Some aspects of intrinsic inference (Q1074255)

The author's belief that the intrinsic dependence among the elements of a population (a dependence that arises from the fact that elements of such a population share a number of common characteristics) should be considered, as well as the sampling variability, in statistical procedures, leads him to what is here termed intrinsic inference. In this paper the latter concept is investigated; for instance, it is found that with independent sampling, the sample probability function is exchangeable no matter what the intrinsic dependence might be. Differences and similarities between models for intrinsic inference and those for classical, Bayesian and random processes inference are explored. Finally it is suggested that predictive intrinsic inference might be of value in the analysis of a certain learning process.