RE: Geo mean vs Mean

Hi Sidonie:
In fact with a non-Gaussian distribution you absolutely should use the
Geometric Mean as it is a weighted mean and less impacted by outliers. With
a Gaussian distribution they will be very similar. I refer you to Geoff
Osborne's wonderful website for a full explanation.
http://jcsmr.anu.edu.au/facslab/statistics.html


Joanne Lannigan, MS
Director, Flow Cytometry Core Facility
University of Virginia
Jordan Hall, Room 7067
P.O. Box 800734
Charlottesville, VA 22908-0734
Office: 434-924-0274
Lab: 434-243-2695
Fax: 434-982-1071
email: joannelannigan@virginia.edu

> -----Original Message-----
> From: Lavergne Sidonie [mailto:sidoniel@svm.vetmed.wisc.edu]
> Sent: Wednesday, April 26, 2006 1:55 PM
> To: Cytometry Mailing List
> Subject: Geo mean vs Mean
> 
> Good afternoon,
> 
> I ran some samples to detect anti-platelet antibodies in
> patient serum.
> 
> I have a very hard time with my stats because some of my
> samples that show a right shift (happy me!) gets a lower
> mean with the cellquest stats (sad me!).
> I tried to analyse my data with another software and
> encountered the same problem.
> 
> However my Geo Means follow what my graphs are showing: an
> increase in geo mean fluorescence when I see a right shift.
> 
> Can I use the Geo Mean?
> 
> Sidonie.