Causal inference when you can't experiment: difference-in-differences and synthetic controls
Linear Digressions
00:00
The Difference in Differences Between Cities
i can't think of any state that's the same or similar enough as california to do this, ye. It would be really hard to come up with another city that's a good pair it's not. Silican valley isn't even really a city. Oes so i would the next step then be you turn off the ads for one of the cities, and you leave the ads on for the other,. And just kind of assume that they're similar enough that you can treat them as almost like an a b test. The difference in those differences is something that you can attribute to the causal impact of your treatment. So we have this notion, the control part comes from the
Play episode from 10:43
Transcript


