Experiment IV · Open
The Tarot Frequency Lab
Rate the day you have just had. Then, and only then, draw a card. Over enough days the question answers itself: do the cards land where their meanings say they should?
How was your day?
The day that has just ended, as you actually found it. Answer before the card, because that ordering is the only thing keeping this experiment meaningful.
1 — Bad · 7 — Very good
Why the order matters
A tarot reading normally runs the other way: draw in the morning, then live the day. Testing it that way proves nothing, because a bad card can spoil a day simply by being drawn, and expectancy is indistinguishable from divination in the results.
So this register reverses it. You rate the day that has ended, the rating is stored, and only then is a card drawn — in the same request, so there is no moment where you hold a card and could still change your answer. The card cannot have influenced the rating, because it did not exist yet. If an association still appeared, it would need the draw to carry information about a day already over.
The cost is honesty about scope: this does not test predictive tarot as it is practised. It tests something narrower and cleaner.
The wager
Twenty-two major arcana, each drawn independently at one in twenty-two. The null says mean day quality is the same behind every card. The tarot claim says The Tower sits with worse days and The Sun with better ones.
Fixed in advance: Kruskal–Wallis across all 22 cards at α = .05, chosen over ANOVA because a 1–7 rating is ordinal. Four contrasts are named in the protocol before any data exists — The Tower, Death, The Sun, The Star — with Holm correction, so a card cannot be chosen for testing after it looks interesting.
The filed protocol
SHA-256 12d79ea7f85b964f520a734b8236883be2c46bea063389fa33c4834df9301975
Known ways this could mislead
- Twenty-two cards means twenty-two chances for one to look remarkable. Roughly one will, at the usual threshold, with nothing behind it — which is what the simulated panel is for, and why the correction is applied to the four named contrasts rather than to whichever card looks best.
- People rate a day after living it, and memory is reconstructive. That adds noise, but it cannot bias the result toward any particular card, because the card is unknown at rating time.
- People are likelier to record an entry on an unusual day. That skews the overall distribution of ratings without touching its relationship to a card drawn afterwards.
- A skewed draw would manufacture exactly the card effect this is looking for, so card frequencies are published and should sit at one in twenty-two each. A deviation there is a defect in our generator, not a discovery about tarot.
Reuse this instrument
Share the current result or embed its compact, accessible summary. The card reads the same public aggregate as this page and carries the current N.
Embed this result
Continue through the research
These links connect the claim to the test, method, or neighboring history that can challenge it.