Select Language

# Calculating the Average CSI/rCSI for multiple samples

Posted by eholmes on 28 Aug 2015

Hi everyone,

I need advice on calculating the CSI/rCSI score for the total sample when there are multiple samples (stratified sampling). I consulted the 2008 WFP CSI Field Methods Manual and I can find how to calculate the CSI/rCSI for an individual household and how to report the percent or mean for each strategy for the full sample. However, I am looking for the calculation for how to report the average CSI/rCSI for the entire sample.

For example, if I have five host households and the rCSI score for these households are 25, 30, 14, 18, 27 and five IDP households and the rCSI scores are 12, 28, 16, 22, 21. I could calculate it in a couple of different ways:

Option 1:
25+30+14+18+27/5= 114/5=22.8 (host)
12+24+16+22+21/5=95/5=19.0 (IDP)
22.8+19.8/2=21.3

Option 2:
25+30+14+18+27/5= 114/5=22.8 (host)
12+24+16+22+21/5=95/5=19.0 (IDP)
114+95/10=20.9

I recognize that the end result is not extremely different, however, I do want to ensure that I am calculating it correctly and consistently.

Thanks for the help!

Emily Holmes
M&E Analyst
Samaritan's Purse #### Average sample versus sub samples

Posted by Edith on 14 Oct 2015

Hi Emily, thank you for your question.
For option one you would need to apply weights (depending on the sample sizes relative to the sub populations). In lieu of weighting, you are better off using option 2.
I hope this helps.