Steve G. Hilliard steve at cb.uga.edu
Mon Nov 17 18:25:10 EST 1997


It's been one of those days.  I'm not about to say everything has gone 
wrong (I still have to make it home safely) but Murphy rules today.
Now the last straw:  I was in the middle of running an endless rack of
samples for a user who was 3 HOURS late, when I got a call from a grad
student who works for one of the hot PI's on campus.  She ran some samples
6 months ago, and now she wants me to explain how to test for statistical
significance between overlapping populations.  Right now.  Like she has a
presentation tomorrow.  Ignoring my gut reaction to this type of
demand, I embarassed to admit that I've been in flow 5 years and I've 
never been hit with this question.  What IS the best way to approach this
type of data?  If I understood her correctly these are all "positive"
samples, but she wants to assure her prof. that their slight differences
in fluorescence are not significant.  The Elite gives us mean and SD, so
can we just use whatever test you would use to test any other pair of 
populations?  Is there a problem in using these means, which are reported
not as channel numbers but as values ranging from 0.1 to 1000 ?  A simple
answer would be great, but I suppose I'm ultimately after the correct
answer, simple or not.  

Actually right now all I'm after is a stiff scotch, but I'd love 
to come in tomorrow and read all your thoughtful replies.  Thanks in
advance, I love you guys!

Steve--who's too tired to think of anything witty for a closer.....

Steve G. Hilliard        steve at habanero.cb.uga.edu        
If you're a runner you should be dead. 

More information about the Cytometry mailing list