There is this question that I could not really find an answer to online. Let's say that we do a simple linear regression model with y and x1, and we compute the slope of x. Then, we make a multiple linear regression model with y, x1, and x2. But now the slope of x1 changed. Why is that?
Asked
Active
Viewed 311 times
0
-
Google for "omitted variable bias." – Matthew Gunn Nov 06 '20 at 14:13
-
@MatthewGunn thanks a lot! never heard of that! – Kerem Tuncer Nov 06 '20 at 14:17
-
A related and highly illustrative example is [Simpson's paradox](https://en.wikipedia.org/wiki/Simpson%27s_paradox). – Matthew Gunn Nov 06 '20 at 14:24
-
This is a FAQ: see https://stats.stackexchange.com/search?tab=relevance&q=regression%20coefficient%20change – whuber Nov 06 '20 at 14:51