Weighting Calculator

This calculator computes a post-stratification weight for each of up to four strata as Population % ÷ Sample % (lines 15-18), the factor you'd multiply that stratum's responses by to make your sample proportions match known population proportions. Each stratum's weight depends only on that stratum's own Sample % and Population %; Stratum 1's inputs never move Stratum 2's weight or vice versa (lines 15-18 read one pair of inputs each). At the defaults, Sample % Stratum 1 has a very slightly larger effect on Stratum 1 Weight (about a 20.2% swing when nudged 10% up and down) than Population % Stratum 1 does (exactly 20%) — not a coincidence of these particular numbers but a general property of any ratio N ÷ D: because 1/D is a convex function, a symmetric 10% nudge to the denominator always produces a slightly larger swing than the same-sized nudge to the numerator.

Weighting Efficiency applies Leslie Kish's design effect formula for unequal weighting — Deff = 1 + the coefficient-of-variation-squared of the nonzero weights, lines 26-34 — the same measure Kish introduced in his 1965 text *Survey Sampling* to summarize how much a weighting scheme inflates the variance of survey estimates relative to a simple random sample of the same size; strata whose weight comes out to exactly 0 — Stratum 3 and Stratum 4 at the defaults, since their Population % is 0 — are excluded from that calculation entirely rather than dragging it toward 0. This calculator does not validate that your Sample % or Population % values actually sum to 100 across strata, and it does not compute a true multi-way design effect beyond this weight-only Kish approximation restricted to whichever of the up-to-four strata currently have a nonzero weight.