One problem I have with the "evidence" for scientific theories in general is the often-imprecise use of the word "prediction". To normal people, when you predict something, that means you say what's going to happen before it happens. In science, it often doesn't mean that. It means something more like "our theory would have predicted this, except we already knew about it before we had the theory." People say, for example, that Relativity predicts some oddities in the orbit of Mercury, but these oddities were known long before there was any Theory of Relativity. The justification for this kind of talk is that the Theory of Relativity was not originally formulated to explain the orbit of Mercury and, in fact, had nothing really to do with Mercury or orbits at all, yet it leads to mathematical formulas that do describe the orbit of Mercury. That's a quite impressive accomplishment even if it isn't really prediction.
The case in evolution is considerably less impressive because evolution isn't a mathematical formula with mathematically-determined conclusions. "Evolution predicts that X" generally means something more along the lines of "we have an explanation of X in evolutionary terms that could have been used to predict X if anyone had had any reason to believe the explanation before we saw X." I'm going to give several examples from this Frequently Asked Questions list on evolution.
From the FAQ:
Prediction 3.1: Anatomical parahomologyFor example, bat wings aren't entirely new structures, they are specialized arms and hands; whale flukes and monkey tails are both just specialized tails.
… When one species branches into two species, one or both of the species may acquire new functions. Since the new species must recruit and modify preexisting structures to perform these new functions, the same structure shared by these two species will now perform a different function in each of the two species. This is parahomology.
The obvious problem with this "prediction" is that the facts were known for centuries before the theory of evolution ever came to be. And unlike the case with Relativity and Mercury (1) evolution was inspired partly by this very phenomenon (making it considerably less impressive that evolution happens to explain it) and (2) the explanation doesn't logically follow from the basic idea of evolution anyway. There is nothing in the basic idea of common descent that logically implies anatomical parahomology. If the universe were a bit different, it might be the case that new functionalities always arose from mutations that created entirely new structures. And in such a universe, scientists would be telling us with just as much confidence that evolution predicts that new functions will usually come from brand new structures.
Also from the FAQ:
Prediction 3.3: Anatomical analogyAgain, this "prediction" was known before the theory, the theory was designed with the facts in mind, and it isn't actually a logical inference from evolution. In a different world, scientists would be informing us confidently that genetics is such that in general a given function can only be performed by a given structure.
... Analogy is the case where different structures perform the same or similar functions in different species. Two distinct species have different histories and different structures; if both species evolve the same new function, they may recruit different structures to perform this new function.
Also notice this: when two different functions are performed by the same structure, it is evidence for evolution; if one function is performed by two different structures, that is also evidence for evolution. Just what could have counted as evidence against evolution? At this point, evolution is starting to look suspiciously like the pseudo-sciences that Popper talked about. He said that the difference between science and pseudo-science is that real science makes predictions that might be falsified while pseudo-sciences have explanations that prove them true no matter what.
Continuing with the FAQ
Prediction 3.5: Anatomical suboptimalityThis has the same three problems as the other two objections and the additional problem of not being an objective fact anyway. Biologists don't know enough to say that another way of doing things would have been better (or even possible) because they don't understand all of the variables. Is there any genetic sequence at all that would lead to the better structures they envision? Without knowing that, they are a bit presumptuous to say that the better structure would have been possible. Note that just because there are genetic sequences that lead to the more optimal structures in different organisms, that doesn't mean that they would have been compatible with this organism.
… in gradually evolving a new function, organisms must make do with what they already have. Thus, functions are likely to be performed by structures that would have been arranged differently (e.g. more efficiently) if the final function were known from the outset. "Suboptimality" does not mean that a structure functions poorly. It simply means that a structure with a more efficient design (usually with less superfluous complexity), could perform the same final function equally well.
And again, if scientists had never been able to find instances of what they viewed as suboptimality, they could have explained this in terms of evolution also. In fact, if genetics are as fluid as this argument requires, then the amount of suboptimality would be a function of the mutation rate and survival pressures. With high-enough values for those two factors, we would see little or no suboptimality today and the lack of suboptimality would be claimed as an evidence for evolution.
Go ahead and read the rest of that chapter and the other "predictions". Isn't it ironic that Popper's mechanism to distinguish science from pseudo-science is being drafted to support exactly the kind of arguments that he was trying to throw out of science?
Of course, just because there are bad, pseudo-science arguments for evolution, that doesn't mean that there aren't also some good arguments for evolution. I've got a better FAQ that I'll get to later, but I think I'll spend a bit more time on this one first.