khalchris
Mechanical
- Apr 17, 2013
- 33
Hi All,
I did some experiments and I used full factorial and fractional factorial as shown below:
Test 1: 2 factors, 3 levels full factorial 3x3=9 points
Test 2: 4 factors, 3 levels fractional factorial 18 points (High-Low and 2 for center points)
I repeated each test 3 times in total.
would somebody please help me with some issues I am facing in my models:
1) when I used factorial design using Minitab, I also got an ANOVA table showing the contribution of the terms and their interactions, I was wondering if it is valid to use ANOVA tables to indicate the contribution of the terms and their interaction on the outputs. I am not interested in the whole statistical analysis, I just want to show how each term is influencing my outputs.
2)It seems that test 1 and test 2 might have quadratic terms in them especially in test 2. the curvature error in ANOVA table is significant, would it be ok to leave the model as it is by only showing linear terms and their linear interactions (1-2-3-4 way interactions) or do I need to do other things to check for quadratic terms and fix the model further.
I have all my dataset and results if required.
I did some experiments and I used full factorial and fractional factorial as shown below:
Test 1: 2 factors, 3 levels full factorial 3x3=9 points
Test 2: 4 factors, 3 levels fractional factorial 18 points (High-Low and 2 for center points)
I repeated each test 3 times in total.
would somebody please help me with some issues I am facing in my models:
1) when I used factorial design using Minitab, I also got an ANOVA table showing the contribution of the terms and their interactions, I was wondering if it is valid to use ANOVA tables to indicate the contribution of the terms and their interaction on the outputs. I am not interested in the whole statistical analysis, I just want to show how each term is influencing my outputs.
2)It seems that test 1 and test 2 might have quadratic terms in them especially in test 2. the curvature error in ANOVA table is significant, would it be ok to leave the model as it is by only showing linear terms and their linear interactions (1-2-3-4 way interactions) or do I need to do other things to check for quadratic terms and fix the model further.
I have all my dataset and results if required.
