|Wealth Index and place of residence [message #3746]
|Wed, 04 February 2015 11:19
Registered: January 2015
Location: United States
Hi, I have been having some trouble to work with Uganda's 2011 data. FYI I am working with STATA.
I am only interested in some questions on the women's questionnaire (only for women that gave birth, since my outcome is birth assistance) and only two questions from the household questionnaire (HV271 and HV025). The individual recode ugir60dt file is too large to open it with STATA, I tried using the "Select" software but I get an error. I opened the file with SPSS to check the data.
My questions are:
1. Since I will be performing my analysis with STATA I used the births recode file (ugbr60dt) instead to get the variables I need for the women. All the variables I need are included in this dataset. Since I am only interested in those women that gave birth, is ok to use this dataset instead of the individual recode?
2. I need to merge HV271 and HV025 from the household recode to the previous dataset and match the women with this two household characteristics (I match using hhid). When I do so I get a lot of missing values. What calls my attention is that I have the data for households with low Household number (<24) but not for the rest. Is this ok? Why is this? I took a look at the Individual's recode on SPSS and saw that there are a lot of missing values too, but couldn't check if they are the same.
I need to know if I am doing the right thing.
Thanks a lot,
|Re: Wealth Index and place of residence [message #3768 is a reply to message #3746]
|Fri, 06 February 2015 14:20
Registered: February 2013
Here is a response from one of our experts, Dr. Sarah Staveteig,
1. We recommend using the dataset appropriate to the universe you wish to study, in this case women. In many versions of Stata you can set maxvar and set memory. If you are using Intercooled Stata it may simply not be possible to open the entire dataset. If the select utility on our website does not work for you, you can also specify only the variable names you want to open in Stata.
use [varlist] [if] [in] using filename [, clear nolabel]
for example you could type:
use v000-v191 b* using UGIR60FL, clear
This will open the dataset with just the respondent's background variables and birth history variables. You can also name a list of variables individually.
Also note, if you are certain that all of the women you want to study are contained in the BR file, you can use that instead, but remember that women who have had more than one birth in past 5 years will appear multiple times. It is better to try opening the IR file as above if possible.
2. Look for variables on wealth in individual datasets, usually v190 and v191. These are pre-merged from the household data. Many other relevant household variables are already added to individual datasets. If you do come across a variable you need to merge from household dataset, see instructions for merging DHS datasets at http://dhsprogram.com/data/Merging-Datasets.cfm. As you discovered, the different ways that statistical programs handle spacing in the values for HHID generally prevent a clean match. Instead you will want to merge on cluster and household number (hv001 with v001, and hv002 with v002 respectively).
Hope this helps.