The DHS Program User Forum: Sampling

Home » Data » Sampling » Two or three stages?

Show: Today's Messages :: Show Polls :: Message Navigator

Two or three stages? [message #28105]

Mon, 13 November 2023 08:25

Lyly
Messages: 2
Registered: November 2023

Member

Hi,
What is better: 1) developing a sampling frame composed of primary units (PU), and each primary unit contains 5 secondary units (SU), and in this case, we conduct a 3-stage sampling, drawing a PU and within the PU drawing an SU, and within the SU drawing a household for a household survey; or 2) developing a sampling frame for households composed only of secondary units and conducting a 2-stage draw: drawing an SU and within the SU drawing a household?

Report message to a moderator

Re: Two or three stages? [message #28106 is a reply to message #28105]

Mon, 13 November 2023 09:27

Bridgette-DHS
Messages: 3203
Registered: February 2013

Senior Member

Following is a response from Senior DHS staff member, Tom Pullum:

This is not a question about DHS or DHS data. You will have to consult with a sampling statistician, and I am sure that they will want more information than you have provided about your proposed survey.

Report message to a moderator

Re: Two or three stages? [message #28111 is a reply to message #28106]

Tue, 14 November 2023 03:49

Lyly
Messages: 2
Registered: November 2023

Member

Thank you, Bridgette, for your response.
In fact, I asked this question because DHS deals with relatively rare phenomena, and I thought that if we directly sample secondary units without going through primary units, we might cover them better. However, I was concerned that estimates from 2 stages, since the master sample is designed for a 3-stage sampling, could negatively impact the estimates.

Report message to a moderator

Previous Topic:	Statistical Power
Next Topic:	Sampling

Goto Forum:

-=] Back to Top [=-

[ Syndicate this forum (XML) ] [

] [

]

Current Time: Tue Dec 3 14:20:05 Coordinated Universal Time 2024