- The MEAN shares the total (dragged by one giant: 5000 no one earns) ; the MEDIAN is the sorted middle (resistant).
- SORT before the median — the raw middle is an accident of entry.
- Quartiles cut into four (Q2 = median) ; Q3 − Q1 (interquartile range) = spread of the middle half, immune to outliers.
Three Summaries That Disagree
One number rarely tells a data set's whole story, so this year gathers THREE summaries that see it differently. The MEAN levels (total shared equally) — but a single giant value drags it: wages 1500, 1500, 2000, 2000, 18 000 give a mean of 5000 that NO ONE earns. The MEDIAN is the MIDDLE of the ORDERED series (sort first!) — resistant to extremes, it reads 2000 here, the typical earner, unmoved by the outlier: mean 6350, median 1500 can describe the same data and tell opposite stories. The QUARTILES cut the sorted series into four: Q1 (a quarter below), Q2 = median, Q3 (three quarters below) — and Q3 − Q1 is the INTERQUARTILE RANGE, the spread of the middle half, immune to outliers. Weighted mean for report cards: (8×2 + 14×3)/5 — coefficients weight the grades, and dividing by the total weight (not the count) is the discipline. The signature error: the MEAN OF MEANS — averaging two class averages ignores their sizes; only the weighted mean (by headcount) is honest. And the ETENDUE (range) max − min: dispersion in one number, though it hears only the extremes.
The Three Summaries
| The Tool | What It Says | The Watchpoint | The Weakness |
|---|---|---|---|
| Mean | Total shared equally | Dragged by one giant | 5000 no one earns |
| Median | The MIDDLE (sort first!) | Resistant to extremes | Says nothing of spread |
| Quartiles | Cut into four | Q1, Q2 = median, Q3 | The middle-half spread |
| Interquartile range | Q3 − Q1 | Spread, immune to outliers | Ignores the tails |
| Weighted mean | Coefficients weight | Divide by total WEIGHT | (8×2 + 14×3)/5 |
| The trap | Mean of means | Ignores group sizes | Only weighted is honest |
Wages: 1500, 1500, 2000, 2000, 18 000. Which tells the typical earner?
The mean shares the total (outlier included): 5000, a wage no one earns. The MEDIAN sits at the middle of the sorted series: 2000, the typical earner, and doubling the outlier wouldn't move it. Tailed data (wages, prices) split the two wide — the gap itself is the diagnosis.
The Weighted Mean, the Quartile Cut
The weighted mean, report-card style: grades 8 (coefficient 2) and 14 (coefficient 3) — NOT (8 + 14)/2 = 11, but (8×2 + 14×3)/(2+3) = (16 + 42)/5 = 58/5 = 11.6. Divide by the TOTAL WEIGHT (5), not the count (2): the coefficients pull the average toward the heavier grade. The plain average ignores that the 14 counts more; the weighted mean respects the coefficients, which is the whole point of coefficients.The quartile cut: sort the series, then Q1 marks the first quarter, Q2 the middle (the median), Q3 the third quarter. The INTERQUARTILE RANGE Q3 − Q1 measures the spread of the MIDDLE HALF — and unlike the range (max − min), it ignores the extremes: a single outlier inflates the range but barely touches the interquartile range. Two dispersions, two sensitivities: the range hears the loudest voice, the interquartile range hears the crowd.
The Mean of Means — and the Unsorted Median

Grades 8 (coeff 2) and 14 (coeff 3). Weighted mean?
Weighted mean: (8×2 + 14×3)/(2+3) = (16 + 42)/5 = 11.6 — divide by the TOTAL WEIGHT (5), not the count. The 11 is the plain average (ignores coefficients: the 14 should count more, pulling above 11). Coefficients weight the grades; the denominator is their sum.
Case File: The Estate Wages










- The MEAN shares the total (dragged by one giant: 5000 no one earns) ; the MEDIAN is the sorted middle (resistant).
- SORT before the median — the raw middle is an accident of entry.
- Quartiles cut into four (Q2 = median) ; Q3 − Q1 (interquartile range) = spread of the middle half, immune to outliers.
- Weighted mean: (8×2 + 14×3)/5 — divide by the total WEIGHT, not the count.
- The trap: MEAN OF MEANS ignores group sizes — only the weighted mean (by headcount) is honest.




