Thanks Sqad -- it's hard to make some of my post clear because I don't do much String Theory work myself so don't understand it perfectly, but I suppose the main point is that it is a mathematical tool more than a scientific theory.
The problem with Statistics is a good one to raise. There are I think two separate problems, actually -- the first thing is that some statistics are just utter rubbish, or have been deliberately manipulated or massaged or otherwise made to give a particular impression when it was known to be a false one. Some people have cynical reasons for lying. It's most obvious in the medical industry, where naturally a drug that looks good will make you more money than one that does not.
There is a second problem, and that is that frankly most people just don't get it. This is not to insult anyone's intelligence, but a full statistical analysis is often very difficult to follow, and uses complicated methods that, while well-understood by those who use them, are almost impossible to follow for everyone else. How then is the layman to evaluate these sorts of tests? They just can't, really, and almost inevitably I suppose a certain amount of "faith" enters -- you trust that the people who perform these tests aren't lying through their teeth and relying on you not to bother looking over the work thoroughly, and usually this is the case.
The issue in the above is not that "most people are thick", by the way -- it's simply a true statement that most people haven't devoted a large amount of time to learning the mathematical techniques necessary to understand how a particular result was achieved. And usually it's hard to explain how it was, too. In one example I saw yesterday it was shown in a very sketchy way how it was possible to extract six different values from the same data, when my own relatively basic knowledge of statistical methods would lead me to expect that it's usually one measurement at a time. How this is achieved I don't entirely follow, but it is at any rate a complicated mathematical procedure. Most people just don't have the time to learn this even if they wanted to.
The end result is that most of the Scientific work going on at the moment is possible to explain to the layman, but not necessarily easy to justify. And as long as that is the case, then it's very hard to distinguish the "real Science", where the statistical tests have been correctly applied to the correct data with nothing embarrassing hidden away deliberately, to the pseudoscience and to the people who are lying to you behind the veil of "we did this properly". And while that remains the case, lots of potentially dangerous myths will hang around. Those people who say that sometimes scientists lie or are dishonest are perfectly correct, of course. But it can then be used to justify other, even more dubious claims. In the world of Physics this isn't all that dangerous -- people will just believe something that's wrong -- but in the world of medicine where lives and livelihoods are at stake, we can see all sorts of horrific consequences. The MMR scandal in this country, the anti-vaccine movement in general, rebellions against Polio in Afghanistan and African countries as some sort of Western conspiracy, the flagrant misuse of statistics in cases of Sudden Infant Death Syndrome in the 1990s that led to innocent mothers being convicted of murder... and so on and so forth.
The answer to the original question might seem a little defeatist, and maybe even a little pompous. But I think that the best way to be able to tell good Science from bad is to have a decent amount of scientific training. Equally, doctors are usually better placed to know good medicine from bad. This doesn't always hold, of course. But usually Scientists know how to do Science better than non-scientists do.