MaxDiff
Best-worst scores from counts or a logit, rescaled 0 to 100.
Copy as MarkdownEdit on GitHub
When to use it
Use MaxDiff when each person saw a set of items and picked the best and the worst. The score is how often an item wins, or a logit of those choices, rescaled from 0 to 100. A full choice-based conjoint, with attributes and levels, is conjoint.
Running it in Tensr
Options
Prop
Type
Reading the output
Counting reports how often each item was chosen best against how often it was shown. MNL reports a utility. Both can be rescaled from 0 to 100 so the top item reads as 100. Individual utilities give one score per item for each person.
Coming from SPSS
SPSS has no MaxDiff item. Tensr’s item is Analyze → Survey techniques → MaxDiff (counting / MNL), with a design export and an analyse step on the returned file.
Related
Conjoint is the same dialog with attributes and levels. Van Westendorp and Gabor-Granger price a single offer.