Hi all, 

I've been working on an experiment that involves participants being in either 2-person groups or 3-person groups across four treatments (2 tasks with 2-person groups and 2 tasks with 3-person groups). I'd hoped to use a within-subjects design with the task order, which means I need to randomize the task order within-session. I know the usual way this is done is via pages (and this is how I usually do it). However, in this case, since different group sizes are necessary by task, that will not work unless I can somehow change the number in group from round to round (which, as far as I'm aware, will not work since it is set in Constants). The other option would be to randomize the order of standalone apps, which there's no straightforward way to do by participant within the same session. The last option would be to set some variable at the player level (or maybe an ExtraModel) to create subgroups (which I've done before in other experiments). However, I'd like to avoid this last option if I can since it's clunky and not at all flexible (especially when working with a variable number of participants). So, 2 questions: 

1) is there a way to change group size between rounds (perhaps using set_group_matrix() or something similar)? 
2) if not, is there a workaround to randomize the app order that works with oTree>5.0?

I suspect the answer will be no to both, but I thought I would ask in case anyone has stumbled across anything. 

Thanks!

Braxton

(1) yes you can use set_group_matrix.
(2) no. There is an app I have seen recently that attempts to do this, but it heavily mucks around with oTree's internal functionality, meaning it is likely to cause compatibility problems in the future.