2
$\begingroup$

The design effect (deff) quantifies the extent to which the expected sampling error in a survey departs from the sampling error that can be expected under simple random sampling .

My question is

  • Why do we compare expected sampling design with simple random sampling instead of other sampling method such as , systematic sampling , stratified sampling ?
$\endgroup$
2
$\begingroup$

Sampling designs are often compared to simple random sampling, mainly because formulas for estimators and variances (and variance estimators) are very simple for SRS :

$ \begin{align*} \mathbb{V}(\hat{Y}_{HT, SRS}) &= (1-\dfrac{n}{N}) \dfrac{S^2}{n} \\ \text{with : } S^2 &= \dfrac{1}{N-1} \sum_{k \in pop} (y_k - \bar{Y})^2 \end{align*} $

It'd be impratical to compare the sampling variance of a particular sampling design with :

  • Stratified sampling, because sampling error depends very much on sample allocation between strata and how strata are designed (imagine badly designed strata and allocation : you could end up with a much higher error than SRS with same sample size, whereas wisely designed strata and Neyman allocation can spectacularly improve precision compared to SRS)
  • Systematic sampling, because its variance is very hard to compute. Very often, people using systematic sampling consider that the sampling variance is roughly equal to the variance of a stratified sampling using proportional allocation and strata formed by the variable on which the sample frame was sorted before the draw.
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.