# Gauge 3 · Comfort Index

**Published research summary · November 5, 2026.** Research status: completed revised exploratory study. The original concentration-restricted test is unavailable. Fragrantica window: 1980–2025; Parfumo comparison window: 1980–2023.

Does economic stress precede a sweeter or more familiar fragrance market? Comfort Index compares annual cataloged launches with changes in consumer sentiment. It tracks gourmand accords, fresh accords, and flankers: extensions of existing fragrance lines. This describes recorded industry output. It does not measure purchases, preferences, or comfort experienced by consumers.

The original design required specific fragrance concentrations. The downloaded Fragrantica v3 source has no concentration field, so that test cannot be evaluated. The delivered study uses all recorded concentrations under a revised exploratory plan frozen before outcome analysis. Annual changes in launch shares are related to falling consumer sentiment, with fixed lag sums of one to three years for gourmand and fresh shares and one to two years for flankers. Those lags are assumptions rather than measured development timelines.

**None of the three directional lag-sum tests passes the Benjamini–Hochberg 5% threshold.** A one-year, one-point stress impulse has an estimated cumulative gourmand-share response of +0.041 percentage points, 95% HAC interval −0.092 to +0.175, adjusted p=0.40185. The flanker estimate is −0.048 points, interval −0.215 to +0.119, adjusted p=0.71688. The fresh estimate is −0.128 points, interval −0.358 to +0.102, adjusted p=0.39891. These findings do not establish the proposed lagged comfort pattern.

A separate recession-event flanker result has an unadjusted p-value near 0.021. It is secondary and particularly exposed to classification error. AI-assisted labels provide weak validation: 90 of 100 holdout rows remained unsure, leaving only ten scored cases and flanker precision of 42.9%. They are not independent human ground truth. Parfumo replication also faces name encoding that undermines the frozen flanker-grouping rule; its selected-concentration gourmand and fresh effects have opposite signs to Fragrantica. Different source taxonomies and substantial missing fields further limit comparison.

The latest descriptive comfort reading is **+3.6 percentage points above its fitted trend for 2025**, classified as steady. A high latest reading can coexist with inconclusive historical tests. Community catalogs are surviving snapshots with enthusiast and coverage biases, rather than complete annual release censuses. Optional Google Trends demand validation is unavailable. The study therefore offers a reproducible exploratory account and a clear record of what the available data cannot establish.

## Chart choices

1. **Lead:** `charts/gauge03/g03_gourmand_share.svg` — label the all-concentration exploratory scope and final launch year.
2. `charts/gauge03/g03_replication.svg` — retain disagreement and unavailable comparisons.
3. `charts/gauge03/g03_flanker_share.svg` — use only with a nearby classification warning; avoid making it the lead result.

## Source and verification note

Fragrantica public Kaggle dataset `ledecanteur/fragrantica-perfumes`, v3, updated September 20, 2026; Parfumo via TidyTuesday’s December 10, 2024 release, upstream scrape date unavailable; FRED consumer sentiment, unemployment, and recession series. Catalog acquisition is recorded October 5–6, 2026 with exact UTC timestamps. `reports/gauge03-results.md` retains all tests, raw and adjusted p-values, robustness, and source notes. Cached deterministic replay is documented in `reports/gauge03-verification.md`. Raw community catalogs and row-level data remain outside the publication package.
