Home » Topics » Service Provision Assessment (SPA) » SPA survey
SPA survey [message #23500] |
Wed, 29 September 2021 00:13 ![Go to next message Go to next message](/theme/default/images/down.png) |
foysal
Messages: 9 Registered: August 2021
|
Member |
|
|
Dear Sara,
Hope you are well.
I want to create a variable named quality assurance activities (o=no, 1=yes) using Malawi SPA 2013-14. I have tried with the following codes in STATA:
tab sw=facwt/10000000
(83 missing values generated)
tab q440 [iw=sw],m
routine qa activities freq. percent
yes 472.936149 48.41
no 496.102203 50.78
don't know 7.9616701 0.81
total 977.000022 100.00
tab q441 [iw=sw],m
official records of qa activities in past year freq. percent
yes 276.25236 28.28
no records not maintained 196.683789 20.13
. 504.063873 51.59
total 977.000022 100.00
tab q442 [iw=sw],m
official records of qa activities in past year-observed freq. percent
observed 132.525358 13.56
reported, not seen 143.727002 14.71
. 700.747661 71.72
total 977.000022 100.00
From the variable q442, it is clear to me that only 13.56% facilities routinely carries out quality assurance activities and had documentation of a recent quality assurance activity. But my question refers to the percentage of facilities having no quality assurance activities. Will I count all missing facilities (n=700.747661) in q442 as "having No quality assurance activities" (i.e. merge with category "reported, not seen") rather than as missing values?
If I count all missing values and merge with category 2 (reported, not seen) then the percentage of facilities having no quality assurance activities would be 86.44%. Is it correct?
If I consider all missing facilities as "No" category, then would it be draw any misleading conclusion particularly in statistical modelling?
I would appreciate any help with this matter?
Sincerely,
Foysal
|
|
|
Re: SPA survey [message #23515 is a reply to message #23500] |
Wed, 29 September 2021 09:58 ![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) |
SaraDHS
Messages: 44 Registered: December 2020
|
Member |
|
|
Hello Foysal,
I think that the response to this question really depends on your research question. If you are particularly interested in the documentation of the QA activities, then it would make sense to limit your sample to the facilities which had any QA activities, or the facilities which had any QA activities and said they had records of them. If you are interested in whether QA activities were implemented, then you would include all facilities. From the first line of your question you ask about creating a binary variable named "quality assurance activities". In my opinion, using the response from q440 would be appropriate for this kind of variable, and dropping any further questions about documentation.
The approach you are suggesting mixes two different categories of responses: "No documentation available" and "No QA activities held". I wouldn't recommend this approach, since these two categories are different and combining them may lead to incorrect estimates.
Best,
Sara
Sara Riese, PhD
Senior Demographic and Health Researcher, DHS Program
|
|
|
|
Re: SPA survey [message #26435 is a reply to message #26336] |
Mon, 20 March 2023 14:40 ![Go to previous message Go to previous message](/theme/default/images/up.png) |
SaraDHS
Messages: 44 Registered: December 2020
|
Member |
|
|
Hello,
The Stata files that end in SP are the raw SPA datasets, so the variables have not been standardized. The variables in the raw files correspond more or less to the question numbers in the questionnaire. So q300 in the raw data set is the availability of services, which is question 300 in the facility inventory on page 417 in Appendix D of the final report, q303 is whether or not FP services are available on the day of the interview, which is question 303 in the questionnaire.
Unfortunately, there is no Guide to SPA statistics, so you'll need to look at the questionnaire, as well as the tables and footnotes in the final report to determine how to calculate them. If you are unable to match the tables in the final report, you can look here in the user forum, where we've answered many questions about specific indicators. If you can't find the information already in the user forum, then feel free to post a question yourself.
Best,
Sara
Sara Riese, PhD
Senior Demographic and Health Researcher, DHS Program
|
|
|
Goto Forum:
Current Time: Mon Aug 26 10:16:34 Coordinated Universal Time 2024
|