I think the overwhelming consensus has generally been "insignificant" or "no preference" type results.
One thing I would say in support of these experiments that Marshall has been doing is this - He has never claimed to be attempting by the book science. He has never claimed to be using the perfect, lab quality protocol. I agree, that in almost all of these experiments there are probably better ways to truly isolate a particular variable or alter processes to focus on that variable.
However, that is not what he is doing (at least that is not the impression I get). The impression I get is this:
"Here is the way I brew beer....... within my basic set up, my basic process, I wonder what happens if I alter this or that?"
He is not going to revamp his process every time to test a particular variable, because he is going to keep brewing the same basic way. He is not publishing in scholarly journals with lab grade experiments. He is representing an average home brewer and reenacting what might happen if that average home brewer brewed the same beer 2 different ways.
I think the one thing that the experiments do tell us is this - if you have good sanitation and brewing practices - any one variable is probably not going to have a profound effect. Good, bad or great beer is likely the result of a broad interaction of many aspects of brewing - not any one in isolation.