[R] predict.glm and continous variables

[line]

Hello.

I am using the predict function to transform logit values from a glm to 
probabilities, (predict(model,type="response",se=T))..
 Everything is going very well for the categorical variables where I get 
one value for the slope of each level and one value for standard error 
for each level.
The problem occurs when I whant to do the same for my continous 
variables. Instead of getting one value for the slope and and one value 
of SE for each variable (like what pops up in the summary in a 
univariate glm) I get a slope and SE value for every datapoint in the 
variable.

I hope someone knows the solution to my problem.

Regards,

Line.