Announcement

Collapse
No announcement yet.
X
  • Filter
  • Time
  • Show
Clear All
new posts

  • svyset with a three-stage sampling design

    Hi all,

    I am wondering how to declare the survey design when there are both multiple stages of sampling and an element of time. So, the setup is like this:

    For a given week, randomly select sample days with probability pdayi
    1. On the given day, select clusteri with probability pclusteri
    2. Within clusteri , select site with probability psitei, where sites are stratified so that psitei = pstratums * (1/ number of sites in stratums)
    3. Within sitei, select one individual per group. The probability of an individual might be 1/N where N=number of groups * avg number of individuals per group
    In this case, what is the best way to describe survey data in Stata?

    So far I have tried simply multiplying all the individual pieces together for one "pweight" that applies to an indivdual and declaring:
    Code:
    svyset [pweight=pweight]
    And also
    Code:
    svyset cluster, vce(linearized) singleunit(missing) || site, strata(stratum) || _n [pweight=pweight]
    Any advice would be appreciated.
Working...
X