I have a very small sample size on drowning data of Bangladesh in following way

enter image description here

Now I want to give sampling weight for further analysis. How can I calculate sampling weight for the data? I have information of total population. I have searched in google but didn't get any concrete idea or calculation procedure.

  • $\begingroup$ Your sampling weight should depend on your sample design. How did you design/collect this sample? $\endgroup$
    – RoryT
    Commented Jul 24, 2017 at 2:28
  • $\begingroup$ It is a national, cross-sectional survey with 300,001 respondents interviewed between March and June 2016. It uses a complex sampling strategy based on probability-proportional-to-size (PPS) methodologies with separate urban and rural samples. $\endgroup$ Commented Jul 25, 2017 at 3:23

1 Answer 1


Your data set should inherit the sampling weights from your original survey. Moreover, your 113 cases should be analyzed as a domain of your survey (svy, subpop(): whatever), keeping (in the worst case, depending on the specifics of the original design) all 300K observations in your data set. See West, Berglund and Heeringa (2008).

  • $\begingroup$ I have to work with only these 113 samples from 300k observations for specific topics. Did you recommend any articles for imbalance data cause I want to explore the effect on drowning on these groups. $\endgroup$ Commented Sep 20, 2017 at 18:48

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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