Skip to content
What Causes What

Methods

The product is an epistemic filter, not a supplement shop. An edge is allowed onto the atlas only when we can name its design class.

The four stamps

MECHANISM

Supported by an experiment or a tightly identified design. Read the caveats.

MENDELIAN

Mendelian randomization. It's a great tool for ruling out environmental factors, though it has its own genetic limitations.

LOOP

Each side influences the other.

SMOKE

The association disappears when you account for other factors, like socioeconomic status or healthy-user bias.

The sign of the arrow

A stamp says whether the link is a mechanism, a genetic instrument, a loop, or smoke. It does not say whether the target goes up or down. That is polarity: + increases, − decreases, ~ non-linear, ? conditional, 0 null. It is not a fifth stamp. Inquire multiplies signs along a path so a chain can be read as a net rise or a net fall.

+ INCREASES

A rise in the source is signed as a rise in the target.

− DECREASES

A rise in the source is signed as a fall in the target.

~ NON-LINEAR

Dose, timing, or a U-shaped curve can flip the sign.

? CONDITIONAL

The sign depends on a moderator. Read the mechanism.

0 NULL

The identified effect is zero, or the association is smoke.

What we will not do

  • Promote a bivariate observational correlation as a treatment effect.
  • Hide an interaction (T × cortisol) inside a slogan.
  • Cite a paper we have not opened. Camerer did not run the testosterone ultimatum RCT; Eisenegger, Fehr and colleagues did.
  • Pretend Inquire is full do-calculus. It is a backdoor scan on a curated graph.
  • Give medical, legal, or financial advice.

How a new edge gets in

Extract X, Y, effect size, p-value if reported, design class, and N. Attach the DOI or PMID. Name confounders. Prefer RCTs and Mendelian randomization when they exist; keep observational work when it is the literature, and stamp it honestly. When popular claims rest on confounded or unproven associations, we do not hide the papers; we cite them under a spurious edge so visitors can trace the evidence themselves. Future ingestion from PubMed and open bibliographic sources writes into the same contracts. The gold set inpackages/contracts/src/data/seed.tsis the exam, not the living catalog. The public bibliography is /sources (papers cited by an edge). The screenshotable grade table is /matrix.