The DHS Program User Forum
Discussions regarding The DHS Program data and results
Home » Topics » Service Provision Assessment (SPA) » TZ SPA 2014/15 - Facility Type and Managing Authority
TZ SPA 2014/15 - Facility Type and Managing Authority [message #15419] Mon, 16 July 2018 08:42 Go to next message
mh-uk is currently offline  mh-uk
Messages: 1
Registered: July 2018
Dear Sir or Madam,

regarding the Tanzania SPA 2014/15, I've got the following questions:

A) Weighting / Representativeness:
How did facility ownership / management influence weighting?

The full report states on p17(pdf p42): The sample was designed to provide nationally representative results by facility type and managing authority and regionally representative results for the 25 Tanzania Mainland regions and the 5 Zanzibar regions (a total of 30 survey regions).

TZSR71.DOC (provided with the data files) states: weights were given for facility type and region.

1) Can I conclude from that, that facilities were first weighted by facility type according to national proportions of facility type, and then weighted according to the regions? Or were facilities weighted by facility type according to regional proportions of facility types?

2) Can I conclude from the statement in the report, that sampling of facilities by managing authority / ownership happened on a proportionate-to-size sampling on a national level and was not incorporated in the weighting process?

Regarding 2): I don't have the original sampling frame, but using the current Health Facility Registry, national percentages of facility type seem to be better reflected by weighted percentages, than by unweighted percentages (using numbers from e.g. Table 3.9 p.48/72, and Table 2.1, p.17/42) which would contradict my assumption, wouldn't it?

B) Management Authorities / Ownership
Do your surveys generally include facilities run by the military, police or prison services? The MoHSW in Tanzania subsumes these facilities in its master list as "government-run". If you included these facilities, could some of the client interviews theoretically have been conducted with prisoners or soldiers?

C) Size of Sampling Frame
How many facilities were on the master list?
The full report states the master list of health facilities for TZ mainland and Zanzibar consisted of 7102 verified active health facilities (e.g. page 17/42). On page 4/28 the report lists the distribution of health facilities in Tanzania Mainland. Can I assume that this is not based on the master list used for sampling? If you add up the numbers of facilities, you have 13,723 facilities in total (the total of 6,790 is actually the total of all clinics, not the total of all facilities).

Thank you very much,

[Updated on: Mon, 16 July 2018 10:27]

Report message to a moderator

Re: TZ SPA 2014/15 - Facility Type and Managing Authority [message #15536 is a reply to message #15419] Wed, 08 August 2018 16:09 Go to previous message
Messages: 1516
Registered: February 2013
Senior Member
Responses from two of our experts:
Senior sampler, Dr. Ruilin Ren:

For Q #1, the facility weight was calculated based on selection probability and with correction of non-response by region and by facility type. None of the mentioned proportions were referenced in the weight calculation.
For Q #2, the health facilities were not selected by proportionate-to-size on a national level, it was selected by a stratified simple random sampling with an explicit stratification by region crossing facility type. Management authority was used as implicit stratification.
For your question B, sample did include military and prison running facilities.
For your question C, the size of the sampling frame was stated in several places and as you cited, it was 7102. We selected 1200 HF, among them, 1188 were eligible, so you see the number 1188 in many tables. Note that the sampling weights are all normalized weights to the sample scale, so you cannot conclude the population size based on the weighted number of cases, for HF, as well as for providers and clients.
Hope this is helpful.

Survey manager, Dr. Michelle Winner:

I would like to add few words for question B.
SPA is a facility survey. It is started in the methodology section that interviews are conducted with the managers and the most knowledgeable persons in each facility/service. Clients are interviewed only during exit interviews. Only FP, ANC and services for sick child are eligible for observations and exit interviews. The mentioned didn't have consultations were a soldier or a prisoner was involved in FP, SC and ANC to be observed and interviewed.

Previous Topic: Exit survey question development
Next Topic: SPA response or non-response rates
Goto Forum:

Current Time: Sun May 16 15:22:00 Coordinated Universal Time 2021