usualcaveats.com

You may be here because someone invoked usualcaveats.com.

It exists to spare conversation the dreary tariff of compulsory caveats and statistical confusion - deliberate or not.

You are allowed to generalise

  1. Population-level claims concern patterns in a distribution, not verdicts on particular individuals.
    A population has shape, frequency, proportion. It does not have a face.
  2. Describing what is typical does not prohibit what is possible.
    “X is not Y” does not mean “X cannot be Y.”
  3. A probabilistic claim is not defeated by a single counterexample.
    One contrary instance refutes a universal claim; it does not dissolve a distribution.
  4. Observation or explanation does not entail advocacy or justification.
    To analyse and explain a phenomenon is not to enlist in its defence.
  5. Unequal outcomes do not imply unequal worth.
    Statistical claims quantify patterns. They do not grade souls.
  6. Some domains are dominated by rare events.
    In such cases, averages conceal fragility and extremes drive outcomes.

Good faith is assumed

  1. Error is presumed before malice.
    Incompetence is common; conspiracy is rarer.
  2. An anecdote is not a verdict.
    It may herald a shift; it does not, unaided, recast a distribution.
  3. Disagreement is not harm.
    To contest a claim is to engage with it. It is not an act of violence.