The statistical argument is just pseudo-scientific woo. It's bad statistics. To make any kind of probability model (and validate it) you need to have sampled multiple universes and go from there. We only have the one, and nothing in our current experimental methods can change that.

It's basically a slightly twist on Pascal's wager, where you take some unexamined pre-existing phylosophycal beliefs and arbitrarily assign "probabilities" to them. Bleh