The credibility of the author is relevant to the author's argument/results/conclusions
to the extent that the argument depends on points whose plausibility requires faith in the author's honesty and ability.
For example, suppose my thesis is, "Running the Miller-Rabin test with a=2 is sufficient to determine primality for all numbers up to 2047." If my argument is, "I have verified this myself", then my argument depends entirely on the reader's trust in me. If my argument is, "Here is a C program that tests all numbers up to 2047; its output shows no exceptions", then my argument depends on the assumptions that a) this C program works as intended and b) when run, this program produces the desired output.
In the second case, (a) can be verified by someone who understands the Miller-Rabin test and can understand C code, and (b) can be verified by anyone with a (trusted) C compiler. These things are not particularly hard to do for an expert (or amateur) in the field. So, even if you don't know anything about the Miller-Rabin test or about C, you can find other people who do, and they will be able to tell you whether my argument is correct; my own credibility is useless for evaluating my argument unless you are too lazy to find yourself a trustworthy mathematician-programmer. And journalists are certainly not lazy in this respect--they interview experts all the time, and do so in this article.
The credibility of the scientist is irrelevant until you find a part in his argument that depends on his credibility. Like a part where he carries out an experiment that requires careful controlling to be accurate, and you have reason to think he is sloppy (inherently or deliberately). Or where he collects a bunch of data and you have reason to think it might be inaccurate (again, due to clumsiness or malice). Here, it seems to be (and I quote from the story--two pages in, I might mention):
"Dr. Jennifer Sass, a senior scientist with the health group at the Natural Resources Defense Council, says that while the Bromenshenk/Army study is interesting, it fails to ask the underlying question "Why are colonies dying? Is it because they're getting weak? People who have HIV don't die of HIV. They die of other diseases they get because their immune systems are knocked off, making them more susceptible." In other words, pesticides could weaken the bees -- and then the virus/fungus combination finishes them off. That notion, however, is not explored in the new study."
There are some mutterings about pesticides maybe killing bees earlier, but it's not clear--wasn't to me, anyway--what this has to do with alleged problems in the author's study. But this is the part where we (finally) see what part of the study depends on his credibility. It is the sort of implied conclusion that the fungus/virus combination is the end of the story, that this fungus and virus just proliferated somehow and the solution is to get rid of them. The flaw, or rather the incomplete part of the study, is that pesticides might somehow make it easier for the virus/fungus to infect the bees, and this possibility isn't given the attention it should get; thus, we see how the author's strong connections with a company that makes pesticides might lead him to keep his mouth shut about it. (Which, I might opine, is not much of a flaw--it says that the author merely didn't go far enough.)
It is not true that, "in the real world", facts that would call a scientist's credibility into question are necessarily significant in evaluating his study. It is only true if there are parts of the study that depend critically on his credibility; and once you've found these parts, you can isolate them and check them yourself (or go ask experts to check them), and either prove the study wrong or validate it.
If you say this is a minor distinction, I say, no, quite the contrary. I think it's important to realize exactly what can and what can't be invalidated by attacks on someone's credibility: it puts a limit on the potential power of character assassination, which should comfort scientists who know they wouldn't have the stomach or resources to resist a campaign against them, and should reduce the power of people with the resources to conduct or threaten such a campaign.
Also, it gives a way for scientists who do have questionable reputations or backgrounds to proceed: present their reports in such a way that there are no parts that fundamentally require faith in the author to believe. A good way to do this is to carefully and generically document everything that you do, so that any other expert in the field could perform the steps himself and verify them. (Come to think of it, this is actually ideal operating practice for any scientist. To take it for granted that a scientist's lack of credibility invalidates his study is actually a grave insult to every scientist who takes pains to produce well-documented and reproducible studies.) We saw how this could be done in the Miller-Rabin test example, where I took "some unspecified calculation I presumably performed" and replaced it with "a C program that anyone should be able to run".
Anyway, for the standards of a news article, it would suffice to mention the questionable, credibility-dependent parts and then explain why we shouldn't take him at his word. It does not suffice to spend a couple of pages attacking him and then mention, halfway through the article, "Oh BTW this is why his credibility makes a difference." It feels very dishonest to me.
I stand by what I said. Personal attacks can be justified, but justify them before you conduct them, dammit. Especially in a news article, where the order of presentation should reflect the order of importance of each point.
By the way, I'm going to run a blanket attack on anyone who responds, "That's all fine and good, but in the real world...". It implies, quite directly, that the speaker's words do not apply to the real world, when usually the speaker intended precisely for them to apply to the real world. In other words, it is a blanket dismissal of everything the speaker said. It might be an expert telling off an annoying novice: "You're completely wrong and I'm not going to bother telling you why."
More commonly--and I'm sure this is how the author of the parent meant it--it's intended as a sort of sympathetic, "Yeah, I wish it were that way too [optionally, "I thought like that when I was idealistic and young"], but, well, you know, the real world is just never as clean and perfect as that." Implied: "The world is a dirty and flawed place, and we who live in it just have to get used to it. And there's not much point in trying to understand it, else I would have explained something instead of merely contradicting you." I think this is profoundly contemptuous of man's ability to understand the world, not to mention pessimistic, and I absolutely reject it.
Maybe you don't mean to imply any of those things. But just bear in mind--I don't think you can argue with this--that it is a way of saying, however politely or sympathetically, "You're completely wrong." With all the implications that go with it; for example, if it is said in response to a serious statement or question, and there is no explanation of why one is completely wrong, then it is pretty dismissive and may be rude or condescending.