> Aaron makes some good comments, but I don't think he has
> really applied his mind to the problem, yet in anything more
> than a superficial way.

Fair 'nuff.  The thing that most impressed me about Bob's response was the
comment about clustered bees in Alberta at this time of year.  Due to the
closeness of the bees and the fact that they've been clustered for many
months now, one can assume that if any bee has a problem then ALL bees in
that cluster will have that problem.  The assumption/assertion that if
present, any malady would be pretty well homogenized through the entire
cluster is reasonable beekeeping, it's not sound statistics which assume
samples are distinct and independent.  The thing that throws the statistical
analysis off is the concept that bees are not the organism, the colony is
the organism.  An analogy could be assuming a body is cancer free based on a
normal white blood cell count.  A safe bet, but by no means assured.

You asked a statistician's view, I gave one.  Looking at it statistically,
one looks at the ratio of sample set to overall population.  What's the
population of a late-winter cluster in Alberta (race issues aside)?  15,000
bees?  What's 5 out of 15,000 represent?  A pretty flimsy sample set to be
using in statistical science.  Now, are those 5 bees distinct, independent
samples (an important concept in stats)?  Bob's observation about bees in a
cluster makes a point that individual bees are NOT distinct and independent.
In fact, the point of bees in a cluster is the sort of thing that sound
statistics strives for, that the sample set is truly representative of the
total population.  Normally this is achieved through sound sampling (large
sample sets).  It can be stated, and I would not argue that 5 bees in a
cluster is more representative of the true population of 15,000 bees than
would be say 5 individuals in a city of 15,000 people.

So, is no Varroa in the total yard sample significant?  "Bees in a cluster"
says yes, distinct and independent sampling says no.  From a numbers point
of view, no varroa is the least supportable conclusion you can draw.
Numbers of tracheal mites would be higher, hence chance of detection would
also be higher.  Same conclusion:  "Bees in a cluster" says no problem,
statistical science (which treats sampling as independent and distinct) says
you need a bigger sample size.  Nosema (as Bob pointed out) is even less of
a concern.  Nosema would be the most homogenized in the population and most
readily detected

> what is the level of confidence?
A statistician would still say 0.  A commercial beekeeper would (does) stand
confidently that level of concern is low.  Someone very busy will drop out
of the discussion.

> What next?
Don't treat, but keep your samples in case something DOES crop up in one of
your yards, and if so, you can look more closely to see if you have missed
something in you statistical quick glance.

> No guessing.  It can be calculated.
Please share.

/Aa