Weighting Namibia 2013 Data (PR) [message #17769] |
Thu, 23 May 2019 23:51 ![Go to next message Go to next message](/theme/default/images/down.png) |
sduklee
Messages: 2 Registered: May 2019
|
Member |
|
|
I have two questions about Namibia 2013 Data use:
1. According to the 2013 Namibia Final Report, the total number of women and men was 41,665. However, only 41,646 observations are available in the PR dataset. Any insights in this will be greatly appreciated!
2. I would like to duplicate the Table 17.5 from the 2013 Namibia Final Report. Am I using the correct weights?
proc surveyfreq data=nm.NMPR61FL2;
tables hv104*SH328;
cluster Hv021;
strata Hv023;
weight newWeight; /*newWeight = hv028/1000000*/
run;
3. Should I use different weights for biomarkers (e.g., fasting blood glucose)?
Thanks!
[Updated on: Fri, 31 May 2019 13:48] Report message to a moderator
|
|
|
Re: Weighting Namibia 2013 Data (PR) [message #17796 is a reply to message #17769] |
Tue, 04 June 2019 15:39 ![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 our Research & Data Analysis Director, Tom Pullum:
You say "According to the 2013 Namibia Final Report, the total number of women and men was 41,665." Apparently you are referring to table 2.1 on page 12, the number in the bottom right. This is the number of individuals, not "women and men" and is limited to de jure residents, for whom hv103=1. I used the Stata command "tab hv025 if hv103==1 [iweight=hv005]". The restriction to de jure residents is mentioned in the title of the table. Most tables in the report, however, refer to de facto residents, for whom hv012=1.
The other question is about table 17.5. It says that 2621 women and 2091 men (weighted) were in the denominators for the percentages 2.6% and 2.6%. I can match the n's and %'s with "tab sh328 hv104 if hv105>=35 & hv105<=64 & hv042==1 [iweight=hv005/1000000],m col". Two things surprise me. One is that these numbers are not restricted to de facto residents (OR de jure residents). The other is that the n's include some cases that are missing on sh328 (that's why the "m" is included after the comma in this Stata command). However, that command matches the bottom row of table 17.5
Some numbers in the reports are indeed difficult to match, and alternative estimates could be just as valid. However, DHS tries to be consistent with WHO guidelines and MICS (UNICEF) practices and there are usually good reasons why the indicators are defined as they are.
|
|
|
Re: Weighting Namibia 2013 Data (PR) [message #17797 is a reply to message #17796] |
Tue, 04 June 2019 15:51 ![Go to previous message Go to previous message](/theme/default/images/up.png) |
sduklee
Messages: 2 Registered: May 2019
|
Member |
|
|
Thank you very much for the feedback - it was very helpful! =D
I just have one more question - when should we be using hv005 vs hv028? For the health module or men's survey, shouldn't hv028 be used? I did notice that using either hv005 or hv028 did not make a huge differences for self-reported diabetes.
Any insights into this will be greatly appreciated.
[Updated on: Tue, 04 June 2019 15:54] Report message to a moderator
|
|
|