The same 21 items, priced at twelve stores.
Nine years of receipts behind them.

1. Totals use effective prices — the sale price when lower. 18 sale rows were recorded across the tracker (5 with zero savings or excluded); 13 real discounts totaling $21.68. Met Fresh alone had $7.83 of sales, worth four rank places.
2. 7 unavailable store-item rows clip both tails of the full-basket ranking; pair the printed 1.21x spread with the fairer common-18 1.29x.
3. Mr. Lemon’s produce carries 3 unit-mismatch comparability flags; Whole Foods’ basket is 81% substitutions (organic/store-brand) — cheapest-at-store, not like-for-like.
4. Single visit per store, collected 2026-07-13/14/17 by YJC student journalists. 67 of 252 rows lack recorded dates (Brooklyn’s Natural, Food Story, Whole Foods).

Reproduce: python3 analysis/scratch2_compute.py

1. Single visit per store over four weeks, different staffers, some items on sale at visit time.
2. Trader Joe’s is in Rego Park — a drive or bike ride, not a neighborhood walk.
3. Whole Foods (Williamsburg, $61.29) was only 27.6% above Trader Joe’s — and $13.54 cheaper than Food Story.

1. Three basket definitions (17-item totals; 6-item budget basket; common-18) — rank only, never a cross-year dollar slope.
2. City Fresh’s 2021 point carries a $18.34–$21.34 sensitivity band from two article-vs-Tableau conflicts; its direction is robust.
3. Mr. Fruit (2021) → Mr. Lemon (2026) is an approximate entity mapping and is excluded from this cohort.

1. Normalization: 2021 = cheapest-available dozen per store; 2026 = the one product each store recorded — constructions differ, so rows are compared to their own year’s BLS line, never to each other item-for-item.
2. Met Fresh’s 2026 18-pack is adjusted ×12/18 to per-dozen and flagged MISMATCH; Billy’s $1.67 is a 3-for-$5 sale (whisker to its $2.79 regular price).
3. 7 of the 9 prices ≥$4.99 are specialty SKUs (cage-free/free-range/organic/farm-fresh — mostly DIFFERENT or UNCLEAR flags) — the band shift mixes real inflation with spec drift.

1. 2021 recorded cheapest-available floors (conservative vs. an average); 2026 recorded one product per store, so 2026 medians overstate a cheapest-available premium — floors are the honest bound.
2. BLS values are national averages over standardized specifications, not NYC-metro item averages; they never share a nominal axis with survey floors without full labeling.
3. Milk is excluded from the 2026 comparison (organic vs. conventional — not comparable).

1. Normalization: 2026 recorded one product per store, so the floor (cheapest observed shelf price) is the honest conservative bound; medians overstate a cheapest-available premium — the reverse of 2021’s construction, and the reason both ends of the range are drawn.
2. Milk is excluded (2026 observations are organic-only — not comparable to the BLS conventional-gallon average); chicken cuts differ (thin-sliced premium at some stores).

1. Normalization: single-brand (Cafe Bustelo 10 oz) regular shelf prices — never compared to 2021’s any-brand cheapest floors ($1.67), which measure a different thing.
2. The 2021 $3.99 pack size is assumed-standard 10 oz (unverified in the source article); Food Story’s 2026 size is flagged UNCLEAR.
3. Hollow dots are visit-day sale prices (Met Fresh $6.99, City Fresh $8.99); the median uses regular prices.

1. The CPI ruler is a benchmark, not observed Bushwick prices — the surveys provide the observations; restated totals are the break-even line, not measured 2026 baskets.
2. Factors: 2017→2021 ×1.1298, 2021→2026 ×1.1898, 2017→2026 ×1.3442 (CUURS12ASAF11).

1. Normalization: each ratio divides a store’s 2026 price for the 15 items overlapping the 2017 basket by its FULL 17-item 2017 total — a deliberate lower bound, so “outpaced inflation” claims are conservative.
2. Hana and Food Story bars are hatched: their 2026 overlap subsets miss 1–2 items; they stay ≥14% below break-even under maximally generous imputation.