|
><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><>
If you reply to this message, it will go to all FISH-SCI members.
><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><>
Hi all,
I admit to most of this discussion being well over my head. The one point
I am sure of was brought up by Trevor. In order to keep clear of
pseudoreplication (Hurlbert 1984) the basic sampling unit needs to be the
trawl catch, with individual shell heights as replicates. It seems (maybe
too simplistically) that ANOVA would be appropriate for the analysis, since
the variability among shell heights would be used in the analysis, rather
than lost in averaging.
Howard Teas
Anchorage, AK
>
>Or would it be necessary to treat the abundance of each size-class as a
>separate independent variable and then use some GLM analogue of MANOVA?
>Should the model use individual shell heights as replicates, nested
>within samples (maybe one sample per landing)? Or would one have to
>reduce each frequency to a mean and analyze just the means? It would
>seem that any of those latter options would lose a lot of the
>information in the original data.
>
>
>Trevor Kenchington
>
>
>--
>Trevor J. Kenchington PhD [log in to unmask]
>Gadus Associates, Office(902) 889-9250
>R.R.#1, Musquodoboit Harbour, Fax (902) 889-9251
>Nova Scotia B0J 2L0, CANADA Home (902) 889-3555
>
> Science Serving the Fisheries
> http://home.istar.ca/~gadus
>
>><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><>
> To leave the Fish-Sci list, Send blank message to:
> mailto:[log in to unmask]
> For information send INFO FISH-SCI to [log in to unmask]
>><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><>
><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><>
To leave the Fish-Sci list, Send blank message to:
mailto:[log in to unmask]
For information send INFO FISH-SCI to [log in to unmask]
><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><> ><>
|
