Anchored MaxDiff

In a MaxDiff exercise, respondents typically choose items from groups of 3 to 6, selecting their “best” and “worst” options. These choices represent relative judgments among items chosen by the researcher.

But what if a respondent doesn’t eat sugar and is asked to pick the best and worst ice cream flavor? They might make choices similar to those of an enthusiastic ice cream eater, even if they don’t personally consume the product. Anchored MaxDiff addresses this issue by adding an absolute reference point, allowing researchers to determine whether respondents find the items important/not important or would buy/not buy them in a more absolute sense.

Anchored MaxDiff achieves this by appending an additional grid question to the end of the MaxDiff exercise. This grid presents a subset of items (typically around 7) that range from the respondent’s most to least favored items, based on their prior choices. Respondents then indicate which of these items they would actually buy or consider important.

This approach to Anchored MaxDiff, known as the Direct Binary Approach Method, was suggested by Kevin Lattery and presented at several Sawtooth Software conferences.

The results are displayed with a utility boundary line between important and unimportant items. Items above the line are considered important (positive utility), while those below are considered unimportant (negative utility) in an absolute sense (e.g., buy/no-buy).

The responses to anchoring question, in addition to the MaxDiff exercise results, contribute the information needed to estimate the anchor line (utility boundary) during analysis. 

Be aware, by turning on the anchored MaxDiff setting, the exercise is considered an anchored MaxDiff exercise in analysis.

1. Open the Advanced tab.
2. Toggle on the Anchored MaxDiff setting.
   

Configuring the anchored MaxDiff question is relatively straightforward.  

1. Configure the Anchor question text as you would any other question text with rich text, images, and more (see text formatting page for more details)  
2. Enter labels for the “Important” and “Not important” labels in the question 
3. Optionally, toggle on the override for the number of items shown to the respondent on the anchor question. By default, 7 items are shown (unless there are only 6 items in your exercise, then all 6 items are shown). At a minimum, 5 items must be shown. The number of anchored items displayed should depend on the length of your MaxDiff items list. As the number of items in the MaxDiff exercise increases, consider increasing the number of items that display in the anchored MaxDiff question to better help the algorithm determine an accurate utility boundary between the important and unimportant items. 

 

View the anchored results by viewing the MaxDiff summary in analysis.

 

Anchored scores are displayed on a positive probability (ratio) scale, where we’ve chosen to index the anchor at 100. Therefore, an item that has a score of 50 is half as preferred/important (selected) as the anchor. An item with a score of 100 has an equal chance of being above or below the anchor (in terms of preference/importance. An item with a score of 200 is twice as preferred/important (selected) as the anchor.