Correcting Incorrect Scale Values

[mg523]

I have been using Media Lab to administer various questionnaires for my research study. I recently learned that when programming Media Lab for my experiment, one of the items in one of my questionnaires had an incorrect value assigned to one of the answer choices.

To explain this more clearly, I am using the questionnaire known as the Children's Depression Inventory (CDI). Each CDI question has three possible answer choices. Each choice is given a number value (0, 1 or 2) that I use as a way to score the questionnaire. I recently learned that when programming my experiment in Media Lab, two of the three answer choices for question 2 were a assigned a numeric value of 2, instead of one choice being assigned 2 and the other being assigned 0 - i.e. I can't tell fom the data output if the subject intended to choose the answer choice with a value of 2 or 0 because both choices were assigned 2.

I was wondering if there in any way to determine which answer choice was actually chosen, other than looking at the straight number values. For example, is there a way to determine which answer choice button was clicked in Media Lab? Any advice and guidance that you can offer would be very helpful.

Thank you.

[jarvis24]

[edited from email follow-up thread]

Could you please send me a copy of the que file and a zipped copy of a typical data folder used during the study for response collection? I'll take a look and see if I can figure anything out. In the meantime, there may be some info related to the issue here (see link below) but its not clear why "2" would be the common response being stored for options of "2" and "0".
www.empirisoft.com/Support/showthread.php?t=334

I have attached the que file, as well as a zipped copy of a data folder. Let me know if you need anything else. I looked at the link that you directed me to and it appears that posts 2 and 3 address my issue - i.e. determining which screen button was clicked in the questionnaire, rather than determining which choice was clicked according to the values that appear in the data files. As I mentioned, the values for item 2 of my questionnaire were programmed incorrectly and I would need to know which screen button was clicked in order to know the true response of the subject. So it appears that posts 2 and 3 address my issue. However, it is not clear to me how to access this information. Thanks again.

Thanks for sending the files--it really helped. I see what you mean about the second item recoding issue (i.e., the first and third options both being recoded and saved as "2"). Am I correct in understanding then that all of the other items are ok?

Yes, your understanding is correct. Is there any hope for saving the data from the second item?

As far as I know, there is not. I don't believe there are any data files saved that include the non-recoded values. Sorry, I wish I could give you better news. Maybe that is an option we could somehow work into the next version.

Thank you for looking into this. For clarification sake...last week you emailed me a link to a few different posts that appeared to be related to my problem (www.empirisoft.com/Support/showthread.php?t=334). The third post seems to imply that there is a way to determine which button was clicked, regardless of the value that is coded (ex. button 2 was pressed, despite the fact that button 4 is the one that codes for 2 children). Is this only referring to keyboard button clicks? Or can it refer to screen buttons that were clicked using the mouse (which is the way that we do it in my paradigm)? If there was a way to determine which screen button was clicked in my paradigm, I believe that would solve my problem.

I think those solutions apply more to issues of labelling than for recoding. e.g., if you simply label buttons 1-3 as 0-2 you may find out what was
actually clicked or pressed. But if two buttons are both recoded as, say
"2", I can't imagine how the original values could be known or restored.

Thanks for all of your help with this. It is much appreciated.