Subscribe Bookmark RSS Feed

SHould I use a blocking factor?

emily_crozier0

Community Member

Joined:

May 26, 2016

I want to complete a design experiment to optimise a reaction.

For 3 different reagents I would like to look at 3 factors; temperature, concentration and time.

Do I run three separate design of experiments for each reagent, do I include the reagent as a categorical factor or do I use reagent as a blocking factor.

 

1 ACCEPTED SOLUTION

Accepted Solutions
Dan_Obermiller

Joined:

Apr 3, 2013

Solution

Are you interested in understanding the effect that reagent has on the response? Do you expect the effect of temperature, concentration, or time to change for different reagents? If so, then it is not a blocking factor. A blocking factor is often called a nuisance factor. Something that must change during the course of running the trials that you wish did not change.

 

For the scenario you described, I would think that you want to use Custom Design, specify reagent as a categorical factor. Specify temperature, concentration, and time as continuous factors. Specify the model you want to estimate (probably a response surface model since you wanted to optimize) and let custom design choose the design for you.

Dan Obermiller
1 REPLY
Dan_Obermiller

Joined:

Apr 3, 2013

Solution

Are you interested in understanding the effect that reagent has on the response? Do you expect the effect of temperature, concentration, or time to change for different reagents? If so, then it is not a blocking factor. A blocking factor is often called a nuisance factor. Something that must change during the course of running the trials that you wish did not change.

 

For the scenario you described, I would think that you want to use Custom Design, specify reagent as a categorical factor. Specify temperature, concentration, and time as continuous factors. Specify the model you want to estimate (probably a response surface model since you wanted to optimize) and let custom design choose the design for you.

Dan Obermiller