Hi Allen:

I am surprised at the quick "not significant" answers from a number of
people who did not present any mathematical analysis of the results. They
are in fact significant:

We have a sample of 1750 bees selected at random from a much larger
population of bees. This is certainly not a small sample size. To decide the
significance of the result (no varroa), reason as follows.

1) Suppose that the general population has an infestation rate of 1 infested
bee per hundred.

The probability of a bee selected at random being mite free is 99/100.

The probability of two being both mite free is 99/100x99/100=.98

The probability of 1750 of them all being mite free
is (.99)x(.99)....x(.99) = .99^1750 = 2.3x10^ -8) = .0000023%

This means that the result is very improbable if there is an infestation
rate as high as 1 mite per 100 bees.

If we suppose an infestation rate of 1 mite per 1000 bees, the probability
of getting a null result is .999 ^ 1750 = .174 = 17.4%

This means that the obtained result has a reasonable chance of happening
with an infestation rate as low as 1 mite per 1000 bees.

As to the probability of error, there are, of course, many possibilities.
The complete absence of any disease is worrisome. One might like to do
another test, doing the work in house.

As to the advisability of treatment, you should do just the same as you did
last year!



                                    Best regards

                                    Donald Aitken