QUB |
Archaeology and Palaeoecology |
The 14Chrono Centre
Manual for psimpoll and pscomb
Menu
Mc.6 produces a shuffle of the samples in the
dataset being analysed. The shuffling of samples
is effective only for zonation, and not for other analyses carried out
in the same run of psimpoll, nor for the main dataset.
The purpose of doing
it is to establish what the background change in variance is between
consecutive splits. You should find that the variance change is more or less
constant, and similar to the change between splits in the unshuffled
dataset when the number of splits is large. It can thus be used to help
determine how many splits are worth considering, and where the splitting is
working on variation that is really noise. Results are written to the
data analysis file, but not to the ZONE file.
Shuffling may be used for all types of zonation in
menu Mc.2, but I am not
convinced of its usefulness for CONISS.
The shuffle is controlled by random numbers picked using a minimal
standard random number generator
(menu N3.1), and uses the same initial
seed as would be used in obtaining random numbers for the purpose of
estimating confidence intervals (menu N4).
Return to menu Mc.
Back to contents page
Copyright © 1995-2007 K.D. Bennett
Archaeology and Palaeoecology | 42 Fitzwilliam St | Belfast BT9 6AX | Northern Ireland | tel +44 28 90 97 5136
Archaeology and Palaeoecology | The 14Chrono Centre | URL http://www.qub.ac.uk/arcpal/ |
WebMaster