Very faulty logic. 1 in 5 logged in with github, and 1 in 7 logged in with twitter, and this implies you don't need facebook? Measuring two other things does not have any meaning to another not-measured metric. One should add a fb button and then measure how many log in via fb to make such a statement.
Did you consider BrowserId? I think this has a lot of advantages over OpenId or various proprietary accounts.<p><a href="https://browserid.org/" rel="nofollow">https://browserid.org/</a>