Posted by Bill McCuistion (198.64.206.92) on October 08, 1999 at 21:47:53:
In Reply to: if not a wordplay joke, read enclosed message posted by gary g on October 08, 1999 at 19:52:06:
There is no risk in collecting a population which can become a future sample, only in using sampled data improperly. There are no perfect samples. There is always a risk in sample selection. Identify the sample risks, later. Deal with them once you have the samples and know better what the samples are.
There are numerous ways to cull the population, once it is collected. We (sorta kinda) know who's who. The anecdotal case histories, the blood chem tests, the all-the-rest.
It will never be possible to collect a statistically valid sample of known CH data as the population of CH is unknown.
There is a sampling technique which involves obtaining a preliminary sample so that you can better plan the next sample. I believe that is what I am aiming for, progress -- not perfection.
Ask you sister how to do this. The real world is not a textbook case. Let's work with what we have. For now.