Actually, from every thing I have read about surveys and statistical accuracy, a survey size of 6000 is more than adequate to come to a conclusion on a population as large as the dog population. It would take less than that simply to achieve the 95% confidence level.
But what also seems to be missing from this discussion is that the people who conducted this public survey also surveyed the national breed clubs,
and got similar results. This survey was aimed to determine from the general population about dogs biting habits, not relying on police reports or hospital records. Smaller dogs may have surprised many, but that is because most of those bites had previously never gone reported.
I tend to agree that the underlying purpose of publishing this study was to let people know the stupidity behind the statistics being used in support of BSL, since the source used would typically rule out pointing at other dogs as being much more susceptable to aggression than the ones being specifically targetted. Then again, there is that other thing about statistics... find me the right interpreter, and I can make that same study say anything I want it to say