You should know if samples have a gaussian distribution or not. If you have
to compare groups with different standard deviation you should use a non
parametric test, if SD are not different you could use a parametric test
Carla
Andrew Hall ha scritto nel messaggio
<7ga64j$d7t$1 at nclient5-gui.server.virgin.net>...
>Anyone know what the correct correlation test is for a 42 sample
>comparison. The samples are random. Normally I would use a non-parametric
>test such as spearmans but I have been told for a group this size (n=42)
>that Pearsons can be used?
>>
