Imputation
RTI statisticians have extensive expertise in imputation, which is the substitution of a reasonable value for a missing value.
Methodologies/Techniques
- Tree-based methodologies for development of imputation classes
- Hot-deck imputation, both classical and weighted sequential methods
- Regression imputation
- Predictive mean neighborhoods imputation
- Logical imputation
- Mean value imputation
- Multiple imputation
Capabilities
Our statisticians have comprehensive knowledge of and experience with commonly used software to construct tree-based imputation classes. We have extensive experience with all forms of hot-deck imputation, especially the weighted sequential hot-deck, which was developed by RTI and is implemented in SUDAAN®.
Before any imputation task, we conduct a detailed analysis of the missing data to determine the most effective method of imputation. This analysis includes reviewing the variables for consistency, level, and pattern of missingness and nonresponse bias analysis.
Applications
- Consistent imputation of sets of related variables
- Highly effective multivariate imputation to preserve complex associations between responses when some are missing
- Weighted sequential hot-deck imputation software to guarantee the expectation over repeated imputations is the weighted respondent mean for each imputation class
- Predictive mean neighborhoods proprietary software combines advantages of weighted hot deck and predictive mean matching
- Software for efficiently imputing values for variables with high levels of nonresponse
- Approaches to imputing values for hundreds of variables
- SUDAAN® for analyzing multiply imputed complex survey data