Core Survey Analysis Infrastructure // A modern, 'S7'-based foundation for survey analysis spanning both probability and non-probability samples. Probability sample designs include Taylor series linearization, replicate weights (BRR, Fay, jackknife, bootstrap), and two-phase estimation, following 'Lumley' (2004) doi:10.18637/jss.v009.i08. Non-probability sample designs support bootstrap and jackknife variance estimation for opt-in panels and convenience samples. Provides a unified estimator interface for means, frequencies, totals, quantiles, ratios, correlations, regression, and t-tests, with weighted 'polychoric' and 'polyserial' correlation following 'Mannan' (2025) doi:10.2139/ssrn.6580480. A metadata system preserves 'haven'-style variable labels, value labels, and question-preface attributes through all operations. Uses a 'tidyselect' interface throughout.