Hello,
I have a question about whether it is appropriate to use sampling weights (pweights) for my dataset. I am using Stata 13.1 on a mac computer.
I am conducting a secondary analysis on survey data collected for an endline survey on women. To achieve the calculated sample size, the sampling strategy was to randomly sample 100 census enumeration areas (EA) using a probability proportional to size technique. For each selected EA, 10 women were randomly selected to be included. However, in a number of EAs, there were less than 10 eligible women. To reach the required sample size, additional EAs were sampled for those regions not initially meeting the target survey number. In the end, there were 120 EAs, with 30 of them having less than 10 eligible women.
After reading examples using pweights, I’m struggling to understand whether sampling weights applies in this situation when setting up my survey characteristics in Stata. Can you please help? My first thought is YES because some EAs had less than 8 women, and so the probability of each individual sampled would differ.
I will be using svy: logistic to create a propensity score. I also will be using PCA for data reduction purposes.
Thank you in advance for your help.
Regards,
Carina
I have a question about whether it is appropriate to use sampling weights (pweights) for my dataset. I am using Stata 13.1 on a mac computer.
I am conducting a secondary analysis on survey data collected for an endline survey on women. To achieve the calculated sample size, the sampling strategy was to randomly sample 100 census enumeration areas (EA) using a probability proportional to size technique. For each selected EA, 10 women were randomly selected to be included. However, in a number of EAs, there were less than 10 eligible women. To reach the required sample size, additional EAs were sampled for those regions not initially meeting the target survey number. In the end, there were 120 EAs, with 30 of them having less than 10 eligible women.
After reading examples using pweights, I’m struggling to understand whether sampling weights applies in this situation when setting up my survey characteristics in Stata. Can you please help? My first thought is YES because some EAs had less than 8 women, and so the probability of each individual sampled would differ.
I will be using svy: logistic to create a propensity score. I also will be using PCA for data reduction purposes.
Thank you in advance for your help.
Regards,
Carina
