Coefficient           Std. Error         t-Statistic        Prob. 
    
C(1)             4.604784          0.092428           49.82044       0.0000
C(2)            -5.46E-05          5.99E-05           -0.912343       0.3622
C(3)             3.41E-08          1.24E-08           2.744363         0.0063
C(4)             -2.67E-12         8.22E-13           -3.250273        0.0013
       
R-squared                    0.727830                   Mean dependent var  4.776104
Adjusted R-squared     0.725726                   S.D. dependent var  0.096750
S.E. of regression        0.050669                   Akaike info criterion  -3.116850
Sum squared resid      0.996135                    Schwarz criterion  -3.076326
Log likelihood              614.9025                   F-statistic  345.8602
Durbin-Watson stat     1.096628                   Prob(F-statistic)  0.000000
    
随着样本个数的增加R-squared和Adjust R增加,但是t-Statistic增加也很快,请问各位大牛这对我的回归方程是否有重要影响.

还有就是我在书上看到拟合度很好的方程式但Log likelihood确是负数. 一般理论不是残差越小，L值越大，越大说明模型越正确吗?