Have an idea?

Visit Sawtooth Software Feedback to share your ideas on how we can improve our products.

Missing Choice Tasks

In most data sets, a small proportion of participants discontinue the survey before completing all choice tasks.   For example, in a study with 15 choice tasks, we may have 25 participants who completed from 1 to 14 choice tasks.   Are there guidelines for deciding what constitutes a minimum number of completed choice tasks?  Thank you!
asked Mar 8, 2013 by cunnic Bronze (1,440 points)
retagged Mar 8, 2013 by Walter Williams

1 Answer

+2 votes
I don't know if there's a hard and fast line here.  I usually allow respondents in if they only missed a few questions - say no more than a quarter of the questions asked.  I might even run some tests - what does my data look like if I kick out respondents who skipped any of the choice sets?  What does it look like if I kick out only those who skipped more than 1, more than 2, more than a quarter?

I usually see that it makes very little difference in terms of aggregate utilities.  It can start to make more of a difference in simulations or in segmentation, but even there the effect is testable.
answered Mar 8, 2013 by Keith Chrzan Platinum Sawtooth Software, Inc. (91,775 points)
Thanks, Keith.  This is good advice.