# Pupils per teacher: comparability audit

Decision date: 2026-09-17. **COMPARABLE_WITH_CAVEATS**.

Retain the Iceland trend unchanged. Add two separate OECD/UOE comparisons, for primary and lower secondary education. Do not splice, average or calibrate these into the Iceland series. No Nordic average is calculated.

## Existing Iceland series

Statistics Iceland table [SKO02105](https://px.hagstofa.is/pxis/pxweb/is/Samfelag/Samfelag__skolamal__2_grunnskolastig__0_gsNemendur/SKO02105.px), frozen in `data/releases/iceland_education_context_2026-09-10/`, provides the published ratio of pupils to teacher full-time equivalents. It covers compulsory schools together (primary and lower secondary), not separate international levels. The denominator includes **department heads and teachers**, not all school employees; it is FTE, not a headcount of teachers. The numerator is the registered pupil count, not an independently reconstructed FTE-enrolment measure.

The source covers 1998–2024. Five-year-olds in grade 0 are included through 2016; the selected chart therefore shows 2017–2024. The table publishes ratios to one decimal. Its national year labels should not be equated mechanically with OECD school-year labels. Source PX notes and API metadata are retained unchanged; no ratio is recalculated.

## Harmonised international source and year

Selected source: OECD, *Education at a Glance 2025*, chapter D2, Table D2.2, [StatLink workbook](https://stat.link/4am3u9), version 1 dated 9 September 2025. This is the latest published EAG edition on the audit date: the [OECD publication calendar](https://www.oecd.org/en/publications/forthcoming.html) schedules EAG 2026 for 29 September 2026. The latest common year in D2.2 for all five Nordics and the official OECD average is **2023**, referring to **school year 2022/23**. Earlier columns are 2013. We select 2023 for both panels.

The live SDMX availability request returned HTTP 403, as did a direct HTTP request for the chapter HTML; failures are recorded in the snapshot manifest. The chosen publication workbook and country-notes workbook were downloaded successfully. This is a dated publication comparison, not a claim to have verified the newest live-API revision. OECD's chapter definitions and forthcoming publication date were checked through the OECD website.

The [chapter's definitions and methodology](https://www.oecd.org/en/publications/2025/09/education-at-a-glance-2025_c58fc9ae/full-report/how-do-student-teacher-ratios-and-class-sizes-vary-across-education-levels-up-to-upper-secondary-education_76b87b21.html) define the ratio as FTE enrolment divided by teaching-staff FTE at the same level and matching institutions. Direct teaching includes classroom, special-education and individual/small-group teaching; this is not a measure of class size or all school staff. Teacher aides are a separate category. We do not assume the national department-head denominator is identical to OECD teaching-staff coverage.

D2.2 provides primary (ISCED 1), lower secondary (ISCED 2), and upper secondary (ISCED 3, including programme splits). We use only the first two. Institution coverage is public and private together for the selected Nordic observations. No selected 2023 cell carries a control code, different-year note, or public-only flag. Denmark's country note 1 concerns **2013**, not the selected 2023 observations. The OECD average is the published aggregate, not a recomputed average of the five Nordics; its contributing countries retain OECD's coverage and year exceptions.

## Country metadata and remaining caveats

The [2025 technical notes](https://doi.org/10.1787/fcfaf2d1-en) link to the [Part D metadata workbook](https://stat.link/57otaz), archived with the data. Its STUD and INST sheets contain no primary/lower-secondary control-code rows for these five countries. METADATA records differing collection dates within school year 2022/23: Denmark 1 October 2022; Finland 31 December 2022; Iceland 15 October 2022 for both selected levels; Norway 1 October 2022; Sweden 15 October 2022. These are not five synchronized calendar-year censuses.

Iceland's general personnel metadata says employees are counted in their main job at ISCED 0–2, work above one full-time job can count above 1.0 FTE, and contractors are excluded. It gives no reconciliation to SKO02105 department heads. Denmark's free-text methodology cell still mentions October 2020, whereas its structured period/date fields and D2.2 identify 2022/23 and 2023. We preserve that inconsistency without changing the published year or inventing a correction. Historical breaks elsewhere in the metadata do not justify constructing an international trend here.

School-stage boundaries, staff assignment and national reporting practices can differ. The two panels provide harmonised contextual comparisons, not an exact decomposition of the Iceland trend. They cannot establish that Iceland has enough teachers, that shortages are absent, or why principals report teaching constraints.

## Exact selected workbook values

| Comparator | Primary: cell / ratio | Lower secondary: cell / ratio |
|---|---:|---:|
| Iceland | E26 / 9.700991 | I26 / 10.37211 |
| Denmark | E19 / 12.15202 | I19 / 11.22587 |
| Finland | E21 / 12.44579 | I21 / 9.155577 |
| Norway | E38 / 9.979199 | I38 / 7.636066 |
| Sweden | E44 / 12.57266 | I44 / 11.11519 |
| OECD average | E50 / 14.099339842105262 | I50 / 12.938765735294115 |

The XLSX stores decimal observations even though its visible cells are formatted as whole numbers. Plot the stored decimals, retain the original XML number strings and cell coordinates, and display one decimal with Icelandic decimal commas. Do not reverse-engineer ratios from rounded table images. The raw workbook, checksum, retrieval UTC, normalized cells, selected CSV and plotting CSV are retained in the self-contained report.

The comparison supports a limited description: Iceland is below the OECD average at both levels. It is the lowest of these five Nordics at primary level, close to Norway; Finland and Norway are lower at lower-secondary level. A blanket statement that Iceland is never the lowest Nordic would be incorrect.

## Presentation decision

Two horizontal dot plots, identical country order and 0–18 scale. Iceland uses the existing navy highlight and a larger point; the official OECD average uses a secondary green diamond; other Nordics are quieter grey. Panels sit side by side at 1100/720 pixels and stack at 360 pixels. The caption distinguishes the national trend from the harmonised comparison and states that the measure is not class size. The original Iceland chart, cohort chart, other public visuals and economic results remain unchanged.

**Final status: COMPARABLE_WITH_CAVEATS.**