Have to agree with Dean here. We sample our honey to buyers by sampling about every 15th drum. The drums are filled from a holding tank which holds about 25 drums. I then blend four of those samples into a single sample which I feel accurately represents the average for a truck load of honey. Sure, there is a risk in doing so but I have confidence in what we produce so I just don't worry about it.
Interestingly enough we got a positive (though quite low) Amitraz reading this fall. I called the packer and asked him if he could double check on that sample as I couldn't imagine how that could be possible. He let me know a few days later that, yes, there had been a mistake and that a lab tech had mislabeled a sample taken from another producer. I haven't had a positive reading for a miticide for 7 years and counting and we haven't had a positive antibiotic reading since discontinuing tylosin use 5 years ago.
Heres my analogy. If you have a room full of people in which only one person has a contagious disease and you take a specimen from each of these people and combine them. Your results will confirm that a contagious disease is present in the room. If, however, you take only a random sampling you may well miss the infected person and get only readings showing no disease.