Hi all,
I am having trouble understanding the output in the results viewer under the "The OPTEX Procedure". I would like to know the d-efficiency of my design, but there are design numbers 1 - 10 listed in order of decreasing d-efficiency. Which design number relates to the experimental design just created? Can anyone share a link to maybe a tutorial or help page with a description of how to interpret the page I am looking at (screenshot below).
Thank you!
@sarah123 wrote:
Thank you so much!
My understanding of Example 15.1 is that I should be seeing the "best" design, and that the more times the highest d-efficiency score is in the table, the more confident I should be that that is the the highest possible d-efficiency of available design runs.
I don't think "confident" is a word that applies here. In example 15.1, the first 6 different designs have equal D-optimality (in other words, by that criterion, there is no difference between these designs), while designs 7 through 9 are just slightly worse than the first 6.
And that if I were programming (rather than using the interface) I might be able to view Designs 2 and 7 and compare them with each other. Design 2 is equally as d-efficient as the one I am looking at already, however, it would have different combinations of attributes and levels. And Design 7 would be less d-efficient than either Design 1 and Design 2.
I think that's correct, although I would probably phrase it as "Design 7 would be very slightly less d-efficient than either Design 1 and Design 2."
See the first example in the PROC OPTEX Documentation.
Thank you so much!
My understanding of Example 15.1 is that I should be seeing the "best" design, and that the more times the highest d-efficiency score is in the table, the more confident I should be that that is the the highest possible d-efficiency of available design runs.
And that if I were programming (rather than using the interface) I might be able to view Designs 2 and 7 and compare them with each other. Design 2 is equally as d-efficient as the one I am looking at already, however, it would have different combinations of attributes and levels. And Design 7 would be less d-efficient than either Design 1 and Design 2.
Please let me know if I misunderstood anything.
@sarah123 wrote:
Thank you so much!
My understanding of Example 15.1 is that I should be seeing the "best" design, and that the more times the highest d-efficiency score is in the table, the more confident I should be that that is the the highest possible d-efficiency of available design runs.
I don't think "confident" is a word that applies here. In example 15.1, the first 6 different designs have equal D-optimality (in other words, by that criterion, there is no difference between these designs), while designs 7 through 9 are just slightly worse than the first 6.
And that if I were programming (rather than using the interface) I might be able to view Designs 2 and 7 and compare them with each other. Design 2 is equally as d-efficient as the one I am looking at already, however, it would have different combinations of attributes and levels. And Design 7 would be less d-efficient than either Design 1 and Design 2.
I think that's correct, although I would probably phrase it as "Design 7 would be very slightly less d-efficient than either Design 1 and Design 2."
Great! That helps a lot. Thank you
SAS Innovate 2025 is scheduled for May 6-9 in Orlando, FL. Sign up to be first to learn about the agenda and registration!
Learn the difference between classical and Bayesian statistical approaches and see a few PROC examples to perform Bayesian analysis in this video.
Find more tutorials on the SAS Users YouTube channel.
Ready to level-up your skills? Choose your own adventure.