S4E08 Missing Data: The New State Of The Art With Craig Enders
Quantitude
00:00
Do You Use Fimble in Computer Simulations?
There's things that fimble does really well. And then there's things that bimbal does pretty terribly where you're going to expect to see some bias. You know, I think anything with a nonlinearity in it, an interaction term or a curvilinear effect or a multilevel model, perhaps with incomplete covariances are the problem. It's easier than ever to apply missing not at random models to data. Those aren't out of our reach anymore. They're simple to use and not appropriate for every situation,. But certainly some are doing sensitivity analyses that examine the impact on our assumptions about why the data is missing.
Transcript
Play full episode
