Quotes are taken from a message written by Ron to Charles on 07/20/96...
RM>RM>If the factors are present, the size of the
RM>RM>sample, it seems to me, would be immaterial....
RM>
RM>CB>If you drop a marble into a bag with one other marble in it, you double 
RM>CB>the weight of the marbles.  If you put a marble into a bag with 
RM>CB>thousands of others, the increase in weight is trivial.
RM>
RM>Heh... We're obviously going to be splitting hairs here,
RM>and I'm quite sure that we're pretty much in agreement...
RM>HOWEVER.... I can't help wanting to interject that if
RM>you fill a bag with marbles, and 20% of the marbles in 
RM>the pile you draw from are, let's say, cracked, then the 
RM>size of the bag you put them into will have little effect 
RM>on the "crackedness" of your test sample.
True, but remember - we're actually filling 2 very large bags (schools 
all across the state participated in the study) and the marbles are 
being randomly assigned to the bags (teachers and students were randomly 
assigned to the control and experimental groups).  The cracked marbles, 
although different than normal, must be included as part of the study as 
they are naturally occuring everywhere - the study should not be 
negatively influenced by their presence.  
RM>The obvious difference in our points of view is that you see the
RM>uncontrolled factors as being fixed in NUMBER, while I'm
RM>inclined to see them as fixed in FREQUENCY...
I don't see where you're headed with this...can you draw me a map?  I 
don't know that I ever considered the number of uncontrolled factors to 
be limited in frequency or number, only that the impact of these factors 
would be inconsequential in a large enough study as there would be 
balance on each side.
RM>Statistics is predicated on frequencies, percentages, etc... Makes a 
RM>huge difference in the outcome!      ;-)
I still need something more concrete in order to follow this.  If i 
understand you at all, it seems that you feel the unpredictability of 
human nature makes studies of this nature to be very unreliable.  If 
that's the case, then how can one explain the degree of success 
experienced by public pollsters in predicting elections when they are 
based on only a few hundred samples?
I think human behavior is more standardized than you do, I guess.
Chuck Beams
Fidonet - 1:2608/70
cbeams@future.dreamscape.com
___
* UniQWK #5290* Some days it's not worth chewing through the restraints.
--- Maximus 2.01wb
---------------