The study compared Travis County, which operates under a risk-assessment model, releases 59 percent of inmates on some form of personal bond, compared to 9 percent in Tarrant County. (Real numbers 60% and 6%; see the update below for clarification from Dottie Carmichael.)
The differences in outcomes were striking.
For starters, the finding that victimization rates were dramatically lower under a risk assessment regimen compared to money bail may be the most persuasive argument they made.
Pretrial detention costs under money bail were 42 percent higher than under a risk assessment model, the study found. At a time when counties are complaining bitterly of unfunded mandates, that's a big potential savings which could be handed to them.
Jail stays - both the initial detention and after rearrest for absconders - were shorter under a risk assessment system.
Poor people get hit much harder under money bail. "Not only are more people detained on a low bond in the financial release system, but a higher proportion of those defendants have a statistically low risk of bond failure." To put a number on it, "three times more people in the financial release system would likely succeed if released but remain in jail because they cannot pay $200 or less for a commercial bond."
MORE: Report author Dottie Carmichael emailed with this clarification:
Tarrant: The correct number for personal bond in Tarrant County is 6%. Travis: The correct number for personal bond in Travis is 60% There must be a rounding difference when you split out PB by supervision category in Figure 4 (no supervision, minimal, and full) because when all 3 groups are combined, 59% becomes 60%. A better summary graphic for this point is on page 33:
As for the guy who thinks it all came from a PIR, we used 3.5 years of individual-level defendant data – 102,000 cases in Tarrant and 61,000 cases in Travis. I am sure you know no one could possibly replicate what we did with a PIR since Travis County had not combined the different data sources or diced the data to answer these specific questions.