Abstract:
|
The vast majority of articles about the theory and practice of formal privacy for statistical disclosure limitation assume either that the records used in the analysis are the entire frame or that sampling occurred with fixed, exogenous probabilities. Adapting the methods to complex sample surveys and information products that combine complex surveys with data from the entire frame, possibly linked to data not acquired through the survey, is a challenging research problem. This talk lays out a conceptual framework for this research program and reports on the state of current knowledge.
|