Create a Conditional Skip Pattern?

Thanks in advance for the help on this one. The logic for the study is a little complex, but I will try to simplify it here.

The study is a learning study where participants are asked to learn and memorize a social network by being exposed to questions about the relationships in the network and then learn from their mistakes. It is important that the participants be able to demonstrate mastery of the social network by being able to answer a series of 12 (randomized) questions with 100% accuracy twice (sequentially).

So, if a participant can answer all 12 questions about the social network with 100% accuracy the first time, they are asked to do so again with a "reshuffled" series of the same questions. Now, if they are 100% successful at this second series, they complete the study. If they are not 100% accurate in EITHER the first or the second series, then they must "go back to the beginning" and start anew.

Thus far, we have successfully created questionaires that will randomly ask the participants the social network questions. We have also attempted to create the Excel responses file that will aggregate the participants' answers to determine if they answered all the questions correctly. What I need help with is how to create these two "loop" structures:

IF, the participant misses one or more questions in the first series, go back to the start. Otherwise, continue to the reshuffled second series.

IF, the participant misses one or more questions in the second series, go back to the start of the first series. Otherwise, continue to the completion of the study.

Attached are the experiment, questionaires, and response Excel spreadsheet. Thanks greatly for any insight you can provide us!

Interesting question, I wonder if Medialab can not accept a conditional skip backwards from the response.xls file?

Yes, backward skipping isn't really doable in this case--even with an [I]responses.xls[/I] file. A bit awkward but you could have repeated blocks of the items within your questionnaires so that if they get the 24 sequential items correct then get skipped over the next 50 copies of the same questions. One nice advantage to that is that every copy of every question would have its own variable name and consequently you could track all the incorrect as well as correct responses.

Finally, you could do this as a custom item if you want a more elegant solution, but the approach above I think would be faster.

For the repeated blocks solution, would you use a response.xls file or could that setup be created within the media lab experiment?

OK. I'm still having trouble getting the skip pattern to work with the responses.xls file. Could you explain exactly what the responses.xls must contain and where it must be located for MediaLab to utilize it?

Thanks greatly.

I would just do it all within a single questionnaire--no xls file. I guess it depends on whether you want to have them start the whole thing over immediately if they get an item wrong or if you want to let them finish.

In this case, it is important that the participants continue through and finish the series of 12 randomized questions before evaluating their success/failure at achieving perfect recall.

If we do not create a responses.xls file to aggregate the participants' score, how can the questionnaire do so with the 12 questions in a randomized order? If we knew ahead of time what the last question would be, then I could create a skip pattern - but there is no way of knowing >_<

Again, thanks greatly for your help!