Sometimes your answers really do come to you in your dreams.
I was the primary investigator in a study comparing two different types of thyroid biopsies, in a population of patients who had both types of biopsies done concurrently. It was a project I 'inherited' from a graduating senior, one who had a pretty notorious reputation for not finishing up what he started. Anyway, I spent about 6 months extracting data, analyzing it and finally wrote up the manuscript. We compared the diagnostic rates of method A to method B separately, so see if method B was really any better than the gold standard A. In the (anticipated) final stage, when I sent my paper to the 2 senior authors to review (you know, the kind of senior authors who don't really say anything or help) one of them suddenly brought up a fatal flaw with my protocol. Apparently pathology frequently just mixes up results between the two; he was concerned the results I had extracted may have been contaminated by the other.
Boy, was I pissed. I don't recall too many times during fellowship when I was as pissed as I was then. I was angry, bitter that I had wasted so much time only to be shot down at the last moment. I was so mad that I had palpitations the whole day, my hands shook. I had a lot of trouble falling asleep that night.
Strangely enough, the answer came to me in my dream that night. I remember it well; I was at work, dressed in my black 3-button suit, and I asked my advisor:
"What I came up with a totally new control group? If the results of A might have been contaminated by B, what if I just looked at the composite diagnostic rate, and compared it to an age-sex matched group who had only biopsy A during the same time period?"
It was so clear I might as well have been there in real life. I woke up the next morning excited. You know, sometimes you have ot drag your sorry ass out of bed at 6 am? Not that day; I leapt out of bed in a hurry, wrote down that idea and then got dressed. I marched right into my advisor's office that morning and laid out that plan.
"You know, that might just work! That's an excellent idea!".
Never mind that I had do find another 380 new patients to make up my control group- at least now my project had a pulse again. 10 months later, my study was finally published (yes, it does take that long sometimes, from the time of first submission to revisions to submissions and then final print).
It's strange sometimes, how answers can sometimes come to you in your dreams.