Memo to OMB for NCRP survey on SSN availability

MEMORANDUM

MEMORANDUM TO: Shelly Wilkie Martinez

Official of Statistical and Science Policy

Office of Management and Budget

THROUGH: Lynn Murray

Clearance Officer

Justice Management Division

William J. Sabol, Ph.D.

Acting Director

Bureau of Justice Statistics

FROM: E. Ann Carson

Statistician, Corrections Unit

Bureau of Justice Statistics

DATE: March 5, 2014

SUBJECT: BJS request for OMB Clearance to conduct agency state department of corrections-level survey to determine states’ ability to provide social security numbers of prison inmates to the National Corrections Reporting Program under the generic clearance agreement OMB Number 1121-0339.

Introduction

The National Corrections Reporting Program (NCRP; OMB 1121-0065, expires 10/31/2015) collects administrative records on annual movements of offenders in five cohorts: Admitted into prison, released from prison, held in prison at yearend, entered post-custody community supervision (PCCS) or parole programs and discharged from PCCS/parole programs. Since 1983 through the NCRP, the Bureau of Justice Statistics (BJS) has annually reported on the movements of offenders through state prison systems. These statistics are part of BJS’s core corrections statistics, as they contribute fundamentally to BJS’s mission of describing transitions and movements of offenders through the criminal justice system.

The NCRP data are provided by state departments of corrections (DOC), although in 5 states with separate parole boards, the DOCs submit prison data and parole boards provide the PCCS information. The District of Columbia only provides PCCS data, for a total of 56 respondents. Participation in NCRP is voluntary, and not every state that submits data is able to give each cohort type. In 1983, 32 states submitted at least one record type, while in 2012, 48 states provided data, representing 92% of prison admissions and 87% each of prison releases and yearend custody population. The individual-level data from NCRP allows BJS to describe the demographic and offense distributions of the state prison population, in addition to identifying recidivists and analyzing sentencing characteristics.

In an attempt to leverage federal agency administrative records and expand BJS’s understanding of the pre-imprisonment environment and reintegration of former prison inmates into society without imposing additional burden on state data providers, BJS is requesting OMB clearance to conduct a survey through semi-structured interviews of NCRP data respondents from states’ departments of corrections (DOC) regarding the ability and feasibility of individual DOCs to provide full or partial social security numbers (SSN) of prisoners.

Collection of SSN will allow BJS to link former prisoners to employment, wage and benefits data from the Social Security Administration both before and after imprisonment, subsidized health insurance from the Centers for Medicare and Medicaid Services, food assistance from the United States Department of Agriculture, past military service through the Department of Veterans Affairs, and fact and cause of death from the Center for Disease Control and Prevention’s National Death Index. While BJS’s prison inmate surveys obtain some information on employment and government assistance prior to imprisonment, the questions are limited in scope because of other goals of the data collections (incidence of sexual assault in the National Inmate Survey, and criminal history, health, and demographic questions in the Survey of Prison Inmates). Data on post-imprisonment employment is limited to three questions on the National Former Prisoner Survey to inform whether the person currently has a job or is looking for one, and whether the job is full time or part time. Linking NCRP data to other federal government datasets using SSN will provide key information on re-entry of former prisoners into society, as well as target interventions toward these persons.

The 9-digit SSN variable is used as a key matching variable between federal administrative datasets, including data from the National Death Index, the Social Security Administration (SSA), and Veterans Affairs Data Information Repository. In some cases, a shortened version of the SSN (final 4 digits) has proved successful in data linkage if additional personally identifiable information (PII) including name, date of birth, and race are included in the matching procedure¹. Recent research, however, has suggested that while truncated SSN may work well for matching records of non-Hispanic whites, it performs less well among other racial and ethnic groups², and even full 9-digit SSN matches are biased against minority populations³. Since the racial and ethnic distribution of the state prison population is more diverse than that of the U.S resident population, biases have the potential to seriously affect SSN linkage rates. While use of SSN in data linkage raises privacy concerns, BJS feels that it needs to request full 9-digit SSNs to obtain the best possible linkage. Additionally, SSA does not perform matches to its datasets using anything less than the full 9 digits. BJS will ask states if they could provide the final 4 digits of SSN only if there is a departmental policy or state law against the release of inmates’ full SSN. Since home address can be used with SSN to improve linkage rates, BJS will also ask states whether they can provide home addresses for inmates prior to imprisonment as well.

In informal conversations with 8 state data providers, BJS received a range of responses to whether the DOCs would be willing and able to provide prison inmates’ SSNs. One data provider commented that BJS would need only to add SSN as another variable in its annual NCRP data request, but another said that it would be impossible for BJS to obtain these data from the DOC under any circumstances. The remaining states’ representatives identified a state official (commissioner of the department of corrections, governor, attorney general) to whom BJS would need to apply for permission to obtain SSNs, or were unsure of the rules and procedures for releasing these data to BJS. None of the respondents could definitively say whether limiting the request to the final four digits of SSN, compared to the whole 9-digit number, would result in a different approval process or increased likelihood of obtaining the data.

Request for developmental work

BJS plans to conduct developmental work for NCRP under the generic clearance agreement (OMB Number 1121-0339). Specifically, BJS is requesting clearance to conduct a department of corrections-level survey of 50 states, plus the separate parole agencies from the District of Columbia, Georgia, Massachusetts, Nevada, Pennsylvania, and South Carolina. The goal of the NCRP survey is to determine which states will be able to provide prison inmates’ full SSNs to BJS through the NCRP, and to define the process(es) by which BJS will need to request these data. BJS considers this short survey critical to deciding whether request permission from OMB to collect SSN from states in NCRP. The survey will assist in BJS’s understanding of the data quality and availability of state DOCs’ SSN data, and in planning how to go about collection if OMB ultimately approves the nationwide collection of SSN under the NCRP’s 3-year clearance (OMB 1121-005, expires 10/31/2015). The survey will be administered to 56 respondents.

Design of the Survey

The survey is designed to measure the availability and quality of SSN data on state prison inmates held by state departments of corrections, and to limit the total burden on each respondent. For example, unless the respondents report that they have a specific law or departmental policy against outside agencies requesting SSN data, they will be skipped out of the question on whether they can provide the final 4 digits of SSN (see Appendix A).

Data Collection Procedures

Due to the sensitive nature of social security numbers, BJS anticipates that some state data providers might need to consult other department of corrections personnel. For this reason, we will first email each data provider a week prior to the phone interview with an explanation of why BJS is seeking to collect SSN, as well as a copy of the questions that will be asked (Appendix B). Data providers will be encouraged to collect any necessary information to answer the questions, including names and contact information for state officials to whom BJS will need to apply for permission to obtain SSN.

Each state NCRP respondent will then be contacted by the Abt Associates’ program manager via telephone and asked the five questions. Responses will be entered into an Excel spreadsheet on a secure server, and the file will be encrypted with password protection prior to Abt making transferring it to BJS via its secure FTP site.

Burden Hours for the Survey

Abt Associates will contact respondents from 50 state departments of corrections and 6 jurisdictions’ independent parole boards (District of Columbia, Georgia, Massachusetts, Nevada, Pennsylvania, and South Carolina). Based on informal conversations with 8 states, we anticipate respondents will need a maximum of 15 minutes to prepare responses to the questions, and a maximum of 20 minutes to respond to the survey. An additional 2 hours per state is included in the burden estimate to cover follow-up contacts with other state DOC officials (DOC administration, legal contacts) identified by the NCRP data respondent on the initial call as the appropriate person to whom our request for SSN data should be made.

Table 1. Burden estimate for survey

Tasks	Average burden per jurisdiction	Total estimated burden hours
Initial email receipt, additional time for respondents to gather information if necessary	15 minutes	56 respondents X 15 minutes = 14.0 hours
Telephone survey	20 minutes	56 respondents X 20 minutes = 18.7 hours
Follow-up with state legal or administrative point of contact identified by the NCRP data respondent	120 minutes	56 respondents X 120 minutes = 112.0 hours
Total respondent burden for all respondents = 144.7 hours

Analysis Plan

The goals of the NCRP SSN survey are to: (a) determine which states will be able to provide full SSN to BJS through the NCRP; and (b) delineate the processes by which BJS will need to request these data. If results from the brief survey reveal that most states can provide full SSN and home address, BJS will include these variables in its modification request to the NCRP OMB package in 2015, since this form of SSN would be more useful in matching to other administrative data. If permission is granted by OMB to collect full SSN and home address through the NCRP 2015 OMB package, BJS will use the information gathered from the survey to contact and request submission from the appropriate officials for each state. For states that could not provide 9-digit SSN due to state law, but could provide 4-digit SSN, BJS will request the truncated form of the variable.

Informed Consent and Data Confidentiality

The data being collected in the NCRP SSN survey will only be used to inform future collection of SSN data. The only personally identifiable information to be collected will be the names and contact information of the state NCRP respondent answering the survey questions, and the identified person to whom BJS should address their request to obtain SSN data through NCRP. All information obtained during the survey will be maintained on secure servers at BJS and its NCRP data collection agent, Abt Associates, Inc., and will not be shared with third parties.

Data Security

BJS has a number of layers of protection to ensure the confidentiality of all data:

• All BJS data are physically stored in a secure building in Washington, DC which houses only Department of Justice's Office of Justice Programs (including BJS). The building is staffed 24 hours a day, 7 days a week by armed guards, and employees must pass through an electronic badge swipe and subsequent acknowledgement of their photograph by a guard. Visitors must be sponsored by Department of Justice employees, and submit to a metal detector test as well as record information in a central log book and wear a visitor's badge. Servers containing BJS data are stored in a locked room with access limited only to information technology personnel from the Department of Justice's Office of Justice Programs, and require a badge swipe to enter. OJP has an intruder detection system in the room housing the OJP servers. Should any data need to be stored on CD-ROMs, they reside in a locked office to which only the director and deputy directors have key access, and all data use in this room is logged.

• Technical control of the data is maintained through a system of firewalls and encryption. Data obtained from the NCRP survey will be stored in a computer file on BJS’s secure hard drive behind the Department of Justice's firewall. OJP has an intruder detection system in the room housing the OJP servers. The secure drive was created specifically for NCRP and limits access to only those BJS staff that worked on the NCRP, and the data are encrypted to FIPS 201 standards. Relevant staff are given permission by the BJS information technology specialist to access the drive when they begin working on the project, and they are removed if the discontinue work on the project. Access to this drive requires a username and password verification to log on to the BJS computer system. Only this limited number of staff will be able to analyze data from the SSN survey and will have access to this file (currently, there are 2 individuals with access). All Department of Justice employees are required to undergo annual computer security training.

• Should BJS decide to destroy data in the future, it will follow all federal government guidelines regarding the technical and physical wiping of data from servers, and any CD-ROMs or paper document that may exist will be cross-cut shredded.

Contact Information

Questions regarding any aspect of this project can be directed to:

E. Ann Carson, Ph.D.

Statistician and NCRP Program Manager, Corrections Unit

Bureau of Justice Statistics

U.S. Department of Justice

810 7th Street NW, Room 2243

Washington, DC 20531

Office Phone: 202-616-3496

Fax: 202-616-1351

E-Mail: [email protected]

Appendix A: NCRP SSN Survey Script

1. Does your agency have electronic records with inmates’ 9-digit social security numbers (SSNs)?

   1. If no, does your agency have the ability to link to inmates’ SSNs in another agency’s database?

2. Are the SSNs based on unverified self-reports by inmates, or are the SSNs validated in some way?

   1. If the SSNs are validated, what is the validation procedure? (example – check with Social Security administration; SSN obtained from other agency documents)

3. Does your agency have a standard procedure by which an external agency like BJS can request to obtain 9-digit SSN data for prison inmates for statistical purposes?

   1. YES. [Obtain a copy of the standard procedure from the NCRP respondent – the actual request form, IAA or MOU example, IRB requirement, and/or name and contact information for the person who is in charge of evaluating the request.]

2. NOT REALLY - It has been requested in the past, but always evaluated on an individual basis. [Obtain information from the NCRP on who to contact next – department chief, legal personnel, etc. – who can tell us how to navigate through the system to get the data].

3. NO, AND WE’VE NEVER PROVIDED THESE DATA TO ANYONE….

   1. Because a law specifically prohibits our release of these data [Ask respondent to give you a link to the legal code that says this]

   2. Because it is DOC policy not to give this information out to anyone [Ask respondent to give contact information for the person in their legal and/or IT/research departments who developed this policy so that we can contact them]

   3. Because no one has ever asked before [Ask respondent for contact information for the person they feel would be the appropriate official from whom to request the data]

   4. Other reason: ___________[Obtain reason, obtain contact information for person to whom we should address our questions about this reason]

4. If there is a law or DOC policy specifically prohibiting the release of 9-digit SSN to external agencies for statistical use, does it also preclude the provision of the final 4 digits of SSN?

   1. YES. [Again, obtain link to legal code or contact information for person who developed policy]

2. NO. [Obtain contact information for appropriate person from whom BJS should request the data]

5. Research suggests that home address, when used with SSN, can improve matches between administrative data sets. Does your agency have access to inmates’ home address prior to imprisonment?

   1. If no, does your agency have the ability to link to inmates’ home address in another agency’s database?

Appendix B. Draft email to respondents in advance of survey

Dear [respondent’s name]:

As a part of a continuing effort to expand and enhance the utility of the National Corrections Reporting Program (NCRP), the Bureau of Justice Statistics (BJS) has recently received approval from the Office of Management and Budget (OMB, Clearance # XXXX-XXXX) to investigate the obstacles involved in having states include offenders’ Social Security Numbers (SSN) with their NCRP data submission. Collection of SSN as part of NCRP would enable BJS to link the NCRP to administrative data sets collected by other Federal agencies, which would further our understanding of the challenges released offenders face upon re-entry. As with all data collected by BJS, the SSN and any linked data would be used for statistical purposes only, and maintained securely. We understand, however, that there can be data quality, legal, and other obstacles to providing reliable SSNs.

We have developed a quick 4-question survey to get information about the feasibility of collecting SSN, the quality of the data, and the process specific to your state by which BJS would go about requesting SSN. Abt Associates, the NCRP data collection agent for BJS, will be calling you in the next week to ask these questions. The questions are attached to this email, in case you need to pull together any information.

If you have any questions regarding this project, please do not hesitate to contact me. We very much appreciate your assistance with this effort, and thank you for your participation in NCRP.

Sincerely,

Ann

E. Ann Carson, Ph.D.

Statistician

Corrections Unit

Bureau of Justice Statistics

U.S. Department of Justice

810 Seventh Street, NW

Washington, DC 20531

---------------------------------------

[email protected]

202/616.3496

1 Abreu DA, Daniel K, Iwig B, Hoge S. Four-digits or no-digit social security numbers – Impact on the National Agricultural Statistics Service Record Linkage Maintenance Processes. National Agricultural Statistics Service, Washington, DC. http://www.nass.usda.gov/Education_and_Outreach/Reports,_Presentations_and_Conferences/reports/conferences/FCSM/fcsm%20pii%20paper%20nov09%20-%20iwig091209.pdf

2 Lariscy JT. 2011. Differential record linkage by Hispanic ethnicity and age in linked mortality studies: implications for the epidemiologic paradox. J Aging Hlth 23:1263-1284.

3 Miller E. Differential linkage quality in population subgroups and the contribution to differences in survival rates. Oral presentation to be given at the 2014 International Health Data Linkage Conference in Vancouver, http://www.ihdlnconference2014.org/abstract/Miller.