Threshold (LNT vs. Hormesis)

[Shonka Research Associates Inc <sra@crl.com>]

If you measure or calculate a distribution of effects based on some
model, the frequency versus severity distribution goes from large,
easily observed effects to smaller ones that may or may not lie in
the statistical "noise".  The "noise" threshold is not an "effect"
threshold but simply a threshold  where the observer's ability to see
the effect disappears because of the noise.  Beneath that, if the
model for the distribution fits the data, it would be accepted by
most that the effect still occurs at the expected frequency, but we
just can't see it.  This conceptual argument is only slightly
applicable to LNT versus hormesis since the debate is on the shape of
the distribution.

Joe Shonka
sra@crl.com