To create runs for a factorial design in Minitab, you select
Stat>DOE>Create Factorial Design. On the main screen, you select
the Number of Factors (the default is 2) and the click on the
Design button. In the Design screen, select the design you want
from the options listed (Full Factorial, 1/2 Fractional Factorial, etc.)
as well as blocks, center points and reps; for now, you'll want to
choose a full factorial design, 1 rep, 1 block and no center points.
After entering your choices
and returning to the main screen, click Options. It's convenient to
have randomized runs when designing an experiment but inconvenient
when entering data for homework--click on Randomize Run Order to
remove this option. After returning to the main screen, you may want
to click Factors in order to change the factor names (under the
Name column) and levels (under the Low and High columns) to something
that is more user friendly.
Return to the main screen once more and click
OK; your worksheet will have all the run levels labeled and your
experiment will be ready to run (or you'll be able to enter your
responses in a response column--be sure to name the response
column--and conduct an analysis).
To analyze the design, you select Stat>DOE>Analyze Factorial Design.
On the main screen, you select the response. At this point, you
could click OK and obtain effects. The effects are under the column
in the Session window labeled Effect and the overall mean is the first
term in the column labeled Coefficient. I encourage you to select
Graphs and confirm that the box for a normal effects plot has been
checked.
A note on unsaturated models and EMRs:
In addition, you can choose Terms and add or delete model terms (the
default model is the saturated model). The ":" notation for main
effects assigns columns names to effects listed A, B, C, etc. Since
the default column names from using "Create Factorial Design" ARE
A, B, C, etc., the notation seems superfluous at first.
In constructing
EMRs, you often want to select only the significant effects as terms. When you first look at the Terms window, all effects will be listed in the Selected Terms window on the right. For each term you want to exclude, click on the term in the Selected Terms window, then on the left arrow (<) to move it back to the Available Terms window. When you are finished, only the terms you want to use for the EMR calculations should be listed in the Selected Terms window. Then
under Storage, you would select Fits. The Fits would be saved in the
worksheet next to your responses (or the first available empty column);
these values are actually the EMRs for all possible combinations of
the important factor effects.
A note on replications: If you have a replicated design and want to
enter averages as your response (similar to the way our text does),
then you should select only 1 rep (since the averages are not
replicated). If you want to enter the raw data (in order to do
t-tests, obtain the MSPE, etc.), indicate the appropriate number
of replications. When the table is stored in your worksheet, you
will see that the reps are stacked one on top of each other.
E.g., for a three factor three rep design, the (-1,-1,-1) run
will be in rows 1, 9 and 17, the (1,-1,-1) run will be in
rows 2, 10 and 18, etc.