# Methods

## Research question

Where does Germany’s overall female majority disappear by age and Land, when does each Land move durably above parity, and which observable regional patterns are most consistent with the imbalance at ages 18–29?

This is a descriptive analysis. It reports official population stocks, transparent arithmetic decompositions and exploratory cross-Land associations. It does not estimate causal effects.

## Source population

Population data come from Statistisches Bundesamt (Destatis) GENESIS table 12411-0013 for 31 December 2025. The table reports the updated population stock by Land, sex and single year of age. Every extracted row carries the GENESIS quality marker `e` (`endgültiger Wert`, or final value). The official release notes that this population update is based on the 2022 census and uses a two-category male/female representation for this analysis; people registered as diverse or without a sex entry are allocated in the source method and are not separately represented here.

Counts are summed before the ratio is calculated. The five broad groups used in the regional atlas are under 18, 18–29, 30–49, 50–64 and 65+.

For every geography and age group:

`women_per_1,000_men = female_population / male_population × 1,000`

Parity is 1,000. Values below parity indicate more men than women; values above parity indicate more women than men.

## Durable crossover

Single-year ratios can cross parity more than once. The first female-majority age may therefore treat a local fluctuation as a demographic transition.

The durable female-majority crossover is the first single year of age `a` for which the ratio exceeds 1,000 at age `a` and at every older age available in the table. The 90+ cell is the terminal open group. A five-year-window sensitivity check aggregates moving five-year windows before testing whether every later window remains above parity.

This durable crossover is a derived analytical definition, not an official Destatis indicator.

## Population pyramids

Single ages are aggregated into five-year groups from 0–4 through 85–89, plus an open 90+ group, for Germany and all 16 Länder. Male values are mirrored left and female values right. Each bar is divided by the geography’s total population. A shared ±5.2% domain lets all 17 profiles compare age-sex composition rather than absolute size.

The interactive comparator overlays Germany and any two selected Länder across all 19 cohorts. The complete small-multiple wall uses the same domain. Hover, tap and keyboard focus reveal the exact male and female shares and cohort-specific female-to-male ratio. The 20–24 and 25–29 cohorts are the closest available five-year representation of the 18–29 focus.

## Nationality decomposition

GENESIS table 12411-0014 is filtered to ages 18–29 and aggregated by Land, sex and nationality group (German or foreign). Separate female-to-male ratios are calculated for German nationals, foreign nationals and the combined population.

The displayed composition difference is:

`composition_difference = combined_ratio − German_national_ratio`

This is an arithmetic comparison of observed population stocks. It is not a causal estimate of migration. Nationality does not identify migration timing, origin, generation or reason for moving.

The Sankey-style chart conserves the observed 18–29 population stock as it partitions residents first by nationality and then by sex. Ribbon width represents residents at the reference date, not migration over time.

## Mechanism proxies

Three Land-level proxies are compared with the 2025 ratio at ages 18–29.

### Historical internal migration

Table 12711-0021 provides annual internal-migration balances by Land, nationality and sex. German-national male and female balances are summed from 2000 through 2025. The primary proxy is the cumulative female-minus-male balance divided by the Land’s 2025 total population and multiplied by 1,000.

This measure covers all ages and 26 calendar years. It is consistent with a historical mobility channel but is not cohort-specific.

### University-location intensity

Table 21311-0005 provides students by Land and sex for winter semester 2025/26. Student intensity is total students divided by the 2025 population aged 18–29, multiplied by 1,000. The student sex ratio is also retained.

Counts follow the higher-education institution’s location, not necessarily the student’s residence. Thüringen is influential because its institution-based count is exceptionally high. The evidence file therefore retains student associations with and without Thüringen.

### Employment

Table 12211-1001 provides population and employed people aged 20–29 by Land and sex in thousands. Sex-specific employment rates are employed people divided by population. The proxy is the female rate minus the male rate.

The table is a Mikrozensus estimate and the 2025 values are provisional. A broad employment rate does not measure occupation-specific opportunity, wages, vacancies, commuting or job quality.

## Association measure

Spearman’s rank correlation is calculated across the 16 Länder. It is used because the sample is small and the proxy scales differ. The scatterplots show every observation and use direct labels for selected or influential Länder. No association is interpreted as a causal effect.

Primary results:

- German internal-migration gender gap: ρ = 0.700
- All-nationality internal-migration gender gap: ρ = 0.571
- Student intensity: ρ = 0.112; without Thüringen, ρ = 0.286
- Student sex ratio: ρ = 0.206; without Thüringen, ρ = 0.393
- Female-minus-male employment-rate gap: ρ = −0.257

## Data lineage and checks

The official GENESIS flat-file extract for table 12411-0013 is preserved at `content/data/germany-young-women/raw/12411-0013_de_flat.zip`. The frozen, publication-ready evidence lives at `content/data/germany-young-women/intermediate/regional-evidence.json`. Its source receipt records byte lengths and SHA-256 checksums.

The generator checks that:

- the five national age-group counts equal the sum of all 16 Länder;
- every reported ratio reproduces from its male and female counts;
- the 18–29 ratios from the population and nationality tables agree;
- every Land has 19 five-year cohorts and Germany reconciles with the Länder;
- foreign-national composition lowers the combined ratio relative to the German-national ratio in all 16 Länder;
- every public headline remains equal to its frozen, reviewed value.

It then writes the website analysis JSON, a compact homepage extract, five exact-value CSV downloads, chart and design contracts, validation results and a checksum manifest. The generated hero is an atmospheric illustration only; its receipt states that it encodes no data.

## Limitations

The outcome is an ecological, cross-sectional population stock. Broad age groups combine multiple cohorts and migration histories. Regional predictors are correlated. The migration proxy is all-age; the university proxy is institution-based; the employment proxy is broad and survey-based. Occupation-specific opportunity, residence-based education, international migration flows by sex and age, urban–rural movement and individual cohort histories are not jointly observed.

The evidence supports regional levels, arithmetic decompositions and descriptive associations. It does not identify counterfactual ratios or the causal contribution of any one factor.

## Reproduction

From the repository root, run:

```bash
npm run germany-young-women:data
```

The generator verifies the frozen inputs before it writes any public artifact. A changed official release should be imported deliberately, with a new source receipt and a review of every affected claim; it should never be absorbed silently during a routine build.