[Response copied to edstat list] Well, if the sampling procedure was badly conducted, as you write, you may not be able to trust any of the proportions/percentages that you can derive from the sample; since even if it was badly done, the proportions observed may be biased away from the "real" proportions in the population, closer to the values you'd have gotten with a balanced sample.
So if I were doing it, I'd _report_ what the sample shows, but for weighting purposes I'd _use_ census data when it's available. On Wed, 18 Feb 2004, Raoul Kamadjeu wrote: > Thank you very much for your explainations and ideas. > In this study, we follow the WHO stepwise diseases for Non > Communicable Disease which recommends a multilevel systematic random > sampling stratified by age group. The fact is data, the sampling > procedure was badly conducted on the field, if not , we would have > had well balanced sample. Which one would be the most statistically > sound; weighing data using age groups proportions or standardizing > using census pop as reference. > Once again, many thanks for your help < snip, copy of my earlier message > ------------------------------------------------------------ Donald F. Burrill [EMAIL PROTECTED] 56 Sebbins Pond Drive, Bedford, NH 03110 (603) 626-0816 . . ================================================================= Instructions for joining and leaving this list, remarks about the problem of INAPPROPRIATE MESSAGES, and archives are available at: . http://jse.stat.ncsu.edu/ . =================================================================
