Problem in Minitab - ERROR * Factor Voltage is highly correlated with other terms

Om Redha

Registered
* NOTE * Design is not orthogonal.
* ERROR * Factor Voltage is highly correlated with other terms. No calculations were done.
 

Miner

Forum Moderator
Leader
Admin
I understand the error, but would have to see your design to pinpoint what is wrong. If you can attach your Minitab file in version 18, I can trouble shoot it. If not, Copy/Paste your worksheet into Excel and attach.
 

Om Redha

Registered
I understand the error, but would have to see your design to pinpoint what is wrong. If you can attach your Minitab file in version 18, I can trouble shoot it. If not, Copy/Paste your worksheet into Excel and attach.
thanks for reply me....I will attach my Minitab file in verision 17 , Ok???
 

Attachments

  • nnn.MPJ
    76 KB · Views: 5

Miner

Forum Moderator
Leader
Admin
The problem is in the design of the experiment, not in the analysis. Factors A and B are 100% aliased (confounded) with each other. This means that you cannot determine which of these factors may be responsible for an effect. And, as the error stated the design is not an orthogonal design, which means there is some degree of aliasing between factors.

Since you are using more than two experimental levels (3 & 5), you cannot use a 2^k factorial design, you must use a general full factorial experiment, which will require 135 experimental runs. I recommend a 2^4 design (attached). With center points this will require 19 runs. This design is orthogonal, and the addition of center points will allow you to see whether there may be curvature in the response. If there is curvature, you will be able to add axial points and expand this into a response surface design.
 

Attachments

  • Problem in Minitab  - ERROR * Factor Voltage is highly correlated with other terms
    nnn.png
    9.1 KB · Views: 3

Om Redha

Registered
The problem is in the design of the experiment, not in the analysis. Factors A and B are 100% aliased (confounded) with each other. This means that you cannot determine which of these factors may be responsible for an effect. And, as the error stated the design is not an orthogonal design, which means there is some degree of aliasing between factors.

Since you are using more than two experimental levels (3 & 5), you cannot use a 2^k factorial design, you must use a general full factorial experiment, which will require 135 experimental runs. I recommend a 2^4 design (attached). With center points this will require 19 runs. This design is orthogonal, and the addition of center points will allow you to see whether there may be curvature in the response. If there is curvature, you will be able to add axial points and expand this into a response surface design.
The problem is in the design of the experiment, not in the analysis. Factors A and B are 100% aliased (confounded) with each other. This means that you cannot determine which of these factors may be responsible for an effect. And, as the error stated the design is not an orthogonal design, which means there is some degree of aliasing between factors.

Since you are using more than two experimental levels (3 & 5), you cannot use a 2^k factorial design, you must use a general full factorial experiment, which will require 135 experimental runs. I recommend a 2^4 design (attached). With center points this will require 19 runs. This design is orthogonal, and the addition of center points will allow you to see whether there may be curvature in the response. If there is curvature, you will be able to add axial points and expand this into a response surface design.
please, can you complete the solution and find analysis of variables (anova) because Iam student and I do not have experience in dealing with Minitab, please....
 

Miner

Forum Moderator
Leader
Admin
It's not that simple. Since the design structure is seriously flawed, the best that you will be able to do is perform a regression analysis instead of an ANOVA. Even then, too many of the factors are highly correlated with each other to trust any results. This experiment should be designed correctly and run again.

Problem in Minitab  - ERROR * Factor Voltage is highly correlated with other terms
 

Om Redha

Registered
It's not that simple. Since the design structure is seriously flawed, the best that you will be able to do is perform a regression analysis instead of an ANOVA. Even then, too many of the factors are highly correlated with each other to trust any results. This experiment should be designed correctly and run again.

View attachment 27939
Well, can you do that? Can you complete the solution as you see fit and correct? I would be grateful for your efforts with me. There is no one to help me, even the supervisor does not know anything about Minitab. I am waiting for your reply
 

Miner

Forum Moderator
Leader
Admin
I assume that A D are the experimental factors? What are GRG, CO, Th, H and R? Five different responses?

Note: The results will be pretty much meaningless.
 

Om Redha

Registered
I assume that A D are the experimental factors? What are GRG, CO, Th, H and R? Five different responses?

Note: The results will be pretty much meaningless.
what do you mean ((The results will be pretty much meaningless))
A D are inputs (factors) and GRG, CO, TH, H and R are output(responses). what do you mean ((The results will be pretty much meaningless)). Do you mean that the results I sent you can't be solved by Mintab???? what I do??? I do not know???
 
Last edited:

Miner

Forum Moderator
Leader
Admin
"To call in the statistician after the experiment is done may be no more than asking him to perform a post-mortem examination: he may be able to say what the experiment died of." Ronald Fisher
 
Top Bottom