S2E02: The Auto-Regressive Cross-Lagged Horror Picture Show
Quantitude
00:00
The Hot Tub Time Machine
In my read of things, there is just a nonstop interest in trying to understand how variables relate over time. And I might say that people aren't spectacularly good at it. If you take six papers that definitively offer guidance and how one should analyze longitudinal data, you can get six diametrically opposed recommendations. Every one of them is either raw score change based or residualized change based. It goes all the way back, Krombach and Furby, where the punchline is, never use raw change scored. There are all these problems. Everything is addressed by residuals.
Transcript
Play full episode
