the modal category of violence in the early 1980s, by 2004 it had decreased to the point at which it was no longer substantially higher than male or female victimization by nonstrangers. Second, despite the low base rate of female violence against strangers, the relatively large percentage increase between 1984 and 1994 seems to warrant further investigation. Third, Figures 3-7a and 3-7b show that, throughout the series, nonstranger violence against men and women has been at roughly comparable levels, but female rates came to exceed those of men by 1992. DISCUSSION This chapter uses pooled NCS-NCVS data to show that violence involv- ing women has come to constitute a greater proportion of violent incidents
GENDER AND VIOLENCE IN THE UNITED STATES 75 over time and that gender clearly matters for understanding U.S. crime trends. First, we present empirical evidence from victimsâ reports of the gender of their assailants that shows meaningful changes in the gender rate ratios of violent offending over time, with some narrowing in the gender gap in aggravated assault, simple assault, and robbery. Second, we present data on the gender of victims that shows that the gender gap in violent victimization has narrowed for aggravated and simple assault. These findings are further illuminated by the changing patterns of female and male victimization and offending across stranger, nonstranger, and intimate partner relationships. The fact that gender rate ratios of offending and victimization have not remained stable indicates that there may well be something unique about gender during this time period. It also suggests that fully understanding crime trends requires consideration of variation across gender and Âvictim-offender relationships. Furthermore, these findings clearly differ from those based on homicide data, which show no narrow- ing of the gender gap in victimization. This shows the need to go beyond homicide data to understand gender and violent victimization. As noted at the outset, the goal of this chapter is to present data on long-term trends in female and male offending and victimization, as well as trends in the gender rate ratios. Examining long-term trends is essential for contextualizing shorter term fluctuations in crime rates, and to date research has not examined gender differences in long-term trends in both victimization and offending. In these conclusions, we compare the pat- terns in offending and victimization and illuminate them further using our findings of patterns across victim-offender relationships. We do not claim to explain the source of gender differences and similarities in these trends, as a time-series analysis that includes a full set of covariates would be questionable given the sample size of 25 years. Rather, because the first step in understanding any phenomenon is thorough description, we seek to highlight select comparisons and offer hypotheses to stimulate future research in this area. A first observation that emerges from the data is that, with the excep- tion of aggravated assault offending, a notable portion of the narrowing of the gender gap in violence can be traced to changing female-to-male ratios before the crime decline of the mid-1990s. Our findings show that large gender differences across the relationship categories occurred before the mid-1990s. During this time, there were increases in both offending against nonstrangers and victimization by nonstrangers among women, yet the corresponding male rates decreased. There also was a notable percentage increase in female violent offending against strangers and victimization by strangers before the mid-1990s, whereas male rates showed smaller per- centage increases. This suggests that gender-specific social changes linked to victimization and offending may have occurred before the onset of the
76 UNDERSTANDING CRIME TRENDS great crime decline of the late 1990s and early 2000s, and these changes appear to hold for both stranger and nonstranger crime. It is interesting that the gender gap decreases not only across types of violence, but also for both victimization and offending. The fact that the largest increases in the gender rate ratios of offending and victimization occur at roughly the same time is not surprising, given research showing that involvement in victimization and offending are correlated and share many of the same predictors (Sampson and Lauritsen, 1994). Yet, because previous studies have not compared the gender gaps in offending and vic- timization over time, this correspondence has been ignored in the literature on crime trends and remains an important area for future research. These shifts in the gender rate ratios of violence may have been asso- ciated with broad social changes due to enhanced social freedoms for women and gender equality that increased before and during the 1980s. For women, these changes may have been accompanied by higher levels of public interactions, in the labor force and elsewhere, thus expanding opportunities for violent victimization and offending. However, changing gender roles may have had two very different con- sequences for violent victimization. On one hand, it may be that displays of interpersonal violence became increasingly less acceptable as women increasingly occupied the public sphere, thus helping to reduce male vic- timization. This comports with the long-term declines that we uncovered in violence against both men and women, by strangers as well as nonstrangers. On the other hand, womenâs increased presence in public life simultane- ously created greater opportunities for nonfamilial victimization. Thus, although rates of violent victimization declined for both genders over the past 25 years, the decline for women was less than the decline for men for aggravated and simple assaults. Viewing the trend data as an indicator of motivation for offend- ing rather than opportunities for victimization suggests an alternative h Â ypothesis: perhaps changing gender roles increased womenâs vulnerability to the effects of the economy. Although the feminization of poverty slowed in the 1980s for all ages combined, the youth of both genders felt the brunt of the decadeâs difficult economic times (Bianchi, 1999). The experience of economic stress may have combined with greater participation in street life by young women to produce relatively greater changes in womenâs than menâs violent encounters. Other hypotheses are certainly plausible. Perhaps increasing incarcera- tion rates or changes in the policing of public spaces over the period studied had a more significant impact on male than female offending rates, thus contributing to the reduction in the gender gap in offending. Because men are more likely than women to be the victims of male-perpetrated violence, the large increases in the numbers of men incarcerated may have had a
GENDER AND VIOLENCE IN THE UNITED STATES 77 larger impact on male victimization than female victimization rates, thus contributing to increasing gender rate ratios of violent victimization. Simi- larly, increases in the policing of public places may have disproportionately decreased male victimization. It is also important to understand why the crime drop after the mid- 1990s affected women and men rather similarly. One hypothesis is that, perhaps by the 1990s, growth in social freedoms for women had slowed and the factors affecting female and male trends became more similar. So, for example, economic prosperity and the growth in imprisonment may have spurred decreases in female as well as male victimization and offend- ing. Another hypothesis is that different factors were associated with the similar rates of decline in female and male violence. For example, the decline in female intimate partner victimization may have been related, in part, to successful policies and programs targeting violence against women, which became more widespread during the 1990s. Declines in other forms of violence against women, such as violence by strangers, and declines in violence against men may have been more associated with other contem- poraneous policies targeting crime more generally, such as increased incar- ceration. Moreover, it could be that policies aimed at reducing violence against women had a spillover effect on other forms of male offending, by bringing men into the criminal justice system when they otherwise might have remained free to commit other types of offenses. One final issue uncovered by our disaggregation of crime trends con- cerns the changing composition of violence over time. While male stranger victimization was by far the modal category of violence in the early 1980s, by 2004 it had decreased to the point at which it was no longer substan- tially higher than male or female nonstranger victimization. Why stranger violence has declined more rapidly than other forms of nonlethal violence is a challenging question for future research. Moreover, nonstranger violence against men and women had occurred at comparable levels, but female nonstranger victimization rates came to exceed male rates by about 1992. This means that nonstranger violence is now a critical part of violence in the United States, and women are now affected at levels similar to those of men. This presents a challenge to criminal justice policy, and it indicates that new efforts to reduce nonlethal violent crime are unlikely to have much effect unless they can affect violence by nonstrangers. If interventions to reduce violence against women have had some impact on violence by men both inside and outside intimate partner relationships, such strategies may offer a place to start the thinking about crafting policy to reduce violence. We note that our analyses cannot speak to race, ethnicity, or age dif- ferences in these gender-specific patterns of offending and victimization. This is a very important issue that requires careful research attention to determine whether the patterns of gender rate ratios that we observed
78 UNDERSTANDING CRIME TRENDS would hold across different race and age groups. Although the study of race and age differences in nonlethal victimization and offending cannot be addressed with UCR data, these important issues can be addressed with careful use of the NCVS data. However, disaggregating the data to address these patterns involves methodological complexities beyond those described here and therefore is beyond the scope of the present analysis. Our goal here has been to take a first step by focusing on long-term trends in the gender rate ratios of offending and victimization as measured by pooled NCS-NCVS data and to link the study of violence against women to the study of crime trends. In conclusion, our findings highlight the complexities inherent in under- standing trends in violence across crime types and gender. The trends that we present indicate that gender-specific trends in violence share similarities but also are sufficiently unique to indicate that female victimization and offending should be part of the consideration of crime trends in the United States. Given that the data reveal some reduction in the gender gap in vio- lent offending and victimization, the situation is akin to that of a narrowing gender gap in mortality from heart disease. Even in a period of an overall decline in crime (or heart disease) the fact that women benefit less than men from the social changes affecting crime rates over the past 25 years signals the need for both scientific research and social policy to address the differ- ences. Gender is therefore an important part of the story of violence in the United States and should not be excluded from analyses of crime trends. REFERENCES Batton, Candice. (2004). Gender differences in lethal violence: Historical trends in the relation- ship between homicide and suicide rates, 1960-2000. Justice Quarterly, 21, 423-461. Bianchi, Suzanne M. (1999). Feminization and juvenilization of poverty: Trends, relative risks, causes and consequences. Annual Review of Sociology, 25, 307-333. Blumstein, Alfred. (2000). Disaggregating the violence trends. In Alfred Blumstein and Joel Wallman (Eds.), The crime drop in America (pp. 13-44). New York: Cambridge Univer- sity Press. Blumstein, Alfred, and Joel Wallman. (2000). The crime drop in America. New York: Cam- bridge University Press. Browne, Angela, and Kirk Williams. (1993). Gender, intimacy, and lethal violence: Trends from 1976 through 1987. Gender and Society, 7, 75-94. Bureau of Justice Statistics. (2006). Criminal victimization in the U.S., 2004. Statistical tables. Washington, DC: U.S. Department of Justice. Catalano, Shannan. (2006). Intimate partner violence in the United States. Bureau of Justice Statistics. Washington, DC: U.S. Department of Justice. Dugan, Laura, Daniel S. Nagin, and Richard Rosenfeld. (1999). Explaining the decline in intimate partner homicide: The effects of changing domesticity, womenâs status, and domestic violence resources. Homicide Studies, 3, 187-214.
GENDER AND VIOLENCE IN THE UNITED STATES 79 Dugan, Laura, Daniel S. Nagin, and Richard Rosenfeld. (2003). Exposure reduction or retali- ation? Effects of domestic violence resources on intimate partner homicide. Law and Society Review, 37, 169-198. Heimer, Karen. (2000). Changes in the gender gap in crime and womenâs economic mar- ginalization. In Gary La Free (Ed.), The nature of crime: Continuity and change (pp. 427-483). Washington, DC: National Institute of Justice. Available: http://www.ncjrs. gov/criminal_justice2000/vol1_2000.html [accessed August 2008]. Kindermann, Charles, James Lynch, and David Cantor. (1997). The effects of the Âredesign on victimization estimates. (Bureau of Justice Statistics Technical Report Series.) ÂWashington, DC: U.S. Department of Justice. Available: http://www.ojp.usdoj.gov/bjs/pub/pdf/erve.pdf [accessed August 2008]. LaFree, Gary, and Gwen Hunnicutt. (2006). Female and male homicide victimization trends: A cross-national context. In Karen Heimer and Candace Kruttschnitt (Eds.), Gender and crime: Patterns in victimization and offending (pp. 195-229). New York: New York University Press. Lauritsen, J.L., and Karen Heimer. (2008). The gender gap in violent victimization, 1973- 2004. Journal of Quantitative Criminology, 24(2), 125-147. Lawlor, Debbie, Alnoor S. Ebrahim, and George Davey Smith. (2001). Sex matters: ÂSecular and geographical trends in sex differences in coronary heart disease mortality. Â British Medical Journal, 323, 541-545. Available: http://www.pubmedcentral.nih.gov/Âpicrender. fcgi?artid=48158&blobtype=pdf [accessed August 2008]. Leach, Susan Llewelyn. (2004). Behind the surge in girl crime. Christian Science Monitor, Sep- tember 15. Available: http://www.csmonitor.com/2004/0915/p16s02-usju.html Â[accessed August 2008]. Lynch, James. (2002). Trends in juvenile violent offending: An analysis of victim survey data. Office of Juvenile Justice and Delinquency Prevention. Washington, DC: U.S. Depart- ment of Justice. Lynch, James, and David Cantor. (1996). Models for adjusting the NCS trend to account for design differences between the NCS and NCVS. Memorandum to the American Statistical Association Committee on Law and Justice Statistics. Miller, Jody. (1998). Up it up: Gender and the accomplishment of street robbery. Criminol- ogy, 36, 37-66. National Research Council. (2004). Advancing the federal research agenda on violence against women. Steering Committee for the Workshop on Issues in Research on Violence Against Women, Candace Kruttschnitt, Brenda L. McLaughlin, and Carol Petrie (Eds.). Com- mittee on Law and Justice, Division of Behavioral and Social Sciences and Education. Washington, DC: The National Academies Press. OâBrien, Robert. (1999). Measuring the convergence/divergence of serious crime arrest rates for males and females: 1960-1995. Journal of Quantitative Criminology, 15, 97-114. Planty, Michael. (2006).Â Series victimizations and divergence.Â In James Lynch and Lynn A Â ddington (Eds.), Understanding crime statistics: Revisiting the divergence of the NCVS and UCR (pp. 156-182). New York: Cambridge University Press. Rand, Michael, and Callie Marie Rennison. (2005).Â Bigger is not necessarily better: An analy- sis of violence against women estimates from the National Crime Victimization Survey and the National Violence Against Women Survey.Â Journal of Quantitative Criminology, 21, 267-291. Rand, Michael, James Lynch, and David Cantor. (1997). Long-term trends in crime victimiza- tion. Bureau of Justice Statistics. Washington, DC: U.S. Department of Justice. Rosenfeld, Richard. (1997). Changing relationships between men and women. Homicide Studies, 1, 72-83.
80 UNDERSTANDING CRIME TRENDS Rosenfeld, Richard. (2000). Patterns in adult homicide: 1980-1995. In Alfred Blumstein and Joel Wallman (Eds.), The crime drop in America (pp. 130-163). New York: Cambridge University Press. Rosenfeld, Richard. (2006). Explaining the divergence between UCR and NCVS aggravated assault trends. In James Lynch and Lynn A. Addington (Eds.), Understanding crime statistics: Revisiting the divergence of the NCVS and UCR (pp. 251-268). New York: Cambridge University Press. Sampson, Robert, and Janet Lauritsen. (1994). Violent victimization and offending: IÂndividual-, situational-, and community-level risk factors. In National Research Council, Under- standing and preventing violence: Social influences on violence (pp. 1-114, vol. 3), A Â lbert Reiss and Jeffrey Roth (Eds.), Panel on the Understanding and Control of Violent B Â ehavior. Washington, DC: National Academy Press. Scelfo, Julie. (2005). Bad girls go wild: A rise in girl-on-girl violence is making headlines nationwide and prompting scientists to ask why. Newsweek, June 13. Available: http:// www.newsweek.com/id/50082 [accessed August 2008]. Simon, Rita James, and Jean Landis. (1991). The crimes women commit and the punishments they receive. Lexington, MA: Lexington Books. Smith, M. Dwayne. (1987). Changes in the victimization of women: Is there a ânew female victim?â Journal of Research in Crime and Delinquency, 24(4), 291-301. Smith, M. Dwayne, and Victoria E. Brewer. (1995). Female status and the gender gap in U.S. homicide victimization. Violence Against Women, 1(4), 339-350. Steffensmeier, Darrell, and Emilie Allan. (1996). Gender and crime: Toward a gendered theory of female offending. Annual Review of Sociology, 22, 459-487. Steffensmeier, Darrell, and Michael J. Cobb. (1981). Sex differences in urban arrest patterns, 1934-1979. Social Problems, 29(1), 37-50. Steffensmeier, Darrell, Jennifer Schwartz, Hua Zhong, and Jeff Ackerman. (2005). An assess- ment of recent trends in girlsâ violence using diverse longitudinal sources: Is the gender gap closing? Criminology, 43(2), 355-405. Steffensmeier, Darrell, Hua Zhong, Jeff Ackerman, Jennifer Schwartz, and Suzanne Agha. (2006). Gender gap trends for violent crimes, 1980-2003. Feminist Criminology, 1, 72-98. Zimring, Franklin E. (2006) The great American crime decline. New York: Oxford University Press.