Definition
A normalized measure showing how far a current value is from its rolling baseline. Used as one of several display formats; the DLI composite itself uses percentile-rank aggregation.
Z-scores appear on individual indicator detail pages as a familiar magnitude framing. Per-indicator z = (Value − Median) / (1.4826 × MAD) over a rolling 10-year baseline, winsorized to [-4, +4]. NOTE: the DLI composite itself does NOT linearly average z-scores — instead it converts each indicator to its tightness percentile in [0, 1] within a rolling 5-year window, then aggregates the four group sub-indices via the CISS-style quadratic form √(s'·(W∘Σ)·s). Bounded percentiles are required by the quadratic aggregation; raw z-scores would not behave numerically. The z-score readout on indicator pages is a UI convenience, not the DLI input. Z above ±1.0 is notable; above ±2.0 is extreme.