A RCT is a random clinical trial, an experimental design that is both (generally) the hardest and most expensive to do, but the strongest scientifically and logically. An example would be giving Asprin to a segment of cancer patients and placebo to the rest. Blinded RCTs are where the patient doens't know that they receive the placebo (or the gold standard, as may be the case). Double-blinded tests are where neither the person giving the drug nor the recipient know what drug they get. (This reduces the possibility that the doctor gives the patient some subtle cue revealing what drug they got.)
I'll chime in with what I've heard. Our surgery chief says that a
developing project is best, that is, a project you take form the bench to the patient is best. This shows repeated cycles of development and improvement. I'd also add that this is virtually impossible for a med student to do given the time it'd take.
Also:
Doing a project in your desired field is better than not, but even if you don't have a "desired field" yet, doing
something is better than nothing.
Basic science, while it does (generally) provide evidence of more independent thought, is harder to get started, funded, designed, etc.
But if you don't deal with patients, then you don't have to deal with the
IRB!
If you're smart:
Choose a project that can be spun into many fields: Random example, find a marker for disposition towards diabetes in blood. Many fields deal with diabetes, and (I presume) would benefit from rapid testing. You can spin the value of the project to residencies in internal medicine, genetics, preventative medicine, endocrinology, pulmonology, surgery, or EM.
Low-hanging fruit: Choose projects that are near completion, or already have funding, or are near the end of data collection. Doing stats and a write-up is much less effort, and much more likely to get your name on the by-line than data collection (which may not be finished by the time you leave!)
Choose a project that can be done in time: Like what I've written above, choose something that fits in the time you have set aside. If that means you can do two clinical projects in the time of one basic science, then maybe you should do it.
Choose a project that gets you a pub: This should go without saying, but you're in the big leagues now, and a pub should be part of the signing bonus. (though it would be nice if good pay was part of the deal...) Make sure you will end up with some line on your resume as a result of working for PI So-and-So.
Lastly, once you choose a plan, regardless of your reasons, stick to your guns and forget what everyone else says. You're doing something beyond the normal student, and that shouldn't be looked down upon by anyone.