Statistical Methods for Confirming That an Intakes Has Accurred

DOE-STD-1121-98

6.2.1 Decisions Based on Individual Monitoring Data

Examples of actions taken following acquisition of a result from an individual monitoring program

are shown in Figure 3. This approach is useful to consider, with individual sites determining the values of

their respective reference levels.

6.3 STATISTICAL METHODS FOR CONFIRMING THAT AN INTAKE HAS

OCCURRED

Beyond the methods described in the IDG, at least two other statistical methods exist for confirming

that an intake has occurred. The first is to simply pool the n bioassay results statistically to achieve

the

improvement in the decision level (Hickey et al. 1993).

The second is to employ Bayesian statistical inference (Miller et al. 1993, 1995), based on Bayes'

theorem (Lindley 1972, 1980, 1985; Martz and Waller 1982; Calvin 1989; Press 1989). The Bayesian

formalism is attractive because it incorporates prior knowledge in addition to the results of a given

measurement, and it results in a distribution of likely outcomes rather than merely a point estimate with

an uncertainty. However, the method has been criticized as being too subjective. At present, the DOE

and this Technical Standard have taken no official position on the use of the Bayesian method. The

appropriateness of Bayesian methods must be decided on a case-by-case basis.