I wouldn't worry too much about such a small discrepancy. It could be as simple as a child moving out of his parents house to his own place across the street. Smith is a common name, but Claude isn't. I think we can be certain it's the same Claude Smith. At worst it'd be a cousin, but assume it's the same and work from there. I don't know what software you use, I use Family TreeMaker. When I make an assumption I write that it's an assumption in the person's record and then if it becomes clear later the the assumption is unwarrented it's easy to go back and correct things. I'll see if I can find the children from the 1920c in an overlapping census.