Survey logistic compare Incremental Explanatory Power

HEB1 — Sun, 18 Feb 2024 15:06:40 GMT

I have 3 models: same dependent variable (indicator variabl with 3 group: 1,2,3- no order among them ). I use proc survey logistic.

Here are the 3 modles, and the refrence group for the dependent variable is "2":

1) proc surveylogistic data=lib.data ;
strata industry fyear loc /list ;
where b_a=0 ;
class ac1 (ref='2') Mandatory (ref=first) Enforcement(ref=first) registration (ref=first) / param = ref ;
model ac1 = registration
log_fvalue debt_to_equity ROA_NI/ link=glogit RSQUARE ;
run;

2) proc surveylogistic data=lib.data;
strata industry fyear loc /list ;
where b_a=0 ;
class ac1 (ref='2') Mandatory (ref=first) Enforcement(ref=first) registration (ref=first) / param = ref ;
model ac1 = registration Enforcement Mandatory
log_fvalue debt_to_equity ROA_NI/ link=glogit RSQUARE ;
run;

3) proc surveylogistic data=lib.data;
strata industry fyear loc /list ;
where b_a=0 ;
class ac1 (ref='2') Mandatory (ref=first) Enforcement(ref=first) registration (ref=first) / param = ref ;
model ac1 = X1 x2
registration Enforcement Mandatory
log_fvalue debt_to_equity ROA_NI/ link=glogit RSQUARE;
run;

As you can see model 3 includes more independent variables. I want to Compare the Incremental Explanatory Power of model 3 over model 2 and model 1.

Now i just comare the R sequre of the 3 models, but I want to Compare the Incremental Explanatory Power, using wald statistic for the model or other measure to compare the explenatory power of the model.

What is the right syntax?

tnk

Re: Survey logistic compare Incremental Explanatory Power

SAS_Rob — Mon, 19 Feb 2024 21:14:14 GMT

You could use CONTRAST statements to test the nested models. There are some good examples in the usage note linked below.

24447 - Examples of writing CONTRAST and ESTIMATE statements (sas.com)

topic Survey logistic compare Incremental Explanatory Power in Statistical Procedures

Survey logistic compare Incremental Explanatory Power

Re: Survey logistic compare Incremental Explanatory Power