Analyze MaxDiff results

MaxDiff (best-worst scaling) gives you a clear ranking of how important individual items really are to respondents – without the agreement bias of classic importance scales. DataLion analyzes MaxDiff data with three methods on the R engine.

Why MaxDiff

With classic scales, (almost) everything often ends up being “very important”. MaxDiff forces respondents into real trade-offs: for each set they choose the most important and the least important item. The result is a scale-free ranking that can be compared cleanly across countries and target groups.

The three analysis methods

Choose the method according to your requirements:

• Count analysis (best−worst) – a fast, intuitive ranking from the difference between best and worst mentions.
• Aggregate logit – a robust overall estimate across all respondents.
• Random-parameter logit – also takes the heterogeneity between respondents into account.

All three run in the background on the R engine – at the click of a button, with no R code and no data export.

In practice

MaxDiff answers prioritization questions: which features, messages, benefits or claims are the most important – and which ones can you leave out? The ranking can be broken down by segment and analyzed further with cross tables.

Tip: This article describes the analysis. You will find the complete use case – from setting up the study through the field phase to the analysis – on the Conjoint & MaxDiff solution page. In addition, the driver analysis explains which factors account for a result.