Part_B_final

B. Collection of Information Employing Statistical Methods

1. Potential Respondent Universe

The potential National Corrections Reporting Program (NCRP) universe is all movements and year-end population status of all offenders in custody in the 50 states’ prison systems and all movements of all offenders under post-custody community supervision (PCCS; formerly known as parole) in the 50 states and the District of Columbia. Movements of federal prisoners are excluded from the NCRP because BJS obtains these data directly from the Federal Bureau of Prisons (BOP) in its Federal Justice Statistics Program (FJSP), a collection of administrative records from federal criminal justice agencies that BJS uses to track federal criminal case processing. Persons sentenced to prison in the District of Columbia Superior Court enter federal prison, and their movements are tracked by the BOP.

The NCRP universe is defined by cohorts, specifically --

• offenders admitted into state prisons during a year,

• offenders released from state prisons during a year,

• offenders held in state prisons at year-end,

• offenders entering PCCS in the 50 states during a year,

• offenders discharged from PCCS in the 50 states during a year.

The NCRP collects administrative records on each prisoner movement (or year-end status) through state departments of corrections. There are 57 total possible respondents in the NCRP data collection universe including the department of corrections (DOC) in each of the 50 states, 6 separate contacts for PCCS data in six states (Alabama, Georgia, Massachusetts, Nevada, Pennsylvania, and South Carolina) where the DOC does not keep data on persons on PCCS, and the Court Services and Offender Supervision Agency for the District of Columbia, which tracks PCCS data for Washington, DC. For data year 2016, 44 states provided data on prison admissions, 46 states provided data on prison releases, 47 provided information on the year-end stock prison population, 31 provided records on entries to PCCS, and 28 on exits from PCCS. These numbers are expected to increase as states are encouraged to provide more NCRP record types (see item B.3., “Methods to Maximize Response,” below).

BJS does not sample states for inclusion in NCRP, but rather tries to obtain data from all states due to wide variation in the laws, sentencing practices, and socioeconomic and racial/ethnic populations of the states. Failure to capture states with smaller prison populations will only enhance the effects of the large states when the offense, time served, admission and release type distributions are presented at a national level. By collecting data from each state, BJS is able to track individual jurisdictional changes in sentencing practices and answer requests from state legislators.

2. Procedures for Information Collection

The data collection agent for NCRP asks states to upload data files in any convenient format to them to a secure server via Secure File Transfer Protocol (SFTP). All respondents currently provide data in electronic format, and no manual submissions are expected in the future. States are contacted during January for data from the previous calendar year, and asked to submit their NCRP data by March 31; however, BJS and its data collection agent work with states to develop a schedule for data submission that meets the states’ individual needs. In general, BJS “closes” data collection at the end of September, after larger states like Texas and California have had time to prepare and submit their data. After processing by the data collection agent, BJS receives the final files in November. BJS archives the NCRP files at the Inter-university Consortium for Political and Social Research (ICPSR) once a year (generally in December or January of the following year), but if states submit NCRP data after the files have been archived, they are updated in the following year’s submission.

3. Methods to Maximize Response

BJS’s data collection agent will continue to make the annual request for NCRP data as clear and concise as possible, and accept data in any format so that states do not need to process the data to meet specific BJS submission standards. BJS and its collection agent contact data providers solely for the purposes of submission follow-up and clarification, and are careful to not place undue burden on respondents. Examples of the follow-up contact from data reporting year 2017 can be found in Appendix L.

Since the previous OMB clearance for NCRP, BJS and its data collection agent have had considerable success in engaging states to participate in NCRP. All fifty states provided at least one type of NCRP record in 2011-2014, with 49 submitting data in 2015 and 47 in 2016. Some states submit after the traditional data collection period, and can lag 1-2 years behind in their submission of annual data. Multiple states have also submitted previous years’ data to extend back their contributions, including Georgia, which in 2015 provided identifiable data back to 1971.

BJS has taken steps to more directly engage with the corrections data providers and researchers about issues that are germane to understanding prison population change. In conjunction with the National Institute of Corrections’ (NIC) Institutional Corrections Research Network (ICRN), in April 2017, BJS and NIC held a data providers and corrections researchers’ workshop in Aurora, Colorado, with a theme of “Using Risk Data to Drive Decision-Making in Corrections”. The meeting was attended by 59 representatives from 46 states. This was the fifth such meeting sponsored by BJS, and has proven to be very popular with the state data providers as a way to share best practices on data collection and information systems, and to get DOC administrators to use the statistics they develop in policy discussions. At the 2017 meeting, BJS presented plans to improve the NCRP through linkage to other administrative datasets, and methods to use NCRP in BJS recidivism research. Providers from several states gave presentations about how they use their data to address policy research questions in their states, including studies on prisoner misconduct, management of sex offenders, development and use of risk assessment tools for particular subpopulations, and the development of digital dashboards to provide more timely data to administrators and legislators. BJS used the opportunity to discuss with state researchers the key policy issues in their states, to learn about their uses of the data and potential issues that NCRP could address, and to more firmly establish the connections between the data providers and BJS’s data collection agent. BJS plans to hold another data providers meeting in the spring of 2019. These meetings have the added bonus of accelerating submission of the NCRP data in the months and weeks leading up to the event.

During the upcoming collection cycle, BJS will work to maintain 50-state participation in the NCRP, and encourage states not submitting certain cohorts to consider doing so. BJS will also work with individual jurisdictions to improve data quality, and is planning to embark on a project to review and reclassify offense codes from each state into a standardized BJS format. The crosswalks between the states’ offenses and BJS codes are updated each year when new laws are passed, but older laws and those that have been modified require review to ensure they are still converted to the correct BJS offense category.

4. Test of Procedures or Methods

As briefly described in Part A, Item 2 of this submission, BJS proposes to add 3 variables to the prison admissions and year-end custody records and PCCS entry records: a yes/no variable for current U.S. citizenship, the prisoner’s country of current citizenship, and the prisoner’s country of birth. If approved, BJS plans to add these data elements in 2019 (request for calendar year 2018 data).

BJS conducted a telephone survey of all 50 DOC NCRP respondents under the OMB generic clearance #1121-0339 (Generic Clearance for Cognitive, Pilot and Field Studies for Bureau of Justice Statistics Data Collection Activities) from December 2017-February 2018 to determine the availability, quality, and feasibility of obtaining citizenship data on prisoners. Overall, the results indicated that most had information on citizenship. Of the 47 jurisdictions that responded to the telephone interviews, 43 reported that their data systems recorded some information that could be converted by BJS into a yes/no variable to indicate current U.S. citizenship (either a flag indicating citizenship or a country of citizenship which could be converted to U.S. citizen versus non-U.S. citizen). Three agencies did not respond to the survey (California, North Dakota, and Oregon), and four jurisdictions (Illinois, Missouri, Oklahoma, and Rhode Island) do not capture citizenship in their DOC database. BJS also asked whether the jurisdictions captured country of current citizenship and country of birth, understanding that country of birth does not necessarily correspond to country of current citizenship. While 43 jurisdictions reported collecting data on country of birth, only 38 could report country of current citizenship.

To assess quality of the data, BJS asked states to report the source of the offender’s current citizenship data. Many states reported that they used multiple sources, with inmate self-report at admission being the most common (43 states), followed by validation by external data sources including Department of Homeland Security (DHS) or Social Security Administration (SSA) (24), and other law enforcement or court documents that accompany prisoners upon intake (17). A similar distribution of responses were received for the sources of country of current citizenship and country of birth (see table 4).

Table 4. Sources of data for citizenship variables as reported by state departments of corrections

Source of data	U.S. citizenship status (yes/no)	Country of current citizenship	Country of birth
Inmate self-report	43	38	43
Court or law enforcement documents	17	14	8
External data sources (DHS, SSA)	24	18	9
Other	0	0	0
Total number of states that could report variable	43	38	43

Note: States could report multiple sources of data.

Respondents pointed out important caveats to these results: the state DOCs do not know the source of the citizenship information that appears on law enforcement or court documents (this could also be based on offender self-report), and in states that used external validation with federal agencies such as DHS, all reported that they did not attempt to validate every single prisoner’s citizenship. Rather, if an inmate self-identifies as a non-U.S. citizen, the court documents indicate the inmate is not a U.S. citizen, or the DOC has some other reason to suspect the prisoner is a non-citizen, the DOC refers the record on to the DHS’s Immigration and Customs Enforcement (ICE) office for verification. Importantly, respondents offered that they rarely if ever hear back from ICE regarding the inmates, and that even if they are notified of the results of the verification, the DOCs do not always update their records with verified results.

In assessing country of current citizenship, all 38 of the states that collected this variable identified the source of the data as inmate self-report, with 18 obtaining validation from external sources and 14 obtaining the information from law enforcement or court documents. For collection of country of birth, all 43 states relied on inmate self-report, with 9 obtaining some external data validation, and 8 using court or law enforcement documents in addition to inmate self-report. The same caveats identified above for a yes/no variable on U.S. citizenship apply to these two variables: not all prisoners’ records are validated even if the DOC does some external verification, the DOCs do not know the source of court document data, and it is unknown whether data in the DOC files are updated should they receive additional information from external sources such as DHS/ICE.

BJS also asked the jurisdictions whether they would be willing to include citizenship information in their NCRP prison admission and year-end custody records. For a yes/no variable on U.S. citizenship, 36 states would be willing to provide, while 4 said they would refuse to submit, and 3 were unsure whether DOC legal counsel or executive leadership would approve the request. The same 7 who said they would refuse were unsure about providing country of current citizenship. Thirty-one said they would be able to provide country of current citizenship. Thirty-seven would submit country of birth. All of the states that refused to provide these variables cited the lack of verification and general unreliability of the data.

Finally, BJS asked states whether the addition of one or more citizenship-related variables would change the willingness of the state to participate in NCRP. While BJS makes it clear to the DOCs each year that submission of any NCRP variable is completely voluntary and states may opt out of item submission at any time, 2 states indicated that the additional request for citizenship data might cause them to stop participating in NCRP. One respondent voiced this in terms of general displeasure with additional burden on the DOC’s budget since the state uses a contract offender management system for data extraction. Another respondent said that it would not provide citizenship data, and the request might cause them to rethink participation. Several other respondents mentioned larger discussions within state government about continued data provision to federal agencies. BJS hopes that it can encourage continued participation of the two states that indicated they may drop NCRP participation, since these states seemed less concerned with providing citizenship data specifically, and more concerned with the burden and provision of all data.

Despite concerns regarding the quality of the citizenship data housed by state DOCs, the Department of Justice has made collection of these data elements a priority. Therefore, with OMB permission, in 2019 (2018 data reference year), BJS will begin to request states to include a yes/no variable to identify U.S. citizenship, prisoner’s country of current citizenship, and prisoner’s country of birth. States that do not have a binary variable in their system will be able to submit whatever they have, and BJS’s data collection agent will convert this to a yes/no response for each record.

All data collected by BJS undergoes strict and continual evaluation for quality and any variation that may exist between respondents due to their conceptual definitions or data collection, storage, and usage processes. The citizenship variables will be no different. BJS plans to compare citizenship data collected in the NCRP to self-report citizenship data from the Survey of Prison Inmates (2016) and the National Inmate Survey (2019-20) collection, and to data collected by other federal agencies (particularly the Social Security Administration) located in the U.S. Census Bureau’s Center for Administrative Records Research and Applications (CARRA) unit to better understand the coverage and quality of the citizenship data in NCRP.

Importantly, BJS practices transparency in the presentation of its data. Limitations to the data are regularly documented in the text, tables, figures, and methodology of BJS reports, data tools, and archived data sets. It is longstanding BJS practice that prior to the first publication of the data, results are never released outside of BJS. This means that the Department of Justice, federal government, and the U.S. public all have simultaneous access to the same data, results, and methodology. As BJS has done in its annual Prisoners bulletins, variations in the collection, reporting, and quality of the citizenship variables will be highlighted in all analyses undertaken with these data, and documentation stating these issues will accompany the archived datasets so that users can judge their fitness of use.

5. Contact Information

The Corrections Statistics Unit at BJS is responsible for the overall design and management of the activities related to the NCRP collection including: data collection; data elements, definitions, and counting rules; and data analysis and dissemination. The BJS contact for the NCRP is --

Elizabeth Ann Carson, Ph.D.

Acting Chief and NCRP Project Manager

Corrections Statistics Unit

Bureau of Justice Statistics

810 Seventh Street, N.W.

Washington, DC 20531

(202) 616-3496

[email protected]

The current data collection agent for NCRP is Abt Associates, Inc. The contact information for the NCRP program manager at Abt is --

Tom Rich

Senior Associate and NCRP Program Manager

Abt Associates, Inc.

10 Fawcett Street

Cambridge, MA 02138

(607) 349-2753

[email protected]

5