The purpose of this study was to examine victims' health and safety following entrance into the criminal justice system as measured by help seeking patterns across several systems; police, courts and Emergency Departments. The study's main question was whether a victim's participation in prosecution increases her safety, defined as a decrease in subsequent documented IPV events. Using administrative data, the study examined the complex relationship between victim participation in the prosecution process for an index event and future help seeking documented through IPV-related police incidents and future emergency department visits for IPV or injury.
This mixed-method longitudinal study followed a cohort of female IPV victims (N=993) with cases adjudicated in criminal court in the year 2000 in a single Midwestern United States county (Kalamazoo County, Michigan) for a four-year period (1999-2002) across multiple systems (police, prosecutor, criminal court, civil court, hospital emergency department). The research team conducted in-depth qualitative analysis of a stratified random sample of index cases to inform the creation of data abstraction forms. Specifically, the data abstraction forms developed and utilized by the research team include a Criminal Court Records Abstraction Form, Police/Prosecutor Long Notes Abstraction Form, and Emergency Department Abstraction Form. These forms were then piloted and iteratively revised and administrative records were reviewed at the offices of the prosecutor, 12 police departments, 2 criminal courts, 1 civil court, and 8 Emergency Departments. Once the data abstraction forms were finalized, the Project Director conducted trainings with the research staff for the first 150 record reviews until 100 percent inter-coder reliability was achieved. Other data sources utilized by the research team include police incident data, protection order data, prosecuting attorney data, the United States Census 2000, and death certificate data.
The study sample consists of 993 female victims of male-perpetrated IPV as identified through police charging requests to the prosecutor's office in the year 2000 in the study site. Inclusion criteria were: female victim; male defendant; intimate partner relationship (defined as current or former spouse, current or former dating partner, or having a child in common) between victim and defendant; and charging request for assault (including simple assault, aggravated assault, felonious assault, assault with intent to commit murder, homicide, sexual assault, kidnapping, stalking or aggravated stalking). Access to, and permission to use, the prosecutor administrative records were granted by the local Prosecuting Attorney's office and facilitated by the Michigan Department of Community Health. The sample cohort was identified through two methods: (1) delivered as an intact dataset, and (2) manual search and abstraction from prosecutor administrative records. The intact dataset containing 964 of the final 993 sample had been compiled previously as part of a larger CDC-funded study, MEDCIIN (Michigan Department of Community Health, 2003, 2000), and contained charging requests that had charge codes tagging them as domestic assault. This database was further supplemented by 29 higher order cases that had not been previously identified but met this study's criteria. These cases included the following charges: homicide, criminal sexual conduct, kidnapping, assault with intent to do great bodily harm, and assault with intent to commit murder. Because these higher-order cases were not flagged as IPV-related, the supplemented cases were identified through a manual search of the administrative records system of the Prosecuting Attorney's office. All cases for each of the higher order charge codes were reviewed and those that met study criteria were added to the study sample.
Longitudinal: Cohort / Event-based
Female intimate partner violence (IPV) victims with cases for which the police presented the case to the prosecutor in the year 2000 in Kalamazoo County, Michigan followed through the administrative records they generated in criminal justice agencies, emergency departments, and court systems.
Police/Prosecutor Long Notes Abstraction Form: This form was used to capture data elements from the police and prosecutor narrative reports, which provided information on victim and defendant demographics, relationship, history of violence, incident characteristics and criminal justice interactions with the victim. For each index incident, police reports and prosecutor case notes were reviewed. Incidents were not abstracted until the full set of reports had been compiled. Prior to abstraction, all reports for an incident were organized chronologically and were read in their entirety before marking on the abstraction form. Then each report was reread line by line, abstracting pertinent information as it occurred onto the abstraction form. In general, if there were conflicting details, the source most directly involved with a particular detail was considered the most accurate (for instance, if a police report noted that the defendant had fled the scene and the prosecutor notes indicated he was present, the police report was considered the primary source for this detail). When necessary, clarifications were obtained through consultation with the study project coordinator and study investigator team.
Criminal Court Records Abstraction Form: The criminal court abstraction form collected information about criminal justice actions, specifically case disposition and the criminal sanctions imposed. Only cases that had been accepted for prosecution had criminal court records. For these cases, the research team accessed court administrative records to obtain copies of the criminal court dockets. For some higher-order cases, adjudication occurs in Circuit court, and there are two dockets, one District court docket and one Circuit court docket. For these cases, both venues' dockets were reviewed, and the data were combined onto a single abstraction sheet for that case incident.
Emergency Department Abstraction Form: The Emergency Department (ED) Abstraction Form was used to identify all emergency department visits occurring during the study period. Hospital ED data was collected on the study sample of 993 female victims from all eight emergency departments (EDs) in the County area. The research team abstracted information on all ED visits during the 1999 to 2002 study period to identify visits that were related to IPV or injury. To determine if a visit was IPV or injury related, the research team reviewed several types of records for each visit: intake and discharge forms, physician dictations, paramedic and nursing notes, injury body map forms, photographs, social work notes, and violent injury report forms. A visit was coded as IPV or injury-related if the Reason-for-Visit code(s) or the Discharge-Diagnosis code(s) indicated assault or injury or if anywhere in the records there was a notation that the patient was experiencing abuse by her intimate partner, even if that visit was not specifically for an injury.
Prosecuting Attorney Dataset: This secondary dataset was provided by the prosecutor's office, from which the research team obtained variables related to a subset of the police incidents that led to intimate partner assault charging requests. This subset of cases comprised the IPV-prosecutor incident database. Originally, these IPV-prosecuted incidents were part of four intact datasets provided by the Prosecutor's office, one for each study year, 1999 through 2002. Each year's dataset contained all the charging requests with domestic assault charge codes for that year. All prosecutor incident charging requests that were linked to the study couple (where the study victim was the victim in the charging request and the study defendant was the defendant in the charging request) were identified and exported into a single IPV-prosecution dataset. Within this dataset, the charging requests that formed the study index events were marked so that they could be included or excluded from analysis as needed. The remaining, non-index prosecution incidents comprised the longitudinal IPV-prosecution variables.
Police Incident datasets: Police data were obtained for the study sample from the 12 police jurisdictions in the County. Using the names, date of birth and addresses of the victim and defendants in our study sample dataset (N=993), data were extracted directly from electronic datasets, when available, or manually retrieved using structured data abstraction forms. Regardless of data source, research staff and police clerks identified police reports for all incidents from 1999-2002 where (1) both the victim and defendant (the study couple) are listed as parties, or (2) property, drug, check welfare, disturbance, trouble with subject incidents included where the study victim is listed as a victim, and there is no identified defendant/perpetrator. The research team included qualifying police incidents regardless of whether the victim was listed in the role of the victim or in the role of the defendant for the incident.
United States Census 2000: The socioeconomic status (SES) variable was identified by using the defendant's address and United States Census on poverty rates. The poverty rate for the defendant's residential block group was used as a proxy for SES. Defendant address, rather than victim address, was used because the defendant address data were more reliable, and there was less missing information.
Protection Order Dataset: Data on protection orders (PO) were abstracted directly from the family court administrative records database. Only PO petitions that involved both members of the study couple (the victim and the defendant) were included in this abstraction. The research team accessed all family court records associated with each name (victim and defendant). Records appearing with the administrative code associated with protection order petitions and initiated between 1999 and 2002 were retrieved and reviewed to determine if the other party was the sample partner. Cases were included as a PO event if both parties were involved, regardless who was the petitioner and who was the respondent. Data abstracted and entered into the PO database included: role of victim (respondent or petitioner), petition date, whether hearing requested or ex parte order, order date, length of order, termination date, termination status (whether natural termination or by motion), violation date(s) and violation filing method (police report or only victim show-cause motion). Some study couples were involved in multiple petitions. All petitions meeting eligibility criteria were included.
Death Certificate Dataset: Data on date of death for victim and defendant were abstracted directly from the Kalamazoo County Clerk's electronic public death certificate database. Each study victim and each study defendant were searched using their name and date of birth. Only those deaths that occurred during the study period (1999-2002) were searched and recorded in the Death Certificate dataset.
administrative records data
The study contains 717 variables including demographic information, case factors, victim characteristics, defendant characteristics, couple characteristics, event characteristics, incident characteristics, adjudication characteristics, and victim participation. Demographic questions include age, race/ethnicity, children, and employment. Case factors information include relationship of victim to defendant, who called police, arrest outcome, and case outcome. Victim characteristics include whether victim was pregnant at time of incident, victim use of alcohol and drugs at time of incident. Defendant characteristics include block socioeconomic status, defendant use of alcohol and drugs at time of incident, and prior domestic violence conviction. Couple characteristics include marital status at time of incident, cohabitation at time of incident, whether couple had children together, and prior intimate partner violence between couple. Event characteristics include charge class (misdemeanor or felony), and violence severity. Incident characteristics include type of violence by either victim or defendant, injuries by either victim or defendant, weapons used by either victim or defendant, and whether others were present during time of incident. Adjudication characteristics include whether prosecutor authorized charges, classification of charges (misdemeanor or felony), date of final disposition, criminal sanctions, and jail time. Victim participation variables include whether victim called police, communication with prosecutor, type of contact, and whether victim expressed prosecution wishes.