WorldMetra Global University Rankings V2.0 · RESEARCH PREVIEW

Seven dimensions. One auditable global standard.

V2.0 separates research, employment, employer evidence, earnings, teaching and international collaboration. Every observation retains its definition, year, source, sample and quality grade; missing evidence is never replaced with a guessed zero.

01 — V2.0 framework

The seven formal weights total exactly 100%.

Academic reputation, international faculty ratio, international student ratio and sustainability may be displayed as context, but each carries 0% and cannot change a score. Commercial relationships and institution-specific adjustments are prohibited.

30% + 20% + 15% + 10% + 10% + 10% + 5% = 100%
DISPLAY-ONLY · 0%
Academic reputation · international faculty ratio · international student ratio · sustainability
02 — Indicators and subweights

Indicators and subweights

01

Citation impact

30%

20% field/year/type-normalised citation impact; 10% citations per verified faculty FTE. A faculty denominator is used only when its definition and full-time-equivalent coverage are credible.

02

Research quality and output

20%

12% share of outputs in the global top 10% for comparable field/year/type cells; 8% contribution-, field-, scale- and subject-mix-adjusted output across the latest five complete years.

03

Graduate employment outcomes

15%

8% sustained employment; 4% skilled or professional employment; 3% evidenced medium- or long-term progression. Undergraduate outcomes are not replaced by MBA results.

04

Employer reputation

10%

10% from a real employer survey with a disclosed method, fieldwork year and employer sample. Proprietary third-party scores and AI judgements are excluded.

05

Graduate earnings and development

10%

2% at year 1, 3% at year 5, 1% at year 8, 3% at year 10 and 1% real earnings growth. “Years after entry” cannot be relabelled as “years after graduation.”

06

Teaching quality

10%

4% student-to-faculty FTE; 3% comparable teaching resources per student; 3% completion and verifiable learning outcomes. Reputation is not used as a teaching proxy.

07

International research collaboration

5%

5% cross-border co-authorship share and verified network contribution using consistent institution-country counting. International student or faculty ratios are not substitutes.

03 — Normalization and fairness

Compare like with like before calculating a score.

Research evidence is adjusted by field, publication year, document type, institution contribution and scale. Outcomes are separated by graduation cohort, subject, qualification, observation horizon and country. Earnings require inflation and purchasing-power treatment; nominal currencies are never compared directly.

04 — Missing evidence and eligibility

Null means unknown—not zero.

A formal comprehensive rank requires at least 80% comparable weight, both core research categories, and at least 2 non-research categories. A rescaled score below that line is labelled provisional with its effective coverage; it is not presented as a complete 100% score.

05 — Evidence standard

Each scored value needs a traceable evidence chain.

The record includes the raw value, unit, statistical definition, cohort, observation year, sample size, source URL, publisher, retrieval time, licence or terms, transformation and quality grade. Search snippets and AI estimates are not primary evidence.

06 — Quality control

Every release must survive repeatability and bias checks.

Checks cover duplicate institutions, branch and hospital attribution, retractions, repeated raw metrics, field and country effects, missingness, outliers, top-100 anomalies and rank sensitivity when citation weight is tested at 25%, 30% and 35%. Fixed inputs and a fixed method version must reproduce the same result.

25% / 30% / 35%

Citation-weight sensitivity scenarios for stability analysis; they do not replace the formal 30% configuration.

NULL ≠ 0

Missing observations remain null with a reason, source coverage and an explicit review field.

07 — Release governance

Ranking research and commercial operations are separated.

Method weights are versioned before calculation. Data corrections enter a review queue, leave an audit trail and never alter a published edition silently. Sponsorship, agency promotion, paid profiles and university participation cannot buy points, suppress evidence or change rank order. A material method change creates a new version and a new comparable series rather than rewriting an old release.

08 — Current research status

The existing 1–1,000 list remains a research index while V2 evidence is completed.

The V2 candidate pool and first-100 identity/research checkpoint are under review. A school does not receive a formal comprehensive rank until its required evidence passes the published eligibility rules. The site does not relabel the older research-only order as a completed V2 comprehensive ranking.