P.M. lived in terror of her husband long before his beatings sent her to the emergency room. They’d had an arranged marriage in India, and he warned her she’d be deported from their new home in the United States if she told anyone he hit her. She had no money of her own, no job, no driver’s license—and no one she could turn to when she was smacked or choked. But what terrified her most was his chilling weekly ritual.
"The scariest times for me were not the times when he pulled out the gun in a rage, but on Sundays, when he would take it apart in complete silence and just stare at me," explained the survivor, who asked VICE to identify her using only her initials. “He would sit on the dining table and get all the cleaning stuff and all the nozzles, and he would take it apart and put it back together and just look at me."
The gun was legally acquired, and its display, however menacing, would never meet the judicial threshold of a threat. Yet, in the context of what P.M. described as routine beatings and sexual assault, its very presence made her 20 times more likely to die, according to one estimate.
Cases like P.M.’s are maddening for cops, who struggle to keep women like her from being killed by their partners. (There are male victims of intimate partner violence, or IPV, but fatalities are overwhelmingly female.) Domestic violence is rarely reported, and when it is, law enforcement often can’t do much to help: Advocates say it takes any given survivor an average of about seven attempts to leave an abuser, and evidence suggests they're 70 times more likely to die in the days after they do. (More than half of female homicide victims in the US are killed by current or former intimate partners.) Police, too, appear to be more likely to get shot and killed responding to domestic violence than to any other type of crime.
Even so, who among the accused pose a clear and present danger is notoriously hard to tell.
“Risk factors for general recidivism are not the same as for future intimate partner violence,” said Dr. Jill Messing, an IPV expert who specializes in risk assessment and femicide at Arizona State. A single spasm of brutal violence can be less predictive than a jealous spiral or a run of more minor assaults, she cautioned. Yet, “when police officers go out to the scene of a domestic violence incident, they’re looking at what happened in that incident. Often, a particular incident doesn’t tell you the whole story.”
To identify the most dangerous abusers, first responders in dozens of cities and a handful of states across the country now rely on a brief questionnaire, often either the Ontario Domestic Assault Risk Assessment or the Lethality Assessment Program (LAP) – Maryland Model. In some jurisdictions, a high score triggers a referral to shelter or services, while in others it requires an arrest. But a seemingly minor administrative rule change in Arizona last week could help reshape how these tools are used by the courts.
Beginning April 2, the state’s judges and magistrates were required to weigh the results of the Arizona Intimate Partner Risk Assessment Instrument (APRAIS) when making bail decisions at arraignment. Advocates expected it to be used to hold many alleged abusers who would previously have been released, since “90 percent of domestic violence arrests are misdemeanors," according to Allie Bones, head of the Arizona Coalition to End Sexual and Domestic Violence. Experts suggested this made Arizona one of only a handful of states to mandate risk assessment in domestic violence investigations, and the only one to give it real legal teeth.
“What they got passed was very radical in a way,” explained Dr. Neil Websdale, director of Northern Arizona University’s Family Violence Institute and the architect of the APRAIS tool. “Other jurisdictions have risk assessments, but they don’t necessarily go anywhere. This risk assessment goes before the judge and the judge is required to consider it.”
Unlike algorithmic criminal risk models such as COMPAS that are used in bail and sentencing decisions in courts across the country, APRAIS relies on subjective answers to just seven (initial) questions, the results of which are tallied by hand. Four or more “yeses” indicate high risk of a future attack.
“I think this is the direction that a lot of people are headed,” Messing, the researcher, said. Criminal risk instruments are already common—and controversial—in courtrooms nationwide, but risk instrument expert Megan Stevenson told me few jurisdictions employ ones purpose-built for domestic violence, and studies suggest more general tools are poor at predicting it. Umbrella assessments don’t test for factors like reported jealousy or how many children live at home, which can inflame potentially deadly abuse, she explained. “It’s the best evidence that we have that something bad is going to happen in the future, whether it be re-assault, severe re-assault, homicide, or rearrest.”
There’s strong scientific evidence that simply being told her score by a cop or a nurse and being offered interventions can help a victim take “protective action” against future abuse. But whether using that information to proactively jail an abuser is actually helpful remains to be seen.
“To the extent we’re using risk assessment to intervene to avoid negative outcomes, I think that’s important,” Messing told me. “To the extent that the criminal justice system can do that, I’m not sure.”
In New Jersey, one of just a handful of states where police are already required to administer domestic violence risk assessments (in this case, the ODARA), judges are specifically prohibited from seeing the offender's scores. In Maine, police have to give the bail commissioner a copy before a recommendation is made, but ODARAs are not transmitted to the judge. Like Oklahoma, which uses a version of the LAP, those states employ risk scores to tailor services and inform arrests. But they don’t let them dictate detention, in part because even the best instruments are only somewhat predictive of future harm.
According to the experts behind the new policy, an Arizona woman with four or more “yeses” on her APRAIS is about ten times more likely to be seriously injured (or reinjured) by her partner than one with fewer than two—and 15 percent of those with high scores will suffer "severe re-assault" or "near-fatal violence" within the next seven months.
“Obviously if we could ask 100 questions, we could get greater resolutions, but we don’t have time for that,” Websdale said during a webinar in January. “The Arizona tool has seven questions because that was the compromise number of questions the community we developed this with said would work,” he told me when we spoke in April. “These were honed out of the white heat of community feedback, including survivor feedback, and a lot of it.”
P.M. was among those survivors.
“I would have said yes to each and every one of those questions,” she told me. “If this document is going to be presentable to the judge, I said, let’s do this. If I had that [in the case against her ex-husband], I don’t know what the [result] would have been, but it couldn’t have been what it was," where her ex was acquitted of the abuse that sent her to the hospital.
Websdale, who is also the country’s foremost expert on domestic violence fatality review, was even more blunt. Though he said risk assessments have protective value on their own, and still more when they’re offered with services, he told me existing interventions are simply too weak and too slow.
In other jurisdictions, “yes the police do it, yes they make a referral to the shelter, but, to be candid, so what? The offender’s still out there and nothing has happened to the offender,” he said. “Unless you’ve got a mechanism for getting it in front of a judge, it’s not going to go anywhere.”
Both Messing, the researcher, and Bones, the victims’ advocate, told me they worried about how the new rule might impact communities of color, which already struggle with disproportionate rates of incarceration. Arizona is one of about two-dozen states with some form of mandatory arrest law in cases of apparent domestic violence, and has one of the highest rates of “dual arrest," where both the victim and the perpetrator are detained, Bones told me. The Arizona approach gives judges the authority to hold a person arrested for a misdemeanor without bail, regardless of the severity of the crime for which they were arrested.
At a minimum, the advocate said she hoped the new system would bring down dual arrest rates. But her organization has also begun asking whether mandatory arrest is even necessary in a state where risk assessment is now universal.
“We’re really trying to address something that we see no longer has any kind of utility,” she told me. “We have all these laws on the books, and people are still dying.”
I asked Dr. Messing whether she thought risk-informed bail decisions could save lives where mandatory arrests had failed.
“I think we’re doing the best that we can within the structure that we have,” the researcher said.
Sign up for our newsletter to get the best of VICE delivered to your inbox daily.
Follow Sonja Sharp on Twitter.