Researchers should take care that they know what questions they're really asking of their data, say two Johns Hopkins scientists.