s‍vy is a Python package for designing and analyzing complex survey data.

Large-scale surveys are often selected using complex random mechanisms such as stratification, clustering, or unequal probabilities. To produce accurate estimates, statistical analyses must account for the sampling design. s‍vy provides a unified toolkit for working with these designs, ensuring that inferences correctly reflect the underlying population.

svy provides a comprehensive set of tools to support quantitative survey research.

Key Features

  • Sample size calculation and allocation Functions to determine required sample sizes and distribute them efficiently across strata or clusters.

  • Sampling methods Built-in algorithms for:

    • Simple random sampling (SRS)
    • Systematic selection (SYS)
    • Probability proportional to size (PPS)

  • Sample weighting Tools to derive design weights and adjust them for nonresponse or to calibrate against population totals.

  • Estimation of population parameters Procedures for means, totals, proportions, and regression estimates using:

    • Taylor-based methods
    • Replication-based methods (Bootstrap, BRR, Jackknife)
    • Regression-based methods (e.g., generalized regression, GREG)

  • Categorical data analysis Support for survey-weighted:

    • Chi-square tests
    • Logistic regression
    • Multinomial regression

Next Steps