Quotes are taken from a message written by Ron to Charles on 07/27/96...
RM>Ok... Now supposed we expand the sample to a larger group.
RM>Your expectation is that all the other factors will even
RM>out; I contend that that is highly unlikely....
In statistics it is recognized that there are uncontrollable factors 
that may influence a study, and as a result a good study establishes, in 
advance, a level of success at which chance is no longer a likely 
factor.  (I did some of these lengthy calculations by hand back in the 
70's - now I can't even remember what the process is called.)  
Mathematical tests are done on the differences in performance between 
the two groups and a determination is made that the findings are either 
significant at the .01 level (differences are likely to be explained by 
uncontrolled factors only 1% of the time) or at the .001 level 
(one-tenth of 1 percent).  When a study reports that there are 
significant differences between the control group and the experimental 
group, that means that the likelihood of the difference being random (or 
created by uncontrolled factors) is 1% or less.  I'm not sure what level 
of significance the STAR study was working with, but the performance 
level of the two groups was different enough that the statisticians were 
comfortable in determining .
Sheila, if you're reading this, do you remember any of the math 
processes involved in this?  My memory is shot.
Chuck
Chuck Beams
Fidonet - 1:2608/70
cbeams@future.dreamscape.com
___
* UniQWK #5290* Of all the things I've lost, I miss my mind the most.
--- Maximus 2.01wb
---------------