Share your ideas for the JMP Scripting Unsession at Discovery Summit by September 17th. We hope to see you there!
Choose Language Hide Translation Bar
Highlighted
PS_Ato
Level III

May manipulation of centre point of DSD DOE (asymmetric nature of factor) influence the results?

We designed a Definitive Screening Design with five factors and three blocks and three centre points.

The range of one factor is assymmetric (With the test we would not produce any results at low concentrations of E (e.g. 25)). Therefore we set the range (35-75) with the lowest possible concentration of E (35) and adjusted the centre point of E after DOE design accordingly (55->50).

After DOE evaluation c-E was not significant for any of the results.

Is this outcome based on the results determined or is it caused by the manipulation of the DOE design?

You can find the file attached. Two sets of results have been produced in sequence: Order: block 1 with t-P=8 and 20 min,  block 2 with t-P=8 and 20 min and block 3 with t-P=8 and 20 min, 

3 ACCEPTED SOLUTIONS

Accepted Solutions
Highlighted
phil_kay
Staff

Re: May manipulation of centre point of DSD DOE (asymmetric nature of factor) influence the results?

I have to say that I dont understand everything about the experiment that you have carried out. It looks like you originally created a 5-factor DSD with 3 blocks. I don't understand what all the other variables are. Are they all responses? Not sure why there are 2 sets of results. Did you carry out 30 runs? Or 15 with duplicate response measurement?

 

I can answer your main question without having to understand any of that.

 

You have a DSD where the centre-point for 1 of the factors is off-centre. The best way to look at the consequence of that is to create a design table for the original design and to create another with the shifted lower limit for 1 of the factors (attached examples). Then use Compare Designs. In this case you should see that it does not make a big difference to power, correlation of effects or any of the other design diagnostics.

 

It will mean that you can't use the Fit Definitive Screening platform.

 

Hope that helps,

Phil

View solution in original post

Highlighted
phil_kay
Staff

Re: May manipulation of centre point of DSD DOE (asymmetric nature of factor) influence the results?

To be clear: you can't use the Fit Definitive Screening platform. That is, the platform for specifically fitting a model to the results of a DSD. You will not be able to use that because the altered design is no longer a DSD.

View solution in original post

Highlighted
phil_kay
Staff

Re: May manipulation of centre point of DSD DOE (asymmetric nature of factor) influence the results?

The important message is that the design properties are not radically altered by the change that you made. Although you should check yourself to ensure that you are comfortable with this.

 

Not being able to use Fit Definitive Screening is a small point and I don't think it is important to you. I get the impression that Fit Definitive Screening is not something that you were aware of anyway. This thread discusses this in more detail.

View solution in original post

4 REPLIES 4
Highlighted
phil_kay
Staff

Re: May manipulation of centre point of DSD DOE (asymmetric nature of factor) influence the results?

I have to say that I dont understand everything about the experiment that you have carried out. It looks like you originally created a 5-factor DSD with 3 blocks. I don't understand what all the other variables are. Are they all responses? Not sure why there are 2 sets of results. Did you carry out 30 runs? Or 15 with duplicate response measurement?

 

I can answer your main question without having to understand any of that.

 

You have a DSD where the centre-point for 1 of the factors is off-centre. The best way to look at the consequence of that is to create a design table for the original design and to create another with the shifted lower limit for 1 of the factors (attached examples). Then use Compare Designs. In this case you should see that it does not make a big difference to power, correlation of effects or any of the other design diagnostics.

 

It will mean that you can't use the Fit Definitive Screening platform.

 

Hope that helps,

Phil

View solution in original post

Highlighted
PS_Ato
Level III

Re: May manipulation of centre point of DSD DOE (asymmetric nature of factor) influence the results?

Hi Phil,

thank you for looking into this.

I'am not quite sure, if I got it.

According to the design comparison, it does not make a difference. - So, the manipulaiton seems to be alright / without consequences.

But, on the other hand you write, that I cannot use the DSD platform.

Isn't this a contradiction to the result of the compare design outcome?

 

Looking forward to your explanation,

Many thanks,

Petra

Highlighted
phil_kay
Staff

Re: May manipulation of centre point of DSD DOE (asymmetric nature of factor) influence the results?

To be clear: you can't use the Fit Definitive Screening platform. That is, the platform for specifically fitting a model to the results of a DSD. You will not be able to use that because the altered design is no longer a DSD.

View solution in original post

Highlighted
phil_kay
Staff

Re: May manipulation of centre point of DSD DOE (asymmetric nature of factor) influence the results?

The important message is that the design properties are not radically altered by the change that you made. Although you should check yourself to ensure that you are comfortable with this.

 

Not being able to use Fit Definitive Screening is a small point and I don't think it is important to you. I get the impression that Fit Definitive Screening is not something that you were aware of anyway. This thread discusses this in more detail.

View solution in original post

Article Labels

    There are no labels assigned to this post.