Multi-Vari Chart

The multi-vari chart shows every individual measurement of a response broken down by several categorical factors at once: factor 1 sits on the X axis, factor 2 as a series (colour/symbol, connected by mean lines), factor 3 as a panel column, and factor 4 as a panel row. It is a purely graphical Analyze-phase tool — it computes no significance, it directs the eye to where the variation actually arises.

Seder's original framework (1950) distinguishes three kinds of variation: positional (within a single part, e.g. several measurement points on one workpiece), cyclical (part to part within a short window, e.g. cavity to cavity or batch to batch) and temporal (over longer spans, e.g. shift to shift, day to day). The order in which factors are chosen is meaningful: the same order also fixes which factor is nested within which in the accompanying variance components model.

The picture alone answers "where", not "how much" and not "whether significant". Quantifying the dominant source of variation — as variance, standard deviation and percentage share per source — is the job of the companion variance components table; the module deliberately keeps the two as separate, complementary views.