He isn't "cooking the data". Or, more accurately, he's the one who isn't cooking the all of the data in one big pot, without regard for whether all of the data are actually describing the same thing.
The point he's making is that there are qualitative distinctions within the data that demand their own discrete analysis. What's misleading here is ignoring distinctions in the data in order to falsely posit a uniform source of causality.
The point he's making is that there are qualitative distinctions within the data that demand their own discrete analysis. What's misleading here is ignoring distinctions in the data in order to falsely posit a uniform source of causality.