Reading a facial morphometry paper: a beginner's guide

Every Facet score traces back to a peer-reviewed paper. Fink 2021 for skin homogeneity. Carruthers 2020 for brow ideal. Goode 1984 for nasal projection. The citations are listed in the methodology page, but "listed" is not "explained." This piece walks through how to find and read a morphometry paper, in case you want to verify Facet's choices yourself.

Where to find papers

• PubMed (pubmed.ncbi.nlm.nih.gov): the canonical free index of biomedical literature. Most aesthetic morphometry papers are indexed here.
• Google Scholar (scholar.google.com): broader, includes some non-indexed journals. Useful for tracking citations forward ("papers that cite this one").
• Direct from journal: many papers are paywalled, but author-uploaded preprints on ResearchGate or institutional pages are common. Author email also works; researchers usually send PDFs on request.

Anatomy of a paper

A morphometry paper typically has six sections. Read them in this order:

1. Abstract. A summary in a paragraph. Read this to decide whether to keep going.
2. Methods. The most important section. How was the data collected? How many subjects? What instruments? What ethnicities and ages? A paper with a sample of 20 European college students cannot speak to global thresholds.
3. Results. The numbers. Tables and figures matter more than the prose around them.
4. Discussion. The authors' interpretation. Often the weakest section. Treat as opinion.
5. Limitations. Frequently buried at the end. Always read. A paper with a strong limitations section is more trustworthy than one without.
6. References. The citations the authors trust. Use these to find adjacent papers.

Key terms

• n: sample size. A study with n = 30 is preliminary. A study with n = 3,000 is a real claim. Be skeptical of small samples.
• p-value: statistical significance. p < 0.05 is the conventional threshold. Smaller p-values are stronger. p > 0.05 means "could be chance."
• Confidence interval (CI): the range the true value probably falls within. A narrow CI is a more precise estimate.
• Mean / median: average / middle. They are not the same. Skewed distributions diverge.
• Standard deviation (SD): how spread the data is. Used in band definitions.

Walking through Fink 2021

Fink et al. 2021 is the source behind Facet's skin homogeneity scoring. The relevant paper measures chromophore variance (variation in hemoglobin and melanin concentration across facial skin) in a large sample of women across age groups, and demonstrates that lower variance correlates with perceived skin health.

If you opened it on PubMed today, you would:

1. Read the abstract: a one-paragraph summary of the finding.
2. Skip to methods: what device measured chromophore variance, on what skin area, in what light conditions, with what sample.
3. Look at the figures: typically a scatter plot of variance vs perceived age, with a fit line.
4. Check the n: a large enough sample to support the claim.
5. Note the limitations: the authors will name them, usually around ethnicity coverage and lighting standardisation.

That is the entire reading. It is one to two hours to genuinely understand a single paper, and it is the most defensible way to evaluate any product that claims to score skin.

Why this matters for Facet

Facet's claim is not that the engine is sophisticated. The claim is that the thresholds are defensible. You can find the papers, you can read the methods, you can disagree with the choices. That is the point. A scoring engine you cannot verify is worth less than no scoring at all.