Market researchers often need respondents to rank lists of items (logo designs, messages, slogans, new product ideas, names, etc). The problem? Most people can’t accurately rank a list of more than a 4-5 items. Their top and bottom ranked items may reflect their true preferences, but it is a toss-up if their middle-ranked items actually reflect their true opinions. Don’t believe me? Try listing in order your favorite 20 movies of all time. It requires some time and effort to give an accurate rank ordering, especially around movies ranked 12 – 16.
Maxdiff is a market research technique that gets around this problem by breaking down the ranking task into sets. Respondents are shown sets of four items. In each set they choose the item they like best and least. After they see enough sets, we can use fancy math to compute accurate rankings of the items.
At FanJuicer, we use this technique to poll sports fans.
We think this is a more efficient and accurate way to understand fan perceptions. This technique has the added benefit of producing interval level data. That just means that the distance between the items ranked first and second doesn’t need to be the same as the distance between the items ranked second and third. This produces natural “tiers” of items, which are another major benefit of this technique over a traditional ranking exercise.
Interpreting The Results
- Results from the MaxDiff are indexed so that the average score is 100.
- Teams with scores above 100 received above average rankings. A score of 120 means that team was ranked 20% greater than the average team in the exercise.
- Teams with scores below 100 received below average rankings. A score of 80 means that team was ranked 20% lower than the average team in the exercise.