Identification of Product Measures by Random Choice of Marginals
Małgorzata Konwerska, Fernando López-Blázquez, Jacek Wesołowski · Communication in Statistics- Theory and Methods · 2007
Three independent random variables are transformed into a bivariate vector by choosing at random one of the variables from two pairs. It appears that such a transformation preserves all information about the parent product measure and the random choice mechanism. Moreover, the original distribution can be explicitly identified. Also identifiability under combinations of random choice with convolution, minimum, and maximum is considered.