Presentation on theme: "Panel on Statistical Methods for Measuring the Group Quarters Population in the ACS APDU Paul R. Voss UNC at Chapel Hill & Panel Chair APDU 2010 Annual."— Presentation transcript:
Panel on Statistical Methods for Measuring the Group Quarters Population in the ACS APDU Paul R. Voss UNC at Chapel Hill & Panel Chair APDU 2010 Annual Conference September 21, 2010
Committee on National Statistics (CNSTAT) Established in 1972 at the National Academies to improve the statistical methods and information on which public policy decisions are based Carries out studies to foster better measures of the economy, crime, poverty, and other domains Evaluates ongoing statistical programs and coordination of the decentralized U.S. federal statistical system Activities generally requested and sponsored by specific federal statistical agencies APDU
Some recent CNSTAT reports addressing census & ACS topics
W.A.V. Clark, UCLA S. Copella, PA SDC D.Dolson, Statistics Canada R. Folsom, RTI International R. Harter, NORC S. Heeringa, University of MI J. Salvo, NYC Dept. of City Planning R. Valliant, Univ. of MD P. Voss, UNC (Chair) K. Marton, CNSTAT (Study Director) APDU Panel on Statistical Methods for Measuring the Group Quarters Population in the ACS
Duration: Approximately months –6 meetings –workshop & other activities –interim & final report Charge: In depth review to advise Census Bureau on measuring GQ in ACS –statistical & operational issues sample frame sampling weighting & population controls –data use/user issues who? how? Alternatives? APDU Panel on Statistical Methods for Measuring the Group Quarters Population in the ACS
We need to know… Why? Basis of objection? If this is you or your organization… we really want to hear from you! Note: –I’m asking this question in my role as chair of the Panel –The Panel has not yet discussed this question in any serious way But we recognize that there are some very serious issues here… APDU WHO WOULD OBJECT IF OUR PANEL RECOMMENDED TO THE CENSUS BUREAU THAT SOME OR ALL GQ SUBPOPULATIONS BE PLACED “OUT-OF-SCOPE” IN THE ACS?
Issues… Early indications suggest that there are few users of the ACS GQ data, per se Why would this be? –not much data available yet from ACS –the eventual ACS data for GQ populations is going to be pretty thin –2010 Census data showing GQ populations by type and age-sex-race characteristics may be sufficient –there are many other special surveys of GQ populations designed to meet federal data & programmatic needs –there are problems with the ACS GQ data But what happens if the GQ were not included in the total population universe? Let me reframe the Q: APDU WHO WOULD OBJECT IF SOME OR ALL GQ POPULATIONS WERE REMOVED FROM THE TOTAL BASE POPULATION FOR ACS DATA ON POPULATION CHARACTERISTICS?
Let’s back up for just a moment …places where people live or stay, in a group living arrangement, that are owned or managed by an entity or organization providing housing and/or services for the residents APDU What are “group quarters”?
7 main types of GQ Correctional facilities for adults Juvenile facilities Nursing facilities/skilled nursing facilities Other health care facilities College/university student housing Military quarters Other noninstitutional facilities APDU
3 principal types of GQ comprise most of GQ universe College/university student housing (25%) Nursing facilities/skilled-nursing facilities (23%) Correctional facilities for adults (19%) Full GQ population approximately 8 million persons (~3% of total population) APDU
Background The census enumeration and the separate identification of persons living in irregular housing dates from 1850 GQ: part of the census sample data collection and reporting since 1940 Comprehensiveness of tabulations and quality of the sample data has varied from census to census GQ in ACS monthly sample since 2006 APDU
Why retain GQ data in the ACS? Because the Bureau promised! Since 1970, the long from sample has provided content rich & statistically very strong data for the full U.S. population; important programmatic needs Because the ACS GQ data, in some respects, are much improved over the 2000 Census long form GQ data –e.g., significantly reduced levels of item imputation Because some data users may need these data –some data users may need certain GQ data for programmatic purposes –many data users want the ACS universe to include the total population –we want to hear from those users APDU
Why consider dropping some or all GQ data from the ACS? The need for sample GQ data is not very well documented Not much is going to be published anyway –But what about the PUMS data? There are problems with the data: –Imputation rates still uncomfortably large –The separation of the sample controls from the census enumeration means that controls are estimate-based –Small populations; smaller samples; “zero” GQ problems Some small, difficult-to-survey, GQ populations have already been dropped from the ACS sample Are there appropriate cost-benefit issues? GQ considerably more expensive than HH portion of ACS (GQ: $89 per person interview; HH: $13 per person interview) APDU
We really want to hear from you! You may write to us directly expressing your concerns or opinions –Paul Voss –Krisztina Marton Survey of data users Data user conference It’s important! APDU