identify correct answer after random pairing

Hi!

(see attached my reduced input-file)

In the presentation phase (here just trial 2 and 3) I combine among other things stim4 (a dichotomous additional info = E or G) with stim5 (a sentence about a person). I have 6 lists of different sentences used as stim5 (here just v_con_j.txt), just to make clear that it is not only this one.
The combination of the sentences in these lists as stim5 with the dichotomous info E or G should be randomly done by directRT. Therefor I use $ and WGR.

Later I want my participants to indicate for each sentence the dichotomous info, i.e. if it was E or G. The sentences should be presented randomly, then the participant should answer. Because I have a jump mid-trial in these recognition-task it doesn’t seem too easy (here trial 5 and 6) ..

Here is my problem: How do I identify, if the answer of the participant is right or wrong?

I would be sooo happy if you have any suggestions!

I’ve found this thread:

[url]http://www.empirisoft.com/Support/showthread.php?t=213[/url]

there it is a similar problem: to use both, jumping/skipping and randomization. If I just use the $, as suggested in this thread, I do not know with which additional info (E or G) this random selected stimulus was combined before. But when I let show one stim, which has been combined with E and one who has been combined with G, then I need the wgr-randomization, but then I can’t jump mid-trail to the next, ‘cause this results in an early ending.

So, how can be wgr and jumping mid-trial be used together?

Thanks for any help!!

Maybe it's just too late in the day but I'm having trouble following this one. What would help, I think, is to have a zipped complete set of files (including stim) so I could run a test session. It would also be ideal if you could simplify the input file even further to that it contains only what is required to demonstrate the problem (e.g., deleting [I]anything/everything[/I] that isn't required to see the problem). Is that possible?

Thanks a lot for helping me!
I would love to run this properly.

Here is a reduced folder so you can run it. I translated some things that it might be easier to understand the structure of this experiment. I’ll also try to explain it again:

There is a learning phase – here trial 2 and 3. There you will be introduced to several persons. You see a name, an age which is young (20-25) or old (75-80), the information if the person is grown up as an only child or with siblings, and something the person is used to do (let’s call this “item”).
All participants hear the tones, for some there is a second task to generate cognitive load.

There are item-lists which are typically young and such which are typically old. So I can combine a young person with typically young behavior or with typically old behavior, etc..
The combination of any item with “only child” or “siblings” should be random. The idea is, that this information (“only child” or “siblings”) is used to explain if a person behaves in a typical or an untypical way – no matter what the information is.

In the recognition task – here trial 6 to 9 – the participants will be presented either a real item which occurred before or a distracter. I’ll ask, and they respond (via 1 or 9):
- do you recognize this behavior from the previous persons or is it a new one?
- was this behavior performed by a young or an old person?
- has this person grown up as an only child or with siblings?

If they answer “new” in the first question, clearly the following 2 would be senseless, therefore I use the skip command (sorry for saying “jump” in the last e-mail) in the 4th time column, as you see.

I let the items and distracters pull from a list where all behaviors (typical and untypical und distracters) are listed. But then I don’t know which combination (of which item with “only child” or “siblings”) was randomly made in the presentation phase by directRT. Well, I see the sentences in the log-files but surely I wont pick each combination and check the answer of the participant myself.

This is my problem! I can’t check if my participants give right answers, since directRT made the combination randomly.

It would be perfectly if I could let pull from a list where only one kind of combination is listed. For example I could write in one list
?b-7t2s5
?b-8t6s5
?b-9t10s5
etc.
and in the log file I would know, this young behavior was combined with an old person and “siblings”. I could write a code number in the column called “!numcodes” and let all such trials be randomized via WGR.
But since I have the skipping in the middle of these recognition-trials this would cause an early ending. Or do you know how I can use both, skipping mid-trial AND wgr-randomization? (So my old question is open again.. : [url]http://www.empirisoft.com/Support/showthread.php?t=1402[/url] )
Or do you have any other ideas?


My last option would be to make a list of these items which will be combined with “only child” and one list for “siblings” and to switch this for half of the participants – but this is not the best solution, since you can always argue afterwards: maybe you used combinations where there is some sense for the explanation of untypical behavior. But there should be no sense for the explanation “s/he is watching the simpsons (= behaving typical) because s/he is grown up an only child” – and I can guarantee that if it is randomly combined.

I appreciate your help very much, thanks a lot!
Julia