Conditional probability holds that all probability statements are implicitly conditioned on the information available, and that inference must account for how observed data was sampled rather than treating it as a representative snapshot of the full population. Selection bias arises when the observed subset of data (survivors, respondents, or otherwise non-randomly retained cases) systematically differs from the unobserved population, so that conclusions drawn from the visible sample misrepresent the underlying truth unless the sampling/conditioning mechanism is explicitly modeled. Regression toward the mean is a related statistical phenomenon in which extreme observations on a variable with a random component tend to be followed by less extreme observations on remeasurement, a consequence of imperfect correlation between repeated measurements rather than any causal or self-correcting process. These concepts belong to the foundations of probability and statistical inference, underlying survival analysis, missing-data problems, and the correct interpretation of longitudinal or repeated-measures data.
Conditional probability holds that all probability statements are implicitly conditioned on the information available, and that inference must account for how observed data was sampled rather than tr…