[solved] Complex skips not working across multiple questionnaires

[natty1220]

I'm working on setting up two complex skip patterns in my media lab experiment and I seem to be doing something wrong. I have set up an experiment file with all my questionnaires in it and have set up a responses excel document.

The two skip patterns are as follows:

I need it to administer a second questionnaire if the score on the first questionnaire is above the score of 12. I have this put into the responses excel, but when I go through the experiment it is administering the second portion even if the score is below 12.

I need it to display a prompt at the very end if a participant scores above a certain score on a questionnaire so the administrator can follow up. I also have this set up in responses and it appears to be recording the responses, but it is displaying the prompt at the end regardless.

[natty1220]

Just an update. The excel responses appears to be working as its is creating reports for the participant number that calculates the values correctly and says the correct information for where to skip to. It just isn't skipping things in the media lab experiment?

[natty1220]

The first skip in the responses file is now working, but the second one is not. Please help, I have been staring at this computer for 3 days trying to tweak things to make it work!

[jarvis24]

Try check the case (upper/lower) on the variables in the responses.xls file versus those in the questionnaire(s). If they DO match, then try converting them all to lower case and see if that helps. Not sure which version of MediaLab you are using but I know some earlier versions had issues with case sensitivity in the skipto values.

[natty1220]

The casing is correct, but I made it lowercase just in case. The second skip pattern in my excel is still is not working. The data is correct in the reports file, but it just won't skip in media lab????????? And i'm using version 2008

[jarvis24]

Natty1220, can you post a zipped folder containing your que files so I can try this out in conjunction with the responses.xls file you posted (which looks fine). Or if you can post a simplified version of the session that includes only what is required to reproduce the problem, that would be great.

[natty1220]

Ok I tried to zip the file. I could not open it before I sent it because i don't have winzip on my computer. Please let me know if this worked.

Natalie

[jarvis24]

Natalie--if I understand, you have the first skip working but not the second. This is referring then to the skip which tells the session assistant to inform you if a participant scores 4 or greater on the DSI-SS? But that if the score is 3 or less, then this notification should be skipped? In the attached (simplified) revision of your files, it seems to be working (at least here on v2010). Can you try it? It simply consists of the DSI-SS and then the followup "Administrator Questions for end" que file. If you select the first option on each of the four questions on the DSI do you still see the alert (since they all contribute "0" points to the score)? Or alternately, three "second" response options and one "first")?

If so, could you send me two resulting files:

\natty1_revision\data\reports\1.xls
\natty1_revision\data\byvariablename\Magnolia_Cree k_Study_Long_Assessment.csv

The first file will show the responses.xls data from the session (in this case, e.g., for subject "1") and *should* show a value of "Administrator" in the skipto value for "INSTEND".

The second file should show a value of "99" for SIPROMPT--but only if it is skipped. Otherwise it will read "ok"

Blair

[natty1220]

Actually the DSI skip pattern was working. Its the FASM questionnaire that was not working. I need it to skip the FASM2 questionnaire and go to the TAS-20, if they score a 0 on the first FASM. If they score over a 0 on the FASM then they need to complete FASM2. I'm sorry if the names aren't exactly right for the questionnaires, I don't have access to that media lab computer at the moment.

[jarvis24]

Hi Natalie, I think I finally figured out what's going on. Until tonight, I was quite certain that any skipto values would always "pass through" any subsequent questionnaires until the target variable was found or the experiment ended. But it appears that they terminate at the start of the next questionnaire. I was so surprised by this that I checked back to some much earlier v2004 code and this was the case even back then. I will have to check out some other posts where I know this has come up as an issue. Regardless, it seems to explain your current problem.

The consequent problem in this case is that the responses.xls file assigns the skipto value for FASM2.que at the end of the FASM.que. This skipto value, consequently, was not being applied because skipto's are apparently reset to nothing at the start of the next que file.

SO..

I think I may have found a work-around for now--I tested it here and it seems to work. Please see the attached revision (where just FASM and FASM2 are administered). The trick--I removed your skipto item from the end of the FASM que file and entered it as the first item in the subsequent FASM2 file. That's pretty much it.

You'll now see that if they answer "no" to all questions in the FASM file, the responses.xls file assigns a skipto value of "skipfasm2" to the first item in FASM2. The item "skipfasm2" is a blank instruction item at the end of the FASM2 file.

You may also notice I placed (d1) parameters on the two bogus instruction items--this sets a 1 second automatic continuation time on them so subjects do not have to actually click continue to pass the blank items.

--

As far as dealing with this issue in the future, I'm not sure yet. Even though skipping through que files should be default bahaviour, I don't want to break existing experiments where this "within que" skipto limit is actually expected by previous users of the software.

I'm thinking what I could do is allow users to add an optional character such as "!" at the end of any skipto value--and that this will tell MediaLab to maintain that skipto value beyond the confines of the current questionnaire. I think that would be a fairly easy way to handle it without affecting the backward compatibility of MediaLab with existing experiments. Any thoughts welcome.

Hope this helps.
Blair

[natty1220]

Ok great. I am not back in the office until monday. I will check things out and let you know. Thank you very much for your help

[natty1220]

THANK YOU SO MUCH FOR YOUR HELP! I tried everything out and its working! I literally jumped for joy. Thank you so much! This is going to have such a great impact on our data collection.

[jarvis24]

Great to hear Natalie--esp. the "literally" Let me know if there's anything else I can do to help.

Blair