How to make household the unit of analysis when using a household survey data

[ESTHER OSADOLOR]

Hello everyone,

I am currently working on LSMS-ISA for Nigeria, however, whilst some sections have unique identifiers as "indiv", that is the individual id, some have "hhid" as the unique identifier, some "item_cd", the id for items and one of the sections on enterprise is uniquely identified by "ent_id". I want to use household as the unit of analysis and also to merge the different sections but I don't know how to go about doing these.

Thank you all in advance.

[Weiwen Ng]

A word of caution: don't assume that everyone knows the survey you refer to, because we come from different countries, and different disciplines in those countries. Please note that I offer this advice every time I respond to someone even from my own country, so I didn't mean this personally.

Some of the surveys I've used have come in two or more parts at different units of analysis. It seems like your survey has multiple parts like this. I'd guess that the household characteristics file is unique by the household ID, hhid. That's fine, you can merge the files. However, if you merge the household file to the person file, what you will have is a dataset that's at the person level, but with all the household characteristics appended to it.

What exactly do you want to get from the person-level file? And it might be helpful for you to give an example of what sort of characteristics you want to use in tabulation or regression at the household level. Remember, if you ask a general question about a complex issue, it's extremely difficult to help you. If your question isn't specific enough, you will get ignored or you will get vague responses, and in other forums you might even get dyspeptic responses.

[ESTHER OSADOLOR]

Thank you Weiwen and sorry for the lack of details in my question. The survey I am using is the World Bank Living Standard and Measurement Survey for Nigeria. Yes, the survey has several parts but unfortunately the sections on household characteristics, such as the household roster, education, labour as well as remittances, are uniquely identified by indiv and not hhid.
#

[Weiwen Ng]

Thank you Weiwen and sorry for the lack of details in my question. The survey I am using is the World Bank Living Standard and Measurement Survey for Nigeria. Yes, the survey has several parts but unfortunately the sections on household characteristics, such as the household roster, education, labour as well as remittances, are uniquely identified by indiv and not hhid.
#

Unfortunately, there's still not quite enough detail here. Let's try this:

1) What dependent and independent variables do you want to create?

2) What data do you have now - do you have a person-level file with all the household characteristics appended? Or do you have something else? Describe the data structure in as much detail as you can.