It’s worth noting that, in other non-scientific (but evidence based) fields, we don’t always have the luxury of direct observation but can still come to a well supported account of past events. I don’t know whether you’ve heard the analogy before, but consider a murder trial:

No-one saw the suspect commit the crime, there’s no experiment to directly test whether he did it. But his fingerprints are on the barrel of a gun that matches the one used against the dead man, he can’t provide an alibi and he was seen just after the murder, 2 blocks away, running like hell and splattered with blood. So we have a theory; this suspect committed the murder. It fits the facts as best we know them. The defence could have another theory that fits the evidence just as well (maybe the real murderer held a gun on the suspect, forced him to shoot the victim, then let him run off, although the question then is why the suspect didn’t mention that). If CCTV footage or a new witness turned up they could overturn the “hedunnit” theory. What they can’t do is say “There’s a gap in your evidence, therefore the suspect was at home all night watching TV”.

To deny that evolution is occurring (macro and micro, they’re the same damn thing) is like the second option there; it doesn’t make sense of the evidence that we have, even if that evidence is limited in some respects. The first option, of an alternate scenario that’s quite similar to the original theory (in terms of what evidence we’d expect to see), that’s like the very real open questions in evolutionary biology.

For those questions, like in the murder trial analogy, we’d need more evidence to differentiate between the possible answers. Science is a great timesulative process, doesn’t claim to have all the answers all at once. What it does offer is the best working model we can come up with based on the evidence that we have so far. Personal incredulity in the suggestions of that evidence is not a valid argument unless you’ve got some observations that don’t accord with the theory.