Difficulty formatting data for SPSS

SuperLab SuperLab 4 Support
1

I’m having difficulty formatting the .txt file output from Superlab to be compatible with SPSS. The participant’s response (e.g., “agree”) and reaction time are in the same row in the output file. When transposed to format the data for SPSS compatibility, the participant input & reaction time wind up in a single column. SPSS requires each column be a single variable. In this case, the column winds up containing two variables (participant input & reaction time).

Someone else posted the same question on this forum: “import file superlab in spss, Posted May 14th, 2008”. Is the only solution to write unix-style code to move the data around? If so, how does that deal with “participant errors”? I used macros in excel to move the data around to a compatible format, but had trouble when participant errors are included in the data. Superlab records all errors and reaction times; therefore, if the participant made an error (selected “f” instead of 1-5 of the presented scale), it is recorded and then the correct answer is recorded as well. This participant’s data winds up being a much longer string in the .txt file, and the macro/unix-code will wind up moving the wrong data points.

I would be very surprised if only one other person is having SPSS comptibility issues. Please let me know how you have dealt with this problem in your data. Thanks!

2

UNIX filters for experimental data

It is true that moving around columns or rows without taking the structure into account will fail in many cases, and the untagged matrix format preferred by Fortran-style programs like SPSS is not very compatible with situations like what you describe.

What I usually do is to create a filter that is sort of a small state machine that “knows” the structure of a trial. Then as the filter scans through the events in a logfile, it can distinguish between multiple responses and format the output accordingly.

One hint: try to avoid using “default” too much in these scripts. It’s generally better to throw an error so that you can gradually develop a specific action for each case the script discovers in the data.

Greg Shenaut

3

Admittedly, we don’t have in-house experience with SPSS. If you think that Data Viewer can be modified to make it easier to move data files to SPSS and are willing to guide us in the effort, please let me know.