Skip to content

Survey Design & Weighting Studio

01 · The finite population and its true mean

Every dot is one person. Three education groups differ in political-knowledge score (0–100). Ȳ is the quantity we want; it is fixed, we just cannot see it in practice.

Primary (A)Secondary (B)Tertiary (C)hover or focus a unit
Population field N = 600; Stage 1. 0 respondents. Hájek mean —. Linearised SE —.

Composition: population vs frame

Populationn = 600

Synthetic finite population (N = 600, 20 clusters, seeded). The data-generating process is known, so every bias shown is real relative to Ȳ. No empirical claims.

Stage 1. 0 respondents. Hájek mean —. Linearised SE —.

Formula

Ȳ = (1/N) Σ_U yₖ

Method: Finite population, known data-generating process · Synthetic, seeded data. Illustrative, not an empirical finding.

Validation: Hájek mean, linearised SE, closed forms and deff match R survey 4.2.1 on samples drawn by this studio (scripts/validation/studio_reference.R). Not validated: the SE after post-stratification or raking treats adjusted weights as fixed (the calibration step is ignored); two-stage samples use the with-replacement ultimate-cluster approximation. · integrated