Imputed Data in SLAITS Microdata Sets

This page contains links to SLAITS microdata sets that include data that have undergone imputation (with related documentation). Imputation is a statistical technique that attempts to address missing data in sample survey datasets through simulation. Data can be missing for a number of reasons: the respondent either did not know the answer to question(s); chose to skip question(s); refused to answer question(s); or question(s) were erroneously not asked. A high level of missing data limits the ability of analysts to draw conclusions from the survey.

To derive the imputed values, an imputation algorithm or model is developed to predict data for the missing variable(s) by taking the observed values into account. In single imputation modeling, the model is run once to predict the missing datum (data). In multiple imputation, the model is run more than once (typically five times) to predict the missing datum (data) and permit more accurate variance estimation.

The highlighted links below connect to the imputed microdata, a report describing the creation and use of the imputed data, and (in some cases) sample SAS programs.

2001 National Survey of Children with Special Health Care Needs

• An indicator variable was developed using single imputation to identify income status below 200% of the Federal poverty level for uninsured children. This variable (POV200_I) is included on the insurance data file (October 2003)
  • Imputation of Low-Income Status Report [PDF - 220 KB]
  • Data Sets
• Detailed income values relative to the Federal poverty level were developed using multiple imputation (June 2007)
  • Multiple Imputation of Missing Household Poverty Values Report [PDF - 320 KB]
  • Datasets and SAS Programs

2003 National Survey of Children's Health

• Detailed income values relative to the Federal poverty level were developed using multiple imputation (June 2007)
  • Multiple Imputation of Missing Household Poverty Values Methodology Report [PDF - 320 KB]
  • Datasets and SAS Programs

2005 - 2006 National Survey of Children with Special Health Care Needs

• Detailed income values relative to the Federal poverty level will be developed using multiple imputation and will be released with the microdata
  • Multiple Imputation of Missing Household Poverty Values Methodology Report
    
  • Datasets and SAS Programs

2007 National Survey of Children's Health

• Detailed income values relative to the Federal poverty level were developed using multiple imputation
  • Multiple Imputation of Missing Household Poverty Values Methodology Report - See Appendix XII [PDF - 640 KB]
  • Datasets and SAS Programs

2009 - 2010 National Survey of Children with Special Health Care Needs

• Values for numerous variables (race, ethnicity, parental education, primary household language, detailed household income values relative to the Federal poverty level, and total number of adults in the household) were developed using multiple imputation.
  • Frequently Asked Questions about Multiple Imputation of Missing Values [PDF - 178 KB]
  • Datasets and SAS Programs [ZIP - 14 MB]

2011 Survey of Pathways to Diagnosis and Services

• Values for numerous variables (race, ethnicity, parental education, primary household language, detailed household income values relative to the Federal poverty level, and total number of adults in the household) were developed for the 2009-2010 National Survey of CSHCN using multiple imputation. The file below is limited to those children with completed Pathways interviews. Users should refer to documentation for the 2009-2010 survey for more information about the development and use of these imputed values.
  • Datasets and SAS Programs [ZIP - 1.2 MB]