The h-index rewards a steady supply of moderately cited papers and says nothing about what kind of work a record is made of. In a field where medical students, residents and attendings publish at very different volumes — and where a large share of that output is editorials, case reports and database dumps — a single count is easy to inflate and hard to read. The yang index answers a different pair of questions: how much credited work is there, and what kind of work carries the weight.

Mass (M) — what each paper is worth

Each paper's mass is w = RCR^1.5 × credit. RCR is the NIH Relative Citation Ratio, which is field- and time-normalised, so a paper is measured against others in its own area and era rather than by raw citations. The 1.5 exponent (gamma) spreads records apart instead of letting everyone average toward the middle — high-impact papers pull more weight, marginal ones less. Credit is your share of the byline (see below). Multiplying the two means a paper only adds real mass when it is both used by the field and genuinely yours.

Credit (c) — conservation across the byline

Credit is a fixed points table by position — sole 5, first 4, senior 3, second 2, middle 1 — normalised so the credit across one paper's byline always sums to one. Nothing is created by adding authors; a paper is a fixed amount of credit divided up. Co-first authorship promotes the second author to a first-author share. Because credit is conserved, padding is invisible: appearing as a middle author on many papers accumulates only small fractions, so it cannot manufacture mass.

Worked example. On a four-author paper you led, the byline scores first 4, second 2, middle 1 and senior 3 — ten points in total — so your first-author share is 4 ÷ 10 = 0.40. The same paper with you sitting in the middle would be 1 ÷ 10 = 0.10, and as senior author 3 ÷ 10 = 0.30. Whoever is on the byline, the four shares always add back to 1.

Evidence level — the CEBM ladder

The horizontal axis is the Oxford Centre for Evidence-Based Medicine (CEBM) Levels of Evidence, scored 0–100. Randomised trials, systematic reviews and meta-analyses sit at the top; prospective cohorts below them; registry and multicentre work in the middle; retrospective and case-control below that; case reports, narrative reviews and editorials near the bottom. These weights are provisional and set in one editable block at the top of the script — they are exactly what the current review with the PIs is meant to settle.

How a paper is assigned to a tier is separate from what that tier is worth, and does not change: a chain runs from the indexed PubMed publication type, to MEDLINE subject headings (Retrospective Studies and similar), to design wording in the title and abstract, and only then falls back to the low registry tier when nothing indicates a design. A "Multicenter Study" that names a recognised administrative database is moved to the registry tier. When nothing at all indicates a design, the paper drops to an assumed no-design tier (weight 50) rather than being credited as a cohort. Every row records which step decided it, so you can correct the ones the record left ambiguous.

Center of Mass (C) — where the weight sits

C brings the two pieces above together into a single balance point. Each ticked paper is placed on the 0–100 CEBM axis at its evidence level e and weighted by its owned mass w, and C is the mass-weighted average of those levels, C = Σ(w · e) ÷ Σ w. It asks not how much a record contains but where its credited weight sits on the evidence ladder.

So a higher C means the mass of a record sits at higher CEBM levels — prospective cohorts, trials, systematic reviews and guidelines rather than editorials and case reports — which lets C stand as a reflection of the methodological quality of the work, where M reflects its quantity. Because C is mass-weighted, a few high-impact, high-evidence papers pull it up more than many marginal ones, and uncited padding (zero mass) does not move it at all.