Family structure [message #8474] |
Sun, 01 November 2015 08:42 ![Go to next message Go to next message](/theme/default/images/down.png) |
Sadhara
Messages: 2 Registered: October 2015 Location: South Africa
|
Member |
|
|
I am intending to create a variable "family structure" with the fpollowing categories 'single family', 'nuclear family' and 'extended family' using DHS data. I need help on how to get the best results. Should I use the individual file only? Or should I merge with the household member file and or the male file? I need help.
|
|
|
|
|
Re: Family structure [message #8548 is a reply to message #8536] |
Thu, 12 November 2015 09:55 ![Go to previous message Go to previous message](/theme/default/images/up.png) ![Go to next message Go to next message](/theme/default/images/down.png) |
Bridgette-DHS
Messages: 3139 Registered: February 2013
|
Senior Member |
|
|
Following is a response from Senior DHS Stata Specialist, Tom Pullum:
Rather than work with what you did, I have started from scratch with the Stata code in the attached text file. This is not a final version--you will have to modify it. I tested it on the PR file for the most recent Ethiopia survey. I get this distribution:
![index.php?t=getfile&id=464&private=0](index.php?t=getfile&id=464&private=0)
The idea is that you must construct binary codes for everyone in the household, based on their value of hv101 (and sometimes also their sex and/or age), then collapse to get the number of people of each type in the household, and then work with these numbers (usually whether they are 0 or greater than 0) to construct the household type. Once you have the type for each household, you can merge it back to the PR file.
The three categories of households that you gave me are not clearly defined. I have tried to implement your instructions but some nearly half of the households fall outside of these three types, as you have defined them.
Strictly from a programming perspective, this program could be re-written to be more elegant, but I think that would just make it harder to understand.
Note that a household is not the same as a family. In the DHS data files, hv002 is the household id code. Usually a household can be interpreted to be a family, but not always. Good luck.
|
|
|
|
Re: Family structure [message #16406 is a reply to message #16362] |
Wed, 09 January 2019 06:39 ![Go to previous message Go to previous message](/theme/default/images/up.png) |
Bridgette-DHS
Messages: 3139 Registered: February 2013
|
Senior Member |
|
|
Following is another response from Senior DHS Stata Specialist, Tom Pullum:
Your December 26 email is almost identical to a query you posted on November 1, 2015. I replied to that query and don't have much to add now. If you are still working on this, I think there are two problems. The first is that you cannot do what you want to do with just the IR file. A typology of household types or family types requires hv101 (relation to head), hv104 (sex) and hv105 (age) in the PR file. You can also get information from hv112 and hv114 (line number of a child's mother or father, if in the household). You would use the PR file to calculate a code for household type or family type and then attach the code to the women in the IR file, using a PR-IR merge. The second problem is that hv101 gives the relation of household members to the household head. Sometimes the head is a woman, but if not, then you have a limitation, especially if there is more than one eligible woman in the household.
You probably can get close to a solution if you are willing to say that some small percentage of cases are ambiguous or indeterminate. In most surveys, really complex household structures are relatively rare.
You provided some Stata code with the 2015 post, but not with the recent one. Did you mean to include some? You say "please find the syntax for the same" but there was no attachment. You can re-post with an attachment.
[Updated on: Wed, 09 January 2019 07:28] Report message to a moderator
|
|
|