Course · 4. MaxDiff

What MaxDiff is

MaxDiff (maximum difference scaling, also called best-worst scaling) ranks a list of items: features to build, claims for an advert, messages, benefits, product names. Each respondent sees a few items at a time, usually 4 or 5, and picks the most and the least appealing. After a dozen such sets, every item has been seen several times next to different others.

Why not just ask for ratings

Rating scales let people say everything matters: most items end up between 4 and 5 out of 5, and people use the scale differently (some never give a 5). A forced choice has neither problem. It cannot be answered "all of them", and it means the same thing for everyone. The result is a sharper ranking from a short survey, even with 20 or 30 items, too many to rank in one go.

What it measures, and what it does not

MaxDiff measures relative preference: how the items compare with each other. It does not say whether anyone wants any of them. If every item on the list is unappealing, one still comes first. Keep this in mind when reading reach or shares from a MaxDiff study: they are within the list.

Compared with conjoint

Conjoint (chapters 2 and 3) describes products by attributes and prices and asks for a choice between whole products, so it gives willingness to pay and a market simulator. MaxDiff has no attributes and no prices: each item is a thing on its own. Use it to rank a long list, often to shortlist what goes into a conjoint study later.

Try it in the tutorial

Next: Your MaxDiff data

Tutorial links open the example project in the app; it is free and needs a free account.