Ubigreen/MeetingNotes > Ubigreen/OpenIssues > Ubigreen/StudyDesign
Yes. That would allow us to track their behavior and make a stronger statement about whether behavior changed (since we don't have time to do an ABA style thing where A=without intervention and B=with intervention). Also, possibly more compelling, the control group could be given access to StepGreen and their data could be used to update that site, so that we could specifically see the impact of the mobile interface.
jen
On Feb 10, 2008 10:14 PM, Jon E. Froehlich <jfroehli@cs.washington.edu> wrote:
What would the control group do? Just carry the technology (running logger software of course) but no interfaces?
j
Jen Mankoff wrote:
Oh I was suggesting *both* between subjects and a 2/1 split. But I just spoke with sue, and she strongly suggested we just do a straight between subjects thing and then show them the other interface (ideally reflecting their own real data) in the interview.
Sue also thought it was really important that we have a control group as well. This would mean something like a 6 6 6 split of people.
jen
On Feb 10, 2008 9:14 PM, James A. Landay <landay@cs.washington.edu> wrote:
I agree and that is what I said (either between subjects or do 2 weeks of 1 and 1 week of another). Another way if we go the first route is to expose them to the SECOND interface in the post interview.
Jen Mankoff wrote: > > How much do we match these interfaces? Do we measure the negative > response as a variable? We use a within subjects design? We'll > use a between subject design. Too long, too many weeks to > design. Tree or polar bear? People are only exposed to one, we > want to test the positive versus the negative. We can have people > point out what they like and dislike. They can't discuss them if > they haven't been exposed to both. > > > I really think we should do a between subjects design with the the > polar bear vs tree and nothing else changing if we only do a two week > study. I'm leaning toward that even if for a 3 week, or maybe 2 wks of > tree, then 1 wk of polar bear (and vice versa) so they can discuss the > differences. With the short timeline we're on, we have to think of > this as a pilot, we can't expect to test more complex stuff. > > jen