You know @Dee, that is a very good point. Plus sometimes, even when you examine data, the interpretation can be very skewed.
I have in mind the www.beeinformed.org database. It is very valuable, and in fact much of the time, it is all we have. However, compared with clinical research studies, it is woefully lacking and full of bias. When you want to ask a question in medicine, you have to carefully design a study to gather all datapoints on every case, and preferably randomize cases to one intervention, or another, or nothing. If you merely look back at old data, that is a retrospective analysis, and is considered very low quality.
Some beekeepers like Randy Oliver, among others in the academic community, are doing much better designed studies, but they are in the minority and not widely quoted. I am very cautious about taking anecdotal, case-study information as reliable. Of course, my own experience is anecdotal and mostly case-study data for bees, but if somebody comes up with a really well-designed experiment that answers a question, I will consider changing my practice, if the new data tells me to do something better. Just a rambling thought from an old, evidence-based physician.