counterbalancing help

ABX

Dear All,

I would like to automate the counterbalancing in an experiment as much as possible and have looked at the counterbalancing tutorial and several posts on this forum but still am unsure.

We have: 2 file pools x 3 audiovisual modalities x 2 music styles (and 24 participants)

I need to satisfy the following:
1) Several video clips will be presented in three modalities: Audiovisual (AV), Visual only (VO) and Audio only (AO), so 3 different orders need to be counterbalanced within-subjects.
2) We are creating 2 separate file pools so that we prevent carryover effects between AV and VO/AO conditions. These will have to be counterbalanced, e.g. odd subject numbers get AV from file pool 1; VO and AO from file pool 2, and vice versa for even numbers.
3) There will be two different music styles: 1 and 2, the order of which would preferably also be counterbalanced within-modality but could also be randomised. (Video files will also be randomised within music style – but that’s easy!)

Could someone please help? I think Edwin’s solutions (April 2013) would be useful if I could understand how to implement them in the experiment procedure (see: http://forum.cogsci.nl/index.php?p=/discussion/415/solved-specific-participants-specific-conditions)

As a really clunky fix, I could hard code the experiment so that there are 12 conditions and then subjects are assigned just one condition only. But I'm still not sure how to do this. I can do an inline script, but how to set up the procedure?

Any help would be most appreciated!

eduard

Hi,

Sorry, but I don't really get what you mean with carry-over effects and why you would need to file pools.

Do I understand correctly, that you your 2x3x2 conditions are all across participants? That is that one participants does a specific order of all 12 possible orderings? Or how exactly is this supposed to work.

In general, Edwin's solution makes sense. It is based on the idea that you use the subject number (var.subject_nr), to decide to which condition the participant is assigned to. That requires that you define the conditions in the beginning of your experiment. Alternatively, you can also generate lists for each condition outside the experiment, and just read a specific list with variables for a particular condition.

Does that make sense? Sorry for not giving more detailed help, but I don't get your description well enough for that.

Eduard