- The mean = total ÷ count: the water level of the flattened waves.
- Weighted: Σ(value × count) ÷ TOTAL count — every copy weighs, and the mean leans toward the crowd (the directional check).
- THE fault: dividing by the distinct values — the divisor is the headcount.
Summarising a Series in Two Numbers
A data series is a crowd; this year hires its first two SUMMARIES. The MEAN levels the values: total ÷ count — grades 8, 12, 13: mean 11 — the image to keep: the WATER LEVEL once the waves flatten (pour all the values into one tank, share equally). When values repeat, the WEIGHTED mean lets each value carry its COUNT: three pupils at 8 and one at 15 — (3 × 8 + 1 × 15) ÷ 4 = 9.75 — dividing by the TOTAL COUNT (4 pupils), never by the number of distinct values (2): each copy weighs, and the mean tilts toward the crowded value. The FREQUENCY re-rates a count against the total: 6 left-handers in a class of 30 — frequency 6/30 = 0.2 = 20% — the percentage lesson in statistical service, and the tool that lets DIFFERENT-SIZED groups compare (15% of one school against 20% of another: frequencies compare, raw counts lie). Closing lucidity: the summary HIDES as much as it shows — two classes can share a mean of 11 with calm seas in one and storms in the other: same level, other tempests.
The Two Summaries
| The Tool | The Formula | The Example | The Watchpoint |
|---|---|---|---|
| The mean | Total of values ÷ count | (8 + 12 + 13) ÷ 3 = 11 | The water level of the flattened waves |
| The weighted mean | Σ(value × count) ÷ TOTAL count | (3×8 + 1×15) ÷ 4 = 9.75 | Divide by 4 pupils, never by 2 values |
| The frequency | Count ÷ total (then ×100 for %) | 6/30 = 20% | Re-rating to compare unequal groups |
| The frequency check | All frequencies sum to 1 (100%) | 20% + 80% = 100% ✓ | The built-in quality control |
| The forbidden mean of means | Means never average without their counts | Two class means need the class sizes | The escort rule |
| What the summary hides | Same mean, other storms | 11-11-11 and 2-11-20: both mean 11 | The lucidity clause |
Three pupils scored 8 and one scored 15. The class mean?
The weighted mean: every copy weighs — 24 + 15 = 39 points shared among FOUR pupils. Dividing by 2 counts the distinct values and forgets the crowd at 8: the mean must tilt toward the crowded value, and 9.75 duly sits near it.
Frequencies That Compare, Escorts That Protect
Frequencies at work across unequal groups: 12 walkers among the 60 pupils of school A (20%), 11 among the 55 of school B (20%) — identical habits, though the raw counts differed: the frequency REMOUNTS everything on one hundred, and only then do groups of different sizes talk to each other. The check rides along: each school's frequencies (walkers + others) must sum to 100% — a hole means a lost pupil.The mean of means, forbidden without escort: class A (20 pupils) averages 10, class B (30 pupils) averages 14 — the year mean is NOT 12: it is (20×10 + 30×14) ÷ 50 = 12.4 — the bigger class weighs more. Means travel WITH their counts (the escort rule): averaging naked means treats a class of 30 like a class of 20, and the weighted machinery exists precisely to stop that.
The Distinct-Values Divisor — and the Naked Mean of Means

Class A (20 pupils) averages 10; class B (30 pupils) averages 14. The overall mean?
The escort rule: means travel with their counts — 200 + 420 = 620 points among 50 pupils. The naked 12 treats both classes as equal crowds: the mean of means is the distinct-values fault, one storey up.
Case File: The Harvest Review










- The mean = total ÷ count: the water level of the flattened waves.
- Weighted: Σ(value × count) ÷ TOTAL count — every copy weighs, and the mean leans toward the crowd (the directional check).
- THE fault: dividing by the distinct values — the divisor is the headcount.
- Means travel with their counts (escort rule): the naked mean of means is forbidden; frequencies, not raw counts, compare unequal groups.
- Same mean, other storms: the summary hides — lucidity is part of the lesson.




