Probation revocations down, but not by much; re-arrest rates among DWI probationers plummets

More evidence that Texas' 2007 probation reforms perhaps contributed less than has been previously estimated to recent prison population declines. The main strategy of the 2007 reforms was to reduce probation and parole revocations to prison by incentivizing diversion and progressive sanctions programs. That's worked better on the parole side than for probation (aka "community supervision"). From the Dec. 2012 "Report to the Governor and Legislative Budget Board on the Monitoring of Community Supervision Diversion Funds" (pdf):

Felony revocations to TDCJ in FY2012 represent a 2.8% decrease from FY2005 (677 fewer felony revocations) and a 1.8% decrease from FY2011 (432 fewer felony revocations). However, the percentage of revocations to TDCJ for a technical violation of community supervision conditions increased from 48.5% in FY2011 to 49.0% in FY2012.

Those are essentially insignificant reductions given the scope of the decline in state prison populations witnessed over the last half decade.* Felony technical revocations among probationers declined 10.9% from 2005 to 2012, TDCJ reported, but they're still awfully high and that small decline was far out-paced by two factors on the parole side: Dramatically reduced parole revocations and marginally increased approval rates by the parole board. Both may be viewed as an expression of legislative policy. Reduced parole revocations stem from greater use of intermediate sanctions facilities (ISFs) and other diversion programs created after 2007. And higher approval rates, particularly for low-risk offenders, resulted in large part from the board finally edging closer to targets under non-binding release guidelines that the Lege mandated they create.

County-level probation revocation trends
By contrast, reducing probation revocations has been a tougher nut to crack, in part because of decentralized local control over the process among various counties and judges. Here are the relative increases and decreases for probation revocations among Texas' largest departments since just before Texas' much-vaunted probation reforms took effect:

Change in Felony Revocations to 

TDCJ among largest counties, 2005-2012

CSCD	Percent change in revocations
Dallas	-22.8%
Harris	-17.8%
Bexar	94.0%
Tarrant	-4.3%
Hidalgo	-5.3%
El Paso	-39.6%
Travis	32.1%
Cameron	22.4%
Nueces	1.8%

**See note below on Collin Co.

.

Travis County's increase in revocations surprised me given their department's reputation for reliance on progressive sanctions, etc.. Cameron County attributes their increase to "more aggressive absconder apprehension and increased monitoring of compliance with community supervision conditions." Otherwise, Bexar County is the most prominent, chronic outlier among large counties, as has been the case since these reports began coming out.

2012 probation revocations compared

to supervised population, large counties

CSCD	% 2012 statewide probation pop	% 2012 statewide felony revocations
Dallas	13.6%	10.5%
Harris	11.5%	12.4%
Bexar	6.7%	6.8%
Tarrant	4.9%	7.1%
Hidalgo	4.0%	2.8%
El Paso	3.7%	1.5%
Travis	3.4%	3.0%
Cameron	2.3%	1.9%
Nueces	1.7%	2.2%
Collin	1.7%	1.9%

This chart perhaps provides a better sense of relative county practices than the previous one. It compares probation populations and revocations among large counties as a proportion of their statewide total. (See this data for all counties in Appendix C to the report.) Counties in which the right-side number is significantly greater than the left-hand column may be considered more aggressive at revoking probationers than their peers. That differential is especially significant in massive Harris County because of the sheer volume they process. Tarrant's numbers here are especially striking, putting their paltry 4.3% decline from the earlier chart in context. Meanwhile, Travis, Hidalgo, and even Bexar don't appear nearly as problematic on this chart as they did in the first table.

Recidivism among probationers declining, especially DWI
According to the Dec. 2012 report, 71.7% of felony probationers revoked back to prison in FY2012 were convicted of nonviolent crimes - drug offenses (32.2%), property offenses (30.4%), and DWI (9.1%), with the rest coming from violent (17.9%) and other (10.4%) felony offenses.

Remarkably, and for the most part unheralded, recidivism rates for felony probationers have been declining. "The overall two-year re-arrest rate for the FY2005 sample was 34.4% (8,914 offenders). The overall two-year re-arrest rate for the FY2010 sample was 31.8% (8,811 offenders), which was a decrease from the FY2005 sample."

The drop in re-arrest rates for DWI offenders in those two studies was especially striking: 16.9% of the 2005 cohort was re-arrested compared to 11.5% of the 2010 cohort - a 32% drop! That's a success story nobody tells much. Re-arrest rates for probationers convicted of drug offenses declined 13% over this period; 10.6% for property offenders. But DWI stands out. Perhaps new treatment resources aimed at that group are helping.

* When and by how much Texas' prison population began declining depends on whose count one chooses to use. Counting only those in prisons and state jails TDCJ has reported steady annual decreases for several years that allowed the state to close three prison units. But the USDOJ Bureau of Justice Statistics, whose count includes thousands of probationers in treatment as well as prisoners in county jails awaiting transfer to TDCJ, parole hearings, etc., did not report declines in Texas' state prisoner population until last year. See here for a more detailed discussion of the differences between DOJ's and TDCJ's numbers. 

** Collin County shows an increase in revocations of 84% in this same report. However, their probation director has persuasively argued that since-rectified data-gathering deficiencies - for many years they didn't even gather revocation data (!) - artificially deflated their initial count to make the increase look larger than it really was. Having dug into it a bit over the years, I agree that's probably true, so I didn't include them in this table.