Hi Jacqui,
I think a multi-celled Chi-square would be the most straightforward for
this if the sample sizes per cell (predicted values ideally >5) merit
it. You could focus on a selection of elements rather than using
everything. I'd be inclined to enter NISP rather than MNE data, but if
both show the same pattern the distinction is of course moot despite the
usual pedantic arguments. If it hasn't been mentioned already these
three books are very useful in addition to Shennan (which I use for my
undergrad stats course):
Baxter, M. 2003. Statistics in Archaeology. London: Arnold.
Drennan, R. D. 1996. Statistics for Archaeologists. New York: Plenum.
Orton, C. 2000. Sampling in Archaeology. Cambridge: Cambridge University
Press.
All the best,
James
www.fishlab.org
-----Original Message-----
From: Analysis of animal remains from archaeological sites
[mailto:[log in to unmask]] On Behalf Of Jacqui Mulville
Sent: 24 November 2004 10:42
To: [log in to unmask]
Subject: [ZOOARCH] statistics and sides
I have an (another) assemblage that seems to be biased to one side of a
species. I have looked back at Windmill Hill (Edwards and Horne 1997),
the only other example I know of sidedness but the claims on an
over-abundance of sides of a couple of particular elements were not
supported by statistical analysis.
I have a range of elements, which I can use to produce a minimum number
for each side - what springs to my mind is a chi-squared test expecting
equal numbers of each element - but can I group all the different
elements together to test them?
So any suggestions please as to how to analyse many elements from a
species to look for statistical significance of the over representation
of a side? Previously I have only had one element (i.e. femur) to query
which seems easier to me.
All/any help welcome - has anyone written a book on statistics for
zooarchaeologists yet? It would be jolly useful.
jacqui
|