Add items without upsetting the existing datafile?

[barlab]

I began my data collection and have data for about 30 participants. I decided recently to add 2 items to one of my questionnaires.

When I entered them and ran the experiment, I noticed it failed to create the additional columns in the existing datafile. I was hoping it would create the columns and display the data as missing for the first 30 participants, and have the new data in for any participants run from that point.


I get an error message when I run the experiment and the data is misplaced (no columns are added on the excel document I viewed, and the data was shifted over by 2 cells). When I deleted the data and let medialab create a new file, the variables rows were added and it recorded the data properly and ran it with no errors.


I am wondering if there is a way around starting a new datafile and having to merge the old datafile (without the additional 2 items) with the new. Is there a way to have it record the new data in the existing datafile without causing problems?

[jarvis24]

Normally, you would merge the two based on common variable names (you can do this in Excel, SPSS or Access). But yes, you could modify the headers of the old CSV files to match the current data structure. Remember the files are appended to so the beginning of the files will not not change--data are just added to the end of the file after each session. If you changed the structure of the old CSV file to accommodate the new variables, you could get what you described (i.e., in terms of what you were expecting/hoping to see).

[sjd220]

I've encountered this a few times myself.

My solution has been to rename the old data folder each time I add or remove variables; e.g. "data before added items". Then, when I run the first participant with the changed variables, a new data folder is created that will have the appropriate variable names/headers.

Merging the files is trivial for me. I copy the data (including header row) all into one file and then add spaces/columns where data is missing a variable that was changed. If you make changes several times, this could become tedious, however.