A problem statistical o±ces and research institutes are faced with by releasing micro-data is the preservation of confidentiality. Traditional methods to avoid disclosure often destroy the structure of data, i.e., information loss is potentially high. In this paper I discuss an alternative technique of creating scientific-use-files, which reproduce the characteristics of the original data quite well. It is based on Fienberg’s (1997 and 1994) [5], [6] idea to estimate and resample from the empirical multivariate cumulative distribution function of the data in order to get synthetic data. The procedure creates datasets - the resample - which have the same characteristics as the original survey data. In this paper I present some applications of this method with (a) simulated data and (b) innovation survey data, the Mannheim Innovation Panel (MIP), and compare resampling with a common method of disclosure control, i.e. disturbance with multiplicative error, concerning confidentiality on the one hand and the appropriateness of the disturbed data for different kinds of analyses on the other. The results show that univariate distributions can be better reproduced by unweighted resampling. Parameter estimates can be reproduced quite well if (a) the resampling procedure implements the correlation structure of the original data as a scale and (b) the data is multiplicative perturbed and a correction term is used. On average, anonymized data with multiplicative perturbed values better protect against re–identification as the various resampling methods used.


resampling, multiplicative data perturbation, Monte Carlo studies, business survey data