Improve the variety of sampling options for timeline variables

[jodeleeuw]

Need to figure out what kinds of options are missing but in demand. Questions on the Google Group are a useful starting place.

Ideas:

• Improve alternate-groups method so that it can sample a subset rather than merely shuffle.

[kupiqu]

Another thing that would be great to have would be allowing associative and complementary sampling (I think it's not currently available, but please let me know if wrong).

Let me describe what I mean with a simple example:

Imagine that you have two blocks of trials A and B and have two sets of instructions I and II, respectively for them.

Now consider that you want to randomize which one (I-A or II-B) appears first across participants.

One way to do that would be randomly sampling instructions I and II, but then the problem is to select the associated block of trials deterministically, A or B respectively.

• EDIT (forgot the second half):

• And not only that, then you need to add the complementary situation, which is deterministically doing II-B if you first did I-A, or vice versa (do I-A if you first did II-B).

As mentioned above, if there is a way to do this already in jsPsych, please let me know (and sorry for the noise).