ADX interface optimal design with blocking

[sarah123]

Hi all,

I am hoping someone can clarify some questions I have about creating d-optimal designs in SAS for discrete choice experiments. I am using a main effects with 4 factors. One 2-level factor, two 3-level factors, and one 4-level factor. I notice that when I do not check the "blocking" box the number of runs is automatically 19. However, I would like to make the choice task easier so I include blocking. When I do this, the number of blocks is automatically equal to 2 and the runs per block equal to 5, making it 10 total runs. Why is there such a large difference between the number of runs with and without blocking?

Thanks for any advice.

[WarrenKuhfeld]

I don't know why, but I would not worry about it, particularly in the context of a choice experiment. You can go here to learn more about software more specifically designed for choice experiments.  http://support.sas.com/techsup/technote/mr2010c.pdf

[sarah123]

Ok, thank you! I understand that when blocking is not selected, the automatic number of runs is 10 more than a saturated design, but what the automatic number of blocks and runs per block are, I'm still pretty confused and have been manually changing the number of blocks and runs per block to be higher. I am trying to increase d-efficiency of the design. If this issue isn't particularly relevant for choice experiment design then I will quit worrying about it.

[WarrenKuhfeld]

I would not worry about it.