The Quality of Information: Errors and Error Propagation

Abstract

In forestry, information is frequently concerned with a quantitative measurement of variables, characterised by a probability distribution. Although point estimates are most frequently used, such estimates do not indicate their precision. In consequence, interval estimates are required in order to make statements about the probability that an interval will indeed contain the unknown population parameter. Normally the assumption is that the sample statistics follow the Gauss distribution, but this assumption does not always hold true. Minimum-variance estimators are most frequently used, but maximum likelihood estimators, although not necessarily producing unbiased estimates; are sometimes preferred to minimum-variance estimators.