Fixed effect regression — WG/DID estimator and their cluster-robust variance
In the early days of recording, the microphone could only cleanly capture a narrow frequency range. So engineers were like: we’ll just record what the microphone handles. Fair enough. Then the microphone got better. Then digital audio arrived. Then we could capture everything. But by then, scholars had already proven theoretically that music only contains those frequencies. The proof was very elegant. It won a Gramophone award. And to this day, grad students learn that a symphony is just a low-frequency hum with robust standard errors. ...