Given the mass of the evidence, it would be easy to handpick a grossly biased basket. Readers must judge how well I’ve countered this ever-present temptation, but I offer one upfront disclosure: I consciously place extra weight on basic statistics over high-tech alternatives. When relevant experimental evidence is thin or nonexistent (as it usually is), I put my trust in Ordinary Least Squares with control variables. When the results seem questionable, I just seek richer data. This approach isn’t perfect, but it’s easy to understand, easy to compare, and hard to manipulate. High-tech
...more
