completely randomized block design: How to include the blocks in JMP?

[#1]

Hello,

Let's say I got an experimental wheat field where I want to investigate how different treatments influence my YIELD. The treatments are:

1) PLOUGH

A: ploughing at 20cm

B: ploughing at 40cm

C: no ploughing

2) FERTILIZATION

1: 100 Kg/ha

0: no fertilization

Unfortunately my field is inclined, therefore I need to add blocks (I, II, III)

Block I	A1	B1	C1	A0	B0	C0
Block II	B1	A0	A1	B0	C0	C1
Block III	C1	B0	C0	A1	A0	B1

If I go to "Fit Model" and put my YELD as Y and use MACRO -> Fullfactorial with PLOUGH and FERTILIZATION I think I can get results just as a completely randomised design. So my question is: How to include the blocks in Fit Model?

Maybe I should use "Fit Y by X" (there is the block option). But then how to include the full-factorial?

Thank you,

Matteo

[#2 · Accepted solution]

I would recommend using Custom Design.

I quickly just chose one three level categorical for Plough, one two-level categorical for Fertilizer and then chose block of size 6

12698_Screen Shot 2016-09-09 at 3.45.24 PM.png

12699_Screen Shot 2016-09-09 at 3.47.09 PM.png

Simulating that experiment gives this output. (I had to reset factor grid to show the block in the Prediction Profiler.

12700_Screen Shot 2016-09-09 at 3.50.25 PM.png

Hope this helps,

PS Perhaps you might want to investigate utilizing a Split-plot design

Lou

[#3]

Thank you very much Lou. That's just an excellent explantation.

[#4]

And what if I want to use repeated measure ove time within the same experiment? For example if I want to monitor CO2 efflux every month? Does somedoby have some idea?