Structural Inference on Reliability in a Lognormal Model
Abstract
**Please note that the full text is embargoed** ABSTRACT: The theory of structural inference, as developed by Fraser (1968), is based on a group-theoretic approach using invariant Haar measures to
Fisher's fiducial theory. Structural inference theory constructs a unique distribution, conditional on the given sample information only, for the parameters of a measurement model. Based on the structural density for the two-parameter lognormal distribution, the structural density and distribution function for the reliability function are derived. Consequently, expressions for structural point and interval estimates of the reliability function are developed. Approximations for large sample sizes and/or moderately reliable components are also discussed. An example based on lognormal data is given to illustrate the theory.