👉 The Cluster Tab is available to customers on the Analyze tier or higher.
The Cluster Tab uses AI to analyze your survey results and group participants into clusters that share the same preferences or priorities.
How The Cluster Tab Works
Clusters
The Cluster Tab groups participants who voted similarly on all ranking blocks in your survey. This is done using k-means clustering, which identifies patterns in the voting data. Each group, or "cluster," is assigned a number and displayed on a card within the Cluster Tab.
Z Score
After forming clusters, a second analysis is done to see which ranking options each cluster felt most strongly about. This is achieved by calculating the Z score, which measures how far the cluster's average score for an option is from the overall average (population mean).
The Z score formula is (μc−μp)/σ, where:
μc = the cluster's mean for a specific option
μp = the population mean for that option
σ = the population standard deviation
To simplify comparison, we convert this Z score into an "absolute Z score" by making it a positive number. This lets us easily rank options by how strongly the cluster's opinion differs from the overall average, regardless of whether it was higher or lower.
Opinions vs Characteristics
Opinions are the options you include in your ranking questions. Clustering uses k-means analysis to group participants based on how they voted on these opinions, putting people together in cluster groups who voted similarly.
Characteristics are additional data points you've collected about participants that are not used in the clustering process. Instead, they help you understand the types of people in each group — particularly if certain types are overrepresented in a cluster compared to the overall population.
-- -- --
How To Run Clustering Analysis On Your Survey Results
You don't need to know any data science or AI programming to run a clustering analysis on your survey results on OpinionX. Just go to the Cluster Tab and hit "Run Clustering". If you have specific criteria for your analysis, you can define this using the Advanced Configuration menu, but this is optional as you can also just run the analysis with the default configuration.
-- -- --
Using Clusters for Segmentation Analysis
When you run a clustering analysis on your survey, every participant will be assigned to a cluster. These clusters can be used throughout OpinionX's segmentation analysis features, from filtering the overview tab to comparing all clusters against each other in the Segment Tab's crosstab heatmaps. Just look for the "Clusters" data category when configuring your segmentation criteria.