Bail Reform, A Failure In Math
Bail reform often declares success by choosing the denominator that produces the best headline. When failures are counted by the defendant rather than by individual court appearance, many taxpayer-funded programs look far less successful.
The same numbers, two opposite stories
Bail reform supporters love to cite court-appearance rates in the 85 to 95 percent range as proof that releasing people without money bail works fine. Those numbers sound airtight. They are also, very often, the product of a counting method built to flatter.
Eric Granof laid this out plainly in his research at Bail Reform Truth, drawing on AIA Surety's analysis. The core point is simple and correct: how you calculate failure to appear dramatically changes the result, and the two sides of this debate usually don't count in the same way.
Per-appearance math versus defendant math
There are two ways to measure failure to appear. The defendant-based method asks how many people failed to appear at any point in their cases. That is the number courts, surety companies, and bail agents actually care about, because a defendant who skips once is a problem no matter how many hearings they attended first. [1][2]
The per-appearance or aggregate method counts every scheduled hearing and asks what share was attended. A defendant with ten hearings who misses one still "attended" 90 percent of the time under that math. Stack thousands of routine appearances together, and the failure rate looks tiny, even if a meaningful share of actual defendants missed court at least once. [2]
Both numbers can be technically true. Only one of them tells you how many people you have to go find.
Missing once is not a footnote
Reform reports often rely on the idea that most missed appearances are eventually resolved. Some do. A person who overslept and reschedules is not a public safety crisis. But rolling every missed date into a tidy final number hides what matters: how often defendants fail in the first place, how long warrants remain active, and who paid to bring anyone back.
A reminder-heavy program can reduce accidental misses. That is worth something. It is not the same as having a recovery mechanism for the person who meant to disappear.
Apples, oranges, and release categories
The counting problem gets worse when programs blend release types. Lump low-risk recognizance cases in with everything else, and the pooled appearance rate climbs, because the easy cases carry the average. That number is then waved around as proof that supervision works as well as secured bail, even though the two were never measured against the same population.
If you want to compare secured release to unsecured release honestly, you have to hold the risk level steady and define failure the same way for both. Most of the glowing reform statistics do neither. [3]
Public safety needs the same discipline
The same sloppiness shows up in rearrest numbers. A program can report a low new-offense rate by using a short follow-up window, excluding certain charges, or counting only convictions instead of arrests. Change any one of those choices and the number moves. When a program controls its own definitions and its own reporting, skepticism is not cynicism. It is basic diligence.
A minimum reporting standard for Minnesota
Minnesota can end this game with a simple rule. Every release model, public or private, should report on the same terms:
Defendant-based failure-to-appear rates,
The length of the follow-up window,
How rearrest is defined,
The share of warrants still active,
The full public cost of the program.
Same definitions, same denominators, same disclosure. [1] Publish the numbers that way, and the debate stops being about whose press release sounds better. It starts being about what the data actually shows.
Data over narrative
This is not an argument that reform advocates are lying. It is an argument that a headline appearance rate means nothing until you know how it was built. Minnesota should not rewrite pretrial policy on the strength of numbers no one is allowed to check.
SOURCES & FURTHER READING
[1] Bureau of Justice Statistics, Pretrial Release of Felony Defendants in State Courts — Government / official source.
[2] The Fuzzy Math Behind Bail Reform — Industry/Advocacy research and analysis.
[3] Helland & Tabarrok, Public versus Private Law Enforcement — Peer-reviewed research.
[4] Harris County O'Donnell monitoring information — Government / official source.
Additional Resources: https://bail-reform-truth-and-propaganda.ghost.io/
Comments