I want to compare results from two SMART surveys conducted at two different times one in 2014 and other in 2015. In 2014, the sample (multi-stage cluster sampling) was selected from three areas but only from 8 segments. In 2015 same methodology and design was used as 2014, but this time, samples have been selected from 26 segments, covering the whole areas. Now, is it acceptable it the results are compared?
(In any particular areas, the segments or sub-divisions are identical, the design effect was calculated as 1.0)
Dear Anonymous 3445:
It sounds from your question like the two surveys did not cover exactly the same population. If this is true, then comparison of the data from the two surveys is like comparing chalk and cheese (or as we say in the US, apples and oranges). I do not know what "area" and "segment" mean in your context, but if the 2014 sampling universe is contained within the 2015 sampling universe, you could select only the data from the 2015 survey which came from the population which was also covered in 2014. This would result in a valid comparison.
Answered:
8 years agoIn addition, the CDC Statistical Calculators (in Excel format) can be used to compare survey results. They are available on the SMART website for free:
Answered:
8 years ago