Data Sources & Methodology

What we use

Subject attainment rates (Not Achieved / Achieved / Merit / Excellence) for NCEA Levels 1, 2, and 3, broken down by ethnicity, gender, school equity group (Q1–Q5), and region. Coverage: English-medium secondary schools, reported at national level and by 16 NZ regions.

Key caveats

• achieved_rate is the Achieved-grade-only band — NOT the overall pass rate. Pass rate = 1 − not_achieved_rate.
• Equity group data (Q1–Q5) is available from 2019 onwards only.
• Each breakdown is single-dimensional — ethnicity, gender, and region data cannot be cross-tabulated.
• Scholarship "Maori" appears without macron in source data.

What we use

Outstanding / Scholarship / No Award rates for NZ Scholarship Calculus and Statistics, broken down by ethnicity, equity group, gender, and region.

Key caveats

• Small cohort sizes in some ethnicity/region breakdowns — treat with caution.
• Scholarship is awarded to the top ~3% of candidates nationally.

What we use

Excellence and Merit endorsement rates for NCEA L1, L2, L3 and University Entrance qualifications, broken down by ethnicity, equity, gender, and region.

Key caveats

• Endorsement is a qualification-level award (not subject-level) — reflects the full year's achievement across all subjects.
• Equity group format changed in 2019: decile bands replaced with equity index groups (Fewer/Moderate/More resources).

What we use

Current-year and cumulative pass rates for NCEA literacy and numeracy co-requisite standards at Year 11, 12, and 13. Broken down by ethnicity, equity group, gender, and region.

Key caveats

• New co-requisite standards (CAAs) replaced old Unit Standards from 2020 — causes a sharp drop in current-year pass rates visible from 2020.
• Equity group format changed in 2019: decile bands replaced by equity index groups.
• Current-year rate = first-time passers. Cumulative rate = ever passed by that year level (includes re-sits).

What we use

NZ Grade 4 (Year 5, age ~9) maths scale scores 1995–2023, by gender. 2023 international country comparison (~58 countries). Coverage: nationally representative sample of Year 5 students in English-medium schools, tested in February each year.

Key caveats

• International average is recalculated each cycle based on participating countries — not directly comparable across years.
• TIMSS scale is NOT the same as NMSSA MS scale. These are completely separate measurement systems.
• AUS/ENG comparison lines are approximate from published reports; exact values may vary slightly by rounding.

What we use

Mean Scale Score (MS units) for Year 4 and Year 8 students, by ethnicity, gender, and school decile band. Coverage: ~2,000 students per year level, English-medium state and integrated schools. Stratified sample by decile, region, and school size.

Key caveats

• MS scale is designed so the combined 2013 average ≈ 100 with SD ≈ 20. Year 4 and Year 8 are NOT on the same sub-scale — a score of 84 at Y4 is not comparable to 84 at Y8.
• 2013 values in our data are reconstructed on the 2018 MS scale via a linking exercise (NMSSA Report 19, Appendix 6). They differ from the original 2013 report figures.
• 95% confidence intervals for 2013 are approximated from 2018 standard errors (similar sample sizes). Treat 2013 CIs as indicative.
• NMSSA assessed at Year 4 and Year 8. The successor programme (Curriculum Insights) assesses at Year 3, Year 6, and Year 8.

What we use

Percentage of students meeting / approaching / behind provisional NZ Curriculum benchmarks, at Year 3, Year 6, and Year 8. Nationally representative sample, successor to NMSSA (launched 2023).

Key caveats

• Uses % meeting benchmarks — NOT the MS scale score used by NMSSA. These two datasets cannot be compared on the same chart.
• Year levels changed: NMSSA measured Year 4 + Year 8; Curriculum Insights measures Year 3 + Year 6 + Year 8.
• No statistically significant change was observed between 2023 and 2024 at any year level.
• Demographic breakdowns (ethnicity, gender) are available in interactive data windows only — not included in our database.