Dismissals, case outcomes often don't make it into state criminal history database

The State Auditor's Office has released a report on criminal justice information systems at the Texas Department of Public Safety and the Department of Criminal Justice, raising questions about the accuracy of criminal history data at both agencies. The Fort Worth Star-Telegram has a brief story on the subject, but let's dig into the audit (pdf) in more detail.

The biggest issue at DPS is a failure to record more than 1/4 of "dispositions," i.e., outcomes, related to arrests logged into their system, primarily because "DPS does not have authorization ... to penalize" prosecutors and courts who don't submit such data. "As of January 2011, prosecutor offices and courts had submitted disposition records to the Computerized Criminal History System for 73.68 percent of arrests made in 2009.

Such omissions can lead to critical errors, as demonstrated by the case of Walter Rothgery, whose false arrest for being a felon in possession of a firearm (he had never been convicted) happened because the outcome of a prior case had never been entered into the system. Lots of people are arrested for things they're never convicted of, and failure to enter disposition data into the system means many of them will be blamed in the future for charges that were dismissed.

According to the audit, of records submitted to DPS between Sept. 1, 2009 and Nov. 30, 2010, an astonishingly low "47,051 (63.61 percent) of 73,967 records for criminal charges that prosecutors dismissed were submitted within 30 days as required." You'd think dismissals - clearing people's names from un-adjudicated charges - would be among the most important information to ensure gets into the system, but more than 1/3 of dismissals aren't reported in a timely fashion.

Poor information gathering on the front end leads to gaps in information for decisionmakers later in the process. Remarkably, examining a one month sampling of data, auditors identified 1,634 offenders who began serving sentences in prison, jail or on probation in November 2010 for whom the Computerized Criminal History System contained no sentencing information from courts or prosecutors. For offenders sent to jail or prison that month, a whopping 19.29 percent of records at TDCJ failed to include an "arrest incident number," meaning the state can't tell for sure the arrest event associated with their incarceration.

Prosecutors and courts, though, aren't the only ones failing to forward sufficient data upstream. The system "lacks arrest records to match with at least 65,424 prosecutor office or court records collected between September 1, 2009 and November 30, 2010. DPS could not match those records because law enforcement agencies had not submitted arrest records appropriately or because the prosecutor offices and courts submitted erroneous data that prevented DPS from matching records."

Counties recording complete data in less than 90% of cases are supposed to submit plans to DPS for improving their record keeping, but that relatively recent requirement doesn't seem to have resulted in substantial improvements in reporting local case data to the state.

Making matters worse, a majority of disposition data submitted by law enforcement is inaccurate. Auditors sampled records from four law enforcement agencies - Garland PD, Houston PD, and the Harris and Kerr County Sheriffs - and found that 59.55% of arrest records "incorrectly showed that the individual was held in the custody of the law enforcement agency when the individual had actually been released." So basically, once you're arrested, there's a good chance state databases will show that you're still caught up in the system even if charges have been dismissed, you've been released from custody, etc..

Other parts of the audit examined information systems at TDCJ and local probation departments that feed them data on community supervision. "Auditors reviewed the error logs for the five largest local probation offices. TDCJ identified errors in 415,453 (22.60 percent) of the 1,838,576 probation records those local probation offices submitted from September 1, 2009 through February 28,2011," meaning more than one in five probationer records contained errors. However TDCJ "was not able to determine how many of those errors remained and how many had been corrected."

I was interested to learn that "TDCJ submits the state identification number of probationers to DPS each day to determine whether probationers were arrested," but nearly half of Texas probation departments do not monitor that data. The notifications to probation departments of probationer arrests are called "flash notices," but "as of May 2011, users representing 120 (47.24 percent) of the 254 counties had not viewed arrest records associated with flash notices in at least 90 days. That included 56 (41.48 percent) of 135 total users in the 254 counties which had not accessed their accounts within a six month period. Those counties are listed in an appendix.

Bexar County, the audit noted, "had not viewed arrest records associated with flash notices in more than a year" and reviewed arrest data in a local system that includes only arrests in that county. "As of May 5, 2011, Bexar County's CSCD did not have a flash coordinator because it was not aware of the flash notice process," which it should be mentioned provides more fuel for critics of that often-dysfunctional department.

Concerns were also raised in the audit that both DPS and TDCJ had staffers authorized to access or alter data who didn't need it for their job description. At DPS, "Twenty-six staff had administrative access that enabled them to modify criminal records, security configurations, and application functionality of the Computerized Criminal History System," but "only one of those individuals required the ability to modify security configurations to perform the individual's job duties." Giving users inappropriate access, said the auditor, "increases the risk of fraud and unauthorized modification of criminal records."

Similarly, TDCJ does not adequately restrict access to its Corrections Tracking System, meaning "a risk still exists that unauthorized changes would not be detected or prevented. In addition, TDCJ does not adequately back up its system, so if "an unprotected database file was deleted, TDCJ could risk losing criminal data and disrupting the availability of the Corrections Tracking System."

Remarkably - and this should be worrisome to local probation directors - one of the auditors' recommendations was that "TDCJ should consider reducing the funds it provides" to probation departments that don't provide all the data on probationers they're supposed to. Somehow I doubt that will happen, but TDCJ said they agreed with the recommendation and at least in theory could do so. DPS, by contrast, has little if any leverage to force courts and prosecutors to improve their data collection.

This is a story reporters in regional markets could localize for the various agencies in their area. See the full report (pdf) and appedices with county level data for more information.