Journal of Official Statistics, Vol.25, No.2, 2009. pp. 269–291
The Quasi-multinomial Distribution as a Tool for Disclosure Risk Assessment
Abstract:To model diverse populations, the present article proposes a family of distributions that are closed under recoding and suppression. This family is supported by an empirical fact known as Zipf’s law and suggests an upper bound of disclosure risk. The quasi-multinomial distribution is an instance from this family and studied in particular for the assessment of disclosure risk. Also reported is an example in which the quasi-multinomial distribution fits better than existing models.
Keywords:Compound Poisson, privacy, size indices, superpopulation, uniqueness
Copyright © Statistics Sweden 1996-2018. Open AccessISSN 0282-423XCreated and Maintained by OKS Group