MaxDiff is a statistical technique that examines the way consumers decide between 3 or more attributes. It is sometimes referred to as best-worst scaling and is similar to conjoint analysis. It was developed in the 1990s.
What are the aims of MaxDiff?
MaxDiff aims to evaluate preferences across a large number of features or benefits. It is superior to asking respondents to ‘tick all the aspects they feel important’ as it prevents them from ticking all aspects!
How is MaxDiff undertaken?
Respondents are shown a sub-set of possible items (typically 4) and asked to indicate the best and worst of them. This process is repeated with different combinations being revealed – each item typically being shown 2+ times. It is typically used for 15-40 items – although advanced models can handle 100s. Large numbers of variables are handled because a respondent only sees a small number at a time. Based on the overall pattern of scores, it is possible to create MaxDiff scores for each respondent.
Outputs are easy to understand because they place each item on a scale of 1 to 100, with all items summing to 100. This makes it an effective tool for international research as there is no opportunity for scale-bias.
With results being calculated for individual respondents it’s possible to create aggregated scores for any sub-group of interest so that preferences can be established.
What’s an example of MaxDiff in use?
We recently used a MaxDiff approach to work out preferences for a home delivery service. The wording of the MaxDiff question went along the following line: ‘please think about home delivery of clothes. Looking at the list of options below, select the MOST important aspect to you, and the LEAST important aspect.’ This approach allowed us to understand the relative importance of price, speed and flexibility.
What is MaxDiff?