An application of canonical expansions to media exposure distributions

Research output: Contribution to journalArticleResearchpeer-review


A model for media exposure probabilities is developed which has the joint probability of exposure proportional to the product of the marginal probabilities. The model is a generalization of Goodhardt & Ehrenberg’s ‘duplication of viewing law’, with the duplication constant computed from a truncated canonical expansion of the joint exposure probability. The proposed model is compared on the basis of estimation accuracy and computation speed with an accurate and quick ‘approximate’ log-linear model (as noted previously) and the popular Metheringham beta-binomial model. Our model is shown to be more accurate than the approximate log-linear model and four times faster. In addition, it is much more accurate than Metheringham’s model.

Original languageEnglish
Pages (from-to)183-192
Number of pages10
JournalJournal of Applied Statistics
Issue number2
Publication statusPublished - 1 Jan 1992
Externally publishedYes

Cite this