Type 4 · Stratification Failure
Level Jump
You treated a small thing like a big rule, or a big rule like a small thing. Knowledge exists in layers — examples, patterns, rules, principles — and jumping between them without justification breaks reasoning.
The move, up close
“Two of them showed up late, so the whole team is unreliable.”
A conclusion about the whole category is drawn from a couple of instances.
Instances really are evidence about the category — a little.
The leap from instance to category needs a sampling warrant that two cases can’t provide.
What it sounds like
You’ve heard this one.
“My uncle smoked until 90, so smoking must be safe.”
“One refugee committed a crime, so all refugees are dangerous.”
“It’s just a theory.” — dismissing established science.
“Someone got vaccinated and still got sick, so vaccines don’t work.”
“I saw one case of welfare fraud, so we should eliminate welfare.”
Classic fallacies
Names you already know.
The tag on each names the move in AOML coordinates — source → target: what got swapped in for what.
A few cases promoted to a universal rule.
One example used to override population data.
A group average applied to a specific case.
Judging past actors by present-day standards.
The test
Did the conclusion stay on the same level as the evidence? One example supports a conclusion about that one case; a general rule needs many systematic observations.
How to respond
Name the move, not the fallacy.
“That’s just anecdotal!”
“That’s a real example. What does the broader data show?”
“Hasty generalization!”
“That happened in these cases. How often does it happen overall?”
“You can’t judge them by our standards!”
“What information and norms were actually available to them at the time?”
The key insight. Match your confidence to your level. The mistake is claiming Level-4 certainty with Level-1 evidence.
The everyday cases here are technically compounds: the jump is repairable with the right evidence (the Type-2 root — the missing license), and the level boundary is just where you catch it. A pure Type 4 is a level failure no evidence can fix — a paradox like “this sentence is false,” where no consistent level assignment exists. For everyday reasoning it changes nothing: spot the jump, ask what justifies it.