Hi Friendly JMPers,
Baking is always a challenge, so I am hoping someone can help me.
I've got a design (I don't know how to classify it - crossed and nested two way split plot design?). And I am not sure how to load the Changes into Custom Design even after reading "Experiments with ‘Hard-to-Change’ and ‘Very-Hard-to-Change’ Variables.
This is a baking process design with four levels (but two of those will be crossed):
Thaw Method (4 levels) - 0, 1, 2, 24
Oven Temp (3 levels) - 345, 360, 375
Muffin Size (2 levels) - 4.5 and 5.4
Rack Position (3 levels) - Top, Middle, Bottom
Thaw Method and Oven Temp are presumably termed "crossed" - there are 12 combinations (4 thaw methods x 3 oven temps)
Then I have an oven rack and I can fit two trays (1 for each of muffin size) on the top, middle, and bottom section of the rack - 6 combinations per rack (2 muffin sizes x 3 rack positions). The rack will always be "nested" within one of the 12 oven temp x thaw method combinations.
And then I can bake a total of 3 or 4 racks per day.
I am having a hard time understanding how to set up the the muffin size x rack position into the oven temp x thaw method in the Custom Design, not getting how "very hard" and "hard" works in this complicated design. And then I am not sure what to put in for whole plot and subplot (because every example I have seen is much simpler).
Ideally, I would like to bake 12 racks total over 3 to 4 days. Each rack contains 3 levels (top middle bottom) and two trays each (4.5 and 5.4 oz) of muffins. Each rack will have a given thaw method (either 0,1, 2, 24) and be baked at one temp (345, 360, 375). I would like those bakes randomized. And if I need to add a few more racks to allow for error that would be ok.
For visualization purposes, I have drawn out and attached my design.
Any help would be much appreciated.
The crux of your experiment, beyond the factor type and level definitions, is the randomization. The way you have it set up now means that you intend to reset Muffin Size and Rack Position in every run (data table row), reset the Oven Temperature less often, and finally reset the Thaw Method least often.
I think that you should not select the option, "Hard to change facts can vary independently of Very Hard to Change factors." If you do, then you won't get the 'nesting' you mention for each oven run.
Then number of whole plots, sub-plots, and runs determines the degrees of freedom available to estimate the variances for hypothesis tests about the effects. The first two levels have to do with the randomization. You show 6 whole plots. Another way to interpret that number is that you will reset Thaw Method 6 times during the experiment. Similarly, you will reset Oven Temperature 12 times during the experiment. The sub-plots (Oven Temperature) will be nested within the whole plots (Thaw Method).
Thank you so much for the help and explanation Mark.
