Andrew Reamer - Keynote Address


Census datasets

The U.S. Census Bureau is the key agency that manages and publishes demographic data, which is in turn used for Congressional apportionment, allocating federal funds, and other government and business decision-making. The Decennial Census is an enormous undertaking, with estimated costs for the 2010 Census at $11 billion and constant pressures to cut back on the costs.

The Decennial Census is one source of data that can provide crucial information for crime mapping and decision-making in law enforcement and criminal justice. For crime mapping, the data and analysis needs focus on small areas.

There are recent and ongoing changes in how the Census is conducted. As well, there are new tools and procedures for accessing data online, web-based tools for integrating data from multiple sources, and more advanced statistical techniques, including synthetic microdata and synthetic aggregate data for small areas.

For the 2010 Decennial Census, the Census Bureau is eliminating the long-form which collects detailed demographic information. The long-form is being replaced by the American Community Survey (ACS), which is conducted annually. Differences between ACS and the long-form are that the ACS surveys group quarters including halfway houses and prisons.

The 2010 census will focus specifically on population counts, whereas ACS does population estimates. Inaccurate population counts, which can happen if addresses are missing from the Census address files, can have consequences for the work of crime analysts. Local governments can help out in this regards, by drawing on resources such as 911 records, to improve the address list, as well as encouraging various hard-to-reach populations to fill out the surveys.

American Community Survey

Other administrative datasets are available from the Federal government, including numerous datasets that provide housing-related information. The HUD Office of Policy Development and Research (PD&R) provides the Government Sponsored Enterprises (GSE) public use database, which has information on mortgages purchased by Freddie Mac and Fannie Mae. The Low Income Housing Tax Credit (LIHTC) database provides information on housing projects and units since 1987. The Federal Financial Institutions Examination Council (FFIEC) provides a database of home mortgages granted and denied for each census tract by type of loan, race, and income.

Synthetic Datasets

Synthetic microdata is data that mimic real, but confidential, microdata and produce the same statistical results in terms of mean, median, variance, etc. The Local Employment Dynamics, developed by the Census Bureau, links large federally-funded databases (business establishment data, employee wage records, and social security files) in a way that protects confidentiality but allows the synthetic microdata to mimic the actual data and allow analysis to be carried out.

Another type of synthetic data is illustrated by the National Household Transportation Survey (NHTS), which surveys local and long distance travel patterns. Researchers at the Federal Highway Administration and the Oak Ridge National Laboratories are working on a transferability model, to allow NHTS data to be married with long form Census data.

Date Modified: May 2, 2011