«Urban Problems and sPatial methods VolUme 17, nUmber 1 • 2015 U.S. Department of Housing and Urban Development | Office of Policy Development and ...»
Author Daniel H. Weinberg is currently a visiting scholar in the Social and Decision Analytics Laboratory, Virginia Bioinformatics Institute at Virginia Polytechnic Institute and State University. He formerly held several positions at the U.S. Census Bureau, including Assistant Director for the Decennial Census and ACS (responsible for the 2010 census and the American Community Survey) and Chief of the Housing and Household Economic Statistics Division (responsible for the American Housing Survey, Housing Vacancy Survey, Residential Finance Survey, and many other housing surveys).
References Abowd, John M., John C. Haltiwanger, and Julia I. Lane. 2004. “Integrated Longitudinal EmployeeEmployer Data for the United States,” American Economic Review 94 (2): 224–229.
Abowd, John M., and Julia I. Lane. 2004. “New Approaches to Confidentiality Protection: Synthetic Data, Remote Access and Research Data Centers.” In Privacy in Statistical Databases, edited by Josep Domingo-Ferrer and Vicenç Torra. New York: Springer-Verlag: 282–289.
Abowd, John M., and Simon D. Woodcock. 2004. “Multiply-Imputing Confidential Characteristics and File Links in Longitudinal Linked Data.” In Privacy in Statistical Databases, edited by Josep Domingo-Ferrer and Vicenç Torra. New York: Springer-Verlag: 290–297.
———. 2001. “Disclosure Limitation in Longitudinal Linked Data.” In Confidentiality, Disclosure, and Data Access: Theory and Practical Applications for Statistical Agencies, edited by Patricia Doyle,
Julia Lane, Laura Zayatz, and Jan Theeuwes. Amsterdam, The Netherlands: North-Holland:
Andersson, Fredrik, John C. Haltiwanger, Mark J. Kutzbach, Giordano Palloni, Henry O.
Pollakowski, and Daniel H. Weinberg. 2013. “Childhood Housing and Adult Earnings: A Between-Siblings Analysis of Housing Vouchers and Public Housing.” U.S. Census Bureau Center for Economic Studies Discussion Paper: 13–48.
Christensen, Gunvor. 2011. “The Building and Housing Register,” Scandinavian Journal of Public Health 39 (7): 106–108.
Clark, Sandra Luckett. 2012. “Understanding Differences in ACS and 2010 Census Information on Occupancy Status—Data Collection Methods.” American Community Survey Research and Evaluation Program (May). http://www.census.gov/acs/www/Downloads/library/2012/2012_Clark_01.pdf.
Fish, Samantha. 2013. “Understanding Inconsistencies in Occupancy Status Between the 2010 Census and the 2010 American Community Survey 1-Year Estimates.” American Community Survey Research and Evaluation Program (July). http://www.census.gov/acs/www/Downloads/ library/2013/2013_Fish_03.pdf.
Griffin, Deborah H., Joan K. Broadwater, Theresa F. Leslie, Susan P. Love, Sally M. Obenski, and David A. Raglin. 2004. Meeting 21st Century Demographic Data Needs—Implementing the American Community Survey. Report 4: “Comparing General Demographic and Housing Characteristics With Census 2000.” Washington, DC: U.S. Census Bureau (May). http://www.census.gov/acs/www/ Downloads/library/2004/2004_Griffin_01.pdf.
Kennickell, Arthur B. 1997. “Multiple Imputation and Disclosure Protection: The Case of the 1995 Survey of Consumer Finances.” Washington, DC: Federal Reserve Board of Governors. http://www.
Little, Roderick J.A. 1993. “Statistical Analysis of Masked Data,” Journal of Official Statistics 9:
Little, Roderick J.A., Fang Liu, and Trivellore E. Raghunathan. 2004. “Statistical Disclosure Techniques Based on Multiple Imputation.” In Applied Bayesian Modeling and Causal Inference From Incomplete-Data Perspectives, edited by Andrew Gelman and Xiao-Li Meng. New York: John Wiley & Sons: 141–152.
Lubell, Jeffrey M., Mark Shroder, and Barry Steffen. 2003. “Work Participation and Length of Stay in HUD-Assisted Housing,” Cityscape 6 (2): 207–223.
Mast, Brent D. 2012. “Comparing Public Housing and Housing Voucher Tenants With Bayesian Propensity Scores,” Cityscape 14 (1): 55–72.
Mills, Gregory, Daniel Gubits, Larry Orr, David Long, Judie Feins, Bulbul Kaul, Michelle Wood, Amy Jones & Associates, Cloudburst Consulting, and the QED Group. 2006. The Effects of Housing Vouchers on Welfare Families. Washington, DC: U.S. Department of Housing and Urban Development, Office of Policy Development and Research; Cambridge, MA: Abt Associates Inc.
National Council of Real Estate Investment Fiduciaries (NCREIF). 2013–2014. “NCREIF Data, Index and Products Guide.” https://www.ncreif.org/public_files/NCREIF_Data_and_Products_Guide.
Olsen, Edgar O., Catherine A. Tyler, Jonathan W. King, and Paul E. Carrillo. 2005. “The Effects of Different Types of Housing Assistance on Earnings and Employment,” Cityscape 8 (2): 163–187.
Reiter, Jerome P. 2005a. “Estimating Identification Risks in Microdata,” Journal of the American Statistical Association 100: 1103–1113.
———. 2005b. “Significance Tests for Multi-Component Estimands From Multiply-Imputed, Synthetic Microdata,” Journal of Statistical Planning and Inference 131: 365–377.
———. 2005c. “Using CART to Generate Partially Synthetic, Public Use Microdata,” Journal of Official Statistics 21: 441–462.
———. 2004a. “New Approaches to Data Dissemination: A Glimpse Into the Future,” Chance 17 (3): 12–16.
———. 2004b. “Simultaneous Use of Multiple Imputation for Missing Data and Disclosure Limitation,” Survey Methodology 30: 235–242.
———. 2003. “Inference for Partially Synthetic, Public Use Microdata Sets,” Survey Methodology 29 (2): 181–189.
Reiter, Jerome P., and Robin Mitra. 2009. “Estimating Risks of Identification Disclosure in Partially Synthetic Data,” The Journal of Privacy and Confidentiality 1 (1): 99–110.
Shroder, Mark. 2002. “Does Housing Assistance Perversely Affect Self-Sufficiency? A Review Essay,” Journal of Housing Economics 11: 381–417.
Tatian, Peter A., and Christopher Snow. 2005. “The Effects of Housing Assistance on Income, Earnings, and Employment,” Cityscape 8 (2): 135–161.
U.S. Census Bureau. 2009. American Community Survey Design and Methodology. Report ACS-DM1 (April). Washington, DC: U.S. Census Bureau.
U.S. Department of Housing and Urban Development, Office of Policy Development and Research (HUD). 2013 (July). HUD Research Roadmap, FY 2014–FY 2018. http://www.huduser.org/portal/pdf/ Research_Roadmap.pdf.
Jessica Cigna HousingWorks RI Rebecca Lee The Providence Plan Abstract Housing conditions can vary greatly from one property to the next, but housing characteristics often are measured at different geographic units because of data limitations.
This article discusses the process of connecting address-level datasets to create meaningful analyses at the property level in the absence of a comprehensive address-to-parcel crosswalk. To demonstrate this process, the authors describe linking child lead screening, lead property compliance, foreclosure, and tax assessors’ property records for a U.S.
Department of Housing and Urban Development-funded Lead Technical Study in four Rhode Island core cities. Using the linked data analysis, robust property-level findings can lead to an effective evaluation of policies that affect properties, particularly for urban communities with high proportions of multifamily housing.
Introduction Connecting existing datasets to conduct policy evaluation is a smart way to make the best use of available resources. Administrative datasets across multiple domains contain addresses and can be linked to gain insight regarding housing conditions and policy. In some situations, however, researchers prefer data about entire properties to address-level data when describing housing issues. Many multiunit residential properties have more than one address and, when researchers
try to collect information about all residential units within properties, address listings are often insufficient. This concern is particularly evident for analysis in urban communities, where a high proportion of the housing stock contains more than one unit.
Robust statewide data systems ideally would exist and would enable researchers and city administrators to easily link address-specific data to property-level data. In Rhode Island, as we suspect in many other states, that ideal is not yet the reality. Therefore, extensive preparatory work was completed to conduct a property-level analysis of childhood lead exposure, lead compliance certificates, and foreclosures in four Rhode Island cities. In this article, we discuss the process of connecting a variety of separate address-level datasets with unique variables and coding systems.
We provide background information that defines the study’s purpose and describe how we created a master lookup table, matched our datasets to it, and analyzed the data. We also share the lessons we learned from this effort.
Context The 2005 Rhode Island Lead Hazard Mitigation Act requires owners of nonowner-occupied properties built before 1978 (when residential lead-based paint was banned in the country) to comply with a series of actions aimed at reducing lead exposure. These requirements include attending a lead hazard awareness class, inspecting rental properties, providing tenants information about lead hazards and a copy of the inspection report, responding to tenants’ concerns about any lead hazards, fixing lead hazards, and using lead-safe work practices when performing any maintenance. After the owners comply with the requirements, they receive a Certificate of Conformance, which needs to be kept current.1 For this U.S. Department of Housing and Urban Development-funded study, we sought to evaluate outcomes associated with the Rhode Island law. We identified the number of residential properties that were in compliance with the law, whether lead-exposed children were more likely to reside in noncompliant homes, and whether foreclosure had an impact on lead exposure and compliance.
The analysis centers on lead exposure and other risks that are likely to pervade entire structures, and the unit of analysis was at the property level rather than address level. In addition, the law has implications for property owners, which bolsters the rationale for a property-based approach.
We studied four cities in Rhode Island that have high risks of substandard housing concerns and lead-exposed children: Central Falls, Pawtucket, Providence, and Woonsocket (Healthy Housing Collaborative, 2012). The first results of our analysis, in which we compared blood lead levels of children with a property owner’s compliance with and exemption from the Lead Hazard Mitigation Act and which included a summary of the methods that we used, were recently published and received notable press coverage in local media (Rogers et al., 2014).
Preparing the Data In the four cities we studied, most residential properties have two or more units (U.S. Census Bureau, 2014), and many of those properties have more than one street address. Thus, to obtain State of Rhode Island General Assembly. 2003. Chapter 23-24.6 Lead Poisoning Prevention Act.
accurate property-level counts, we first determined which distinct addresses were part of the same property and then, based on the knowledge of all addresses for each property, aggregated all the address data from various sources to the property level. For example, if three children were exposed to lead at one address and two at another, but those addresses were both part of the same multifamily property, that property housed five lead-exposed children.
Creating a Master Lookup Table To overcome the obstacles associated with having multiple addresses per property, we created a crosswalk tool called the master lookup table, or MLT, which links each address to its property identifier code as well as other basic descriptive data about the property. Our method for creating the MLT differed between Providence and the other three cities. For Providence, the largest city in the state, the MLT was developed to be a more robust resource (as described further in the Providence MLT Online Tool section that follows). Two key pieces that enabled the work for Providence were (1) the availability of an up-to-date parcel shapefile, which identifies the plat and lot numbers for properties and the size and shape of the parcel of land, and (2) a cooperative relationship with the city tax assessor’s office. Having staff members with Geographic Information System (GIS) experience and interns to assist in the time-consuming portions of creating a reliable MLT were also integral to the process.
Another indispensable resource for this work was a dataset of all addresses for occupied and unoccupied structures in Rhode Island, which was initially developed for the emergency 911 telephone and response system. These addresses are available through the state’s GIS data website as a point shapefile (Rhode Island Geographic Information System, 2014). In ArcGIS software, we were able to join the Providence parcel data shapefile with the emergency 911 addresses to create a citywide map of all properties. To increase accuracy, we consulted paper maps from the tax assessor’s office and, on occasion, staff members physically visited properties to verify the address-to-property crosswalk. Today, where available, Internet-based streetview maps can serve to validate addresses or other property information by zooming in on the address in question. Parcels with more than one street address can be easily identified through this combination of data—any address that matches to a given parcel identifier is then linked to that parcel.