DOE Question: 3 factors with 2 levels and 1 factor with 3 levels

B

Blacksnow

Here is a question in ASQ CSSBB CD bank:
***
An experiment has 3 factors with 2 levels and 1 factor with 3 levels. What is the minimum number of trials necessary if all interactions are ignored?
***
Anyone know how to determine the answer? I cannot find any idea though i have read the whole CSSBB primer...:confused:
 

reynald

Quite Involved in Discussions
Here is a question in ASQ CSSBB CD bank:
***
An experiment has 3 factors with 2 levels and 1 factor with 3 levels. What is the minimum number of trials necessary if all interactions are ignored?
***
Anyone know how to determine the answer? I cannot find any idea though i have read the whole CSSBB primer...:confused:

----My take----
This question has two parts.
First part is on the design:
The design is 2 x 2 x 2 x 3 which means for a single replicate run you would need 2*2*2*3=24 runs. Also since this is not a 2^k design you can't do fractional factorials so doing less than 24 runs is not possible.

Second part is the resolution:
Resolution is typically a concept that means how are your able to distinguish the main effects from the interaction effects. Basically resolution is highest for the full factorial design but is also somewhat related to the number of replicates. For a single replicate run of a full factorial (such as 2*2*2*3=24) it is not possible to estimate the error/SSE term. However it is said that "all interactions are ignored", in this case 24 runs is enough if you will sacrifice knowing the interaction effects and use it instead to estimate the error term SSE.

Bottom line I would answer = 24.
 
Top Bottom