Should I use factorial or randomized complete block?

[nathan.haag@villanova.edu]

Hi, I'm doing research and I would ask my professor this, but he's busy.
Basically my experiment is as such

I have 44 urchins in a sea tank. Replicates are grouped together in 4's. Each of these groups were selected by picking urchins of the same basic size. Each of the 4 urchins within that group gets a different treatment based on two factors (spine clipping and living surface) with two levels each (clip/no clip and flat block/pitted block).

I realize that saying factors and levels instantly cries out FACTORIAL!! but the urchins are also in size blocks of 4, aren't they??

I showed RCB to my prof and he didnt say it was wrong, but I dont know if he was paying attention.

any help would be greatly appreciated.

[fed1]

This is RCB design as you suggested.

Be sure that when you analyze the data yo include size as a factor in the ANCOVA model. If you do not include size you will overestimate the sampling variance, and get to conservative results. I think not including stratifying covariates, like size in your case, is top ten analysitc mistake people make.

[nathan.haag@villanova.edu]

Thanks, I'll go with the block design