On 18 Mar 2002 15:33:54 -0800, [EMAIL PROTECTED] (Ronny Richardson) wrote:
> Statistics textbooks typically show a finite population correction factor > for hypothesis testing and confidence intervals for means and proportions > (sqrt((N-n)/(N-1)). Is there a similar adjustment for the variance? You can google sci.stat.* for previous discussion of FPCs. Here's a rehash of what I say. Statistics textbooks which do mention the FPC typically don't stress enough that they are proper and appropriate for *almost nobody* - especially if you are doing 'research' that you expect to publish, instead of some sort of administration. (Some 'administrators' do research of a special sort; presumably they know what they are doing.) If you are one of the people who have real data to apply, such as -- prognosticator of voting returns for the rest of election evening -- then the simple, textbook formula is almost never going to apply, anyway. You have to start from the basic principles, for a real problem. Thus - Do you have a context where it is useful to have FPC and still refer to variance? If so - for the way you mean it - the answer is yes. But do not try to use that knowledge until you have studied FPCs far more. -- Rich Ulrich, [EMAIL PROTECTED] http://www.pitt.edu/~wpilib/index.html . . ================================================================= Instructions for joining and leaving this list, remarks about the problem of INAPPROPRIATE MESSAGES, and archives are available at: . http://jse.stat.ncsu.edu/ . =================================================================
