Manual backward regression?

[SAS93]

I'm working on a project I need to use SAS-callable SUDAAN to look at marginal predicted prevalence ratios (relative risk estimates) using weighted survey data. Plan to use backward elimination to build initial models.

SUDAAN does not support stepwise/forward/backward regression, at least not for survey procedures. So we're planning to do it "manually" (which I really didn't know was a thing--I was taught using the SELECTION = option with SAS), so I wanted to practice using normal SAS survey regression procedures.

So my question: when looking at -pvalues, which table am I looking at? I assumed I would be looking at the "Type 3 Analysis of Effects" table, selecting one-by-one the variable to remove that has the highest p-value, but I'm not 100% sure. 

[Reeza]

Look at the CHOOSE and SLECTION options in PROC GLMSELECT. Note that you can have different evaluation criterias, your methodology of selecting the variable is to go backwards, but the criteria of how to choose that last is also something that varies.