In an effort to combat intimate partner violence, state laws governing the warrantless arrest powers of the police in domestic violence cases have been greatly expanded over the past thirty years. All states have empowered the police to make warrantless arrests in cases of domestic violence, and some state statutes have sought to reduce police discretion by mandating specific actions be taken when responding to such incidents. The extent to which states have permitted the police to retain discretion varies considerably. While some states allow police a great deal of discretion, many states require more aggressive intervention. While a mandatory arrest law states that an officer must make an arrest if (s)he finds probable cause to believe that an offense has been committed, a preferred arrest law instructs the responding officer that arrest is the preferred response.
Current research indicates that the passage of mandatory and preferred arrest domestic violence laws has resulted in an increase in arrests for intimate partner violence as well as other relationships included under such statutes. (1) This research also suggests that the increased arrest rate is in part attributable to a disproportionate increase in arrests for females either as a single offender or as part of what is known as a "dual arrest," the situation that occurs when the police arrest both parties involved in an incident for offenses committed against each other. The findings from these studies are limited in that they often include a single jurisdiction or small sample sizes. (2)
There may also be a "net widening" of domestic violence arrest practices because more recent legislation has considerably expanded the scope of relationships covered under such statutes. While initial domestic violence statutes typically only addressed violence between married couples, definitions have been expanded to encompass a broader range of domestic relationships, such as couples with a child in common, (3) persons in a dating relationship, (4) and adults related by blood or marriage. (5) No research has yet evaluated the impact of the new laws on these other types of domestic violence cases. The question to be addressed in cases of nondomestic violence is whether the increased attention legislation gives to intimate partner violence has a similar impact on non-domestic assaults, or, alternatively, limits the resources or willingness needed to provide a similarly aggressive response to those cases.
In this study we investigate the impact of domestic violence legislation on arrest practices in police agencies in nineteen states. In the sections that follow, we: (1) describe what is currently known about arrest practices; (2) describe our research approach; (3) present research findings; and (4) discuss the policy implications and future research needs as a result of these findings.
II. THE INCREASE IN DOMESTIC VIOLENCE ARRESTS
Beginning in the 1970s, political pressure exerted by women's groups, lawsuits brought against police departments for negligence and failure to provide equal protection to female victims in domestic violence situations, (6) and the findings reported by the Minneapolis domestic violence experiment, (7) resulted in a nationwide movement toward arrest as the preferred response to domestic violence. (8) At the core of this movement have been legislative mandates aimed at modifying police behavior. The fulfillment of legislative goals has been evidenced by research reporting increased rates of arrests, prosecution, and conviction, as well as improved responsiveness to victims. (9)
Prior research indicates that the raw numbers of domestic violence arrests increased in many police departments after the implementation of mandatory or pro-arrest laws and policies. Arrest rates from data collected in the 1970s and 1980s were generally in the 7% to 15% range. (10) More recently, however, these rates have been observed to be 30% or greater. (11)
Legislative mandates can be expected to promote the desired change, but compliance may impact organizational behavior in unexpected ways. (12) There is an acknowledged need to determine how change impacts police behavior both in intended and unintended ways. (13) Of particular concern has been research suggesting that domestic violence laws have resulted in an increase in female arrests.
A. THE INCREASE IN FEMALE ARRESTS
An increase in the number of females arrested for assault accompanied the general increase in domestic violence arrests following the implementation of a preferred or mandatory arrest law. (14) Part of the increase in female arrests may be the result of the increase in cases where the police arrest both parties. (15) The first detailed study of dual arrests examined the disposition of domestic violence cases handled by the criminal courts in Connecticut just after implementation of a mandatory arrest policy in 1988; the study found the dual arrest rate in adult intimate family violence cases to be 33%. (16) More recent research has shown wide variations in dual arrest rates. Where statewide data are available for domestic violence cases, dual arrest rates are as high as 23% in Connecticut, (17) are as low as 4.9% in neighboring Rhode Island, (18) and are 8% in Arizona. (19) The percentage of domestic violence offenders arrested who are women also varies. In these three jurisdictions, the percentages are 30.8%, (20) 17.4%, (21) and 28% respectively. (22)
In some cases, dual arrests may be the result of legislation, department policies, or both failing to require officers to identify the primary aggressor. In addition, when such provisions are present, police may lack the training or information needed to identify the primary aggressor when responding to a domestic violence assault. This situation may be compounded by batterers who have become increasingly adept at manipulating the criminal justice system, and may make efforts to "pre-empt" victims from notifying police in order to further control or retaliate against them. (23)
Current political and organizational pressure may discourage officers from arresting women as aggressors, and, unsure what to do, the officers may arrest both parties. This observation is supported by some of the existing research. A 1999 study conducted in Boulder found that male victims were three times more likely than female victims to be arrested along with the offender. (24) Similarly, a study of three Massachusetts towns revealed that male victims were five times more likely than female victims to be the subjects of a dual arrest. (25)
There are other possible explanations for high rates of female single and dual arrests. Police officers, inclined to assume that adult male against female violence involves a male primary aggressor, may find that they are in a situation where the female (according to both parties' admissions and evidence upon arrival) is the primary aggressor. Research suggests that women do in fact commit a considerable number of violent acts in intimate relationships that do not constitute self defense; the same research has emphasized that the women's rates of violence are considerably lower and their acts are less severe than those perpetrated by males. (26) In addition, a comparison of 1980 through 2003 Uniform Crime Reports ("UCR") arrest and National Crime Victimization Survey ("NCVS") data leads to the conclusion that women have not become more violent. (27) While the UCR data showed that females constituted an increasingly higher percentage of arrests for both simple and aggravated assault, the NCVS data did not reveal a concomitant increase in female offending.
B. ARREST IN NON-DOMESTIC CASES
In cases of domestic and non-domestic assault, arrest has generally been infrequent and considered a last resort. (28) Statutes mandating arrest in cases of domestic assault are likely to result in an increase in a more "legalistic" approach to domestic assault resulting in a greater likelihood for arrest in a domestic compared to a non-domestic assault. Because the vast majority of domestic violence incidents involve a female as one of the parties while the majority of non-domestic assaults involve males only, there may be a disproportionate increase in the proportion of females arrested for assault overall as a result.
Research regarding leniency toward domestic violence compared to non-domestic violence cases has resulted in mixed findings. While some studies indicate that the police are less likely to arrest in domestic violence cases, (29) other studies show a consistent police response to domestic and non-domestic violence cases. (30) A critique of this research is beyond the scope of this Article, but it should be noted that major differences in methodological strategies make it difficult to draw any conclusive results. (31) Further, research attempting to examine data nationally has relied on NCVS data, (32) an approach that cannot account for potentially major, and possibly conflicting, practices among police departments. Moreover, prior research suggests that dual arrest may occur less frequently in intimate partner relationships than in other types of domestic violence situations, such as disputes between siblings or a parent and child. (33)
In sum, there are several explanations suggested for the increase in domestic arrests observed in the literature. Most hinge on changes in domestic violence legislation. In this Article, we elaborate on prior research by undertaking a more in-depth examination of domestic violence legislation and by conducting a large-scale empirical analysis of police arrest practice and investigating how the structure of domestic violence arrest laws impacts the decision to arrest.
III. RESEARCH APPROACH
The primary focus of this study is to examine the police response to intimate partner violence and the impact of mandatory and preferred arrest legislation on the police response. However, examining arrest decisions in intimate partner cases without reference to what is occurring in other domestic, and in non-domestic, violence situations poses the risk of concluding that particular arrest patterns are unique to intimate partner violence. We made the decision to limit the study to incidents in which the most serious offense reported to the police was aggravated assault, simple assault, or intimidation (34) because the vast majority of domestic violence cases involve assault. (35) In order to understand patterns and variations unique to domestic violence more fully, we included all cases of assault and intimidation, regardless of relationship. The period chosen for the study was calendar year 2000.
Our primary objective in this Article is to examine the effect of state arrest laws on the police decision to arrest. Consequently, we begin by examining the statutory framework under which the nineteen states included in the database operate and how the statutory framework may affect the police response to both domestic and non-domestic calls. Next, we discern the percentage of cases to which the police respond that result in either arrest or dual arrest, and the circumstances under which such arrests occur. These two tasks are accomplished in the following two sections. The descriptive analysis in those sections provides the background for subsequent multivariate analysis which focuses on the following four questions:
Are arrests, dual arrests, or both more likely to be made in intimate partner violence cases compared to other victim-offender relationship categories (i.e., other domestic, acquaintance, stranger)?
To what extent do mandatory and preferred arrest laws influence the likelihood of arrest and dual arrest in intimate partner cases controlling for other agency, situational, and offender characteristics?
How have domestic violence laws affected arrest outcomes for females after controlling for other agency, situational, and offender characteristics?
Does the impact of domestic violence laws affect arrest circumstances for intimate partner cases only or do effects extend to other victim-offender relationship types?
IV. POLICE DATA SOURCE
The police data for this study are taken from NIBRS. NIBRS provides a rich data source for addressing the goals of this research as it contains incident-based, victim-based, and offender-based information obtained by the police in every jurisdiction that contributes to the NIBRS. As a result, we are able to analyze how the characteristics of incidents, as well as those of the involved parties, impact the decision to arrest one of the involved parties, both of the involved parties (dual arrests), or none of the involved parties.
The non-legal data elements required for this study are contained in various data segments of the hierarchical NIBRS structure. The nested structure of the NIBRS allows up to six segments of information (administrative, offense, property, victim, offender, and arrestee) on each incident reported to the police. (36) The hierarchical structure of NIBRS allows for the study of different units of analysis. For example, it is possible to examine victims and offenders in separate data files with the victim data file including each victim as a case, and the offender data file including each offender as a case. It is also possible to examine the interaction between victims and offenders as the unit of analysis.
In the NIBRS data, a single incident can have multiple records in all of the segments except the administrative data segment. For example, the offense data segment can contain up to ten types of offenses, each of which will have a separate offense segment record. The FBI's UCR hierarchy rule for selecting only the most serious offense in an incident for summary reporting is not used in NIBRS. Therefore, in multiple crime incidents all offenses (up to a maximum of the ten most serious) are reported. The victim data segment in an incident report can contain up to 999 records with each record including detailed information pertinent to each victim. Similarly, the offender segment can contain up to ninety-nine unique offender records, and the arrestee segment can have up to ninety-nine unique arrestee records.
As with any relational database, the hierarchical structure of NIBRS permits linkages between segments. For instance, offenders can be linked with victims and victims can be linked with details related to the offense segment or property segment and so on. These links are important for developing a better understanding of the circumstances associated with intimate partner violence. For example, incident circumstance information, contained within the offense segment, can be linked with the victim-offender relationship details in the victim segment. The victim segment also provides details on race, injury, and specific offenses committed against each victim in the incident. For dual arrest incidents, this detail becomes important for examining the different types of offenses and weapons used on one partner against the other. The linkages between all segments within the NIBRS data also provide details on additional victims and additional offenses within the incident that can help in defining a typology of intimate partner dual arrest incidents.
The victim segment of the NIBRS data provides the offender-victim relationship codes for defining both intimate partner relationships and dual arrests. NIBRS includes a special code to define if the victim is also arrested. When this code is used, a second offender-victim relationship code is also included indicating the actual relationship between the offender and the victim. Consequently, identifying dual arrest intimate partner incidents involves selecting those incidents where the "victim was offender" code is used with one of several codes for defining intimate partners. In incidents involving multiple offenders or victims the closest victim-offender relationship is used for analytic purposes.
V. STATUTORY FRAMEWORK
State law provides the outside parameters within which the police must operate within a particular state by prescribing general police powers and duties. In order to understand the variations that exist among states in police policy and practice, it is important to examine the guidelines provided by individual state laws. As discussed above, there has been a major move toward states enacting mandatory and preferred arrest laws in domestic violence cases since the 1980s. These laws seek to govern police practice in responding to domestic violence calls and enforcing suspected violations of restraining orders.
The NIBRS database was supplemented by variables taken from analysis of state statutes. The variable upon which this article focuses is …