Have an idea?

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

MaxDiff – How to proceed with incomplete/partial responses?

Hi everyone!

As you probably all know, many participants drop out during one of the MaxDiff sets. How would you recommend to proceed with incomplete answers?

We have 7 attributes/items in our MaxDiff, show 4 attributes/items per set and have 5 sets in total.

Some participants dropped out after the 1st set, while some others dropped out after the 4th set. Since, for instance, in case of the latter, there is only one (unanswered) set missing, I was wondering whether I could include those?

Any views on this would be greatly appreciated!

Many thanks in advance.

asked Aug 6, 2020 by o4034695 (200 points)

1 Answer

0 votes
Whether to include respondents who drop out early is a good question without a clear answer.   One concern is about potential information content being sufficient.  A second question is whether to cast greater suspicion on respondents who terminate without finishing MaxDiff.  Were they not engaged and were likely giving poor data before they quit?

Our software allows you to include respondents with partial MaxDiff data in the analysis.  I suspect you're using HB estimation for individual-level scores, and in that case I'd want to know that enough questions were completed such that each respondent typically saw each item at least 1x.

Also...you probably already saw the warning when you tried to create a survey that showing more than half the total items in each MaxDiff set is inefficient.   Just a heads-up for next time.
answered Aug 6, 2020 by Bryan Orme Platinum Sawtooth Software, Inc. (187,915 points)