what test to use?

pajo

New Member
#1
I have a data set consiting of 50 vegetation plots. I have five possible categories for vegetation cover. Each category can range form 0 to 100% with a total coverage for each plot equalling 100%. I have three years of data, that is each plot has been sampled 3 times. How do I determine if there is a change or trend in vegetation cover?

Second question how do I determine if this study has enough power to detect a change?

Thanks
 

bugman

Super Moderator
#2
It sounds to me like you need to consider a repeat mesures block ANOVA. With plot being the block and veg type nested within the blocks.

Commonly we use a braun-blanquet scale as the response - a score corresponding to the frequncy of occurence or abundance (by ground cover). This would require a non-parametric version of the above.

Do you have pilot study data? Most stat packages have a power test based on SD, marging of error and current sample size. I use SYSTAT for this.

Phil
 

pajo

New Member
#3
Thanks Phil,

that get's me going in the right direction, I was thinking repeated measures but it was the nesting part that had me confused