# F4-dev-01 — LFS September 2026: headline figures and assertion check

## Question
From the provided extract of Statistics Canada, *The Daily*, "Labour Force Survey, September 2026" (released 2026-10-09), extract (a) the national seasonally adjusted unemployment rate and monthly employment change for September 2026, with section locators, and (b) judge whether the release supports: "In September 2026, the unemployment rate among core-aged women (aged 25 to 54) rose 0.3 percentage points to 5.3%, and the unemployment rate among core-aged men also rose, by 0.2 percentage points, to 5.8%."

## Data and exclusions
Inputs: `data/sources/statcan-daily-lfs-2026-09.txt` (164 lines) and `data/README.md` (9 lines) (`n_input_rows`). The extract's SHA-256 matches the README (`source_sha256`). Nothing was excluded (`out/exclusions.csv`). Two minor input issues are recorded in `data_issues`: the README mentions tables ("RESEA", "London") that are not part of this extract, and chart/infographic titles appear as `##` headings. Neither affects the results.

## Method
`out/analysis.py` checks the source hash, links each line to the most recent `## ` heading, and pulls the figures out with regular expressions. The sign comes from the wording ("declined" gives a negative number). The script checks that every national statement of each figure agrees, and it prints the source lines it matched. For the assertion, the script parses the core-aged women and men sentences and compares direction, size and level with the assertion. No randomness is involved (`random_seed` = null).

## Results
- National unemployment rate, September 2026: **6.5%** (`unemployment_rate_sep2026`). Locator: section "Highlights" (`locators.unemployment_rate_sep2026`). The section "Unemployment rate increases to 6.5%" gives the same value (`baseline_consistency.alternate_locator_unemployment_rate`).
- Monthly employment change: **−68,000 persons** (`employment_change_sep2026_persons`). Locator: section "Highlights" (`locators.employment_change_sep2026_persons`). The section "Employment falls for second consecutive month" gives the same value (`baseline_consistency.alternate_locator_employment_change`).
- Assertion verdict: **not_supported** (`assertion_verdict`). Locator: section "Unemployment rate up for core-aged women" (`assertion_locator`). The women's part matches the release, but the release says the core-aged men's rate "fell 0.2 percentage points to 5.8%", which contradicts "also rose" (`assertion_reason`).
- Items assessed: 3 (`n_analysed`). Feasibility: complete (`feasibility`).

## Uncertainty
The values are printed point estimates, reported with no re-computation. The release itself does not give sampling uncertainty. The Note to readers says only that the analysis focuses on changes that are significant at the 68% confidence level. No extraction uncertainty remains: each national figure is stated twice, and the statements agree (`baseline_consistency`).

## Limitations and open issues
- The only evidence used is the provided extract. Its data tables are not reproduced, so nothing was checked against them.
- Each figure is printed in two sections. The locator reports "Highlights"; the other valid locators are recorded in `baseline_consistency`.
- The README's mention of tables from other documents looks like a template artifact. It was ignored (`data_issues`).