It's tempting to trust (and share) stats that reinforce one's preexisting beliefs. But that's precisely when you should dig into and find the original source. And if there is no original source, heavily discount information from whoever "cited" it. There should be a social cost for misinformation.