How has the Violence Against Women Act affected the response of the criminal justice system to domestic violence?

This study uses an interrupted time series design to examine the association between the Violence Against Women Act of 1994 (VAWA) and several different dimensions of the criminal justice system's involvement in violence against women. These include examining the domestic violence incidence rate, and rates of police notification, arrest, and judicial authorities' involvement. Data from the National Crime Victimization Survey from 1992 to 2003 is used. Results suggest that overall the incidence of domestic violence has decreased while police notification and perpetrator arrest have increased over time. Further, victim involvement with judicial authorities significantly increased after enactment of the VAWA. Interpretations and potential explanations of the results are discussed.

Keywords: domestic violence, National Crime Victimization Survey, Violence Against Women Act, incidence rate, arrest, police notification

Introduction

Since the 1970s when public awareness of violence against women increased largely due to the dedicated efforts of advocates for battered women, society has responded to violence occurring in intimate relationships. Supportive services for victims such as emergency shelters and legal consultation, aggressive batterer interventions such as pro-arrest policies and treatment programs for the batterer, and legislation to institutionally address violence against women such as the Violence Against Women Act of 1994 (VAWA) have been used to respond to violence in intimate relationships.

Beginning in the 1980s, many individual states actively addressed domestic violence through legal reforms including defining domestic violence as crime, pro- or mandated arrest policies, expanding the definition of intimate partners to include cohabiting couples and same sex couples, and introducing civil protection orders (Burt, Dyer, Newmark, Norris, & Harrell, 1996). Legal reforms at the state level led to federal legislation, the Violence Against Women Act of 1994, which was designed to improve interstate criminal justice enforcement and provide adequate funding for criminal justice interventions and social services for victims.

The VAWA focuses on six distinct areas: safe streets for women (e.g., grants to combat violence against women in public), safe homes for women (e.g., grants for domestic violence hotlines and battered women's shelters), equal justice for women in the courts (e.g., grants to develop education and training programs for judges), stalker and domestic violence reduction (e.g., grants to improve processes for data collection regarding stalking and domestic violence into crime information databases), protection for battered immigrant women and children (e.g., rights for battered immigrant women to file legal petitions), and provisions for strengthening existing laws. The VAWA consolidates almost all of states' legal reforms responding to domestic violence and was expected to effectively facilitate and strengthen existing state policies to reduce and intervene in domestic violence. The VAWA does this through grants, education and training programs, and pro-arrest policies.

Given the scope of the VAWA, it is natural to expect positive changes in the criminal justice response following its enactment. This study is interested in whether, and how, the VAWA has affected domestic violence, specifically with regard to the incidence of violence between intimate partners, and interactions between victims and criminal justice system.

Review of the Literature

The rate of domestic violence appears to be declining. From 1993 to 1997 the domestic violence incidence rate fell from 9.8 to 7.5 per 1,000 women (Bureau of Justice Statistics, 2000). Increased legal services for victims and improvements in women's economic status seem to have contributed to the decline (Farmer & Tiefenthaler, 2003), although the results of studies on the effect of arrest have been inconclusive (e.g., Schmidt & Sherman, 1993; Sherman & Berk, 1984; Sherman, Schmidt, & Rogan, 1992). For example, Sherman and colleagues (1984) significantly contributed to nation-wide adoption of pro-arrest policies by arguing that arrest and incarceration were effective ways to deter domestic violence, but then subsequently questioned their effectiveness (Schmidt & Sherman, 1993).

Some studies have reported economic outcomes associated with the VAWA. Burt et al. (1996) evaluated federal and state implementation of a variety of federal programs and found that more than $23.5 million had been provided to states through federal grants during the first year of the implementation in 1995. The Violence Against Women Office reports that federal funds for domestic violence programs have dramatically increased to $1.6 billion in the 5 years since passage of the VAWA (as cited in Clark, Biddle, & Martin, 2002). Clark et al. (2002) conducted a cost-effectiveness evaluation of the VAWA comparing implementation costs with potential benefits such as reductions in costs associated with direct property loss, health care, police response, services for victims, productivity, reduced quality of life, and mortality. Having estimated the economic benefits associated with the VAWA to be approximately $16.4 billion in the 5 years since its inception, the authors concluded that its provisions are a cost-effective and beneficial public policy.

Many studies have been conducted to describe the nature and characteristics of domestic violence using the National Crime Victimization Survey, or NCVS, including changes in victims' reports to law enforcement. According to the Bureau of Justice Statistics (2000), between 1993 and 1998 only about half of all victims of domestic violence reported the crime to the police. However, reporting increased from 48% of victims in 1993 to 59% of victims in 1998. Similarly, since the 1970s the rates of reporting rape cases to the police increased (Baumer, Felson, & Messner, 2003), and accelerated during the 1990s.

Other studies using the NCVS have examined factors associated with law enforcement's response to cases of domestic violence. For example, Felson and Ackerman (2001) reported that in cases of minor violence, the police are less likely to make an arrest, and victims are less likely to sign complaints when it occurred in an intimate relationship than when the batterer is a stranger. This indicates that the relationship between victim and perpetrator has a significant effect on police arrest decisions. Further, victims are often reluctant to report to the police because of privacy concerns, fear of reprisal, and desire to protect the perpetrator (Felson, 2002). On the other hand, police are more likely to be called and perpetrators are more likely to be arrested in first time incidents of physical assaults against women and incidents that involve injury (Jasinski, 2003).

One of the most controversial issues regarding criminal justice intervention with cases of domestic violence is aggressive law enforcement such as mandatory arrest policies. For example, some advocates argue that arrest might assist a victim in the short term but make the victim more vulnerable to further violence in the long term (Schmidt & Sherman, 1993). Other authors point out that the assumption of aggressive prosecution to protect victims is largely an untested ideological assertion which may facilitate violence because it deprives a victim of control and increases the risk of retaliation from batterers (Danis, 2003; Ford, 2003; Travis, 1998). On the other hand, proponents of aggressive law enforcement argue that the problem lies with lenient law enforcement and sentencing and suggest that the laws should be enforced more strictly (Bohmer, Brandt, Bronson, & Hartnett, 2002; Buzawa & Austin, 1993; Stark, 1996; Zorza, 1994). In sum, while victims' reporting to the police seems to have increased and particularly accelerated during the 1990s, the effect of subsequent criminal justice interventions is unclear.

Very few studies have attempted to examine domestic violence using trend analyses. Lawrenz, Lembo, and Schade (1988) used time series analysis to examine the effect of a mandatory arrest policy and found a slight increase in the number of arrests per day. There appears to be no studies utilizing time series analysis with the NCVS data to describe and examine domestic violence trends. Therefore, the present study, which utilized interrupted time series analyses of data from the NCVS, can provide some insight into the relationship between passage of the VAWA and its subsequent impact on reducing domestic violence, and can also facilitate future studies to enhance the understanding of long-term trends of domestic violence. Particularly, this study seeks to answer the following questions: 1) Did passage of the VAWA reduce the incidence of domestic violence? 2) Did passage of the VAWA increase the rate of reporting violence to the police? 3) Did passage of the VAWA increase the rate of perpetrator arrest? 4) Did passage of the VAWA increase the rate of victims' contact with other criminal justice authorities (e.g., District Attorney, etc.)?

Methods

Data

The analyses are based on data from incident files of the NCVS. The U.S. Census Bureau conducts the NCVS annually on behalf of the Bureau of Justice Statistics, gathering detailed crime data from a nationally representative sample of households. The NCVS, previously called the National Crime Survey (NCS), was revised in 1992 and consequently, NCS data are not compatible with NCVS. Therefore, data for the present study is limited to NCVS data from 1992. The NCVS asked respondents if they were criminally victimized during the previous six months. For each victimization incident, the respondent is asked detailed questions, the answers to which comprise the NCVS incident files.

While the full NCVS sample includes individuals age 12 and older, our sample excludes females under the age of 18 as these events are primarily addressed by juvenile- and school-related interventions rather than by the VAWA. Incidents involving violent crime perpetrated by a current or former spouse or boyfriend, including rape, sexual assault, robbery, aggravated assault, and simple assault, were selected for analysis. The final sample includes 2,368 female victims of violence within an intimate relationship.

Dependent variables

Domestic violence incidence rate. The domestic violence incidence rate is defined by the Bureau of Justice Statistics as the rate of females over the age of 18 who were victims of violent crime, including rape, sexual assault, robbery, aggravated assault, and simple assault, committed by a current or former spouse or boyfriend (Bureau of Justice Statistics, 2000). To examine change in the domestic violence incidence rate following enactment of the VAWA, population estimates were generated from the NCVS. Estimates of the incidence rate of domestic violence were obtained by dividing the population estimate provided by the Bureau of Justice Statistics by the population provided by U.S. Census data.

Rate of reporting to the police. To determine if police became involved in the incident, the NCVS asked victims, "Were the police informed or did they find out about this incident in any way?" The rate of reporting to the police was obtained by dividing the number of victims who indicated they reported the violence to the police by the total number of victims.

Arrest rate. For victims who involved the police, the NCVS asked, "What did they do while they were (there/here)?" The question consisted of a range of response categories including arrest. An affirmative response to the category of arrest was included. Arrest rates were obtained by dividing the number of arrests made by the number of police responses.

Rate of contact with other authorities. To see whether other authorities except the police became involved in the incident, the NCVS asked respondents, "Have you had contact with any other authorities about this incident (such as prosecutor, court, or juvenile officer)?" The rate of contact with other authorities was obtained by dividing the number of respondents who had contact with other authorities by the total number of victims.

Hypotheses

The present study examines the effect of the VAWA on the interaction between victims and the criminal justice system. Specifically, the study tests the following hypotheses:

1. The domestic violence incidence rate will decrease following enactment of the VAWA.

2. The rate of reporting to the police will increase following enactment of the VAWA.

3. The perpetrator arrest rate will increase following enactment of the VAWA.

4. The rate of contact with other authorities will increase following enactment of the VAWA.

Data Analysis

The study used an interrupted time series design (Campbell & Stanley, 1963) and used SPSS v.11.5 for data analysis. The intervention is defined as the date of enactment of VAWA. This study uses quarterly data on domestic violence beginning in 1992 and ending in 2003. The VAWA was enacted in September of 1994 so that the quarters prior to the fourth quarter of 1994 serve as the control group and the quarters after serve as the treatment group. As a result, the control group has 9 observations and the experimental group 35 observations for total of 44 observations. The intervention is represented as a 1-0 binary variable. It is important to note that the study is not attempting to explain the totality of trends in domestic violence and its related aspects in 1994 to 2003, but to look for a discrete change in them. Gradual changes are removed by differencing in an interrupted time series analysis. In other words, possible effects of changes over time in the proportion of the population can be controlled by differencing. In addition, frequency distributions of the sample and related comparisons with the U.S. population were also computed.

Results

Sample characteristics

Table 1 compares characteristics of the current sample of victims of domestic violence with victims of other crimes in the NCVS incident files. Domestic violence victims are younger, less likely to be married, more likely to be divorced or separated, and have a lower educational attainment than other victims. The NCVS data reflect marital status at the time of the interview, so it is unknown if the violent episode occurred while the victim was separated or divorced, or if the victim separated or divorced following the incident.

Time series analyses for dependent variables

Figure 1 shows the time series trends for the four dependent variables. First, the time series for the domestic violence incidence rate shows an apparent decrease, in which the rates fell from 1.9 victims per 1,000 women to 0.3 victims between the third quarter of 1992 and the second quarter of 2003. (Note: the data used in this study are quarterly, not annual. Quarterly data should be summed to get annual estimates. Using this formula, there is a decrease in the annual estimate of domestic violence from 7.7 victims per 1,000 women to 3.7 victims per 1,000 women between 1993 and 2002). The time series for the rate of reporting to the police shows a slight increase from 54% to 61% over the same period, although the trend is not as apparent as that seen for the domestic violence incidence rate. The arrest rate dramatically increased from 23% to 51% and the rate of contact with other authorities shows a small increase. This reflects an increasingly aggressive criminal justice response to violence against women.

[FIGURE 1 OMITTED]

The intervention analysis followed conventional procedures. Each of the time series was first analyzed using an autoregressive integrated moving average (ARIMA) model. All of them white noise without any differencing, in which both the autoregressive and moving average parameters are equal to 1. In other words, all of the four time series are well identified by ARIMA (1, 0, 1) model.

The VAWA (intervention) was enacted in the fourth quarter of 1994. One dummy variable was created to represent the timing of the intervention. Prior to the fourth quarter of 1994, the dummy variable was coded as 0, and starting with the fourth quarter of 1994 it was coded as 1. The model was reestimated, including the dummy variable, for the entire time series. Table 2 shows the model results.

Overall, for each of the variables, small intervention effects were demonstrated. However, a significant finding was only seen for contact with other authorities indicating that the VAWA appears to have a positive impact on victims' contact with authorities such as a prosecutor or court official. The rate of contact with other authorities increased 5.5% following enactment of the VAWA. For the remaining three analyses, results suggest that the VAWA had no significant impact on the domestic violence incidence rate, rate of reporting to the police, and arrest rate. Therefore the first three hypotheses were rejected while the last hypothesis was accepted.

Additional descriptions for dependent variables

It is important to note that failing to find an intervention effect does not mean that there is no change at all in the interested variables. Instead, it means that there is no difference in the slope of the variables before and after the intervention. Therefore, it is useful to examine the correlations between time and the three nonsignificant variables for some insights into the overall trends. First, the correlation between the domestic violence incidence rate and time is -.63. This means that domestic violence has been decreasing over time, which one can plainly see in Figure 1. In other words, although the VAWA did not have a unique impact on the domestic violence incidence rate, a variety of efforts, including criminal justice interventions, state legislation, and social services, have cumulatively reduced the incidence of domestic violence.

Similarly, the correlation between the rate of reporting to the police and time is .41, which means that there has been an increase over time in the proportion of victims who report the violent incident to the police. However, descriptive data indicate that the type of crime is related to reporting to the police. For example, rape and sexual assault were reported less often to the police (25.8%) than robbery (74.3%) or assault (61.2%). The correlation between arrest rate and time is .40, which indicates that there has been an increase over time of police making arrests. Similarly, type of crime is related to arrest of the perpetrator. In contrast to the rate of reporting to the police, incidents involving rape and sexual assault most often lead to arrest (94%) compared to robbery (73.2%) and assault (82.7%).

Discussion

Overall, contrary to expectations, the VAWA enactment did not appear to have a unique impact on most of the existing trends of domestic violence measured for this study, including the domestic violence incidence rate, rate of reporting to the police, and arrest rate. However, a positive change in the rate of contact with other authorities following the VAWA was found. On the other hand, both of the time series graphs for all four variables, and the correlation coefficients show the expected changes; that is, the domestic violence incidence rate decreased over time while the other variables increased since the early 1990s.

Why did enactment of the VAWA, the most comprehensive federal legislation addressing domestic violence, not appear to impact the existing trends of domestic violence? First, one might expect the impact of the VAWA enactment to have occurred in later years (i.e., at longer lags) rather than immediately after the fourth quarter of 1994. It is normal that it takes time from the law enactment to its enforcement and implementation. To examine this possibility, additional time series analyses were conducted beginning with the second quarter of 1995 (six months after the VAWA enactment) and the fourth quarter of 1995 (one year after the enactment). There were still no significant changes in the existing trends using this strategy. Therefore this explanation appears not to be plausible. In fact, it may be difficult to detect the lagged impacts of the intervention when there is no "strong background theory (that) permits us to predict a specific lag ... (because) the interpretation of a delayed effect is obscured by historical events between treatment ... and ... effect" (Shadish, Cook, & Campbell, 2002, p. 197).

A second possible explanation is contrary to the first; if the expected changes for the interested variables were already occurring before enactment of the VAWA, the impact of the VAWA would have been minimized. For example, the number of emergency shelters for battered women has increased nationwide since the early 1970s, batterer treatment programs have been adopted since the late 1970s, and state mandatory or pro-arrest policies have been implemented since the mid 1980s. These pre-VAWA interventions might have had a positive impact on domestic violence. Some studies suggest positive trends in domestic violence before the VAWA. According to the Bureau of Justice Statistics (2000), the number of female victims of intimate partner homicide fell an average 1% per year between 1976 and 1998. For rape cases, police notification has increased since the early 1970s and the increase in rates of police notification accelerated since the early 1990s (Baumer et al., 2003). The arrest rates for assaults including domestic violence increased up to 70% from 1984 to 1989 following implementation of pro-arrest policies. However, these trends could not be examined in this study due to the lack of data before the 1990s.

Although the study results reveal little unique impact of the VAWA, it is encouraging that all variables in this study show desirable trends over time. The domestic violence incidence has been decreasing over time, while reporting to the police, perpetrator arrest, and contact with other authorities have been increasing. This indicates that the variety of efforts to address violence against women, including the VAWA, state legislation, and criminal justice interventions have reinforced each other to have positively affected domestic violence. In addition, it must not be overlooked that the VAWA may have intangible and indirect impacts that were not measured in this study such as increased public awareness, which also importantly contribute to positive changes in domestic violence and response by the criminal justice system to victims.

Limitations

These results should be viewed within the context of the study's limitations. First, the study did not include data before the early 1990s. This prevented examining whether any significant change occurred before the early 1990s. In addition, in this context, having unequal time intervals before and after the intervention (9 observations for the control group and 35 observations for the experimental group) may be another weak point of the study. If there had been considerable change in the time series slope before 1992, the analysis based on data beginning in 1992 might have led to erroneous results.

A second limitation is that the date of enactment and when the laws are substantially implemented are undoubtedly different, and could be different in each state. A nationwide survey such as the NCVS could not capture the time difference among states, which might obscure the effects of the federal law enactment.

Conclusion

This study suggests that the Violence Against Women Act has significantly strengthened victims' involvement with criminal justice authorities such as prosecutors and court officers. In addition, study results suggest that the domestic violence rate has been declining and victims' responses to domestic violence have increasingly involved the criminal justice system. This may reflect an emphasis on victim empowerment (e.g. assistance with restraining orders) to the extent that victims have been actively participating in the criminal justice system on their own behalf (Hart, 1996). In that context, the positive effects of the VAWA demonstrated in this study should be seen as evidence that efforts to address domestic violence have had some success, though much work remains.

References

Baumer, E. P., Felson, R. B., & Messner, S. E (2003). Changes in police notification for rape, 1973-2000. Criminology, 41 (3), 841-872.

Bohmer, C., Brandt, J., Bronson, D., & Hartnett, H. (2002). Domestic violence law reforms: Reactions from the trenches. Journal of Sociology and Social Welfare, 29(3), 71-87.

Bureau of Justice Statistics. (2000). Intimate partner violence (No. NCJ-178247). Washington, DC: U.S. Department of Justice.

Burt, M. R., Dyer, D., Newmark, L. C., Norris, M., & Harrell, A. V. (1996). 1996 report: Evaluation of stop grants to combat violence against women under the violence against women act of 1994.Washington, DC: The Urban Institute.

Buzawa, E. S. & Austin, T. (1993). Determining police response to domestic violence victims. American Behavioral Scientist, 36, 610-623.

Campbell, D. T. & Stanley, J. C. (1963). Experimental and quasi-experimental designs for research. Chicago: Rand McNally.

Clark, K. A., Biddle, A. K., & Martin, S. L. (2002). A cost-benefit analysis of the Violence Against Women Act of 1994. Violence Against Women, 8(4), 417-428.

Danis, F. S. (2003). The criminalization of domestic violence: What social workers need to know. Social Work, 48(2), 237-246.

Farmer, A. & Tiefenthaler, J. (2003). Explaining the recent decline in domestic violence. Contemporary Economic Policy, 21 (2), 158-172.

Felson, R. B. & Ackerman, J. (2001). Arrest for domestic and other assaults. Criminology, 39(3), 655-675.

Felson, R. B., Messner, S. E, Hoskin, A. W., & Deane, G. (2002). Reasons for reporting and not reporting domestic violence to the police. Criminology, 40(3), 617-647.

Ford, D. A. (2003). Coercing victim participation in domestic violence prosecutions. Journal of Interpersonal Violence, 18(6), 669-684.

Hart, B. (1996). Battered women and the criminal justice system. In E. S. Buzawa & C. G. Buzawa (Eds.), Do arrests and restraining orders work? (pp. 98-114). Thousand Oaks, CA: Sage.

Jasinski, J. L. (2003). Police involvement in incidents of physical assault: Analysis of the redesigned national crime victimization survey. Journal of Family Violence, 18(3), 143-150.

Lawrenz, F., Lembo, J. E, & Schade, T. (1988). Time series analysis of the effect of a domestic violence directive on the number of arrests per day. Journal of Criminal Justice, 16(6), 493-498.

Lewis, R., Dobash, R. E., Dobash, R. P., & Cavanagh, K. (2001). Law's progressive potential: The value of engagement with the law for domestic violence. Social and Legal Studies, 10(1), 105-130.

Meyer-Emerick, N. (2002). Policy makers, practitioners, citizens: Perceptions of the Violence Against Women Act of 1994. Administration & Society, 33(6), 629-663.

Rand, M. R. & Saltzman, L. E. (2003). The nature and extent of recurring intimate partner violence against women in the United States. Journal of Comparative Family Studies, 34(1), 137-149.

Salazar, L. F., Baker, C. K., Price, A. W., & Carlin, K. (2003). Moving beyond the individual: Examining the effects of domestic violence policies on social norms. American Journal of Community Psychology, 32(3/4), 253-264.

Schmidt, J. D. & Sherman, L. W. (1993). Does arrest deter domestic violence? American Behavioral Scientist, 36(5), 601-609.

Shadish, W. R., Cook, T. D., & Campbell, D. T. (2002). Experimental and quasi-experimental designs for generalized causal inference. Boston: Houghton Mifflin Company.

Sherman, L. W. & Berk, R. A. (1984). The specific deterrent effects of arrest for domestic assault. American Sociological Review, 49, 261-272.

Sherman, L. W., Schmidt, J. D., & Rogan, D. P. (1992). Policing domestic violence: Experiments and dilemmas. New York: Free Press.

Stark, E. (1996). Mandatory arrest of batterers: A reply to its critics. In E. S. Buzawa & C. G. Buzawa (Eds.), Do arrests and restraining orders work? (pp. 115-149). Thousand Oaks, CA: Sage.

Travis, J. (1998). Civil protection orders: Victims' views on effectiveness: National Institute of Justice: Research Preview. Washington, DC: U.S. Department of Justice.

Zorza, J. (1994). Must we stop arresting batterers? Analysis and policy implications of new police domestic violence studies. New England Law Review, 28(4), 929-990.

HYUNKAG CHO

DINA J. WILKE

Florida State University

College of Social Work

Table 1
Sample Characteristics (N = 2,368)

                                     Study          NCVS
                                    Sample         Sample
                                  (N = 2,368)   (N = 67,042)

Age                                  31.3           38.9
Marital status (%)
  Married                            18.2           43.9
  Widowed                             0.9            6.3
  Divorced                           23.5           16.8
  Separated                          24.5            6.1
  Never married                      32.5           26.6
Educational attainment (%)
  High school                        59.2           46.5
  College                            38.2           51.1
  Other                               2.6            2.4
Race (%)
  White                              82.0           83.0
  Black/African American             15.4           13.8
  American Indian/Alaska native       1.7            1.0
  Asian/Pacific Islander              0.9            2.3

Table 2
Impact Assessment Results

Variable                           Effect coefficient    p-value

Domestic violence incidence rate        -0.0003           .330
Rate of reporting to the police          0.0762           .087
Arrest rate                              0.0554           .172
Contact with other authorities           0.0554           .022 *

* [alpha] = .05