Hello
I am using -glm- in gamma distribution with the canonical link, which is 1/u
(1/mu). In interpreting the results from this, for a unit rise in predictor x1,
will the predicted value of the response variable change by 1/u2-1/u1, assuming
u1 is the intial value and u2 is the resulting value corresponding to a unit
rise in x1. Or is it just a simple inverse of the coefficient obtained (say
beta1) as we do when a log link is used, when exp(beta1) is calculated. The
numbers I get from simple inversion are too high, so am a little confused.
Appreciate any help.
Arti
*
* For searches and help try:
* http://www.stata.com/help.cgi?search
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
