relative survival poisson model

[Matt Piercy]

HI Statalist,
I am trying to model rs using a poisson model as per Royston and Lambert's book Flexible Paremetric Survival Analysis using Stata (2011). I have found the mean/variance is not equal. Does this matter? Is there a way of correcting for this in rs models?
Thanks,
Matt

[Carlo Lazzaro]

Matt:
I would go -vce(cluster id)-.
In addition, you may want to take a look at #4, page 974, -glm- entry, Stata .pdf manual.

[Matt Piercy]

Hi Carlo,
Thanks for your response. I tried that. The problem is that when you use grouped data in rs it removes the id, which is stored as individ.dta. I could you vce(robust) which does work. I also tried scale(x2) which uses pearson to scale se's, I believe both of these are valid. I could also use a Wooldridge residual dispersion parameter, but that would leave me with a QML poisson rather than poisson, but I think this would still be OK (not sure if it breaks the rs decompostion of baseline mort + excess mort). Thanks for your time. Cheers, Matt

[Carlo Lazzaro]

Matt:
I would give -scale(x2)- a shot.

[Matt Piercy]

Thanks Carlo. Great help. Matt