The DHS Program User Forum
Discussions regarding The DHS Program data and results
Home » Topics » Service Provision Assessment (SPA) » weights for Malawi 2013-14 facility questionnaire
weights for Malawi 2013-14 facility questionnaire [message #9701] Mon, 09 May 2016 08:56 Go to next message
Ogriv is currently offline  Ogriv
Messages: 16
Registered: March 2015
Member
Hello
I am about to analyse the facility questionnaire, in particular some items from the inventory which I have extracted.

I want to apply weights.
I'm using Stata, which, in order to use the svyset command, needs:
- PSU variable
- sampling weight
- strata variable

I can see that facwt (divided by 1,000,000) is the facility weight variable.
Presumably that is the sampling weight?

But what should I use for PSU and strata variables?

Like in the household surveys, should I use region divided by urban and rural for strata?

And for PSU, it would normally be cluster hv021 in the household surveys...
But what is the equivalent in this SPA raw dataset?

Kind Regards
Sandra
Re: weights for Malawi 2013-14 facility questionnaire [message #9717 is a reply to message #9701] Tue, 10 May 2016 06:40 Go to previous messageGo to next message
Ogriv is currently offline  Ogriv
Messages: 16
Registered: March 2015
Member
In response to my own message, I just want to say that I am also aware that the Malawi SPA was intended as a census.

In theory that means that no weights should apply.

However, they only managed to sample 977 out of 1060 facilities.

So I am still unsure whether to use weights, especially as FACWT is given as a variable in the facility dataset.

Best Wishes
Re: weights for Malawi 2013-14 facility questionnaire [message #9721 is a reply to message #9717] Tue, 10 May 2016 10:28 Go to previous messageGo to next message
Liz-DHS
Messages: 1393
Registered: February 2013
Senior Member
Dear User,
A response from Senior Sampling Expert, Dr. Ruilin Ren:
Quote:

It is true that the Malawi SPA 2014 was a census of all health facilities (HF), the HF weight FACWT is a non-response weight for correction of non-responses. It is better to use the weight variable in your analysis. The SPA survey is slightly different from the DHS survey. For HF survey, there is no cluster variable. But you can declare the HF ID (FACIL) as PSU variable, there is no effect. For providers and clients survey, there is a cluster effect, HF is a cluster of providers and clients. So you should declare HF as cluster. The stratification is defined by crossing region by HF type (REGION crossing FACTYPE). Urban/rural was not taken into account in survey design in SPA surveys.
Hope this is helpful.
Thanks

Re: weights for Malawi 2013-14 facility questionnaire [message #9724 is a reply to message #9721] Tue, 10 May 2016 12:27 Go to previous messageGo to next message
Ogriv is currently offline  Ogriv
Messages: 16
Registered: March 2015
Member
Thanks very much.
That is really helpful.

Sandra
Re: weights for Malawi 2013-14 facility questionnaire [message #11588 is a reply to message #9724] Thu, 12 January 2017 14:20 Go to previous messageGo to next message
kpw9
Messages: 1
Registered: January 2017
Member
In the response from Senior Sampling Expert, Dr. Ruillen Ren on 10MAY2016, the recommendation for the Malawi 2013-2014 SPA survey was to treat facility as the PSU when analyzing the provider and client surveys. Was this decision based on the desire to be conservative with the standard errors? The facilities were not sampled so it is unclear to me why you would treat them as the PSU. Yes, providers and clients are nested within facilities, but if all facilities were selected then it would seem that facility could be treated as a strata, rather than a PSU.
Re: weights for Malawi 2013-14 facility questionnaire [message #11618 is a reply to message #11588] Tue, 17 January 2017 09:33 Go to previous message
Liz-DHS
Messages: 1393
Registered: February 2013
Senior Member
Dear User, A response from Dr. Ruilin Ren, Sampling Expert:
Quote:

In the Malawi case, since it was a census of all health facilities, health facilities can be used as sampling stratum rather than PSU, for providers and clients. In my original answer to the user's question, I did not realize that the Malawi SPA was a census of all health facilities. My answer was a general answer for SPA surveys when health facilities are sampled.
Thanks
Ruilin

Previous Topic: Kenya SPA - missing data
Next Topic: Malaria Service Readiness Index
Goto Forum:
  


Current Time: Fri May 24 01:56:42 Eastern Daylight Time 2019