Re: Signor-Lipps points

[Stefan Bengtson <Stefan.Bengtson@xxxxxxxxx>]

Chris: In anticipation of Jere's awakening on the currently dark side of
the globe, let me try:

As long as what you call 'random effects' are just that, i.e.  random
(i.e. they do not introduce any bias in the data), then they are part of
the Signor-Lipps Effect in the same way as 'randomly' incomplete sampling
is. The key word is 'random': if your first season's sampling did not at
all include, or underrepresented, the top interval, then I think that
reference to Signor-Lipps is misleading. In the same way, if the top unit
for one or another reason does not preserve fossils, Signor-Lipps does not
apply at this level of analysis. Only if the factors that prevent 100%
discovery are largely random does the concept of Signor-Lipps make any
sense to me. Signor-Lipps is not about bias, it's about blur resulting
from incomplete data. 

This also means that for a the same data set, Signor-Lipps can apply on a
large scale, whereas on a more detailed scale it doesn't. A barren bed
could be part of a largely 'random' pattern when observed over a hundred
million years, but if you go straight to the bed and bash it to pieces,
you can't blame Phil and Jere when you don't find any fossils. 

Doug: We agree on the value of confidence limits for null hypotheses. 
What I'm saying is only that if you have a truly barren interval cutting
of the ranges of a number of fossils, no statistical juggling in the world
can tell you whether it's barren because nothing lived there or because
nothing is preserved there. What the statistics can possibly tell you is
whether the absences are real, i.e.  it just looks barren because you
haven't sampled enough. But if the absences are real, their cause(s) must
be analyzed by other means: they have nothing to do with the S&L Effect. 

(N.B. 'Real' in the above paragraph does not refer to Erwinian reality,
i.e. having an age no younger than Triassic.)

^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
Stefan Bengtson
Institute of Earth Sciences
(Historical Geology & Palaeontology)
Norbyvagen 22
S-752 36 Uppsala
Sweden

tel. +46-18 18 27 62 (work)
     +46-18 54 99 06 (home)
     +46-18 18 27 49 (fax)