# Independent review record

Example project: the café, people and data are invented.

The reviewer is an approved survey specialist who did not take part in the analysis.

## Version 1: correction requested

- **Reproduction:** ran `analysis.py` on the original file in a clean Python 3.12 environment. Tables 1–3 matched the report exactly.
- **Data changes:** compared 20 rows of the checked file with the original. Every difference matches `changes.md`.
- **Scope:** all three agreed questions are answered, and the limitations are stated.
- **Correction requested:** the weekend gap could be read as a fact about weekends alone, but weekend waits were longer. The reviewer asked for the median wait by day and for the share satisfied within each waiting-time band, so readers can see how much of the gap waiting explains.

## Version 2: cleared

- The median wait by day was added to Table 1, and the within-band comparison was added under Table 2.
- Reran `analysis.py`: all tables matched.
- **Methods:** proportions with a confidence interval and a rank correlation suit these data. The association is not described as a cause. Small groups are flagged.
- **Decision:** cleared for the customer.