0
$\begingroup$

My model is as follows:

model2 <- glm(cbind(ncases, ncontrols) ~ agegp + unclass(tobgp)
                                         + unclass(alcgp),
              data = esoph, family = binomial())
summary(model2)

and my result is:

Call:
glm(formula = cbind(ncases, ncontrols) ~ agegp + unclass(tobgp) + 
     unclass(alcgp), family = binomial(), data = esoph)
 
 Deviance Residuals: 
     Min       1Q   Median       3Q      Max  
 -1.7628  -0.6426  -0.2709   0.3043   2.0421  
 
 Coefficients:
                Estimate Std. Error z value Pr(>|z|)    
 (Intercept)    -4.01097    0.31224 -12.846  < 2e-16 ***
 agegp.L         2.96113    0.65092   4.549 5.39e-06 ***
 agegp.Q        -1.33735    0.58918  -2.270  0.02322 *  
 agegp.C         0.15292    0.44792   0.341  0.73281    
 agegp^4         0.06668    0.30776   0.217  0.82848    
 agegp^5        -0.20288    0.19523  -1.039  0.29872    
 unclass(tobgp)  0.26162    0.08198   3.191  0.00142 ** 
 unclass(alcgp)  0.65308    0.08452   7.727 1.10e-14 ***
 ---
 Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
 
 (Dispersion parameter for binomial family taken to be 1)
 
     Null deviance: 227.241  on 87  degrees of freedom
 Residual deviance:  59.277  on 80  degrees of freedom
 AIC: 222.76
 
 Number of Fisher Scoring iterations: 6

With this results of null deviance being higher than residual deviance, can I conclude that it is a good sign since more than a single parameter explains the model better? Or how else can i interpret this result?

$\endgroup$
0