When I use the i.command to run the regression as follows:
Code:
reg coverage i.year i.planning i.finance i.performance,r
coverage | Coef. | St.Err. | t-value | p-value | [95% Conf | Interval] | Sig |
Planning | |||||||
1 planning | Baseline | ||||||
2 planning | -1.088 | .545 | -2.00 | .046 | -2.158 | -.018 | ** |
3 planning | 1.815 | .568 | 3.19 | .001 | .7 | 2.93 | *** |
4 planning | 1.277 | .617 | 2.07 | .039 | .067 | 2.488 | ** |
Finance | |||||||
1 finance | Baseline | ||||||
2 finance | -2.253 | .48 | -4.69 | .000 | -3.195 | -1.31 | *** |
3 finance | -2.326 | .515 | -4.52 | .000 | -3.335 | -1.316 | *** |
4 finance | Omitted | ||||||
Performance | |||||||
1 performance | Baseline | ||||||
2 performance | -3.117 | .517 | -6.03 | .000 | -4.131 | -2.103 | *** |
3 performance | -1.68 | .48 | -3.50 | .000 | -2.623 | -.738 | *** |
*** p<.01, ** p<.05, * p<.1 | |||||||
I am confused about why Finance 4 is omitted. My explanation is because of a high similarity or correlation of coverage between level 1 and level 4.
I try the matrix diagram: and I still cannot infer the clues or reasonings behind this collinearity.
I try the line of mean coverage by 4 levels of Finance, and it seems that the trend of level 1 and level 4 are very common. Can this support the collinearity of Finance 4?
Thanks in advance, and really looking forward to your ideas!
Matrix diagram and line diagram are attached in attachments. Array
Array
0 Response to "i." Command: Omitted dummy variable due to collinearity (NOT the baseline omission)
Post a Comment