S4E08 Missing Data: The New State Of The Art With Craig Enders
Quantitude
00:00
How Do We Deal With the Missing Not at Random?
The two major frameworks that we could work from are selection modeling and pattern mixture models. You can think about those models as like a mediation process, almost like x influences y and then y influences missingness. And essentially, we're saying that subgroups of people, people with and without data on our dependent variable potentially have different parameter estimates. Maybe the effect of an intervention works differently depending on whether you have data or not on the dependent variable.
Transcript
Play full episode
