Combining Methods to Create Synthetic Microdata: Quantile Regression, Hot Deck, and Rank Swapping

Government agencies must simultaneously disseminate useful microdata and maintain confidentiality of individual records. Releasing synthetic data is one approach. We propose to create synthetic data using a combination of quantile regression, hot deck imputation, and rank swapping. The result is a releasable data set containing original values for a few key variables, synthetic quantile regression predictions for several variables, and imputed and perturbed values for remaining variables. The procedure should provide quality data to the user and simultaneously protect the confidentiality of respondents. The methods is illustrated by creating synthetic data for a Public Use Microdata Set from the American Community Survey.