For this report, The Times examined electronic records provided by the Los Angeles County Sheriff of about 2.2 million bookings into county jails from 1999 through 2005. Sheriff's department officials declined to release the names and court case numbers of inmates, citing laws prohibiting the release of criminal histories. The department did assign each inmate a unique number, allowing The Times to track inmates as they came in and out of the jail.
For each inmate sentenced to serve time in county jail, The Times calculated the number of days between the projected release date and the actual release date. Projected release dates were based on sheriff's calculations, and take into account time already served and time off for good behavior.
The Times only considered a release date to be early if it was at least three days prior to the projected release date. This was because the sheriff's department, after having to pay $15 million in 2002 to compensate inmates kept past their release date, has a policy of letting all inmates out 48 hours early.
If a judge sentenced an inmate only to time already served, that inmate was not included in the analysis, because he was never eligible for early release.
If an inmate was released early, The Times searched the data for a subsequent arrest of that inmate on or before the projected release date. For each release, only one subsequent arrest was counted, even though, in some cases, former inmates were arrested more than once during a period covered by a single prior sentence.
Inmates released on bail, bond, citation or other promise to return for a future court appearance were not counted as either having served a full term or having been released early.
The analysis was broken into two equal periods representing the 3-½ years before and after reductions in jail capacity began in mid 2002. The number of release records was unusually low during one three-month period in 2001. The Sheriff's Department said the anomaly probably reflected missing records.
To compare the recidivism of those released early and those serving full term, The Times searched for subsequent arrests of all inmates within 90 days of release. Only one subsequent arrest was counted per release.
The comparison of criminal activity by those released early and those serving full term was based on charges filed after an arrest. The Sheriff's data did not include conviction data.
The Sheriff's data also did not include the date of offense, only the date of arrest. In some cases, a former inmate may have been arrested during an early-release window for a crime actually committed before the arrest leading to the early release. This is especially likely with murder charges which often involve lengthy investigations. Any case cited in the article was independently verified.
The Times used court records and interviews to identify by name 16 men who were re-arrested for murders that occurred during time they would have still been behind bars if they served full sentences.
As part of its analysis, The Times classified more than 5,000 statute citations into 31 crime types using the California Attorney General's crime classifications a guideline. Crimes were further collapsed into six categories. The Times did not count immigration charges as a rearrest because they almost always occurred on the date of release, indicating that the person was booked out of the jail but then held for federal authorities.