Further to my last post, a more important fact than education is of course what you actually say and do with that. If people make demonstrably mistaken statements, then they are mistaken no matter what educational background they have.
One mistake I think you have made, for example, is in your description of what an experiment should do:
"...experiments should be conducted in circumstances that replicate those of the claim..."
Very few scientific experiments meet this statement. The basic reason for this is that the world is too complicated to study all at once. While such an experiment might be useful to confirm whether or not svejk might be telling the truth about the outcomes of his dowsing, it would not say anything about the process behind it.
A more reasonable goal of experiment is not just to establish whether or not someone is lying, but to see why it is that they are "telling the truth" -- if not, as seems likely, the whole truth. So, an experimental programme to investigate dowsing realistically would need to be far more controlled than just copying what svejk does. In particular, it should distinguish between my hypotheses, that dowsing is basically an ideomotor effect, and his, that there is more to it than that.
How does my suggested program achieve this? Firstly, it would need some refining, of course -- you can hardly expect me to design a perfect experiment in a period of about ten minutes when I was very tired. You shouldn't dismiss it out of hand either, because one part of the tests described in that link I gave you:
http://undeceivingourselves.org/S-divi.htm
included a control phase to show that the dowsers could find water when dowsing with " ... containers that had their contents clearly visible. All diviners confirmed that their divining abilities were working well." In principle, then, there should be no reason why dowsing doesn't work equally well over bath water (although one refinement would be to take it outside and bury the container under a few feet of soil) -- because those dowsers certainly didn't think otherwise.
With that in mind, my programme up to some refinements is a lot more reasonable, and has the benefit that it can test one part of the theory at a time.
More generally, all experiments do this -- that is, very few experiments in science ever take place in the full "real world". The real world is uncontrollable, and inherently unanalysable, so to make any real progress you take a theory to the lab and remove as many uncontrollable factors as possible, and focus on just one or two factor that you can control. In our case, we can try to isolate, examine and discard or confirm the possibilities of:
-- any physical connections between the dowsing rod and the water;
-- whether the material of the rod has any influence;
-- importance of the height of the rod;
-- importance of the person using the equipment;
-- importance of the terrain to results;
and so on, and so forth. You could not hope to test all this in a single experiment and come up with any meaningful conclusions about dowsing, other than the aforementioned fact that you would confirm svejk's observations to some extent (that he found water), but not learning why he was able to make that observation. Instead, by separating possible contributing factors from each other and testing each in turn, you might be able not only to see whether or not dowsing contributes to the process of finding water, but also how it does, if it does.
In summary, I think you haven't fully understood how the scientific method works, or how to apply it, and I'm not sure I agree with your interpretation of the data either. Both of these considerations are far more important than educational background.