Statistics is best understood not as a branch of mathematics but as a technology for extracting meaning from data, turning raw numbers into understanding. Its tools are justified by what they reveal about the world, not by mathematical elegance alone.
No statistical technique, however sophisticated, can rescue conclusions drawn from poor data; errors, missing values, and distortions in collection propagate through every subsequent step. Garbage in yields garbage out, so scrutiny of how data were gathered matters more than the choice of method.
A properly drawn random sample allows reliable inference about a large population without examining every member, because randomness controls how representative the sample is. The precision of the estimate depends chiefly on the sample's size, not on the fraction of the population it covers.
A statistical significance test asks how probable the observed data would be if only chance were operating; a small probability is taken as evidence against that chance hypothesis. The result never proves the hypothesis true or false, it only quantifies how surprising the data are under the assumption of no real effect.
Two variables moving together may reflect a direct cause, a reverse cause, or a hidden third factor driving both, and the correlation itself cannot distinguish these. Establishing causation generally requires controlled intervention or careful elimination of confounders, not mere observed association.
Probability provides the formal language for reasoning about uncertainty, whether interpreted as long-run frequency of events or as a degree of belief that can be updated by evidence. These two interpretations underlie rival schools of statistical inference, yet both treat uncertainty as something that can be measured rather than merely feared.