SREL Reprint #2872
Uncertainty and power at low levels of incurred radiation dose
M Wilson1 and D Jackson2
1Ecological
Statistics, University of Georgia, Savannah River Ecology Lab,
PO Drawer E, Aiken, SC 9802, USA
2Enviros, 61 The Shore, Leith, Edinburgh EH6 6RA, UK
Abstract:
It is common practice when calculating dose to exposed populations
to average the variables that go into the dose calculation (e.g. environmental
concentrations, air kerma, consumption rates, occupancy rates). This approach
is simple and can be useful where data are obtained over different periods
(weekly, monthly, quarterly), where samples may be bulked for some analyses
but not others and where gaps in the data are present. However, such an
approach does not yield information on the degree of uncertainty around
the average dose calculated. An alternative approach is to estimate the
dose to each individual and to obtain an average from this data set, which
can then also be used to derive a measure of uncertainty around the central
dose estimate.
In this study, we demonstrate the variability in dose estimates using
a hypothetical data set and consider the implications for sample size
to achieve fixed confidence or resolving power. We recommend calculating
the dose to every individual sampled, in order both to obtain the average
dose and to estimate its variability. We argue that it is best practice
to obtain information as complete as possible from the available sample
of individuals.
SREL Reprint #2872
Wilson, M. and D. Jackson. 2005. Uncertainty and power at low levels of incurred radiation dose. Journal of Radiological Protection 25:51-66.