Pulse Documentation

name: op-overlay-chisq-row
kind: operator
category: OVERLAY
operator: OVERLAY_CHISQ_ROW
description: Per-row χ² goodness-of-fit test across the host crosstab's contingency table.
type: reference
applies_to: process, compose
examples_tags: [overlay, cross-tabulation, hypothesis-test]

Overlays decorate the host; no Response.Components.

Use when

Chi-square test per crosstab row: checks whether each row’s spread across the columns departs from the table’s overall column mix.

Questions it answers:

• Which regions have an answer mix unlike the overall mix?
• Which product lines get a different spread of complaint types?

Use something else:

• OVERLAY_CHISQ_COL when the groups are the columns rather than the rows.

Params

Scope (enum, required) — must be row. Ref (object, empty) — implicit-margin — leave empty. Level/Within must be 0.

• multiplicity — optional {method, family, alpha}; adds p_adjusted figures, raw p untouched (multiplicity-correction).

Host shape

MATRIX crosstab (Response.Crosstab.Matrix). Family: implicit-margin χ² (no Ref). Compatible with any crosstab regardless of cell aggregator; row-axis twin of the column χ².

Output

SERIES — OverlayLayer.Payload.Shape = "series". One SeriesEntry per row key carrying Summary.Statistic (row χ²), Summary.PValue, Summary.Parameters["df"] = cols - 1. Entries align element-for-element with host RowKeys. Layer Baseline unset.

Reading the output

• summary.statistic: In each series entry (one per crosstab row): chi-square for that row’s counts against the counts its row total would give at the table’s overall column shares. 0 means the row’s mix matches the overall mix; read the entry’s p_value rather than the raw value.
• summary.p_value: a p-value. Below alpha (0.05 unless the request sets another) the result is called significant; that is not the same as important, so read the effect size for how big it is.

Gotchas

• Weighted host (both kinds, any source): cells are Σw; under probability each row is scaled to its margin’s Kish n_eff (expected-low on the scaled row) — first-order Kish, not Rao-Scott. Each entry’s Parameters adds that row’s sum_weights (+ n_eff). Probability host with components disabled ⇒ PROCESSING_CONFIG (no n_eff to scale by; predict AND runtime).
• Reuses chiSquareSurvival — byte-equal p-values to every χ² test and overlay on the same contingency.
• Any expected < 5 in a row emits ONE PULSE_OVERLAY_EXPECTED_LOW per offending row.
• Absent host cell treated as observed count of 0.
• Scope MUST be row. Populated Ref arm → PULSE_OVERLAY_REF_INCOMPATIBLE_WITH_SHAPE.
• Buffered (inherent).

See

• Skills: overlay-system, crosstab-guide, op-overlay-chisq-col, op-test-chisq.