results reporting in ENA with non-cluster but SRS sampling

[Gudrun (Concern)Participant]

Hi – we’re considering implementing a survey where it’d be possible to use systematic random sampling (or even simple random sampling) as an overall sampling method (not at 2nd stage). This would obviously help to keep the sample size small.
It’s possible to use the Planning function for random sampling (vs cluster) but what should we do during the analysis stage? We’ll be able to calculate Z-scores for the data set for ENA.
how about the plausibility test? I’d expect some parts should be ok but obviously anything that’d involve the cluster variable would be useless in such a survey.

Also, how will CIs be calculated because they’d usually consider the sampling design (at least that’s what I’d expect) to apply some sort of complex sampling calculation. This again would not be necessary in a survey without clusters.

concretely – we’re looking at a survey set within POC/ camps in S Sudan. If anybody has thoughts about a good sampling strategy there, I’d greatly appreciate this also! (I[m sure there are good examples, I just haven’t searched for them yet.)

Many thanks!
Gudrun

[Blessing MureverwiParticipant]

Dear Gudrun,

It seems that in your setting you are indeed looking at using either simple random sampling (if a complete list of households is available) or systematic random sampling (if households are arranged in a linear pattern, which I doubt in the POC camp). This would be similar to what we often use in some refugee camps where there are not enough zones/blocks to generate clusters. The plausibility test will run in the same way as it would run with a cluster survey. CIs will be calculated by ENA taking into account the design, ie. they will not be complex CIs.The formula is CI=+/-Z x SE, where Z-1.96, and SE is the standard error. I hope this helps.

[SMARTKeymaster]

Thank you for your response Blessing. I would also like to reiterate that in your example you will select ‘Random’ sampling in the ‘Planning’ Tab and the plausibility check will automatically be adjusted accordingly. Kindly note that the Poisson distribution will appear in the Plausibility Check summary table (page 1) but will not be calculated as it is used for cluster sampling. Also, the Design Effect for your survey will be 1.

[Author]

Posts