topic Re: Getting the regression equation for a Poisson regression model in Statistical Procedures

Getting the regression equation for a Poisson regression model

ammarhm — Sat, 11 Apr 2020 22:58:13 GMT

Hi everyone,

I have an unusual request from a paper reviewer who wanted to see 'the poisson model equation' generated by our model.

The code I used to construct the model is

proc glimmix data=have;
class disease sex;
effect yearspline=spline(year / naturalcubic knotmethod=equal(5));
effect agespline=spline(age / naturalcubic knotmethod=equal(5));
model events=disease sex agespline yearspline / dist=poisson offset=logtime s cl link=log; 
run;

Now I dont really know how to write the regression equation based on the output generated from the code above, the output includes estimates for disease group (4 levels) as well as the year and age spline (5 levels each).

Could anyone please direct and help me?

Re: Getting the regression equation for a Poisson regression model

SteveDenham — Mon, 13 Apr 2020 16:25:02 GMT

It is going to be a bit messy because of the two splines, but all of the information you need is provided in the SOLUTION section of the output. If it isn't apparent how the regression equation, with parameter estimates follows, then post the pertinent section and we should be able to figure it out.

Is there any chance the reviewer was asking for the form of the equation, without the parameter estimates, but as a vector or long-hand equation?

Re: Getting the regression equation for a Poisson regression model

Rick_SAS — Tue, 14 Apr 2020 14:41:08 GMT

I wrote an article that might help you to state how the parameter estimates in the spline model relate to the original variables.