Timeline for Non-significant in correlation, but significant predictor in regression. How to explain suppression?
Current License: CC BY-SA 4.0
8 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Aug 8, 2020 at 7:01 | history | bumped | CommunityBot | This question has answers that may be good or bad; the system has marked it active so that they can be reviewed. | |
| Apr 6, 2020 at 5:02 | history | bumped | CommunityBot | This question has answers that may be good or bad; the system has marked it active so that they can be reviewed. | |
| Dec 7, 2019 at 20:02 | history | bumped | CommunityBot | This question has answers that may be good or bad; the system has marked it active so that they can be reviewed. | |
| Apr 7, 2019 at 12:24 | answer | added | Peter Flom | timeline score: 1 | |
| Apr 7, 2019 at 1:40 | review | Close votes | |||
| Apr 8, 2019 at 11:48 | |||||
| Apr 6, 2019 at 23:27 | comment | added | behold | Can you print out the correlation matrix for these variables? As your tags suggest, I do think this is because of multicollinearity. Try running a stepwise regression with variables in different orders, e.g. 1st try running A,C,B,D,E and then try running D,E,A,C,B. If correlations are high and also the coeffs change based on ordering a lot, then try using Lasso/LAR to get rid of correlated independent variables. If intuition of variables are not of prime importance, then just take top few principal components for regressing. | |
| Apr 6, 2019 at 22:00 | review | First posts | |||
| Apr 7, 2019 at 1:24 | |||||
| Apr 6, 2019 at 21:59 | history | asked | Callum H | CC BY-SA 4.0 |