Mostly Harmless Econometrics: An Empiricist’s Companion

Key takeaways

content: "@angristMostlyHarmlessEconometrics2008" -file:@angristMostlyHarmlessEconometrics2008

Reading notes

Weights:

A simple rule of thumb for waiting regression is to use weights when they make it more likely that the regression you are estimating is close to the population target. You are trying to estimate.

Waiting by the inverse sampling probability generates estimates that are consistent for the population regression function, even if the sample you have to work with is not a simple random sample.

Finally, an old caution comes to mind. If it ain't broke, don't fix it. The interpretation of the population regression vector is unaffected by heteroscedasticity, so why worry about it? Any efficiency gained from waiting is likely to be modest and incorrect or poorly estimated weights can do more harm than good.