Preparing Texas for a statewide record clearing program
Last month, I saw something that I can’t shake. It happened during a criminal record clearing conference in Philadelphia, in one of those law firm conference rooms with giant windows that make you feel like you’re floating above the city.
We were discussing the technical aspects of record clearing. A group of about one hundred lawyers from across the country were discussing their state policies, getting down into the details that make this such a complicated topic.
But then a woman stood up, cleared her throat, and pulled us down from the clouds.
“I applied for my dream job,” she said, her voice trembling. “I went through the interviews. They loved me. I got the job. And then they took it away.”
The reason? A mistake on her background check. There was a case that should have been expunged, but it wasn’t. The employer, faced with this information, chose what they saw as the safe option: rescinding the job offer.
When record clearing falls short
Record clearing is not just about the paperwork and policies. People’s livelihoods hang in the balance, along with their dignity and their right to move forward with their lives.
The irony is that record clearing exists to prevent situations like this. It is supposed to give people a genuine second chance. But when the system fails, as it often does, people pay the price.
January Advisors has been working on record clearing technology for several years, developing eligibility algorithms, case management software, and self-service websites with partners across the United States.
Texas represents the biggest opportunity to identify cases eligible for record clearing, eliminating roadblocks to employment and hundreds of other collateral consequences.
Indeed, the state is at the cusp of a transformative opportunity. Working with Texas Appleseed and Senator Royce West, we looked behind the curtain at the Texas Department of Public Safety’s statewide case database. We found that there are over five million cases potentially eligible for expunction or non-disclosure.
This post is a summary of what we found, and how this database could be used to facilitate record clearing programs across the state.
The rules for record clearing are confusing
Record clearing, which includes processes like expunction and nondisclosure, helps individuals remove or seal eligible criminal records. This helps people improve access to employment and housing, and reduce the impact of collateral consequences.
But the process of record clearing isn’t easy. In Texas, there are so many different rules that might apply based on the charge and the length of time since the case disposition.
For example, a DWI from 2017 might be eligible for nondisclosure if it was a first offense and the blood alcohol level was under 0.15. A misdemeanor marijuana charge from 2018 might be eligible for nondisclosure if the person successfully completed deferred adjudication and waited the mandatory five-year period after discharge, but the same charge wouldn’t qualify if they pled guilty and received a sentence of probation.
These rules create a maze that’s difficult for lawyers to navigate, let alone individuals trying to clear their records on their own.
Over the last few years, our team has written code to parse various datasets and determine whether a case is eligible for record clearing. These eligibility algorithms drastically reduce the time an attorney spends on preparing a record clearing petition, since the system does a lot of the qualification work ahead of time.
Through this initial record clearing work, we became acquainted with the conviction database offered for sale by the Texas Department of Public Safety (DPS). This conviction database is a report from all the counties in Texas of cases that resulted in a conviction.
The conviction database is just a subset of a larger Computerized Criminal History (CCH) database. This database also includes cases where the initial disposition was not guilty or dismissed. These cases are crucial to our work in record clearing: they are most likely to be eligible for expunction, which outnumber nondisclosures four to one.
But this larger database is not available to the public. So together with Texas Appleseed and Senator West, we obtained a copy of the full CCH database to analyze for feasibility in statewide record clearing. We rewrote our algorithms for this new dataset and identified over five million cases eligible for expunction or non-disclosure.
This is the good news. However, the data needs improvements in order to scale a record-clearing program across the state. Our analysis is limited to the data that we were provided, so it is possible that many these improvements are simple to address.
Critical fixes needed for a statewide record clearing program
The data provided by DPS is comprehensive, but it needs several critical fixes in order to be useful for a statewide record clearing program:
- Include arresting agency codes. Each case should have a corresponding Originating Agency Identifier (ORI) code. In the CCH database, it refers to the criminal court that processed the case. But we need the arresting agency ORI. It is essential to the record-clearing process, since the agency name is included in the petition for expunction or non-disclosure. Without the arresting agency ORI code, there is no way to determine the arresting agency at scale. It would require reviewing the original record of each case.
- Include tracking incident numbers (TRN) where available. The tracking incident number is required by statute to be included in expunction petitions. However, in order to obtain the TRN, it requires the petitioner to submit fingerprints to the FBI and pay a $30 fee. This is a logistical and financial barrier that prevents record sealing at scale. In order to address this issue, DPS may have to petition the FBI’s Criminal Justice Information Services (CJIS) division and submit a formal request outlining the use case of record clearing, the potential public benefit, and how this effort aligns with existing permissible use guidelines.
- Connect charge codes and charge descriptions to statutes. Almost every case in the CCH database contains a charge description, and many cases also contain a corresponding offense code. Yet the crosswalk provided for these offense codes does not correspond to the records in the CCH database. This connection is crucial because elements essential to eligibility, such as the waiting period, are determined by the specific charge. While we were able to match many of the charges to the statutes, there are still over 8,000 cases without a match. A charge/statute crosswalk from DPS would address this gap, as well as provide higher confidence in our estimates.
- Include the defendant’s name to allow for information validation. The CCH database did not supply defendants’ names. The System Prisoner (“spin”) number is supposed to represent a unique individual. However, the spin numbers provided in the CCH database did not correspond to the spin numbers provided by Harris County. As a result, there was no easy way to match cases provided by CCH with cases provided by the Harris County District Clerk. This is critical because the CCH database contained over 850,000 fewer cases than the Harris County District Clerk data. Without the ability to cross-reference cases, we cannot be assured that the CCH database is complete.
- Include date of birth. Knowing the defendant’s date of birth helps positively identify defendants who have the same name. Additionally, by including the date of birth, we can determine the cases where the defendant is likely deceased. Date of birth also enables researchers to analyze cases where a juvenile was prosecuted as an adult.
Improving data quality to enable better research
There are other enhancements that would provide higher confidence in the data and enable better program evaluations in the future:
- Evaluate race/ethnicity data quality. 32% of the cases that were eligible for non-disclosure did not identify ethnicity (marked in the data as Hispanic or non-Hispanic). This is common in criminal court datasets, where “race” is often noted, but “ethnicity” is not. In Harris County, for example, Latino people are classified as white.
- Understand discrepancies by county. Because the database is a compilation of reports provided by each county, it covers different time periods. For example, Harris County data begins in 1955, while Bexar County data begins in 1967. While it is unlikely that a case processed 60+ years ago will be processed for expunction or non-disclosure, there may be additional discrepancies because of different data collection and historical reporting practices by each county. For example, a county with incomplete data may be missing cases that affect eligibility, potentially creating false positives if there are separate, unreported, and disqualifying offenses.
Three key changes to chart a path forward
When that woman stood up in Philadelphia, she wasn’t just telling her story. She was showing us what’s at stake when our systems fail. Here in Texas, we have the potential to help millions of people avoid that same heartbreak and take advantage of the opportunities they have earned.
But to turn these opportunities into reality, we need three key changes:
- Better data infrastructure. The data provided by DPS is useful, but needs improvements. We need reliable connections between charge codes, descriptions, and statutes. And we need to address potential gaps in the data, including names, dates of birth, and demographic data.
- Stronger coordination among agencies. DPS must include essential identifiers like arresting agency codes and tracking incident numbers. That will involve working with the FBI and NCIC to facilitate easier processing and/or amending what constitutes permissible use. DPS must also ensure that the counties who report their data up to DPS have consistent reporting standards and practices. And legal aid organizations should have consistent, easy access to eligibility data.
- Sustainable funding. Even if the process were streamlined, record clearing takes time and effort. Texas needs to dedicate funding for data system improvements, legal aid organizations need grants to fund this work, and new technology must be developed and maintained to determine eligibility and manage the record clearing process.
My takeaway from this experience: every time a background check shows a case that should have been cleared, we’re not just dealing with a data error. We’re looking at a rescinded job offer and a series of unfortunate misunderstandings.
Texas has the opportunity to lead the nation and use technology to deliver on the promise of second chances. We have the data. Now we need the will to make it happen.