Advanced Sampling

Objectifs

After a few reminders of the general principles of inference and estimation in the context of a finite population, unequal probability sampling methods will be presented, with concrete examples of their use.

In the second part, the notion of balanced sampling will be introduced. We will study the Cube method for selecting balanced samples, and present the associated variance estimation method. We’ll also look at examples of applications to INSEE surveys. We’ll look at how the Cube method can be used to limit imputation variance, in the case of random imputation to deal with partial non-response.

The third part of the course looks at spatial sampling methods.

The various techniques presented in the course will be worked on in practical exercises.

Plan

Part 1: Reminders on sampling methods

Part 2: Unequal probability sampling methods

– Poisson sampling

– Fixed-size unequal-probability sampling

– Application to self-weighted sampling in household surveys

Part 3: Balanced sampling

– The principle

– How it works

– Variance estimation

– Application 1: The new Census

– Application 2: Balanced random imputation

Part 4: Spatial sampling methods

Prérequis

Not indicated