Back to Annual Meeting Page		American Public Health Association 133rd Annual Meeting & Exposition December 10-14, 2005 Philadelphia, PA

Abstract #117330

Deborah D. Ingram, PhD, National Center for Health Statistics, 3311 Toledo Rd., MS6208, Hyattsville, MD 20782, (301) 458-4733, DDIngram@cdc.gov

Census 2000 allowed respondents to report more than one race,; most birth and death certificates only permit the reporting of one race. To permit the calculation of race-specific birth and death rates, Census 2000 multiple-race groups are being “bridged” to single-race categories. Logistic and multi-logit models were fit to pooled 1997-2000 National Health Interview Survey data to predict preferred single-race category for multiple-race respondents. The resulting parameter estimates have been used to obtain county-age-sex-Hispanic origin –specific probabilities of selecting each of the possible single-race categories for the multiple-race groups and these probabilities have been applied to Census 2000 population files to obtain bridged-race population counts. The Census Quality Survey (CQS) used a split panel design to obtain race data using both “Mark 1 or more races” and “Mark 1 race” questions. The resulting multiple-race sample is considerably larger than the NHIS multiple-race sample. The NHIS bridging models have been replicated using the CQS. Now the impact on the model parameters and on resulting population counts of using additional Census 2000 contextual variables (such as segregation indexes) is examined.

Learning Objectives: