Have an idea?

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

Anchoring a MaxDiff on a subset of items

I'm interested in fielding a normal MaxDiff with 28 items (N~2000) and anchoring using a top-2-box approach of a 1-5 scaled series of items which are asked after the MaxDiff exercise. Inherently, this will be using the direct binary anchoring approach, though the catch is that the scaled items are only asked of a subset (14 of the 28) of the items. Does implementing this partial-like anchor have a negative impact on our results if we're still planning to report at an aggregate level?
asked Nov 20, 2015 by Matt

1 Answer

0 votes
If the 14 of the 28 items you plan to us in the anchoring questions are well spread among the full range of bad to good items, then you should be in very good shape.  But, imagine if the 14 happen to be the 14 on average best for the sample...or the 14 worst.  That could be a bad situation for finding the anchor point if on average the anchor utility was found in a region you hadn't been able to measure vs. the subset of items used in the anchor questions.
answered Nov 20, 2015 by Bryan Orme Platinum Sawtooth Software, Inc. (172,690 points)