Here's a tough question that you'd think would be easier to answer: how do you know that someone you're interested in hiring is any good? The problem, naturally, is that you can't just turn them loose in the lab or in an office for the afternoon to see how they perform. There's a whole industry that claims to help you out with this problem, which tells you how serious it is. The fact that this industry has persisted for decades, though, tells you that the problem does not admit to any simple solution.
Most of these HR consultants have one-size-fits-all techniques that they can (they hope) sell to anyone: pharmaceutical companies, cat litter manufacturers, denture labs, auto body shops, whatever. A favorite is "Behavioral Event Interviewing." where you're supposed to ask the prospective candidate about some tough situation they've been through and how they handled it. "So then," the interviewer is supposed to say, "can you think of a time when you've had a problem getting what you needed from one of your co-workers?" (Just for once, wouldn't you like to hear someone respond: "Those losers? Every day of the week! Why do you think I'm out interviewing?") As you can see, these questions seem to be telegraphed by huge, flashing signals that only the dimmest job candidate could be taken unawares by.
Does anyone actually get any value out of this? Why such methods are supposed to generate any kind of useful (or honest) answer is a mystery. To me, it seems like the old "What's your greatest weakness?" question, to which the traditional answer is, "Darn it all, but I have this tendency to just work too much, you know?" Asking someone about a difficult situation that they handled well is an invitation for them to spin a thrilling tale of obstacles overcome, troubles vanquished, goals attained. Pass the popcorn. Interview techniques that favor the best storytellers don't seem useful to me, unless one of the key requirements of the job you're filling is the ability to spin convincing tall ones. Like, say, marketing or investor relations.
BEI is an example of the attempt to find out how your candidate has actually performed in the past. But there's only so much you can learn, and asking the candidate seems like the least likely method to yield any useful information. There are always personal references, which these days have almost to be given out in hiding, but they're of limited use as well. The biggest problem is that they're almost invariably good. If you ever do call someone who gives out an unfavorable reference, it counts double against the candidate: once for the substance of the recommendation, and again for being clueless enough to pick someone who would give them a bad review.
So if you can't see how the person has been performing, how about measuring their abilities instead? Back in academic chemistry labs, there's always the technique of having a new student try a standard group preparation or something useful from Organic Syntheses. These are all well-tested procedures that a competent organic chemist should be able to reproduce. Even in academia, though, you don't usually have the option of running people through such a test before taking them into the group, but if the new addition struggles through a standard prep, you at least know where to rank them. Unfortunately, in industry there's almost no way to put someone to a practical test like this.
That probably accounts for the puzzle interviews, which have been a feature of the IT field and Wall Street, to the point that they seem to be killing themselves off through overuse. These are the ones where the candidate gets tricky logic problems to solve - you know, where you have a goat, a stick of dynamite, and a greedy pirate and you're trying to get them all through a round manhole cover. (Well, something like that - perhaps I've read too many of these myself.) The idea is that you're testing the interviewee's innate reasoning abilities under pressure. The result, though, is that you end up selecting for the people who've read the most about tricky interview logic problems. If this is a key part of the job you're trying to fill, then feel free. I can't think of a position that could possibly depend on a candidate's ability to game your interview questions, though.
The chemistry equivalent of the puzzle interview is the "Let's go the board" one. These can be massively unpleasant for the interviewee, as they're put through drawing arrows and proposing mechanisms for whatever bizarre reactions have been saved up for them. I've never cared for that sort of thing, actually, even on the giving end. Reaction mechanisms are only a fair surrogate for chemical knowledge, and in my experience some people who excel at electron-pushing have gotten that way through deficiencies in other areas (such as other people being able to stand them, for one). I wouldn't hire someone who just stands their chewing on the chalk, either, but I don't rank mechanistic whizzes as high as some people do.
For a candidate with some experience, I think the better sort of blackboard interview is something more to do with the job requirements. For medicinal chemistry, I'd put up a few structures with their assay numbers, and ask what would be good to try next. Or I'd show some closely related compounds with widely varying blood levels, and ask what to do about them. This sort of thing has the advantage of being tied to the real world. If you want to be cruel about it, you could put up something that no one ever figured out how to solve. Most candidates aren't confident enough to say that they don't know the answer to a question like that, but I'd seriously consider someone who would. You're probably better off with someone who can admit that as a possibility, rather than hiring someone who's convinced that their every idea is golden.
Of course, you can always give up on trying to find out what your candidate has done or what they know, and go straight to the, "If you were a tree, what kind of tree would you be?" questions. I'm sorry to say that I've heard of that one really being asked. The upside of these questions is that they'll help you weed out unsuitable people very quickly; anyone who puts up with one is unsuitable. A quick stunned expression followed by a polite recovery is OK, but a smooth, serious, unfazed response is as big a danger sign as I know. Try it if you dare! Maybe, "If you were a round-bottomed flask . . ." might do the trick. (I'm not sure how you get the interview back on track after a blast on the trumpet like that, but no doubt there are advanced techniques that I haven't yet encountered). And who knows? If it actually works, you can go and start your own consulting firm.