perlman@capybara.cis.ohio-state.edu (Gary Perlman) (03/14/89)
I am working on a new program (for the PC) to help code and analyze
overlapping events such as those you might get from a videotape of a
user learning how to use a program. I would appreciate any references
to appropriate data analysis techniques. The next diagram should
provide a bit more detail about the sort of data I am working with.
Codes of possible events:
H=The user is asking for help from the system.
W=The user is working productively.
E=The user has made an error and is correcting it.
Q=The user asked a question.
C=The urser made some comment or complaint.
1,2,3=Working on Task 1, 2, or 3
Time ---->
H W E H W
|-----------| |---------| |---------| |-----| |---------|
1 2 3
|---------| |------| |--------------------------------|
^ ^ ^ ^ ^ ^ ^ ^
Q Q C Q Q C C Q
Notes: Some events are momentary or instantaneous while
others have durations. Some tasks are mutually exclusive
while others can happen at the same time (for example, two
people might be talking at the same time). Right now, I
have the interactive coding system made and am looking for
suggestions for the data analysis part.
Obvious analyses are sums of counts and times for events.
Only slightly less obvious are conditional probabilities
between events. Are there standard analyses for these
data? Are there analyses that you have used (or wanted to)
and wanted to see in a program? (I have Ericsson & Simon.)
The program will be available free of charge when it is done.
It should be able to run on UNIX systems eventually.
-=-
Gary Perlman Department of Computer and Information Science
perlman@cis.ohio-state.edu The Ohio State University
614-292-2566 2036 Neil Avenue Mall
Columbus, OH 43210-1277