Do identical variable names cause lost data?
Do identical variable names cause lost data?
[edited from support email]
After collecting all my data, I realized that I made a serious error in naming my variables. I had two questionnaires, each which were exactly the same, but I showed different pictures in each. The variable names were exactly the same. They were placed in the same experiment, with questionnaire 1 always coming before questionnaire 2. The program did not record one set of responses. I need to know whether Medialab would have overwritten the responses to questionnaire 1 or whether it would simply not have recorded the reponses to questionnaire 2. I cannot tell by simply looking at the data as the theoretic responses should be the same.
Check out the data section in the users guide--especially with respect to the byVariablename versus byQuestionnaire data formats. In your case, I think the data you are looking for will be found in the byQuestionnaire data folder. But yes, in general, it's a good idea to use different variable names when the questions are presented to the same subjects. Let me know if that helps.
-Blair
Something interesting (not in a good way) happened in my experiment. The experiment had 4 questionnaires in it, but data did not collect for the last questionnaire. Even though the person went through the entire experiment, their final questionnaire data did not appear in either the ByVariable or ByQuestionnaire folders...they didn't appear in the .txt file either. After the first three questionnaires and their data, it lists the last one, but with no data. I found this odd because the WriteData box was definitely checked. I created this questionnaire by copying the third one and modifying some file names, but the variable names were the same. When I tried to run the questionnaire by itself later, it did not collect any data. Seems like it got corrupted somehow. Anyway, now I need to rerun half my experiment! Arrrgh :(
That sounds pretty weird. Can you zip your entire data folder and send it to me along with a copy of your exp and que files? I'll take a look and see if I can tell what happened.
Update 8/22/06:
Variables that were not being written to the byQuestionnaire data files were being skipped because they did not have a valid value in the 'WriteText' column. The WriteText column is supposed to have a 1 if the variable is numeric and a 2 if the variable is text. For whatever reason, these variables all had a 3 in this column. Very strange. We'll add a trap to detect that just in case anyone else enters or otherwise ends up with values other than 1 or 2 there.
