[WikiEN-l] Age fabrication and original research

[Rob]

This may have come up before so if there's a previous discussion on en
or here, please direct me to it.

Do we have an official stance on using primary sources like the US
census and the Social Security Death Index to prove a case of [[age
fabrication]]?  My take on it is that it is prohibited original
research, using primary sources to disprove secondary ones, compounded
by the fact that we could easily confuse the subject of the article
with another person of the same or similar name.

If you want to be specific, here it is:  Every published source has a
birthdate of 1918 for the late psychic Jeane Dixon.  However the SSDI
has her birthdate as 1904 and the brother-in-law of her nephew swears
on the talk page that the 1904 date is the correct one.  I think the
1904 is correct, and it's frustrating because likely no journalist or
historian is going to bother publishing something about such a minor
matter, but my opinion is irrelevant and we should defer to published
sources.  Verifiability not truth and all that.  Or should we IAR in
cases like this and go with the "correct" date?