Experimental Design

[Hawnk]

Hello all,

 

I am looking into designing an experiment and am considerable stumped/ confused. I have limited knowledge on statistics, consisting of one undergrad and one graduate course, not covering DOE. Basically I am doing my experiment to investigate 10 factors on a particular process and its corresponding output. The complicating issue in this case is the fact that the factors have different levels (two, three, or four). As well, complicating things greater, three of the factors are nested within one factor. This issue is slightly hard to explain so I'll explain it differently:

 

a,b,c are nested within x; 'a' under x1, and 'b', 'c' under x2.

 

x = x1,x2

 

a=a1,a2,a3

 

b=b1,b2,b3

 

c=c1,c2

 

So if x1 is chosen then 'a' can be three levels.

 

if x2 is chosen 'b' can be three levels, and 'c' can be two levels.

 

I have looked into different types of experimental design such as fractional factorial, taguchi method etc. and this experiment seems too complex to apply those methods. Any recommendations on ways I should approach this? I am unsure if I am explaining this coherently, so if any further clarification is required let me know.

 

Thanks!

[LaurieAG]

Hi Hawnk,

 

You can probably do it in MS Excel with 3 lookup tables or worksheets depending on the complexity. That way you can set up your data in consistent rows and columns on one worksheet for statistical analysis or copy and past to other applications for the analysis. The MS Excel help files have examples of lookup tables and worksheets.