L.A. Police Enlist Computers To Prevent Domestic Abuse

When a Los Angeles man shot his two children and then took his own life after a domestic dispute, the tragedy earlier this month reverberated through his community.

It was also deeply felt in local precinct houses because domestic violence is a source of enduring frustration. In L.A. County alone, 12,000 domestic-violence felony arrests are made annually. More so than any other kind of homicide, domestic-violence deaths follow a predictable, escalating pattern.

"If people follow the right warning signs, many needless deaths could be prevented," says Deputy District Attorney Scott Gordon, who chairs the county committee on domestic violence.

In a nationwide first, the L.A. County Sheriff's Department, Carson division, last week began using a tool that can help decipher and gauge those warning signs: a new computer program that can assess how dangerous a domestic abuser is to his household and help authorities determine what action should be taken.

The "Mosaic-20" software was developed by Gavin de Becker, a nationally recognized violence prediction expert. His programs for the US Marshall's office, California Highway Patrol, CIA, and US Supreme Court have helped analyze the danger posed to governors, legislators, and justices by obsessed fans and stalkers.

The new program will compare known batterers with more than 4,000 other abusers whose actions have led to homicides. By asking first-time and repeat victims 48 questions - including, "Has your batterer recently used drugs?" and "Has he recently purchased a firearm?" - authorities can obtain a printed data analysis that rates the level of danger.

Officials and case managers can then coordinate with other social agencies such as help hot lines, social counselors, or shelters on appropriate strategies.

'THIS brings to each victim the collective wisdom of all the experts and all the research that has been done," says Robert Martin, vice president of Gavin de Becker Inc. A minimum of 13 questions are asked at arrests and other domestic-violence calls. Victims are queried in more depth later at police headquarters.

"Officers on the beat and others who may have no experience in the area can draw on the expertise of the prevailing research," says Mr. Martin, who cautions that the program does not suggest what to do about the pattern of abuse, but merely waves a red flag. Police and social workers can track the history of escalating cases and obtain more nuanced information than is available from police reports.

"The strength of this idea is in bringing together disciplines that usually are not communicating with one another," says Lt. Sue Tyler of the L.A. Sheriff's Department. Often police, court and social workers, are working independently of one another.

The history and context of individual episodes of violence provided by Mosaic-20 will also aid prosecutors and judges in determining court strategies and sentencing. "This is a triage tool in helping courts and cops figure out which cases are worth working on," says Deputy D.A. Gordon.

He notes that there is a domestic-violence homicide once every nine days nationally and that a million women annually complain of some kind of abuse. In L.A. county, domestic violence felony filings have increased 220 percent since 1991 and the sheriff's department and LAPD combined field 30,000 calls annually.

The Mosaic-20 program will become a pilot at the North Hollywood branch of the Los Angeles Police Department in the next few weeks as well. Other divisions in Lancaster and Walnut are slated to introduce the program.

The California Highway Patrol has been using a Mosaic program to assess danger to Gov. Pete Wilson and other prominent figures for about four years. "[It's] been very useful in determining where to place our resources," notes Sgt. Steve Weston.

In Connecticut, Yale University Police use Mosaic programs to protect professors who get threatening letters. "It's great at flagging ... cases before they get to crisis stage," says Assistant Police Chief James Perrotti.