Menu ENN Search
Language: English Français

Sampling households from a community

This question was posted the Assessment and Surveillance forum area and has 2 replies. You can also reply via email – be sure to leave the subject unchanged.

» Post a reply

Anonymous 31952

Normal user

24 Mar 2019, 08:36

Hi there!
I have less than a month to conduct a study. My sample size is 60 and my target population is non pregnant women between 18-49 years. I am recruiting women from a specific community that has between 3000-4000 households. I plan to randomly select 60 households from the community however with no household list to use I am not sure how to do to do this sampling. What would be the best method to use in this case in terms of having a representative sample? Thank you

Mark Myatt

Frequent user

25 Mar 2019, 09:35

Your sample size (n = 60) seems quite small. This size of sample will limit you to simple analyses that will usually lack precision. Anyway ...

You could make your own HH list using mapping and door counting. This may take longer than taking you small sample.

A number of other techniques can be used. The basic EPI selection method can be representative or a wide area if a small sample is taken from a large number of communities (e.g. 7 subjects from 30 communities). Such a sample tends to lack variability compared to a random sample. This can be addressed by increasing the sample size (usually double, sometimes more) and using special analysis methods.

A number of methods can be used to make and EPI type sample representative at the community level. This usually involve both "segmentation" and "broadening" the EPI sample.

Segmentation means dividing the community into non-overlapping contiguous areas and taking part of the sample from each area. If (e.g.) there were 15 city block to sample you would take 60 / 15 = 4 women from each block. This ensures that you sample is taken from all over the community.

Broadening the EPI sample means having sampling rules to avoid selecting immediate neighbours so as to capture more variability than the basic EPI sampling method. Methods like EPI3 and EPI5 (e.g.) use the basic EPI method but take every third or fifth house in a "random walk".

Both segmentation and broadening make the sample capture more variation than a simple EPI cluster sample and used together will give you a good sample.

If you have a large area with many blocks (i.e. more blocks than your sample size) then you can take a sample of blocks. The best way to do this is to take a systematic spatial sample. This ensures that the sample is evenly spread across the community.

Here is the RAM-OP manual. This design uses map-segment-sample techniques. Tou may find something useful in it.

I hope this is of some use.

Anonymous 31952

Normal user

26 Mar 2019, 13:46

Thank you! This is really helpful. I am conducting a pilot study thus the reason for the small sample size.

Back to top

» Post a reply