Population pyramids: read structure and check ratios
A population pyramid shows how a population is distributed across age groups and recorded sex categories at a stated time. Its shape helps reveal structure, but it does not by itself explain births, deaths, migration or employment. First check the bars and denominator; only then interpret the pattern.
A and B below are fictional populations observed on the invented date 1 January 2026. The figures were created for calculation practice and contain no actual people’s records. They are not forecasts or descriptions of Kenya, Sweden or any other country. “Male” and “female” are the two supplied statistical categories in these tables; the exercise does not make claims about individuals’ identities.
| Age in years | Male | Female | Combined |
|---|---|---|---|
| 0–4 | 610 | 590 | 1200 |
| 5–9 | 550 | 550 | 1100 |
| 10–14 | 510 | 490 | 1000 |
| 15–19 | 440 | 460 | 900 |
| 20–24 | 410 | 390 | 800 |
| 25–29 | 340 | 360 | 700 |
| 30–34 | 320 | 330 | 650 |
| 35–39 | 300 | 300 | 600 |
| 40–44 | 270 | 280 | 550 |
| 45–49 | 250 | 250 | 500 |
| 50–54 | 220 | 230 | 450 |
| 55–59 | 190 | 210 | 400 |
| 60–64 | 170 | 180 | 350 |
| 65–69 | 140 | 160 | 300 |
| 70–74 | 110 | 140 | 250 |
| 75–79 | 60 | 90 | 150 |
| 80+ | 40 | 60 | 100 |
| Total | 4930 | 5070 | 10000 |
| Age in years | Male | Female | Combined |
|---|---|---|---|
| 0–4 | 610 | 590 | 1200 |
| 5–9 | 600 | 600 | 1200 |
| 10–14 | 710 | 690 | 1400 |
| 15–19 | 820 | 780 | 1600 |
| 20–24 | 1220 | 980 | 2200 |
| 25–29 | 1150 | 1050 | 2200 |
| 30–34 | 740 | 760 | 1500 |
| 35–39 | 690 | 710 | 1400 |
| 40–44 | 590 | 610 | 1200 |
| 45–49 | 540 | 560 | 1100 |
| 50–54 | 480 | 520 | 1000 |
| 55–59 | 430 | 470 | 900 |
| 60–64 | 380 | 420 | 800 |
| 65–69 | 360 | 440 | 800 |
| 70–74 | 290 | 360 | 650 |
| 75–79 | 210 | 290 | 500 |
| 80+ | 140 | 210 | 350 |
| Total | 9960 | 10040 | 20000 |
Construct both sides on the same basis
Put the youngest age group at the bottom and the oldest at the top. Use equal bar heights for each five-year band and identical horizontal magnitude scales on the two sides. The open-ended 80+ group is visibly labelled; it covers an unknown number of years, unlike a five-year band.
For this comparison, every bar is the count in that age–sex cell divided by the total population of its own area, multiplied by 100. Thus A’s denominator is 10,000 on both sides and B’s is 20,000 on both sides. Do not divide male counts by the male subtotal while dividing female counts by the female subtotal and then present the result as percentages of the whole population.
For example, A’s male 0–4 bar is 610 ÷ 10,000 × 100 = 6.1%; its female partner is 590 ÷ 10,000 × 100 = 5.9%. Together they represent 12% of A. Draw each from its own zero line beside the central age labels. Bars for the two sexes together sum to 100% across all ages.
Worked case 1: describe the structure with values
A’s combined 0–4 group contains 1,200 people, compared with 800 in 20–24 and 100 in 80+. Its youngest three groups total 1,200 + 1,100 + 1,000 = 3,300 people, or 33%. This supports a broad younger-age share in the supplied snapshot.
B’s combined 20–24 and 25–29 groups each contain 2,200 people, exceeding its 1,200 in 0–4. Calling this a young-adult bulge is a description supported by the counts. Calling it proof of a recent rise in births would be a causal claim requiring other records: current young adults were born many years before this snapshot.
The U.S. Census Bureau’s population-pyramid explanation discusses age composition and migration. Its historical examples are background for the method; the numbers here are entirely independent.
Worked case 2: compare proportions rather than unlike totals
Both A and B contain 1,200 people aged 0–4. Those equal counts represent 12% of A but only 6% of B, because B’s total is twice as large. A count-based chart with different axis ranges could conceal that distinction. The displayed common percentage scale makes it explicit.
B has 3,800 people aged 0–14, more than A’s 3,300, but B’s share is 3,800 ÷ 20,000 × 100 = 19%, below A’s 33%. State whether “more young people” means number or proportion. The difference in shares is 33 − 19 = 14 percentage points, not simply “14% more”.
The open-ended 80+ group also needs care. Its total can be compared as a share of population, but its bar cannot be interpreted as the number per year of age on the same basis as a five-year band. A detailed single-year distribution would be needed for that comparison.
Worked case 3: calculate an explicitly defined dependency ratio
Use this age convention: (ages 0–14 plus ages 65+) ÷ ages 15–64 × 100. For A, the groups are 3,300, 800 and 5,900 respectively. Its ratio is (3,300 + 800) ÷ 5,900 × 100 = 69.5 per 100, rounded to one decimal place.
For B, the corresponding totals are 3,800, 2,300 and 13,900. Its ratio is 6,100 ÷ 13,900 × 100 = 43.9 per 100. Check that numerator plus denominator equals the whole population: 4,100 + 5,900 = 10,000 and 6,100 + 13,900 = 20,000.
This is an age-defined ratio. It does not count actual workers or people receiving financial support. Some people aged 15–64 do not have paid work, and some outside that range do. The Census Bureau’s dependency-ratio definition supplies this age convention; a different convention would require relabelling and recalculation.
Test a cause with a supplementary record
A fictional migration card states that 1,200 of B’s current 20–29-year-olds arrived from elsewhere during the previous five years, compared with 100 arrivals among its current 0–9 group. This record supports migration as a contributor to the young-adult bulge. It does not establish that migration explains the entire bulge: earlier births, deaths and departures also affect the snapshot.
The card concerns current age groups, not the ages of arrival. Do not subtract arrivals from a current band and call the result its birth cohort size; people age between dates and may move or die. A proper historical explanation needs records with compatible periods, age definitions and populations.
Three solved practice tasks
- Check a bar. B’s 20–24 male count is 1,220. Its whole-population percentage is 1,220 ÷ 20,000 × 100 = 6.1%. It is drawn leftward from zero, with a positive magnitude.
- Repair a denominator. A learner computes A’s dependency measure as 4,100 ÷ 10,000 × 100 = 41%. That is the share in the specified younger-and-older categories, not the requested ratio. Divide by 5,900 aged 15–64 to obtain 69.5 per 100.
- Handle missing data. A new table omits the female 65–69 cell and supplies no total allowing it to be recovered. Can its exact dependency ratio be completed by entering zero? No. Missing is not zero. Report that the numerator is incomplete and obtain the missing count before calculating the exact ratio.
Before interpreting any pyramid, check the date, units, age boundaries, denominator and missing-data symbols. Then separate description from explanation. A well-labelled limited conclusion is more useful than a confident story that the table cannot establish.
Read settlement growth through time and across space
The pyramids above compare age structure at one date. Growth needs comparable observations at more than one date. Luma is a fictional settlement with a fixed 12 km² reporting boundary in every record below. Its dates, residents, roads and land-use changes are invented, not a census, current land-rights record or forecast. It is separate from populations A and B above.
| Date | Residents | Built-up cells | Built-up area |
|---|---|---|---|
| 1 January 2020 | 2,400 | 2 | 2 km² |
| 1 January 2025 | 3,600 | 5 | 5 km² |
The original map is a four-column by three-row grid. Each cell is exactly 1 km by 1 km, hence 1 km². North is up; A–D run west to east and rows 1–3 north to south. For this deliberately simplified classification, a filled cell is wholly built-up and an open cell wholly other land. Cell codes are local addresses in the model, not six-figure map references. No building counts, individual homes or access rights can be inferred.
| Row / column | A | B | C | D |
|---|---|---|---|---|
| 1 (north) | · | · | · | · |
| 2 | B | B | · | · |
| 3 (south) | · | · | · | · |
| Row / column | A | B | C | D |
|---|---|---|---|---|
| 1 (north) | · | · | · | · |
| 2 | B | B | B | B |
| 3 (south) | · | B | · | · |
Equivalent cell lists: 2020 = A2 and B2; 2025 = A2, B2, C2, D2 and B3. A supplied chronology records a new east–west road along row 2 opening in 2022. This chronology is supplementary fictional evidence; the letter B in a cell means built-up, not a road symbol.
Worked interpretation: population and footprint change differently
Population increases by 3,600 − 2,400 = 1,200 residents, or 1,200/2,400 × 100 = 50% over five years. Whole-boundary density changes from 2,400/12 = 200 to 3,600/12 = 300 residents/km². These are average densities over the same reporting area, not the number living in every individual cell. The calculation must not switch silently from the 12 km² boundary to the built-up area.
The mapped footprint increases by 5 − 2 = 3 km², or 150% of the initial 2 km². New built-up cells occur eastward at C2 and D2 and southward at B3. Thus the area expanded faster proportionally than the resident count in this model. Neither map reveals building height, vacant rooms or average household size; do not invent a reason for the different growth rates from area alone.
Worked explanation: separate a population balance from a spatial influence
A complete model register for the interval gives 500 births, 150 deaths, 1,100 arrivals and 250 departures. Natural increase is 500 − 150 = 350; net migration is 1,100 − 250 = 850. Together they add 1,200, matching the change between the two population totals. Net migration contributes 850/1,200 ≈ 70.8% of the net increase. The denominator here is the change, not the 2025 total; arrivals alone are not net migration.
The new road predates the 2025 snapshot, and two new built-up cells follow its east–west alignment. Better connection is a plausible influence on development there. Yet the record does not identify when each building was constructed, why residents moved, jobs, planning permissions or land transactions. Chronology plus spatial association strengthens a hypothesis; it does not isolate road building as the only cause. For practice identifying static arrangements without a growth claim, the prepared Settlement Maps module in the Form 3 notes owner supplies the complete Q285 maps and answers. Verify that intended module destination before publication.
Independent next-snapshot check
A fictional 1 January 2030 record uses the same 12 km² boundary and conventions. It reports 4,000 residents, six built-up cells, and a complete 2025–2030 balance of 450 births, 200 deaths, 600 arrivals and 450 departures. Calculate population growth and density, verify the balance, compare the population and footprint percentage changes, and state whether the road caused all growth. Suggested classroom allocation: 7 marks.
Explained answer
The increase is 400 residents (1), or 400/3,600 × 100 ≈ 11.1% (1). Density is 4,000/12 ≈ 333.3 residents/km² (1). Natural increase is 250 and net migration 150; their 400 total matches the change (1). Built-up area grows from 5 to 6 km², a 1/5 × 100 = 20% rise (1), larger than population’s 11.1% proportional rise (1). The balance accounts for people but does not establish their motives; the road’s full causal effect remains unmeasured (1). These five-year percentages are not annual rates.
Frequently asked questions
Why are male bars drawn to the left?
It is a display convention that makes the two age–sex distributions easy to compare. The counts remain positive.
Which total is the percentage denominator?
In this lesson, each cell is divided by its own area’s whole population: 10,000 for A or 20,000 for B, on both sides.
Can equal counts give different bar percentages?
Yes. The same 1,200 people are 12% of A and 6% of B because their population totals differ.
Does a young-adult bulge prove a high birth rate?
No. Birth history, survival and migration can contribute. The supplied migration card supports one contribution without isolating every cause.
Does age dependency count actual unemployment?
No. It compares stated age categories and does not measure paid work, unemployment or actual financial support.
How is 80+ different from the other bands?
It has no upper age limit. It can be compared as a population share, but it is not a five-year interval.
May I treat a missing cell as zero?
No. Zero is a known count. Missing data must be recovered from valid information or reported as a limitation.
For other materials, see the Geography notes resource guide. Some linked resources may require membership; this lesson does not verify their contents or access.
WhatsApp Us Now