Statistics turns raw data into tested conclusions
Statistics gathers, organizes, analyzes, interprets, and presents data, starting from a population or model. The German Statistik once meant describing a state; by the 1790s John Sinclair helped give English the modern sense of working with uncertain numbers and samples.
Statistics concerns every stage of data: planning collection through survey and experiment design, summarising samples, and drawing inferences about populations. When a full census is impractical, random sampling lets conclusions extend reasonably from sample to whole. Experimental studies manipulate a system and remeasure; observational studies correlate predictors and responses without manipulation. Fitted models also let analysts project likely outcomes beyond the observed record.
Two broad methods dominate. Descriptive statistics compress data via indexes like mean and standard deviation, characterising central tendency and dispersion. Inferential statistics uses probability theory to conclude from data subject to random variation—observational error, sampling fluctuation. Hypothesis testing proposes relationships, compares them to a null of no effect, and recognises Type I errors (false positives) and Type II errors (false negatives). Multiple problems arise around obtaining sufficient sample size and specifying adequate null hypotheses.
The discipline sits between probability and applied fields. Probability deduces sample behaviour from known population parameters; statistics inductively infers population parameters from samples—the reverse direction. Though once taught together, they are conceptually distinct. One encyclopedia entry calls statistics both a science of uncertainty and a toolkit for wringing meaning from numbers.
Word history traces to Latin status ("condition"). Gottfried Achenwall coined German Statistik in the 1740s; English adopted it by 1770 for political arrangements, gaining its modern analytical sense in John Sinclair's 1790s works. Classic pitfalls include the Hawthorne effect—workers at Western Electric's Hawthorne plant improved productivity simply because they knew they were observed, not because lighting changed. Measurement itself can alter the system being measured.
Source: Statistics
