# Chapter 13 · Draft answer checks

Constructed examples; subject and author review pending.

1. January: 35 orders, five late, 14.2857%. February: 65, sixteen, 24.6154%.
2. The erroneous $16/50=32\%$ is 7.3846 percentage points above the correct February rate. Reconcile the monthly denominator to batches B04–B06.
3. A useful caption names both months, six constructed batches, numerators/denominators, rates and source-coverage assumption.
4. Identical point rates conceal different denominators and uncertainty. Show counts and appropriate intervals if population inference is intended.
5. The uncertainty of a difference depends on covariance and design; display the estimated contrast with an appropriate interval.
6. Independent scales show within-week rank. A fixed scale supports cross-week magnitude comparison.
7. A gap says no observation; an interpolated point is a labelled model estimate; zero claims an observed zero and is wrong for a missing week.
8. Use distinguishable dashes, symbols, direct labels and a grayscale export check at print size.
9. The observed overall rate fell from 83% to 43%, while an equal-mix descriptive comparison rose from 55% to 62.5%. Neither non-randomised comparison isolates a process effect.
10. Show the last validated timestamp and stale status. Alert the named source/report owner, with severity and affected outputs.
11. Restrict or revise the public artifact, preserve an access-controlled correction record, and tell authorised recipients what changed.
12. A review packet needs source identity, dictionary, extraction scope, transformations, test outputs, figure data, report text and limitations.
