National Health Interview Survey
2012 Imputed Family Income/Personal Earnings Files
The missing data on family income and personal earnings in the 2012 NHIS were imputed using multiple-imputation methodology. Five ASCII data sets containing imputed values for the 2012 survey year are included in the compressed data file (INCMIMP.ZIP), which can be downloaded via the Datasets link below. For analyses involving other variables in addition to family income or personal earnings, each set of imputed values can be merged with other data from the 2012 NHIS to create a single completed data set.
To increase the usability of NHIS income data, the 2012 NHIS imputed income files contain income dollar amount point estimates rather than income dollar intervals. Because respondent confidentiality must be balanced against providing more detailed information, the variables containing the dollar amounts for personal earnings and family income have been top-coded to the 95th percentile of the appropriate distribution. The 95th percentile was calculated separately for each of the 5 imputed family income/personal earnings datasets and then a weighted average of the 5 individual 95th percentile amounts was calculated. The weighted average was rounded to the nearest $1,000 and this weighted average was used to top-code all 5 supplemental imputed personal earnings and family income datasets. The same procedure was used for family income and personal earnings dollar amounts. For all observations which were top-coded, the family income or personal earnings dollar amounts were replaced with the top-coded value. Also, since the 1997 NHIS, poverty ratio intervals have been provided on the NHIS public-use data files. The poverty ratio is a ratio of the family’s income to the appropriate Federal poverty threshold. In the 2012 NHIS imputed income files, the Federal poverty ratio has been calculated using top-coded family income and the final calculated poverty ratio value is truncated to 3 decimal places. Note that all top-coded point estimates contained in the 2012 NHIS Imputed Family Income/Personal Earnings files are analogous to the top-coded point estimates contained in the 1997-2008 Supplemental Imputed Family Income/Personal Earnings files and the 2009-2011 Imputed Family Income/Personal Earnings file. No analogue of the 1997-2008 Imputed Family Income/Personal Earnings files (with categorical income data) was produced in 2009-2012.
Multiple imputation is a technique that allows analysts to incorporate the extra variability due to imputation into their analyses. This is accomplished by analyzing each of the five completed data sets separately using methods and software that are appropriate for survey data, and then combining the estimates and standard errors using the combining rules described in Section 2.2 and Appendix A of the document available via the Technical Documentation link below. The extra variability due to imputation cannot be incorporated by simply analyzing a single completed data set as if the imputed values were true values. Moreover, analysts should not create a single completed data set using the average of the five sets of imputed values. Examples of correct data analyses using SAS-callable SUDAAN and SAS-callable IVEware are provided in Section 4 of the document available via the Technical Documentation link below; the document also provides information on the procedures used to create the imputations.
The Dataset Documentation link below opens to a document containing both the file layout description and the frequency counts (in the last page) of the variables in the data sets containing imputed values for the 2012 survey year. Users interested in data for several years should note that to date, multiple imputation has been carried out for the 1997-2012 NHIS, and that the file layout description is identical for years 1997-2003. Since 2004, there have been several changes to the imputed income file layout:
- Beginning with 2004, the person number variable changed to FPX which is unique within each family.
- For the years 2007-2008, the variables INCGRP_F, INCGRP_I, RAT_CATF, and ERNYRG_I changed to INCGRPF2, INCGRPI2, RATCATF2, and ERNYRGI2 respectively due to questionnaire and response category changes.
- Beginning with 2009, the variables INCGRPF2, INCGRPI2, RATCATF2, RAT_CATI, and ERNYRGI2 were removed and the variables FAMINCF2, TCINCM_F, FAMINCI2, POVRATI2, TCEARN_F, and ERNYR_I2 were added, replacing family income and personal earnings dollar amount intervals with dollar amount point estimates.
- Beginning with 2010, the variable POVRATI2 changed to POVRATI3, adding a third decimal place to the poverty ratio estimate.