Sublime

finally realized that the messages were conditional probabilities in one direction and likelihood ratios in the other.

statistical quantity to be estimated from the data that, once estimated, can legitimately represent the answer to our query.

In a multivariate test, the individual changes are served up at random to create many possible combinations.

the intuition: The probability has gone up because evidence from past experiments (two-thirds have not increased performance) is now included in the analysis.

Thus, people do not learn the relation between sample size and sampling variability, although the data for such learning are abundant.

hypothesis test evaluates two mutually exclusive statements about the population and determines which statement the data support.

Small data is big data in disguise. The reason we can often make good predictions from a small number of observations—or just a single one—is that our priors are so rich.

statistics isn’t about data but distillation, rigor, and avoiding being fooled by randomness