Here’s an interesting example showing how technical choices in a regression model can make a big difference in the result. And I’m not talking about “statistical significance,” I’m talking about substantive interpretation. David “should have his own weekly column in … Continue reading