Home » Data » Weighting data » SSU variables in the Mauritania DHS2019-2021 (The number of pregnant women who delivered their last child in the past five years preceding to the survey in Mauritania)
| SSU variables in the Mauritania DHS2019-2021 [message #28730] |
Thu, 29 February 2024 13:56  |
kinden
Messages: 5
Registered: February 2024
|
Member |
|
|
Dear DHS team,
According to the DHS Household and Sampling Manual, the DHS surveys are two stage surveys. The sampling units for the first stage of selection is called the Primary Sampling Unit (PSU) with probability sampling of EAs; the sampling unit for the second stage of selection is called the Secondary Sampling Unit (SSU) with random sampling of households in the selected EAs.
I try to extract the data to identify the following categories of pregnant women who delivered their last child in the past five years preceding to the Mauritania DHS2019-2021 survey.
(1) The number of pregnant women who delivered their last child after August 2019 in the regions of Hodh Gharbi and Guidimakha
(2) The number of pregnant women who delivered their last child before August 2019 in the regions of Hodh Gharbi and Guidimakha
(3) The number of pregnant women who delivered their last child after August 2019 in other regions than Hodh Gharbi and Guidimakha
(4) The number of pregnant women who delivered their last child before August 2019 in other regions than Hodh Gharbi and Guidimakha
However, I could only get the data regarding the number of women in the sample in the PSU: 39,793 women in 39,793 households.
What is the secondary sampling unit(ssu) variable in DHS data set?
My objective is not to produce sampling error but rather to identify the sample size of the above-mentioned different categories of pregnant women.
Any suggestions would be appreciated.
Thank you!
|
|
|
|
| Re: SSU variables in the Mauritania DHS2019-2021 [message #28735 is a reply to message #28730] |
Fri, 01 March 2024 07:45  |
Bridgette-DHS
Messages: 3199
Registered: February 2013
|
Senior Member |
|
|
Following is a response from Senior DHS Stata Specialist, Tom Pullum:
The variable to identify PSUs is v001 (duplicated as v021) and the variable for households, which are the SSUs, is v002. However, you don't need either of them. The following lines will give a table for the number of births that the sampled women had before August 2019, during August 2019, or after August 2019, for each region (v024), not just the two regions you mentioned.
Note that these are numbers of births to women in the sample. The number of births after August 2019 is affected by when the woman was interviewed. The fieldwork started in November 2019 and was stopped in March 2020 (there were 2 interviews in April 2020), and then resumed for two months in February-March 2021. If you are trying to get at the impact of COVID-19 on fertility, you will have to take the date of interview into account. Only the 5,016 women interviewed in 2021 could possibly show a COVID-19 effect.
* Number of births in the Mauritania 2019-21 survey before/after August 2019, by region
use "C:\Users\26216\ICF\Analysis - Shared Resources\Data\DHSdata\MRBR71FL.DTA", clear
* Find the cmc for August 2019
summarize b3 if b1==8 & b2==2019
* August 2019 is b3=1436
* Construct a variable for the timing of births in the birth histories
gen birthdate=1
replace birthdate=2 if b3==1436
replace birthdate=3 if b3>1436
label define bd 1 "Before August 2019" 2 "In August 2019" 3 "After August 2019"
label values birthdate bd
tab v024 birthdate [iweight=v005/1000000]
Here is the table of weighted frequencies:
| birthdate
region | Before Au In August After Aug | Total
----------------------+---------------------------------+----------
hodh echargui | 5,398.926 40.237882 165.757235 |5,604.9211
hodh gharbi |4,306.5666 28.994562 137.206754 | 4,472.768
assaba | 3,542.996 19.488518 85.435286 | 3,647.92
gorgol | 3,866.476 21.88072 100.96081 | 3,989.317
brakna | 2,995.976 21.032225 97.846587 | 3,114.855
trarza | 1,992.851 16.385234 51.722557 | 2,060.959
adrar |676.115868 2.597507 16.480042 |695.193417
dakhlet nouadhibou | 1,097.09 4.328784 28.115727 | 1,129.534
tagant |867.836632 5.258122 23.18794 |896.282694
guidimagha | 3,860.71 23.182374 104.898378 | 3,988.791
tiris zemour et inchi |589.395148 6.91581 15.450279 |611.761237
nouakchott ouest | 1,339.084 8.220457 146.107834 | 1,493.412
nouakchott nord | 4,009.97 17.350836 347.069054 | 4,374.39
nouakchott sud | 3,567.589 18.178068 302.901025 | 3,888.668
----------------------+---------------------------------+----------
Total | 38,111.58 234.051099 1,623.14 | 39,968.77
|
|
|
|
|
|
| Re: SSU variables in the Mauritania DHS2019-2021 [message #28841 is a reply to message #28782] |
Mon, 18 March 2024 08:15 |
Bridgette-DHS
Messages: 3199
Registered: February 2013
|
Senior Member |
|
|
Following is a response from Senior DHS staff member, Tom Pullum:
We apologize for the delay in this response.
The variable for region, v024, has the following categories:
. label list V024
V024:
1 hodh echargui
2 hodh gharbi
3 assaba
4 gorgol
5 brakna
6 trarza
7 adrar
8 dakhlet nouadhibou
9 tagant
10 guidimagha
11 tiris zemour et inchiri
12 nouakchott ouest
13 nouakchott nord
14 nouakchott sud
Nouakshott is divided into west, north, and south. The weighted and unweighted numbers of women in the regions is obtained with the IR file:
. use "...MRIR71FL.DTA"
. tab v024
region | Freq. Percent Cum.
------------------------+-----------------------------------
hodh echargui | 1,327 8.44 8.44
hodh gharbi | 1,402 8.92 17.37
assaba | 1,380 8.78 26.15
gorgol | 1,449 9.22 35.37
brakna | 1,454 9.25 44.62
trarza | 1,320 8.40 53.02
adrar | 808 5.14 58.16
dakhlet nouadhibou | 643 4.09 62.26
tagant | 920 5.85 68.11
guidimagha | 1,786 11.37 79.48
tiris zemour et inchiri | 787 5.01 84.49
nouakchott ouest | 728 4.63 89.12
nouakchott nord | 910 5.79 94.91
nouakchott sud | 800 5.09 100.00
------------------------+-----------------------------------
Total | 15,714 100.00
. tab v024 [iweight=v005/1000000]
region | Freq. Percent Cum.
------------------------+-----------------------------------
hodh echargui |2,033.68582 12.94 12.94
hodh gharbi | 1,579.5074 10.05 22.99
assaba | 1,249.2951 7.95 30.94
gorgol | 1,292.7495 8.23 39.17
brakna | 1,281.9262 8.16 47.33
trarza | 961.483433 6.12 53.45
adrar | 297.682976 1.89 55.34
dakhlet nouadhibou | 538.786626 3.43 58.77
tagant | 349.098656 2.22 60.99
guidimagha | 1,243.5337 7.91 68.91
tiris zemour et inchiri | 273.47023 1.74 70.65
nouakchott ouest | 792.569719 5.04 75.69
nouakchott nord | 2,073.0319 13.19 88.88
nouakchott sud | 1,747.1787 11.12 100.00
------------------------+-----------------------------------
Total | 15,714 100.00
However, in order to get the 5016 that you refer to, I have to go to the BR file, in which the cases are all the children in the birth histories:
. use "...MRBR71FL.DTA", clear
. tab v024
region | Freq. Percent Cum.
------------------------+-----------------------------------
hodh echargui | 3,560 8.95 8.95
hodh gharbi | 3,865 9.71 18.66
assaba | 3,902 9.81 28.46
gorgol | 4,355 10.94 39.41
brakna | 3,467 8.71 48.12
trarza | 2,841 7.14 55.26
adrar | 1,832 4.60 59.86
dakhlet nouadhibou | 1,388 3.49 63.35
tagant | 2,300 5.78 69.13
guidimagha | 5,467 13.74 82.87
tiris zemour et inchiri | 1,800 4.52 87.39
nouakchott ouest | 1,279 3.21 90.61
nouakchott nord | 1,936 4.87 95.47
nouakchott sud | 1,801 4.53 100.00
------------------------+-----------------------------------
Total | 39,793 100.00
. tab v024 [iweight=v005/1000000]
region | Freq. Percent Cum.
------------------------+-----------------------------------
hodh echargui | 5,604.9211 14.02 14.02
hodh gharbi | 4,472.7679 11.19 25.21
assaba | 3,647.9203 9.13 34.34
gorgol | 3,989.3172 9.98 44.32
brakna | 3,114.8548 7.79 52.12
trarza | 2,060.9592 5.16 57.27
adrar | 695.193417 1.74 59.01
dakhlet nouadhibou | 1,129.5344 2.83 61.84
tagant | 896.282694 2.24 64.08
guidimagha |3,988.79101 9.98 74.06
tiris zemour et inchiri | 611.761237 1.53 75.59
nouakchott ouest | 1,493.412 3.74 79.33
nouakchott nord |4,374.38955 10.94 90.27
nouakchott sud | 3,888.6678 9.73 100.00
------------------------+-----------------------------------
Total | 39,968.772 100.00
By adding the numbers in regions 12, 13, and 14, weighted and unweighted, in the IR and BR files, I found that the 5016 is the sum of the weighted frequencies in the BR file. I have highlighted those numbers in yellow. In other words, I can confirm that there are 5016 BIRTHS, NOT WOMEN, and these are WEIGHTED frequencies, NOT UNWEIGHTED.
You also ask for "the number of the last child born before/after August 2019 among the classified pregnant women by region". I have tried to understand your question but can you clarify it? Are you trying to match a table in the final report, or are you doing something new?
As before, I do not know why you are particularly interested in August 2019, and I don't know whether you intend to include that month 2019 in "before" or "after". I don't know whey you are limiting yourself to women who are pregnant, which I interpret to mean pregnant at the time of the survey.
If I interpret your question literally, you want to restrict yourself to women who are pregnant at the time of the survey (v213=1), and to the most recent birth (bidx=1) and you want to know how many of those births were before or after August 2019, by region. Here are the Stata lines for doing that. Let us know if you want something else.
use "...MRBR71FL.DTA", clear
* Find the cmc for August 2019
summarize b3 if b1==8 & b2==2019
* August 2019 is b3=1436
* Construct a variable for the timing of births in the birth histories
gen birthdate=1
replace birthdate=2 if b3==1436
replace birthdate=3 if b3>1436
label define bd 1 "Before August 2019" 2 "In August 2019" 3 "After August 2019"
label values birthdate bd
tab v024 birthdate if bidx==1 & v213==1 [iweight=v005/1000000]
. tab v024 birthdate if bidx==1 & v213==1 [iweight=v005/1000000]
| birthdate
region | Before Au In August After Aug | Total
----------------------+---------------------------------+----------
hodh echargui |149.223643 2.148514 0 |151.372157
hodh gharbi |151.271238 0 0 |151.271238
assaba | 76.345168 0 0 | 76.345168
gorgol |102.743707 0 0 |102.743707
brakna | 42.965639 0 0 | 42.965639
trarza | 53.324355 0 0 | 53.324355
adrar | 13.528143 0 0 | 13.528143
dakhlet nouadhibou | 26.032095 0 0 | 26.032095
tagant | 23.939958 0 0 | 23.939958
guidimagha | 81.070053 0 .723028 | 81.793081
tiris zemour et inchi | 15.290715 0 0 | 15.290715
nouakchott ouest | 17.218033 0 .792849 | 18.010882
nouakchott nord | 91.418925 2.031406 6.741004 |100.191335
nouakchott sud | 83.416651 5.80855 5.440926 | 94.666127
----------------------+---------------------------------+----------
Total |927.788323 9.98847 13.697807 | 951.4746
|
|
|
|
| Re: SSU variables in the Mauritania DHS2019-2021 [message #28914 is a reply to message #28841] |
Wed, 27 March 2024 17:45 |
kinden
Messages: 5
Registered: February 2024
|
Member |
|
|
Thank you so much once again, Bridgette.
It is very useful to confirm that 5016 is the sum of the weighted frequencies in the BR file.
And yes, I would like to restrict the data to women who are pregnant at the time of the survey (v213=1), and to the most recent birth (bidx=1) and I want to know how many of those births were before or after August 2019, by region, assuming that most pregnant women delivered one baby at their last birth.
The reasons why I am interested in before/after August 2019 are that the intervention of my interest has started since December 2018 (9 months before August 2019) and that I would like to analyze key issues around ANC service delivery.
What I still do not understand from your two useful responses is why the number of livebirths after August 2019 is different between the data from the first response (1,623.14 - weighted) and those from the second response (13.697 - unweighted?)?
These numbers are supposed to be similar as no women will be able to deliver another baby from August 2019 onwards during the survey period, non?
* Number of births in the Mauritania 2019-21 survey before/after August 2019, by region
use "C:\Users\26216\ICF\Analysis - Shared Resources\Data\DHSdata\MRBR71FL.DTA", clear
* Find the cmc for August 2019
summarize b3 if b1==8 & b2==2019
* August 2019 is b3=1436
* Construct a variable for the timing of births in the birth histories
gen birthdate=1
replace birthdate=2 if b3==1436
replace birthdate=3 if b3>1436
label define bd 1 "Before August 2019" 2 "In August 2019" 3 "After August 2019"
label values birthdate bd
tab v024 birthdate [iweight=v005/1000000]
Here is the table of weighted frequencies:
| birthdate
region | Before Au In August After Aug | Total
----------------------+---------------------------------+--- -------
hodh echargui | 5,398.926 40.237882 165.757235 |5,604.9211
hodh gharbi |4,306.5666 28.994562 137.206754 | 4,472.768
assaba | 3,542.996 19.488518 85.435286 | 3,647.92
gorgol | 3,866.476 21.88072 100.96081 | 3,989.317
brakna | 2,995.976 21.032225 97.846587 | 3,114.855
trarza | 1,992.851 16.385234 51.722557 | 2,060.959
adrar |676.115868 2.597507 16.480042 |695.193417
dakhlet nouadhibou | 1,097.09 4.328784 28.115727 | 1,129.534
tagant |867.836632 5.258122 23.18794 |896.282694
guidimagha | 3,860.71 23.182374 104.898378 | 3,988.791
tiris zemour et inchi |589.395148 6.91581 15.450279 |611.761237
nouakchott ouest | 1,339.084 8.220457 146.107834 | 1,493.412
nouakchott nord | 4,009.97 17.350836 347.069054 | 4,374.39
nouakchott sud | 3,567.589 18.178068 302.901025 | 3,888.668
----------------------+---------------------------------+--- -------
Total | 38,111.58 234.051099 1,623.14| 39,968.77
use "...MRBR71FL.DTA", clear
* Find the cmc for August 2019
summarize b3 if b1==8 & b2==2019
* August 2019 is b3=1436
* Construct a variable for the timing of births in the birth histories
gen birthdate=1
replace birthdate=2 if b3==1436
replace birthdate=3 if b3>1436
label define bd 1 "Before August 2019" 2 "In August 2019" 3 "After August 2019"
label values birthdate bd
tab v024 birthdate if bidx==1 & v213==1 [iweight=v005/1000000]
. tab v024 birthdate if bidx==1 & v213==1 [iweight=v005/1000000]
| birthdate
region | Before Au In August After Aug | Total
----------------------+---------------------------------+--- -------
hodh echargui |149.223643 2.148514 0 |151.372157
hodh gharbi |151.271238 0 0 |151.271238
assaba | 76.345168 0 0 | 76.345168
gorgol |102.743707 0 0 |102.743707
brakna | 42.965639 0 0 | 42.965639
trarza | 53.324355 0 0 | 53.324355
adrar | 13.528143 0 0 | 13.528143
dakhlet nouadhibou | 26.032095 0 0 | 26.032095
tagant | 23.939958 0 0 | 23.939958
guidimagha | 81.070053 0 .723028 | 81.793081
tiris zemour et inchi | 15.290715 0 0 | 15.290715
nouakchott ouest | 17.218033 0 .792849 | 18.010882
nouakchott nord | 91.418925 2.031406 6.741004 |100.191335
nouakchott sud | 83.416651 5.80855 5.440926 | 94.666127
----------------------+---------------------------------+--- -------
Total |927.788323 9.98847 13.697807 | 951.4746
Kind regards,
Kazumi
|
|
|
|
| Re: SSU variables in the Mauritania DHS2019-2021 [message #28920 is a reply to message #28914] |
Thu, 28 March 2024 13:07 |
Bridgette-DHS
Messages: 3199
Registered: February 2013
|
Senior Member |
|
|
Following is a response from Senior DHS staff member, Tom Pullum:
The table with 1, 623.14 births after August 2019 was constructed with this command: "tab v024 birthdate [iweight=v005/1000000]". The table with just 13.697 births after August 2019 was constructed with "tab v024 birthdate if bidx==1 & v213==1 [iweight=v005/1000000]". Both frequencies were weighted, but the second and much smaller frequency was limited to the most recent birth (bidx=1) and to women who are pregnant at the time of the survey (v213=1). They are a very small subset of the births in the first table. These tables were what I understood you to be requesting.
You say that you are trying to measure the potential impact of an intervention related to ANC care. I don't see how either of these tables could be used to describe such an impact. This is an interesting question, but this survey may not be appropriate for answering it. I would like to help but the question is outside the scope of the user forum. I hope other users can help.
Here is a journal article that may help, but it uses two successive surveys:
Mallick, Lindsay, Trinadh Dontamsetti, Thomas Pullum, and Julia Fleuret. 2019. Using the Uganda Demographic and Health Surveys from 2011 and 2016 to assess changes in Saving Mothers, Giving Life intervention districts. J. of Global Health Research 3. doi:10.29392/joghr.3.e2019026.
|
|
|
|
|
|
| Re: SSU variables in the Mauritania DHS2019-2021 [message #28945 is a reply to message #28936] |
Mon, 01 April 2024 08:20 |
Bridgette-DHS
Messages: 3199
Registered: February 2013
|
Senior Member |
|
|
Following is a response from Senior DHS staff member, Tom Pullum:
My previous response said what was the difference between the two tables. I'm still not quite sure what your hypothesis is, about the difference between Nouakchott and the rest of Mauritania, but the comparisons that I believe you are trying to make would have to take account of the date of interview, which affects the length of time after August 2019 that the woman could have had a birth.
As you would know, the data collection for Mauritania was spread over 3 calendar years, 2019-2021. Most of the fieldwork was between late November 2019 and March of 2020. Because of Covid, fieldwork was suspended, and was not resumed and completed until February-March 2021. As it happens, the three areas that make up Nouakchott were the only areas that were surveyed in 2021.
Here is a table that gives the century month code (cmc, hv008) and the number of interviews in each month of data collection. This table has households as units and is unweighted. Ignore the "totals" row and column:
. tab hv006 hv007, summarize(hv008) means freq
Means and Frequencies of date of interview (cmc)
month of | year of interview
interview | 2019 2020 2021 | Total
-----------+---------------------------------+----------
1 | . 1441 . | 1441
| 0 2540 0 | 2540
-----------+---------------------------------+----------
2 | . 1442 1454 | 1443.591
| 0 2545 389 | 2934
-----------+---------------------------------+----------
3 | . 1443 1455 | 1447.8632
| 0 2009 1369 | 3378
-----------+---------------------------------+----------
11 | 1439 . . | 1439
| 115 0 0 | 115
-----------+---------------------------------+----------
12 | 1440 . . | 1440
| 2691 0 0 | 2691
-----------+---------------------------------+----------
Total | 1439.959 1441.9251 1454.7787 | 1443.3902
| 2806 7094 1758 | 11658
The cmc's for February-March 2021 are 1454 and 1455. In those months, 389 and 1369 households, respectively, were interviewed.
Next, still with households as units, I get the number of interviews by region and cmc as follows:
tab hv024 hv008
| date of interview (cmc)
region | 1439 1440 1441 1442 1443 1454 1455 | Total
----------------------+-----------------------------------------------------------------------------+----------
hodh echargui | 14 498 281 218 136 0 0 | 1,147
hodh gharbi | 6 133 266 402 219 0 0 | 1,026
assaba | 9 325 279 283 198 0 0 | 1,094
gorgol | 14 263 263 267 129 0 0 | 936
brakna | 6 246 298 263 239 0 0 | 1,052
trarza | 15 254 275 292 190 0 0 | 1,026
adrar | 7 230 157 134 140 0 0 | 668
dakhlet nouadhibou | 8 141 147 141 231 0 0 | 668
tagant | 14 226 241 197 51 0 0 | 729
guidimagha | 16 279 169 118 245 0 0 | 827
tiris zemour et inchi | 6 96 164 230 231 0 0 | 727
nouakchott ouest | 0 0 0 0 0 103 479 | 582
nouakchott nord | 0 0 0 0 0 174 417 | 591
nouakchott sud | 0 0 0 0 0 112 473 | 585
----------------------+-----------------------------------------------------------------------------+----------
Total | 115 2,691 2,540 2,545 2,009 389 1,369 | 11,658
To me, it is notable that the date collection is neatly divided, geographically, into pre-covid and post-covid, giving a kind of natural experimental design for potential effects of covid. But I don't see this as a good design for assessing the impact of an intervention in August 2019 in Nouakchott. Any potential effect of that intervention will be completely confounded with the potential effect of covid during the interruption to fieldwork.
If you have other questions, perhaps other users can help.
|
|
|
|
Goto Forum:
Current Time: Sun Nov 24 20:46:46 Coordinated Universal Time 2024
|
The DHS Program User Forum
Members
Search
Help
Register
Login
Home
