CODEWITHSUNDEEP
rglmcoefficients
Another.Chemist
Another.Chemist

Reputation: 2559

In R, Are valid the given coefficients of glm()?

I got in R after glm():

Coefficients:
         Estimate Std. Error t value Pr(>|t|)    
D_N    -1.405e+05  3.451e+04  -4.072 0.000166 ***
D_q     1.405e+05  3.451e+04   4.072 0.000166 ***
D_Rho  -9.368e-01  2.455e-01  -3.815 0.000375 ***
Dn_N    4.958e+05  1.265e+05   3.919 0.000271 ***
Dn_q   -4.958e+05  1.265e+05  -3.919 0.000271 ***
Dn_Rho  3.777e+00  5.567e-01   6.785  1.3e-08 ***

The question is: The coefficient of D_N and D_q or Dn_N and Dn_q have the same values but with opposite sign.

Is this a valid model? Still both coefficients are the same with opposite sign?

More info: D_N and D_q or Dn_N and Dn_q in the database have the same values but with opposite sign.

Upvotes: 0

Views: 70

Answers (1)

Diego
Diego

Reputation: 356

I'd do the following to verify the model after getting what you just showed:

fit2 <- glm(Y~I(D_q-D_N)+I(Dn_N-Dn_q)+D_Rho+Dn_Rho)

Perhaps then, you could compare both models using Akaike's coefficient:

AIC(yourfit, fit2)

And draw conclussions... Only you know if that makes physical sense... Also, check if the independent variables present some sort of colinerity...

Upvotes: 1

Related Questions