Hi everyone,
I am conducting an analysis on the trends of breastfeeding using MICS 4, 5, and 6 data in Vietnam. The population of interest is women aged 15-49 who have given birth in the last 2 years. I plan to use the svyset command, incorporating the PSU, stratum, and weight variables.
However, I have encountered an issue: in MICS 4 and 5, the strata are defined by urban and rural areas, whereas in MICS 6, they are defined by ethnic groups. Given this difference, how should I approach weighting the data when combining the years?
Thank you very much in advance for your help!
When pooling multiple IPUMS MICS samples (e.g., Vietnam 2010, 2013 and 2020), you should use IMICSSTRATUM as the stratum variable. This variable reports the unique cross-sample sampling stratum. IPUMS creates this variable by making an adjustment to strata codes to ensure that strata identifiers are unique when pooling samples that may reuse strata identifiers. This solves the problem of strata not being comparable across samples, or the same strata code identifying different strata in different samples. When pooling samples, it is generally rare that the strata would be the same between different samples, and using IMICSSTRATUM ensures that the sample-specific strata are appropriately retained and identified in the data.