Millisecond Forums

Hi Dave, 

Thank you for letting me know. To clarify, I cannot have Option A and B set to the original values, plus the manipulations that were selected (mentioned below)? 


 I understand that you are indicating I need to have the <include> to do the combinations. 

<list manipulations_a>
/ items = (
1,2,3,4,5,1,2,3,4,5,1,2,3,4,5,1,2,3,4,5,
1,2,3,4,5,1,2,3,4,5,1,2,3,4,5,1,2,3,4,5,
1,2,3,4,5,1,2,3,4,5,1,2,3,4,5,1,2,3,4,5,
1,2,3,4,5,1,2,3,4,5,1,2,3,4,5,1,2,3,4,5)
/ selectionmode = random
/ not = (values.manip_b)
</list>

<list manipulations_b>
/ items = (
1,2,3,4,5,1,2,3,4,5,1,2,3,4,5,1,2,3,4,5,
1,2,3,4,5,1,2,3,4,5,1,2,3,4,5,1,2,3,4,5,
1,2,3,4,5,1,2,3,4,5,1,2,3,4,5,1,2,3,4,5,
1,2,3,4,5,1,2,3,4,5,1,2,3,4,5,1,2,3,4,5)
/ selectionmode = random
/ not = (values.manip_a)
</list>

<list distributions_orig>
/ items = (1,2,3,4)
/ selectionmode = random
/ replace = true
</list>

If you want to ensure that all 80 combinations occur with equal frequency -- either all within a single participant or split across multiple participants (you talked of sets of 20 split across 4 consecutive participants before) -- then you need explicitly define the combinations and sample from those combinations. If you don't link the lists / don't explicitly define the combinations, you will have no guarantee whatsoever that all combinations occur across participants or that they occur with equal frequency across participants. Some will occur more often than others just by chance, in the extreme some might not occur at all.

As I've explained before: Take your 80 combinations as defined here https://www.millisecond.com/forums/FindPost27573.aspx and split them into four sets of 20 unique combinations each (in whichever way you want). Those four sets of linked lists go into different files -- then you just <include> one out of the four sets as a between-subjects factor.