I ran two multiple regressions. The first with alm my variables. The second I removed one variable for which I had missing data for a lot of my cases. How can I explain why in the second multiple regression, an additional two variables become 'negative' in direction?
Asked
Active
Viewed 12 times
0
-
Welcome to CV. This is a faq. If those posts don't help, look at this site search. Or this one about sign changes. – whuber Mar 21 '23 at 22:53
-
See the first two pictures here: https://en.wikipedia.org/wiki/Simpson's_paradox and also see https://en.wikipedia.org/wiki/Omitted_variable_bias – Glen_b Mar 21 '23 at 23:36
-
Can Simpson's Paradox and ommitted variable bias apply even when the coefficients are all insignificant? – LLGG Mar 21 '23 at 23:39