Year
1983
Abstract
This paper describes an extension of the AMASS methodology which was previously presented at the 1981 INMM annual meeting. The main thrust of the current effort is to develop procedures and a computer program for estimating the variance of an Inventory Difference when many sources of variability, other than measurement error, are admitted in the model. Procedures also are included for the estimation of the variances associated with measurement error estimates and their effect on the estimated limit of error of the inventory difference (LEID). The algorithm for the LEID measurement component uncertainty involves the propagated component measurement variance estimates as well as their associated degrees of freedom. The methodology and supporting computer software is referred to as the augmented Automated Material Accounting Statistics System (AMASS). Specifically, AMASS accommodates five source effects. These are: (1) measurement errors (2) known but unmeasured effects (3) measurement adjustment effects (4) unmeasured process hold-up effects (5) residual process variation A major result of this effort is a procedure for determining the effect of bias correction on LEID, properly taking into account all the covariances that exist. This paper briefly describes the basic models that are assumed; some of the estimation procedures consistent with the model; data requirements, emphasizing availability and other practical considerations; discusses implications for bias corrections; and concludes by briefly describing the supporting computer program.