S4E19 The Dark Art of Regression Diagnostics
Quantitude
00:00
How Much Would My Regression Coefficient Change if I Drop This Person?
With D F fits, there's one value for each person. And that is if I were to drop this person, how much would my regression coefficient change? If you don't have elevated free radical opioids, right, you want those very small,. Doesn't matter. This is a random sample from a homogeneous population. It doesn't matter. But what if you get a case where it's like, holy crap, if I drop this case, my significant predictor becomes non significant? So I have two questions about all of this. One is a very severous, Snape question. How do you look at this?
Transcript
Play full episode
