Methodology

In July 2019, in partnership with the Correctional Leaders Association (CLA) and Arnold Ventures, The Council of State Governments (CSG) Justice Center released a report detailing the prevalence of supervision violations among state prison admissions and populations for nearly every state in 2017. This study found that probation and parole violations made up as much as 45 percent of state prison admissions nationwide, with wide variation across states. The findings underscored the importance of supervision policies and practices in states and the broad impact of supervision populations on prison admissions and populations.

In 2020, these partners reengaged state corrections and research directors to build on these findings. Their intention was to (1) develop a routinized data collection mechanism that replicates the prior study annually over the next three years to track progress and trends; (2) help to build state capacity to report on the impact of supervision violations on prison admissions and populations more accurately; and (3) understand how supervision revocation populations change year-to-year.

Updated Survey Collection Instrument

Part of reengaging state leaders involved moving to a new survey platform that improved the efficiency of data collection efforts. Starting with the same questions and definitions established in the report released in 2019, the CSG Justice Center programmed the survey using Qualtrics software. It was then divided into sections:

1. Project Goals, Survey Instructions, and Definitions
2. Prison Admissions
3. Prison Populations

The first section introduces the survey, its goals, how to use the survey platform, and provides definitions. We kept the same definitions from the first survey:

• Total Admissions to Prison/Total Prison Population: The number of people admitted to or in state-funded incarceration as a result of a sentence to incarceration of any length or a violation of supervision conditions.
  • Include people sentenced to a term of incarceration regardless of sentence length (include jail sentences in unified systems). This may include people held in private facilities as well as county-run facilities, provided they have been sentenced or formally sanctioned by a court or supervision authority and their incarceration is state funded.
  • Include people incarcerated for short sanctions in state-funded custody (may include county jails) while on probation or parole/post-prison supervision.
  • Exclude movements such as transfers, returns from bench warrants, and furloughs.
  • Exclude people awaiting trial (i.e., pretrial admissions) or civil commitments.
• Admissions to Prison/Population in Prison on a Violation of Probation Supervision Conditions:
  • Include people admitted on new offense violations as well as technical violations of probation supervision.
  • Include people who have violated probation conditions where probation functions as a post-prison supervision term (e.g., split sentence).
  • Include people incarcerated for short sanctions in state-funded custody (may include county jails) while on probation supervision.
  • Exclude people incarcerated in a county jail or other non-state-funded custody while awaiting a probation revocation hearing.
• Admissions to Prison/Population in Prison on a Violation of Parole/Post-Prison Supervision Conditions:
  • Include people admitted on new offense violations as well as technical violations of parole/post-prison supervision.
  • Include people incarcerated for short sanctions in state-funded custody (may include county jails) while on parole/post-prison supervision.
  • Exclude people who have violated post-prison probation supervision conditions (as they are to be counted in the previous category).
  • Exclude people incarcerated in a county jail or other non-state-funded custody while awaiting a parole/post-prison revocation hearing.

For Sections two and three, respondents were asked to provide aggregate counts of the number of people admitted to or in prison on a given day across numerous categories for calendar years 2018, 2019, and on December 31, for 2020. Respondents were given the option to provide the data according to fiscal year if that is the standard reporting method for the state and indicate if so. The categories requested include total admissions/population; admissions to/population in prison on a violation of probation supervision (total of new offense and technical violations); admissions to/population in prison on a technical violation of probation supervision (technical violations only); admissions to/population in prison on a violation of parole and/or post-prison supervision (total of new offense and technical violations); and admissions to/population in prison on a technical violation of parole and/or post-prison supervision (technical violations only). In these sections, we also provided respondents the opportunity to revise the numbers they submitted in the last survey.

2020 Survey Administration (for 2019 Data)

An initial reengagement survey was sent to the departments of corrections of all 50 states, usually to their research directors in June 2020. Each department received multiple reminders and follow up calls and emails to encourage participation. Due to requests for additional time based on higher than usual workloads and competing priorities related to the COVID-19 pandemic, we kept the survey in the field for much longer than we originally anticipated. Therefore, all departments were allowed to submit data through October 2020.

2021 Survey Administration (for 2020 Data)

In January 2021, another survey was sent to the departments of corrections of all 50 states, usually to their research directors. This additional survey was an effort to obtain more 2020 data, given the special circumstances that the pandemic presented. Multiple follow ups were conducted to departments over several months, concluding data collection in May 2021. Two departments declined to participate (New Mexico and New York), and Tennessee reported their available data for this reporting period would not be available until the end of their fiscal year.

Missing Data

Where available, data provided in the survey was supplemented with publicly available data or data that we routinely collect for other purposes. We note on each state’s page when we used publicly available data.

Data Validation

After the data were collected, the CSG Justice Center developed state profiles that included their survey responses to provide an opportunity for state leaders to review the data and encourage their research departments to fill in missing data elements. CSG Justice Center staff also reached out to individual states when data discrepancies were noticed to confirm the accuracy the survey results. Additionally, an advanced copy of the data was provided to all the departments of corrections prior to releasing it for review.

Caution on Comparing States

States define admissions and populations due to supervision violations differently:

• Some states reported that they include quick dips and supervision sanctions in their admissions numbers, while others do not.
• Some states reported that they classify new felony offenses as technical violations, while others do not.
• States do not necessarily have the infrastructure or institutional memory to pull data in a consistent way over time.
• When provided the opportunity, 19 states revised data that they submitted in 2018.
• States may continue to revise their survey responses during future validation processes.

Final Sample and Data

The first survey wave in 2018 included at least one data point on the prison population from 49 states. The two follow up surveys yielded slightly lower response rates; however, the CSG Justice Center was able to obtain at least one data point for 48 states for 2019 and 45 for 2020.

National Level Estimates

When developing admissions and populations estimates, we used all available data from each state for each year they reported it. Data was not missing at random, so we replaced missing values using the MICE package with multiple imputation when we summed up admissions and violations to the national level. Note that the overall violations admissions and populations do not always sum up to the admissions and populations for the two subtypes, which is due to inconsistencies in the way data was reported across sites. To estimate the averted costs associated with the reduction in the supervision violator populations between 2019 and 2020, we summed the state-by-state differences in populations multiplied by their average yearly costs. We only used imputed values when making national level estimates and never report them at the state level. The data and code can be found in our GitHub repository. The estimates along with their 95% Confidence Intervals were as follows:

Admissions

Populations