MaxDiff Analysis is shorthand for Maximum Differential Analysis, is also known as ‘Best-Worst scaling,’ where survey respondents choose the best value and worst value from a set of attributes (and the attributes repeat in intentional permutations based on the logit model). It assumes that respondents evaluate all possible pairs of items within the displayed set and choose the pair that reflects the maximum difference in preference or importance.