Home » Data » Merging data files » merge IR and KR: duplicate line number in IR
| merge IR and KR: duplicate line number in IR [message #23352] |
Thu, 26 August 2021 13:52  |
bun_2019fall
Messages: 9
Registered: August 2021
|
Member |
|
|
Hi DHS user forum,
I was trying to merge IR and KR (one observation in IR correspond to many obs in KR), based on v000, v001, and v002, v003, as suggested by the DHS website. However, I failed to do so because there are duplicate line numbers (v003) in IR files. Thus, my Stata code "1:m" does not work.
I tried to search the web and I did not find any information mentioning if there exists duplicate line number in the same IR files, and why? I wonder if I missed anything here? Any suggestion would be greatly appreciated!
|
|
|
|
|
|
|
|
|
|
| Re: merge IR and KR: duplicate line number in IR [message #23357 is a reply to message #23356] |
Fri, 27 August 2021 11:51  |
bun_2019fall
Messages: 9
Registered: August 2021
|
Member |
|
|
Hi Bruno,
For example, I used "BOIR01FL", and below the dup variable (dup by v000 - country v001-cluster v002-household v003-line number) that I created have values > 0.
use "V:\DHS raw survey/BOIR01DT/BOIR01FL", clear
duplicates tag v000 v001 v002 v003, generate(dup)
tab dup
And below are all the IR files that I found the duplicates (note, I cut the "IR" string from the filename, so the orginal filename should be, for exxample, BR21FL=BRIR21FL). These are all raw DHS files and I did not modified anything in the raw data.
data_name | Freq. Percent Cum.
------------+-----------------------------------
BO01FL.DTA | 6,669 7.13 7.13
BR21FL.DTA | 54 0.06 7.19
CI51FL.DTA | 7,442 7.96 15.14
CM22FL.DTA | 3,329 3.56 18.70
CM31FL.DTA | 4,259 4.55 23.25
CO01FL.DTA | 4,721 5.05 28.30
DR21FL.DTA | 6,354 6.79 35.09
HN52FL.DTA | 196 0.21 35.30
HT31FL.DTA | 4,438 4.74 40.05
IA42FL.DTA | 34,892 37.30 77.35
LK02FL.DTA | 2 0.00 77.35
ML32FL.DTA | 589 0.63 77.98
ML41FL.DTA | 9,300 9.94 87.92
NG21FL.DTA | 4 0.00 87.92
NI22FL.DTA | 4,608 4.93 92.85
NI31FL.DTA | 47 0.05 92.90
PE01FL.DTA | 197 0.21 93.11
PE6IFL.DTA | 272 0.29 93.40
SN02FL.DTA | 237 0.25 93.65
SN4AFL.DTA | 4 0.00 93.66
TD31FL.DTA | 5,923 6.33 99.99
TN02FL.DTA | 8 0.01 100.00
UG01FL.DTA | 2 0.00 100.00
------------+-----------------------------------
Total | 93,547 100.00
I wonder if I missed anything here, as to whether I correctly identified the unique women in each IR files, thus to link back to women who have children in the KR files? Thank you very much again!
[Updated on: Fri, 27 August 2021 11:54] Report message to a moderator |
|
|
|
| Re: merge IR and KR: duplicate line number in IR [message #23359 is a reply to message #23357] |
Fri, 27 August 2021 16:26 |
schoumaker
Messages: 66
Registered: May 2013
Location: Belgium
|
Senior Member |
|
|
Hello,
I looked at the Bolivia file. It seems the v002 variable is indeed always equal to 1. Maybe a dwelling variable was used for creating the caseid and not available in the data set.
I think you can just use the caseid variable (the woman's id) that is available in the IR and the KR file to merge these two files. It seems to work in Bolivia, but I did not check in the other countries.
Best,
Bruno
***
use BOIR01FL.DTA, clear
merge 1:m caseid using BOKR01FL.DTA
***
Bruno Schoumaker
Centre for Demographic Research
Université catholique de Louvain
[Updated on: Fri, 27 August 2021 16:45] Report message to a moderator |
|
|
|
| Re: merge IR and KR: duplicate line number in IR [message #23360 is a reply to message #23359] |
Fri, 27 August 2021 22:45 |
bun_2019fall
Messages: 9
Registered: August 2021
|
Member |
|
|
Hi Bruno,
Thank you for much for looking into the file, and I really appreciated it!
The "caseid" solution is almost perfect, except for one file "DRIR21FL":
use "V:\DHS raw survey/DRIR21DT/DRIR21FL.DTA", clear
*duplicates by caseid
duplicates tag caseid, generate(dup)
sort caseid
tab dup /*4 duplicates*/
*complete duplicates
duplicates tag, gen(dup1)
tab dup1 /*no duplicates*/
The "caseid" itself has 4 duplicates in this dataset, out of all DHS IR files that I have downloaded as of July 2021. I further checked, these observations are not complete duplicates. That said, to proceed with data analysis, I wonder if I should just drop the 4 observations? I wonder if DHS has any best practice regarding this issue? Thank you, again!
|
|
|
|
Goto Forum:
Current Time: Mon Dec 30 13:51:36 Coordinated Universal Time 2024
|
The DHS Program User Forum
Members
Search
Help
Register
Login
Home
