I had a recent discussion with a big data-crunching colleague about statistical validity and hypothesis testing in clinical research. He remarked that if all the data resided in a large dataset, we wouldn't need hypothesis testing because we would have the answers from direct observation. Of course, he works with large public databases analyzing trends in public policy, so he really means "big data." And yes, if we were equivalent to the National Security Agency and could access all the data in every electronic medical record and had the super-computer crunching power, we could probably answer many clinical questions directly and obviate the need for clinical research.
展开▼
