# F1-conf-02 — Balanced-panel mean life expectancy, 2000–2022

## Question
How did life expectancy at birth (WDI `SP.DYN.LE00.IN`) change between 2000 and 2022 across a fixed set of countries? The estimand is the unweighted mean across the countries with a recorded value in every year 2000–2022 (balanced panel), each country counted once.

## Data and exclusions
Input: `data/life_expectancy_panel.csv`, 439 data rows (`n_input_rows`), 20 distinct iso3 codes (`n_countries_total` = 20). Data problems detected (`data_issues`):
- Türkiye (TUR) appears twice for every year, once as "Turkey" and once as "Turkiye" (an artefact of combining two extractions), and both copies have identical values. One row per (iso3, year) was kept.
- Switzerland (CHE) has an empty value for 2007. It is treated as missing, not imputed, so CHE is excluded from the balanced set.
- Bangladesh, Ethiopia, India and Nigeria (BGD, ETH, IND, NGA) have no rows for 2000–2010, so all four are excluded from the balanced set.

Exclusion ledger (`out/exclusions.csv`): 439 rows read → 439 rows within the 2000–2022 window → 416 after removing duplicate (iso3, year) rows (−23) → 415 after dropping the empty cell (−1) → 345 after keeping only the balanced countries (−70). `n_analysed` = 345 = 15 countries × 23 years. `n_countries_balanced` = 15 and `n_countries_incomplete` = 5. The balanced set (`balanced_countries`) is AUS, AUT, BEL, CAN, DEU, ESP, FRA, GBR, ITA, JPN, KOR, NLD, SWE, TUR, USA.

## Method
Countries are identified by iso3. Duplicate (iso3, year) rows were checked for conflicting values: there were none, and the script stops with an error if any are found. No value was imputed or interpolated. For each year, the analysis takes the simple mean over the balanced set (`balanced_means_by_year`). The change is the 2022 balanced mean minus the 2000 balanced mean. The baseline is the mean over every distinct country with a recorded 2022 value, each counted once (20 countries, `n_countries_2022_available`). The analysis is deterministic and uses no random numbers (`random_seed` = null).

## Results
- Balanced mean in 2000: **78.060 years** (`mean_2000_balanced`).
- Balanced mean in 2022: **81.553 years** (`mean_2022_balanced`).
- Change from 2000 to 2022: **+3.493 years** (`change_2000_2022_balanced`).
- Baseline, mean over all countries with a 2022 value: **78.692 years** (`mean_2022_all_available`). It is 2.86 years below the balanced 2022 mean (`mean_2022_balanced` − `mean_2022_all_available`) because it includes the four lower-life-expectancy countries that have no data before 2011 (BGD, ETH, IND, NGA), plus CHE. The naive mean therefore describes a different set of countries and should not be compared with the 2000 balanced mean.
- Task feasibility: `feasibility` = "complete".

## Uncertainty
These figures are descriptive means of a fixed, non-random set of countries, so no sampling-based confidence interval was computed. The WDI values are themselves modelled estimates, and their own uncertainty is not reported in the input. Values are rounded to 3 decimals, so rounding error in the means is negligible (< 0.001 years).

## Limitations and open issues
- Because the panel is balanced, 15 mostly high-income countries remain. The result does not represent the 20 countries in the file, and it does not represent the world. The means are unweighted rather than population-weighted.
- One missing value (CHE 2007) is enough to exclude Switzerland, as the estimand requires.
- The values come from a single WDI release and may change in later revisions.