At 09:30 PM 3/5/02 -0500, you wrote:
>  1000 is only 0.0004% of the approx. 250 million
>americans.  Five bees out of a hive of 60,000 is 0.0083%.  This is a
>much larger sample than we get in our political polls, etc.  Smaller
>sample sizes do have a higher margin of error but what really drives the
>sample size in the consistancy of varience (square of the standard
>deviation) of the data.

It is important to note that proper selection of these small samples is
extremely important.  I can hardly poll 1000 individual randomly from
a given county and extrapolate to the entire country.

I don't recall the original message indicating the method of selecting
the 5 random individuals from a given colony.  Were they randomly
selected from the entire cluster (ie. some from the edge, middle, etc.)
or were they simply the first 5 bees to the entrance when you knock
on the hive?  It makes a difference.

There is also such a thing as polling to small a sample.  Calling up
5 random people in the town I work in (population around 30,000) isn't
likely to give me a good idea of income levels, local politics etc.  You
need more than a simple random sample,  you need an evenly
distributed sample.  Normally we take random samples with populations
of people because choosing a evenly distributed sample of people is
difficult, but random selection of a large enough sample with give us
an approximation of an even distribution.

-Tim