topic Re: Can one factor overpower another when analyzing full factorial RCBD? in Statistical Procedures

Can one factor overpower another when analyzing full factorial RCBD?

dsuchoff — Tue, 08 Aug 2017 15:03:23 GMT

My apologies for the somewhat convoluted title. Let me explain:

We are conducting a greenhouse study in which we are looking at plant genotype (factor A) and irrigation (factor B). Plants are grown in pots to which we randomly assigned both genotype and irrigation. The study is a full factorial RCBD with five blocks. In analyzing the data, I notice that the effect of irrigation is very strong, and I wonder if the variability associated with it overshadows potential interactions? Here is an example of analyzed data for stomatal conductance in which we took two subsamples per plant.

SAS Output

SAS Output
Type III Tests of Fixed Effects

Effect Num DF Den DF F Value Pr > F
Genotype 2 50 3.06 0.0559

Irrigation 1 50 1168.62 <.0001
Interaction 2 50 1.97 0.1506

Any input would be much appreciated!

David

Re: Can one factor overpower another when analyzing full factorial RCBD?

PaigeMiller — Tue, 08 Aug 2017 15:09:03 GMT

If you truly have a full factorial, then the answer to your question is NO. Each factor and interaction explains a certain amount of variability of the response, independently of the others, and so the big value you get is because the effect of irrigation is huge, and the effect of genotype and interaction is quite small.

Re: Can one factor overpower another when analyzing full factorial RCBD?

dsuchoff — Tue, 08 Aug 2017 15:12:02 GMT

Thank you, Paige. That's what I was thinking, but wanted more expert advice.
Thanks again.

Re: Can one factor overpower another when analyzing full factorial RCBD?

Rick_SAS — Tue, 08 Aug 2017 15:53:00 GMT

That's the beauty of designed experiments: the factors are not correlated