I think he just started with that assumption because that's what 538 did. His last analysis assumed one distribution, so my guess is he thought that might be m

I think he just started with that assumption because that's what 538 did. His last analysis assumed one distribution, so my guess is he thought that might be more accurate. Although he didn't give data either way.