In: Statistics and Probability
4. (a) Describe the advantages and disadvantages of using a stratified sample design in a survey. Also, what information would you need to create the strata.
(b) Describe the advantages and disadvantages of using a cluster sample design in a survey.
(c) There are various ways of doing cluster sampling. For example, there are one-or two-stage sampling and clusters can be selected with equal or unequal (pps) selection probabilities. The choice of which design is better depends on the relative sources of variation and the relative costs of sampling within and between clusters. i. What would the population have to be like for a two-stage cluster sample design to give better results than a one-stage cluster sample? ii. Describe an example where unequal selection of clusters would give more precise estimates than an equal selection of clusters.
(d) What is the design effect and in general terms (that is, an exact formula is not required) how is it calculated?
(e) For stratified and cluster designs would you expect the design effect to be greater, less than or approximately equal to zero? Clearly explain why you expect this.