Dear Statalisters,

I'm facing a problem with variables omitted in the regressions. I have listed my data as a long format.

I'm running the following two regressions:
Xtreg dep after yearafter year, fe vce(r) and
Reg dep after yearafter year

The dependent variable "dep" is either 0 or 1. "After" is a dummy(0 or 1), and "year" has a value from 1 to 5. I also made a interactive term, "yearafter," which is either 0 or 4.

When I run the following regression (fixed effect), I see that yearafter is omitted because of collinearity.


xtreg dep after yearafter year, fe vce(r)
note: yearafter omitted because of collinearity

Fixed-effects (within) regression Number of obs = 4,548
Group variable: pid Number of groups = 3,278

R-sq: Obs per group:
within = 0.0477 min = 1
between = 0.0101 avg = 1.4
overall = 0.0012 max = 4

F(2,3277) = 32.80
corr(u_i, Xb) = -0.1730 Prob > F = 0.0000

(Std. Err. adjusted for 3,278 clusters in pid)
------------------------------------------------------------------------------
| Robust
dep | Coef. Std. Err. t P>|t| [95% Conf. Interval]
-------------+----------------------------------------------------------------
after | -.1523807 .0419597 -3.63 0.000 -.2346506 -.0701108
yearafter | 0 (omitted)
year | .1195637 .0168031 7.12 0.000 .086618 .1525093
_cons | .3755438 .0347421 10.81 0.000 .3074253 .4436623
-------------+----------------------------------------------------------------
sigma_u | .43909427
sigma_e | .47539416
rho | .46036824 (fraction of variance due to u_i)
------------------------------------------------------------------------------


When I run the following regression, however, I see that "after" is omitted.


. reg dep after yearafter year
note: after omitted because of collinearity

Source | SS df MS Number of obs = 4,548
-------------+---------------------------------- F(2, 4545) = 12.97
Model | 5.91446943 2 2.95723471 Prob > F = 0.0000
Residual | 1036.1759 4,545 .227981496 R-squared = 0.0057
-------------+---------------------------------- Adj R-squared = 0.0052
Total | 1042.09037 4,547 .229181959 Root MSE = .47747

------------------------------------------------------------------------------
dep | Coef. Std. Err. t P>|t| [95% Conf. Interval]
-------------+----------------------------------------------------------------
after | 0 (omitted)
yearafter | -.0328489 .0064511 -5.09 0.000 -.0454962 -.0202016
year | .0419783 .0105605 3.98 0.000 .0212745 .0626821
_cons | .5723782 .0232908 24.58 0.000 .5267168 .6180396
------------------------------------------------------------------------------


I don't understand why "after" or "yearafter" is omitted in those regressions. Could someone help me solve this issue? Thanks in advance.