Millisecond Forums

If you get stuck, post a small concrete example [1] that illustrates your intended procedure here. Then I can give you a more specific response (including example code) on how to go about implementing it.

[1] I.e. using fewer items, types and/or categories; just enough to get a sense of your intended procedure and the stimulus interdependencies.

Try to use conditional branching as you suggested. Thank you very much for your help!

> The problem is that stimulus pairing wouldn't work, would it?

The stimulus pairing works the same -- regardless of whether you have both stimuli in a single <trial> or in two separate <trial> elements.

And for "trial"-pairing, you can use conditional branching (cf. /branch attributes) to invoke the proper "testboard" <trial> element from the respective "studyboard" <trial> element.

> I could build a bunch of two-trial blocks [...]

That seems entirely unnecessary.

The problem is that stimulus pairing wouldn't work, would it? what I need is rather trial pairing - as both of the stimuli (study board and test board) require participant's response and thus two separate <trial> elements are needed. I could build a bunch of two-trial blocks and then build an experiment out of that but it does not seem to be a very neat solution as it would yield a lot of blocks...

By counterbalancing i meant within- subject design so the number of test boards of each type presented to a participant. 

> I want counterbalancing to be based on test board types rather than study boards.

I'm not sure what that means in concrete terms. Counter-balance what and how exactly? Between-subjects? Within-subjects? etc. A specific example would be helpful here.

> Also, for a given test board there is a particular study board that needs to be presented before (they cannot be randomly drawn).

You can pair your study- and testboard items as needed. See the "How to present stimulus pairs" topic in the documentation for details.

Also, I don't see why having a single <trial> element to display both study- and testboard would somehow avoid those two issues. If you have worked those out for the single -<trial>-element case, the solution should apply to the two-<trial>-case as well.

Dave, thank you very much for your reply! I thought it would probably solve the problem. But now I have 4 types of study boards and for each study board there are 6 possible types of test boards. I want counterbalancing to be based on test board types rather than study boards. Also, for a given test board there is a particular study board that needs to be presented before (they cannot be randomly drawn). Do you have any suggestions how blocks should be defined to take these into account?

You need two <trial> elements to do that:

<trial a>
/ stimulustimes = [0=fixation_cross; 500=study_board]
/ validresponse = (" ")
/ branch = [trial.b]
</trial>

<trial b>
/ stimulustimes = [0=test_board]
/ validresponse = ("B", "N")
 ...
</trial>