The DHS Program User Forum
Discussions regarding The DHS Program data and results
Home » Data » Merging data files » Problem with gender variable merging household recode with women's recode
Problem with gender variable merging household recode with women's recode [message #5636] Fri, 19 June 2015 03:23 Go to next message
tiampalermo@gmail.com is currently offline  tiampalermo@gmail.com
Messages: 3
Registered: June 2015
Member
I am using the Tanzania AIS 2011-2012 data and trying to merge the household recode (PR) with the women's recode (IR). I created a gender variable based on hv104 in the PR dataset, created cluster, household and indiv (hvidx) variables with consistent names to merge across datasets, and then merged 1:1 on cluster hh id with the women's recode (IR). However, when I tab the _merge variable based on gender, there are n=8352 matches (_merge==3) among those who are supposedly male according to the PR dataset. Why is the gender information in the PR dataset inconsistent with who is interviewed in the women's recode (IR)? I would appreciate any insight on where I'm going wrong.

Thank you. Code is below:

. use "$z\TZPR6AFL.DTA", clear

. *gender
. tab hv104, missing

Sex of |
household |
member | Freq. Percent Cum.
------------+-----------------------------------
Male | 26,132 48.37 48.37
Female | 27,888 51.63 100.00
------------+-----------------------------------
Total | 54,020 100.00

. tab hv104, nol

Sex of |
household |
member | Freq. Percent Cum.
------------+-----------------------------------
1 | 26,132 48.37 48.37
2 | 27,888 51.63 100.00
------------+-----------------------------------
Total | 54,020 100.00

. gen female=hv104==2

. gen male=hv104==1

.
. tab female

female | Freq. Percent Cum.
------------+-----------------------------------
0 | 26,132 48.37 48.37
1 | 27,888 51.63 100.00
------------+-----------------------------------
Total | 54,020 100.00

.
.
.
. ***generate variables for merging with same names across datasets
. gen cluster=hv001

. gen hh=hv002

. gen id=hvidx

.
. /*****MERGES********/
.
. *women's questionnaire
. preserve

. use "$z\TZIR6AFL.DTA", clear

. count
19319

. gen cluster=v001

. gen hh=v002

. gen id=v003


. gen female2=1

. sort cluster hh id

. save "$z\women_temp.DTA", replace


. restore

.
. tab female

female | Freq. Percent Cum.
------------+-----------------------------------
0 | 26,132 48.37 48.37
1 | 27,888 51.63 100.00
------------+-----------------------------------
Total | 54,020 100.00

. sort cluster hh id

. merge 1:1 cluster hh id using "$z\women_temp.DTA"

Result # of obs.
-----------------------------------------
not matched 34,701
from master 34,701 (_merge==1)
from using 0 (_merge==2)

matched 19,319 (_merge==3)
-----------------------------------------

. tab _m

_merge | Freq. Percent Cum.
------------------------+-----------------------------------
master only (1) | 34,701 64.24 64.24
matched (3) | 19,319 35.76 100.00
------------------------+-----------------------------------
Total | 54,020 100.00

. bys female: tab _m

------------------------------------------------------------ ------------------------------------------------------------ -----------
-> female = 0

_merge | Freq. Percent Cum.
------------------------+-----------------------------------
master only (1) | 17,780 68.04 68.04
matched (3) | 8,352 31.96 100.00
------------------------+-----------------------------------
Total | 26,132 100.00

------------------------------------------------------------ ------------------------------------------------------------ -----------
-> female = 1

_merge | Freq. Percent Cum.
------------------------+-----------------------------------
master only (1) | 16,921 60.67 60.67
matched (3) | 10,967 39.33 100.00
------------------------+-----------------------------------
Total | 27,888 100.00


. bys female2: tab _m

------------------------------------------------------------ ------------------------------------------------------------ -----------
-> female2 = 1

_merge | Freq. Percent Cum.
------------------------+-----------------------------------
matched (3) | 19,319 100.00 100.00
------------------------+-----------------------------------
Total | 19,319 100.00

------------------------------------------------------------ ------------------------------------------------------------ -----------
-> female2 = .

_merge | Freq. Percent Cum.
------------------------+-----------------------------------
master only (1) | 34,701 100.00 100.00
------------------------+-----------------------------------
Total | 34,701 100.00


.
. tab female female2, missing

| female2
female | 1 . | Total
-----------+----------------------+----------
0 | 8,352 17,780 | 26,132
1 | 10,967 16,921 | 27,888
-----------+----------------------+----------
Total | 19,319 34,701 | 54,020


. drop _m

.
.
.* n=8352 matched among supposed males---investigate


Tia Palermo
Re: Problem with gender variable merging household recode with women's recode [message #5637 is a reply to message #5636] Fri, 19 June 2015 03:44 Go to previous message
tiampalermo@gmail.com is currently offline  tiampalermo@gmail.com
Messages: 3
Registered: June 2015
Member
I think I realized the issue: In AIS, the IR file can include both men and women (identified through the aidsex variable), which is not the case for DHS files, where IR indicates women's recode. That would be why there's no male recode (MR) in the AIS file. So the merge is working as it should.

Tia Palermo
Previous Topic: How to merge women and child data set?
Next Topic: Computing the stunting prevalence among U5 children according to the BMI of their mothers
Goto Forum:
  


Current Time: Sat Mar 28 19:59:34 Eastern Daylight Time 2020