I would agree with you that all facts should have some source identified to as to where the "fact" came from. However, not all "facts" are produced directly from an event that correspondes to the "fact". Many times you will find evidence that an individual was marrried, burried, enlisted in the army, dead, born, ... but have no event based documentation. So in these cases a fact can be entered, with a source connected that says that this occured but then note that you have not found any "primary" evidence to back this up. This is one of the reason that an attribute of the source_citation is a Quality (aka CERTAINTY_ASSESSMENT)indicator.
In this case I would enter: 1 = Questionable reliability of evidence (interviews, census, oral genealogies, or potential for bias for example, an autobiography)