A 1921 divorce case I'm looking at for an upcoming issue of Casefile Clues indicates several times that the couple was married in 1908. Just because this year is repeated in several places does not make it more likely to be correct. Frequency does not mean accuracy. Errors can easily be repeated and a divorce record ten years after the marriage is not a primary source.
Just because you see something repeated numerous places does not mean it's right. That's true even if legal documents are being used.