| Senegal 2005 Source of Family planning method [message #6740] |
Wed, 08 July 2015 07:08  |
lbenova
Messages: 4
Registered: April 2014
|
Member |
|
|
In the 2005 Senegal survey (the source of FP method - question 328 on questionnaire, variable v326 in IR), there are no labels assigned to the various locations in the Stata data file. I tried to go back to the questionnaire to see what the response options were. Here is where more questions arose:
1. The questionnaire lists public sector sources from 11 to 19. In the dataset, there are no observations in code 18 and 3 observations in 19. Does this mean that there were 0 women responding "Strat.Avancee/Equ.Mobile/Agent sante" and the 3 responses in 19 are "Other public"?
2. In the private sector, the first response option on the questionnaire is "Hopital/clinique/cabinet prive" - but it seems that in order to select this option, an enumerator may have circled code 21 or code 22 (the response option breaks over two lines and the lead "..." goes to 22, but 21 is also possible).
In the dataset, codes 21, 22, 23, 25, 26 and 27 have some observations. Were responses 21 and 22 merged during data processing since they both refer to the same response option? Does code 22 in the dataset then refer to response option 23 on the questionnaire (Pharmacy)? Does this also mean that code 24 on the dataset does not exist because there were no women who chose response option 25 on the questionnaire ("Dispensaire relig")? Or was this response code left empty in order to match questionnaire and dataset response code so that codes 25, 26 and 27 on the dataset refer to "dispensaire relig", "agent sante" and "Other private medical", respectively?
Do you know how to match the correct labels to the codes on the dataset?
3. The issue with response codes 21/22 being the same response option in this question also appears on the Senegal surveys from 2010/11, 2012/13 and 2014. How was this issue handled in the IR recode?
It seems for example from the 2012/13 dataset that "pharmacy" (response 23 on questionnaire) was assigned code 22 in the dataset, and so on. Response code 25 in the dataset then does not exist (not in the labelbook) which means that 26 is the same in the questionnaire and the dataset (other private medical). Is this what happened in the affected surveys?
|
|
|
|
| Re: Senegal 2005 Source of Family planning method [message #6786 is a reply to message #6740] |
Wed, 15 July 2015 11:38  |
lbenova
Messages: 4
Registered: April 2014
|
Member |
|
|
I am writing to follow up on this issue as some clarification has been provided by the DHS team through email in the meantime.
1.The 2005 survey labels for variable v326 are
Last source for current users
PUBLIC 10
Government hospital 11
Government health center 12
Government health post 13
Government family planning center 14
Rural maternity clinic (maternité rurale) 15
Government health house (case de santé) 16
Community pharmacy 17
Mobile field workers 18
Other public 19
PRIVATE MEDICAL 20
Private hospital 21
Pharmacy 22
Private doctor 23
Mosque/church dispensary 24
Private Health worker 25
Other medical private 26
OTHER PRIVATE 30
Shop 31
Mosque/church 32
Friend/relative 33
Bar 34
OTHER 96
DK 98
Missing 99
My original question 3 - collapsing of codes 21 and 22 seems to be happening on all surveys beginning in 2005. The v326 response code 21 which contains this collapsed value is labelled "private clinic" but in fact reflects the questionnaire response which are broader than that "Hopital/clinique/cabinet prive".
Hope this helps other Senegal data analysts!
|
|
|
|
The DHS Program User Forum
Members
Search
Help
Register
Login
Home
