Question on Taguchi Method

#1
Hi,

I had a question on the dummy treatment technique in the Taguchi method. Suppose I am downgrading a 3 level column into a 2 level column by changing the 3rd level of the column to (say) the first level.

While analyzing the results, when I am trying to calculate the average effect of the first level of that factor, should I take into consideration the results obtained from the trial conditions in which the 3rd level had been changed to the first level?

Or, for the purpose of calculating the effect of the first level, should I only include those trial conditions where the first level had originally been present as such in the array?