I'm interested in an archaeology project which includes 50 'trial plots',
each 1m square, over an irregularly shaped area of perhaps 20 hectares.
The current design has these trial plots placed regularly along two straight
lines.
This design seems to be widely used among archaeologists. I suspect it is
sub-optimal in terms of the information it would provide - but can anybody
please advise me how to put some formal meat on this idea?
I suppose I am thinking of a small-dimensional response surface: I recall
from work of Fedorov (?) that the optimal designs are related somehow to
vertices of polygons. But of course the dimensionality is unknown - and will
be different for different variables of interest.
So would a random allocation of trial plots be most sensible? How much
'better' would this be than the current design? Does the plot size and shape
(1m square) affect things - indeed, what is the 'optimal' size and shape?
Thanks for any advice offered.
JOHN BIBBY
You may leave the list at any time by sending the command
SIGNOFF allstat
to [log in to unmask], leaving the subject line blank.
|