You aren't missing anything. The author's article isn't about the optimal ways to do statistics, it's a criticism of the informal economic discussion that dominates policy, blogs, and pop-economics--and, according to the post, an unfortunate number of actual econ papers--that says things like "economic plan X doesn't work because there hasn't been a change".

The criticism generalizes to any statistical methods, though. You always need to think about the underlying model before doing any kind of statistics in any science.