Multi-Vari Chart: An Intuitive Anatomy of Variation
The Principle of the Multi-Vari Chart
A multi-vari chart groups data by classification dimensions (position, time, batch, operator) and connects each group's points into a vertical line, showing visually whether variation comes mainly from within groups (short-cycle) or between groups (long-cycle). This helps locate the dominant source of variation and points improvement efforts in the right direction, making it a classic graphical tool for stratified variation analysis. It decomposes a complex variation structure into visible layers and is the tool of choice early in a quality-improvement project.
When to Use It
Use a multi-vari chart early in quality improvement to locate the source of variation; to judge whether machine-to-machine, batch-to-batch or time drift dominates; as a preliminary exploration before Gage R&R or ANOVA; and to show the variation structure to management. The chart makes the variation source immediately visible and reduces blind analysis; it is a common tool in the Analyze phase of DMAIC. Looking at the chart first is faster than jumping straight into ANOVA, and it also provides a basis for gage studies and sampling-plan design.
How to Read the Results
The length of each group's line represents within-group variation, while the differences between group means represent between-group variation. Large between-group differences mean the key variation lies in the classification factor, so improvement should target that factor; large within-group differences mean short-cycle fluctuation dominates. The tool draws multi-dimensional groupings automatically and supports AI interpretation to help non-statisticians understand the variation structure. Combine the chart with business judgment on whether the between-group differences are worth improving, rather than focusing only on statistical significance.
Precautions When Using It
Design the classification dimensions around the actual suspected sources of variation (time, equipment, people, raw material). Keep group sample sizes balanced for fair comparison. The multi-vari chart is an exploratory tool, so confirm conclusions with ANOVA before finalizing them; when there are many dimensions, draw several charts focusing on one or two dimensions at a time to avoid overload. Dimensions should cover all practically possible sources of variation.