Minimize donor variability in studies

[#1]

Hi, 

In our studies, we always use multiple donors and we know that there's variability from the donors. Can we use fit y by x and choose "donor" for blocking?

Thanks.

[#2 · Accepted solution]

It seems like you want to treat an effect in your model as a 'block' effect...that is donor. Whether or not an effect is treated as a 'block' depends on how the data was collected wrt to the experimental design. If 'donor' was treated as a blocking factor wrt to the experimental design then you can incorporate that effect into the model in a way that is appropriate wrt to how the design was constructed. Here's a JMP On Demand video you may find informative regarding blocking and DOE.

Blocking in DOE with JMP 

And generally speaking I'd stick with the Fit Model platform for your modeling work...not Fit Y by X for inclusion of block effects in a model.