Abstract
SUMMARY Surveys of behaviour often question respondents about events in the recent past, with a view to making inferences about underlying parameters describing longer-term temporal aspects of the phenomena under investigation, or estimating behaviour over past or future time periods of varying lengths. To address these issues in a particular context, this paper develops the class of binomial mixtures arising from transformation of the binomial parameter p as 1 - exp(-A) where A is treated as a random variable. It is shown that this formulation provides closed forms for the marginal probabilities in the compound distribution if the Laplace transform of the mixing distribution may be written in closed form. Although illustrated by examples arising from particular consumption processes, the class is of general application in empirical modelling of binomial mixtures of unknown genesis.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call