1992 Hispanic Oversample


The 1992 core file includes an oversample of the Hispanic population, sponsored by the National Cancer Institute. The oversample was created by re-contacting Hispanic respondents who participated in the 1991 NHIS survey. Those respondents were assigned to complete a second interview during the same interview week in 1992 as they had been assigned in 1991. The PERWEIGHT value for 1992 assumes that the Hispanic oversample is included in any analysis of the 1992 sample. Members of the Hispanic oversample in the 1992 sample have NHISPID values that begin with 1991.

Implications for Analysis

If users pool 1992 with 1991 they will include members of the 1992 Hispanic oversample in the analytic sample twice, effectively double counting them. Users who wish to include both the 1991 and 1992 samples in a pooled analysis are advised to drop the Hispanic oversample records in 1992 (identifiable by NHISPID values beginning with 1991) and use the 1992 core-only weight (PERWEIGHT92 for person-level analyses or HHWEIGHT92 for household-level analyses) instead of PERWEIGHT and HHWEIGHT, respectively.

One of the most common reasons for pooling the 1991 and 1992 samples is for analyses of the linked NHIS-NDI data. Users are typically advised to use MORTWT for analyses using the linked NHIS-NDI data. The value of MORTWT in 1992 assumes the inclusion of the Hispanic oversample. However, for analyses of pooled NHIS-NDI data that include the 1991 and 1992 samples, users are advised to drop the Hispanic oversample in 1992 and adjust PERWEIGHT92 for linkage ineligibility using the NCHS guidance from Appendix III of Linkage of NCHS Population Health Surveys to Administrative Records From Social Security Administration and Centers for Medicare & Medicaid Services.