Survey Weighting: Post-Stratification, Raking (IPF) & Design Effects
Compare sample and population margins, build post-stratification cell weights, rake a 3 × 3 table to two margins by iterative proportional fitting, trim and rescale extreme weights, compute weighted estimates with SUMPRODUCT, measure Kish's design effect, widen the margin of error and chart the before/after gaps: 8 graded labs on one fictional transit survey. Synthetic data for practice.
The first lesson is free — no account needed.
About this course
A survey weighting course that weights the 1,200 interviews of one fictional transit survey in a spreadsheet, for survey researchers, market and public opinion analysts, data analysts and statistics students. You compare the sample with the population margins for age, gender and region, build post-stratification cell weights on age by region, rake a 3 × 3 table to two margins by iterative proportional fitting until the sums converge, trim and cap extreme weights and rescale them, calculate weighted means and shares with SUMPRODUCT, measure Kish's effective sample size and design effect, widen the margin of error for the weights and for a small subgroup, and chart the gaps to the population before and after weighting. Synthetic data for practice. The first lab is free.
Lessons
- Sample vs Population: Comparing the Margins (Age, Gender, Region)Free
- Cell Weighting: Post-Stratification on Age × Region Pro
- Raking by Iterative Proportional Fitting (IPF): Two Margins, Four Steps Pro
- Weight Trimming & Capping: Limits, Rescaling and Who Is Affected Pro
- Weighted Estimates with SUMPRODUCT vs Unweighted Results Pro
- Kish Effective Sample Size & Design Effect: Raked vs Trimmed Weights Pro
- Weighted Margin of Error: z, √deff and Subgroups Pro
- Before vs After Weighting: Gap Table & Comparison Chart Pro