1
$\begingroup$

I ran a model with 2 fixed factors that are categorial:

Ind: 10 levels
CRENEAU: 4 levels

For the parameter nu (I am running a gamlss of family BEZI), I get this summary:

Nu link function:  logit   
Nu Coefficients:
                 Estimate  
(Intercept)     2.180e+00  
IndF2           9.259e-01     
IndF3           9.259e-01      
IndF4          -4.184e-01     
IndF5           4.126e-15      
IndT1          -6.434e-01   
IndT2           9.259e-01      
IndT3          -1.623e+00  
IndT4           3.763e-01     
IndT5          -1.412e+00  
CRENEAUEVENING -3.071e-01    
CRENEAUMORNING  2.379e-01   
CRENEAUNOON     5.507e-01   

CRENEAU and Ind are 2 different factors linked to my response. I understand that the intercept is the level of IndF1. But what is the estimate for the missing level of CRENEAU? Is there only one intercept for both factors that is equal ?

$\endgroup$
1
  • $\begingroup$ Yes, the intercept is the value for all of your reference levels. $\endgroup$
    – PBulls
    Oct 18, 2023 at 10:26

1 Answer 1

0
$\begingroup$

If you were to write out the equation, the intercept is the log odds of your outcome variable (Y) when all predictors included in the model are set to zero.

So it is the log odds of Y when none of the nine included IND categories apply and none of the three included CRENEAU categories apply.

Hence, it is the log-odds for individuals in the base category for both categories! So for those in INDF1 and CRENEAUNIGHT (this is me guessing the base categories).

$\endgroup$
2
  • $\begingroup$ Ok thanks. The calculation for the estimate of InfF2 for exemple is inverse logit (estimate IndF2) or inverse logit( intercept + estimate IndF2) ? $\endgroup$ Oct 24, 2023 at 12:36
  • 1
    $\begingroup$ The marginal effect on the log odds of InfF2 is the estimate for IndF2. Because you have another categorical variable, you can't really look at the overall estimate for IndF2 without considering the other variable. For example, the overall log odds for IndF2 and CRENEAUNIGHT (base) is (intercept + estimate IndF2). Likewise, the log odds for IndF2 and CRENEAUEVENING is (intercept + estimate IndF2 + estimate CRENEAUEVENING). $\endgroup$ Oct 24, 2023 at 13:20

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.