L.A. County computer screening could produce get-out-of-jail card

In a jail system facing overcrowding and under growing pressure to release inmates early, one of the most difficult questions confronting the Los Angeles County Sheriff’s Department is: Who do you let out?

Officials hope a key part of the answer is computer software that can sift through a matrix of “psychometric” data, including a 137-question survey, and help identify inmates who seem least likely to commit new crimes.

The questionnaire delves into personal histories: Were your parents divorced or separated? Have you or your friends ever used drugs? Have you ever belonged to a gang? The questions also probe inmates’ personalities and emotional makeup, including their ability to manage anger: “Some people see me as a violent person. Do you ‘strongly disagree,’ ‘disagree,’ ‘not sure,’ ‘agree,’ ‘strongly agree?’ ”

The department has already surveyed about 3,400 inmates, and the screening program placed about 45% in low-risk categories, meaning that they could be eligible to serve their sentences outside of jail with electronic monitoring.


Reducing the inmate population is crucial because jails are filling up, mostly due to the state’s prison realignment program that is shifting responsibility for more criminals to local lockups. Los Angeles County could begin releasing more low-risk inmates — perhaps thousands — to house arrest in the coming months.

The high-tech screening approach marks a sharp departure from the way jailers have historically decided which inmates should be released early. The computers, officials say, add a scientific security blanket of sorts to a process that now relies mostly on simple guidelines with a dose of gut instinct.

Before the software can be deployed, however, sheriff’s officials hope to convince the county’s elected Board of Supervisors that the change can be made without creating new threats to public safely.

“Willie Horton is still one of the most powerful political symbols out there,” said UCLA public policy professor Mark Kleiman, referring to the Massachusetts murderer who in 1986 was let out of prison on a weekend furlough only to rape a woman and kill her fiance. Horton’s release was used to help dash the 1988 presidential hopes of Democratic nominee Michael Dukakis, then Massachusetts’ governor.

The Sheriff’s Department has come under fire in the past for releasing inmates before their sentences were up to avoid overcrowding, only to have them rearrested for new crimes, including murders.

About 100 Los Angeles County inmates now participate in a house arrest program for low-grade offenders who are willing to wear electronic monitors. Sheriff’s officials say the new software could help them shift hundreds, even thousands more to various forms of house arrest and slash the jails’ roughly 19,000-inmate population.

The same system is being used in Broward County, Fla., where it helped trim the inmate population so dramatically that officials were able to close one of five jails and save taxpayers money.

Sheriff’s officials stress that whatever data the new computer program spits out, decisions about releasing convicts would still be individually weighed and reviewed by law enforcement personnel and possibly others.

“I can almost guarantee that if we take the proper precautions that public safety will not be compromised,” said Los Angeles County Sheriff’s Correctional Services Division Chief Alexander Yim, who oversees inmate-release programs for the nation’s largest jail system.

Dubbed COMPAS — an acronym for Correctional Offender Management Profiling for Alternative Sanctions — the screening program uses proprietary software developed by Northpointe Inc., a Colorado-based criminal justice and research consultant. The Sheriff’s Department has signed an initial $75,000 contract to use the program.

Figuring out how to implement it is primarily the work of Sheriff’s Sgt. Ryan Vienna, a 28-year-old custody division deputy with a background in computer programming. He has been testing COMPAS, which uses scripted interviews, the inmate’s criminal history and jailer observations, in addition to the questionnaire.

Northpointe recommends each agency tailor the program to the skill level of its jail examiners.

So far, Vienna and other sheriff’s employees have administered the surveys to select groups, among them prisoners in wheelchairs and females convicted of petty theft and drunk driving, as well as a random sample of inmates from the jail’s general population.

Once inmates are questioned, the Northpointe software uses a 10-point scale to gauge the probability that they would commit a new crime or another violent act if released to house arrest. Inmates with the highest numbers would remain behind bars, while those with the lowest would qualify to be sent home with an electronic monitoring device, or be subject to frequent unannounced visits or curfew checks by authorities.

On its website, Northpointe notes that there are “a number of interpersonal nuances … such as demeanor, eye contact [and] body language” that its software cannot pick up. But the program does contain a “lie test” designed to alert jail examiners when an inmate gives inconsistent responses.

Northpointe acknowledges its program has limitations in assessing inmates who suffer from mental illness. The company recommends that jail examiners seek advice from medical professionals, but generally contends its program “works well between gender and ethnicities.”

Last fall, the Vera Institute of Justice, a nonprofit research firm hired by the county to review jail crowding issues, raised a question about that last Northpointe claim. Vera cited a 2008 research study that found that COMPAS showed weak results when predicting whether black male inmates were likely to be arrested again.

In Los Angeles County’s jail system, where nearly 30% of the inmates are black, that could be significant. But the county consultants also found the study period and sample size was small and more recent reviews did not identify similar problems.

Before any prisoner could get out of jail, his or her case would be reviewed by at least three more people, officials stress. Who they would be — sheriff’s officials, elected leaders, independent experts — has not been determined.

Los Angeles County Supervisor Zev Yaroslavsky said he would oppose any program that could endanger public safety. Still, he recently asked the Sheriff’s Department to report back on the COMPAS program.

“The current situation is one that forces the sheriff to kick people out the door,” Yaroslavsky said. “Necessity is the mother of invention.”