Re: Basic caution about cancer clusters

["Sandy Perle" <sandyfl@earthlink.net>]

Chris Alston asked:

> Aren't you talking about the median, rather than the arithmetic mean?
> Simply knowing the average doesn't tell you anything about the distribution
> of the data points, does it?

You are correct. In a statistical test it's the variance that most 
often determines whether or not a process is stable or not. If the 
variances between two samples isn't considered equal (in statistical 
terms), then there is no reason to even look at the mean of the 
samples. Many tests can assess this, but we only need to look at the 
simple tests, such as "F" test, "Z" test and "t" test to show than 
many of these studies have no statistical validity. 

------------------------------------------------------------------------
Sandy Perle					Tel:(714) 545-0100 / (800) 548-5100   				    	
Director, Technical				Extension 2306 				     	
ICN Worldwide Dosimetry Division		Fax:(714) 668-3149 	                   		    
ICN Biomedicals, Inc.				E-Mail: sandyfl@earthlink.net 				                           
ICN Plaza, 3300 Hyland Avenue  		E-Mail: sperle@icnpharm.com          	          
Costa Mesa, CA 92626                                      

Personal Website:  http://www.geocities.com/capecanaveral/1205
ICN Worldwide Dosimetry Website: http://www.dosimetry.com

************************************************************************
The RADSAFE Frequently Asked Questions list, archives and subscription
information can be accessed at http://www.ehs.uiuc.edu/~rad/radsafe.html