U.S. Census Bureau Protections and Controls in the Use of ... · U.S. Census Bureau Protections and...
Transcript of U.S. Census Bureau Protections and Controls in the Use of ... · U.S. Census Bureau Protections and...
U.S. Census Bureau
Protections and Controls in
the Use of Administrative
Records Data
Mark A. Leach, PhDCenter for Administrative Records Research and Applications
(CARRA)U.S. Census Bureau
ICRN/NCRP Data Providers MeetingApril 5-6, 2017
.
1
Agenda About CARRA
Data protection and controls
Acquisition phase
Ingest/processing phase
Provisioning phase
Product release phase
Penalties
Data Protection Summary
Data Requested and Potential Uses at Census
2
About CARRA
Acquires, processes and provisions administrative data from federal, state and third party providers
Enhances Census Bureau operations and demographic and socio economic research capabilities through record linkage and statistical matching
Increases the use of administrative records and third party data in the Federal Statistical System
3
Examples of Administrative Records
and Third Party Data Sources
Federal Administrative Records
• Centers for Medicare and Medicaid Services
• Health and Human Services
• Housing and Urban Development
• Internal Revenue Service
• Social Security Administration
• U.S. Postal Service
State Administrative Records
• Supplemental Nutrition Assistance Program (SNAP)
• Temporary Assistance for Needy Families (TANF)
• Special Supplemental Nutrition Program for Women, Infants, and Children (WIC)
• Low Income Home Energy Assistance Program
Third Party Data
• Corelogic
• Experian
• InfoGroup
• MCH
• MDR
• Melissa Data
• RealtyTrac
• Targus/Neustar
• VSGI
4
Examples of
Administrative Records Data Sources
Federal Administrative Records
• Centers for Medicare and Medicaid Services
• Health and Human Services
• Housing and Urban Development
• Internal Revenue Service
• Social Security Administration
• U.S. Postal Service
State Administrative Records
• Supplemental Nutrition Assistance Program (SNAP)
• Temporary Assistance for Needy Families (TANF)
• Special Supplemental Nutrition Program for Women, Infants, and Children (WIC)
• Low Income Home Energy Assistance Program
Third Party Data
• Corelogic
• Experian
• InfoGroup
• MCH
• MDR
• Melissa Data
• RealtyTrac
• Targus/Neustar
• VSGI
5
Nutritional Assistance Data Project
Joint project with USDA Food and Nutrition Service and Economic Research Service
Supports the acquisition of data from the administration of: SNAP and WIC
Also request TANF data (administered by HHS)
Requires state-level agreements with each program
State participation is voluntary Table package and data visualizations provided to states
Reimbursement for data extraction costs offered
6
Data Protection and Controls –
Acquisition Authority to Acquire Data is Title 13 of the U.S. Code
Section 6 authorizes the Census Bureau to acquire information to assist the Census Bureau in the performance of T-13 duties for statistical purposes only
Section 8(b) authorizes the Census Bureau to
Engage in joint statistical projects with nonprofit agencies or organizations. on matters of mutual interest, upon equitable apportionment of costs
Furnish copies of tabulations and other statistical materials which do not disclose the information reported by, or on behalf of, any particular respondent, upon payment of the actual or estimated cost of such work
7
Acquisition (continued)
Agreement Types
Data acquisition only
Nutritional Assistance Data Project
Data acquisition and joint statistical project
Oregon WIC
Data acquisition and reimbursable work
Requests for proposal and contracts (for-profit entities)
8
Acquisition (continued)
Agreements contain terms and conditions to protect the data
Authorized secure modes of data transfer
Retention period for use of the data
Authorized uses of the data
Provisions to authorize additional uses
9
State
Agreements and
Data Delivered to
CARRA
STATE SNAP TANF WIC
Alabama 2014 – 2016
Arizona * * 2014 – 2016
CA – LA County * *
Colorado 2012 – 2013 * 2011 – 2015
Florida * *
Hawaii 2013 – 2015
Idaho * * *
Illinois 2008 – 2015
Indiana 2004 – 2015 2004 – 2015
Kentucky 2014 – 2015 2014 – 2015
Maryland 2009 – 2015 2009 – 2015
Michigan * * *
Mississippi * *
New Jersey 2006 – 2016 2006 – 2016
New York 2007 – 2012 2007 – 2012
Nevada * * 2006 – 2014
North Dakota * *
Oregon 2009 – 2014 * 2008 – 2016 (JSP)
Pennsylvania *
Tennessee 2004 – 2015 2004 – 2015
Utah *
Virginia 2009 – 2013
Washington 2004 – 2008
Wisconsin 2008 – 2009
• SNAP– 19 agreements
– 11 delivered data
• TANF– 17 agreements
– 7 delivered data
• WIC– 10 agreements
– 6 delivered data
Modes of ingest
Secure FTP
Encrypted disk/drive
Data clearance
Quality check to ensure full data receipt
Record count
Cross-check against metadata
Data Protection and Controls -
Ingest
11
INPUT FILE
Data Protection and Controls – Processing
INPUT FILESSN, N, A, DOB
V1, V2,……VN
Reference FileSourcesInput and Reference File Matched
SSAHHS
IRSHUD
CMSMAF
TIGER
USER FILEPIK, Sequence #
V1, V2,……VN
12
Data Protection and Controls -
Provisioning
Centralized data management system
Each dataset is registered and assigned an information owner
Project requests for datasets are reviewed by information owner to ensure access is authorized (according to acquisition agreement)
If access is approved, server controls are set to make data available to only authorized users
13
Data Protection and Controls -
Product Review and Release
Examples of pre-release reviews
Disclosure review (all products)
Sensitivity review (all products)
Statistical review Ex. conference presentations and working papers
Peer review and fact checking Ex. Working papers and manuscripts
Theoretical and methodological review Ex. Working papers and manuscripts
14
Penalties
Disclosure of Title 13 information without authorization is subject to a federal prison sentence of up to 5 years and a fine of up to $250,000, or both
Depending on source of data, additional penalties may apply (i.e. Title 26 federal tax information data)
15
Data Protection and Controls - Summary
Protections and controls in-place for all phases from acquisition to product deliverable
Data management system ensures approved uses only
Information owner has critical role
Protections and controls are critical to continue expansion of data acquisition and linkage projects
16
Data Requested
Case Unit Data Unit identifiers
Monthly address history
Monthly benefit amounts
Eligibility and denial information
17
Individual data PII -> Census unique identifier (PIK) Relationship to primary recipient Race Hispanic origin/Ethnicity Sex Date of birth Education Employment Monthly income amounts and
sources (earnings, TANF, SSI, SSA, UI, other)
Monthly history of membership in case unit
Potential Uses of Administrative Records
(as agreement allows)
18
Before data collection
• Frame
• Predict
• Contact
During data collection
• Edit
• Impute
• Replace
After data release
• Evaluate
• Model
• Innovate
19
Table 1a. <State> WIC Modeled Eligibility and Participation (Among Eligible)
Rates (%) by Infant and Child Characteristics, Infants and Children, <Year>.
Sources: <State> WIC administrative records (<Years>) linked to 1-year American
Community Survey data (<Year>). Note: A standard error for each estimate is
provided in parentheses in the column labeled SE.
Estimate SE Estimate SE
Total 53.5 (1.0) 55.1 (1.4)
Age
0 54.9 (1.8) 67.6 (2.5)
1 52.9 (2.2) 62.2 (2.8)
2 53.7 (1.8) 57.2 (2.8)
3 53.8 (2.0) 47.7 (2.7)
4 52.5 (1.8) 41.3 (3.3)
Race / Hispanic Origin
Non-Hispanic White alone 38.2 (1.4) 45.5 (2.3)
Non-Hispanic Black alone 81.7 (4.0) 72.8 (5.4)
Non-Hispanic AIAN* alone 83.3 (10.0) 46.6 (11.0)
Non-Hispanic Asian alone 43.0 (7.3) 40.0 (8.0)
Other Non-Hispanic 46.4 (4.2) 60.1 (6.2)
Hispanic 78.8 (1.7) 61.4 (2.1)
Migration in Past Year
No Migration or Age 0 51.2 (1.1) 55.6 (1.7)
Intra-State Migration 65.5 (2.6) 57.8 (3.2)
Inter-State or Intl Migration 52.0 (4.8) 33.4 (8.2)
Eligibility Participation